Scouring

Non-Equilibrium Scour Evolution around an Emerged Structure Exposed to a Transient Wave

일시적인 파도에 노출된 구조에서의 비평형 세굴 결과

Deniz Velioglu Sogut ,Erdinc Sogut ,Ali Farhadzadeh,Tian-Jian Hsu 

Abstract


The present study evaluates the performance of two numerical approaches in estimating non-equilibrium scour patterns around a non-slender square structure subjected to a transient wave, by comparing numerical findings with experimental data. This study also investigates the impact of the structure’s positioning on bed evolution, analyzing configurations where the structure is either attached to the sidewall or positioned at the centerline of the wave flume. The first numerical method treats sediment particles as a distinct continuum phase, directly solving the continuity and momentum equations for both sediment and fluid phases. The second method estimates sediment transport using the quadratic law of bottom shear stress, yielding robust predictions of bed evolution through meticulous calibration and validation. The findings reveal that both methods underestimate vortex-induced near-bed vertical velocities. Deposits formed along vortex trajectories are overestimated by the first method, while the second method satisfactorily predicts the bed evolution beneath these paths. Scour holes caused by wave impingement tend to backfill as the flow intensity diminishes. The second method cannot sufficiently capture this backfilling, whereas the first method adequately reflects the phenomenon. Overall, this study highlights significant variations in the predictive capabilities of both methods in regard to the evolution of non-equilibrium scour at low Keulegan–Carpenter numbers.

Keywords


Keulegan-Carpenter number, Solitary wave, non slender, wave-structure interaction, FLOW-3D, WedWaveFoam

Image_Sacrificial_Pier

Sacrificial Piles as Scour Countermeasures in River Bridges A Numerical Study using FLOW-3D

하천 교량의 파괴 대책으로서 희생파일에 대한 FLOW-3D를 이용한 수치 연구

Mohammad Nazari-Sharabian, Aliasghar Nazari-Sharabian, Moses Karakouzian, Mehrdad Karami

Abstract

Scour is defined as the erosive action of flowing water, as well as the excavating and carrying away materials from beds and banks of streams, and from the vicinity of bridge foundations, which is one of the main causes of river bridge failures. In the present study, implementing a numerical approach, and using the FLOW-3D model that works based on the finite volume method (FVM), the applicability of using sacrificial piles in different configurations in front of a bridge pier as countermeasures against scouring is investigated. In this regard, the numerical model was calibrated based on an experimental study on scouring around an unprotected circular river bridge pier. In simulations, the bridge pier and sacrificial piles were circular, and the riverbed was sandy. In all scenarios, the flow rate was constant and equal to 45 L/s. Furthermore, one to five sacrificial piles were placed in front of the pier in different locations for each scenario. Implementation of the sacrificial piles proved to be effective in substantially reducing the scour depths. The results showed that although scouring occurred in the entire area around the pier, the maximum and minimum scour depths were observed on the sides (using three sacrificial piles located upstream, at three and five times the pier diameter) and in the back (using five sacrificial piles located upstream, at four, six, and eight times the pier diameter) of the pier. Moreover, among scenarios where single piles were installed in front of the pier, installing them at a distance of five times the pier diameter was more effective in reducing scour depths. For other scenarios, in which three piles and five piles were installed, distances of six and four times the pier diameter for the three piles scenario, and four, six, and eight times the pier diameter for the five piles scenario were most effective.

 

Keywords

Scouring; River Bridges; Sacrificial Piles; Finite Volume Method (FVM); FLOW-3D.

 

References


Karakouzian, Chavez, Hayes, and Nazari-Sharabian. “Bulbous Pier: An Alternative to Bridge Pier Extensions as a Countermeasure Against Bridge Deck Splashing.” Fluids 4, no. 3 (July 24, 2019): 140. doi:10.3390/fluids4030140.

Karami, Mehrdad, Abdorreza Kabiri-Samani, Mohammad Nazari-Sharabian, and Moses Karakouzian. “Investigating the Effects of Transient Flow in Concrete-Lined Pressure Tunnels, and Developing a New Analytical Formula for Pressure Wave Velocity.” Tunnelling and Underground Space Technology 91 (September 2019): 102992. doi:10.1016/j.tust.2019.102992.

Karakouzian, Moses, Mohammad Nazari-Sharabian, and Mehrdad Karami. “Effect of Overburden Height on Hydraulic Fracturing of Concrete-Lined Pressure Tunnels Excavated in Intact Rock: A Numerical Study.” Fluids 4, no. 2 (June 19, 2019): 112. doi:10.3390/fluids4020112.

Chiew, Yee-Meng. “Scour protection at bridge piers.” Journal of Hydraulic Engineering 118, no. 9 (1992): 1260-1269. doi:10.1061/(ASCE)0733-9429(1992)118:9(1260).

Shen, Hsieh Wen, Verne R. Schneider, and Susumu Karaki. “Local scour around bridge piers.” Journal of the Hydraulics Division (1969): 1919-1940.

Richardson, E.V., and Davis, S.R. “Evaluating Scour at Bridges”. Hydraulic Engineering Circular. (2001), 18 (HEC-18), Report no. FHWA NHI 01–001, U.S. Department of Transportation, Federal Highway Administration, Washington, DC, USA.

Elsaeed, Gamal, Hossam Elsersawy, and Mohammad Ibrahim. “Scour Evaluation for the Nile River Bends on Rosetta Branch.” Advances in Research 5, no. 2 (January 10, 2015): 1–15. doi:10.9734/air/2015/17380.

Chang, Wen-Yi, Jihn-Sung Lai, and Chin-Lien Yen. “Evolution of scour depth at circular bridge piers.” Journal of Hydraulic Engineering 130, no. 9 (2004): 905-913. doi:10.1061/(ASCE)0733-9429(2004)130:9(905).

Unger, Jens, and Willi H. Hager. “Riprap failure at circular bridge piers.” Journal of Hydraulic Engineering 132, no. 4 (2006): 354-362. doi:10.1061/(ASCE)0733-9429(2006)132:4(354).

Abdeldayem, Ahmed W., Gamal H. Elsaeed, and Ahmed A. Ghareeb. “The effect of pile group arrangements on local scour using numerical models.” Advances in Natural and Applied Sciences 5, no. 2 (2011): 141-146.

Sheppard, D. M., B. Melville, and H. Demir. “Evaluation of Existing Equations for Local Scour at Bridge Piers.” Journal of Hydraulic Engineering 140, no. 1 (January 2014): 14–23. doi:10.1061/(asce)hy.1943-7900.0000800.

Melville, Bruce W., and Anna C. Hadfield. “Use of sacrificial piles as pier scour countermeasures.” Journal of Hydraulic Engineering 125, no. 11 (1999): 1221-1224. doi:10.1061/(ASCE)0733-9429(1999)125:11(1221).

Yao, Weidong, Hongwei An, Scott Draper, Liang Cheng, and John M. Harris. “Experimental Investigation of Local Scour Around Submerged Piles in Steady Current.” Coastal Engineering 142 (December 2018): 27–41. doi:10.1016/j.coastaleng.2018.08.015.

Link, Oscar, Marcelo García, Alonso Pizarro, Hernán Alcayaga, and Sebastián Palma. “Local Scour and Sediment Deposition at Bridge Piers During Floods.” Journal of Hydraulic Engineering 146, no. 3 (March 2020): 04020003. doi:10.1061/(asce)hy.1943-7900.0001696.

Khan, Mujahid, Mohammad Tufail, Muhammad Fahad, Hazi Muhammad Azmathullah, Muhammad Sagheer Aslam, Fayaz Ahmad Khan, and Asif Khan. “Experimental analysis of bridge pier scour pattern.” Journal of Engineering and Applied Sciences 36, no. 1 (2017): 1-12.

Yang, Yifan, Bruce W. Melville, D. M. Sheppard, and Asaad Y. Shamseldin. “Clear-Water Local Scour at Skewed Complex Bridge Piers.” Journal of Hydraulic Engineering 144, no. 6 (June 2018): 04018019. doi:10.1061/(asce)hy.1943-7900.0001458.

Moussa, Yasser Abdallah Mohamed, Tarek Hemdan Nasr-Allah, and Amera Abd-Elhasseb. “Studying the Effect of Partial Blockage on Multi-Vents Bridge Pier Scour Experimentally and Numerically.” Ain Shams Engineering Journal 9, no. 4 (December 2018): 1439–1450. doi:10.1016/j.asej.2016.09.010.

Guan, Dawei, Yee-Meng Chiew, Maoxing Wei, and Shih-Chun Hsieh. “Characterization of Horseshoe Vortex in a Developing Scour Hole at a Cylindrical Bridge Pier.” International Journal of Sediment Research 34, no. 2 (April 2019): 118–124. doi:10.1016/j.ijsrc.2018.07.001.

Dougherty, E.M. “CFD Analysis of Bridge Pier Geometry on Local Scour Potential” (2019). LSU Master’s Theses. 5031.

Vijayasree, B. A., T. I. Eldho, B. S. Mazumder, and N. Ahmad. “Influence of Bridge Pier Shape on Flow Field and Scour Geometry.” International Journal of River Basin Management 17, no. 1 (November 10, 2017): 109–129. doi:10.1080/15715124.2017.1394315.

Farooq, Rashid, and Abdul Razzaq Ghumman. “Impact Assessment of Pier Shape and Modifications on Scouring Around Bridge Pier.” Water 11, no. 9 (August 23, 2019): 1761. doi:10.3390/w11091761.

Link, Oscar, Cristian Castillo, Alonso Pizarro, Alejandro Rojas, Bernd Ettmer, Cristián Escauriaza, and Salvatore Manfreda. “A Model of Bridge Pier Scour During Flood Waves.” Journal of Hydraulic Research 55, no. 3 (November 18, 2016): 310–323. doi:10.1080/00221686.2016.1252802.

Karakouzian, Moses, Mehrdad Karami, Mohammad Nazari-Sharabian, and Sajjad Ahmad. “Flow-Induced Stresses and Displacements in Jointed Concrete Pipes Installed by Pipe Jacking Method.” Fluids 4, no. 1 (February 21, 2019): 34. doi:10.3390/fluids4010034.

Flow Science, Inc. FLOW-3D User’s Manual, Flow Science (2018).

Brethour, J. Modeling Sediment Scour. Flow Science, Santa Fe, NM. (2003).

Brethour, James, and Jeff Burnham. “Modeling sediment erosion and deposition with the FLOW-3D sedimentation & scour model.” Flow Science Technical Note, FSI-10-TN85 (2010): 1-22.

Balouchi, M., and Chamani, M.R. “Investigating the Effect of using a Collar around a Bridge Pier, on the Shape of the Scour Hole”. Proceedings of the First International Conference on Dams and Hydropower (2012) (In Persian).

Bayon, Arnau, Daniel Valero, Rafael García-Bartual, Francisco José Vallés-Morán, and P. Amparo López-Jiménez. “Performance Assessment of OpenFOAM and FLOW-3D in the Numerical Modeling of a Low Reynolds Number Hydraulic Jump.” Environmental Modelling & Software 80 (June 2016): 322–335. doi:10.1016/j.envsoft.2016.02.018.

Aminoroayaie Yamini, O., S. Hooman Mousavi, M. R. Kavianpour, and Azin Movahedi. “Numerical Modeling of Sediment Scouring Phenomenon Around the Offshore Wind Turbine Pile in Marine Environment.” Environmental Earth Sciences 77, no. 23 (November 24, 2018). doi:10.1007/s12665-018-7967-4.

Nazari-Sharabian, Mohammad, Masoud Taheriyoun, Sajjad Ahmad, Moses Karakouzian, and Azadeh Ahmadi. “Water Quality Modeling of Mahabad Dam Watershed–Reservoir System under Climate Change Conditions, Using SWAT and System Dynamics.” Water 11, no. 2 (February 24, 2019): 394. doi:10.3390/w11020394.

DOI: 10.28991/cej-2020-03091531

Estimating maximum initial wave amplitude of subaerial landslide tsunamis: A three-dimensional modelling approach

Estimating maximum initial wave amplitude of subaerial landslide tsunamis: A three-dimensional modelling approach

해저 산사태 쓰나미의 최대 초기 파동 진폭 추정: 3차원 모델링 접근법

Ramtin Sabeti a, Mohammad Heidarzadeh ab

aDepartment of Architecture and Civil Engineering, University of Bath, Bath BA27AY, UK
bHydroCoast Consulting Engineers Ltd, Bath, UK

https://doi.org/10.1016/j.ocemod.2024.102360

Highlights

  • •Landslide travel distance is considered for the first time in a predictive equation.
  • •Predictive equation derived from databases using 3D physical and numerical modeling.
  • •The equation was successfully tested on the 2018 Anak Krakatau tsunami event.
  • •The developed equation using three-dimensional data exhibits a 91 % fitting quality.

Abstract

Landslide tsunamis, responsible for thousands of deaths and significant damage in recent years, necessitate the allocation of sufficient time and resources for studying these extreme natural hazards. This study offers a step change in the field by conducting a large number of three-dimensional numerical experiments, validated by physical tests, to develop a predictive equation for the maximum initial amplitude of tsunamis generated by subaerial landslides. We first conducted a few 3D physical experiments in a wave basin which were then applied for the validation of a 3D numerical model based on the Flow3D-HYDRO package. Consequently, we delivered 100 simulations using the validated model by varying parameters such as landslide volume, water depth, slope angle and travel distance. This large database was subsequently employed to develop a predictive equation for the maximum initial tsunami amplitude. For the first time, we considered travel distance as an independent parameter for developing the predictive equation, which can significantly improve the predication accuracy. The predictive equation was tested for the case of the 2018 Anak Krakatau subaerial landslide tsunami and produced satisfactory results.

Keywords

Tsunami, Subaerial landslide, Physical modelling, Numerical simulation, FLOW-3D HYDRO

1. Introduction and literature review

The Anak Krakatau landslide tsunami on 22nd December 2018 was a stark reminder of the dangers posed by subaerial landslide tsunamis (Ren et al., 2020Mulia et al. 2020a; Borrero et al., 2020Heidarzadeh et al., 2020Grilli et al., 2021). The collapse of the volcano’s southwest side into the ocean triggered a tsunami that struck the Sunda Strait, leading to approximately 450 fatalities (Syamsidik et al., 2020Mulia et al., 2020b) (Fig. 1). As shown in Fig. 1, landslide tsunamis (both submarine and subaerial) have been responsible for thousands of deaths and significant damage to coastal communities worldwide. These incidents underscored the critical need for advanced research into landslide-generated waves to aid in hazard prediction and mitigation. This is further emphasized by recent events such as the 28th of November 2020 landslide tsunami in the southern coast mountains of British Columbia (Canada), where an 18 million m3 rockslide generated a massive tsunami, with over 100 m wave run-up, causing significant environmental and infrastructural damage (Geertsema et al., 2022).

Fig 1

Physical modelling and numerical simulation are crucial tools in the study of landslide-induced waves due to their ability to replicate and analyse the complex dynamics of landslide events (Kim et al., 2020). In two-dimensional (2D) modelling, the discrepancy between dimensions can lead to an artificial overestimation of wave amplification (e.g., Heller and Spinneken, 2015). This limitation is overcome with 3D modelling, which enables the scaled-down representation of landslide-generated waves while avoiding the simplifications inherent in 2D approaches (Erosi et al., 2019). Another advantage of 3D modelling in studying landslide-generated waves is its ability to accurately depict the complex dynamics of wave propagation, including lateral and radial spreading from the slide impact zone, a feature unattainable with 2D models (Heller and Spinneken, 2015).

Physical experiments in tsunami research, as presented by authors such as Romano et al. (2020), McFall and Fritz (2016), and Heller and Spinneken (2015), have supported 3D modelling works through validation and calibration of the numerical models to capture the complexities of wave generation and propagation. Numerical modelling has increasingly complemented experimental approach in tsunami research due to the latter’s time and resource-intensive nature, particularly for 3D models (Li et al., 2019; Kim et al., 2021). Various numerical approaches have been employed, from Eulerian and Lagrangian frameworks to depth-averaged and Navier–Stokes models, enhancing our understanding of tsunami dynamics (Si et al., 2018Grilli et al., 2019Heidarzadeh et al., 20172020Iorio et al., 2021Zhang et al., 2021Kirby et al., 2022Wang et al., 20212022Hu et al., 2022). The sophisticated numerical techniques, including the Particle Finite Element Method and the Immersed Boundary Method, have also shown promising results in modelling highly dynamic landslide scenarios (Mulligan et al., 2020Chen et al., 2020). Among these methods and techniques, FLOW-3D HYDRO stands out in simulating landslide-generated tsunami waves due to its sophisticated technical features such as offering Tru Volume of Fluid (VOF) method for precise free surface tracking (e.g., Sabeti and Heidarzadeh 2022a). TruVOF distinguishes itself through a split Lagrangian approach, adeptly reducing cumulative volume errors in wave simulations by dynamically updating cell volume fractions and areas with each time step. Its intelligent adaptation of time step size ensures precise capture of evolving free surfaces, offering unparalleled accuracy in modelling complex fluid interfaces and behaviour (Flow Science, 2023).

Predictive equations play a crucial role in assessing the potential hazards associated with landslide-generated tsunami waves due to their ability to provide risk assessment and warnings. These equations can offer swift and reasonable evaluations of potential tsunami impacts in the absence of detailed numerical simulations, which can be time-consuming and expensive to produce. Among multiple factors and parameters within a landslide tsunami generation, the initial maximum wave amplitude (Fig. 1) stands out due to its critical role. While it is most likely that the initial wave generated by a landslide will have the highest amplitude, it is crucial to clarify that the term “initial maximum wave amplitude” refers to the highest amplitude within the first set of impulse waves. This parameter is essential in determining the tsunami’s impact severity, with higher amplitudes signalling a greater destructive potential (Sabeti and Heidarzadeh 2022a). Additionally, it plays a significant role in tsunami modelling, aiding in the prediction of wave propagation and the assessment of potential impacts.

In this study, we initially validate the FLOW-3D HYDRO model through a series of physical experiments conducted in a 3D wave tank at University of Bath (UK). Upon confirmation of the model’s accuracy, we use it to systematically vary parameters namely landslide volume, water depth, slope angle, and travel distance, creating an extensive database. Alongside this, we perform a sensitivity analysis on these variables to discern their impacts on the initial maximum wave amplitude. The generated database was consequently applied to derive a non-dimensional predictive equation aimed at estimating the initial maximum wave amplitude in real-world landslide tsunami events.

Two innovations of this study are: (i) The predictive equation of this study is based on a large number of 3D experiments whereas most of the previous equations were based on 2D results, and (ii) For the first time, the travel distance is included in the predictive equation as an independent parameter. To evaluate the performance of our predictive equation, we applied it to a previous real-world subaerial landslide tsunami, i.e., the Anak Krakatau 2018 event. Furthermore, we compare the performance of our predictive equation with other existing equations.

2. Data and methods

The methodology applied in this research is a combination of physical and numerical modelling. Limited physical modelling was performed in a 3D wave basin at the University of Bath (UK) to provide data for calibration and validation of the numerical model. After calibration and validation, the numerical model was employed to model a large number of landslide tsunami scenarios which allowed us to develop a database for deriving a predictive equation.

2.1. Physical experiments

To validate our numerical model, we conducted a series of physical experiments including two sets in a 3D wave basin at University of Bath, measuring 2.50 m in length (WL), 2.60 m in width (WW), and 0.60 m in height (WH) (Fig. 2a). Conducting two distinct sets of experiments (Table 1), each with different setups (travel distance, location, and water depth), provided a robust framework for validation of the numerical model. For wave measurement, we employed a twin wire wave gauge from HR Wallingford (https://equipit.hrwallingford.com). In these experiments, we used a concrete prism solid block, the dimensions of which are outlined in Table 2. In our experiments, we employed a concrete prism solid block with a density of 2600 kg/m3, chosen for its similarity to the natural density of landslides, akin to those observed with the 2018 Anak Krakatau tsunami, where the landslide composition is predominantly solid rather than granular. The block’s form has also been endorsed in prior studies (Watts, 1998Najafi-Jilani and Ataie-Ashtiani, 2008) as a suitable surrogate for modelling landslide-induced waves. A key aspect of our methodology was addressing scale effects, following the guidelines proposed by Heller et al. (2008) as it is described in Table 1. To enhance the reliability and accuracy of our experimental data, we conducted each physical experiment three times which revealed all three experimental waveforms were identical. This repetition was aimed at minimizing potential errors and inconsistencies in laboratory measurements.

Fig 2

Table 1. The locations and other information of the laboratory setups for making landslide-generated waves in the physical wave basin. This table details the specific parameters for each setup, including slope range (α), slide volume (V), kinematic viscosity (ν), water depth (h), travel distance (D), surface tension coefficient of water (σ), Reynolds number (R), Weber number (W), and the precise coordinates of the wave gauges (WG).

Labα(°)V (m³)h (m)D (m)WG’s Location(ν) (m²/s)(σ) (N/m)Acceptable range for avoiding scale effects*Observed values of W and R ⁎⁎
Lab 1452.60 × 10−30.2470.070X1=1.090 m1.01 × 10−60.073R > 3.0 × 105R1 = 3.80 × 105
Y1=1.210 m
W1 = 8.19 × 105
Z1=0.050mW >5.0 × 103
Lab 2452.60 × 10−30.2460.045X2=1.030 m1.01 × 10−60.073R2 = 3.78 × 105
Y2=1.210 mW2 = 8.13 × 105
Z2=0.050 m

The acceptable ranges for avoiding scale effects are based on the study by Heller et al. (2008).⁎⁎

The Reynolds number (R) is given by g0.5h1.5/ν, with ν denoting the kinematic viscosity. The Weber number (W) is W = ρgh2/σ, where σ represents surface tension coefficient and ρ = 1000kg/m3 is the density of water. In our experiments, conducted at a water temperature of approximately 20 °C, the kinematic viscosity (ν) and the surface tension coefficient of water (σ) are 1.01 × 10−6 m²/s and 0.073 N/m, respectively (Kestin et al., 1978).

Table 2. Specifications of the solid block used in physical experiments for generating subaerial landslides in the laboratory.

Solid-block attributesProperty metricsGeometric shape
Slide width (bs)0.26 mImage, table 2
Slide length (ls)0.20 m
Slide thickness (s)0.10 m
Slide volume (V)2.60 × 10−3 m3
Specific gravity, (γs)2.60
Slide weight (ms)6.86 kg

2.2. Numerical simulations applying FLOW-3D hydro

The detailed theoretical framework encompassing the governing equations, the computational methodologies employed, and the specific techniques used for tracking the water surface in these simulations are thoroughly detailed in the study by Sabeti et al. (2024). Here, we briefly explain some of the numerical details. We defined a uniform mesh for our flow domain, carefully crafted with a fine spatial resolution of 0.005 m (i.e., grid size). The dimensions of the numerical model directly matched those of our wave basin used in the physical experiment, being 2.60 m wide, 0.60 m deep, and 2.50 m long (Fig. 2). This design ensures comprehensive coverage of the study area. The output intervals of the numerical model are set at 0.02 s. This timing is consistent with the sampling rates of wave gauges used in laboratory settings. The friction coefficient in the FLOW-3D HYDRO is designated as 0.45. This value corresponds to the Coulombic friction measurements obtained in the laboratory, ensuring that the simulation accurately reflects real-world physical interactions.

In order to simulate the landslide motion, we applied coupled motion objects in FLOW-3D-HYDRO where the dynamics are predominantly driven by gravity and surface friction. This methodology stands in contrast to other models that necessitate explicit inputs of force and torque. This approach ensures that the simulation more accurately reflects the natural movement of landslides, which is heavily reliant on gravitational force and the interaction between sliding surfaces. The stability of the numerical simulations is governed by the Courant Number criterion (Courant et al., 1928), which dictates the maximum time step (Δt) for a given mesh size (Δx) and flow speed (U). According to Courant et al. (1928), this number is required to stay below one to ensure stability of numerical simulations. In our simulations, the Courant number is always maintained below one.

In alignment with the parameters of physical experiments, we set the fluid within the mesh to water, characterized by a density of 1000 kg/m³ at a temperature of 20 °C. Furthermore, we defined the top, front, and back surfaces of the mesh as symmetry planes. The remaining surfaces are designated as wall types, incorporating no-slip conditions to accurately simulate the interaction between the fluid and the boundaries. In terms of selection of an appropriate turbulence model, we selected the k–ω model that showed a better performance than other turbulence methods (e.g., Renormalization-Group) in a previous study (Sabeti et al., 2024). The simulations are conducted using a PC Intel® Core™ i7-10510U CPU with a frequency of 1.80 GHz, and a 16 GB RAM. On this PC, completion of a 3-s simulation required approximately 12.5 h.

2.3. Validation

The FLOW-3D HYDRO numerical model was validated using the two physical experiments (Fig. 3) outlined in Table 1. The level of agreement between observations (Oi) and simulations (Si) is examined using the following equation:(1)�=|��−����|×100where ε represents the mismatch error, Oi denotes the observed laboratory values, and Si represents the simulated values from the FLOW-3D HYDRO model. The results of this validation process revealed that our model could replicate the waves generated in the physical experiments with a reasonable degree of mismatch (ε): 14 % for Lab 1 and 8 % for Lab 2 experiments, respectively (Fig. 3). These values indicate that while the model is not perfect, it provides a sufficiently close approximation of the real-world phenomena.

Fig 3

In terms of mesh efficiency, we varied the mesh size to study sensitivity of the numerical results to mesh size. First, by halving the mesh size and then by doubling it, we repeated the modelling by keeping other parameters unchanged. This analysis guided that a mesh size of ∆x = 0.005 m is the most effective for the setup of this study. The total number of computational cells applying mesh size of 0.005 m is 9.269 × 106.

2.4. The dataset

The validated numerical model was employed to conduct 100 simulations, incorporating variations in four key landslide parameters namely water depth, slope angle, slide volume, and travel distance. This methodical approach was essential for a thorough sensitivity analysis of these variables, and for the creation of a detailed database to develop a predictive equation for maximum initial tsunami amplitude. Within the model, 15 distinct slide volumes were established, ranging from 0.10 × 10−3 m3 to 6.25 × 10−3 m3 (Table 3). The slope angle varied between 35° and 55°, and water depth ranged from 0.24 m to 0.27 m. The travel distance of the landslides was varied, spanning from 0.04 m to 0.07 m. Detailed configurations of each simulation, along with the maximum initial wave amplitudes and dominant wave periods are provided in Table 4.

Table 3. Geometrical information of the 15 solid blocks used in numerical modelling for generating landslide tsunamis. Parameters are: ls, slide length; bs, slide width; s, slide thickness; γs, specific gravity; and V, slide volume.

Solid blockls (m)bs (m)s (m)V (m3)γs
Block-10.3100.2600.1556.25 × 10−32.60
Block-20.3000.2600.1505.85 × 10−32.60
Block-30.2800.2600.1405.10 × 10−32.60
Block-40.2600.2600.1304.39 × 10−32.60
Block-50.2400.2600.1203.74 × 10−32.60
Block-60.2200.2600.1103.15 × 10−32.60
Block-70.2000.2600.1002.60 × 10−32.60
Block-80.1800.2600.0902.11 × 10−32.60
Block-90.1600.2600.0801.66 × 10−32.60
Block-100.1400.2600.0701.27 × 10−32.60
Block-110.1200.2600.0600.93 × 10−32.60
Block-120.1000.2600.0500.65 × 10−32.60
Block-130.0800.2600.0400.41 × 10−32.60
Block-140.0600.2600.0300.23 × 10−32.60
Block-150.0400.2600.0200.10 × 10−32.60

Table 4. The numerical simulation for the 100 tests performed in this study for subaerial solid-block landslide-generated waves. Parameters are aM, maximum wave amplitude; α, slope angle; h, water depth; D, travel distance; and T, dominant wave period. The location of the wave gauge is X=1.030 m, Y=1.210 m, and Z=0.050 m. The properties of various solid blocks are presented in Table 3.

Test-Block Noα (°)h (m)D (m)T(s)aM (m)
1Block-7450.2460.0290.5100.0153
2Block-7450.2460.0300.5050.0154
3Block-7450.2460.0310.5050.0156
4Block-7450.2460.0320.5050.0158
5Block-7450.2460.0330.5050.0159
6Block-7450.2460.0340.5050.0160
7Block-7450.2460.0350.5050.0162
8Block-7450.2460.0360.5050.0166
9Block-7450.2460.0370.5050.0167
10Block-7450.2460.0380.5050.0172
11Block-7450.2460.0390.5050.0178
12Block-7450.2460.0400.5050.0179
13Block-7450.2460.0410.5050.0181
14Block-7450.2460.0420.5050.0183
15Block-7450.2460.0430.5050.0190
16Block-7450.2460.0440.5050.0197
17Block-7450.2460.0450.5050.0199
18Block-7450.2460.0460.5050.0201
19Block-7450.2460.0470.5050.0191
20Block-7450.2460.0480.5050.0217
21Block-7450.2460.0490.5050.0220
22Block-7450.2460.0500.5050.0226
23Block-7450.2460.0510.5050.0236
24Block-7450.2460.0520.5050.0239
25Block-7450.2460.0530.5100.0240
26Block-7450.2460.0540.5050.0241
27Block-7450.2460.0550.5050.0246
28Block-7450.2460.0560.5050.0247
29Block-7450.2460.0570.5050.0248
30Block-7450.2460.0580.5050.0249
31Block-7450.2460.0590.5050.0251
32Block-7450.2460.0600.5050.0257
33Block-1450.2460.0450.5050.0319
34Block-2450.2460.0450.5050.0294
35Block-3450.2460.0450.5050.0282
36Block-4450.2460.0450.5050.0262
37Block-5450.2460.0450.5050.0243
38Block-6450.2460.0450.5050.0223
39Block-7450.2460.0450.5050.0196
40Block-8450.2460.0450.5050.0197
41Block-9450.2460.0450.5050.0198
42Block-10450.2460.0450.5050.0184
43Block-11450.2460.0450.5050.0173
44Block-12450.2460.0450.5050.0165
45Block-13450.2460.0450.4040.0153
46Block-14450.2460.0450.4040.0124
47Block-15450.2460.0450.5050.0066
48Block-7450.2020.0450.4040.0220
49Block-7450.2040.0450.4040.0219
50Block-7450.2060.0450.4040.0218
51Block-7450.2080.0450.4040.0217
52Block-7450.2100.0450.4040.0216
53Block-7450.2120.0450.4040.0215
54Block-7450.2140.0450.5050.0214
55Block-7450.2160.0450.5050.0214
56Block-7450.2180.0450.5050.0213
57Block-7450.2200.0450.5050.0212
58Block-7450.2220.0450.5050.0211
59Block-7450.2240.0450.5050.0208
60Block-7450.2260.0450.5050.0203
61Block-7450.2280.0450.5050.0202
62Block-7450.2300.0450.5050.0201
63Block-7450.2320.0450.5050.0201
64Block-7450.2340.0450.5050.0200
65Block-7450.2360.0450.5050.0199
66Block-7450.2380.0450.4040.0196
67Block-7450.2400.0450.4040.0194
68Block-7450.2420.0450.4040.0193
69Block-7450.2440.0450.4040.0192
70Block-7450.2460.0450.5050.0190
71Block-7450.2480.0450.5050.0189
72Block-7450.2500.0450.5050.0187
73Block-7450.2520.0450.5050.0187
74Block-7450.2540.0450.5050.0186
75Block-7450.2560.0450.5050.0184
76Block-7450.2580.0450.5050.0182
77Block-7450.2590.0450.5050.0183
78Block-7450.2600.0450.5050.0191
79Block-7450.2610.0450.5050.0192
80Block-7450.2620.0450.5050.0194
81Block-7450.2630.0450.5050.0195
82Block-7450.2640.0450.5050.0195
83Block-7450.2650.0450.5050.0197
84Block-7450.2660.0450.5050.0197
85Block-7450.2670.0450.5050.0198
86Block-7450.2700.0450.5050.0199
87Block-7300.2460.0450.5050.0101
88Block-7350.2460.0450.5050.0107
89Block-7360.2460.0450.5050.0111
90Block-7370.2460.0450.5050.0116
91Block-7380.2460.0450.5050.0117
92Block-7390.2460.0450.5050.0119
93Block-7400.2460.0450.5050.0121
94Block-7410.2460.0450.5050.0127
95Block-7420.2460.0450.4040.0154
96Block-7430.2460.0450.4040.0157
97Block-7440.2460.0450.4040.0162
98Block-7450.2460.0450.5050.0197
99Block-7500.2460.0450.5050.0221
100Block-7550.2460.0450.5050.0233

In all these 100 simulations, the wave gauge was consistently positioned at coordinates X=1.09 m, Y=1.21 m, and Z=0.05 m. The dominant wave period for each simulation was determined using the Fast Fourier Transform (FFT) function in MATLAB (MathWorks, 2023). Furthermore, the classification of wave types was carried out using a wave categorization graph according to Sorensen (2010), as shown in Fig. 4a. The results indicate that the majority of the simulated waves are on the border between intermediate and deep-water waves, and they are categorized as Stokes waves (Fig. 4a). Four sample waveforms from our 100 numerical experiments are provided in Fig. 4b.

Fig 4

The dataset in Table 4 was used to derive a new predictive equation that incorporates travel distance for the first time to estimate the initial maximum tsunami amplitude. In developing this equation, a genetic algorithm optimization technique was implemented using MATLAB (MathWorks 2023). This advanced approach entailed the use of genetic algorithms (GAs), an evolutionary algorithm type inspired by natural selection processes (MathWorks, 2023). This technique is iterative, involving selection, crossover, and mutation processes to evolve solutions over several generations. The goal was to identify the optimal coefficients and powers for each landslide parameter in the predictive equation, ensuring a robust and reliable model for estimating maximum wave amplitudes. Genetic Algorithms excel at optimizing complex models by navigating through extensive combinations of coefficients and exponents. GAs effectively identify highly suitable solutions for the non-linear and complex relationships between inputs (e.g., slide volume, slope angle, travel distance, water depth) and the output (i.e., maximum initial wave amplitude, aM). MATLAB’s computational environment enhances this process, providing robust tools for GA to adapt and evolve solutions iteratively, ensuring the precision of the predictive model (Onnen et al., 1997). This approach leverages MATLAB’s capabilities to fine-tune parameters dynamically, achieving an optimal equation that accurately estimates aM. It is important to highlight that the nondimensionalized version of this dataset is employed to develop a predictive equation which enables the equation to reproduce the maximum initial wave amplitude (aM) for various subaerial landslide cases, independent of their dimensional differences (e.g., Heler and Hager 2014Heller and Spinneken 2015Sabeti and Heidarzadeh 2022b). For this nondimensionalization, we employed the water depth (h) to nondimensionalize the slide volume (V/h3) and travel distance (D/h). The slide thickness (s) was applied to nondimensionalize the water depth (h/s).

2.5. Landslide velocity

In discussing the critical role of landslide velocity for simulating landslide-generated waves, we focus on the mechanisms of landslide motion and the techniques used to record landslide velocity in our simulations (Fig. 5). Also, we examine how these methods were applied in two distinct scenarios: Lab 1 and Lab 2 (see Table 1 for their details). Regarding the process of landslide movement, a slide starts from a stationary state, gaining momentum under the influence of gravity and this acceleration continues until the landslide collides with water, leading to a significant reduction in its speed before eventually coming to a stop (Fig. 5) (e.g., Panizzo et al. 2005).

Fig 5

To measure the landslide’s velocity in our simulations, we attached a probe at the centre of the slide, which supplied a time series of the velocity data. The slide’s velocity (vs) peaks at the moment it enters the water (Fig. 5), a point referred to as the impact time (tImp). Following this initial impact, the slides continue their underwater movement, eventually coming to a complete halt (tStop). Given the results in Fig. 5, it can be seen that Lab 1, with its longer travel distance (0.070 m), exhibits a higher peak velocity of 1.89 m/s. This increase in velocity is attributed to the extended travel distance allowing more time for the slide to accelerate under gravity. Whereas Lab 2, featuring a shorter travel distance (0.045 m), records a lower peak velocity of 1.78 m/s. This difference underscores how travel distance significantly influences the dynamics of landslide motion. After reaching the peak, both profiles show a sharp decrease in velocity, marking the transition to submarine motion until the slides come to a complete stop (tStop). There are noticeable differences observable in Fig. 5 between the Lab-1 and Lab-2 simulations, including the peaks at 0.3 s . These variations might stem from the placement of the wave gauge, which differs slightly in each scenario, as well as the water depth’s minor discrepancies and, the travel distance.

2.6. Effect of air entrainment

In this section we examine whether it is required to consider air entrainment for our modelling or not as the FLOW-3D HYDRO package is capable of modelling air entrainment. The process of air entrainment in water during a landslide tsunami and its subsequent transport involve two key components: the quantification of air entrainment at the water surface, and the simulation of the air’s transport within the fluid (Hirt, 2003). FLOW-3D HYDRO employs the air entrainment model to compute the volume of air entrained at the water’s surface utilizing three approaches: a constant density model, a variable density model accounting for bulking, and a buoyancy model that adds the Drift-FLUX mechanism to variable density conditions (Flow Science, 2023). The calculation of the entrainment rate is based on the following equation:(2)�������=������[2(��−�����−2�/���)]1/2where parameters are: Vair, volume of air; Cair, entrainment rate coefficient; As, surface area of fluid; ρ, fluid density; k, turbulent kinetic energy; gn, gravity normal to surface; Lt, turbulent length scale; and σ, surface tension coefficient. The value of k is directly computed from the Reynolds-averaged Navier-Stokes (RANS) (kw) calculations in our model.

In this study, we selected the variable density + Drift-FLUX model, which effectively captures the dynamics of phase separation and automatically activates the constant density and variable density models. This method simplifies the air-water mixture, treating it as a single, homogeneous fluid within each computational cell. For the phase volume fractions f1and f2​, the velocities are expressed in terms of the mixture and relative velocities, denoted as u and ur, respectively, as follows:(3)��1��+�.(�1�)=��1��+�.(�1�)−�.(�1�2��)=0(4)��2��+�.(�2�)=��2��+�.(�2�)−�.(�1�2��)=0

The outcomes from this simulation are displayed in Fig. 6, which indicates that the influence of air entrainment on the generated wave amplitude is approximately 2 %. A value of 0.02 for the entrained air volume fraction means that, in the simulated fluid, approximately 2 % of the volume is composed of entrained air. In other words, for every unit volume of the fluid-air mixture at that location, 2 % is air and the remaining 98 % is water. The configuration of Test-17 (Table 4) was employed for this simulation. While the effect of air entrainment is anticipated to be more significant in models of granular landslide-generated waves (Fritz, 2002), in our simulations we opted not to incorporate this module due to its negligible impact on the results.

Fig 6

3. Results

In this section, we begin by presenting a sequence of our 3D simulations capturing different time steps to illustrate the generation process of landslide-generated waves. Subsequently, we derive a new predictive equation to estimate the maximum initial wave amplitude of landslide-generated waves and assess its performance.

3.1. Wave generation and propagation

To demonstrate the wave generation process in our simulation, we reference Test-17 from Table 4, where we employed Block-7 (Tables 34). In this configuration, the slope angle was set to 45°, with a water depth of 0.246 m and a travel distance at 0.045 m (Fig. 7). At 0.220 s, the initial impact of the moving slide on the water is depicted, marking the onset of the wave generation process (Fig. 7a). Disturbances are localized to the immediate area of impact, with the rest of the water surface remaining undisturbed. At this time, a maximum water particle velocity of 1.0 m/s – 1.2 m/s is seen around the impact zone (Fig. 7d). Moving to 0.320 s, the development of the wave becomes apparent as energy transfer from the landslide to the water creates outwardly radiating waves with maximum water particle velocity of up to around 1.6 m/s – 1.8 m/s (Fig. 7b, e). By the time 0.670 s, the wave has fully developed and is propagating away from the impact point exhibiting maximum water particle velocity of up to 2.0 m/s – 2.1 m/s. Concentric wave fronts are visible, moving outwards in all directions, with a colour gradient signifying the highest wave amplitude near the point of landslide entry, diminishing with distance (Fig. 7c, f).

Fig 7

3.2. Influence of landslide parameters on tsunami amplitude

In this section, we investigate the effects of various landslide parameters namely slide volume (V), water depth (h), slipe angle (α) and travel distance (D) on the maximum initial wave amplitude (aM). Fig. 8 presents the outcome of these analyses. According to Fig. 8, the slide volume, slope angle, and travel distance exhibit a direct relationship with the wave amplitude, meaning that as these parameters increase, so does the amplitude. Conversely, water depth is inversely related to the maximum initial wave amplitude, suggesting that the deeper the water depth, the smaller the maximum wave amplitude will be (Fig. 8b).

Fig 8

Fig. 8a highlights the pronounced impact of slide volume on the aM, demonstrating a direct correlation between the two variables. For instance, in the range of slide volumes we modelled (Fig. 8a), The smallest slide volume tested, measuring 0.10 × 10−3 m3, generated a low initial wave amplitude (aM= 0.0066 m) (Table 4). In contrast, the largest volume tested, 6.25 × 10−3 m3, resulted in a significantly higher initial wave amplitude (aM= 0.0319 m) (Table 4). The extremities of these results emphasize the slide volume’s paramount impact on wave amplitude, further elucidated by their positions as the smallest and largest aM values across all conducted tests (Table 4). This is corroborated by findings from the literature (e.g., Murty, 2003), which align with the observed trend in our simulations.

The slope angle’s influence on aM was smooth. A steady increase of wave amplitude was observed as the slope angle increased (Fig. 8c). In examining travel distance, an anomaly was identified. At a travel distance of 0.047 m, there was an unexpected dip in aM, which deviates from the general increasing trend associated with longer travel distances. This singular instance could potentially be attributed to a numerical error. Beyond this point, the expected pattern of increasing aM with longer travel distances resumes, suggesting that the anomaly at 0.047 m is an outlier in an otherwise consistent trend, and thus this single data point was overlooked while deriving the predictive equation. Regarding the inverse relationship between water depth and wave amplitude, our result (Fig. 8b) is consistent with previous reports by Fritz et al. (2003), (2004), and Watts et al. (2005).

The insights from Fig. 8 informed the architecture of the predictive equation in the next Section, with slide volume, travel distance, and slope angle being multiplicatively linked to wave amplitude underscoring their direct correlations with wave amplitude. Conversely, water depth is incorporated as a divisor, representing its inverse relationship with wave amplitude. This structure encapsulates the dynamics between the landslide parameters and their influence on the maximum initial wave amplitude as discussed in more detail in the next Section.

3.3. Predictive equation

Building on our sensitivity analysis of landslide parameters, as detailed in Section 3.2, and utilizing our nondimensional dataset, we have derived a new predictive equation as follows:(5)��/ℎ=0.015(tan�)0.10(�ℎ3)0.90(�ℎ)0.10(ℎ�)−0.11where, V is sliding volume, h is water depth, α is slope angle, and s is landslide thickness. It is important to note that this equation is valid only for subaerial solid-block landslide tsunamis as all our experiments were for this type of waves. The performance of this equation in predicting simulation data is demonstrated by the satisfactory alignment of data points around a 45° line, indicating its accuracy and reliability with regard to the experimental dataset (Fig. 9). The quality of fit between the dataset and Eq. (5) is 91 % indicating that Eq. (5) represents the dataset very well. Table 5 presents Eq. (5) alongside four other similar equations previously published. Two significant distinctions between our Eq. (5) and these others are: (i) Eq. (5) is derived from 3D experiments, whereas the other four equations are based on 2D experiments. (ii) Unlike the other equations, our Eq. (5) incorporates travel distance as an independent parameter.

Fig 9

Table 5. Performance comparison among our newly-developed equation and existing equations for estimating the maximum initial amplitude (aM) of the 2018 Anak Krakatau subaerial landslide tsunami. Parameters: aM, initial maximum wave amplitude; h, water depth; vs, landslide velocity; V, slide volume; bs, slide width; ls, slide length; s, slide thickness; α, slope angle; and ����, volume of the final immersed landslide. We considered ����= V as the slide volume.

EventPredictive equationsAuthor (year)Observed aM (m) ⁎⁎Calculated aM (m)Error, ε (%) ⁎⁎⁎⁎
2018 Anak Krakatau tsunami (Subaerial landslide) *��/ℎ=1.32���ℎNoda (1970)1341340
��/ℎ=0.667(0.5(���ℎ)2)0.334(���)0.754(���)0.506(�ℎ)1.631Bolin et al. (2014) ⁎⁎⁎13459424334
��/ℎ=0.25(������ℎ2)0.8Robbe-Saule et al. (2021)1343177
��/ℎ=0.4545(tan�)0.062(�ℎ3)0.296(ℎ�)−0.235Sabeti and Heidarzadeh (2022b)1341266
��/ℎ=0.015(tan�)0.10(�ℎ3)0.911(�ℎ)0.10(ℎ�)−0.11This study1341302.9

Geometrical and kinematic parameters of the 2018 Anak Krakatau subaerial landslide based on Heidarzadeh et al. (2020)Grilli et al. (2019) and Grilli et al. (2021)V=2.11 × 107 m3h= 50 m; s= 114 m; α= 45°; ls=1250 m; bs= 2700 m; vs=44.9 m/s; D= 2500 m; aM= 100 m −150 m.⁎⁎

aM= An average value of aM = 134 m is considered in this study.⁎⁎⁎

The equation of Bolin et al. (2014) is based on the reformatted one reported by Lindstrøm (2016).⁎⁎⁎⁎

Error is calculated using Eq. (1), where the calculated aM is assumed as the simulated value.

Additionally, we evaluated the performance of this equation using the real-world data from the 2018 Anak Krakatau subaerial landslide tsunami. Based on previous studies (Heidarzadeh et al., 2020Grilli et al., 20192021), we were able to provide a list of parameters for the subaerial landslide and associated tsunami for the 2018 Anak Krakatau event (see footnote of Table 5). We note that the data of the 2018 Anak Krakatau event was not used while deriving Eq. (5). The results indicate that Eq. (5) predicts the initial amplitude of the 2018 Anak Krakatau tsunami as being 130 m indicating an error of 2.9 % compared to the reported average amplitude of 134 m for this event. This performance indicates an improvement compared to the previous equation reported by Sabeti and Heidarzadeh (2022a) (Table 5). In contrast, the equations from Robbe-Saule et al. (2021) and Bolin et al. (2014) demonstrate higher discrepancies of 4200 % and 77 %, respectively (Table 5). Although Noda’s (1970) equation reproduces the tsunami amplitude of 134 m accurately (Table 5), it is crucial to consider its limitations, notably not accounting for parameters such as slope angle and travel distance.

It is essential to recognize that both travel distance and slope angle significantly affect wave amplitude. In our model, captured in Eq. (5), we integrate the slope angle (α) through the tangent function, i.e., tan α. This choice diverges from traditional physical interpretations that often employ the cosine or sine function (e.g., Heller and Hager, 2014Watts et al., 2003). We opted for the tangent function because it more effectively reflects the direct impact of slope steepness on wave generation, yielding superior estimations compared to conventional methods.

The significance of this study lies in its application of both physical and numerical 3D experiments and the derivation of a predictive equation based on 3D results. Prior research, e.g. Heller et al. (2016), has reported notable discrepancies between 2D and 3D wave amplitudes, highlighting the important role of 3D experiments. It is worth noting that the suitability of applying an equation derived from either 2D or 3D data depends on the specific geometry and characteristics inherent in the problem being addressed. For instance, in the case of a long, narrow dam reservoir, an equation derived from 2D data would likely be more suitable. In such contexts, the primary dynamics of interest such as flow patterns and potential wave propagation are predominantly two-dimensional, occurring along the length and depth of the reservoir. This simplification to 2D for narrow dam reservoirs allows for more accurate modelling of these dynamics.

This study specifically investigates waves initiated by landslides, focusing on those characterized as solid blocks instead of granular flows, with slope angles confined to a range of 25° to 60°. We acknowledge the additional complexities encountered in real-world scenarios, such as dynamic density and velocity of landslides, which could affect the estimations. The developed equation in this study is specifically designed to predict the maximum initial amplitude of tsunamis for the aforementioned specified ranges and types of landslides.

4. Conclusions

Both physical and numerical experiments were undertaken in a 3D wave basin to study solid-block landslide-generated waves and to formulate a predictive equation for their maximum initial wave amplitude. At the beginning, two physical experiments were performed to validate and calibrate a 3D numerical model, which was subsequently utilized to generate 100 experiments by varying different landslide parameters. The generated database was then used to derive a predictive equation for the maximum initial wave amplitude of landslide tsunamis. The main features and outcomes are:

  • •The predictive equation of this study is exclusively derived from 3D data and exhibits a fitting quality of 91 % when applied to the database.
  • •For the first time, landslide travel distance was considered in the predictive equation. This inclusion provides more accuracy and flexibility for applying the equation.
  • •To further evaluate the performance of the predictive equation, it was applied to a real-world subaerial landslide tsunami (i.e., the 2018 Anak Krakatau event) and delivered satisfactory performance.

CRediT authorship contribution statement

Ramtin Sabeti: Conceptualization, Methodology, Validation, Software, Visualization, Writing – review & editing. Mohammad Heidarzadeh: Methodology, Data curation, Software, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Funding

RS is supported by the Leverhulme Trust Grant No. RPG-2022-306. MH is funded by open funding of State Key Lab of Hydraulics and Mountain River Engineering, Sichuan University, grant number SKHL2101. We acknowledge University of Bath Institutional Open Access Fund. MH is also funded by the Great Britain Sasakawa Foundation grant no. 6217 (awarded in 2023).

Acknowledgements

Authors are sincerely grateful to the laboratory technician team, particularly Mr William Bazeley, at the Faculty of Engineering, University of Bath for their support during the laboratory physical modelling of this research. We appreciate the valuable insights provided by Mr. Brian Fox (Senior CFD Engineer at Flow Science, Inc.) regarding air entrainment modelling in FLOW-3D HYDRO. We acknowledge University of Bath Institutional Open Access Fund.

Data availability

  • All data used in this study are given in the body of the article.

References

Coupled CFD-DEM simulation of interfacial fluid–particle interaction during binder jet 3D printing

Coupled CFD-DEM simulation of interfacial fluid–particle interaction during binder jet 3D printing

바인더 제트 3D 프린팅 중 계면 유체-입자 상호 작용에 대한 CFD-DEM 결합 시뮬레이션

Joshua J. Wagner, C. Fred Higgs III

https://doi.org/10.1016/j.cma.2024.116747

Abstract

The coupled dynamics of interfacial fluid phases and unconstrained solid particles during the binder jet 3D printing process govern the final quality and performance of the resulting components. The present work proposes a computational fluid dynamics (CFD) and discrete element method (DEM) framework capable of simulating the complex interfacial fluid–particle interaction that occurs when binder microdroplets are deposited into a powder bed. The CFD solver uses a volume-of-fluid (VOF) method for capturing liquid–gas multifluid flows and relies on block-structured adaptive mesh refinement (AMR) to localize grid refinement around evolving fluid–fluid interfaces. The DEM module resolves six degrees of freedom particle motion and accounts for particle contact, cohesion, and rolling resistance. Fully-resolved CFD-DEM coupling is achieved through a fictitious domain immersed boundary (IB) approach. An improved method for enforcing three-phase contact lines with a VOF-IB extension technique is introduced. We present several simulations of binder jet primitive formation using realistic process parameters and material properties. The DEM particle systems are experimentally calibrated to reproduce the cohesion behavior of physical nickel alloy powder feedstocks. We demonstrate the proposed model’s ability to resolve the interdependent fluid and particle dynamics underlying the process by directly comparing simulated primitive granules with one-to-one experimental counterparts obtained from an in-house validation apparatus. This computational framework provides unprecedented insight into the fundamental mechanisms of binder jet 3D printing and presents a versatile new approach for process parameter optimization and defect mitigation that avoids the inherent challenges of experiments.

바인더 젯 3D 프린팅 공정 중 계면 유체 상과 구속되지 않은 고체 입자의 결합 역학이 결과 구성 요소의 최종 품질과 성능을 좌우합니다. 본 연구는 바인더 미세액적이 분말층에 증착될 때 발생하는 복잡한 계면 유체-입자 상호작용을 시뮬레이션할 수 있는 전산유체역학(CFD) 및 이산요소법(DEM) 프레임워크를 제안합니다.

CFD 솔버는 액체-가스 다중유체 흐름을 포착하기 위해 VOF(유체량) 방법을 사용하고 블록 구조 적응형 메쉬 세분화(AMR)를 사용하여 진화하는 유체-유체 인터페이스 주위의 그리드 세분화를 국지화합니다. DEM 모듈은 6개의 자유도 입자 운동을 해결하고 입자 접촉, 응집력 및 구름 저항을 설명합니다.

완전 분해된 CFD-DEM 결합은 가상 도메인 침지 경계(IB) 접근 방식을 통해 달성됩니다. VOF-IB 확장 기술을 사용하여 3상 접촉 라인을 강화하는 향상된 방법이 도입되었습니다. 현실적인 공정 매개변수와 재료 특성을 사용하여 바인더 제트 기본 형성에 대한 여러 시뮬레이션을 제시합니다.

DEM 입자 시스템은 물리적 니켈 합금 분말 공급원료의 응집 거동을 재현하기 위해 실험적으로 보정되었습니다. 우리는 시뮬레이션된 기본 과립과 내부 검증 장치에서 얻은 일대일 실험 대응물을 직접 비교하여 프로세스의 기본이 되는 상호 의존적인 유체 및 입자 역학을 해결하는 제안된 모델의 능력을 보여줍니다.

이 계산 프레임워크는 바인더 제트 3D 프린팅의 기본 메커니즘에 대한 전례 없는 통찰력을 제공하고 실험에 내재된 문제를 피하는 공정 매개변수 최적화 및 결함 완화를 위한 다용도의 새로운 접근 방식을 제시합니다.

Introduction

Binder jet 3D printing (BJ3DP) is a powder bed additive manufacturing (AM) technology capable of fabricating geometrically complex components from advanced engineering materials, such as metallic superalloys and ultra-high temperature ceramics [1], [2]. As illustrated in Fig. 1(a), the process is comprised of many repetitive print cycles, each contributing a new cross-sectional layer on top of a preceding one to form a 3D CAD-specified geometry. The feedstock material is first delivered from a hopper to a build plate and then spread into a thin layer by a counter-rotating roller. After powder spreading, a print head containing many individual inkjet nozzles traverses over the powder bed while precisely jetting binder microdroplets onto select regions of the spread layer. Following binder deposition, the build plate lowers by a specified layer thickness, leaving a thin void space at the top of the job box that the subsequent powder layer will occupy. This cycle repeats until the full geometries are formed layer by layer. Powder bed fusion (PBF) methods follow a similar procedure, except they instead use a laser or electron beam to selectively melt and fuse the powder material. Compared to PBF, binder jetting offers several distinct advantages, including faster build rates, enhanced scalability for large production volumes, reduced machine and operational costs, and a wider selection of suitable feedstock materials [2]. However, binder jetted parts generally possess inferior mechanical properties and reduced dimensional accuracy [3]. As a result, widescale adoption of BJ3DP to fabricate high-performance, mission-critical components, such as those common to the aerospace and defense sectors, is contingent on novel process improvements and innovations [4].

A major obstacle hindering the advancement of BJ3DP is our limited understanding of how various printing parameters and material properties collectively influence the underlying physical mechanisms of the process and their effect on the resulting components. To date, the vast majority of research efforts to uncover these relationships have relied mainly on experimental approaches [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], which are often expensive and time-consuming and have inherent physical restrictions on what can be measured and observed. For these reasons, there is a rapidly growing interest in using computational models to circumvent the challenges of experimental investigations and facilitate a deeper understanding of the process’s fundamental phenomena. While significant progress has been made in developing and deploying numerical frameworks aimed at powder spreading [20], [21], [22], [23], [24], [25], [26], [27] and sintering [28], [29], [30], [31], [32], simulating the interfacial fluid–particle interaction (IFPI) in the binder deposition stage is still in its infancy. In their exhaustive review, Mostafaei et al. [2] point out the lack of computational models capable of resolving the coupled fluid and particle dynamics associated with binder jetting and suggest that the development of such tools is critical to further improving the process and enhancing the quality of its end-use components.

We define IFPI as a multiphase flow regime characterized by immiscible fluid phases separated by dynamic interfaces that intersect the surfaces of moving solid particles. As illustrated in Fig. 1(b), an elaborate IFPI occurs when a binder droplet impacts the powder bed in BJ3DP. The momentum transferred from the impacting droplet may cause powder compaction, cratering, and particle ejection. These ballistic disturbances can have deleterious effects on surface texture and lead to the formation of large void spaces inside the part [5], [13]. After impact, the droplet spreads laterally on the bed surface and vertically into the pore network, driven initially by inertial impact forces and then solely by capillary action [33]. Attractive capillary forces exerted on mutually wetted particles tend to draw them inward towards each other, forming a packed cluster of bound particles referred to as a primitive [34]. A single-drop primitive is the most fundamental building element of a BJ3DP part, and the interaction leading to its formation has important implications on the final part characteristics, such as its mechanical properties, resolution, and dimensional accuracy. Generally, binder droplets are deposited successively as the print head traverses over the powder bed. The traversal speed and jetting frequency are set such that consecutive droplets coalesce in the bed, creating a multi-drop primitive line instead of a single-drop primitive granule. The binder must be jetted with sufficient velocity to penetrate the powder bed deep enough to provide adequate interlayer binding; however, a higher impact velocity leads to more pronounced ballistic effects.

A computational framework equipped to simulate the interdependent fluid and particle dynamics in BJ3DP would allow for unprecedented observational and measurement capability at temporal and spatial resolutions not currently achievable by state-of-the-art imaging technology, namely synchrotron X-ray imaging [13], [14], [18], [19]. Unfortunately, BJ3DP presents significant numerical challenges that have slowed the development of suitable modeling frameworks; the most significant of which are as follows:

  • 1.Incorporating dynamic fluid–fluid interfaces with complex topological features remains a nontrivial task for standard mesh-based CFD codes. There are two broad categories encompassing the methods used to handle interfacial flows: interface tracking and interface capturing [35]. Interface capturing techniques, such as the popular volume-of-fluid (VOF) [36] and level-set methods [37], [38], are better suited for problems with interfaces that become heavily distorted or when coalescence and fragmentation occur frequently; however, they are less accurate in resolving surface tension and boundary layer effects compared to interface tracking methods like front-tracking [39], arbitrary Lagrangian–Eulerian [40], and space–time finite element formulations [41]. Since interfacial forces become increasingly dominant at decreasing length scales, inaccurate surface tension calculations can significantly deteriorate the fidelity of IFPI simulations involving <100 μm droplets and particles.
  • 2.Dynamic powder systems are often modeled using the discrete element method (DEM) introduced by Cundall and Strack [42]. For IFPI problems, a CFD-DEM coupling scheme is required to exchange information between the fluid and particle solvers. Fully-resolved CFD-DEM coupling suggests that the flow field around individual particle surfaces is resolved on the CFD mesh [43], [44]. In contrast, unresolved coupling volume averages the effect of the dispersed solid phase on the continuous fluid phases [45], [46], [47], [48]. Comparatively, the former is computationally expensive but provides detailed information about the IFPI in question and is more appropriate when contact line dynamics are significant. However, since the pore structure of a powder bed is convoluted and evolves with time, resolving such solid–fluid interfaces on a computational mesh presents similar challenges as fluid–fluid interfaces discussed in the previous point. Although various algorithms have been developed to deform unstructured meshes to accommodate moving solid surfaces (see Bazilevs et al. [49] for an overview of such methods), they can be prohibitively expensive when frequent topology changes require mesh regeneration rather than just modification through nodal displacement. The pore network in a powder bed undergoes many topology changes as particles come in and out of contact with each other, constantly closing and opening new flow channels. Non-body-conforming structured grid approaches that rely on immersed boundary (IB) methods to embed the particles in the flow field can be better suited for such cases [50]. Nevertheless, accurately representing these complex pore geometries on Cartesian grids requires extremely high mesh resolutions, which can impose significant computational costs.
  • 3.Capillary effects depend on the contact angle at solid–liquid–gas intersections. Since mesh nodes do not coincide with a particle surface when using an IB method on structured grids, imposing contact angle boundary conditions at three-phase contact lines is not straightforward.

While these issues also pertain to PBF process modeling, resolving particle motion is generally less crucial for analyzing melt pool dynamics compared to primitive formation in BJ3DP. Therefore, at present, the vast majority of computational process models of PBF assume static powder beds and avoid many of the complications described above, see, e.g., [51], [52], [53], [54], [55], [56], [57], [58], [59]. Li et al. [60] presented the first 2D fully-resolved CFD-DEM simulations of the interaction between the melt pool, powder particles, surrounding gas, and metal vapor in PBF. Following this work, Yu and Zhao [61], [62] published similar melt pool IFPI simulations in 3D; however, contact line dynamics and capillary forces were not considered. Compared to PBF, relatively little work has been published regarding the computational modeling of binder deposition in BJ3DP. Employing the open-source VOF code Gerris [63], Tan [33] first simulated droplet impact on a powder bed with appropriate binder jet parameters, namely droplet size and impact velocity. However, similar to most PBF melt pool simulations described in the current literature, the powder bed was fixed in place and not allowed to respond to the interacting fluid phases. Furthermore, a simple face-centered cubic packing of non-contacting, monosized particles was considered, which does not provide a realistic pore structure for AM powder beds. Building upon this approach, we presented a framework to simulate droplet impact on static powder beds with more practical particle size distributions and packing arrangements [64]. In a study similar to [33], [64], Deng et al. [65] used the VOF capability in Ansys Fluent to examine the lateral and vertical spreading of a binder droplet impacting a fixed bimodal powder bed with body-centered packing. Li et al. [66] also adopted Fluent to conduct 2D simulations of a 100 μm diameter droplet impacting substrates with spherical roughness patterns meant to represent the surface of a simplified powder bed with monosized particles. The commercial VOF-based software FLOW-3D offers an AM module centered on process modeling of various AM technologies, including BJ3DP. However, like the above studies, particle motion is still not considered in this codebase. Ur Rehman et al. [67] employed FLOW-3D to examine microdroplet impact on a fixed stainless steel powder bed. Using OpenFOAM, Erhard et al. [68] presented simulations of different droplet impact spacings and patterns on static sand particles.

Recently, Fuchs et al. [69] introduced an impressive multipurpose smoothed particle hydrodynamics (SPH) framework capable of resolving IFPI in various AM methods, including both PBF and BJ3DP. In contrast to a combined CFD-DEM approach, this model relies entirely on SPH meshfree discretization of both the fluid and solid governing equations. The authors performed several prototype simulations demonstrating an 80 μm diameter droplet impacting an unconstrained powder bed at different speeds. While the powder bed responds to the hydrodynamic forces imparted by the impacting droplet, the particle motion is inconsistent with experimental time-resolved observations of the process [13]. Specifically, the ballistic effects, such as particle ejection and bed deformation, were drastically subdued, even in simulations using a droplet velocity ∼ 5× that of typical jetting conditions. This behavior could be caused by excessive damping in the inter-particle contact force computations within their SPH framework. Moreover, the wetted particles did not appear to be significantly influenced by the strong capillary forces exerted by the binder as no primitive agglomeration occurred. The authors mention that the objective of these simulations was to demonstrate their codebase’s broad capabilities and that some unrealistic process parameters were used to improve computational efficiency and stability, which could explain the deviations from experimental observations.

In the present paper, we develop a novel 3D CFD-DEM numerical framework for simulating fully-resolved IFPI during binder jetting with realistic material properties and process parameters. The CFD module is based on the VOF method for capturing binder–air interfaces. Surface tension effects are realized through the continuum surface force (CSF) method with height function calculations of interface curvature. Central to our fluid solver is a proprietary block-structured AMR library with hierarchical octree grid nesting to focus enhanced grid resolution near fluid–fluid interfaces. The GPU-accelerated DEM module considers six degrees of freedom particle motion and includes models based on Hertz-Mindlin contact, van der Waals cohesion, and viscoelastic rolling resistance. The CFD and DEM modules are coupled to achieve fully-resolved IFPI using an IB approach in which Lagrangian solid particles are mapped to the underlying Eulerian fluid mesh through a solid volume fraction field. An improved VOF-IB extension algorithm is introduced to enforce the contact angle at three-phase intersections. This provides robust capillary flow behavior and accurate computations of the fluid-induced forces and torques acting on individual wetted particles in densely packed powder beds.

We deploy our integrated codebase for direct numerical simulations of single-drop primitive formation with powder beds whose particle size distributions are generated from corresponding laboratory samples. These simulations use jetting parameters similar to those employed in current BJ3DP machines, fluid properties that match commonly used aqueous polymeric binders, and powder properties specific to nickel alloy feedstocks. The cohesion behavior of the DEM powder is calibrated based on the angle of repose of the laboratory powder systems. The resulting primitive granules are compared with those obtained from one-to-one experiments conducted using a dedicated in-house test apparatus. Finally, we demonstrate how the proposed framework can simulate more complex and realistic printing operations involving multi-drop primitive lines.

Section snippets

Mathematical description of interfacial fluid–particle interaction

This section briefly describes the governing equations of fluid and particle dynamics underlying the CFD and DEM solvers. Our unified framework follows an Eulerian–Lagrangian approach, wherein the Navier–Stokes equations of incompressible flow are discretized on an Eulerian grid to describe the motion of the binder liquid and surrounding gas, and the Newton–Euler equations account for the positions and orientations of the Lagrangian powder particles. The mathematical foundation for

CFD solver for incompressible flow with multifluid interfaces

This section details the numerical methodology used in our CFD module to solve the Navier–Stokes equations of incompressible flow. First, we introduce the VOF method for capturing the interfaces between the binder and air phases. This approach allows us to solve the fluid dynamics equations considering only a single continuum field with spatial and temporal variations in fluid properties. Next, we describe the time integration procedure using a fractional-step projection algorithm for

DEM solver for solid particle dynamics

This section covers the numerical procedure for tracking the motion of individual powder particles with DEM. The Newton–Euler equations (Eqs. (10), (11)) are ordinary differential equations (ODEs) for which many established numerical integrators are available. In general, the most challenging aspects of DEM involve processing particle collisions in a computationally efficient manner and dealing with small time step constraints that result from stiff materials, such as metallic AM powders. The

Unified CFD-DEM solver

The preceding sections have introduced the CFD and DEM solution algorithms separately. Here, we discuss the integrated CFD-DEM solution algorithm and related details.

Binder jet process modeling and validation experiments

In this section, we deploy our CFD-DEM framework to simulate the IFPI occurring during the binder droplet deposition stage of the BJ3DP process. The first simulations attempt to reproduce experimental single-drop primitive granules extracted from four nickel alloy powder samples with varying particle size distributions. The experiments are conducted with a dedicated in-house test apparatus that allows for the precision deposition of individual binder microdroplets into a powder bed sample. The

Conclusions

This paper introduces a coupled CFD-DEM framework capable of fully-resolved simulation of the interfacial fluid–particle interaction occurring in the binder jet 3D printing process. The interfacial flow of binder and surrounding air is captured with the VOF method and surface tension effects are incorporated using the CSF technique augmented by height function curvature calculations. Block-structured AMR is employed to provide localized grid refinement around the evolving liquid–gas interface.

CRediT authorship contribution statement

Joshua J. Wagner: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Visualization, Writing – original draft, Writing – review & editing. C. Fred Higgs III: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work was supported by a NASA Space Technology Research Fellowship, United States of America, Grant No. 80NSSC19K1171. Partial support was also provided through an AIAA Foundation Orville, USA and Wilbur Wright Graduate Award, USA . The authors would like to gratefully acknowledge Dr. Craig Smith of NASA Glenn Research Center for the valuable input he provided on this project.

References (155)

그림 12: 시간 경과에 따른 속도 카운터: 30초 그림 13: 시간 경과에 따른 속도 카운터: 20초

Gemelo digital del puente de Kalix: cargas estructurales de futuros eventos climáticos extremos

Kalix Bridge 디지털 트윈: 미래 극한 기후 현상으로 인한 구조적 부하

Este documento está relacionado con un proyecto en curso para el cual se está desarrollando e implementando un gemelo digital estructural del puente de Kalix en Suecia.
이 문서는 스웨덴 Kalix 교량의 구조적 디지털 트윈이 개발 및 구현되고 있는 진행 중인 프로젝트와 관련이 있습니다.

Autores: Mahyar Kazemian1, Sajad Nikdel2, Mehrnaz MohammadEsmaeili3, Vahid Nik4, Kamyab Zandi*5

RESUMEN Las cargas ambientales, como el viento y el caudal de los ríos, juegan un papel esencial en el diseño y evaluación estructural de puentes de grandes luces. El cambio climático y los eventos climáticos extremos son amenazas para la confiabilidad y seguridad de la red de transporte.

Esto ha llevado a una creciente demanda de modelos de gemelos digitales para investigar la resistencia de los puentes en condiciones climáticas extremas. El puente de Kalix, construido sobre el río Kalix en Suecia en 1956, se utiliza como banco de pruebas en este contexto.

La estructura del puente, realizada en hormigón postensado, consta de cinco vanos, siendo el más largo de 94 m. En este estudio, las características aerodinámicas y los valores extremos de la simulación numérica del viento, como la presión en la superficie, se obtienen utilizando la simulación de remolinos desprendidos retardados (DDES) de Spalart-Allmaras como un enfoque de turbulencia RANS-LES híbrido que es práctico y computacionalmente eficiente para cerca de la pared densidad de malla impuesta por el método LES.

La presión del viento en la superficie se obtiene para tres escenarios climáticos extremos, que incluyen un clima con mucho viento, un clima extremadamente frío y el valor de cálculo para un período de retorno de 3000 años. El resultado indica diferencias significativas en la presión del viento en la superficie debido a las capas de tiempo que provienen de la simulación del flujo de viento transitorio. Para evaluar el comportamiento estructural en el escenario de viento crítico, se considera el valor más alto de presión en la superficie para cada escenario.

Además, se realiza un estudio hidrodinámico en los pilares del puente, en el que se simula el flujo del río por el método VOF, y se examina el proceso de movimiento del agua alrededor de los pilares de forma transitoria y en diferentes momentos. En cada una de las superficies del pilar se calcula la presión superficial aplicada por el caudal del río con el caudal volumétrico más alto registrado.

Para simular el flujo del río, se ha utilizado la información y las condiciones meteorológicas registradas en períodos anteriores. Los resultados muestran que la presión en la superficie en el momento en que el flujo del río golpea los pilares es mucho mayor que en los momentos posteriores. Esta cantidad de presión se puede usar como carga crítica en los cálculos de interacción fluido-estructura (FSI).

Finalmente, para ambas secciones, la presión en la superficie del viento, el campo de velocidades con respecto a las líneas de sondas auxiliares, los contornos del movimiento circunferencial del agua alrededor de los pilares y el diagrama de presión en ellos se informan en diferentes intervalos de tiempo.

요약 바람, 강의 흐름과 같은 환경 하중은 장대 교량의 설계 및 구조 평가에 필수적인 역할을 합니다. 기후 변화와 기상 이변은 교통 네트워크의 신뢰성과 보안에 위협이 됩니다.

이로 인해 극한 기상 조건에서 교량의 복원력을 조사하기 위한 디지털 트윈 모델에 대한 수요가 증가했습니다. 1956년 스웨덴 칼릭스 강 위에 건설된 칼릭스 다리는 이러한 맥락에서 테스트베드로 사용됩니다.

포스트텐션 콘크리트로 만들어진 교량 구조는 5개 경간으로 구성되며 가장 긴 길이는 94m입니다. 본 연구에서는 하이브리드 RANS-LES 난류 접근 방식인 Spalart-Allmaras 지연 분리 와류 시뮬레이션(DDES)을 사용하여 수치적 바람 시뮬레이션의 공기역학적 특성과 표면압 등 극한값을 얻습니다. LES 방법으로 부과된 벽 근처 메쉬 밀도.

바람이 많이 부는 기후, 극도로 추운 기후, 그리고 3000년의 반환 기간에 대해 계산된 값을 포함한 세 가지 극한 기후 시나리오에 대해 표면 풍압을 얻습니다. 결과는 과도 풍류 시뮬레이션에서 나오는 시간 레이어로 인해 표면 풍압에 상당한 차이가 있음을 나타냅니다. 임계 바람 시나리오에서 구조적 거동을 평가하기 위해 각 시나리오에 대해 가장 높은 표면 압력 값이 고려됩니다.

또한 교량 기둥에 대한 유체 역학 연구를 수행하여 하천의 흐름을 VOF 방법으로 시뮬레이션하고 기둥 주변의 물 이동 과정을 일시적이고 다른 시간에 조사합니다. 각 기둥 표면에서 기록된 체적 유량이 가장 높은 강의 흐름에 의해 적용되는 표면 압력이 계산됩니다.

강의 흐름을 시뮬레이션하기 위해 이전 기간에 기록된 정보와 기상 조건이 사용되었습니다. 결과는 강의 흐름이 기둥에 닿는 순간의 표면 압력이 나중에 순간보다 훨씬 높다는 것을 보여줍니다. 이 압력의 양은 유체-구조 상호작용(FSI) 계산에서 임계 하중으로 사용될 수 있습니다.

마지막으로 두 섹션 모두 바람 표면의 압력, 보조 프로브 라인에 대한 속도장, 기둥 주위 물의 원주 운동 윤곽 및 압력 다이어그램이 서로 다른 시간 간격으로 보고됩니다.

키워드: 디지털 트윈 , 풍력 공학, 콘크리트 교량, 유체역학, CFD 시뮬레이션, DDES 난류 모델, Kalix 교량

Palabras clave: Gemelo digital , Ingeniería eólica, Puente de hormigón, Hidrodinámica, Simulación CFD, Modelo de turbulencia DDES, Puente Kalix

1. Introducción

Las infraestructuras de transporte son la columna vertebral de nuestra sociedad y los puentes son el cuello de botella de la red de transporte [1]. Además, el cambio climático que da como resultado tasas de deterioro más altas y los eventos climáticos extremos son amenazas importantes para la confiabilidad y seguridad de las redes de transporte. Durante la última década, muchos puentes se han dañado o fallado por condiciones climáticas extremas como tifones e inundaciones.

Wang et al. analizó los impactos del cambio climático y mostró que se espera que el deterioro de los puentes de hormigón sea aún peor que en la actualidad, y se prevé que los eventos climáticos extremos sean más frecuentes y con mayor gravedad [2].

Además, la demanda de capacidad de carga a menudo aumenta con el tiempo, por ejemplo, debido al uso de camiones más pesados para el transporte de madera en el norte de Europa y América del Norte. Por lo tanto, existe una necesidad creciente de métodos confiables para evaluar la resistencia estructural de la red de transporte en condiciones climáticas extremas que tengan en cuenta los escenarios futuros de cambio climático.

Los activos de transporte por carretera se diseñan, construyen y explotan basándose en numerosas fuentes de datos y varios modelos. Por lo tanto, los ingenieros de diseño usan modelos establecidos proporcionados por las normas; ingenieros de construccion
documentar los datos en el material real y proporcionar planos según lo construido; los operadores recopilan datos sobre el tráfico, realizan inspecciones y planifican el mantenimiento; los científicos del clima combinan datos y modelos climáticos para
predecir eventos climáticos futuros, y los ingenieros de evaluación calculan el impacto de la carga climática extrema en la estructura.

Dadas las fuentes abrumadoras y la complejidad de los datos y modelos, es posible que la información y los cálculos actualizados no estén disponibles para decisiones cruciales, por ejemplo, con respecto a la seguridad estructural y la operabilidad de la infraestructura durante episodios de eventos extremos. La falta de una integración perfecta entre los datos de la infraestructura, los modelos estructurales y la toma de decisiones a nivel del sistema es una limitación importante de las soluciones actuales, lo que conduce a la inadaptación e incertidumbre y crea costos e ineficiencias.

El gemelo digital estructural de la infraestructura es una simulación estructural viva que reúne todos los datos y modelos y se actualiza desde múltiples fuentes para representar su contraparte física. El Digital Twin estructural, mantenido durante todo el ciclo de vida de un activo y fácilmente accesible en cualquier momento, proporciona al propietario/usuarios de la infraestructura una idea temprana de los riesgos potenciales para la movilidad inducidos por eventos climáticos, cargas de vehículos pesados e incluso el envejecimiento de un infraestructura de transporte.

En un proyecto en curso, estamos desarrollando e implementando un gemelo digital estructural para el puente de Kalix en Suecia. El objetivo general del presente artículo es presentar un método y estudiar los resultados de la cuantificación de las cargas estructurales resultantes de eventos climáticos extremos basados en escenarios climáticos futuros para el puente de Kalix. El puente de Kalix, construido sobre el río Kalix en Suecia en 1956, está hecho de una viga cajón de hormigón postensado. El puente se utiliza como banco de pruebas para la demostración de métodos de evaluación y control de la salud estructural (SHM) de última generación.

El objetivo específico de la investigación actual es dar cuenta de parámetros climáticos como el viento y el flujo de agua, que imponen cargas estáticas y dinámicas en las estructuras. Nuestro método, en el primer paso, consiste en simulaciones de flujo de viento y simulaciones de flujo de agua utilizando un modelado CFD transitorio basado en el modelo de turbulencia LES/DES para cuantificar las cargas de viento e hidráulicas; esto constituye el punto focal principal de este artículo.

En el siguiente paso, se estudiará la respuesta estructural del puente mediante la transformación de los perfiles de carga eólica e hidráulica en cargas estructurales en el análisis de EF estructural no lineal. Por último, el modelo estructural se actualizará incorporando sin problemas los datos del SHM y, por lo tanto, creando un gemelo digital estructural que refleje la verdadera respuesta de la estructura. Los dos primeros enfoques de investigación permanecen fuera del alcance inmediato del presente artículo.

2. Descripción del puente de Kalix

El puente de Kalix consta de 5 vanos largos de los cuales el más largo tiene unos 94 metros y el más corto 43,85 m. El puente es de hormigón postensado, el cual se cuela in situ de forma segmentaria y una viga cajón no prismática como se muestra en la Fig. 1. El puente es simétrico en geometría y hay una bisagra en el punto medio. El ancho del tablero del puente en la losa superior e inferior es de aproximadamente 13 my 7,5 m, respectivamente. El espesor del muro es de 45 cm y el espesor de la losa inferior varía de 20 cm a
50 cm.

Fig. 1. Geometría y secciones del puente

Fig. 1. Geometría y secciones del puente

3. Simulación de viento

Las pruebas en túnel de viento solían ser la única forma de examinar la reacción de los puentes a las cargas de viento Consulte [3]; sin embargo, estos experimentos requieren mucho tiempo y son costosos. Se requieren cerca de 6 a 8 semanas para realizar una prueba típica en un túnel de viento Consulte [4]. Los últimos logros en la capacidad computacional de las computadoras brindan oportunidades para la simulación práctica del viento alrededor de puentes utilizando la dinámica de fluidos computacional (CFD).

Es beneficioso investigar la presión del viento en los componentes del puente utilizando una simulación por computadora. Es necesario determinar los parámetros de simulación del puente y el campo de viento a su alrededor; por lo tanto, se pueden evaluar con precisión sus impactos en las fuerzas aplicadas en el puente.

Las demandas de diseño de las estructuras de puentes requieren una investigación rigurosa de la acción del viento, especialmente en condiciones climáticas extremas. Garantizar la estabilidad de los puentes de grandes luces, ya que sus características y formaciones son más propensas a la carga de viento, se encuentra entre las principales consideraciones de diseño [3].

3.1. Parámetros de simulación

La velocidad básica del viento se elige 22 m/s según el mapa de viento de Suecia y la ubicación del puente de Kalix según EN 1991-1-4 [5] y el código sueco BFS 2019: 1 EKS 11; ver figura 1. La superficie libre sobre el agua se considera un área expuesta a la carga de viento. La dirección del ataque del viento dominante se considera perpendicular al tablero del puente.

Las simulaciones actuales se basan en tres escenarios que incluyen: viento extremo, frío extremo y valor de diseño para un período de retorno de 3000 años. Cada condición tiene diferentes valores de temperatura, viento básico
velocidad, viscosidad cinemática y densidad del aire, como se muestra en la Tabla 1. Los conjuntos de datos meteorológicos se sintetizaron para dos semanas meteorológicas extremas durante el período de 30 años de 2040-2069, considerando 13 escenarios climáticos futuros diferentes con diferentes modelos climáticos globales (GCM) y rutas de concentración representativas (RCP).

Se seleccionaron una semana de frío extremo y una semana de viento extremo utilizando el enfoque desarrollado
de Nik [7]. El planteamiento se adaptó a las necesidades de este trabajo, considerando el horario semanal en lugar de mensual. Se ha verificado la aplicación del enfoque para simulaciones complejas, incluidos los sistemas de energía Consulte [7] Consulte [8], hidrotermal Consulte [ 9] y simulaciones de microclimas Consulte [10].

Para considerar las condiciones climáticas extremas de una infraestructura muy importante, el valor de la velocidad básica del viento debe transferirse del período de retorno de 50 años a 3000 años como se indica en la ecuación 1 [6]. El perfil de velocidad y turbulencia se crea en base a EN 1991-1-4 [5] para la categoría de terreno 0 (Z0 = 0,003 my Zmín = 1 m), donde Z0 y Zmín son la longitud de rugosidad y la altura mínima, respectivamente. La variación de la velocidad del viento con la altura se define en la ecuación 2, donde co (z) es el factor de orografía tomado como 1, vm (z) es la velocidad media del viento a la altura z, kr es el factor del terreno que depende de la longitud de la rugosidad , e Iv (z) es la intensidad de la turbulencia; ver ecuación 3.���50=[0.36+0.1ln12�]     1�����=��·ln��0·���  [2]���=�����=�1�0�·ln�/�0  ��� ����≤�≤����  [3]���=������                                ��� �<����                   [4]

Velocidad del viento, variación de la velocidad del viento con la altura, intensidad de la turbulencia

Se calcula que el valor de la velocidad del viento para T = período de retorno de 3000 años es de 31 m/s; por lo tanto, los diagramas de velocidad del viento e intensidad de turbulencia se obtienen como se muestra en la figura 2.

Tabla. 1. Información meteorológica para tres escenarios

Tabla. 1. Información meteorológica para tres escenarios

Fig.  2. Valor de cálculo para la información del periodo de retorno de 3000 años: (a) Velocidad del viento y (b) Perfil de intensidad de turbulencia, y (c) Especificaciones del modelo

Fig. 2. Valor de cálculo para la información del periodo de retorno de 3000 años: (a) Velocidad del viento y (b) Intensidad de la turbulencia perfil, y (c) Especificaciones del modelo

3.2. Modelo de turbulencia

Para que las investigaciones sean precisas en el flujo alrededor de estructuras importantes como puentes, se aplica un enfoque híbrido que incluye simulaciones de remolinos desprendidos retardados (DDES) y es computacionalmente eficiente [11] [12]. Este modelo de turbulencia usa un método RANS cerca de las capas límite y el método LES lejos de las capas límite y en el área del flujo de la región separada ‘.

En el primer paso, el enfoque de simulación de remolinos separados se ha ampliado para adquirir predicciones de fuerza fiables en los modelos con un gran impacto del flujo separado. Hay varios ejemplos en la parte de revisión de Spalart Consulte [11] para varios casos que usan la aplicación del modelo de turbulencia de simulación de remolino separado (DES).

La formulación DES inicial [13] se desarrolla utilizando el enfoque de Spalart-Allmaras. Con respecto a la transición del enfoque RANS al LES, se revisa el término de destrucción en la ecuación de transporte de viscosidad modificada: la distancia entre un punto en el dominio y la superficie sólida más cercana (d) se sustituye por el factor introducido por:�~=���(�.����·∆)

Factor que sustituye la distancia entre el punto en el dominio y la superficie sólida más cercana (d)

donde CDES es un coeficiente, se considera como 0,65 y Δ es una escala de longitud asociada con el espaciado de la rejilla local:�=���(��.��.��)

Escala de longitud asociada con el espaciado de rejilla local

Se ha empleado un enfoque modificado de DES, conocido como simulación de remolinos desprendidos retardados (DDES), para dominar el probable problema de la “separación inducida por la rejilla” (GIS) que está relacionado con la geometría de la rejilla. El objetivo de este nuevo enfoque es confirmar que el modelado de turbulencia se mantiene en modo RANS en todas las capas de contorno [14]. Por lo tanto, la definición del parámetro se modifica como se define:�~=�-�����(0. �-����·�)   6

Modificación del parámetro d

donde fd es una función de filtro que considera un valor de 0 en las capas límite cercanas al muro (zona RANS) y un valor de 1 en las áreas donde se realizó la separación del flujo (zona LES).

3.3. Rejilla computacional y resultados

RWIND 2.01 Pro se emplea para la simulación de viento CFD, que usa el código CFD externo OpenFOAM® versión 17.10. La simulación CFD tridimensional se realiza como una simulación de viento transitorio para flujo turbulento incompresible utilizando el algoritmo SIMPLE (Método semi-implícito para ecuaciones vinculadas a presión).

En la simulación actual, el solucionador de estado estacionario se considera como la condición inicial, lo que significa que cuando se está calculando el flujo transitorio, el cálculo del estado estacionario de la condición inicial comienza en la primera parte de la simulación y tan pronto como se calcula. completado, el cálculo de transitorios se iniciará automáticamente.

Fig.  3. Dominio del túnel de viento y rejilla computacional de referencia (8.057.279 celdas)

Fig. 3. Dominio del túnel de viento y rejilla computacional de referencia (8.057.279 celdas)

La cuadrícula computacional se realiza mediante 8.057.279 celdas tridimensionales y 8.820.901 nudos, también se consideran las dimensiones del dominio del túnel de viento 2000 m * 1000 m * 100 m (largo, ancho, alto) como se muestra en la figura 3. El volumen mínimo de la celda es de 6,34 * 10-5 m3, el volumen máximo es de 812,30 m3 y la desviación máxima es de 1,80.

La presión residual final se considera 5 * 10-5. El proceso de generación de mallas e independencia de la rejilla se ha realizado utilizando los cuatro tamaños de malla que se muestran en la figura 4 para la malla de referencia, y finalmente se ha conseguido la independencia de la rejilla.

Fig.  4. Estudio de rejilla de cuatro tamaños de malla computacional a través de la línea de sondeo.

Fig. 4. Estudio de rejilla de cuatro tamaños de malla computacional a través de la línea de sondeo.

Se han realizado tres simulaciones para obtener el valor de la presión del viento para condiciones climáticas extremas y el valor de cálculo del viento que se muestra en la Fig. 5. Para cada escenario, el resultado de la presión del viento se obtiene utilizando el modelo de turbulencia transitoria DDES con respecto a 30 (s) de duración que incluye 60 capas de tiempo (Δt = 0,5 s).

Se puede observar que el área frontal del puente está expuesta a la presión del viento positiva y la cantidad de presión aumenta en la altura cerca del borde del tablero para todos los escenarios. Además, la Fig. 5. ilustra los valores negativos de la presión del viento en su totalidad en la superficie de la cubierta. El valor de pertenencia para el período de 3000 años es mucho más alto que los otros escenarios.

Es importante tener en cuenta que el intervalo de la velocidad del viento de entrada tiene un gran impacto en el valor de la presión en la superficie más que en los otros parámetros. Además, para cada escenario, el intervalo más alto de presión del viento y succión durante el tiempo total debe considerarse como una carga de viento crítica impuesta a la estructura. El valor más bajo de la presión en la superficie se obtiene en el escenario de condiciones de frío extremo, mientras que en condiciones de mucho viento, el valor de la presión se vuelve un orden de magnitud más alto.

Fig.  5. Contorno de presión superficial y diagrama para 60 capas de tiempo (Δt = 0.5 s) a través de una línea de sondeo para tres escenarios.

Fig. 5. Contorno de presión superficial y diagrama para 60 capas de tiempo (Δt = 0.5 s) a través de una línea de sondeo para tres escenarios.

Además, es importante tener en cuenta que el comportamiento del puente sería completamente diferente debido a las diferentes temperaturas del aire, y puede ocurrir un posible caso crítico en el escenario que experimente una presión menor. Con respecto al valor de entrada de cada escenario, el rango más alto de presión del viento pertenece al nivel de diseño debido al período de retorno de 3000 años, que ha recibido la velocidad del viento más alta como velocidad de entrada.

4. Simulación hidráulica

Los pilares de los puentes a través del río pueden bloquear el flujo al reducir la sección transversal del río, crear corrientes parásitas locales y cambiar la velocidad del flujo, lo que puede ejercer presión en las superficies de los pilares. Cuando el río fluye hacia los pilares del puente, el proceso del flujo de agua alrededor de la base se puede dividir en dos partes: aplicando presión en el momento en que el agua golpea el pilar del puente y después de la presión inicial cuando el agua fluye alrededor de los pilares [15].

Cuando el agua alcanza los pilares del puente a una cierta velocidad, el efecto de la presión sobre los pilares es mucho mayor que la presión del fluido que queda a su alrededor. Debido a los desarrollos de la ciencia de la computación, así como al desarrollo cada vez mayor de los códigos dinámicos de fluidos computacionales, se han utilizado ampliamente varias simulaciones numéricas y se ha demostrado que los resultados de muchas simulaciones son consistentes con los resultados experimentales [16].

Por ello, en esta investigación se ha utilizado el método de la dinámica de fluidos computacional para simular los fenómenos que gobiernan el comportamiento del flujo de los ríos. Para este estudio se ha seleccionado una solución tridimensional basada en cálculos numéricos utilizando el modelo de turbulencia LES. La simulación tridimensional del flujo del río en diferentes direcciones y velocidades nos permite calcular y analizar todas las presiones en la superficie de los pilares del puente en diferentes intervalos de tiempo.

4.1. Parámetros de simulación

El flujo del río se puede definir como un flujo de dos fases, que incluye agua y aire, en un canal abierto. El flujo de canal abierto es un flujo de fluido con una superficie libre en la que la presión atmosférica se distribuye uniformemente y se crea por el peso del fluido. Para simular este tipo de flujo se utiliza el método multifase VOF.

El programa Flow3D, disponible en el mercado, utiliza los métodos de fracciones volumétricas VOF y FAVOF. En el método VOF, el dominio de modelado se divide primero en celdas de elementos o volúmenes de controles más pequeños. Para los elementos que contienen fluidos, se mantienen valores numéricos para cada una de las variables de flujo dentro de ellos.

Estos valores representan la media volumétrica de los valores en cada elemento. En las corrientes superficiales libres, no todas las celdas están llenas de líquido; algunas celdas en la superficie de flujo están medio llenas. En este caso, se define una cantidad llamada volumen de fluido, F, que representa la parte de la celda que se llena con el fluido.

Después de determinar la posición y el ángulo de la superficie del flujo, será posible aplicar las condiciones de contorno apropiadas en la superficie del flujo para calcular el movimiento del fluido. A medida que se mueve el fluido, el valor de F también cambia con él. Las superficies libres son monitoreadas automáticamente por el movimiento de fluido dentro de una red fija. El método FAVOR se usa para determinar la geometría.

También se puede usar otra cantidad de fracción volumétrica para determinar el nivel de un cuerpo rígido desocupado ( Vf ). Cuando se conoce el volumen que ocupa el cuerpo rígido en cada celda, el límite del fluido dentro de la red fija se puede determinar como VOF. Este límite se usa para determinar las condiciones de contorno del muro que sigue el arroyo. En general, la ecuación de continuidad de masa es la siguiente:��𝜕�𝜕�+𝜕𝜕�(����)+�𝜕𝜕�(����)+𝜕𝜕�(����)+������=����   10

Ecuación de continuidad de masa

Las ecuaciones de movimiento para los componentes de la velocidad de un fluido en coordenadas 3D, o en otras palabras, las ecuaciones de Navier-Stokes, son las siguientes:𝜕�𝜕�+1�����𝜕�𝜕�+���𝜕�𝜕�+���𝜕�𝜕�+��2�����=-1�𝜕�𝜕�+��+��-��-��������-��-���    11𝜕�𝜕�+1�����𝜕�𝜕�+���𝜕�𝜕�+���𝜕�𝜕�+��������=-�1�𝜕�𝜕�+��+��-��-��������-��-���  12𝜕�𝜕�+1�����𝜕�𝜕�+���𝜕�𝜕�+���𝜕�𝜕�=-1�𝜕�𝜕�+��+��-��-��������-��-���              13

Ecuaciones de Navier-Stokes

Donde VF es la relación del volumen abierto al flujo, ρ es la densidad del fluido, (u, v, w) son las componentes de la velocidad en las direcciones x, y y z, respectivamente, R SOR es la función de la fuente, (Ax, Ay, Az ) son las áreas fraccionales, (Gx, Gy, Gz ) son las fuerzas gravitacionales, (fx, fy, fz ) son las aceleraciones de la viscosidad y (bx, by, bz ) son las pérdidas de flujo en medios porosos en las direcciones x, y, z, respectivamente [17].

La zona de captación del río Kalix es grande y amplia, por lo que tiene un clima subpolar con inviernos fríos y largos y veranos suaves y cortos. Aproximadamente el 50% de las precipitaciones en esta zona es nieve. En mayo, por lo general, el deshielo provoca un aumento significativo en el caudal del río. Las condiciones climáticas del río se resumen en la Tabla 2, [18].

Contrariamente a la tendencia general de este estudio, la previsión de las condiciones meteorológicas mencionadas está utilizando la información meteorológica registrada en los períodos pasados. En función de la información meteorológica disponible, definimos las condiciones de contorno al realizar los cálculos.

Tabla 2: Parámetros del modelo y tabla 3:Condiciones de contorno del modelo

Tabla 2: Parámetros del modelo y tabla 3:Condiciones de contorno del modelo

4.2 Cuadrícula computacional y resultados

Primero, según las dimensiones de los pilares en tres direcciones X, Y, Z, y según la dimensión longitudinal de los pilares (D = 8,5 m; véase la figura 7), el dominio se extiende 10D aguas arriba y 20D aguas abajo. Se ha utilizado el método de mallado estructurado (cartesiano) y el software Flow3D para resolver este problema. Para una cuadrícula correcta, el dominio se debe dividir en diferentes secciones.

Esta división se basa en lugares con fuertes pendientes. Usando la creación de una nueva superficie, el dominio se puede dividir en varias secciones para crear una malla regular con las dimensiones correctas y apropiadas, se puede especificar el número de celdas en cada superficie.

Fig. 6: Estudio de rejilla para el dominio hídrico

Fig. 6: Estudio de rejilla para el dominio hídrico

Esto aumenta el volumen final de las células. Por esta razón, hemos dividido este dominio en tres niveles: Grueso, medio y fino. Los resultados de los estudios de independencia de la red se muestran en la figura 6. Para comprobar los resultados calculados, primero debemos asegurarnos de que la corriente de entrada sea la correcta. Para hacer esto, el caudal de entrada se mide en el dominio de la solución y se compara con el valor base. Las dimensiones del dominio de la solución se especifican en la figura 7. Esta figura también contribuye al reconocimiento de los pilares del puente y su denominación de superficies.

Como se muestra en la Fig. 8, el caudal del río se encuentra dentro del intervalo admisible durante el 90% del tiempo de simulación y el caudal de entrada se ha simulado correctamente. Además, en la Fig. 9, la velocidad media del río se calcula en función del caudal y del área de la sección transversal del río.

Para extraer la cantidad de presión aplicada a los diferentes lados de las columnas, hemos seleccionado el intervalo de tiempo de simulación de 10 a 25 segundos (tiempo de estabilización de descarga en la cantidad de 1800 metros cúbicos por segundo). Los resultados calculados para cada lado se muestran en la Fig. 10 y 11. Los contornos de velocidad también se muestran en las Figuras 12 y 13. Estos contornos se ajustan en función de la velocidad del fluido en un momento dado.

Debido a las dimensiones del dominio de la solución y al caudal del río, el flujo de agua llega a los pilares del puente en el décimo segundo y la presión inicial del flujo del río afecta las superficies de los pilares del puente. Esta presión inicial decrece con el tiempo y se estabiliza en un rango determinado para cada lado según el área y el porcentaje de interacción con el flujo. Para los cálculos de interacción fluido-estructura (FSI), se puede usar la presión crítica calculada en el momento en que la corriente golpea los pilares.

Fig. 7: Dibujo del dominio hidrostático

Fig. 7: Dibujo del dominio hidrostático

Fig. 8: caudal del río; La figura 9: Caudal de la velocidad del río; La figura 10: Presión en la pila del puente - I; La figura 11: Presión en la pila del puente – II

Fig. 8: caudal del río; La figura 9: Caudal de la velocidad del río; La figura 10: Presión en la pila del puente – I; La figura 11: Presión en la pila del puente – II

Fig. 12: Contador de velocidad en el tiempo: 30s Fig. 13: Contador de velocidad en el tiempo: 20 s

Fig. 12: Contador de velocidad en el tiempo: 30s Fig. 13: Contador de velocidad en el tiempo: 20 s

5. Conclusión

Los efectos de las condiciones meteorológicas extremas, incluido el viento dinámico y el flujo de agua, se investigaron numéricamente para el puente de Kalix. Se definieron tres escenarios para las simulaciones dinámicas de viento, incluido el clima con mucho viento, el clima extremadamente frío y el valor de diseño para un período de retorno de 3.000 años. Aprovechando las simulaciones CFD, se determinaron las presiones del viento en pasos de 60 tiempos (30 segundos) utilizando el modelo de turbulencia transitoria DDES.

Los resultados indican diferencias significativas entre los escenarios, lo que implica la importancia de los datos de entrada, especialmente el diagrama de velocidades del viento. Se observó que el valor de diseño para el período de devolución de 3000 años tiene un impacto mucho mayor que los otros escenarios. Además, se mostró la importancia de considerar el rango más alto de presión del viento en la superficie a través de los pasos de tiempo para evaluar el comportamiento estructural del puente en la condición más crítica.

Además, se consideró el caudal máximo del río para una simulación transitoria según las condiciones meteorológicas registradas, y los pilares del puente se sometieron al caudal máximo del río durante 30 segundos. Por lo tanto, además de las condiciones físicas del flujo del río y cómo cambia la dirección del flujo aguas abajo, se cuantificaron las presiones máximas del agua en el momento en que el flujo golpea los pilares.

En el trabajo futuro, el rendimiento estructural del puente de Kalix será evaluado por
imposición de la carga del viento, la presión del agua y la carga del tráfico, creando así un gemelo digital estructural que refleja la verdadera respuesta de la estructura.

6. Reconocimiento

Los autores agradecen enormemente el apoyo de Dlubal Software por proporcionar la licencia de RWIND Simulation, así como de Flow Sciences Inc. por proporcionar la licencia de FLOW-3D.

Autores: Mahyar Kazemian1, Sajad Nikdel2, Mehrnaz MohammadEsmaeili3, Vahid Nik4, Kamyab Zandi*5

Candidato a doctorado, becario en el Departamento de Ingeniería de Timezyx Inc., Canadá.

M.Sc. estudiante, pasante en el Departamento de Ingeniería, Timezyx Inc., Canadá.

Estudiante de licenciatura, pasante en el Departamento de Ingeniería, Timezyx Inc., Canadá.

4 Profesor adjunto en la división de Física de la construcción de la Universidad de Lund y la Universidad Tecnológica de Chalmers, Suecia.

* 5 Director, Timezyx Inc., Vancouver, BC V6N 2R2, Canadá. E-mail: kamyab.zandi@timezyx.com


Referencias

  1. Jančula, M., Jošt, J., & Gocál, J. (2021). Influencia de las acciones ambientales agresivas en las estructuras de los puentes. Transportation Research Procedia, 55 , 1229–1235. https://doi.org/10.1016/j.trpro.2021.07.104
  2. Wang, X., Nguyen, M., Stewart, MG, Syme, M. y Leitch, A. (2010). Análisis de los impactos del cambio climático en el deterioro de la infraestructura de hormigón – Informe de síntesis. CSIRO, Canberra.
  3. Kemayou, BTM (2016). Análisis de secciones de tableros de puentes por el método de la pseudocompresibilidad basado en FDM y LES: Mejora del rendimiento mediante la implementación de la computación en paralelo (tesis). Universidad de Arkansas.
  4. Larsen, A. y Walther, JH (1997). Análisis aeroelástico de secciones de vigas de puentes basado en simulaciones discretas de vórtices. Journal of Wind Engineering and Industrial Aerodynamics, 67–68 , 253–265. https://doi.org/10.1016/s0167-6105(97)00077-9
  5. Eurocódigo 1: Acciones en estructuras. (2006). Instituto Británico de Normas.
  6. ASCE. Cargas mínimas de cálculo para edificios y otras estructuras. (2013). Sociedad Estadounidense de Ingenieros Civiles.
  7. Nik, VM (2016). Facilitación de la simulación energética para el clima futuro: síntesis de conjuntos de datos meteorológicos típicos y extremos a partir de modelos climáticos regionales (RCM). Applied Energy, 177 , 204–226. https://doi.org/10.1016/j.apenergy.2016.05.107
  8. Perera, AT, Nik, VM, Chen, D., Scartezzini, J.‑L. y Hong, T. (2020). Cuantificación de los impactos del cambio climático y los eventos climáticos extremos en los sistemas energéticos. Nature Energy, 5 (2), 150–159. https://doi.org/10.1038/s41560-020-0558-0
  9. Nik, VM (2017). Aplicación de conjuntos de datos meteorológicos típicos y extremos en la simulación higrotérmica de componentes de construcción para el clima futuro: un estudio de caso para un muro de entramado de madera. Energy and Buildings, 154 , 30–45. https://doi.org/10.1016/j.enbuild.2017.08.042
  10. Hosseini, M., Javanroodi, K. y Nik, VM (2022). Evaluación de impacto de alta resolución del cambio climático en el rendimiento energético de los edificios considerando los eventos meteorológicos extremos y el microclima – Investigando las variaciones en el confort térmico interior y los grados-día. Ciudades sostenibles y sociedad, 78 , 103634. https://doi.org/10.1016/j.scs.2021.103634
  11. Spalart, P. R. (2009). Simulación de remolinos separados. Revisión anual de mecánica de fluidos, 41 , 181–202. https://doi.org/10.1146/annurev.fluid.010908.165130
  12. Spalart, PR, et al. (2006) Una nueva versión de simulación de remolinos separados, resistente a densidades de rejilla ambiguas. Dinámica de fluidos teórica y computacional, 2006. 20 (3), 181-195. https://doi.org/10.1007/s00162-006-0015-0
  13. Spalart, PR (1997). Comentarios sobre la viabilidad de LES para alas y sobre una aproximación híbrida RANS/LES. En Actas de la Primera Conferencia Internacional de AFOSR sobre DNS/LES. Prensa de Greyden.
  14. Boudreau, M., Dumas, G. y Veilleux, J.-C. (2017). Evaluación de la capacidad del enfoque de modelado de turbulencia DDES para simular la estela de un cuerpo de farol. Aeroespacial, 4 (3), 41. https://doi.org/10.3390/aerospace4030041
  15. Wang, Y., Zou, Y., Xu, L. y Luo, Z. (2015). Análisis de la presión del flujo de agua en pilas de puentes considerando el efecto del impacto. Problemas matemáticos en ingeniería, 2015 , 1–8. https://doi.org/10.1155/2015/687535
  16. Qi, H., Zheng, J. y Zhang, C. (2020). Simulación numérica del campo de velocidades alrededor de dos pilares de pilas en tándem del puente longitudinal. Fluidos, 5 (1), 32. https://doi.org/10.3390/fluids5010032
  17. Jalal, H. K. y Hassan, W. H (2020). Simulación numérica tridimensional de la socavación local alrededor de la pila de un puente circular utilizando el software flow-3d. Ciclo de conferencias de IOP: Ciencia e ingeniería de materiales, 745 , 012150. https://doi.org/10.1088/1757-899x/745/1/012150
  18. Herzog, S. D., Conrad, S., Ingri, J., Persson, P. y Kritzberg, E. S (2019). Cambios inducidos por crecidas de primavera en la especiación y destino del Fe a mayor salinidad. Geoquímica aplicada, 109 , 104385. https://doi.org/10.1016/j.apgeochem.2019.104385

Figure 1 | Schematic of the present research model with dimensions and macro-roughnesses installed.

On the hydraulic performance of the inclined drops: the effect of downstreammacro-roughness elements

경사 낙하의 수력학적 성능: 하류 거시 거칠기 요소의 영향

Farhoud Kalateh a,*, Ehsan Aminvash a and Rasoul Daneshfaraz b
a Faculty of Civil Engineering, University of Tabriz, Tabriz, Iran
b Faculty of Engineering, University of Maragheh, Maragheh, Iran
*Corresponding author. E-mail: f.kalateh@gmail.com

ABSTRACT

The main goal of the present study is to investigate the effects of macro-roughnesses downstream of the inclined drop through numerical models. Due to the vital importance of geometrical properties of the macro-roughnesses in the hydraulic performance and efficient energy dissipation downstream of inclined drops, two different geometries of macro-roughnesses, i.e., semi-circular and triangular geometries, have been investigated using the Flow-3D model. Numerical simulation showed that with the flow rate increase and relative critical depth, the flow energy consumption has decreased. Also, relative energy dissipation increases with the increase in height and slope angle, so that this amount of increase in energy loss compared to the smooth bed in semi-circular and triangular elements is 86.39 and 76.80%, respectively, in the inclined drop with a height of 15 cm and 86.99 and 65.78% in the drop with a height of 20 cm. The Froude number downstream on the uneven bed has been dramatically reduced, so this amount of reduction has been approximately 47 and 54% compared to the control condition. The relative depth of the downstream has also increased due to the turbulence of the flow on the uneven bed with the increase in the flow rate.

본 연구의 주요 목표는 수치 모델을 통해 경사 낙하 하류의 거시 거칠기 효과를 조사하는 것입니다. 수력학적 성능과 경사 낙하 하류의 효율적인 에너지 소산에서 거시 거칠기의 기하학적 특성이 매우 중요하기 때문에 두 가지 서로 다른 거시 거칠기 형상, 즉 반원형 및 삼각형 형상이 Flow를 사용하여 조사되었습니다.

3D 모델 수치 시뮬레이션을 통해 유량이 증가하고 상대 임계 깊이가 증가함에 따라 유동 에너지 소비가 감소하는 것으로 나타났습니다. 또한, 높이와 경사각이 증가함에 따라 상대적인 에너지 소산도 증가하는데, 반원형 요소와 삼각형 요소에서 평활층에 비해 에너지 손실의 증가량은 경사낙하에서 각각 86.39%와 76.80%입니다.

높이 15cm, 높이 20cm의 드롭에서 86.99%, 65.78%입니다. 고르지 못한 베드 하류의 프루드 수가 극적으로 감소하여 이 감소량은 대조 조건에 비해 약 47%와 54%였습니다. 유속이 증가함에 따라 고르지 못한 층에서의 흐름의 난류로 인해 하류의 상대적 깊이도 증가했습니다.

Key words

flow energy dissipation, Froude number, inclined drop, numerical simulation

Figure 1 | Schematic of the present research model with dimensions and macro-roughnesses installed.
Figure 1 | Schematic of the present research model with dimensions and macro-roughnesses installed.
Figure 2 | Meshing, boundary condition, and solution field network
Figure 2 | Meshing, boundary condition, and solution field network

REFERENCES

Abbaspour, A., Taghavianpour, T. & Arvanaghi, H. 2019 Experimental study of the hydraulic jump on the reverse bed with porous screens.
Applied Water Science 9, 155.
Abbaspour, A., Shiravani, P. & Hosseinzadeh Dalir, A. 2021 Experimental study of the energy dissipation on rough ramps. ISH Journal of
Hydraulic Engineering 27, 334–342.
Akib, S., Ahmed, A. A., Imran, H. M., Mahidin, M. F., Ahmed, H. S. & Rahman, S. 2015 Properties of a hydraulic jump over apparent
corrugated beds. Dam Engineering 25, 65–77.
AlTalib, A. N., Mohammed, A. Y. & Hayawi, H. A. 2015 Hydraulic jump and energy dissipation downstream stepped weir. Flow
Measurement and Instrumentation 69, 101616.
Bayon-Barrachina, A. & Lopez-Jimenez, P. A. 2015 Numerical analysis of hydraulic jumps using OpenFOAM. Journal of Hydroinformatics
17, 662–678.
Canovaro, F. & Solari, L. 2007 Dissipative analogies between a schematic macro-roughness arrangement and step–pool morphology. Earth
Surface Processes and Landforms: The Journal of the British Geomorphological Research Group 32, 1628–1640.
Daneshfaraz, R., Ghaderi, A., Akhtari, A. & Di Francesco, S. 2020 On the effect of block roughness in ogee spill-ways with flip buckets. Fluids
5, 182.
Daneshfaraz, R., Aminvash, E., Di Francesco, S., Najibi, A. & Abraham, J. 2021a Three-dimensional study of the effect of block roughness
geometry on inclined drop. Numerical Methods in Civil Engineering 6, 1–9.
Daneshfaraz, R., Aminvash, E., Ghaderi, A., Abraham, J. & Bagherzadeh, M. 2021b SVM performance for predicting the effect of horizontal
screen diameters on the hydraulic parameters of a vertical drop. Applied Science 11, 4238.
Daneshfaraz, R., Aminvash, E., Ghaderi, A., Kuriqi, A. & Abraham, J. 2021c Three-dimensional investigation of hydraulic properties of
vertical drop in the presence of step and grid dissipators. Symmetry 13, 895.
Dey, S. & Sarkar, A. 2008 Characteristics of turbulent flow in submerged jumps on rough beds. Journal of Engineering Mechanics 134, 49–59.
Ead, S. A. & Rajaratnam, N. 2002 Hydraulic jumps on corrugated beds. Journal of Hydraulic Engineering 128, 656–663.
Fang, H., Han, X., He, G. & Dey, S. 2018 Influence of permeable beds on hydraulically macro-rough flow. Journal of Fluid Mechanics 847,
552–590.
Federico, I., Marrone, S., Colagrossi, A., Aristodemo, F. & Antuono, M. 2019 Simulating 2D open-channel flows through an SPH model.
European Journal of Mechanics-B/Fluids 34, 35–46.
Ghaderi, A., Dasineh, M., Aristodemo, F. & Aricò, C. 2021 Numerical simulations of the flow field of a submerged hydraulic jump over
triangular macroroughnesses. Water 13, 674.
Ghare, A. D., Ingl, R. N., Porey, P. D. & Gokhale, S. S. 2010 Block ramp design for efficient energy dissipation. Journal of Energy Dissipation
136, 1–5.
Habibzadeh, A., Rajaratnam, N. & Loewen, M. 2019 Characteristics of the flow field downstream of free and submerged hydraulic jumps.
Proceedings of the Institution of Civil Engineers-Water Management 172, 180–194.
Hajiahmadi, A., Ghaeini-Hessaroeyeh, M. & Khanjani, M. J. 2021 Experimental evaluation of vertical shaft efficiency in vortex flow energy
dissipation. International Journal of Civil Engineering 19, 1445–1455.

Katourani, S. & Kashefipour, S. M. 2012 Effect of the geometric characteristics of baffle on hydraulic flow condition in baffled apron drop.
Irrigation Sciences and Engineering 37, 51–59.
Kurdistani, S. M., Varaki, M. E. & Moayedi Moshkaposhti, M. 2024 Apron and macro roughness as scour countermeasures downstream of
block ramps. ISH Journal of Hydraulic Engineering 1–9.
Lopardo, R. A. 2013 Extreme velocity fluctuations below free hydraulic jumps. Journal of Engineering 1–5.
Mahmoudi-Rad, M. & Najafzadeh, M. 2023 Experimental evaluation of the energy dissipation efficiency of the vortex flow section of drop
shafts. Scientific Reports 13, 1679.
Matin, M. A., Hasan, M. & Islam, M. R. 2018 Experiment on hydraulic jump in sudden expansion in a sloping rectangular channel. Journal of
Civil Engineering 36, 65–77.
Moghadam, K. F., Banihashemi, M. A., Badiei, P. & Shirkavand, A. 2019 A numerical approach to solve fluid-solid two-phase flows using time
splitting projection method with a pressure correction technique. Progress in Computational Fluid Dynamics, an International Journal
19, 357–367.
Moghadam, K. F., Banihashemi, M. A., Badiei, P. & Shirkavand, A. 2020 A time-splitting pressure-correction projection method for complete
two-fluid modeling of a local scour hole. International Journal of Sediment Research 35, 395–407.
Moradi-SabzKoohi, A., Kashefipour, S. M. & Bina, M. 2011 Experimental comparison of energy dissipation on drop structures. JWSS –
Isfahan University of Technology 15, 209–223. (in Persian).
Mouaze, D., Murzyn, F. & Chaplin, J. R. 2005 Free surface length scale estimation in hydraulic jumps. Journal of Fluids Engineering 127,
1191–1193.
Nicosia, A., Carollo, F. G. & Ferro, V. 2023 Effects of boulder arrangement on flow resistance due to macro-scale bed roughness. Water 15,
349.
Ohtsu, I. & Yasuda, Y. 1991 Hydraulic jump in sloping channel. Journal of Hydraulic Engineering 117, 905–921.
Pagliara, S. & Palermo, M. 2012 Effect of stilling basin geometry on the dissipative process in the presence of block ramps. Journal of
Irrigation and Drainage Engineering 138, 1027–1031.
Pagliara, S., Das, R. & Palermo, M. 2008 Energy dissipation on submerged block ramps. Journal of Irrigation and Drainage Engineering 134,
527–532.
Pagliara, S., Roshni, T. & Palermo, M. 2015 Energy dissipation over large-scale roughness for both transition and uniform flow conditions.
International Journal of Civil Engineering 13, 341–346.
Parsaie, A., Dehdar-Behbahani, S. & Haghiabi, A. H. 2016 Numerical modeling of cavitation on spillway’s flip bucket. Frontiers of Structural
and Civil Engineering 10, 438–444.
Pourabdollah, N., Heidarpour, M. & Abedi Koupai, J. 2018 Characteristics of free and submerged hydraulic jumps in different stilling basins.
In: Proceedings of the Institution of Civil Engineers-Water Management. Thomas Telford Ltd, pp. 1–11.
Roushangar, K. & Ghasempour, R. 2019 Evaluation of the impact of channel geometry and rough elements arrangement in hydraulic jump
energy dissipation via SVM. Journal of Hydroinformatics 21, 92–103.
Samadi-Boroujeni, H., Ghazali, M., Gorbani, B. & Nafchi, R. F. 2013 Effect of triangular corrugated beds on the hydraulic jump
characteristics. Canadian Journal of Civil Engineering 40, 841–847.
Shekari, Y., Javan, M. & Eghbalzadeh, A. 2014 Three-dimensional numerical study of submerged hydraulic jumps. Arabian Journal for
Science and Engineering 39, 6969–6981.
Tokyay, N. D., Evcimen, T. U. & Şimsek, Ç. 2011 Forced hydraulic jump on non-protruding rough beds. Canadian Journal of Civil
Engineering 38, 1136–1144.
Wagner, W. E. 1956 Hydraulic model studies of the check intake structure-potholes East canal. Bureau of Reclamation Hydraulic Laboratory
Report Hyd, 411.
Witt, A., Gulliver, J. S. & Shen, L. 2018 Numerical investigation of vorticity and bubble clustering in an air-entraining hydraulic jump.
Computers & Fluids 172, 162–180.

Lab-on-a-Chip 시스템의 혈류 역학에 대한 검토: 엔지니어링 관점

Review on Blood Flow Dynamics in Lab-on-a-Chip Systems: An Engineering Perspective

  • Bin-Jie Lai
  • Li-Tao Zhu
  • Zhe Chen*
  • Bo Ouyang*
  • , and 
  • Zheng-Hong Luo*

Abstract

다양한 수송 메커니즘 하에서, “LOC(lab-on-a-chip)” 시스템에서 유동 전단 속도 조건과 밀접한 관련이 있는 혈류 역학은 다양한 수송 현상을 초래하는 것으로 밝혀졌습니다.

본 연구는 적혈구의 동적 혈액 점도 및 탄성 거동과 같은 점탄성 특성의 역할을 통해 LOC 시스템의 혈류 패턴을 조사합니다. 모세관 및 전기삼투압의 주요 매개변수를 통해 LOC 시스템의 혈액 수송 현상에 대한 연구는 실험적, 이론적 및 수많은 수치적 접근 방식을 통해 제공됩니다.

전기 삼투압 점탄성 흐름에 의해 유발되는 교란은 특히 향후 연구 기회를 위해 혈액 및 기타 점탄성 유체를 취급하는 LOC 장치의 혼합 및 분리 기능 향상에 논의되고 적용됩니다. 또한, 본 연구는 보다 정확하고 단순화된 혈류 모델에 대한 요구와 전기역학 효과 하에서 점탄성 유체 흐름에 대한 수치 연구에 대한 강조와 같은 LOC 시스템 하에서 혈류 역학의 수치 모델링의 문제를 식별합니다.

전기역학 현상을 연구하는 동안 제타 전위 조건에 대한 보다 실용적인 가정도 강조됩니다. 본 연구는 모세관 및 전기삼투압에 의해 구동되는 미세유체 시스템의 혈류 역학에 대한 포괄적이고 학제적인 관점을 제공하는 것을 목표로 한다.

KEYWORDS: 

1. Introduction

1.1. Microfluidic Flow in Lab-on-a-Chip (LOC) Systems

Over the past several decades, the ability to control and utilize fluid flow patterns at microscales has gained considerable interest across a myriad of scientific and engineering disciplines, leading to growing interest in scientific research of microfluidics. 

(1) Microfluidics, an interdisciplinary field that straddles physics, engineering, and biotechnology, is dedicated to the behavior, precise control, and manipulation of fluids geometrically constrained to a small, typically submillimeter, scale. 

(2) The engineering community has increasingly focused on microfluidics, exploring different driving forces to enhance working fluid transport, with the aim of accurately and efficiently describing, controlling, designing, and applying microfluidic flow principles and transport phenomena, particularly for miniaturized applications. 

(3) This attention has chiefly been fueled by the potential to revolutionize diagnostic and therapeutic techniques in the biomedical and pharmaceutical sectorsUnder various driving forces in microfluidic flows, intriguing transport phenomena have bolstered confidence in sustainable and efficient applications in fields such as pharmaceutical, biochemical, and environmental science. The “lab-on-a-chip” (LOC) system harnesses microfluidic flow to enable fluid processing and the execution of laboratory tasks on a chip-sized scale. LOC systems have played a vital role in the miniaturization of laboratory operations such as mixing, chemical reaction, separation, flow control, and detection on small devices, where a wide variety of fluids is adapted. Biological fluid flow like blood and other viscoelastic fluids are notably studied among the many working fluids commonly utilized by LOC systems, owing to the optimization in small fluid sample volumed, rapid response times, precise control, and easy manipulation of flow patterns offered by the system under various driving forces. 

(4)The driving forces in blood flow can be categorized as passive or active transport mechanisms and, in some cases, both. Under various transport mechanisms, the unique design of microchannels enables different functionalities in driving, mixing, separating, and diagnosing blood and drug delivery in the blood. 

(5) Understanding and manipulating these driving forces are crucial for optimizing the performance of a LOC system. Such knowledge presents the opportunity to achieve higher efficiency and reliability in addressing cellular level challenges in medical diagnostics, forensic studies, cancer detection, and other fundamental research areas, for applications of point-of-care (POC) devices. 

(6)

1.2. Engineering Approach of Microfluidic Transport Phenomena in LOC Systems

Different transport mechanisms exhibit unique properties at submillimeter length scales in microfluidic devices, leading to significant transport phenomena that differ from those of macroscale flows. An in-depth understanding of these unique transport phenomena under microfluidic systems is often required in fluidic mechanics to fully harness the potential functionality of a LOC system to obtain systematically designed and precisely controlled transport of microfluids under their respective driving force. Fluid mechanics is considered a vital component in chemical engineering, enabling the analysis of fluid behaviors in various unit designs, ranging from large-scale reactors to separation units. Transport phenomena in fluid mechanics provide a conceptual framework for analytically and descriptively explaining why and how experimental results and physiological phenomena occur. The Navier–Stokes (N–S) equation, along with other governing equations, is often adapted to accurately describe fluid dynamics by accounting for pressure, surface properties, velocity, and temperature variations over space and time. In addition, limiting factors and nonidealities for these governing equations should be considered to impose corrections for empirical consistency before physical models are assembled for more accurate controls and efficiency. Microfluidic flow systems often deviate from ideal conditions, requiring adjustments to the standard governing equations. These deviations could arise from factors such as viscous effects, surface interactions, and non-Newtonian fluid properties from different microfluid types and geometrical layouts of microchannels. Addressing these nonidealities supports the refining of theoretical models and prediction accuracy for microfluidic flow behaviors.

The analytical calculation of coupled nonlinear governing equations, which describes the material and energy balances of systems under ideal conditions, often requires considerable computational efforts. However, advancements in computation capabilities, cost reduction, and improved accuracy have made numerical simulations using different numerical and modeling methods a powerful tool for effectively solving these complex coupled equations and modeling various transport phenomena. Computational fluid dynamics (CFD) is a numerical technique used to investigate the spatial and temporal distribution of various flow parameters. It serves as a critical approach to provide insights and reasoning for decision-making regarding the optimal designs involving fluid dynamics, even prior to complex physical model prototyping and experimental procedures. The integration of experimental data, theoretical analysis, and reliable numerical simulations from CFD enables systematic variation of analytical parameters through quantitative analysis, where adjustment to delivery of blood flow and other working fluids in LOC systems can be achieved.

Numerical methods such as the Finite-Difference Method (FDM), Finite-Element-Method (FEM), and Finite-Volume Method (FVM) are heavily employed in CFD and offer diverse approaches to achieve discretization of Eulerian flow equations through filling a mesh of the flow domain. A more in-depth review of numerical methods in CFD and its application for blood flow simulation is provided in Section 2.2.2.

1.3. Scope of the Review

In this Review, we explore and characterize the blood flow phenomena within the LOC systems, utilizing both physiological and engineering modeling approaches. Similar approaches will be taken to discuss capillary-driven flow and electric-osmotic flow (EOF) under electrokinetic phenomena as a passive and active transport scheme, respectively, for blood transport in LOC systems. Such an analysis aims to bridge the gap between physical (experimental) and engineering (analytical) perspectives in studying and manipulating blood flow delivery by different driving forces in LOC systems. Moreover, the Review hopes to benefit the interests of not only blood flow control in LOC devices but also the transport of viscoelastic fluids, which are less studied in the literature compared to that of Newtonian fluids, in LOC systems.

Section 2 examines the complex interplay between viscoelastic properties of blood and blood flow patterns under shear flow in LOC systems, while engineering numerical modeling approaches for blood flow are presented for assistance. Sections 3 and 4 look into the theoretical principles, numerical governing equations, and modeling methodologies for capillary driven flow and EOF in LOC systems as well as their impact on blood flow dynamics through the quantification of key parameters of the two driving forces. Section 5 concludes the characterized blood flow transport processes in LOC systems under these two forces. Additionally, prospective areas of research in improving the functionality of LOC devices employing blood and other viscoelastic fluids and potentially justifying mechanisms underlying microfluidic flow patterns outside of LOC systems are presented. Finally, the challenges encountered in the numerical studies of blood flow under LOC systems are acknowledged, paving the way for further research.

2. Blood Flow Phenomena

ARTICLE SECTIONS

Jump To


2.1. Physiological Blood Flow Behavior

Blood, an essential physiological fluid in the human body, serves the vital role of transporting oxygen and nutrients throughout the body. Additionally, blood is responsible for suspending various blood cells including erythrocytes (red blood cells or RBCs), leukocytes (white blood cells), and thrombocytes (blood platelets) in a plasma medium.Among the cells mentioned above, red blood cells (RBCs) comprise approximately 40–45% of the volume of healthy blood. 

(7) An RBC possesses an inherent elastic property with a biconcave shape of an average diameter of 8 μm and a thickness of 2 μm. This biconcave shape maximizes the surface-to-volume ratio, allowing RBCs to endure significant distortion while maintaining their functionality. 

(8,9) Additionally, the biconcave shape optimizes gas exchange, facilitating efficient uptake of oxygen due to the increased surface area. The inherent elasticity of RBCs allows them to undergo substantial distortion from their original biconcave shape and exhibits high flexibility, particularly in narrow channels.RBC deformability enables the cell to deform from a biconcave shape to a parachute-like configuration, despite minor differences in RBC shape dynamics under shear flow between initial cell locations. As shown in Figure 1(a), RBCs initiating with different resting shapes and orientations displaying display a similar deformation pattern 

(10) in terms of its shape. Shear flow induces an inward bending of the cell at the rear position of the rim to the final bending position, 

(11) resulting in an alignment toward the same position of the flow direction.

Figure 1. Images of varying deformation of RBCs and different dynamic blood flow behaviors. (a) The deforming shape behavior of RBCs at four different initiating positions under the same experimental conditions of a flow from left to right, (10) (b) RBC aggregation, (13) (c) CFL region. (18) Reproduced with permission from ref (10). Copyright 2011 Elsevier. Reproduced with permission from ref (13). Copyright 2022 The Authors, under the terms of the Creative Commons (CC BY 4.0) License https://creativecommons.org/licenses/by/4.0/. Reproduced with permission from ref (18). Copyright 2019 Elsevier.

The flexible property of RBCs enables them to navigate through narrow capillaries and traverse a complex network of blood vessels. The deformability of RBCs depends on various factors, including the channel geometry, RBC concentration, and the elastic properties of the RBC membrane. 

(12) Both flexibility and deformability are vital in the process of oxygen exchange among blood and tissues throughout the body, allowing cells to flow in vessels even smaller than the original cell size prior to deforming.As RBCs serve as major components in blood, their collective dynamics also hugely affect blood rheology. RBCs exhibit an aggregation phenomenon due to cell to cell interactions, such as adhesion forces, among populated cells, inducing unique blood flow patterns and rheological behaviors in microfluidic systems. For blood flow in large vessels between a diameter of 1 and 3 cm, where shear rates are not high, a constant viscosity and Newtonian behavior for blood can be assumed. However, under low shear rate conditions (0.1 s

–1) in smaller vessels such as the arteries and venules, which are within a diameter of 0.2 mm to 1 cm, blood exhibits non-Newtonian properties, such as shear-thinning viscosity and viscoelasticity due to RBC aggregation and deformability. The nonlinear viscoelastic property of blood gives rise to a complex relationship between viscosity and shear rate, primarily influenced by the highly elastic behavior of RBCs. A wide range of research on the transient behavior of the RBC shape and aggregation characteristics under varied flow circumstances has been conducted, aiming to obtain a better understanding of the interaction between blood flow shear forces from confined flows.

For a better understanding of the unique blood flow structures and rheological behaviors in microfluidic systems, some blood flow patterns are introduced in the following section.

2.1.1. RBC Aggregation

RBC aggregation is a vital phenomenon to be considered when designing LOC devices due to its impact on the viscosity of the bulk flow. Under conditions of low shear rate, such as in stagnant or low flow rate regions, RBCs tend to aggregate, forming structures known as rouleaux, resembling stacks of coins as shown in Figure 1(b). 

(13) The aggregation of RBCs increases the viscosity at the aggregated region, 

(14) hence slowing down the overall blood flow. However, when exposed to high shear rates, RBC aggregates disaggregate. As shear rates continue to increase, RBCs tend to deform, elongating and aligning themselves with the direction of the flow. 

(15) Such a dynamic shift in behavior from the cells in response to the shear rate forms the basis of the viscoelastic properties observed in whole blood. In essence, the viscosity of the blood varies according to the shear rate conditions, which are related to the velocity gradient of the system. It is significant to take the intricate relationship between shear rate conditions and the change of blood viscosity due to RBC aggregation into account since various flow driving conditions may induce varied effects on the degree of aggregation.

2.1.2. Fåhræus-Lindqvist Effect

The Fåhræus–Lindqvist (FL) effect describes the gradual decrease in the apparent viscosity of blood as the channel diameter decreases. 

(16) This effect is attributed to the migration of RBCs toward the central region in the microchannel, where the flow rate is higher, due to the presence of higher pressure and asymmetric distribution of shear forces. This migration of RBCs, typically observed at blood vessels less than 0.3 mm, toward the higher flow rate region contributes to the change in blood viscosity, which becomes dependent on the channel size. Simultaneously, the increase of the RBC concentration in the central region of the microchannel results in the formation of a less viscous region close to the microchannel wall. This region called the Cell-Free Layer (CFL), is primarily composed of plasma. 

(17) The combination of the FL effect and the following CFL formation provides a unique phenomenon that is often utilized in passive and active plasma separation mechanisms, involving branched and constriction channels for various applications in plasma separation using microfluidic systems.

2.1.3. Cell-Free Layer Formation

In microfluidic blood flow, RBCs form aggregates at the microchannel core and result in a region that is mostly devoid of RBCs near the microchannel walls, as shown in Figure 1(c). 

(18) The region is known as the cell-free layer (CFL). The CFL region is often known to possess a lower viscosity compared to other regions within the blood flow due to the lower viscosity value of plasma when compared to that of the aggregated RBCs. Therefore, a thicker CFL region composed of plasma correlates to a reduced apparent whole blood viscosity. 

(19) A thicker CFL region is often established following the RBC aggregation at the microchannel core under conditions of decreasing the tube diameter. Apart from the dependence on the RBC concentration in the microchannel core, the CFL thickness is also affected by the volume concentration of RBCs, or hematocrit, in whole blood, as well as the deformability of RBCs. Given the influence CFL thickness has on blood flow rheological parameters such as blood flow rate, which is strongly dependent on whole blood viscosity, investigating CFL thickness under shear flow is crucial for LOC systems accounting for blood flow.

2.1.4. Plasma Skimming in Bifurcation Networks

The uneven arrangement of RBCs in bifurcating microchannels, commonly termed skimming bifurcation, arises from the axial migration of RBCs within flowing streams. This uneven distribution contributes to variations in viscosity across differing sizes of bifurcating channels but offers a stabilizing effect. Notably, higher flow rates in microchannels are associated with increased hematocrit levels, resulting in higher viscosity compared with those with lower flow rates. Parametric investigations on bifurcation angle, 

(20) thickness of the CFL, 

(21) and RBC dynamics, including aggregation and deformation, 

(22) may alter the varying viscosity of blood and its flow behavior within microchannels.

2.2. Modeling on Blood Flow Dynamics

2.2.1. Blood Properties and Mathematical Models of Blood Rheology

Under different shear rate conditions in blood flow, the elastic characteristics and dynamic changes of the RBC induce a complex velocity and stress relationship, resulting in the incompatibility of blood flow characterization through standard presumptions of constant viscosity used for Newtonian fluid flow. Blood flow is categorized as a viscoelastic non-Newtonian fluid flow where constitutive equations governing this type of flow take into consideration the nonlinear viscometric properties of blood. To mathematically characterize the evolving blood viscosity and the relationship between the elasticity of RBC and the shear blood flow, respectively, across space and time of the system, a stress tensor (τ) defined by constitutive models is often coupled in the Navier–Stokes equation to account for the collective impact of the constant dynamic viscosity (η) and the elasticity from RBCs on blood flow.The dynamic viscosity of blood is heavily dependent on the shear stress applied to the cell and various parameters from the blood such as hematocrit value, plasma viscosity, mechanical properties of the RBC membrane, and red blood cell aggregation rate. The apparent blood viscosity is considered convenient for the characterization of the relationship between the evolving blood viscosity and shear rate, which can be defined by Casson’s law, as shown in eq 1.

𝜇=𝜏0𝛾˙+2𝜂𝜏0𝛾˙⎯⎯⎯⎯⎯⎯⎯√+𝜂�=�0�˙+2��0�˙+�

(1)where τ

0 is the yield stress–stress required to initiate blood flow motion, η is the Casson rheological constant, and γ̇ is the shear rate. The value of Casson’s law parameters under blood with normal hematocrit level can be defined as τ

0 = 0.0056 Pa and η = 0.0035 Pa·s. 

(23) With the known property of blood and Casson’s law parameters, an approximation can be made to the dynamic viscosity under various flow condition domains. The Power Law model is often employed to characterize the dynamic viscosity in relation to the shear rate, since precise solutions exist for specific geometries and flow circumstances, acting as a fundamental standard for definition. The Carreau and Carreau–Yasuda models can be advantageous over the Power Law model due to their ability to evaluate the dynamic viscosity at low to zero shear rate conditions. However, none of the above-mentioned models consider the memory or other elastic behavior of blood and its RBCs. Some other commonly used mathematical models and their constants for the non-Newtonian viscosity property characterization of blood are listed in Table 1 below. 

(24−26)Table 1. Comparison of Various Non-Newtonian Models for Blood Viscosity 

(24−26)

ModelNon-Newtonian ViscosityParameters
Power Law(2)n = 0.61, k = 0.42
Carreau(3)μ0 = 0.056 Pa·s, μ = 0.00345 Pa·s, λ = 3.1736 s, m = 2.406, a = 0.254
Walburn–Schneck(4)C1 = 0.000797 Pa·s, C2 = 0.0608 Pa·s, C3 = 0.00499, C4 = 14.585 g–1, TPMA = 25 g/L
Carreau–Yasuda(5)μ0 = 0.056 Pa·s, μ = 0.00345 Pa·s, λ = 1.902 s, n = 0.22, a = 1.25
Quemada(6)μp = 0.0012 Pa·s, k = 2.07, k0 = 4.33, γ̇c = 1.88 s–1

The blood rheology is commonly known to be influenced by two key physiological factors, namely, the hematocrit value (H

t) and the fibrinogen concentration (c

f), with an average value of 42% and 0.252 gd·L

–1, respectively. Particularly in low shear conditions, the presence of varying fibrinogen concentrations affects the tendency for aggregation and rouleaux formation, while the occurrence of aggregation is contingent upon specific levels of hematocrit. 

(27) The study from Apostolidis et al. 

(28) modifies the Casson model through emphasizing its reliance on hematocrit and fibrinogen concentration parameter values, owing to the extensive knowledge of the two physiological blood parameters.The viscoelastic response of blood is heavily dependent on the elasticity of the RBC, which is defined by the relationship between the deformation and stress relaxation from RBCs under a specific location of shear flow as a function of the velocity field. The stress tensor is usually characterized by constitutive equations such as the Upper-Convected Maxwell Model 

(29) and the Oldroyd-B model 

(30) to track the molecule effects under shear from different driving forces. The prominent non-Newtonian features, such as shear thinning and yield stress, have played a vital role in the characterization of blood rheology, particularly with respect to the evaluation of yield stress under low shear conditions. The nature of stress measurement in blood, typically on the order of 1 mPa, is challenging due to its low magnitude. The occurrence of the CFL complicates the measurement further due to the significant decrease in apparent viscosity near the wall over time and a consequential disparity in viscosity compared to the bulk region.In addition to shear thinning viscosity and yield stress, the formation of aggregation (rouleaux) from RBCs under low shear rates also contributes to the viscoelasticity under transient flow 

(31) and thixotropy 

(32) of whole blood. Given the difficulty in evaluating viscoelastic behavior of blood under low strain magnitudes and limitations in generalized Newtonian models, the utilization of viscoelastic models is advocated to encompass elasticity and delineate non-shear components within the stress tensor. Extending from the Oldroyd-B model, Anand et al. 

(33) developed a viscoelastic model framework for adapting elasticity within blood samples and predicting non-shear stress components. However, to also address the thixotropic effects, the model developed by Horner et al. 

(34) serves as a more comprehensive approach than the viscoelastic model from Anand et al. Thixotropy 

(32) typically occurs from the structural change of the rouleaux, where low shear rate conditions induce rouleaux formation. Correspondingly, elasticity increases, while elasticity is more representative of the isolated RBCs, under high shear rate conditions. The model of Horner et al. 

(34) considers the contribution of rouleaux to shear stress, taking into account factors such as the characteristic time for Brownian aggregation, shear-induced aggregation, and shear-induced breakage. Subsequent advancements in the model from Horner et al. often revolve around refining the three aforementioned key terms for a more substantial characterization of rouleaux dynamics. Notably, this has led to the recently developed mHAWB model 

(35) and other model iterations to enhance the accuracy of elastic and viscoelastic contributions to blood rheology, including the recently improved model suggested by Armstrong et al. 

(36)

2.2.2. Numerical Methods (FDM, FEM, FVM)

Numerical simulation has become increasingly more significant in analyzing the geometry, boundary layers of flow, and nonlinearity of hyperbolic viscoelastic flow constitutive equations. CFD is a powerful and efficient tool utilizing numerical methods to solve the governing hydrodynamic equations, such as the Navier–Stokes (N–S) equation, continuity equation, and energy conservation equation, for qualitative evaluation of fluid motion dynamics under different parameters. CFD overcomes the challenge of analytically solving nonlinear forms of differential equations by employing numerical methods such as the Finite-Difference Method (FDM), Finite-Element Method (FEM), and Finite-Volume Method (FVM) to discretize and solve the partial differential equations (PDEs), allowing for qualitative reproduction of transport phenomena and experimental observations. Different numerical methods are chosen to cope with various transport systems for optimization of the accuracy of the result and control of error during the discretization process.FDM is a straightforward approach to discretizing PDEs, replacing the continuum representation of equations with a set of finite-difference equations, which is typically applied to structured grids for efficient implementation in CFD programs. 

(37) However, FDM is often limited to simple geometries such as rectangular or block-shaped geometries and struggles with curved boundaries. In contrast, FEM divides the fluid domain into small finite grids or elements, approximating PDEs through a local description of physics. 

(38) All elements contribute to a large, sparse matrix solver. However, FEM may not always provide accurate results for systems involving significant deformation and aggregation of particles like RBCs due to large distortion of grids. 

(39) FVM evaluates PDEs following the conservation laws and discretizes the selected flow domain into small but finite size control volumes, with each grid at the center of a finite volume. 

(40) The divergence theorem allows the conversion of volume integrals of PDEs with divergence terms into surface integrals of surface fluxes across cell boundaries. Due to its conservation property, FVM offers efficient outcomes when dealing with PDEs that embody mass, momentum, and energy conservation principles. Furthermore, widely accessible software packages like the OpenFOAM toolbox 

(41) include a viscoelastic solver, making it an attractive option for viscoelastic fluid flow modeling. 

(42)

2.2.3. Modeling Methods of Blood Flow Dynamics

The complexity in the blood flow simulation arises from deformability and aggregation that RBCs exhibit during their interaction with neighboring cells under different shear rate conditions induced by blood flow. Numerical models coupled with simulation programs have been applied as a groundbreaking method to predict such unique rheological behavior exhibited by RBCs and whole blood. The conventional approach of a single-phase flow simulation is often applied to blood flow simulations within large vessels possessing a moderate shear rate. However, such a method assumes the properties of plasma, RBCs and other cellular components to be evenly distributed as average density and viscosity in blood, resulting in the inability to simulate the mechanical dynamics, such as RBC aggregation under high-shear flow field, inherent in RBCs. To accurately describe the asymmetric distribution of RBC and blood flow, multiphase flow simulation, where numerical simulations of blood flows are often modeled as two immiscible phases, RBCs and blood plasma, is proposed. A common assumption is that RBCs exhibit non-Newtonian behavior while the plasma is treated as a continuous Newtonian phase.Numerous multiphase numerical models have been proposed to simulate the influence of RBCs on blood flow dynamics by different assumptions. In large-scale simulations (above the millimeter range), continuum-based methods are wildly used due to their lower computational demands. 

(43) Eulerian multiphase flow simulations offer the solution of a set of conservation equations for each separate phase and couple the phases through common pressure and interphase exchange coefficients. Xu et al. 

(44) utilized the combined finite-discrete element method (FDEM) to replicate the dynamic behavior and distortion of RBCs subjected to fluidic forces, utilizing the Johnson–Kendall–Roberts model 

(45) to define the adhesive forces of cell-to-cell interactions. The iterative direct-forcing immersed boundary method (IBM) is commonly employed in simulations of the fluid–cell interface of blood. This method effectively captures the intricacies of the thin and flexible RBC membranes within various external flow fields. 

(46) The study by Xu et al. 

(44) also adopts this approach to bridge the fluid dynamics and RBC deformation through IBM. Yoon and You utilized the Maxwell model to define the viscosity of the RBC membrane. 

(47) It was discovered that the Maxwell model could represent the stress relaxation and unloading processes of the cell. Furthermore, the reduced flexibility of an RBC under particular situations such as infection is specified, which was unattainable by the Kelvin–Voigt model 

(48) when compared to the Maxwell model in the literature. The Yeoh hyperplastic material model was also adapted to predict the nonlinear elasticity property of RBCs with FEM employed to discretize the RBC membrane using shell-type elements. Gracka et al. 

(49) developed a numerical CFD model with a finite-volume parallel solver for multiphase blood flow simulation, where an updated Maxwell viscoelasticity model and a Discrete Phase Model are adopted. In the study, the adapted IBM, based on unstructured grids, simulates the flow behavior and shape change of the RBCs through fluid-structure coupling. It was found that the hybrid Euler–Lagrange (E–L) approach 

(50) for the development of the multiphase model offered better results in the simulated CFL region in the microchannels.To study the dynamics of individual behaviors of RBCs and the consequent non-Newtonian blood flow, cell-shape-resolved computational models are often adapted. The use of the boundary integral method has become prevalent in minimizing computational expenses, particularly in the exclusive determination of fluid velocity on the surfaces of RBCs, incorporating the option of employing IBM or particle-based techniques. The cell-shaped-resolved method has enabled an examination of cell to cell interactions within complex ambient or pulsatile flow conditions 

(51) surrounding RBC membranes. Recently, Rydquist et al. 

(52) have looked to integrate statistical information from macroscale simulations to obtain a comprehensive overview of RBC behavior within the immediate proximity of the flow through introduction of respective models characterizing membrane shape definition, tension, bending stresses of RBC membranes.At a macroscopic scale, continuum models have conventionally been adapted for assessing blood flow dynamics through the application of elasticity theory and fluid dynamics. However, particle-based methods are known for their simplicity and adaptability in modeling complex multiscale fluid structures. Meshless methods, such as the boundary element method (BEM), smoothed particle hydrodynamics (SPH), and dissipative particle dynamics (DPD), are often used in particle-based characterization of RBCs and the surrounding fluid. By representing the fluid as discrete particles, meshless methods provide insights into the status and movement of the multiphase fluid. These methods allow for the investigation of cellular structures and microscopic interactions that affect blood rheology. Non-confronting mesh methods like IBM can also be used to couple a fluid solver such as FEM, FVM, or the Lattice Boltzmann Method (LBM) through membrane representation of RBCs. In comparison to conventional CFD methods, LBM has been viewed as a favorable numerical approach for solving the N–S equations and the simulation of multiphase flows. LBM exhibits the notable advantage of being amenable to high-performance parallel computing environments due to its inherently local dynamics. In contrast to DPD and SPH where RBC membranes are modeled as physically interconnected particles, LBM employs the IBM to account for the deformation dynamics of RBCs 

(53,54) under shear flows in complex channel geometries. 

(54,55) However, it is essential to acknowledge that the utilization of LBM in simulating RBC flows often entails a significant computational overhead, being a primary challenge in this context. Krüger et al. 

(56) proposed utilizing LBM as a fluid solver, IBM to couple the fluid and FEM to compute the response of membranes to deformation under immersed fluids. This approach decouples the fluid and membranes but necessitates significant computational effort due to the requirements of both meshes and particles.Despite the accuracy of current blood flow models, simulating complex conditions remains challenging because of the high computational load and cost. Balachandran Nair et al. 

(57) suggested a reduced order model of RBC under the framework of DEM, where the RBC is represented by overlapping constituent rigid spheres. The Morse potential force is adapted to account for the RBC aggregation exhibited by cell to cell interactions among RBCs at different distances. Based upon the IBM, the reduced-order RBC model is adapted to simulate blood flow transport for validation under both single and multiple RBCs with a resolved CFD-DEM solver. 

(58) In the resolved CFD-DEM model, particle sizes are larger than the grid size for a more accurate computation of the surrounding flow field. A continuous forcing approach is taken to describe the momentum source of the governing equation prior to discretization, which is different from a Direct Forcing Method (DFM). 

(59) As no body-conforming moving mesh is required, the continuous forcing approach offers lower complexity and reduced cost when compared to the DFM. Piquet et al. 

(60) highlighted the high complexity of the DFM due to its reliance on calculating an additional immersed boundary flux for the velocity field to ensure its divergence-free condition.The fluid–structure interaction (FSI) method has been advocated to connect the dynamic interplay of RBC membranes and fluid plasma within blood flow such as the coupling of continuum–particle interactions. However, such methodology is generally adapted for anatomical configurations such as arteries 

(61,62) and capillaries, 

(63) where both the structural components and the fluid domain undergo substantial deformation due to the moving boundaries. Due to the scope of the Review being blood flow simulation within microchannels of LOC devices without deformable boundaries, the Review of the FSI method will not be further carried out.In general, three numerical methods are broadly used: mesh-based, particle-based, and hybrid mesh–particle techniques, based on the spatial scale and the fundamental numerical approach, mesh-based methods tend to neglect the effects of individual particles, assuming a continuum and being efficient in terms of time and cost. However, the particle-based approach highlights more of the microscopic and mesoscopic level, where the influence of individual RBCs is considered. A review from Freund et al. 

(64) addressed the three numerical methodologies and their respective modeling approaches of RBC dynamics. Given the complex mechanics and the diverse levels of study concerning numerical simulations of blood and cellular flow, a broad spectrum of numerical methods for blood has been subjected to extensive review. 

(64−70) Ye at al. 

(65) offered an extensive review of the application of the DPD, SPH, and LBM for numerical simulations of RBC, while Rathnayaka et al. 

(67) conducted a review of the particle-based numerical modeling for liquid marbles through drawing parallels to the transport of RBCs in microchannels. A comparative analysis between conventional CFD methods and particle-based approaches for cellular and blood flow dynamic simulation can be found under the review by Arabghahestani et al. 

(66) Literature by Li et al. 

(68) and Beris et al. 

(69) offer an overview of both continuum-based models at micro/macroscales and multiscale particle-based models encompassing various length and temporal dimensions. Furthermore, these reviews deliberate upon the potential of coupling continuum-particle methods for blood plasma and RBC modeling. Arciero et al. 

(70) investigated various modeling approaches encompassing cellular interactions, such as cell to cell or plasma interactions and the individual cellular phases. A concise overview of the reviews is provided in Table 2 for reference.

Table 2. List of Reviews for Numerical Approaches Employed in Blood Flow Simulation

ReferenceNumerical methods
Li et al. (2013) (68)Continuum-based modeling (BIM), particle-based modeling (LBM, LB-FE, SPH, DPD)
Freund (2014) (64)RBC dynamic modeling (continuum-based modeling, complementary discrete microstructure modeling), blood flow dynamic modeling (FDM, IBM, LBM, particle-mesh methods, coupled boundary integral and mesh-based methods, DPD)
Ye et al. (2016) (65)DPD, SPH, LBM, coupled IBM-Smoothed DPD
Arciero et al. (2017) (70)LBM, IBM, DPD, conventional CFD Methods (FDM, FVM, FEM)
Arabghahestani et al. (2019) (66)Particle-based methods (LBM, DPD, direct simulation Monte Carlo, molecular dynamics), SPH, conventional CFD methods (FDM, FVM, FEM)
Beris et al. (2021) (69)DPD, smoothed DPD, IBM, LBM, BIM
Rathnayaka (2022) (67)SPH, CG, LBM

3. Capillary Driven Blood Flow in LOC Systems

ARTICLE SECTIONS

Jump To


3.1. Capillary Driven Flow Phenomena

Capillary driven (CD) flow is a pivotal mechanism in passive microfluidic flow systems 

(9) such as the blood circulation system and LOC systems. 

(71) CD flow is essentially the movement of a liquid to flow against drag forces, where the capillary effect exerts a force on the liquid at the borders, causing a liquid–air meniscus to flow despite gravity or other drag forces. A capillary pressure drops across the liquid–air interface with surface tension in the capillary radius and contact angle. The capillary effect depends heavily on the interaction between the different properties of surface materials. Different values of contact angles can be manipulated and obtained under varying levels of surface wettability treatments to manipulate the surface properties, resulting in different CD blood delivery rates for medical diagnostic device microchannels. CD flow techniques are appealing for many LOC devices, because they require no external energy. However, due to the passive property of liquid propulsion by capillary forces and the long-term instability of surface treatments on channel walls, the adaptability of CD flow in geometrically complex LOC devices may be limited.

3.2. Theoretical and Numerical Modeling of Capillary Driven Blood Flow

3.2.1. Theoretical Basis and Assumptions of Microfluidic Flow

The study of transport phenomena regarding either blood flow driven by capillary forces or externally applied forces under microfluid systems all demands a comprehensive recognition of the significant differences in flow dynamics between microscale and macroscale. The fundamental assumptions and principles behind fluid transport at the microscale are discussed in this section. Such a comprehension will lay the groundwork for the following analysis of the theoretical basis of capillary forces and their role in blood transport in LOC systems.

At the macroscale, fluid dynamics are often strongly influenced by gravity due to considerable fluid mass. However, the high surface to volume ratio at the microscale shifts the balance toward surface forces (e.g., surface tension and viscous forces), much larger than the inertial force. This difference gives rise to transport phenomena unique to microscale fluid transport, such as the prevalence of laminar flow due to a very low Reynolds number (generally lower than 1). Moreover, the fluid in a microfluidic system is often assumed to be incompressible due to the small flow velocity, indicating constant fluid density in both space and time.Microfluidic flow behaviors are governed by the fundamental principles of mass and momentum conservation, which are encapsulated in the continuity equation and the Navier–Stokes (N–S) equation. The continuity equation describes the conservation of mass, while the N–S equation captures the spatial and temporal variations in velocity, pressure, and other physical parameters. Under the assumption of the negligible influence of gravity in microfluidic systems, the continuity equation and the Eulerian representation of the incompressible N–S equation can be expressed as follows:

∇·𝐮⇀=0∇·�⇀=0

(7)

−∇𝑝+𝜇∇2𝐮⇀+∇·𝝉⇀−𝐅⇀=0−∇�+�∇2�⇀+∇·�⇀−�⇀=0

(8)Here, p is the pressure, u is the fluid viscosity, 

𝝉⇀�⇀ represents the stress tensor, and F is the body force exerted by external forces if present.

3.2.2. Theoretical Basis and Modeling of Capillary Force in LOC Systems

The capillary force is often the major driving force to manipulate and transport blood without an externally applied force in LOC systems. Forces induced by the capillary effect impact the free surface of fluids and are represented not directly in the Navier–Stokes equations but through the pressure boundary conditions of the pressure term p. For hydrophilic surfaces, the liquid generally induces a contact angle between 0° and 30°, encouraging the spread and attraction of fluid under a positive cos θ condition. For this condition, the pressure drop becomes positive and generates a spontaneous flow forward. A hydrophobic solid surface repels the fluid, inducing minimal contact. Generally, hydrophobic solids exhibit a contact angle larger than 90°, inducing a negative value of cos θ. Such a value will result in a negative pressure drop and a flow in the opposite direction. The induced contact angle is often utilized to measure the wall exposure of various surface treatments on channel walls where different wettability gradients and surface tension effects for CD flows are established. Contact angles between different interfaces are obtainable through standard values or experimental methods for reference. 

(72)For the characterization of the induced force by the capillary effect, the Young–Laplace (Y–L) equation 

(73) is widely employed. In the equation, the capillary is considered a pressure boundary condition between the two interphases. Through the Y–L equation, the capillary pressure force can be determined, and subsequently, the continuity and momentum balance equations can be solved to obtain the blood filling rate. Kim et al. 

(74) studied the effects of concentration and exposure time of a nonionic surfactant, Silwet L-77, on the performance of a polydimethylsiloxane (PDMS) microchannel in terms of plasma and blood self-separation. The study characterized the capillary pressure force by incorporating the Y–L equation and further evaluated the effects of the changing contact angle due to different levels of applied channel wall surface treatments. The expression of the Y–L equation utilized by Kim et al. 

(74) is as follows:

𝑃=−𝜎(cos𝜃b+cos𝜃tℎ+cos𝜃l+cos𝜃r𝑤)�=−�(cos⁡�b+cos⁡�tℎ+cos⁡�l+cos⁡�r�)

(9)where σ is the surface tension of the liquid and θ

bθ

tθ

l, and θ

r are the contact angle values between the liquid and the bottom, top, left, and right walls, respectively. A numerical simulation through Coventor software is performed to evaluate the dynamic changes in the filling rate within the microchannel. The simulation results for the blood filling rate in the microchannel are expressed at a specific time stamp, shown in Figure 2. The results portray an increasing instantaneous filling rate of blood in the microchannel following the decrease in contact angle induced by a higher concentration of the nonionic surfactant treated to the microchannel wall.

Figure 2. Numerical simulation of filling rate of capillary driven blood flow under various contact angle conditions at a specific timestamp. (74) Reproduced with permission from ref (74). Copyright 2010 Elsevier.

When in contact with hydrophilic or hydrophobic surfaces, blood forms a meniscus with a contact angle due to surface tension. The Lucas–Washburn (L–W) equation 

(75) is one of the pioneering theoretical definitions for the position of the meniscus over time. In addition, the L–W equation provides the possibility for research to obtain the velocity of the blood formed meniscus through the derivation of the meniscus position. The L–W equation 

(75) can be shown below:

𝐿(𝑡)=𝑅𝜎cos(𝜃)𝑡2𝜇⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯√�(�)=��⁡cos(�)�2�

(10)Here L(t) represents the distance of the liquid driven by the capillary forces. However, the generalized L–W equation solely assumes the constant physical properties from a Newtonian fluid rather than considering the non-Newtonian fluid behavior of blood. Cito et al. 

(76) constructed an enhanced version of the L–W equation incorporating the power law to consider the RBC aggregation and the FL effect. The non-Newtonian fluid apparent viscosity under the Power Law model is defined as

𝜇=𝑘·(𝛾˙)𝑛−1�=�·(�˙)�−1

(11)where γ̇ is the strain rate tensor defined as 

𝛾˙=12𝛾˙𝑖𝑗𝛾˙𝑗𝑖⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯√�˙=12�˙���˙��. The stress tensor term τ is computed as τ = μγ̇

ij. The updated L–W equation by Cito 

(76) is expressed as

𝐿(𝑡)=𝑅[(𝑛+13𝑛+1)(𝜎cos(𝜃)𝑅𝑘)1/𝑛𝑡]𝑛/𝑛+1�(�)=�[(�+13�+1)(�⁡cos(�)��)1/��]�/�+1

(12)where k is the flow consistency index and n is the power law index, respectively. The power law index, from the Power Law model, characterizes the extent of the non-Newtonian behavior of blood. Both the consistency and power law index rely on blood properties such as hematocrit, the appearance of the FL effect, the formation of RBC aggregates, etc. The updated L–W equation computes the location and velocity of blood flow caused by capillary forces at specified time points within the LOC devices, taking into account the effects of blood flow characteristics such as RBC aggregation and the FL effect on dynamic blood viscosity.Apart from the blood flow behaviors triggered by inherent blood properties, unique flow conditions driven by capillary forces that are portrayed under different microchannel geometries also hold crucial implications for CD blood delivery. Berthier et al. 

(77) studied the spontaneous Concus–Finn condition, the condition to initiate the spontaneous capillary flow within a V-groove microchannel, as shown in Figure 3(a) both experimentally and numerically. Through experimental studies, the spontaneous Concus–Finn filament development of capillary driven blood flow is observed, as shown in Figure 3(b), while the dynamic development of blood flow is numerically simulated through CFD simulation.

Figure 3. (a) Sketch of the cross-section of Berthier’s V-groove microchannel, (b) experimental view of blood in the V-groove microchannel, (78) (c) illustration of the dynamic change of the extension of filament from FLOW 3D under capillary flow at three increasing time intervals. (78) Reproduced with permission from ref (78). Copyright 2014 Elsevier.

Berthier et al. 

(77) characterized the contact angle needed for the initiation of the capillary driving force at a zero-inlet pressure, through the half-angle (α) of the V-groove geometry layout, and its relation to the Concus–Finn filament as shown below:

𝜃<𝜋2−𝛼sin𝛼1+2(ℎ2/𝑤)sin𝛼<cos𝜃{�<�2−�sin⁡�1+2(ℎ2/�)⁡sin⁡�<cos⁡�

(13)Three possible regimes were concluded based on the contact angle value for the initiation of flow and development of Concus–Finn filament:

𝜃>𝜃1𝜃1>𝜃>𝜃0𝜃0no SCFSCF without a Concus−Finn filamentSCF without a Concus−Finn filament{�>�1no SCF�1>�>�0SCF without a Concus−Finn filament�0SCF without a Concus−Finn filament

(14)Under Newton’s Law, the force balance with low Reynolds and Capillary numbers results in the neglect of inertial terms. The force balance between the capillary forces and the viscous force induced by the channel wall is proposed to derive the analytical fluid velocity. This relation between the two forces offers insights into the average flow velocity and the penetration distance function dependent on time. The apparent blood viscosity is defined by Berthier et al. 

(78) through Casson’s law, 

(23) given in eq 1. The research used the FLOW-3D program from Flow Science Inc. software, which solves transient, free-surface problems using the FDM in multiple dimensions. The Volume of Fluid (VOF) method 

(79) is utilized to locate and track the dynamic extension of filament throughout the advancing interface within the channel ahead of the main flow at three progressing time stamps, as depicted in Figure 3(c).

4. Electro-osmotic Flow (EOF) in LOC Systems

ARTICLE SECTIONS

Jump To


The utilization of external forces, such as electric fields, has significantly broadened the possibility of manipulating microfluidic flow in LOC systems. 

(80) Externally applied electric field forces induce a fluid flow from the movement of ions in fluid terms as the “electro-osmotic flow” (EOF).Unique transport phenomena, such as enhanced flow velocity and flow instability, induced by non-Newtonian fluids, particularly viscoelastic fluids, under EOF, have sparked considerable interest in microfluidic devices with simple or complicated geometries within channels. 

(81) However, compared to the study of Newtonian fluids and even other electro-osmotic viscoelastic fluid flows, the literature focusing on the theoretical and numerical modeling of electro-osmotic blood flow is limited due to the complexity of blood properties. Consequently, to obtain a more comprehensive understanding of the complex blood flow behavior under EOF, theoretical and numerical studies of the transport phenomena in the EOF section will be based on the studies of different viscoelastic fluids under EOF rather than that of blood specifically. Despite this limitation, we believe these studies offer valuable insights that can help understand the complex behavior of blood flow under EOF.

4.1. EOF Phenomena

Electro-osmotic flow occurs at the interface between the microchannel wall and bulk phase solution. When in contact with the bulk phase, solution ions are absorbed or dissociated at the solid–liquid interface, resulting in the formation of a charge layer, as shown in Figure 4. This charged channel surface wall interacts with both negative and positive ions in the bulk sample, causing repulsion and attraction forces to create a thin layer of immobilized counterions, known as the Stern layer. The induced electric potential from the wall gradually decreases with an increase in the distance from the wall. The Stern layer potential, commonly termed the zeta potential, controls the intensity of the electrostatic interactions between mobile counterions and, consequently, the drag force from the applied electric field. Next to the Stern layer is the diffuse mobile layer, mainly composed of a mobile counterion. These two layers constitute the “electrical double layer” (EDL), the thickness of which is directly proportional to the ionic strength (concentration) of the bulk fluid. The relationship between the two parameters is characterized by a Debye length (λ

D), expressed as

𝜆𝐷=𝜖𝑘B𝑇2(𝑍𝑒)2𝑐0⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯√��=��B�2(��)2�0

(15)where ϵ is the permittivity of the electrolyte solution, k

B is the Boltzmann constant, T is the electron temperature, Z is the integer valence number, e is the elementary charge, and c

0 is the ionic density.

Figure 4. Schematic diagram of an electro-osmotic flow in a microchannel with negative surface charge. (82) Reproduced with permission from ref (82). Copyright 2012 Woodhead Publishing.

When an electric field is applied perpendicular to the EDL, viscous drag is generated due to the movement of excess ions in the EDL. Electro-osmotic forces can be attributed to the externally applied electric potential (ϕ) and the zeta potential, the system wall induced potential by charged walls (ψ). As illustrated in Figure 4, the majority of ions in the bulk phase have a uniform velocity profile, except for a shear rate condition confined within an extremely thin Stern layer. Therefore, EOF displays a unique characteristic of a “near flat” or plug flow velocity profile, different from the parabolic flow typically induced by pressure-driven microfluidic flow (Hagen–Poiseuille flow). The plug-shaped velocity profile of the EOF possesses a high shear rate above the Stern layer.Overall, the EOF velocity magnitude is typically proportional to the Debye Length (λ

D), zeta potential, and magnitude of the externally applied electric field, while a more viscous liquid reduces the EOF velocity.

4.2. Modeling on Electro-osmotic Viscoelastic Fluid Flow

4.2.1. Theoretical Basis of EOF Mechanisms

The EOF of an incompressible viscoelastic fluid is commonly governed by the continuity and incompressible N–S equations, as shown in eqs 7 and 8, where the stress tensor and the electrostatic force term are coupled. The electro-osmotic body force term F, representing the body force exerted by the externally applied electric force, is defined as 

𝐹⇀=𝑝𝐸𝐸⇀�⇀=���⇀, where ρ

E and 

𝐸⇀�⇀ are the net electric charge density and the applied external electric field, respectively.Numerous models are established to theoretically study the externally applied electric potential and the system wall induced potential by charged walls. The following Laplace equation, expressed as eq 16, is generally adapted and solved to calculate the externally applied potential (ϕ).

∇2𝜙=0∇2�=0

(16)Ion diffusion under applied electric fields, together with mass transport resulting from convection and diffusion, transports ionic solutions in bulk flow under electrokinetic processes. The Nernst–Planck equation can describe these transport methods, including convection, diffusion, and electro-diffusion. Therefore, the Nernst–Planck equation is used to determine the distribution of the ions within the electrolyte. The electric potential induced by the charged channel walls follows the Poisson–Nernst–Plank (PNP) equation, which can be written as eq 17.

∇·[𝐷𝑖∇𝑛𝑖−𝑢⇀𝑛𝑖+𝑛𝑖𝐷𝑖𝑧𝑖𝑒𝑘𝑏𝑇∇(𝜙+𝜓)]=0∇·[��∇��−�⇀��+����������∇(�+�)]=0

(17)where D

in

i, and z

i are the diffusion coefficient, ionic concentration, and ionic valence of the ionic species I, respectively. However, due to the high nonlinearity and numerical stiffness introduced by different lengths and time scales from the PNP equations, the Poisson–Boltzmann (PB) model is often considered the major simplified method of the PNP equation to characterize the potential distribution of the EDL region in microchannels. In the PB model, it is assumed that the ionic species in the fluid follow the Boltzmann distribution. This model is typically valid for steady-state problems where charge transport can be considered negligible, the EDLs do not overlap with each other, and the intrinsic potentials are low. It provides a simplified representation of the potential distribution in the EDL region. The PB equation governing the EDL electric potential distribution is described as

∇2𝜓=(2𝑒𝑧𝑛0𝜀𝜀0)sinh(𝑧𝑒𝜓𝑘b𝑇)∇2�=(2���0��0)⁡sinh(����b�)

(18)where n

0 is the ion bulk concentration, z is the ionic valence, and ε

0 is the electric permittivity in the vacuum. Under low electric potential conditions, an even further simplified model to illustrate the EOF phenomena is the Debye–Hückel (DH) model. The DH model is derived by obtaining a charge density term by expanding the exponential term of the Boltzmann equation in a Taylor series.

4.2.2. EOF Modeling for Viscoelastic Fluids

Many studies through numerical modeling were performed to obtain a deeper understanding of the effect exhibited by externally applied electric fields on viscoelastic flow in microchannels under various geometrical designs. Bello et al. 

(83) found that methylcellulose solution, a non-Newtonian polymer solution, resulted in stronger electro-osmotic mobility in experiments when compared to the predictions by the Helmholtz–Smoluchowski equation, which is commonly used to define the velocity of EOF of a Newtonian fluid. Being one of the pioneers to identify the discrepancies between the EOF of Newtonian and non-Newtonian fluids, Bello et al. attributed such discrepancies to the presence of a very high shear rate in the EDL, resulting in a change in the orientation of the polymer molecules. Park and Lee 

(84) utilized the FVM to solve the PB equation for the characterization of the electric field induced force. In the study, the concept of fractional calculus for the Oldroyd-B model was adapted to illustrate the elastic and memory effects of viscoelastic fluids in a straight microchannel They observed that fluid elasticity and increased ratio of viscoelastic fluid contribution to overall fluid viscosity had a significant impact on the volumetric flow rate and sensitivity of velocity to electric field strength compared to Newtonian fluids. Afonso et al. 

(85) derived an analytical expression for EOF of viscoelastic fluid between parallel plates using the DH model to account for a zeta potential condition below 25 mV. The study established the understanding of the electro-osmotic viscoelastic fluid flow under low zeta potential conditions. Apart from the electrokinetic forces, pressure forces can also be coupled with EOF to generate a unique fluid flow behavior within the microchannel. Sousa et al. 

(86) analytically studied the flow of a standard viscoelastic solution by combining the pressure gradient force with an externally applied electric force. It was found that, at a near wall skimming layer and the outer layer away from the wall, macromolecules migrating away from surface walls in viscoelastic fluids are observed. In the study, the Phan-Thien Tanner (PTT) constitutive model is utilized to characterize the viscoelastic properties of the solution. The approach is found to be valid when the EDL is much thinner than the skimming layer under an enhanced flow rate. Zhao and Yang 

(87) solved the PB equation and Carreau model for the characterization of the EOF mechanism and non-Newtonian fluid respectively through the FEM. The numerical results depict that, different from the EOF of Newtonian fluids, non-Newtonian fluids led to an increase of electro-osmotic mobility for shear thinning fluids but the opposite for shear thickening fluids.Like other fluid transport driving forces, EOF within unique geometrical layouts also portrays unique transport phenomena. Pimenta and Alves 

(88) utilized the FVM to perform numerical simulations of the EOF of viscoelastic fluids considering the PB equation and the Oldroyd-B model, in a cross-slot and flow-focusing microdevices. It was found that electroelastic instabilities are formed due to the development of large stresses inside the EDL with streamlined curvature at geometry corners. Bezerra et al. 

(89) used the FDM to numerically analyze the vortex formation and flow instability from an electro-osmotic non-Newtonian fluid flow in a microchannel with a nozzle geometry and parallel wall geometry setting. The PNP equation is utilized to characterize the charge motion in the EOF and the PTT model for non-Newtonian flow characterization. A constriction geometry is commonly utilized in blood flow adapted in LOC systems due to the change in blood flow behavior under narrow dimensions in a microchannel. Ji et al. 

(90) recently studied the EOF of viscoelastic fluid in a constriction microchannel connected by two relatively big reservoirs on both ends (as seen in Figure 5) filled with the polyacrylamide polymer solution, a viscoelastic fluid, and an incompressible monovalent binary electrolyte solution KCl.

Figure 5. Schematic diagram of a negatively charged constriction microchannel connected to two reservoirs at both ends. An electro-osmotic flow is induced in the system by the induced potential difference between the anode and cathode. (90) Reproduced with permission from ref (90). Copyright 2021 The Authors, under the terms of the Creative Commons (CC BY 4.0) License https://creativecommons.org/licenses/by/4.0/.

In studying the EOF of viscoelastic fluids, the Oldroyd-B model is often utilized to characterize the polymeric stress tensor and the deformation rate of the fluid. The Oldroyd-B model is expressed as follows:

𝜏=𝜂p𝜆(𝐜−𝐈)�=�p�(�−�)

(19)where η

p, λ, c, and I represent the polymer dynamic viscosity, polymer relaxation time, symmetric conformation tensor of the polymer molecules, and the identity matrix, respectively.A log-conformation tensor approach is taken to prevent convergence difficulty induced by the viscoelastic properties. The conformation tensor (c) in the polymeric stress tensor term is redefined by a new tensor (Θ) based on the natural logarithm of the c. The new tensor is defined as

Θ=ln(𝐜)=𝐑ln(𝚲)𝐑Θ=ln(�)=�⁡ln(�)�

(20)in which Λ is the diagonal matrix and R is the orthogonal matrix.Under the new conformation tensor, the induced EOF of a viscoelastic fluid is governed by the continuity and N–S equations adapting the Oldroyd-B model, which is expressed as

∂𝚯∂𝑡+𝐮·∇𝚯=𝛀Θ−ΘΩ+2𝐁+1𝜆(eΘ−𝐈)∂�∂�+�·∇�=�Θ−ΘΩ+2�+1�(eΘ−�)

(21)where Ω and B represent the anti-symmetric matrix and the symmetric traceless matrix of the decomposition of the velocity gradient tensor ∇u, respectively. The conformation tensor can be recovered by c = exp(Θ). The PB model and Laplace equation are utilized to characterize the charged channel wall induced potential and the externally applied potential.The governing equations are numerically solved through the FVM by RheoTool, 

(42) an open-source viscoelastic EOF solver on the OpenFOAM platform. A SIMPLEC (Semi-Implicit Method for Pressure Linked Equations-Consistent) algorithm was applied to solve the velocity-pressure coupling. The pressure field and velocity field were computed by the PCG (Preconditioned Conjugate Gradient) solver and the PBiCG (Preconditioned Biconjugate Gradient) solver, respectively.Ranging magnitudes of an applied electric field or fluid concentration induce both different streamlines and velocity magnitudes at various locations and times of the microchannel. In the study performed by Ji et al., 

(90) notable fluctuation of streamlines and vortex formation is formed at the upper stream entrance of the constriction as shown in Figure 6(a) and (b), respectively, due to the increase of electrokinetic effect, which is seen as a result of the increase in polymeric stress (τ

xx). 

(90) The contraction geometry enhances the EOF velocity within the constriction channel under high E

app condition (600 V/cm). Such phenomena can be attributed to the dependence of electro-osmotic viscoelastic fluid flow on the system wall surface and bulk fluid properties. 

(91)

Figure 6. Schematic diagram of vortex formation and streamlines of EOF depicting flow instability at (a) 1.71 s and (b) 1.75 s. Spatial distribution of the elastic normal stress at (c) high Eapp condition. Streamline of an electro-osmotic flow under Eapp of 600 V/cm (90) for (d) non-Newtonian and (e) Newtonian fluid through a constriction geometry. Reproduced with permission from ref (90). Copyright 2021 The Authors, under the terms of the Creative Commons (CC BY 4.0) License https://creativecommons.org/licenses/by/4.0/.

As elastic normal stress exceeds the local shear stress, flow instability and vortex formation occur. The induced elastic stress under EOF not only enhances the instability of the flow but often generates an irregular secondary flow leading to strong disturbance. 

(92) It is also vital to consider the effect of the constriction layout of microchannels on the alteration of the field strength within the system. The contraction geometry enhances a larger electric field strength compared with other locations of the channel outside the constriction region, resulting in a higher velocity gradient and stronger extension on the polymer within the viscoelastic solution. Following the high shear flow condition, a higher magnitude of stretch for polymer molecules in viscoelastic fluids exhibits larger elastic stresses and enhancement of vortex formation at the region. 

(93)As shown in Figure 6(c), significant elastic normal stress occurs at the inlet of the constriction microchannel. Such occurrence of a polymeric flow can be attributed to the dominating elongational flow, giving rise to high deformation of the polymers within the viscoelastic fluid flow, resulting in higher elastic stress from the polymers. Such phenomena at the entrance result in the difference in velocity streamline as circled in Figure 6(d) compared to that of the Newtonian fluid at the constriction entrance in Figure 6(e). 

(90) The difference between the Newtonian and polymer solution at the exit, as circled in Figure 6(d) and (e), can be attributed to the extrudate swell effect of polymers 

(94) within the viscoelastic fluid flow. The extrudate swell effect illustrates that, as polymers emerge from the constriction exit, they tend to contract in the flow direction and grow in the normal direction, resulting in an extrudate diameter greater than the channel size. The deformation of polymers within the polymeric flow at both the entrance and exit of the contraction channel facilitates the change in shear stress conditions of the flow, leading to the alteration in streamlines of flows for each region.

4.3. EOF Applications in LOC Systems

4.3.1. Mixing in LOC Systems

Rather than relying on the micromixing controlled by molecular diffusion under low Reynolds number conditions, active mixers actively leverage convective instability and vortex formation induced by electro-osmotic flows from alternating current (AC) or direct current (DC) electric fields. Such adaptation is recognized as significant breakthroughs for promotion of fluid mixing in chemical and biological applications such as drug delivery, medical diagnostics, chemical synthesis, and so on. 

(95)Many researchers proposed novel designs of electro-osmosis micromixers coupled with numerical simulations in conjunction with experimental findings to increase their understanding of the role of flow instability and vortex formation in the mixing process under electrokinetic phenomena. Matsubara and Narumi 

(96) numerically modeled the mixing process in a microchannel with four electrodes on each side of the microchannel wall, which generated a disruption through unstable electro-osmotic vortices. It was found that particle mixing was sensitive to both the convection effect induced by the main and secondary vortex within the micromixer and the change in oscillation frequency caused by the supplied AC voltage when the Reynolds number was varied. Qaderi et al. 

(97) adapted the PNP equation to numerically study the effect of the geometry and zeta potential configuration of the microchannel on the mixing process with a combined electro-osmotic pressure driven flow. It was reported that the application of heterogeneous zeta potential configuration enhances the mixing efficiency by around 23% while the height of the hurdles increases the mixing efficiency at most 48.1%. Cho et al. 

(98) utilized the PB model and Laplace equation to numerically simulate the electro-osmotic non-Newtonian fluid mixing process within a wavy and block layout of microchannel walls. The Power Law model is adapted to describe the fluid rheological characteristic. It was found that shear-thinning fluids possess a higher volumetric flow rate, which could result in poorer mixing efficiency compared to that of Newtonian fluids. Numerous studies have revealed that flow instability and vortex generation, in particular secondary vortices produced by barriers or greater magnitudes of heterogeneous zeta potential distribution, enhance mixing by increasing bulk flow velocity and reducing flow distance.To better understand the mechanism of disturbance formed in the system due to externally applied forces, known as electrokinetic instability, literature often utilize the Rayleigh (Ra) number, 

(1) as described below:

𝑅𝑎𝑣=𝑢ev𝑢eo=(𝛾−1𝛾+1)2𝑊𝛿2𝐸el2𝐻2𝜁𝛿Ra�=�ev�eo=(�−1�+1)2��2�el2�2��

(22)where γ is the conductivity ratio of the two streams and can be written as 

𝛾=𝜎el,H𝜎el,L�=�el,H�el,L. The Ra number characterizes the ratio between electroviscous and electro-osmotic flow. A high Ra

v value often results in good mixing. It is evident that fluid properties such as the conductivity (σ) of the two streams play a key role in the formation of disturbances to enhance mixing in microsystems. At the same time, electrokinetic parameters like the zeta potential (ζ) in the Ra number is critical in the characterization of electro-osmotic velocity and a slip boundary condition at the microchannel wall.To understand the mixing result along the channel, the concentration field can be defined and simulated under the assumption of steady state conditions and constant diffusion coefficient for each of the working fluid within the system through the convection–diffusion equation as below:

∂𝑐𝒊∂𝑡+∇⇀(𝑐𝑖𝑢⇀−𝐷𝑖∇⇀𝑐𝒊)=0∂��∂�+∇⇀(���⇀−��∇⇀��)=0

(23)where c

i is the species concentration of species i and D

i is the diffusion coefficient of the corresponding species.The standard deviation of concentration (σ

sd) can be adapted to evaluate the mixing quality of the system. 

(97) The standard deviation for concentration at a specific portion of the channel may be calculated using the equation below:

𝜎sd=∫10(𝐶∗(𝑦∗)−𝐶m)2d𝑦∗∫10d𝑦∗⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯�sd=∫01(�*(�*)−�m)2d�*∫01d�*

(24)where C*(y*) and C

m are the non-dimensional concentration profile and the mean concentration at the portion, respectively. C* is the non-dimensional concentration and can be calculated as 

𝐶∗=𝐶𝐶ref�*=��ref, where C

ref is the reference concentration defined as the bulk solution concentration. The mean concentration profile can be calculated as 

𝐶m=∫10(𝐶∗(𝑦∗)d𝑦∗∫10d𝑦∗�m=∫01(�*(�*)d�*∫01d�*. With the standard deviation of concentration, the mixing efficiency 

(97) can then be calculated as below:

𝜀𝑥=1−𝜎sd𝜎sd,0��=1−�sd�sd,0

(25)where σ

sd,0 is the standard derivation of the case of no mixing. The value of the mixing efficiency is typically utilized in conjunction with the simulated flow field and concentration field to explore the effect of geometrical and electrokinetic parameters on the optimization of the mixing results.

5. Summary

ARTICLE SECTIONS

Jump To


5.1. Conclusion

Viscoelastic fluids such as blood flow in LOC systems are an essential topic to proceed with diagnostic analysis and research through microdevices in the biomedical and pharmaceutical industries. The complex blood flow behavior is tightly controlled by the viscoelastic characteristics of blood such as the dynamic viscosity and the elastic property of RBCs under various shear rate conditions. Furthermore, the flow behaviors under varied driving forces promote an array of microfluidic transport phenomena that are critical to the management of blood flow and other adapted viscoelastic fluids in LOC systems. This review addressed the blood flow phenomena, the complicated interplay between shear rate and blood flow behaviors, and their numerical modeling under LOC systems through the lens of the viscoelasticity characteristic. Furthermore, a theoretical understanding of capillary forces and externally applied electric forces leads to an in-depth investigation of the relationship between blood flow patterns and the key parameters of the two driving forces, the latter of which is introduced through the lens of viscoelastic fluids, coupling numerical modeling to improve the knowledge of blood flow manipulation in LOC systems. The flow disturbances triggered by the EOF of viscoelastic fluids and their impact on blood flow patterns have been deeply investigated due to their important role and applications in LOC devices. Continuous advancements of various numerical modeling methods with experimental findings through more efficient and less computationally heavy methods have served as an encouraging sign of establishing more accurate illustrations of the mechanisms for multiphase blood and other viscoelastic fluid flow transport phenomena driven by various forces. Such progress is fundamental for the manipulation of unique transport phenomena, such as the generated disturbances, to optimize functionalities offered by microdevices in LOC systems.

The following section will provide further insights into the employment of studied blood transport phenomena to improve the functionality of micro devices adapting LOC technology. A discussion of the novel roles that external driving forces play in microfluidic flow behaviors is also provided. Limitations in the computational modeling of blood flow and electrokinetic phenomena in LOC systems will also be emphasized, which may provide valuable insights for future research endeavors. These discussions aim to provide guidance and opportunities for new paths in the ongoing development of LOC devices that adapt blood flow.

5.2. Future Directions

5.2.1. Electro-osmosis Mixing in LOC Systems

Despite substantial research, mixing results through flow instability and vortex formation phenomena induced by electro-osmotic mixing still deviate from the effective mixing results offered by chaotic mixing results such as those seen in turbulent flows. However, recent discoveries of a mixing phenomenon that is generally observed under turbulent flows are found within electro-osmosis micromixers under low Reynolds number conditions. Zhao 

(99) experimentally discovered a rapid mixing process in an AC applied micromixer, where the power spectrum of concentration under an applied voltage of 20 V

p-p induces a −5/3 slope within a frequency range. This value of the slope is considered as the O–C spectrum in macroflows, which is often visible under relatively high Re conditions, such as the Taylor microscale Reynolds number Re > 500 in turbulent flows. 

(100) However, the Re value in the studied system is less than 1 at the specific location and applied voltage. A secondary flow is also suggested to occur close to microchannel walls, being attributed to the increase of convective instability within the system.Despite the experimental phenomenon proposed by Zhao et al., 

(99) the range of effects induced by vital parameters of an EOF mixing system on the enhanced mixing results and mechanisms of disturbance generated by the turbulent-like flow instability is not further characterized. Such a gap in knowledge may hinder the adaptability and commercialization of the discovery of micromixers. One of the parameters for further evaluation is the conductivity gradient of the fluid flow. A relatively strong conductivity gradient (5000:1) was adopted in the system due to the conductive properties of the two fluids. The high conductivity gradients may contribute to the relatively large Rayleigh number and differences in EDL layer thickness, resulting in an unusual disturbance in laminar flow conditions and enhanced mixing results. However, high conductivity gradients are not always achievable by the working fluids due to diverse fluid properties. The reliance on turbulent-like phenomena and rapid mixing results in a large conductivity gradient should be established to prevent the limited application of fluids for the mixing system. In addition, the proposed system utilizes distinct zeta potential distributions at the top and bottom walls due to their difference in material choices, which may be attributed to the flow instability phenomena. Further studies should be made on varying zeta potential magnitude and distribution to evaluate their effect on the slip boundary conditions of the flow and the large shear rate condition close to the channel wall of EOF. Such a study can potentially offer an optimized condition in zeta potential magnitude through material choices and geometrical layout of the zeta potential for better mixing results and manipulation of mixing fluid dynamics. The two vital parameters mentioned above can be varied with the aid of numerical simulation to understand the effect of parameters on the interaction between electro-osmotic forces and electroviscous forces. At the same time, the relationship of developed streamlines of the simulated velocity and concentration field, following their relationship with the mixing results, under the impact of these key parameters can foster more insight into the range of impact that the two parameters have on the proposed phenomena and the microfluidic dynamic principles of disturbances.

In addition, many of the current investigations of electrokinetic mixers commonly emphasize the fluid dynamics of mixing for Newtonian fluids, while the utilization of biofluids, primarily viscoelastic fluids such as blood, and their distinctive response under shear forces in these novel mixing processes of LOC systems are significantly less studied. To develop more compatible microdevice designs and efficient mixing outcomes for the biomedical industry, it is necessary to fill the knowledge gaps in the literature on electro-osmotic mixing for biofluids, where properties of elasticity, dynamic viscosity, and intricate relationship with shear flow from the fluid are further considered.

5.2.2. Electro-osmosis Separation in LOC Systems

Particle separation in LOC devices, particularly in biological research and diagnostics, is another area where disturbances may play a significant role in optimization. 

(101) Plasma analysis in LOC systems under precise control of blood flow phenomena and blood/plasma separation procedures can detect vital information about infectious diseases from particular antibodies and foreign nucleic acids for medical treatments, diagnostics, and research, 

(102) offering more efficient results and simple operating procedures compared to that of the traditional centrifugation method for blood and plasma separation. However, the adaptability of LOC devices for blood and plasma separation is often hindered by microchannel clogging, where flow velocity and plasma yield from LOC devices is reduced due to occasional RBC migration and aggregation at the filtration entrance of microdevices. 

(103)It is important to note that the EOF induces flow instability close to microchannel walls, which may provide further solutions to clogging for the separation process of the LOC systems. Mohammadi et al. 

(104) offered an anti-clogging effect of RBCs at the blood and plasma separating device filtration entry, adjacent to the surface wall, through RBC disaggregation under high shear rate conditions generated by a forward and reverse EOF direction.

Further theoretical and numerical research can be conducted to characterize the effect of high shear rate conditions near microchannel walls toward the detachment of binding blood cells on surfaces and the reversibility of aggregation. Through numerical modeling with varying electrokinetic parameters to induce different degrees of disturbances or shear conditions at channel walls, it may be possible to optimize and better understand the process of disrupting the forces that bind cells to surface walls and aggregated cells at filtration pores. RBCs that migrate close to microchannel walls are often attracted by the adhesion force between the RBC and the solid surface originating from the van der Waals forces. Following RBC migration and attachment by adhesive forces adjacent to the microchannel walls as shown in Figure 7, the increase in viscosity at the region causes a lower shear condition and encourages RBC aggregation (cell–cell interaction), which clogs filtering pores or microchannels and reduces flow velocity at filtration region. Both the impact that shear forces and disturbances may induce on cell binding forces with surface walls and other cells leading to aggregation may suggest further characterization. Kinetic parameters such as activation energy and the rate-determining step for cell binding composition attachment and detachment should be considered for modeling the dynamics of RBCs and blood flows under external forces in LOC separation devices.

Figure 7. Schematic representations of clogging at a microchannel pore following the sequence of RBC migration, cell attachment to channel walls, and aggregation. (105) Reproduced with permission from ref (105). Copyright 2018 The Authors under the terms of the Creative Commons (CC BY 4.0) License https://creativecommons.org/licenses/by/4.0/.

5.2.3. Relationship between External Forces and Microfluidic Systems

In blood flow, a thicker CFL suggests a lower blood viscosity, suggesting a complex relationship between shear stress and shear rate, affecting the blood viscosity and blood flow. Despite some experimental and numerical studies on electro-osmotic non-Newtonian fluid flow, limited literature has performed an in-depth investigation of the role that applied electric forces and other external forces could play in the process of CFL formation. Additional studies on how shear rates from external forces affect CFL formation and microfluidic flow dynamics can shed light on the mechanism of the contribution induced by external driving forces to the development of a separate phase of layer, similar to CFL, close to the microchannel walls and distinct from the surrounding fluid within the system, then influencing microfluidic flow dynamics.One of the mechanisms of phenomena to be explored is the formation of the Exclusion Zone (EZ) region following a “Self-Induced Flow” (SIF) phenomenon discovered by Li and Pollack, 

(106) as shown in Figure 8(a) and (b), respectively. A spontaneous sustained axial flow is observed when hydrophilic materials are immersed in water, resulting in the buildup of a negative layer of charges, defined as the EZ, after water molecules absorb infrared radiation (IR) energy and break down into H and OH

+.

Figure 8. Schematic representations of (a) the Exclusion Zone region and (b) the Self Induced Flow through visualization of microsphere movement within a microchannel. (106) Reproduced with permission from ref (106). Copyright 2020 The Authors under the terms of the Creative Commons (CC BY 4.0) License https://creativecommons.org/licenses/by/4.0/.

Despite the finding of such a phenomenon, the specific mechanism and role of IR energy have yet to be defined for the process of EZ development. To further develop an understanding of the role of IR energy in such phenomena, a feasible study may be seen through the lens of the relationships between external forces and microfluidic flow. In the phenomena, the increase of SIF velocity under a rise of IR radiation resonant characteristics is shown in the participation of the external electric field near the microchannel walls under electro-osmotic viscoelastic fluid flow systems. The buildup of negative charges at the hydrophilic surfaces in EZ is analogous to the mechanism of electrical double layer formation. Indeed, research has initiated the exploration of the core mechanisms for EZ formation through the lens of the electrokinetic phenomena. 

(107) Such a similarity of the role of IR energy and the transport phenomena of SIF with electrokinetic phenomena paves the way for the definition of the unknown SIF phenomena and EZ formation. Furthermore, Li and Pollack 

(106) suggest whether CFL formation might contribute to a SIF of blood using solely IR radiation, a commonly available source of energy in nature, as an external driving force. The proposition may be proven feasible with the presence of the CFL region next to the negatively charged hydrophilic endothelial glycocalyx layer, coating the luminal side of blood vessels. 

(108) Further research can dive into the resonating characteristics between the formation of the CFL region next to the hydrophilic endothelial glycocalyx layer and that of the EZ formation close to hydrophilic microchannel walls. Indeed, an increase in IR energy is known to rapidly accelerate EZ formation and SIF velocity, depicting similarity to the increase in the magnitude of electric field forces and greater shear rates at microchannel walls affecting CFL formation and EOF velocity. Such correlation depicts a future direction in whether SIF blood flow can be observed and characterized theoretically further through the lens of the relationship between blood flow and shear forces exhibited by external energy.

The intricate link between the CFL and external forces, more specifically the externally applied electric field, can receive further attention to provide a more complete framework for the mechanisms between IR radiation and EZ formation. Such characterization may also contribute to a greater comprehension of the role IR can play in CFL formation next to the endothelial glycocalyx layer as well as its role as a driving force to propel blood flow, similar to the SIF, but without the commonly assumed pressure force from heart contraction as a source of driving force.

5.3. Challenges

Although there have been significant improvements in blood flow modeling under LOC systems over the past decade, there are still notable constraints that may require special attention for numerical simulation applications to benefit the adaptability of the designs and functionalities of LOC devices. Several points that require special attention are mentioned below:

1.The majority of CFD models operate under the relationship between the viscoelasticity of blood and the shear rate conditions of flow. The relative effect exhibited by the presence of highly populated RBCs in whole blood and their forces amongst the cells themselves under complex flows often remains unclearly defined. Furthermore, the full range of cell populations in whole blood requires a much more computational load for numerical modeling. Therefore, a vital goal for future research is to evaluate a reduced modeling method where the impact of cell–cell interaction on the viscoelastic property of blood is considered.
2.Current computational methods on hemodynamics rely on continuum models based upon non-Newtonian rheology at the macroscale rather than at molecular and cellular levels. Careful considerations should be made for the development of a constructive framework for the physical and temporal scales of micro/nanoscale systems to evaluate the intricate relationship between fluid driving forces, dynamic viscosity, and elasticity.
3.Viscoelastic fluids under the impact of externally applied electric forces often deviate from the assumptions of no-slip boundary conditions due to the unique flow conditions induced by externally applied forces. Furthermore, the mechanism of vortex formation and viscoelastic flow instability at laminar flow conditions should be better defined through the lens of the microfluidic flow phenomenon to optimize the prediction of viscoelastic flow across different geometrical layouts. Mathematical models and numerical methods are needed to better predict such disturbance caused by external forces and the viscoelasticity of fluids at such a small scale.
4.Under practical situations, zeta potential distribution at channel walls frequently deviates from the common assumption of a constant distribution because of manufacturing faults or inherent surface charges prior to the introduction of electrokinetic influence. These discrepancies frequently lead to inconsistent surface potential distribution, such as excess positive ions at relatively more negatively charged walls. Accordingly, unpredicted vortex formation and flow instability may occur. Therefore, careful consideration should be given to these discrepancies and how they could trigger the transport process and unexpected results of a microdevice.

Author Information

ARTICLE SECTIONS

Jump To


  • Corresponding Authors
    • Zhe Chen – Department of Chemical Engineering, School of Chemistry and Chemical Engineering, State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P. R. China;  Email: zaccooky@sjtu.edu.cn
    • Bo Ouyang – Department of Chemical Engineering, School of Chemistry and Chemical Engineering, State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P. R. China;  Email: bouy93@sjtu.edu.cn
    • Zheng-Hong Luo – Department of Chemical Engineering, School of Chemistry and Chemical Engineering, State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P. R. China;  Orcidhttps://orcid.org/0000-0001-9011-6020; Email: luozh@sjtu.edu.cn
  • Authors
    • Bin-Jie Lai – Department of Chemical Engineering, School of Chemistry and Chemical Engineering, State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P. R. China;  Orcidhttps://orcid.org/0009-0002-8133-5381
    • Li-Tao Zhu – Department of Chemical Engineering, School of Chemistry and Chemical Engineering, State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P. R. China;  Orcidhttps://orcid.org/0000-0001-6514-8864
  • NotesThe authors declare no competing financial interest.

Acknowledgments

ARTICLE SECTIONS

Jump To


This work was supported by the National Natural Science Foundation of China (No. 22238005) and the Postdoctoral Research Foundation of China (No. GZC20231576).

Vocabulary

ARTICLE SECTIONS

Jump To


Microfluidicsthe field of technological and scientific study that investigates fluid flow in channels with dimensions between 1 and 1000 μm
Lab-on-a-Chip Technologythe field of research and technological development aimed at integrating the micro/nanofluidic characteristics to conduct laboratory processes on handheld devices
Computational Fluid Dynamics (CFD)the method utilizing computational abilities to predict physical fluid flow behaviors mathematically through solving the governing equations of corresponding fluid flows
Shear Ratethe rate of change in velocity where one layer of fluid moves past the adjacent layer
Viscoelasticitythe property holding both elasticity and viscosity characteristics relying on the magnitude of applied shear stress and time-dependent strain
Electro-osmosisthe flow of fluid under an applied electric field when charged solid surface is in contact with the bulk fluid
Vortexthe rotating motion of a fluid revolving an axis line

References

ARTICLE SECTIONS

Jump To


This article references 108 other publications.

  1. 1Neethirajan, S.; Kobayashi, I.; Nakajima, M.; Wu, D.; Nandagopal, S.; Lin, F. Microfluidics for food, agriculture and biosystems industries. Lab Chip 201111 (9), 1574– 1586,  DOI: 10.1039/c0lc00230eViewGoogle Scholar
  2. 2Whitesides, G. M. The origins and the future of microfluidics. Nature 2006442 (7101), 368– 373,  DOI: 10.1038/nature05058ViewGoogle Scholar
  3. 3Burklund, A.; Tadimety, A.; Nie, Y.; Hao, N.; Zhang, J. X. J. Chapter One – Advances in diagnostic microfluidics; Elsevier, 2020; DOI:  DOI: 10.1016/bs.acc.2019.08.001 .ViewGoogle Scholar
  4. 4Abdulbari, H. A. Chapter 12 – Lab-on-a-chip for analysis of blood. In Nanotechnology for Hematology, Blood Transfusion, and Artificial Blood; Denizli, A., Nguyen, T. A., Rajan, M., Alam, M. F., Rahman, K., Eds.; Elsevier, 2022; pp 265– 283.ViewGoogle Scholar
  5. 5Vladisavljević, G. T.; Khalid, N.; Neves, M. A.; Kuroiwa, T.; Nakajima, M.; Uemura, K.; Ichikawa, S.; Kobayashi, I. Industrial lab-on-a-chip: Design, applications and scale-up for drug discovery and delivery. Advanced Drug Delivery Reviews 201365 (11), 1626– 1663,  DOI: 10.1016/j.addr.2013.07.017ViewGoogle Scholar
  6. 6Kersaudy-Kerhoas, M.; Dhariwal, R.; Desmulliez, M. P. Y.; Jouvet, L. Hydrodynamic blood plasma separation in microfluidic channels. Microfluid. Nanofluid. 20108 (1), 105– 114,  DOI: 10.1007/s10404-009-0450-5ViewGoogle Scholar
  7. 7Popel, A. S.; Johnson, P. C. Microcirculation and Hemorheology. Annu. Rev. Fluid Mech. 200537 (1), 43– 69,  DOI: 10.1146/annurev.fluid.37.042604.133933ViewGoogle Scholar
  8. 8Fedosov, D. A.; Peltomäki, M.; Gompper, G. Deformation and dynamics of red blood cells in flow through cylindrical microchannels. Soft Matter 201410 (24), 4258– 4267,  DOI: 10.1039/C4SM00248BViewGoogle Scholar
  9. 9Chakraborty, S. Dynamics of capillary flow of blood into a microfluidic channel. Lab Chip 20055 (4), 421– 430,  DOI: 10.1039/b414566fViewGoogle Scholar
  10. 10Tomaiuolo, G.; Guido, S. Start-up shape dynamics of red blood cells in microcapillary flow. Microvascular Research 201182 (1), 35– 41,  DOI: 10.1016/j.mvr.2011.03.004ViewGoogle Scholar
  11. 11Sherwood, J. M.; Dusting, J.; Kaliviotis, E.; Balabani, S. The effect of red blood cell aggregation on velocity and cell-depleted layer characteristics of blood in a bifurcating microchannel. Biomicrofluidics 20126 (2), 24119,  DOI: 10.1063/1.4717755ViewGoogle Scholar
  12. 12Nader, E.; Skinner, S.; Romana, M.; Fort, R.; Lemonne, N.; Guillot, N.; Gauthier, A.; Antoine-Jonville, S.; Renoux, C.; Hardy-Dessources, M.-D. Blood Rheology: Key Parameters, Impact on Blood Flow, Role in Sickle Cell Disease and Effects of Exercise. Frontiers in Physiology 201910, 01329,  DOI: 10.3389/fphys.2019.01329ViewGoogle Scholar
  13. 13Trejo-Soto, C.; Lázaro, G. R.; Pagonabarraga, I.; Hernández-Machado, A. Microfluidics Approach to the Mechanical Properties of Red Blood Cell Membrane and Their Effect on Blood Rheology. Membranes 202212 (2), 217,  DOI: 10.3390/membranes12020217ViewGoogle Scholar
  14. 14Wagner, C.; Steffen, P.; Svetina, S. Aggregation of red blood cells: From rouleaux to clot formation. Comptes Rendus Physique 201314 (6), 459– 469,  DOI: 10.1016/j.crhy.2013.04.004ViewGoogle Scholar
  15. 15Kim, H.; Zhbanov, A.; Yang, S. Microfluidic Systems for Blood and Blood Cell Characterization. Biosensors 202313 (1), 13,  DOI: 10.3390/bios13010013ViewGoogle Scholar
  16. 16Fåhræus, R.; Lindqvist, T. THE VISCOSITY OF THE BLOOD IN NARROW CAPILLARY TUBES. American Journal of Physiology-Legacy Content 193196 (3), 562– 568,  DOI: 10.1152/ajplegacy.1931.96.3.562ViewGoogle Scholar
  17. 17Ascolese, M.; Farina, A.; Fasano, A. The Fåhræus-Lindqvist effect in small blood vessels: how does it help the heart?. J. Biol. Phys. 201945 (4), 379– 394,  DOI: 10.1007/s10867-019-09534-4ViewGoogle Scholar
  18. 18Bento, D.; Fernandes, C. S.; Miranda, J. M.; Lima, R. In vitro blood flow visualizations and cell-free layer (CFL) measurements in a microchannel network. Experimental Thermal and Fluid Science 2019109, 109847,  DOI: 10.1016/j.expthermflusci.2019.109847ViewGoogle Scholar
  19. 19Namgung, B.; Ong, P. K.; Wong, Y. H.; Lim, D.; Chun, K. J.; Kim, S. A comparative study of histogram-based thresholding methods for the determination of cell-free layer width in small blood vessels. Physiological Measurement 201031 (9), N61,  DOI: 10.1088/0967-3334/31/9/N01ViewGoogle Scholar
  20. 20Hymel, S. J.; Lan, H.; Fujioka, H.; Khismatullin, D. B. Cell trapping in Y-junction microchannels: A numerical study of the bifurcation angle effect in inertial microfluidics. Phys. Fluids (1994) 201931 (8), 082003,  DOI: 10.1063/1.5113516ViewGoogle Scholar
  21. 21Li, X.; Popel, A. S.; Karniadakis, G. E. Blood-plasma separation in Y-shaped bifurcating microfluidic channels: a dissipative particle dynamics simulation study. Phys. Biol. 20129 (2), 026010,  DOI: 10.1088/1478-3975/9/2/026010ViewGoogle Scholar
  22. 22Yin, X.; Thomas, T.; Zhang, J. Multiple red blood cell flows through microvascular bifurcations: Cell free layer, cell trajectory, and hematocrit separation. Microvascular Research 201389, 47– 56,  DOI: 10.1016/j.mvr.2013.05.002ViewGoogle Scholar
  23. 23Shibeshi, S. S.; Collins, W. E. The Rheology of Blood Flow in a Branched Arterial System. Appl. Rheol 200515 (6), 398– 405,  DOI: 10.1515/arh-2005-0020ViewGoogle Scholar
  24. 24Sequeira, A.; Janela, J. An Overview of Some Mathematical Models of Blood Rheology. In A Portrait of State-of-the-Art Research at the Technical University of Lisbon; Pereira, M. S., Ed.; Springer Netherlands: Dordrecht, 2007; pp 65– 87.ViewGoogle Scholar
  25. 25Walburn, F. J.; Schneck, D. J. A constitutive equation for whole human blood. Biorheology 197613, 201– 210,  DOI: 10.3233/BIR-1976-13307ViewGoogle Scholar
  26. 26Quemada, D. A rheological model for studying the hematocrit dependence of red cell-red cell and red cell-protein interactions in blood. Biorheology 198118, 501– 516,  DOI: 10.3233/BIR-1981-183-615ViewGoogle Scholar
  27. 27Varchanis, S.; Dimakopoulos, Y.; Wagner, C.; Tsamopoulos, J. How viscoelastic is human blood plasma?. Soft Matter 201814 (21), 4238– 4251,  DOI: 10.1039/C8SM00061AViewGoogle Scholar
  28. 28Apostolidis, A. J.; Moyer, A. P.; Beris, A. N. Non-Newtonian effects in simulations of coronary arterial blood flow. J. Non-Newtonian Fluid Mech. 2016233, 155– 165,  DOI: 10.1016/j.jnnfm.2016.03.008ViewGoogle Scholar
  29. 29Luo, X. Y.; Kuang, Z. B. A study on the constitutive equation of blood. J. Biomech. 199225 (8), 929– 934,  DOI: 10.1016/0021-9290(92)90233-QViewGoogle Scholar
  30. 30Oldroyd, J. G.; Wilson, A. H. On the formulation of rheological equations of state. Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences 1950200 (1063), 523– 541,  DOI: 10.1098/rspa.1950.0035ViewGoogle Scholar
  31. 31Prado, G.; Farutin, A.; Misbah, C.; Bureau, L. Viscoelastic transient of confined red blood cells. Biophys J. 2015108 (9), 2126– 2136,  DOI: 10.1016/j.bpj.2015.03.046ViewGoogle Scholar
  32. 32Huang, C. R.; Pan, W. D.; Chen, H. Q.; Copley, A. L. Thixotropic properties of whole blood from healthy human subjects. Biorheology 198724 (6), 795– 801,  DOI: 10.3233/BIR-1987-24630ViewGoogle Scholar
  33. 33Anand, M.; Kwack, J.; Masud, A. A new generalized Oldroyd-B model for blood flow in complex geometries. International Journal of Engineering Science 201372, 78– 88,  DOI: 10.1016/j.ijengsci.2013.06.009ViewGoogle Scholar
  34. 34Horner, J. S.; Armstrong, M. J.; Wagner, N. J.; Beris, A. N. Investigation of blood rheology under steady and unidirectional large amplitude oscillatory shear. J. Rheol. 201862 (2), 577– 591,  DOI: 10.1122/1.5017623ViewGoogle Scholar
  35. 35Horner, J. S.; Armstrong, M. J.; Wagner, N. J.; Beris, A. N. Measurements of human blood viscoelasticity and thixotropy under steady and transient shear and constitutive modeling thereof. J. Rheol. 201963 (5), 799– 813,  DOI: 10.1122/1.5108737ViewGoogle Scholar
  36. 36Armstrong, M.; Tussing, J. A methodology for adding thixotropy to Oldroyd-8 family of viscoelastic models for characterization of human blood. Phys. Fluids 202032 (9), 094111,  DOI: 10.1063/5.0022501ViewGoogle Scholar
  37. 37Crank, J.; Nicolson, P. A practical method for numerical evaluation of solutions of partial differential equations of the heat-conduction type. Mathematical Proceedings of the Cambridge Philosophical Society 194743 (1), 50– 67,  DOI: 10.1017/S0305004100023197ViewGoogle Scholar
  38. 38Clough, R. W. Original formulation of the finite element method. Finite Elements in Analysis and Design 19907 (2), 89– 101,  DOI: 10.1016/0168-874X(90)90001-UViewGoogle Scholar
  39. 39Liu, W. K.; Liu, Y.; Farrell, D.; Zhang, L.; Wang, X. S.; Fukui, Y.; Patankar, N.; Zhang, Y.; Bajaj, C.; Lee, J.Immersed finite element method and its applications to biological systems. Computer Methods in Applied Mechanics and Engineering 2006195 (13), 1722– 1749,  DOI: 10.1016/j.cma.2005.05.049ViewGoogle Scholar
  40. 40Lopes, D.; Agujetas, R.; Puga, H.; Teixeira, J.; Lima, R.; Alejo, J. P.; Ferrera, C. Analysis of finite element and finite volume methods for fluid-structure interaction simulation of blood flow in a real stenosed artery. International Journal of Mechanical Sciences 2021207, 106650,  DOI: 10.1016/j.ijmecsci.2021.106650ViewGoogle Scholar
  41. 41Favero, J. L.; Secchi, A. R.; Cardozo, N. S. M.; Jasak, H. Viscoelastic flow analysis using the software OpenFOAM and differential constitutive equations. J. Non-Newtonian Fluid Mech. 2010165 (23), 1625– 1636,  DOI: 10.1016/j.jnnfm.2010.08.010ViewGoogle Scholar
  42. 42Pimenta, F.; Alves, M. A. Stabilization of an open-source finite-volume solver for viscoelastic fluid flows. J. Non-Newtonian Fluid Mech. 2017239, 85– 104,  DOI: 10.1016/j.jnnfm.2016.12.002ViewGoogle Scholar
  43. 43Chee, C. Y.; Lee, H. P.; Lu, C. Using 3D fluid-structure interaction model to analyse the biomechanical properties of erythrocyte. Phys. Lett. A 2008372 (9), 1357– 1362,  DOI: 10.1016/j.physleta.2007.09.067ViewGoogle Scholar
  44. 44Xu, D.; Kaliviotis, E.; Munjiza, A.; Avital, E.; Ji, C.; Williams, J. Large scale simulation of red blood cell aggregation in shear flows. J. Biomech. 201346 (11), 1810– 1817,  DOI: 10.1016/j.jbiomech.2013.05.010ViewGoogle Scholar
  45. 45Johnson, K. L.; Kendall, K.; Roberts, A. Surface energy and the contact of elastic solids. Proceedings of the royal society of London. A. mathematical and physical sciences 1971324 (1558), 301– 313,  DOI: 10.1098/rspa.1971.0141ViewGoogle Scholar
  46. 46Shi, L.; Pan, T.-W.; Glowinski, R. Deformation of a single red blood cell in bounded Poiseuille flows. Phys. Rev. E 201285 (1), 016307,  DOI: 10.1103/PhysRevE.85.016307ViewGoogle Scholar
  47. 47Yoon, D.; You, D. Continuum modeling of deformation and aggregation of red blood cells. J. Biomech. 201649 (11), 2267– 2279,  DOI: 10.1016/j.jbiomech.2015.11.027ViewGoogle Scholar
  48. 48Mainardi, F.; Spada, G. Creep, relaxation and viscosity properties for basic fractional models in rheology. European Physical Journal Special Topics 2011193 (1), 133– 160,  DOI: 10.1140/epjst/e2011-01387-1ViewGoogle Scholar
  49. 49Gracka, M.; Lima, R.; Miranda, J. M.; Student, S.; Melka, B.; Ostrowski, Z. Red blood cells tracking and cell-free layer formation in a microchannel with hyperbolic contraction: A CFD model validation. Computer Methods and Programs in Biomedicine 2022226, 107117,  DOI: 10.1016/j.cmpb.2022.107117ViewGoogle Scholar
  50. 50Aryan, H.; Beigzadeh, B.; Siavashi, M. Euler-Lagrange numerical simulation of improved magnetic drug delivery in a three-dimensional CT-based carotid artery bifurcation. Computer Methods and Programs in Biomedicine 2022219, 106778,  DOI: 10.1016/j.cmpb.2022.106778ViewGoogle Scholar
  51. 51Czaja, B.; Závodszky, G.; Azizi Tarksalooyeh, V.; Hoekstra, A. G. Cell-resolved blood flow simulations of saccular aneurysms: effects of pulsatility and aspect ratio. J. R Soc. Interface 201815 (146), 20180485,  DOI: 10.1098/rsif.2018.0485ViewGoogle Scholar
  52. 52Rydquist, G.; Esmaily, M. A cell-resolved, Lagrangian solver for modeling red blood cell dynamics in macroscale flows. J. Comput. Phys. 2022461, 111204,  DOI: 10.1016/j.jcp.2022.111204ViewGoogle Scholar
  53. 53Dadvand, A.; Baghalnezhad, M.; Mirzaee, I.; Khoo, B. C.; Ghoreishi, S. An immersed boundary-lattice Boltzmann approach to study the dynamics of elastic membranes in viscous shear flows. Journal of Computational Science 20145 (5), 709– 718,  DOI: 10.1016/j.jocs.2014.06.006ViewGoogle Scholar
  54. 54Krüger, T.; Holmes, D.; Coveney, P. V. Deformability-based red blood cell separation in deterministic lateral displacement devices─A simulation study. Biomicrofluidics 20148 (5), 054114,  DOI: 10.1063/1.4897913ViewGoogle Scholar
  55. 55Takeishi, N.; Ito, H.; Kaneko, M.; Wada, S. Deformation of a Red Blood Cell in a Narrow Rectangular Microchannel. Micromachines 201910 (3), 199,  DOI: 10.3390/mi10030199ViewGoogle Scholar
  56. 56Krüger, T.; Varnik, F.; Raabe, D. Efficient and accurate simulations of deformable particles immersed in a fluid using a combined immersed boundary lattice Boltzmann finite element method. Computers & Mathematics with Applications 201161 (12), 3485– 3505,  DOI: 10.1016/j.camwa.2010.03.057ViewGoogle Scholar
  57. 57Balachandran Nair, A. N.; Pirker, S.; Umundum, T.; Saeedipour, M. A reduced-order model for deformable particles with application in bio-microfluidics. Computational Particle Mechanics 20207 (3), 593– 601,  DOI: 10.1007/s40571-019-00283-8ViewGoogle Scholar
  58. 58Balachandran Nair, A. N.; Pirker, S.; Saeedipour, M. Resolved CFD-DEM simulation of blood flow with a reduced-order RBC model. Computational Particle Mechanics 20229 (4), 759– 774,  DOI: 10.1007/s40571-021-00441-xViewGoogle Scholar
  59. 59Mittal, R.; Iaccarino, G. IMMERSED BOUNDARY METHODS. Annu. Rev. Fluid Mech. 200537 (1), 239– 261,  DOI: 10.1146/annurev.fluid.37.061903.175743ViewGoogle Scholar
  60. 60Piquet, A.; Roussel, O.; Hadjadj, A. A comparative study of Brinkman penalization and direct-forcing immersed boundary methods for compressible viscous flows. Computers & Fluids 2016136, 272– 284,  DOI: 10.1016/j.compfluid.2016.06.001ViewGoogle Scholar
  61. 61Akerkouch, L.; Le, T. B. A Hybrid Continuum-Particle Approach for Fluid-Structure Interaction Simulation of Red Blood Cells in Fluid Flows. Fluids 20216 (4), 139,  DOI: 10.3390/fluids6040139ViewGoogle Scholar
  62. 62Barker, A. T.; Cai, X.-C. Scalable parallel methods for monolithic coupling in fluid-structure interaction with application to blood flow modeling. J. Comput. Phys. 2010229 (3), 642– 659,  DOI: 10.1016/j.jcp.2009.10.001ViewGoogle Scholar
  63. 63Cetin, A.; Sahin, M. A monolithic fluid-structure interaction framework applied to red blood cells. International Journal for Numerical Methods in Biomedical Engineering 201935 (2), e3171  DOI: 10.1002/cnm.3171ViewGoogle Scholar
  64. 64Freund, J. B. Numerical Simulation of Flowing Blood Cells. Annu. Rev. Fluid Mech. 201446 (1), 67– 95,  DOI: 10.1146/annurev-fluid-010313-141349ViewGoogle Scholar
  65. 65Ye, T.; Phan-Thien, N.; Lim, C. T. Particle-based simulations of red blood cells─A review. J. Biomech. 201649 (11), 2255– 2266,  DOI: 10.1016/j.jbiomech.2015.11.050ViewGoogle Scholar
  66. 66Arabghahestani, M.; Poozesh, S.; Akafuah, N. K. Advances in Computational Fluid Mechanics in Cellular Flow Manipulation: A Review. Applied Sciences 20199 (19), 4041,  DOI: 10.3390/app9194041ViewGoogle Scholar
  67. 67Rathnayaka, C. M.; From, C. S.; Geekiyanage, N. M.; Gu, Y. T.; Nguyen, N. T.; Sauret, E. Particle-Based Numerical Modelling of Liquid Marbles: Recent Advances and Future Perspectives. Archives of Computational Methods in Engineering 202229 (5), 3021– 3039,  DOI: 10.1007/s11831-021-09683-7ViewGoogle Scholar
  68. 68Li, X.; Vlahovska, P. M.; Karniadakis, G. E. Continuum- and particle-based modeling of shapes and dynamics of red blood cells in health and disease. Soft Matter 20139 (1), 28– 37,  DOI: 10.1039/C2SM26891DViewGoogle Scholar
  69. 69Beris, A. N.; Horner, J. S.; Jariwala, S.; Armstrong, M. J.; Wagner, N. J. Recent advances in blood rheology: a review. Soft Matter 202117 (47), 10591– 10613,  DOI: 10.1039/D1SM01212FViewGoogle Scholar
  70. 70Arciero, J.; Causin, P.; Malgaroli, F. Mathematical methods for modeling the microcirculation. AIMS Biophysics 20174 (3), 362– 399,  DOI: 10.3934/biophy.2017.3.362ViewGoogle Scholar
  71. 71Maria, M. S.; Chandra, T. S.; Sen, A. K. Capillary flow-driven blood plasma separation and on-chip analyte detection in microfluidic devices. Microfluid. Nanofluid. 201721 (4), 72,  DOI: 10.1007/s10404-017-1907-6ViewGoogle Scholar
  72. 72Huhtamäki, T.; Tian, X.; Korhonen, J. T.; Ras, R. H. A. Surface-wetting characterization using contact-angle measurements. Nat. Protoc. 201813 (7), 1521– 1538,  DOI: 10.1038/s41596-018-0003-zViewGoogle Scholar
  73. 73Young, T., III. An essay on the cohesion of fluids. Philosophical Transactions of the Royal Society of London 180595, 65– 87,  DOI: 10.1098/rstl.1805.0005ViewGoogle Scholar
  74. 74Kim, Y. C.; Kim, S.-H.; Kim, D.; Park, S.-J.; Park, J.-K. Plasma extraction in a capillary-driven microfluidic device using surfactant-added poly(dimethylsiloxane). Sens. Actuators, B 2010145 (2), 861– 868,  DOI: 10.1016/j.snb.2010.01.017ViewGoogle Scholar
  75. 75Washburn, E. W. The Dynamics of Capillary Flow. Physical Review 192117 (3), 273– 283,  DOI: 10.1103/PhysRev.17.273ViewGoogle Scholar
  76. 76Cito, S.; Ahn, Y. C.; Pallares, J.; Duarte, R. M.; Chen, Z.; Madou, M.; Katakis, I. Visualization and measurement of capillary-driven blood flow using spectral domain optical coherence tomography. Microfluid Nanofluidics 201213 (2), 227– 237,  DOI: 10.1007/s10404-012-0950-6ViewGoogle Scholar
  77. 77Berthier, E.; Dostie, A. M.; Lee, U. N.; Berthier, J.; Theberge, A. B. Open Microfluidic Capillary Systems. Anal Chem. 201991 (14), 8739– 8750,  DOI: 10.1021/acs.analchem.9b01429ViewGoogle Scholar
  78. 78Berthier, J.; Brakke, K. A.; Furlani, E. P.; Karampelas, I. H.; Poher, V.; Gosselin, D.; Cubizolles, M.; Pouteau, P. Whole blood spontaneous capillary flow in narrow V-groove microchannels. Sens. Actuators, B 2015206, 258– 267,  DOI: 10.1016/j.snb.2014.09.040ViewGoogle Scholar
  79. 79Hirt, C. W.; Nichols, B. D. Volume of fluid (VOF) method for the dynamics of free boundaries. J. Comput. Phys. 198139 (1), 201– 225,  DOI: 10.1016/0021-9991(81)90145-5ViewGoogle Scholar
  80. 80Chen, J.-L.; Shih, W.-H.; Hsieh, W.-H. AC electro-osmotic micromixer using a face-to-face, asymmetric pair of planar electrodes. Sens. Actuators, B 2013188, 11– 21,  DOI: 10.1016/j.snb.2013.07.012ViewGoogle Scholar
  81. 81Zhao, C.; Yang, C. Electrokinetics of non-Newtonian fluids: A review. Advances in Colloid and Interface Science 2013201-202, 94– 108,  DOI: 10.1016/j.cis.2013.09.001ViewGoogle Scholar
  82. 82Oh, K. W. 6 – Lab-on-chip (LOC) devices and microfluidics for biomedical applications. In MEMS for Biomedical Applications; Bhansali, S., Vasudev, A., Eds.; Woodhead Publishing, 2012; pp 150– 171.ViewGoogle Scholar
  83. 83Bello, M. S.; De Besi, P.; Rezzonico, R.; Righetti, P. G.; Casiraghi, E. Electroosmosis of polymer solutions in fused silica capillaries. ELECTROPHORESIS 199415 (1), 623– 626,  DOI: 10.1002/elps.1150150186ViewGoogle Scholar
  84. 84Park, H. M.; Lee, W. M. Effect of viscoelasticity on the flow pattern and the volumetric flow rate in electroosmotic flows through a microchannel. Lab Chip 20088 (7), 1163– 1170,  DOI: 10.1039/b800185eViewGoogle Scholar
  85. 85Afonso, A. M.; Alves, M. A.; Pinho, F. T. Analytical solution of mixed electro-osmotic/pressure driven flows of viscoelastic fluids in microchannels. J. Non-Newtonian Fluid Mech. 2009159 (1), 50– 63,  DOI: 10.1016/j.jnnfm.2009.01.006ViewGoogle Scholar
  86. 86Sousa, J. J.; Afonso, A. M.; Pinho, F. T.; Alves, M. A. Effect of the skimming layer on electro-osmotic─Poiseuille flows of viscoelastic fluids. Microfluid. Nanofluid. 201110 (1), 107– 122,  DOI: 10.1007/s10404-010-0651-yViewGoogle Scholar
  87. 87Zhao, C.; Yang, C. Electro-osmotic mobility of non-Newtonian fluids. Biomicrofluidics 20115 (1), 014110,  DOI: 10.1063/1.3571278ViewGoogle Scholar
  88. 88Pimenta, F.; Alves, M. A. Electro-elastic instabilities in cross-shaped microchannels. J. Non-Newtonian Fluid Mech. 2018259, 61– 77,  DOI: 10.1016/j.jnnfm.2018.04.004ViewGoogle Scholar
  89. 89Bezerra, W. S.; Castelo, A.; Afonso, A. M. Numerical Study of Electro-Osmotic Fluid Flow and Vortex Formation. Micromachines (Basel) 201910 (12), 796,  DOI: 10.3390/mi10120796ViewGoogle Scholar
  90. 90Ji, J.; Qian, S.; Liu, Z. Electroosmotic Flow of Viscoelastic Fluid through a Constriction Microchannel. Micromachines (Basel) 202112 (4), 417,  DOI: 10.3390/mi12040417ViewGoogle Scholar
  91. 91Zhao, C.; Yang, C. Exact solutions for electro-osmotic flow of viscoelastic fluids in rectangular micro-channels. Applied Mathematics and Computation 2009211 (2), 502– 509,  DOI: 10.1016/j.amc.2009.01.068ViewGoogle Scholar
  92. 92Gerum, R.; Mirzahossein, E.; Eroles, M.; Elsterer, J.; Mainka, A.; Bauer, A.; Sonntag, S.; Winterl, A.; Bartl, J.; Fischer, L. Viscoelastic properties of suspended cells measured with shear flow deformation cytometry. Elife 202211, e78823,  DOI: 10.7554/eLife.78823ViewGoogle Scholar
  93. 93Sadek, S. H.; Pinho, F. T.; Alves, M. A. Electro-elastic flow instabilities of viscoelastic fluids in contraction/expansion micro-geometries. J. Non-Newtonian Fluid Mech. 2020283, 104293,  DOI: 10.1016/j.jnnfm.2020.104293ViewGoogle Scholar
  94. 94Spanjaards, M.; Peters, G.; Hulsen, M.; Anderson, P. Numerical Study of the Effect of Thixotropy on Extrudate Swell. Polymers 202113 (24), 4383,  DOI: 10.3390/polym13244383ViewGoogle Scholar
  95. 95Rashidi, S.; Bafekr, H.; Valipour, M. S.; Esfahani, J. A. A review on the application, simulation, and experiment of the electrokinetic mixers. Chemical Engineering and Processing – Process Intensification 2018126, 108– 122,  DOI: 10.1016/j.cep.2018.02.021ViewGoogle Scholar
  96. 96Matsubara, K.; Narumi, T. Microfluidic mixing using unsteady electroosmotic vortices produced by a staggered array of electrodes. Chemical Engineering Journal 2016288, 638– 647,  DOI: 10.1016/j.cej.2015.12.013ViewGoogle Scholar
  97. 97Qaderi, A.; Jamaati, J.; Bahiraei, M. CFD simulation of combined electroosmotic-pressure driven micro-mixing in a microchannel equipped with triangular hurdle and zeta-potential heterogeneity. Chemical Engineering Science 2019199, 463– 477,  DOI: 10.1016/j.ces.2019.01.034ViewGoogle Scholar
  98. 98Cho, C.-C.; Chen, C.-L.; Chen, C. o.-K. Mixing enhancement in crisscross micromixer using aperiodic electrokinetic perturbing flows. International Journal of Heat and Mass Transfer 201255 (11), 2926– 2933,  DOI: 10.1016/j.ijheatmasstransfer.2012.02.006ViewGoogle Scholar
  99. 99Zhao, W.; Yang, F.; Wang, K.; Bai, J.; Wang, G. Rapid mixing by turbulent-like electrokinetic microflow. Chemical Engineering Science 2017165, 113– 121,  DOI: 10.1016/j.ces.2017.02.027ViewGoogle Scholar
  100. 100Tran, T.; Chakraborty, P.; Guttenberg, N.; Prescott, A.; Kellay, H.; Goldburg, W.; Goldenfeld, N.; Gioia, G. Macroscopic effects of the spectral structure in turbulent flows. Nat. Phys. 20106 (6), 438– 441,  DOI: 10.1038/nphys1674ViewGoogle Scholar
  101. 101Toner, M.; Irimia, D. Blood-on-a-chip. Annu. Rev. Biomed Eng. 20057, 77– 103,  DOI: 10.1146/annurev.bioeng.7.011205.135108ViewGoogle Scholar
  102. 102Maria, M. S.; Rakesh, P. E.; Chandra, T. S.; Sen, A. K. Capillary flow of blood in a microchannel with differential wetting for blood plasma separation and on-chip glucose detection. Biomicrofluidics 201610 (5), 054108,  DOI: 10.1063/1.4962874ViewGoogle Scholar
  103. 103Tripathi, S.; Varun Kumar, Y. V. B.; Prabhakar, A.; Joshi, S. S.; Agrawal, A. Passive blood plasma separation at the microscale: a review of design principles and microdevices. Journal of Micromechanics and Microengineering 201525 (8), 083001,  DOI: 10.1088/0960-1317/25/8/083001ViewGoogle Scholar
  104. 104Mohammadi, M.; Madadi, H.; Casals-Terré, J. Microfluidic point-of-care blood panel based on a novel technique: Reversible electroosmotic flow. Biomicrofluidics 20159 (5), 054106,  DOI: 10.1063/1.4930865ViewGoogle Scholar
  105. 105Kang, D. H.; Kim, K.; Kim, Y. J. An anti-clogging method for improving the performance and lifespan of blood plasma separation devices in real-time and continuous microfluidic systems. Sci. Rep 20188 (1), 17015,  DOI: 10.1038/s41598-018-35235-4ViewGoogle Scholar
  106. 106Li, Z.; Pollack, G. H. Surface-induced flow: A natural microscopic engine using infrared energy as fuel. Science Advances 20206 (19), eaba0941  DOI: 10.1126/sciadv.aba0941ViewGoogle Scholar
  107. 107Mercado-Uribe, H.; Guevara-Pantoja, F. J.; García-Muñoz, W.; García-Maldonado, J. S.; Méndez-Alcaraz, J. M.; Ruiz-Suárez, J. C. On the evolution of the exclusion zone produced by hydrophilic surfaces: A contracted description. J. Chem. Phys. 2021154 (19), 194902,  DOI: 10.1063/5.0043084ViewGoogle Scholar
  108. 108Yalcin, O.; Jani, V. P.; Johnson, P. C.; Cabrales, P. Implications Enzymatic Degradation of the Endothelial Glycocalyx on the Microvascular Hemodynamics and the Arteriolar Red Cell Free Layer of the Rat Cremaster Muscle. Front Physiol 20189, 168,  DOI: 10.3389/fphys.2018.00168ViewGoogle Scholar
비선형 파력의 영향에 따른 잔해 언덕 방파제 형상의 효과에 대한 수치 분석

비선형 파력의 영향에 따른 잔해 언덕 방파제 형상의 효과에 대한 수치 분석

Numerical Analysis of the Effects of Rubble Mound Breakwater Geometry Under the Effect of Nonlinear Wave Force

Arabian Journal for Science and EngineeringAims and scopeSubmit manuscript

Cite this article

Abstract

Assessing the interaction of waves and porous offshore structures such as rubble mound breakwaters plays a critical role in designing such structures optimally. This study focused on the effect of the geometric parameters of a sloped rubble mound breakwater, including the shape of the armour, method of its arrangement, and the breakwater slope. Thus, three main design criteria, including the wave reflection coefficient (Kr), transmission coefficient (Kt), and depreciation wave energy coefficient (Kd), are discussed. Based on the results, a decrease in wavelength reduced the Kr and increased the Kt and Kd. The rubble mound breakwater with the Coreloc armour layer could exhibit the lowest Kr compared to other armour geometries. In addition, a decrease in the breakwater slope reduced the Kr and Kd by 3.4 and 1.25%, respectively. In addition, a decrease in the breakwater slope from 33 to 25° increased the wave breaking height by 6.1% on average. Further, a decrease in the breakwater slope reduced the intensity of turbulence depreciation. Finally, the armour geometry and arrangement of armour layers on the breakwater with its different slopes affect the wave behaviour and interaction between the wave and breakwater. Thus, layering on the breakwater and the correct use of the geometric shapes of the armour should be considered when designing such structures.

파도와 잔해 더미 방파제와 같은 다공성 해양 구조물의 상호 작용을 평가하는 것은 이러한 구조물을 최적으로 설계하는 데 중요한 역할을 합니다. 본 연구는 경사진 잔해 둔덕 방파제의 기하학적 매개변수의 효과에 초점을 맞추었는데, 여기에는 갑옷의 형태, 배치 방법, 방파제 경사 등이 포함된다. 따라서 파동 반사 계수(Kr), 투과 계수(Kt) 및 감가상각파 에너지 계수(Kd)에 대해 논의합니다. 결과에 따르면 파장이 감소하면 K가 감소합니다.r그리고 K를 증가시켰습니다t 및 Kd. Coreloc 장갑 층이 있는 잔해 언덕 방파제는 가장 낮은 K를 나타낼 수 있습니다.r 다른 갑옷 형상과 비교했습니다. 또한 방파제 경사가 감소하여 K가 감소했습니다.r 및 Kd 각각 3.4%, 1.25% 증가했다. 또한 방파제 경사가 33°에서 25°로 감소하여 파도 파쇄 높이가 평균 6.1% 증가했습니다. 또한, 방파제 경사의 감소는 난류 감가상각의 강도를 감소시켰다. 마지막으로, 경사가 다른 방파제의 장갑 형상과 장갑 층의 배열은 파도 거동과 파도와 방파제 사이의 상호 작용에 영향을 미칩니다. 따라서 이러한 구조를 설계 할 때 방파제에 층을 쌓고 갑옷의 기하학적 모양을 올바르게 사용하는 것을 고려해야합니다.

Keywords

  • Rubble mound breakwater
  • Computational fluid dynamics
  • Armour layer
  • Wave reflection coefficient
  • Wave transmission coefficient
  • Wave energy dissipation coefficient

References

  1. Sollitt, C.K.; Cross, R.H.: Wave transmission through permeable breakwaters. In Coastal Engineering. pp. 1827–1846. (1973)
  2. Sulisz, W.: Wave reflection and transmission at permeable breakwaters of arbitrary cross-section. Coast. Eng. 9(4), 371–386 (1985)Article  Google Scholar 
  3. Kobayashi, N.; Wurjanto, A.: Numerical model for waves on rough permeable slopes. J. Coast. Res.149–166. (1990)
  4. Wurjanto, A.; Kobayashi, N.: Irregular wave reflection and runup on permeable slopes. J. Waterw. Port Coast. Ocean Eng. 119(5), 537–557 (1993)Article  Google Scholar 
  5. van Gent, M.R.: Numerical modelling of wave interaction with dynamically stable structures. In Coastal Engineering 1996. pp. 1930–1943. (1997)
  6. Liu, P.L.F.; Wen, J.: Nonlinear diffusive surface waves in porous media. J. Fluid Mech. 347, 119–139 (1997)Article  MathSciNet  MATH  Google Scholar 
  7. Troch, P.; De Rouck, J.: Development of two-dimensional numerical wave flume for wave interaction with rubble mound breakwaters. In Coastal Engineering. pp. 1638–1649. (1999)
  8. Liu, P.L.F.; Lin, P.; Chang, K.A.; Sakakiyama, T.: Numerical modeling of wave interaction with porous structures. J. Waterw. Port Coast. Ocean Eng. 125(6), 322–330 (1999)Article  Google Scholar 
  9. Abdolmaleki, K.; Thiagarajan, K.P.; Morris-Thomas, M.T.: Simulation of the dam break problem and impact flows using a Navier-Stokes solver. Simulation 13, 17 (2004)Google Scholar 
  10. Higuera, P.; Lara, J.L.; Losada, I.J.: Realistic wave generation and active wave absorption for Navier-Stokes models: application to OpenFOAM®. Coast. Eng. 71, 102–118 (2013)Article  Google Scholar 
  11. Higuera, P.; Lara, J.L.; Losada, I.J.: Three-dimensional interaction of waves and porous coastal structures using OpenFOAM®. Part II: application. Coast. Eng. 83, 259–270 (2014)Article  Google Scholar 
  12. Gui, Q.; Dong, P.; Shao, S.; Chen, Y.: Incompressible SPH simulation of wave interaction with porous structure. Ocean Eng. 110, 126–139 (2015)Article  Google Scholar 
  13. Dentale, F.; Donnarumma, G.; Carratelli, E.P.; Reale, F.: A numerical method to analyze the interaction between sea waves and rubble mound emerged breakwaters. WSEAS Trans. Fluid Mech 10, 106–116 (2015)Google Scholar 
  14. Dentale, F.; Reale, F.; Di Leo, A.; Carratelli, E.P.: A CFD approach to rubble mound breakwater design. Int. J. Naval Archit. Ocean Eng. 10(5), 644–650 (2018)Article  Google Scholar 
  15. Koley, S.: Wave transmission through multilayered porous breakwater under regular and irregular incident waves. Eng. Anal. Bound. Elem. 108, 393–401 (2019)Article  MathSciNet  MATH  Google Scholar 
  16. Koley, S.; Panduranga, K.; Almashan, N.; Neelamani, S.; Al-Ragum, A.: Numerical and experimental modeling of water wave interaction with rubble mound offshore porous breakwaters. Ocean Eng. 218, 108218 (2020)Article  Google Scholar 
  17. Pourteimouri, P.; Hejazi, K.: Development of an integrated numerical model for simulating wave interaction with permeable submerged breakwaters using extended Navier-Stokes equations. J. Mar. Sci. Eng. 8(2), 87 (2020)Article  Google Scholar 
  18. Cao, D.; Yuan, J.; Chen, H.: Towards modelling wave-induced forces on an armour layer unit of rubble mound coastal revetments. Ocean Eng. 239, 109811 (2021)Article  Google Scholar 
  19. Díaz-Carrasco, P.; Eldrup, M.R.; Andersen, T.L.: Advance in wave reflection estimation for rubble mound breakwaters: the importance of the relative water depth. Coast. Eng. 168, 103921 (2021)Article  Google Scholar 
  20. Vieira, F.; Taveira-Pinto, F.; Rosa-Santos, P.: Damage evolution in single-layer cube armoured breakwaters with a regular placement pattern. Coast. Eng. 169, 103943 (2021)Article  Google Scholar 
  21. Booshi, S.; Ketabdari, M.J.: Modeling of solitary wave interaction with emerged porous breakwater using PLIC-VOF method. Ocean Eng. 241, 110041 (2021)Article  Google Scholar 
  22. Aristodemo, F.; Filianoti, P.; Tripepi, G.; Gurnari, L.; Ghaderi, A.: On the energy transmission by a submerged barrier interacting with a solitary wave. Appl. Ocean Res. 122, 103123 (2022)Article  Google Scholar 
  23. Teixeira, P.R.; Didier, E.: Numerical analysis of performance of an oscillating water column wave energy converter inserted into a composite breakwater with rubble mound foundation. Ocean Eng. 278, 114421 (2023)Article  Google Scholar 
  24. Burgan, H.I.: Numerical modeling of structural irregularities on unsymmetrical buildings. Tehnički vjesnik 28(3), 856–861 (2021)Google Scholar 
  25. Jones, I.P.: CFDS-Flow3D user guide. (1994)
  26. Al Shaikhli, H.I.; Khassaf, S.I.: Stepped mound breakwater simulation by using flow 3D. Eurasian J. Eng. Technol. 6, 60–68 (2022)Google Scholar 
  27. Hirt, C.W.; Nichols, B.D.: Volume of fluid (VOF) method for the dynamics of free boundaries. J. Comput. Phys. 39(1), 201–225 (1981)Article  MATH  Google Scholar 
  28. Ghaderi, A.; Dasineh, M.; Aristodemo, F.; Aricò, C.: Numerical simulations of the flow field of a submerged hydraulic jump over triangular macroroughnesses. Water 13(5), 674 (2021)Article  Google Scholar 
  29. Yakhot, V.; Orszag, S.A.; Thangam, S.; Gatski, T.B.; Speziale, C.G.: Development of turbulence models for shear flows by a double expansion technique. Phys. Fluids A 4(7), 1510–1520 (1992)Article  MathSciNet  MATH  Google Scholar 
  30. Van der Meer, J.W.; Stam, C.J.M.: Wave runup on smooth and rock slopes of coastal structures. J. Waterw. Port Coast. Ocean Eng. 118(5), 534–550 (1992)Article  Google Scholar 
  31. Goda, Y.; Suzuki, Y. Estimation of incident and reflected waves in random wave experiments. In: ASCE, Proceedings of 15th International Conference on Coastal Engineering, (Honolulu, Hawaii). vol. 1, pp. 828–845. (1976)
  32. Zanuttigh, B.; Van der Meer, J.W.: Wave reflection from coastal structures. In: AA.VV., Proceedings of the XXX International Conference on Coastal Engineering, World Scientific, (San Diego, CA, USA, September 2006). pp. 4337–4349. (2006)
  33. Seelig W.N.; Ahrens J.P.: Estimation of wave reflection and energy dissipation coefficients for beaches, revetments, and breakwaters. CERC, Technical Paper, Fort Belvoir. vol. 81, p. 41 (1981)
  34. Mase, H.: Random wave runup height on gentle slope. J. Waterw. Port Coast. Ocean Eng. 115(5), 649–661 (1989)Article  Google Scholar 
Figure 11. Sketch of scour mechanism around USAF under random waves.

Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves

by Ruigeng Hu 1,Hongjun Liu 2,Hao Leng 1,Peng Yu 3 andXiuhai Wang 1,2,*

1College of Environmental Science and Engineering, Ocean University of China, Qingdao 266000, China

2Key Lab of Marine Environment and Ecology (Ocean University of China), Ministry of Education, Qingdao 266000, China

3Qingdao Geo-Engineering Survering Institute, Qingdao 266100, China

*Author to whom correspondence should be addressed.

J. Mar. Sci. Eng. 20219(8), 886; https://doi.org/10.3390/jmse9080886

Received: 6 July 2021 / Revised: 8 August 2021 / Accepted: 13 August 2021 / Published: 17 August 2021

(This article belongs to the Section Ocean Engineering)

Download 

Abstract

A series of numerical simulation were conducted to study the local scour around umbrella suction anchor foundation (USAF) under random waves. In this study, the validation was carried out firstly to verify the accuracy of the present model. Furthermore, the scour evolution and scour mechanism were analyzed respectively. In addition, two revised models were proposed to predict the equilibrium scour depth Seq around USAF. At last, a parametric study was carried out to study the effects of the Froude number Fr and Euler number Eu for the Seq. The results indicate that the present numerical model is accurate and reasonable for depicting the scour morphology under random waves. The revised Raaijmakers’s model shows good agreement with the simulating results of the present study when KCs,p < 8. The predicting results of the revised stochastic model are the most favorable for n = 10 when KCrms,a < 4. The higher Fr and Eu both lead to the more intensive horseshoe vortex and larger Seq.

Keywords: 

scournumerical investigationrandom wavesequilibrium scour depthKC number

1. Introduction

The rapid expansion of cities tends to cause social and economic problems, such as environmental pollution and traffic jam. As a kind of clean energy, offshore wind power has developed rapidly in recent years. The foundation of offshore wind turbine (OWT) supports the upper tower, and suffers the cyclic loading induced by waves, tides and winds, which exerts a vital influence on the OWT system. The types of OWT foundation include the fixed and floating foundation, and the fixed foundation was used usually for nearshore wind turbine. After the construction of fixed foundation, the hydrodynamic field changes in the vicinity of the foundation, leading to the horseshoe vortex formation and streamline compression at the upside and sides of foundation respectively [1,2,3,4]. As a result, the neighboring soil would be carried away by the shear stress induced by vortex, and the scour hole would emerge in the vicinity of foundation. The scour holes increase the cantilever length, and weaken the lateral bearing capacity of foundation [5,6,7,8,9]. Moreover, the natural frequency of OWT system increases with the increase of cantilever length, causing the resonance occurs when the system natural frequency equals the wave or wind frequency [10,11,12]. Given that, an innovative foundation called umbrella suction anchor foundation (USAF) has been designed for nearshore wind power. The previous studies indicated the USAF was characterized by the favorable lateral bearing capacity with the low cost [6,13,14]. The close-up of USAF is shown in Figure 1, and it includes six parts: 1-interal buckets, 2-external skirt, 3-anchor ring, 4-anchor branch, 5-supporting rod, 6-telescopic hook. The detailed description and application method of USAF can be found in reference [13].

Jmse 09 00886 g001 550

Figure 1. The close-up of umbrella suction anchor foundation (USAF).

Numerical and experimental investigations of scour around OWT foundation under steady currents and waves have been extensively studied by many researchers [1,2,15,16,17,18,19,20,21,22,23,24]. The seabed scour can be classified as two types according to Shields parameter θ, i.e., clear bed scour (θ < θcr) or live bed scour (θ > θcr). Due to the set of foundation, the adverse hydraulic pressure gradient exists at upstream foundation edges, resulting in the streamline separation between boundary layer flow and seabed. The separating boundary layer ascended at upstream anchor edges and developed into the horseshoe vortex. Then, the horseshoe vortex moved downstream gradually along the periphery of the anchor, and the vortex shed off continually at the lee-side of the anchor, i.e., wake vortex. The core of wake vortex is a negative pressure center, liking a vacuum cleaner. Hence, the soil particles were swirled into the negative pressure core and carried away by wake vortexes. At the same time, the onset of scour at rear side occurred. Finally, the wake vortex became downflow when the turbulence energy could not support the survival of wake vortex. According to Tavouktsoglou et al. [25], the scale of pile wall boundary layer is proportional to 1/ln(Rd) (Rd is pile Reynolds), which means the turbulence intensity induced by the flow-structure interaction would decrease with Rd increases, but the effects of Rd can be neglected only if the flow around the foundation is fully turbulent [26]. According to previous studies [1,15,27,28,29,30,31,32], the scour development around pile foundation under waves was significantly influenced by Shields parameter θ and KC number simultaneously (calculated by Equation (1)). Sand ripples widely existed around pile under waves in the case of live bed scour, and the scour morphology is related with θ and KC. Compared with θKC has a greater influence on the scour morphology [21,27,28]. The influence mechanism of KC on the scour around the pile is reflected in two aspects: the horseshoe vortex at upstream and wake vortex shedding at downstream.

KC=UwmTD��=�wm��(1)

where, Uwm is the maximum velocity of the undisturbed wave-induced oscillatory flow at the sea bottom above the wave boundary layer, T is wave period, and D is pile diameter.

There are two prerequisites to satisfy the formation of horseshoe vortex at upstream pile edges: (1) the incoming flow boundary layer with sufficient thickness and (2) the magnitude of upstream adverse pressure gradient making the boundary layer separating [1,15,16,18,20]. The smaller KC results the lower adverse pressure gradient, and the boundary layer cannot separate, herein, there is almost no horseshoe vortex emerging at upside of pile. Sumer et al. [1,15] carried out several sets of wave flume experiments under regular and irregular waves respectively, and the experiment results show that there is no horseshoe vortex when KC is less than 6. While the scale and lifespan of horseshoe vortex increase evidently with the increase of KC when KC is larger than 6. Moreover, the wake vortex contributes to the scour at lee-side of pile. Similar with the case of horseshoe vortex, there is no wake vortex when KC is less than 6. The wake vortex is mainly responsible for scour around pile when KC is greater than 6 and less than O(100), while horseshoe vortex controls scour nearly when KC is greater than O(100).

Sumer et al. [1] found that the equilibrium scour depth was nil around pile when KC was less than 6 under regular waves for live bed scour, while the equilibrium scour depth increased with the increase of KC. Based on that, Sumer proposed an equilibrium scour depth predicting equation (Equation (2)). Carreiras et al. [33] revised Sumer’s equation with m = 0.06 for nonlinear waves. Different with the findings of Sumer et al. [1] and Carreiras et al. [33], Corvaro et al. [21] found the scour still occurred for KC ≈ 4, and proposed the revised equilibrium scour depth predicting equation (Equation (3)) for KC > 4.

Rudolph and Bos [2] conducted a series of wave flume experiments to investigate the scour depth around monopile under waves only, waves and currents combined respectively, indicting KC was one of key parameters in influencing equilibrium scour depth, and proposed the equilibrium scour depth predicting equation (Equation (4)) for low KC (1 < KC < 10). Through analyzing the extensive data from published literatures, Raaijmakers and Rudolph [34] developed the equilibrium scour depth predicting equation (Equation (5)) for low KC, which was suitable for waves only, waves and currents combined. Khalfin [35] carried out several sets of wave flume experiments to study scour development around monopile, and proposed the equilibrium scour depth predicting equation (Equation (6)) for low KC (0.1 < KC < 3.5). Different with above equations, the Khalfin’s equation considers the Shields parameter θ and KC number simultaneously in predicting equilibrium scour depth. The flow reversal occurred under through in one wave period, so sand particles would be carried away from lee-side of pile to upside, resulting in sand particles backfilled into the upstream scour hole [20,29]. Considering the backfilling effects, Zanke et al. [36] proposed the equilibrium scour depth predicting equation (Equation (7)) around pile by theoretical analysis, and the equation is suitable for the whole range of KC number under regular waves and currents combined.

S/D=1.3(1−exp([−m(KC−6)])�/�=1.3(1−exp(−�(��−6))(2)

where, m = 0.03 for linear waves.

S/D=1.3(1−exp([−0.02(KC−4)])�/�=1.3(1−exp(−0.02(��−4))(3)

S/D=1.3γKwaveKhw�/�=1.3��wave�ℎw(4)

where, γ is safety factor, depending on design process, typically γ = 1.5, Kwave is correction factor considering wave action, Khw is correction factor considering water depth.

S/D=1.5[tanh(hwD)]KwaveKhw�/�=1.5tanh(ℎw�)�wave�ℎw(5)

where, hw is water depth.

S/D=0.0753(θθcr−−−√−0.5)0.69KC0.68�/�=0.0753(��cr−0.5)0.69��0.68(6)

where, θ is shields parameter, θcr is critical shields parameter.

S/D=2.5(1−0.5u/uc)xrelxrel=xeff/(1+xeff)xeff=0.03(1−0.35ucr/u)(KC−6)⎫⎭⎬⎪⎪�/�=2.5(1−0.5�/��)��������=����/(1+����)����=0.03(1−0.35�cr/�)(��−6)(7)

where, u is near-bed orbital velocity amplitude, uc is critical velocity corresponding the onset of sediment motion.

S/D=1.3{1−exp[−0.03(KC2lnn+36)1/2−6]}�/�=1.31−exp−0.03(��2ln�+36)1/2−6(8)

where, n is the 1/n’th highest wave for random waves

For predicting equilibrium scour depth under irregular waves, i.e., random waves, Sumer and Fredsøe [16] found it’s suitable to take Equation (2) to predict equilibrium scour depth around pile under random waves with the root-mean-square (RMS) value of near-bed orbital velocity amplitude Um and peak wave period TP to calculate KC. Khalfin [35] recommended the RMS wave height Hrms and peak wave period TP were used to calculate KC for Equation (6). References [37,38,39,40] developed a series of stochastic theoretical models to predict equilibrium scour depth around pile under random waves, nonlinear random waves plus currents respectively. The stochastic approach thought the 1/n’th highest wave were responsible for scour in vicinity of pile under random waves, and the KC was calculated in Equation (8) with Um and mean zero-crossing wave period Tz. The results calculated by Equation (8) agree well with experimental values of Sumer and Fredsøe [16] if the 1/10′th highest wave was used. To author’s knowledge, the stochastic approach proposed by Myrhaug and Rue [37] is the only theoretical model to predict equilibrium scour depth around pile under random waves for the whole range of KC number in published documents. Other methods of predicting scour depth under random waves are mainly originated from the equation for regular waves-only, waves and currents combined, which are limited to the large KC number, such as KC > 6 for Equation (2) and KC > 4 for Equation (3) respectively. However, situations with relatively low KC number (KC < 4) often occur in reality, for example, monopile or suction anchor for OWT foundations in ocean environment. Moreover, local scour around OWT foundations under random waves has not yet been investigated fully. Therefore, further study are still needed in the aspect of scour around OWT foundations with low KC number under random waves. Given that, this study presents the scour sediment model around umbrella suction anchor foundation (USAF) under random waves. In this study, a comparison of equilibrium scour depth around USAF between this present numerical models and the previous theoretical models and experimental results was presented firstly. Then, this study gave a comprehensive analysis for the scour mechanisms around USAF. After that, two revised models were proposed according to the model of Raaijmakers and Rudolph [34] and the stochastic model developed by Myrhaug and Rue [37] respectively to predict the equilibrium scour depth. Finally, a parametric study was conducted to study the effects of the Froude number (Fr) and Euler number (Eu) to equilibrium scour depth respectively.

2. Numerical Method

2.1. Governing Equations of Flow

The following equations adopted in present model are already available in Flow 3D software. The authors used these theoretical equations to simulate scour in random waves without modification. The incompressible viscous fluid motion satisfies the Reynolds-averaged Navier-Stokes (RANS) equation, so the present numerical model solves RANS equations:

∂u∂t+1VF(uAx∂u∂x+vAy∂u∂y+wAz∂u∂z)=−1ρf∂p∂x+Gx+fx∂�∂�+1��(���∂�∂�+���∂�∂�+���∂�∂�)=−1�f∂�∂�+��+��(9)

∂v∂t+1VF(uAx∂v∂x+vAy∂v∂y+wAz∂v∂z)=−1ρf∂p∂y+Gy+fy∂�∂�+1��(���∂�∂�+���∂�∂�+���∂�∂�)=−1�f∂�∂�+��+��(10)

∂w∂t+1VF(uAx∂w∂x+vAy∂w∂y+wAz∂w∂z)=−1ρf∂p∂z+Gz+fz∂�∂�+1��(���∂�∂�+���∂�∂�+���∂�∂�)=−1�f∂�∂�+��+��(11)

where, VF is the volume fraction; uv, and w are the velocity components in xyz direction respectively with Cartesian coordinates; Ai is the area fraction; ρf is the fluid density, fi is the viscous fluid acceleration, Gi is the fluid body acceleration (i = xyz).

2.2. Turbulent Model

The turbulence closure is available by the turbulent model, such as one-equation, the one-equation k-ε model, the standard k-ε model, RNG k-ε turbulent model and large eddy simulation (LES) model. The LES model requires very fine mesh grid, so the computational time is large, which hinders the LES model application in engineering. The RNG k-ε model can reduce computational time greatly with high accuracy in the near-wall region. Furthermore, the RNG k-ε model computes the maximum turbulent mixing length dynamically in simulating sediment scour model. Therefore, the RNG k-ε model was adopted to study the scour around anchor under random waves [41,42].

∂kT∂T+1VF(uAx∂kT∂x+vAy∂kT∂y+wAz∂kT∂z)=PT+GT+DiffkT−εkT∂��∂�+1��(���∂��∂�+���∂��∂�+���∂��∂�)=��+��+������−���(12)

∂εT∂T+1VF(uAx∂εT∂x+vAy∂εT∂y+wAz∂εT∂z)=CDIS1εTkT(PT+CDIS3GT)+Diffε−CDIS2ε2TkT∂��∂�+1��(���∂��∂�+���∂��∂�+���∂��∂�)=����1����(��+����3��)+�����−����2��2��(13)

where, kT is specific kinetic energy involved with turbulent velocity, GT is the turbulent energy generated by buoyancy; εT is the turbulent energy dissipating rate, PT is the turbulent energy, Diffε and DiffkT are diffusion terms associated with VFAiCDIS1CDIS2 and CDIS3 are dimensionless parameters, and CDIS1CDIS3 have default values of 1.42, 0.2 respectively. CDIS2 can be obtained from PT and kT.

2.3. Sediment Scour Model

The sand particles may suffer four processes under waves, i.e., entrainment, bed load transport, suspended load transport, and deposition, so the sediment scour model should depict the above processes efficiently. In present numerical simulation, the sediment scour model includes the following aspects:

2.3.1. Entrainment and Deposition

The combination of entrainment and deposition determines the net scour rate of seabed in present sediment scour model. The entrainment lift velocity of sand particles was calculated as [43]:

ulift,i=αinsd0.3∗(θ−θcr)1.5∥g∥di(ρi−ρf)ρf−−−−−−−−−−−−√�lift,i=�����*0.3(�−�cr)1.5���(��−�f)�f(14)

where, αi is the entrainment parameter, ns is the outward point perpendicular to the seabed, d* is the dimensionless diameter of sand particles, which was calculated by Equation (15), θcr is the critical Shields parameter, g is the gravity acceleration, di is the diameter of sand particles, ρi is the density of seabed species.

d∗=di(∥g∥ρf(ρi−ρf)μ2f)1/3�*=��(��f(��−�f)�f2)1/3(15)

where μf is the fluid dynamic viscosity.

In Equation (14), the entrainment parameter αi confirms the rate at which sediment erodes when the given shear stress is larger than the critical shear stress, and the recommended value 0.018 was adopted according to the experimental data of Mastbergen and Von den Berg [43]. ns is the outward pointing normal to the seabed interface, and ns = (0,0,1) according to the Cartesian coordinates used in present numerical model.

The shields parameter was obtained from the following equation:

θ=U2f,m(ρi/ρf−1)gd50�=�f,m2(��/�f−1)��50(16)

where, Uf,m is the maximum value of the near-bed friction velocity; d50 is the median diameter of sand particles. The detailed calculation procedure of θ was available in Soulsby [44].

The critical shields parameter θcr was obtained from the Equation (17) [44]

θcr=0.31+1.2d∗+0.055[1−exp(−0.02d∗)]�cr=0.31+1.2�*+0.0551−exp(−0.02�*)(17)

The sand particles begin to deposit on seabed when the turbulence energy weaken and cann’t support the particles suspending. The setting velocity of the particles was calculated from the following equation [44]:

usettling,i=νfdi[(10.362+1.049d3∗)0.5−10.36]�settling,�=�f��(10.362+1.049�*3)0.5−10.36(18)

where νf is the fluid kinematic viscosity.

2.3.2. Bed Load Transport

This is called bed load transport when the sand particles roll or bounce over the seabed and always have contact with seabed. The bed load transport velocity was computed by [45]:

ubedload,i=qb,iδicb,ifb�bedload,�=�b,����b,��b(19)

where, qb,i is the bed load transport rate, which was obtained from Equation (20), δi is the bed load thickness, which was calculated by Equation (21), cb,i is the volume fraction of sand i in the multiple species, fb is the critical packing fraction of the seabed.

qb,i=8[∥g∥(ρi−ρfρf)d3i]1/2�b,�=8�(��−�f�f)��31/2(20)

δi=0.3d0.7∗(θθcr−1)0.5di��=0.3�*0.7(��cr−1)0.5��(21)

2.3.3. Suspended Load Transport

Through the following transport equation, the suspended sediment concentration could be acquired.

∂Cs,i∂t+∇(us,iCs,i)=∇∇(DfCs,i)∂�s,�∂�+∇(�s,��s,�)=∇∇(�f�s,�)(22)

where, Cs,i is the suspended sand particles mass concentration of sand i in the multiple species, us,i is the sand particles velocity of sand iDf is the diffusivity.

The velocity of sand i in the multiple species could be obtained from the following equation:

us,i=u¯¯+usettling,ics,i�s,�=�¯+�settling,��s,�(23)

where, u¯�¯ is the velocity of mixed fluid-particles, which can be calculated by the RANS equation with turbulence model, cs,i is the suspended sand particles volume concentration, which was computed from Equation (24).

cs,i=Cs,iρi�s,�=�s,���(24)

3. Model Setup

The seabed-USAF-wave three-dimensional scour numerical model was built using Flow-3D software. As shown in Figure 2, the model includes sandy seabed, USAF model, sea water, two baffles and porous media. The dimensions of USAF are shown in Table 1. The sandy bed (210 m in length, 30 m in width and 11 m in height) is made up of uniform fine sand with median diameter d50 = 0.041 cm. The USAF model includes upper steel tube with the length of 20 m, which was installed in the middle of seabed. The location of USAF is positioned at 140 m from the upstream inflow boundary and 70 m from the downstream outflow boundary. Two baffles were installed at two ends of seabed. In order to eliminate the wave reflection basically, the porous media was set at the outflow side on the seabed.

Jmse 09 00886 g002 550

Figure 2. (a) The sketch of seabed-USAF-wave three-dimensional model; (b) boundary condation:Wv-wave boundary, S-symmetric boundary, O-outflow boundary; (c) USAF model.

Table 1. Numerical simulating cases.

Table

3.1. Mesh Geometric Dimensions

In the simulation of the scour under the random waves, the model includes the umbrella suction anchor foundation, seabed and fluid. As shown in Figure 3, the model mesh includes global mesh grid and nested mesh grid, and the total number of grids is 1,812,000. The basic procedure for building mesh grid consists of two steps. Step 1: Divide the global mesh using regular hexahedron with size of 0.6 × 0.6. The global mesh area is cubic box, embracing the seabed and whole fluid volume, and the dimensions are 210 m in length, 30 m in width and 32 m in height. The details of determining the grid size can see the following mesh sensitivity section. Step 2: Set nested fine mesh grid in vicinity of the USAF with size of 0.3 × 0.3 so as to shorten the computation cost and improve the calculation accuracy. The encryption range is −15 m to 15 m in x direction, −15 m to 15 m in y direction and 0 m to 32 m in z direction, respectively. In order to accurately capture the free-surface dynamics, such as the fluid-air interface, the volume of fluid (VOF) method was adopted for tracking the free water surface. One specific algorithm called FAVORTM (Fractional Area/Volume Obstacle Representation) was used to define the fractional face areas and fractional volumes of the cells which are open to fluid flow.

Jmse 09 00886 g003 550

Figure 3. The sketch of mesh grid.

3.2. Boundary Conditions

As shown in Figure 2, the initial fluid length is 210 m as long as seabed. A wave boundary was specified at the upstream offshore end. The details of determining the random wave spectrum can see the following wave parameters section. The outflow boundary was set at the downstream onshore end. The symmetry boundary was used at the top and two sides of the model. The symmetric boundaries were the better strategy to improve the computation efficiency and save the calculation cost [46]. At the seabed bottom, the wall boundary was adopted, which means the u = v = w= 0. Besides, the upper steel tube of USAF was set as no-slip condition.

3.3. Wave Parameters

The random waves with JONSWAP wave spectrum were used for all simulations as realistic representation of offshore conditions. The unidirectional JONSWAP frequency spectrum was described as [47]:

S(ω)=αg2ω5exp[−54(ωpω)4]γexp[−(ω−ωp)22σ2ω2p]�(�)=��2�5exp−54(�p�)4�exp−(�−�p)22�2�p2(25)

where, α is wave energy scale parameter, which is calculated by Equation (26), ω is frequency, ωp is wave spectrum peak frequency, which can be obtained from Equation (27). γ is wave spectrum peak enhancement factor, in this study γ = 3.3. σ is spectral width factor, σ equals 0.07 for ω ≤ ωp and 0.09 for ω > ωp respectively.

α=0.0076(gXU2)−0.22�=0.0076(���2)−0.22(26)

ωp=22(gU)(gXU2)−0.33�p=22(��)(���2)−0.33(27)

where, X is fetch length, U is average wind velocity at 10 m height from mean sea level.

In present numerical model, the input key parameters include X and U for wave boundary with JONSWAP wave spectrum. The objective wave height and period are available by different combinations of X and U. In this study, we designed 9 cases with different wave heights, periods and water depths for simulating scour around USAF under random waves (see Table 2). For random waves, the wave steepness ε and Ursell number Ur were acquired form Equations (28) and (29) respectively

ε=2πgHsT2a�=2���s�a2(28)

Ur=Hsk2h3w�r=�s�2ℎw3(29)

where, Hs is significant wave height, Ta is average wave period, k is wave number, hw is water depth. The Shield parameter θ satisfies θ > θcr for all simulations in current study, indicating the live bed scour prevails.

Table 2. Numerical simulating cases.

Table

3.4. Mesh Sensitivity

In this section, a mesh sensitivity analysis was conducted to investigate the influence of mesh grid size to results and make sure the calculation is mesh size independent and converged. Three mesh grid size were chosen: Mesh 1—global mesh grid size of 0.75 × 0.75, nested fine mesh grid size of 0.4 × 0.4, and total number of grids 1,724,000, Mesh 2—global mesh grid size of 0.6 × 0.6, nested fine mesh grid size of 0.3 × 0.3, and total number of grids 1,812,000, Mesh 3—global mesh grid size of 0.4 × 0.4, nested fine mesh grid size of 0.2 × 0.2, and total number of grids 1,932,000. The near-bed shear velocity U* is an important factor for influencing scour process [1,15], so U* at the position of (4,0,11.12) was evaluated under three mesh sizes. As the Figure 4 shown, the maximum error of shear velocity ∆U*1,2 is about 39.8% between the mesh 1 and mesh 2, and 4.8% between the mesh 2 and mesh 3. According to the mesh sensitivity criterion adopted by Pang et al. [48], it’s reasonable to think the results are mesh size independent and converged with mesh 2. Additionally, the present model was built according to prototype size, and the mesh size used in present model is larger than the mesh size adopted by Higueira et al. [49] and Corvaro et al. [50]. If we choose the smallest cell size, it will take too much time. For example, the simulation with Mesh3 required about 260 h by using a computer with Intel Xeon Scalable Gold 4214 CPU @24 Cores, 2.2 GHz and 64.00 GB RAM. Therefore, in this case, considering calculation accuracy and computation efficiency, the mesh 2 was chosen for all the simulation in this study.

Jmse 09 00886 g004 550

Figure 4. Comparison of near-bed shear velocity U* with different mesh grid size.

The nested mesh block was adopted for seabed in vicinity of the USAF, which was overlapped with the global mesh block. When two mesh blocks overlap each other, the governing equations are by default solved on the mesh block with smaller average cell size (i.e., higher grid resolution). It is should be noted that the Flow 3D software used the moving mesh captures the scour evolution and automatically adjusts the time step size to be as large as possible without exceeding any of the stability limits, affecting accuracy, or unduly increasing the effort required to enforce the continuity condition [51].

3.5. Model Validation

In order to verify the reliability of the present model, the results of present study were compared with the experimental data of Khosronejad et al. [52]. The experiment was conducted in an open channel with a slender vertical pile under unidirectional currents. The comparison of scour development between the present results and the experimental results is shown in Figure 5. The Figure 5 reveals that the present results agree well with the experimental data of Khosronejad et al. [52]. In the first stage, the scour depth increases rapidly. After that, the scour depth achieves a maximum value gradually. The equilibrium scour depth calculated by the present model is basically corresponding with the experimental results of Khosronejad et al. [52], although scour depth in the present model is slightly larger than the experimental results at initial stage.

Jmse 09 00886 g005 550

Figure 5. Comparison of time evolution of scour between the present study and Khosronejad et al. [52], Petersen et al. [17].

Secondly, another comparison was further conducted between the results of present study and the experimental data of Petersen et al. [17]. The experiment was carried out in a flume with a circular vertical pile in combined waves and current. Figure 4 shows a comparison of time evolution of scour depth between the simulating and the experimental results. As Figure 5 indicates, the scour depth in this study has good overall agreement with the experimental results proposed in Petersen et al. [17]. The equilibrium scour depth calculated by the present model is 0.399 m, which equals to the experimental value basically. Overall, the above verifications prove the present model is accurate and capable in dealing with sediment scour under waves.

In addition, in order to calibrate and validate the present model for hydrodynamic parameters, the comparison of water surface elevation was carried out with laboratory experiments conducted by Stahlmann [53] for wave gauge No. 3. The Figure 6 depicts the surface wave profiles between experiments and numerical model results. The comparison indicates that there is a good agreement between the model results and experimental values, especially the locations of wave crest and trough. Comparison of the surface elevation instructs the present model has an acceptable relative error, and the model is a calibrated in terms of the hydrodynamic parameters.

Jmse 09 00886 g006 550

Figure 6. Comparison of surface elevation between the present study and Stahlmann [53].

Finally, another comparison was conducted for equilibrium scour depth or maximum scour depth under random waves with the experimental data of Sumer and Fredsøe [16] and Schendel et al. [22]. The Figure 7 shows the comparison between the numerical results and experimental data of Run01, Run05, Run21 and Run22 in Sumer and Fredsøe [16] and test A05 and A09 in Schendel et al. [22]. As shown in Figure 7, the equilibrium scour depth or maximum scour depth distributed within the ±30 error lines basically, meaning the reliability and accuracy of present model for predicting equilibrium scour depth around foundation in random waves. However, compared with the experimental values, the present model overestimated the equilibrium scour depth generally. Given that, a calibration for scour depth was carried out by multiplying the mean reduced coefficient 0.85 in following section.

Jmse 09 00886 g007 550

Figure 7. Comparison of equilibrium (or maximum) scour depth between the present study and Sumer and Fredsøe [16], Schendel et al. [22].

Through the various examination for hydrodynamic and morphology parameters, it can be concluded that the present model is a validated and calibrated model for scour under random waves. Thus, the present numerical model would be utilized for scour simulation around foundation under random waves.

4. Numerical Results and Discussions

4.1. Scour Evolution

Figure 8 displays the scour evolution for case 1–9. As shown in Figure 8a, the scour depth increased rapidly at the initial stage, and then slowed down at the transition stage, which attributes to the backfilling occurred in scour holes under live bed scour condition, resulting in the net scour decreasing. Finally, the scour reached the equilibrium state when the amount of sediment backfilling equaled to that of scouring in the scour holes, i.e., the net scour transport rate was nil. Sumer and Fredsøe [16] proposed the following formula for the scour development under waves

St=Seq(1−exp(−t/Tc))�t=�eq(1−exp(−�/�c))(30)

where Tc is time scale of scour process.

Jmse 09 00886 g008 550

Figure 8. Time evolution of scour for case 1–9: (a) Case 1–5; (b) Case 6–9.

The computing time is 3600 s and the scour development curves in Figure 8 kept fluctuating, meaning it’s still not in equilibrium scour stage in these cases. According to Sumer and Fredsøe [16], the equilibrium scour depth can be acquired by fitting the data with Equation (30). From Figure 8, it can be seen that the scour evolution obtained from Equation (30) is consistent with the present study basically at initial stage, but the scour depth predicted by Equation (30) developed slightly faster than the simulating results and the Equation (30) overestimated the scour depth to some extent. Overall, the whole tendency of the results calculated by Equation (30) agrees well with the simulating results of the present study, which means the Equation (30) is applicable to depict the scour evolution around USAF under random waves.

4.2. Scour Mechanism under Random Waves

The scour morphology and scour evolution around USAF are similar under random waves in case 1~9. Taking case 7 as an example, the scour morphology is shown in Figure 9.

Jmse 09 00886 g009 550

Figure 9. Scour morphology under different times for case 7.

From Figure 9, at the initial stage (t < 1200 s), the scour occurred at upstream foundation edges between neighboring anchor branches. The maximum scour depth appeared at the lee-side of the USAF. Correspondingly, the sediments deposited at the periphery of the USAF, and the location of the maximum accretion depth was positioned at an angle of about 45° symmetrically with respect to the wave propagating direction in the lee-side of the USAF. After that, when t > 2400 s, the location of the maximum scour depth shifted to the upside of the USAF at an angle of about 45° with respect to the wave propagating direction.

According to previous studies [1,15,16,19,30,31], the horseshoe vortex, streamline compression and wake vortex shedding were responsible for scour around foundation. The Figure 10 displays the distribution of flow velocity in vicinity of foundation, which reflects the evolving processes of horseshoe vertex.

Jmse 09 00886 g010a 550
Jmse 09 00886 g010b 550

Figure 10. Velocity profile around USAF: (a) Flow runup and down stream at upstream anchor edges; (b) Horseshoe vortex at upstream anchor edges; (c) Flow reversal during wave through stage at lee side.

As shown in Figure 10, the inflow tripped to the upstream edges of the USAF and it was blocked by the upper tube of USAF. Then, the downflow formed the horizontal axis clockwise vortex and rolled on the seabed bypassing the tube, that is, the horseshoe vortex (Figure 11). The Figure 12 displays the turbulence intensity around the tube on the seabed. From Figure 12, it can be seen that the turbulence intensity was high-intensity with respect to the region of horseshoe vortex. This phenomenon occurred because of drastic water flow momentum exchanging in the horseshoe vortex. As a result, it created the prominent shear stress on the seabed, causing the local scour at the upstream edges of USAF. Besides, the horseshoe vortex moved downstream gradually along the periphery of the tube and the wake vortex shed off continually at the lee-side of the USAF, i.e., wake vortex.

Jmse 09 00886 g011 550

Figure 11. Sketch of scour mechanism around USAF under random waves.

Jmse 09 00886 g012 550

Figure 12. Turbulence intensity: (a) Turbulence intensity of horseshoe vortex; (b) Turbulence intensity of wake vortex; (c) Turbulence intensity of accretion area.

The core of wake vortex is a negative pressure center, liking a vacuum cleaner [11,42]. Hence, the soil particles were swirled into the negative pressure core and carried away by wake vortex. At the same time, the onset of scour at rear side occurred. Finally, the wake vortex became downflow at the downside of USAF. As is shown in Figure 12, the turbulence intensity was low where the downflow occurred at lee-side, which means the turbulence energy may not be able to support the survival of wake vortex, leading to accretion happening. As mentioned in previous section, the formation of horseshoe vortex was dependent with adverse pressure gradient at upside of foundation. As shown in Figure 13, the evaluated range of pressure distribution is −15 m to 15 m in x direction. The t = 450 s and t = 1800 s indicate that the wave crest and trough arrived at the upside and lee-side of the foundation respectively, and the t = 350 s was neither the wave crest nor trough. The adverse gradient pressure reached the maximum value at t = 450 s corresponding to the wave crest phase. In this case, it’s helpful for the wave boundary separating fully from seabed, which leads to the formation of horseshoe vortex with high turbulence intensity. Therefore, the horseshoe vortex is responsible for the local scour between neighboring anchor branches at upside of USAF. What’s more, due to the combination of the horseshoe vortex and streamline compression, the maximum scour depth occurred at the upside of the USAF with an angle of about 45° corresponding to the wave propagating direction. This is consistent with the findings of Pang et al. [48] and Sumer et al. [1,15] in case of regular waves. At the wave trough phase (t = 1800 s), the pressure gradient became positive at upstream USAF edges, which hindered the separating of wave boundary from seabed. In the meantime, the flow reversal occurred (Figure 10) and the adverse gradient pressure appeared at downstream USAF edges, but the magnitude of adverse gradient pressure at lee-side was lower than the upstream gradient pressure under wave crest. In this way, the intensity of horseshoe vortex behind the USAF under wave trough was low, which explains the difference of scour depth at upstream and downstream, i.e., the scour asymmetry. In other words, the scour asymmetry at upside and downside of USAF was attributed to wave asymmetry for random waves, and the phenomenon became more evident for nonlinear waves [21]. Briefly speaking, the vortex system at wave crest phase was mainly related to the scour process around USAF under random waves.

Jmse 09 00886 g013 550

Figure 13. Pressure distribution around USAF.

4.3. Equilibrium Scour Depth

The KC number is a key parameter for horseshoe vortex emerging and evolving under waves. According to Equation (1), when pile diameter D is fixed, the KC depends on the maximum near-bed velocity Uwm and wave period T. For random waves, the Uwm can be denoted by the root-mean-square (RMS) value of near-bed velocity amplitude Uwm,rms or the significant value of near-bed velocity amplitude Uwm,s. The Uwm,rms and Uwm,s for all simulating cases of the present study are listed in Table 3 and Table 4. The T can be denoted by the mean up zero-crossing wave period Ta, peak wave period Tp, significant wave period Ts, the maximum wave period Tm, 1/10′th highest wave period Tn = 1/10 and 1/5′th highest wave period Tn = 1/5 for random waves, so the different combinations of Uwm and T will acquire different KC. The Table 3 and Table 4 list 12 types of KC, for example, the KCrms,s was calculated by Uwm,rms and Ts. Sumer and Fredsøe [16] conducted a series of wave flume experiments to investigate the scour depth around monopile under random waves, and found the equilibrium scour depth predicting equation (Equation (2)) for regular waves was applicable for random waves with KCrms,p. It should be noted that the Equation (2) is only suitable for KC > 6 under regular waves or KCrms,p > 6 under random waves.

Table 3. Uwm,rms and KC for case 1~9.

Table

Table 4. Uwm,s and KC for case 1~9.

Table

Raaijmakers and Rudolph [34] proposed the equilibrium scour depth predicting model (Equation (5)) around pile under waves, which is suitable for low KC. The format of Equation (5) is similar with the formula proposed by Breusers [54], which can predict the equilibrium scour depth around pile at different scour stages. In order to verify the applicability of Raaijmakers’s model for predicting the equilibrium scour depth around USAF under random waves, a validation of the equilibrium scour depth Seq between the present study and Raaijmakers’s equation was conducted. The position where the scour depth Seq was evaluated is the location of the maximum scour depth, and it was depicted in Figure 14. The Figure 15 displays the comparison of Seq with different KC between the present study and Raaijmakers’s model.

Jmse 09 00886 g014 550

Figure 14. Sketch of the position where the Seq was evaluated.

Jmse 09 00886 g015a 550
Jmse 09 00886 g015b 550

Figure 15. Comparison of the equilibrium scour depth between the present model and the model of Raaijmakers and Rudolph [34]: (aKCrms,sKCrms,a; (bKCrms,pKCrms,m; (cKCrms,n = 1/10KCrms,n = 1/5; (dKCs,sKCs,a; (eKCs,pKCs,m; (fKCs,n = 1/10KCs,n = 1/5.

As shown in Figure 15, there is an error in predicting Seq between the present study and Raaijmakers’s model, and Raaijmakers’s model underestimates the results generally. Although the error exists, the varying trend of Seq with KC obtained from Raaijmakers’s model is consistent with the present study basically. What’s more, the error is minimum and the Raaijmakers’s model is of relatively high accuracy for predicting scour around USAF under random waves by using KCs,p. Based on this, a further revision was made to eliminate the error as much as possible, i.e., add the deviation value ∆S/D in the Raaijmakers’s model. The revised equilibrium scour depth predicting equation based on Raaijmakers’s model can be written as

S′eq/D=1.95[tanh(hD)](1−exp(−0.012KCs,p))+ΔS/D�eq′/�=1.95tanh(ℎ�)(1−exp(−0.012��s,p))+∆�/�(31)

As the Figure 16 shown, through trial-calculation, when ∆S/D = 0.05, the results calculated by Equation (31) show good agreement with the simulating results of the present study. The maximum error is about 18.2% and the engineering requirements have been met basically. In order to further verify the accuracy of the revised model for large KC (KCs,p > 4) under random waves, a validation between the revised model and the previous experimental results [21]. The experiment was conducted in a flume (50 m in length, 1.0 m in width and 1.3 m in height) with a slender vertical pile (D = 0.1 m) under random waves. The seabed is composed of 0.13 m deep layer of sand with d50 = 0.6 mm and the water depth is 0.5 m for all tests. The significant wave height is 0.12~0.21 m and the KCs,p is 5.52~11.38. The comparison between the predicting results by Equation (31) and the experimental results of Corvaro et al. [21] is shown in Figure 17. From Figure 17, the experimental data evenly distributes around the predicted results and the prediction accuracy is favorable when KCs,p < 8. However, the gap between the predicting results and experimental data becomes large and the Equation (31) overestimates the equilibrium scour depth to some extent when KCs,p > 8.

Jmse 09 00886 g016 550

Figure 16. Comparison of Seq between the simulating results and the predicting values by Equation (31).

Jmse 09 00886 g017 550

Figure 17. Comparison of Seq/D between the Experimental results of Corvaro et al. [21] and the predicting values by Equation (31).

In ocean environment, the waves are composed of a train of sinusoidal waves with different frequencies and amplitudes. The energy of constituent waves with very large and very small frequencies is relatively low, and the energy of waves is mainly concentrated in a certain range of moderate frequencies. Myrhaug and Rue [37] thought the 1/n’th highest wave was responsible for scour and proposed the stochastic model to predict the equilibrium scour depth around pile under random waves for full range of KC. Noteworthy is that the KC was denoted by KCrms,a in the stochastic model. To verify the application of the stochastic model for predicting scour depth around USAF, a validation between the simulating results of present study and predicting results by the stochastic model with n = 2,3,5,10,20,500 was carried out respectively.

As shown in Figure 18, compared with the simulating results, the stochastic model underestimates the equilibrium scour depth around USAF generally. Although the error exists, the varying trend of Seq with KCrms,a obtained from the stochastic model is consistent with the present study basically. What’s more, the gap between the predicting values by stochastic model and the simulating results decreases with the increase of n, but for large n, for example n = 500, the varying trend diverges between the predicting values and simulating results, meaning it’s not feasible only by increasing n in stochastic model to predict the equilibrium scour depth around USAF.

Jmse 09 00886 g018 550

Figure 18. Comparison of Seq between the simulating results and the predicting values by Equation (8).

The Figure 19 lists the deviation value ∆Seq/D′ between the predicting values and simulating results with different KCrms,a and n. Then, fitted the relationship between the ∆S′and n under different KCrms,a, and the fitting curve can be written by Equation (32). The revised stochastic model (Equation (33)) can be acquired by adding ∆Seq/D′ to Equation (8).

ΔSeq/D=0.052*exp(−n/6.566)+0.068∆�eq/�=0.052*exp(−�/6.566)+0.068(32)

S′eq¯/D=S′eq/D+0.052*exp(−n/6.566)+0.068�eq′¯/�=�eq′/�+0.052*exp(−�/6.566)+0.068(33)

Jmse 09 00886 g019 550

Figure 19. The fitting line between ∆S′and n.

The comparison between the predicting results by Equation (33) and the simulating results of present study is shown in Figure 20. According to the Figure 20, the varying trend of Seq with KCrms,a obtained from the stochastic model is consistent with the present study basically. Compared with predicting results by the stochastic model, the results calculated by Equation (33) is favorable. Moreover, comparison with simulating results indicates that the predicting results are the most favorable for n = 10, which is consistent with the findings of Myrhaug and Rue [37] for equilibrium scour depth predicting around slender pile in case of random waves.

Jmse 09 00886 g020 550

Figure 20. Comparison of Seq between the simulating results and the predicting values by Equation (33).

In order to further verify the accuracy of the Equation (33) for large KC (KCrms,a > 4) under random waves, a validation was conducted between the Equation (33) and the previous experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21]. The details of experiments conducted by Corvaro et al. [21] were described in above section. Sumer and Fredsøe [16] investigated the local scour around pile under random waves. The experiments were conducted in a wave basin with a slender vertical pile (D = 0.032, 0.055 m). The seabed is composed of 0.14 m deep layer of sand with d50 = 0.2 mm and the water depth was maintained at 0.5 m. The JONSWAP wave spectrum was used and the KCrms,a was 5.29~16.95. The comparison between the predicting results by Equation (33) and the experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21] are shown in Figure 21. From Figure 21, contrary to the case of low KCrms,a (KCrms,a < 4), the error between the predicting values and experimental results increases with decreasing of n for KCrms,a > 4. Therefore, the predicting results are the most favorable for n = 2 when KCrms,a > 4.

Jmse 09 00886 g021 550

Figure 21. Comparison of Seq between the experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21] and the predicting values by Equation (33).

Noteworthy is that the present model was built according to prototype size, so the errors between the numerical results and experimental data of References [16,21] may be attribute to the scale effects. In laboratory experiments on scouring process, it is typically impossible to ensure a rigorous similarity of all physical parameters between the model and prototype structure, leading to the scale effects in the laboratory experiments. To avoid a cohesive behaviour, the bed material was not scaled geometrically according to model scale. As a consequence, the relatively large-scaled sediments sizes may result in the overestimation of bed load transport and underestimation of suspended load transport compared with field conditions. What’s more, the disproportional scaled sediment presumably lead to the difference of bed roughness between the model and prototype, and thus large influences for wave boundary layer on the seabed and scour process. Besides, according to Corvaro et al. [21] and Schendel et al. [55], the pile Reynolds numbers and Froude numbers both affect the scour depth for the condition of non fully developed turbulent flow in laboratory experiments.

4.4. Parametric Study

4.4.1. Influence of Froude Number

As described above, the set of foundation leads to the adverse pressure gradient appearing at upstream, leading to the wave boundary layer separating from seabed, then horseshoe vortex formatting and the horseshoe vortex are mainly responsible for scour around foundation (see Figure 22). The Froude number Fr is the key parameter to influence the scale and intensity of horseshoe vortex. The Fr under waves can be calculated by the following formula [42]

Fr=UwgD−−−√�r=�w��(34)

where Uw is the mean water particle velocity during 1/4 cycle of wave oscillation, obtained from the following formula. Noteworthy is that the root-mean-square (RMS) value of near-bed velocity amplitude Uwm,rms is used for calculating Uwm.

Uw=1T/4∫0T/4Uwmsin(t/T)dt=2πUwm�w=1�/4∫0�/4�wmsin(�/�)��=2��wm(35)

Jmse 09 00886 g022 550

Figure 22. Sketch of flow field at upstream USAF edges.

Tavouktsoglou et al. [25] proposed the following formula between Fr and the vertical location of the stagnation y

yh∝Fer�ℎ∝�r�(36)

where e is constant.

The Figure 23 displays the relationship between Seq/D and Fr of the present study. In order to compare with the simulating results, the experimental data of Corvaro et al. [21] was also depicted in Figure 23. As shown in Figure 23, the equilibrium scour depth appears a logarithmic increase as Fr increases and approaches the mathematical asymptotic value, which is also consistent with the experimental results of Corvaro et al. [21]. According to Figure 24, the adverse pressure gradient pressure at upstream USAF edges increases with the increase of Fr, which is benefit for the wave boundary layer separating from seabed, resulting in the high-intensity horseshoe vortex, hence, causing intensive scour around USAF. Based on the previous study of Tavouktsoglou et al. [25] for scour around pile under currents, the high Fr leads to the stagnation point is closer to the mean sea level for shallow water, causing the stronger downflow kinetic energy. As mentioned in previous section, the energy of downflow at upstream makes up the energy of the subsequent horseshoe vortex, so the stronger downflow kinetic energy results in the more intensive horseshoe vortex. Therefore, the higher Fr leads to the more intensive horseshoe vortex by influencing the position of stagnation point y presumably. Qi and Gao [19] carried out a series of flume tests to investigate the scour around pile under regular waves, and proposed the fitting formula between Seq/D and Fr as following

lg(Seq/D)=Aexp(B/Fr)+Clg(�eq/�)=�exp(�/�r)+�(37)

where AB and C are constant.

Jmse 09 00886 g023 550

Figure 23. The fitting curve between Seq/D and Fr.

Jmse 09 00886 g024 550

Figure 24. Sketch of adverse pressure gradient at upstream USAF edges.

Took the Equation (37) to fit the simulating results with A = −0.002, B = 0.686 and C = −0.808, and the results are shown in Figure 23. From Figure 23, the simulating results evenly distribute around the Equation (37) and the varying trend of Seq/D and Fr in present study is consistent with Equation (37) basically, meaning the Equation (37) is applicable to express the relationship of Seq/D with Fr around USAF under random waves.

4.4.2. Influence of Euler Number

The Euler number Eu is the influencing factor for the hydrodynamic field around foundation. The Eu under waves can be calculated by the following formula. The Eu can be represented by the Equation (38) for uniform cylinders [25]. The root-mean-square (RMS) value of near-bed velocity amplitude Um,rms is used for calculating Um.

Eu=U2mgD�u=�m2��(38)

where Um is depth-averaged flow velocity.

The Figure 25 displays the relationship between Seq/D and Eu of the present study. In order to compare with the simulating results, the experimental data of Sumer and Fredsøe [16] and Corvaro et al. [21] were also plotted in Figure 25. As shown in Figure 25, similar with the varying trend of Seq/D and Fr, the equilibrium scour depth appears a logarithmic increase as Eu increases and approaches the mathematical asymptotic value, which is also consistent with the experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21]. According to Figure 24, the adverse pressure gradient pressure at upstream USAF edges increases with the increasing of Eu, which is benefit for the wave boundary layer separating from seabed, inducing the high-intensity horseshoe vortex, hence, causing intensive scour around USAF.

Jmse 09 00886 g025 550

Figure 25. The fitting curve between Seq/D and Eu.

Therefore, the variation of Fr and Eu reflect the magnitude of adverse pressure gradient pressure at upstream. Given that, the Equation (37) also was used to fit the simulating results with A = 8.875, B = 0.078 and C = −9.601, and the results are shown in Figure 25. From Figure 25, the simulating results evenly distribute around the Equation (37) and the varying trend of Seq/D and Eu in present study is consistent with Equation (37) basically, meaning the Equation (37) is also applicable to express the relationship of Seq/D with Eu around USAF under random waves. Additionally, according to the above description of Fr, it can be inferred that the higher Fr and Eu both lead to the more intensive horseshoe vortex by influencing the position of stagnation point y presumably.

5. Conclusions

A series of numerical models were established to investigate the local scour around umbrella suction anchor foundation (USAF) under random waves. The numerical model was validated for hydrodynamic and morphology parameters by comparing with the experimental data of Khosronejad et al. [52], Petersen et al. [17], Sumer and Fredsøe [16] and Schendel et al. [22]. Based on the simulating results, the scour evolution and scour mechanisms around USAF under random waves were analyzed respectively. Two revised models were proposed according to the model of Raaijmakers and Rudolph [34] and the stochastic model developed by Myrhaug and Rue [37] to predict the equilibrium scour depth around USAF under random waves. Finally, a parametric study was carried out with the present model to study the effects of the Froude number Fr and Euler number Eu to the equilibrium scour depth around USAF under random waves. The main conclusions can be described as follows.(1)

The packed sediment scour model and the RNG k−ε turbulence model were used to simulate the sand particles transport processes and the flow field around UASF respectively. The scour evolution obtained by the present model agrees well with the experimental results of Khosronejad et al. [52], Petersen et al. [17], Sumer and Fredsøe [16] and Schendel et al. [22], which indicates that the present model is accurate and reasonable for depicting the scour morphology around UASF under random waves.(2)

The vortex system at wave crest phase is mainly related to the scour process around USAF under random waves. The maximum scour depth appeared at the lee-side of the USAF at the initial stage (t < 1200 s). Subsequently, when t > 2400 s, the location of the maximum scour depth shifted to the upside of the USAF at an angle of about 45° with respect to the wave propagating direction.(3)

The error is negligible and the Raaijmakers’s model is of relatively high accuracy for predicting scour around USAF under random waves when KC is calculated by KCs,p. Given that, a further revision model (Equation (31)) was proposed according to Raaijmakers’s model to predict the equilibrium scour depth around USAF under random waves and it shows good agreement with the simulating results of the present study when KCs,p < 8.(4)

Another further revision model (Equation (33)) was proposed according to the stochastic model established by Myrhaug and Rue [37] to predict the equilibrium scour depth around USAF under random waves, and the predicting results are the most favorable for n = 10 when KCrms,a < 4. However, contrary to the case of low KCrms,a, the predicting results are the most favorable for n = 2 when KCrms,a > 4 by the comparison with experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21].(5)

The same formula (Equation (37)) is applicable to express the relationship of Seq/D with Eu or Fr, and it can be inferred that the higher Fr and Eu both lead to the more intensive horseshoe vortex and larger Seq.

Author Contributions

Conceptualization, H.L. (Hongjun Liu); Data curation, R.H. and P.Y.; Formal analysis, X.W. and H.L. (Hao Leng); Funding acquisition, X.W.; Writing—original draft, R.H. and P.Y.; Writing—review & editing, X.W. and H.L. (Hao Leng); The final manuscript has been approved by all the authors. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Fundamental Research Funds for the Central Universities (grant number 202061027) and the National Natural Science Foundation of China (grant number 41572247).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available on request from the corresponding author.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Sumer, B.M.; Fredsøe, J.; Christiansen, N. Scour Around Vertical Pile in Waves. J. Waterw. Port. Coast. Ocean Eng. 1992118, 15–31. [Google Scholar] [CrossRef]
  2. Rudolph, D.; Bos, K. Scour around a monopile under combined wave-current conditions and low KC-numbers. In Proceedings of the 6th International Conference on Scour and Erosion, Amsterdam, The Netherlands, 1–3 November 2006; pp. 582–588. [Google Scholar]
  3. Nielsen, A.W.; Liu, X.; Sumer, B.M.; Fredsøe, J. Flow and bed shear stresses in scour protections around a pile in a current. Coast. Eng. 201372, 20–38. [Google Scholar] [CrossRef]
  4. Ahmad, N.; Bihs, H.; Myrhaug, D.; Kamath, A.; Arntsen, Ø.A. Three-dimensional numerical modelling of wave-induced scour around piles in a side-by-side arrangement. Coast. Eng. 2018138, 132–151. [Google Scholar] [CrossRef]
  5. Li, H.; Ong, M.C.; Leira, B.J.; Myrhaug, D. Effects of Soil Profile Variation and Scour on Structural Response of an Offshore Monopile Wind Turbine. J. Offshore Mech. Arct. Eng. 2018140, 042001. [Google Scholar] [CrossRef]
  6. Li, H.; Liu, H.; Liu, S. Dynamic analysis of umbrella suction anchor foundation embedded in seabed for offshore wind turbines. Géoméch. Energy Environ. 201710, 12–20. [Google Scholar] [CrossRef]
  7. Fazeres-Ferradosa, T.; Rosa-Santos, P.; Taveira-Pinto, F.; Vanem, E.; Carvalho, H.; Correia, J.A.F.D.O. Editorial: Advanced research on offshore structures and foundation design: Part 1. Proc. Inst. Civ. Eng. Marit. Eng. 2019172, 118–123. [Google Scholar] [CrossRef]
  8. Chavez, C.E.A.; Stratigaki, V.; Wu, M.; Troch, P.; Schendel, A.; Welzel, M.; Villanueva, R.; Schlurmann, T.; De Vos, L.; Kisacik, D.; et al. Large-Scale Experiments to Improve Monopile Scour Protection Design Adapted to Climate Change—The PROTEUS Project. Energies 201912, 1709. [Google Scholar] [CrossRef][Green Version]
  9. Wu, M.; De Vos, L.; Chavez, C.E.A.; Stratigaki, V.; Fazeres-Ferradosa, T.; Rosa-Santos, P.; Taveira-Pinto, F.; Troch, P. Large Scale Experimental Study of the Scour Protection Damage Around a Monopile Foundation Under Combined Wave and Current Conditions. J. Mar. Sci. Eng. 20208, 417. [Google Scholar] [CrossRef]
  10. Sørensen, S.P.H.; Ibsen, L.B. Assessment of foundation design for offshore monopiles unprotected against scour. Ocean Eng. 201363, 17–25. [Google Scholar] [CrossRef]
  11. Prendergast, L.; Gavin, K.; Doherty, P. An investigation into the effect of scour on the natural frequency of an offshore wind turbine. Ocean Eng. 2015101, 1–11. [Google Scholar] [CrossRef][Green Version]
  12. Fazeres-Ferradosa, T.; Chambel, J.; Taveira-Pinto, F.; Rosa-Santos, P.; Taveira-Pinto, F.; Giannini, G.; Haerens, P. Scour Protections for Offshore Foundations of Marine Energy Harvesting Technologies: A Review. J. Mar. Sci. Eng. 20219, 297. [Google Scholar] [CrossRef]
  13. Yang, Q.; Yu, P.; Liu, Y.; Liu, H.; Zhang, P.; Wang, Q. Scour characteristics of an offshore umbrella suction anchor foundation under the combined actions of waves and currents. Ocean Eng. 2020202, 106701. [Google Scholar] [CrossRef]
  14. Yu, P.; Hu, R.; Yang, J.; Liu, H. Numerical investigation of local scour around USAF with different hydraulic conditions under currents and waves. Ocean Eng. 2020213, 107696. [Google Scholar] [CrossRef]
  15. Sumer, B.M.; Christiansen, N.; Fredsøe, J. The horseshoe vortex and vortex shedding around a vertical wall-mounted cylinder exposed to waves. J. Fluid Mech. 1997332, 41–70. [Google Scholar] [CrossRef]
  16. Sumer, B.M.; Fredsøe, J. Scour around Pile in Combined Waves and Current. J. Hydraul. Eng. 2001127, 403–411. [Google Scholar] [CrossRef]
  17. Petersen, T.U.; Sumer, B.M.; Fredsøe, J. Time scale of scour around a pile in combined waves and current. In Proceedings of the 6th International Conference on Scour and Erosion, Paris, France, 27–31 August 2012. [Google Scholar]
  18. Petersen, T.U.; Sumer, B.M.; Fredsøe, J.; Raaijmakers, T.C.; Schouten, J.-J. Edge scour at scour protections around piles in the marine environment—Laboratory and field investigation. Coast. Eng. 2015106, 42–72. [Google Scholar] [CrossRef]
  19. Qi, W.; Gao, F. Equilibrium scour depth at offshore monopile foundation in combined waves and current. Sci. China Ser. E Technol. Sci. 201457, 1030–1039. [Google Scholar] [CrossRef][Green Version]
  20. Larsen, B.E.; Fuhrman, D.R.; Baykal, C.; Sumer, B.M. Tsunami-induced scour around monopile foundations. Coast. Eng. 2017129, 36–49. [Google Scholar] [CrossRef][Green Version]
  21. Corvaro, S.; Marini, F.; Mancinelli, A.; Lorenzoni, C.; Brocchini, M. Hydro- and Morpho-dynamics Induced by a Vertical Slender Pile under Regular and Random Waves. J. Waterw. Port. Coast. Ocean Eng. 2018144, 04018018. [Google Scholar] [CrossRef]
  22. Schendel, A.; Welzel, M.; Schlurmann, T.; Hsu, T.-W. Scour around a monopile induced by directionally spread irregular waves in combination with oblique currents. Coast. Eng. 2020161, 103751. [Google Scholar] [CrossRef]
  23. Fazeres-Ferradosa, T.; Taveira-Pinto, F.; Romão, X.; Reis, M.; das Neves, L. Reliability assessment of offshore dynamic scour protections using copulas. Wind. Eng. 201843, 506–538. [Google Scholar] [CrossRef]
  24. Fazeres-Ferradosa, T.; Welzel, M.; Schendel, A.; Baelus, L.; Santos, P.R.; Pinto, F.T. Extended characterization of damage in rubble mound scour protections. Coast. Eng. 2020158, 103671. [Google Scholar] [CrossRef]
  25. Tavouktsoglou, N.S.; Harris, J.M.; Simons, R.R.; Whitehouse, R.J.S. Equilibrium Scour-Depth Prediction around Cylindrical Structures. J. Waterw. Port. Coast. Ocean Eng. 2017143, 04017017. [Google Scholar] [CrossRef][Green Version]
  26. Ettema, R.; Melville, B.; Barkdoll, B. Scale Effect in Pier-Scour Experiments. J. Hydraul. Eng. 1998124, 639–642. [Google Scholar] [CrossRef]
  27. Umeda, S. Scour Regime and Scour Depth around a Pile in Waves. J. Coast. Res. Spec. Issue 201164, 845–849. [Google Scholar]
  28. Umeda, S. Scour process around monopiles during various phases of sea storms. J. Coast. Res. 2013165, 1599–1604. [Google Scholar] [CrossRef]
  29. Baykal, C.; Sumer, B.; Fuhrman, D.R.; Jacobsen, N.; Fredsøe, J. Numerical simulation of scour and backfilling processes around a circular pile in waves. Coast. Eng. 2017122, 87–107. [Google Scholar] [CrossRef][Green Version]
  30. Miles, J.; Martin, T.; Goddard, L. Current and wave effects around windfarm monopile foundations. Coast. Eng. 2017121, 167–178. [Google Scholar] [CrossRef][Green Version]
  31. Miozzi, M.; Corvaro, S.; Pereira, F.A.; Brocchini, M. Wave-induced morphodynamics and sediment transport around a slender vertical cylinder. Adv. Water Resour. 2019129, 263–280. [Google Scholar] [CrossRef]
  32. Yu, T.; Zhang, Y.; Zhang, S.; Shi, Z.; Chen, X.; Xu, Y.; Tang, Y. Experimental study on scour around a composite bucket foundation due to waves and current. Ocean Eng. 2019189, 106302. [Google Scholar] [CrossRef]
  33. Carreiras, J.; Larroudé, P.; Seabra-Santos, F.; Mory, M. Wave Scour Around Piles. In Proceedings of the Coastal Engineering 2000, American Society of Civil Engineers (ASCE), Sydney, Australia, 16–21 July 2000; pp. 1860–1870. [Google Scholar]
  34. Raaijmakers, T.; Rudolph, D. Time-dependent scour development under combined current and waves conditions—Laboratory experiments with online monitoring technique. In Proceedings of the 4th International Conference on Scour and Erosion, Tokyo, Japan, 5–7 November 2008; pp. 152–161. [Google Scholar]
  35. Khalfin, I.S. Modeling and calculation of bed score around large-diameter vertical cylinder under wave action. Water Resour. 200734, 357. [Google Scholar] [CrossRef][Green Version]
  36. Zanke, U.C.; Hsu, T.-W.; Roland, A.; Link, O.; Diab, R. Equilibrium scour depths around piles in noncohesive sediments under currents and waves. Coast. Eng. 201158, 986–991. [Google Scholar] [CrossRef]
  37. Myrhaug, D.; Rue, H. Scour below pipelines and around vertical piles in random waves. Coast. Eng. 200348, 227–242. [Google Scholar] [CrossRef]
  38. Myrhaug, D.; Ong, M.C.; Føien, H.; Gjengedal, C.; Leira, B.J. Scour below pipelines and around vertical piles due to second-order random waves plus a current. Ocean Eng. 200936, 605–616. [Google Scholar] [CrossRef]
  39. Myrhaug, D.; Ong, M.C. Random wave-induced onshore scour characteristics around submerged breakwaters using a stochastic method. Ocean Eng. 201037, 1233–1238. [Google Scholar] [CrossRef]
  40. Ong, M.C.; Myrhaug, D.; Hesten, P. Scour around vertical piles due to long-crested and short-crested nonlinear random waves plus a current. Coast. Eng. 201373, 106–114. [Google Scholar] [CrossRef]
  41. Yakhot, V.; Orszag, S.A. Renormalization group analysis of turbulence. I. Basic theory. J. Sci. Comput. 19861, 3–51. [Google Scholar] [CrossRef]
  42. Yakhot, V.; Smith, L.M. The renormalization group, the e-expansion and derivation of turbulence models. J. Sci. Comput. 19927, 35–61. [Google Scholar] [CrossRef]
  43. Mastbergen, D.R.; Berg, J.V.D. Breaching in fine sands and the generation of sustained turbidity currents in submarine canyons. Sedimentology 200350, 625–637. [Google Scholar] [CrossRef]
  44. Soulsby, R. Dynamics of Marine Sands; Thomas Telford Ltd.: London, UK, 1998. [Google Scholar] [CrossRef]
  45. Van Rijn, L.C. Sediment Transport, Part I: Bed Load Transport. J. Hydraul. Eng. 1984110, 1431–1456. [Google Scholar] [CrossRef][Green Version]
  46. Zhang, Q.; Zhou, X.-L.; Wang, J.-H. Numerical investigation of local scour around three adjacent piles with different arrangements under current. Ocean Eng. 2017142, 625–638. [Google Scholar] [CrossRef]
  47. Yu, Y.X.; Liu, S.X. Random Wave and Its Applications to Engineering, 4th ed.; Dalian University of Technology Press: Dalian, China, 2011. [Google Scholar]
  48. Pang, A.; Skote, M.; Lim, S.; Gullman-Strand, J.; Morgan, N. A numerical approach for determining equilibrium scour depth around a mono-pile due to steady currents. Appl. Ocean Res. 201657, 114–124. [Google Scholar] [CrossRef]
  49. Higuera, P.; Lara, J.L.; Losada, I.J. Three-dimensional interaction of waves and porous coastal structures using Open-FOAM®. Part I: Formulation and validation. Coast. Eng. 201483, 243–258. [Google Scholar] [CrossRef]
  50. Corvaro, S.; Crivellini, A.; Marini, F.; Cimarelli, A.; Capitanelli, L.; Mancinelli, A. Experimental and Numerical Analysis of the Hydrodynamics around a Vertical Cylinder in Waves. J. Mar. Sci. Eng. 20197, 453. [Google Scholar] [CrossRef][Green Version]
  51. Flow3D User Manual, version 11.0.3; Flow Science, Inc.: Santa Fe, NM, USA, 2013.
  52. Khosronejad, A.; Kang, S.; Sotiropoulos, F. Experimental and computational investigation of local scour around bridge piers. Adv. Water Resour. 201237, 73–85. [Google Scholar] [CrossRef]
  53. Stahlmann, A. Experimental and Numerical Modeling of Scour at Foundation Structures for Offshore Wind Turbines. Ph.D. Thesis, Franzius-Institute for Hydraulic, Estuarine and Coastal Engineering, Leibniz Universität Hannover, Hannover, Germany, 2013. [Google Scholar]
  54. Breusers, H.N.C.; Nicollet, G.; Shen, H. Local Scour Around Cylindrical Piers. J. Hydraul. Res. 197715, 211–252. [Google Scholar] [CrossRef]
  55. Schendel, A.; Hildebrandt, A.; Goseberg, N.; Schlurmann, T. Processes and evolution of scour around a monopile induced by tidal currents. Coast. Eng. 2018139, 65–84. [Google Scholar] [CrossRef]
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

Share and Cite

      

MDPI and ACS Style

Hu, R.; Liu, H.; Leng, H.; Yu, P.; Wang, X. Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves. J. Mar. Sci. Eng. 20219, 886. https://doi.org/10.3390/jmse9080886

AMA Style

Hu R, Liu H, Leng H, Yu P, Wang X. Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves. Journal of Marine Science and Engineering. 2021; 9(8):886. https://doi.org/10.3390/jmse9080886Chicago/Turabian Style

Hu, Ruigeng, Hongjun Liu, Hao Leng, Peng Yu, and Xiuhai Wang. 2021. “Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves” Journal of Marine Science and Engineering 9, no. 8: 886. https://doi.org/10.3390/jmse9080886

Find Other Styles

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

For more information on the journal statistics, click here.

Multiple requests from the same IP address are counted as one view.

Figure 5. Schematic view of flap and support structure [32]

Design Optimization of Ocean Renewable Energy Converter Using a Combined Bi-level Metaheuristic Approach

결합된 Bi-level 메타휴리스틱 접근법을 사용한 해양 재생 에너지 변환기의 설계 최적화

Erfan Amini a1, Mahdieh Nasiri b1, Navid Salami Pargoo a, Zahra Mozhgani c, Danial Golbaz d, Mehrdad Baniesmaeil e, Meysam Majidi Nezhad f, Mehdi Neshat gj, Davide Astiaso Garcia h, Georgios Sylaios i

Abstract

In recent years, there has been an increasing interest in renewable energies in view of the fact that fossil fuels are the leading cause of catastrophic environmental consequences. Ocean wave energy is a renewable energy source that is particularly prevalent in coastal areas. Since many countries have tremendous potential to extract this type of energy, a number of researchers have sought to determine certain effective factors on wave converters’ performance, with a primary emphasis on ambient factors. In this study, we used metaheuristic optimization methods to investigate the effects of geometric factors on the performance of an Oscillating Surge Wave Energy Converter (OSWEC), in addition to the effects of hydrodynamic parameters. To do so, we used CATIA software to model different geometries which were then inserted into a numerical model developed in Flow3D software. A Ribed-surface design of the converter’s flap is also introduced in this study to maximize wave-converter interaction. Besides, a Bi-level Hill Climbing Multi-Verse Optimization (HCMVO) method was also developed for this application. The results showed that the converter performs better with greater wave heights, flap freeboard heights, and shorter wave periods. Additionally, the added ribs led to more wave-converter interaction and better performance, while the distance between the flap and flume bed negatively impacted the performance. Finally, tracking the changes in the five-dimensional objective function revealed the optimum value for each parameter in all scenarios. This is achieved by the newly developed optimization algorithm, which is much faster than other existing cutting-edge metaheuristic approaches.

Keywords

Wave Energy Converter

OSWEC

Hydrodynamic Effects

Geometric Design

Metaheuristic Optimization

Multi-Verse Optimizer

1Introduction

The increase in energy demand, the limitations of fossil fuels, as well as environmental crises, such as air pollution and global warming, are the leading causes of calling more attention to harvesting renewable energy recently [1][2][3]. While still in its infancy, ocean wave energy has neither reached commercial maturity nor technological convergence. In recent decades, remarkable progress has been made in the marine energy domain, which is still in the early stage of development, to improve the technology performance level (TPL) [4][5]and technology readiness level (TRL) of wave energy converters (WECs). This has been achieved using novel modeling techniques [6][7][8][9][10][11][12][13][14] to gain the following advantages [15]: (i) As a source of sustainable energy, it contributes to the mix of energy resources that leads to greater diversity and attractiveness for coastal cities and suppliers. [16] (ii) Since wave energy can be exploited offshore and does not require any land, in-land site selection would be less expensive and undesirable visual effects would be reduced. [17] (iii) When the best layout and location of offshore site are taken into account, permanent generation of energy will be feasible (as opposed to using solar energy, for example, which is time-dependent) [18].

In general, the energy conversion process can be divided into three stages in a WEC device, including primary, secondary, and tertiary stages [19][20]. In the first stage of energy conversion, which is the subject of this study, the wave power is converted to mechanical power by wave-structure interaction (WSI) between ocean waves and structures. Moreover, the mechanical power is transferred into electricity in the second stage, in which mechanical structures are coupled with power take-off systems (PTO). At this stage, optimal control strategies are useful to tune the system dynamics to maximize power output [10][13][12]. Furthermore, the tertiary energy conversion stage revolves around transferring the non-standard AC power into direct current (DC) power for energy storage or standard AC power for grid integration [21][22]. We discuss only the first stage regardless of the secondary and tertiary stages. While Page 1 of 16 WECs include several categories and technologies such as terminators, point absorbers, and attenuators [15][23], we focus on oscillating surge wave energy converters (OSWECs) in this paper due to its high capacity for industrialization [24].

Over the past two decades, a number of studies have been conducted to understand how OSWECs’ structures and interactions between ocean waves and flaps affect converters performance. Henry et al.’s experiment on oscillating surge wave energy converters is considered as one of the most influential pieces of research [25], which demonstrated how the performance of oscillating surge wave energy converters (OSWECs) is affected by seven different factors, including wave period, wave power, flap’s relative density, water depth, free-board of the flap, the gap between the tubes, gap underneath the flap, and flap width. These parameters were assessed in their two models in order to estimate the absorbed energy from incoming waves [26][27]. In addition, Folly et al. investigated the impact of water depth on the OSWECs performance analytically, numerically, and experimentally. According to this and further similar studies, the average annual incident wave power is significantly reduced by water depth. Based on the experimental results, both the surge wave force and the power capture of OSWECs increase in shallow water [28][29]. Following this, Sarkar et al. found that under such circumstances, the device that is located near the coast performs much better than those in the open ocean [30]. On the other hand, other studies are showing that the size of the converter, including height and width, is relatively independent of the location (within similar depth) [31]. Subsequently, Schmitt et al. studied OSWECs numerically and experimentally. In fact, for the simulation of OSWEC, OpenFOAM was used to test the applicability of Reynolds-averaged Navier-Stokes (RANS) solvers. Then, the experimental model reproduced the numerical results with satisfying accuracy [32]. In another influential study, Wang et al. numerically assessed the effect of OSWEC’s width on their performance. According to their findings, as converter width increases, its efficiency decreases in short wave periods while increases in long wave periods [33]. One of the main challenges in the analysis of the OSWEC is the coupled effect of hydrodynamic and geometric variables. As a result, numerous cutting-edge geometry studies have been performed in recent years in order to find the optimal structure that maximizes power output and minimizes costs. Garcia et al. reviewed hull geometry optimization studies in the literature in [19]. In addition, Guo and Ringwood surveyed geometric optimization methods to improve the hydrodynamic performance of OSWECs at the primary stage [14]. Besides, they classified the hull geometry of OSWECs based on Figure 1. Subsequently, Whittaker et al. proposed a different design of OSWEC called Oyster2. There have been three examples of different geometries of oysters with different water depths. Based on its water depth, they determined the width and height of the converter. They also found that in the constant wave period the less the converter’s width, the less power captures the converter has [34]. Afterward, O’Boyle et al. investigated a type of OSWEC called Oyster 800. They compared the experimental and numerical models with the prototype model. In order to precisely reproduce the shape, mass distribution, and buoyancy properties of the prototype, a 40th-scale experimental model has been designed. Overall, all the models were fairly accurate according to the results [35].

Inclusive analysis of recent research avenues in the area of flap geometry has revealed that the interaction-based designs of such converters are emerging as a novel approach. An initiative workflow is designed in the current study to maximizing the wave energy extrication by such systems. To begin with, a sensitivity analysis plays its role of determining the best hydrodynamic values for installing the converter’s flap. Then, all flap dimensions and characteristics come into play to finalize the primary model. Following, interactive designs is proposed to increase the influence of incident waves on the body by adding ribs on both sides of the flap as a novel design. Finally, a new bi-level metaheuristic method is proposed to consider the effects of simultaneous changes in ribs properties and other design parameters. We hope this novel approach will be utilized to make big-scale projects less costly and justifiable. The efficiency of the method is also compared with four well known metaheuristic algorithms and out weight them for this application.

This paper is organized as follows. First, the research methodology is introduced by providing details about the numerical model implementation. To that end, we first introduced the primary model’s geometry and software details. That primary model is later verified with a benchmark study with regard to the flap angle of rotation and water surface elevation. Then, governing equations and performance criteria are presented. In the third part of the paper, we discuss the model’s sensitivity to lower and upper parts width (we proposed a two cross-sectional design for the flap), bottom elevation, and freeboard. Finally, the novel optimization approach is introduced in the final part and compared with four recent metaheuristic algorithms.

2. Numerical Methods

In this section, after a brief introduction of the numerical software, Flow3D, boundary conditions are defined. Afterwards, the numerical model implementation, along with primary model properties are described. Finally, governing equations, as part of numerical process, are discussed.

2.1Model Setup

FLOW-3D is a powerful and comprehensive CFD simulation platform for studying fluid dynamics. This software has several modules to solve many complex engineering problems. In addition, modeling complex flows is simple and effective using FLOW-3D’s robust meshing capabilities [36]. Interaction between fluid and moving objects might alter the computational range. Dynamic meshes are used in our modeling to take these changes into account. At each time step, the computational node positions change in order to adapt the meshing area to the moving object. In addition, to choose mesh dimensions, some factors are taken into account such as computational accuracy, computational time, and stability. The final grid size is selected based on the detailed procedure provided in [37]. To that end, we performed grid-independence testing on a CFD model using three different mesh grid sizes of 0.01, 0.015, and 0.02 meters. The problem geometry and boundary conditions were defined the same, and simulations were run on all three grids under the same conditions. The predicted values of the relevant variable, such as velocity, was compared between the grids. The convergence behavior of the numerical solution was analyzed by calculating the relative L2 norm error between two consecutive grids. Based on the results obtained, it was found that the grid size of 0.02 meters showed the least error, indicating that it provided the most accurate and reliable solution among the three grids. Therefore, the grid size of 0.02 meters was selected as the optimal spatial resolution for the mesh grid.

In this work, the flume dimensions are 10 meters long, 0.1 meters wide, and 2.2 meters high, which are shown in figure2. In addition, input waves with linear characteristics have a height of 0.1 meters and a period of 1.4 seconds. Among the linear wave methods included in this software, RNGk-ε and k- ε are appropriate for turbulence model. The research of Lopez et al. shows that RNGk- ε provides the most accurate simulation of turbulence in OSWECs [21]. We use CATIA software to create the flap primary model and other innovative designs for this project. The flap measures 0.1 m x 0.65 m x 0.360 m in x, y and z directions, respectively. In Figure 3, the primary model of flap and its dimensions are shown. In this simulation, five boundaries have been defined, including 1. Inlet, 2. Outlet, 3. Converter flap, 4. Bed flume, and 5. Water surface, which are shown in figure 2. Besides, to avoid wave reflection in inlet and outlet zones, Flow3D is capable of defining some areas as damping zones, the length of which has to be one to one and a half times the wavelength. Therefore, in the model, this length is considered equal to 2 meters. Furthermore, there is no slip in all the boundaries. In other words, at every single time step, the fluid velocity is zero on the bed flume, while it is equal to the flap velocity on the converter flap. According to the wave theory defined in the software, at the inlet boundary, the water velocity is called from the wave speed to be fed into the model.

2.2Verification

In the current study, we utilize the Schmitt experimental model as a benchmark for verification, which was developed at the Queen’s University of Belfast. The experiments were conducted on the flap of the converter, its rotation, and its interaction with the water surface. Thus, the details of the experiments are presented below based up on the experimental setup’s description [38]. In the experiment, the laboratory flume has a length of 20m and a width of 4.58m. Besides, in order to avoid incident wave reflection, a wave absorption source is devised at the end of the left flume. The flume bed, also, includes two parts with different slops. The flap position and dimensions of the flume can be seen in Figure4. In addition, a wave-maker with 6 paddles is installed at one end. At the opposite end, there is a beach with wire meshes. Additionally, there are 6 indicators to extract the water level elevation. In the flap model, there are three components: the fixed support structure, the hinge, and the flap. The flap measures 0.1m x 0.65m x 0.341m in x, y and z directions, respectively. In Figure5, the details are given [32]. The support structure consists of a 15 mm thick stainless steel base plate measuring 1m by 1.4m, which is screwed onto the bottom of the tank. The hinge is supported by three bearing blocks. There is a foam centerpiece on the front and back of the flap which is sandwiched between two PVC plates. Enabling changes of the flap, three metal fittings link the flap to the hinge. Moreover, in this experiment, the selected wave is generated based on sea wave data at scale 1:40. The wave height and the wave period are equal to 0.038 (m) and 2.0625 (s), respectively, which are tantamount to a wave with a period of 13 (s) and a height of 1.5 (m).

Two distinct graphs illustrate the numerical and experi-mental study results. Figure6 and Figure7 are denoting the angle of rotation of flap and surface elevation in computational and experimental models, respectively. The two figures roughly represent that the numerical and experimental models are a good match. However, for the purpose of verifying the match, we calculated the correlation coefficient (C) and root mean square error (RMSE). According to Figure6, correlation coefficient and RMSE are 0.998 and 0.003, respectively, and in Figure7 correlation coefficient and RMSE are respectively 0.999 and 0.001. Accordingly, there is a good match between the numerical and empirical models. It is worth mentioning that the small differences between the numerical and experimental outputs may be due to the error of the measuring devices and the calibration of the data collection devices.

Including continuity equation and momentum conserva- tion for incompressible fluid are given as [32][39]:(1)

where P represents the pressure, g denotes gravitational acceleration, u represents fluid velocity, and Di is damping coefficient. Likewise, the model uses the same equation. to calculate the fluid velocity in other directions as well. Considering the turbulence, we use the two-equation model of RNGK- ε. These equations are:

(3)��t(��)+����(����)=����[�eff�������]+��-��and(4)���(��)+����(����)=����[�eff�������]+�1�∗����-��2��2�Where �2� and �1� are constants. In addition, �� and �� represent the turbulent Prandtl number of � and k, respectively.

�� also denote the production of turbulent kinetic energy of k under the effect of velocity gradient, which is calculated as follows:(5)��=�eff[�����+�����]�����(6)�eff=�+��(7)�eff=�+��where � is molecular viscosity,�� represents turbulence viscosity, k denotes kinetic energy, and ∊∊ is energy dissipation rate. The values of constant coefficients in the two-equation RNGK ∊-∊ model is as shown in the Table 1 [40].Table 2.

Table 1. Constant coefficients in RNGK- model

Factors�0�1�2������
Quantity0.0124.381.421.681.391.390.084

Table 2. Flap properties

Joint height (m)0.476
Height of the center of mass (m)0.53
Weight (Kg)10.77

It is worth mentioning that the volume of fluid method is used to separate water and air phases in this software [41]. Below is the equation of this method [40].(8)����+����(���)=0where α and 1 − α are portion of water phase and air phase, respectively. As a weighting factor, each fluid phase portion is used to determine the mixture properties. Finally, using the following equations, we calculate the efficiency of converters [42][34][43]:(9)�=14|�|2�+�2+(�+�a)2(�n2-�2)2where �� represents natural frequency, I denotes the inertia of OSWEC, Ia is the added inertia, F is the complex wave force, and B denotes the hydrodynamic damping coefficient. Afterward, the capture factor of the converter is calculated by [44]:(10)��=�1/2��2����gw where �� represents the capture factor, which is the total efficiency of device per unit length of the wave crest at each time step [15], �� represent the dimensional amplitude of the incident wave, w is the flap’s width, and Cg is the group velocity of the incident wave, as below:(11)��=��0·121+2�0ℎsinh2�0ℎwhere �0 denotes the wave number, h is water depth, and H is the height of incident waves.

According to previous sections ∊,����-∊ modeling is used for all models simulated in this section. For this purpose, the empty boundary condition is used for flume walls. In order to preventing wave reflection at the inlet and outlet of the flume, the length of wave absorption is set to be at least one incident wavelength. In addition, the structured mesh is chosen, and the mesh dimensions are selected in two distinct directions. In each model, all grids have a length of 2 (cm) and a height of 1 (cm). Afterwards, as an input of the software for all of the models, we define the time step as 0.001 (s). Moreover, the run time of every simulation is 30 (s). As mentioned before, our primary model is Schmitt model, and the flap properties is given in table2. For all simulations, the flume measures 15 meters in length and 0.65 meters in width, and water depth is equal to 0.335 (m). The flap is also located 7 meters from the flume’s inlet.

Finally, in order to compare the results, the capture factor is calculated for each simulation and compared to the primary model. It is worth mentioning that capture factor refers to the ratio of absorbed wave energy to the input wave energy.

According to primary model simulation and due to the decreasing horizontal velocity with depth, the wave crest has the highest velocity. Considering the fact that the wave’s orbital velocity causes the flap to move, the contact between the upper edge of the flap and the incident wave can enhance its performance. Additionally, the numerical model shows that the dynamic pressure decreases as depth increases, and the hydrostatic pressure increases as depth increases.

To determine the OSWEC design, it is imperative to understand the correlation between the capture factor, wave period, and wave height. Therefore, as it is shown in Figure8, we plot the change in capture factor over the variations in wave period and wave height in 3D and 2D. In this diagram, the first axis features changes in wave period, the second axis displays changes in wave height, and the third axis depicts changes in capture factor. According to our wave properties in the numerical model, the wave period and wave height range from 2 to 14 seconds and 2 to 8 meters, respectively. This is due to the fact that the flap does not oscillate if the wave height is less than 2 (m), and it does not reverse if the wave height is more than 8 (m). In addition, with wave periods more than 14 (s), the wavelength would be so long that it would violate the deep-water conditions, and with wave periods less than 2 (s), the flap would not oscillate properly due to the shortness of wavelength. The results of simulation are shown in Figure 8. As it can be perceived from Figure 8, in a constant wave period, the capture factor is in direct proportion to the wave height. It is because of the fact that waves with more height have more energy to rotate the flap. Besides, in a constant wave height, the capture factor increases when the wave period increases, until a given wave period value. However, the capture factor falls after this point. These results are expected since the flap’s angular displacement is not high in lower wave periods, while the oscillating motion of that is not fast enough to activate the power take-off system in very high wave periods.

As is shown in Figure 9, we plot the change in capture factor over the variations in wave period (s) and water depth (m) in 3D. As it can be seen in this diagram, the first axis features changes in water depth (m), the second axis depicts the wave period (s), and the third axis displays OSWEC’s capture factor. The wave period ranges from 0 to 10 seconds based on our wave properties, which have been adopted from Schmitt’s model, while water depth ranges from 0 to 0.5 meters according to the flume and flap dimensions and laboratory limitations. According to Figure9, for any specific water depth, the capture factor increases in a varying rate when the wave period increases, until a given wave period value. However, the capture factor falls steadily after this point. In fact, the maximum capture factor occurs when the wave period is around 6 seconds. This trend is expected since, in a specific water depth, the flap cannot oscillate properly when the wavelength is too short. As the wave period increases, the flap can oscillate more easily, and consequently its capture factor increases. However, the capture factor drops in higher wave periods because the wavelength is too large to move the flap. Furthermore, in a constant wave period, by changing the water depth, the capture factor does not alter. In other words, the capture factor does not depend on the water depth when it is around its maximum value.

3Sensitivity Analysis

Based on previous studies, in addition to the flap design, the location of the flap relative to the water surface (freeboard) and its elevation relative to the flume bed (flap bottom elevation) play a significant role in extracting energy from the wave energy converter. This study measures the sensitivity of the model to various parameters related to the flap design including upper part width of the flap, lower part width of the flap, the freeboard, and the flap bottom elevation. Moreover, as a novel idea, we propose that the flap widths differ in the lower and upper parts. In Figure10, as an example, a flap with an upper thickness of 100 (mm) and a lower thickness of 50 (mm) and a flap with an upper thickness of 50 (mm) and a lower thickness of 100 (mm) are shown. The influence of such discrepancy between the widths of the upper and lower parts on the interaction between the wave and the flap, or in other words on the capture factor, is evaluated. To do so, other parameters are remained constant, such as the freeboard, the distance between the flap and the flume bed, and the wave properties.

In Figure11, models are simulated with distinct upper and lower widths. As it is clear in this figure, the first axis depicts the lower part width of the flap, the second axis indicates the upper part width of the flap, and the colors represent the capture factor values. Additionally, in order to consider a sufficient range of change, the flap thickness varies from half to double the value of the primary model for each part.

According to this study, the greater the discrepancy in these two parts, the lower the capture factor. It is on account of the fact that when the lower part of the flap is thicker than the upper part, and this thickness difference in these two parts is extremely conspicuous, the inertia against the motion is significant at zero degrees of rotation. Consequently, it is difficult to move the flap, which results in a low capture factor. Similarly, when the upper part of the flap is thicker than the lower part, and this thickness difference in these two parts is exceedingly noticeable, the inertia is so great that the flap can not reverse at the maximum degree of rotation. As the results indicate, the discrepancy can enhance the performance of the converter if the difference between these two parts is around 20%. As it is depicted in the Figure11, the capture factor reaches its own maximum amount, when the lower part thickness is from 5 to 6 (cm), and the upper part thickness is between 6 and 7 (cm). Consequently, as a result of this discrepancy, less material will be used, and therefore there will be less cost.

As illustrated in Figure12, this study examines the effects of freeboard (level difference between the flap top and water surface) and the flap bottom elevation (the distance between the flume bed and flap bottom) on the converter performance. In this diagram, the first axis demonstrates the freeboard and the second axis on the left side displays the flap bottom elevation, while the colors indicate the capture factor. In addition, the feasible range of freeboard is between -15 to 15 (cm) due to the limitation of the numerical model, so that we can take the wave slamming and the overtopping into consideration. Additionally, based on the Schmitt model and its scaled model of 1:40 of the base height, the flap bottom should be at least 9 (cm) high. Since the effect of surface waves is distributed over the depth of the flume, it is imperative to maintain a reasonable flap height exposed to incoming waves. Thus, the maximum flap bottom elevation is limited to 19 (cm). As the Figure12 pictures, at constant negative values of the freeboard, the capture factor is in inverse proportion with the flap bottom elevation, although slightly.

Furthermore, at constant positive values of the freeboard, the capture factor fluctuates as the flap bottom elevation decreases while it maintains an overall increasing trend. This is on account of the fact that increasing the flap bottom elevation creates turbulence flow behind the flap, which encumbers its rotation, as well as the fact that the flap surface has less interaction with the incoming waves. Furthermore, while keeping the flap bottom elevation constant, the capture factor increases by raising the freeboard. This is due to the fact that there is overtopping with adverse impacts on the converter performance when the freeboard is negative and the flap is under the water surface. Besides, increasing the freeboard makes the wave slam more vigorously, which improves the converter performance.

Adding ribs to the flap surface, as shown in Figure13, is a novel idea that is investigated in the next section. To achieve an optimized design for the proposed geometry of the flap, we determine the optimal number and dimensions of ribs based on the flap properties as our decision variables in the optimization process. As an example, Figure13 illustrates a flap with 3 ribs on each side with specific dimensions.

Figure14 shows the flow velocity field around the flap jointed to the flume bed. During the oscillation of the flap, the pressure on the upper and lower surfaces of the flap changes dynamically due to the changing angle of attack and the resulting change in the direction of fluid flow. As the flap moves upwards, the pressure on the upper surface decreases, and the pressure on the lower surface increases. Conversely, as the flap moves downwards, the pressure on the upper surface increases, and the pressure on the lower surface decreases. This results in a cyclic pressure variation around the flap. Under certain conditions, the pressure field around the flap can exhibit significant variations in magnitude and direction, forming vortices and other flow structures. These flow structures can affect the performance of the OSWEC by altering the lift and drag forces acting on the flap.

4Design Optimization

We consider optimizing the design parameters of the flap of converter using a nature-based swarm optimization method, that fall in the category of metaheuristic algorithms [45]. Accordingly, we choose four state-of-the-art algorithms to perform an optimization study. Then, based on their performances to achieve the highest capture factor, one of them will be chosen to be combined with the Hill Climb algorithm to carry out a local search. Therefore, in the remainder of this section, we discuss the search process of each algorithm and visualize their performance and convergence curve as they try to find the best values for decision variables.

4.1. Metaheuristic Approaches

As the first considered algorithm, the Gray Wolf Optimizer (GWO) algorithm simulates the natural leadership and hunting performance of gray wolves which tend to live in colonies. Hunters must obey the alpha wolf, the leader, who is responsible for hunting. Then, the beta wolf is at the second level of the gray wolf hierarchy. A subordinate of alpha wolf, beta stands under the command of the alpha. At the next level in this hierarchy, there are the delta wolves. They are subordinate to the alpha and beta wolves. This category of wolves includes scouts, sentinels, elders, hunters, and caretakers. In this ranking, omega wolves are at the bottom, having the lowest level and obeying all other wolves. They are also allowed to eat the prey just after others have eaten. Despite the fact that they seem less important than others, they are really central to the pack survival. Since, it has been shown that without omega wolves, the entire pack would experience some problems like fighting, violence, and frustration. In this simulation, there are three primary steps of hunting including searching, surrounding, and finally attacking the prey. Mathematically model of gray wolves’ hunting technique and their social hierarchy are applied in determined by optimization. this study. As mentioned before, gray wolves can locate their prey and surround them. The alpha wolf also leads the hunt. Assuming that the alpha, beta, and delta have more knowledge about prey locations, we can mathematically simulate gray wolf hunting behavior. Hence, in addition to saving the top three best solutions obtained so far, we compel the rest of the search agents (also the omegas) to adjust their positions based on the best search agent. Encircling behavior can be mathematically modeled by the following equations: [46].(12)�→=|�→·��→(�)-�→(�)|(13)�→(�+1)=��→(�)-�→·�→(14)�→=2.�2→(15)�→=2�→·�1→-�→Where �→indicates the position vector of gray wolf, ��→ defines the vector of prey, t indicates the current iteration, and �→and �→are coefficient vectors. To force the search agent to diverge from the prey, we use �→ with random values greater than 1 or less than -1. In addition, C→ contains random values in the range [0,2], and �→ 1 and �2→ are random vectors in [0,1]. The second considered technique is the Moth Flame Optimizer (MFO) algorithm. This method revolves around the moths’ navigation mechanism, which is realized by positioning themselves and maintaining a fixed angle relative to the moon while flying. This effective mechanism helps moths to fly in a straight path. However, when the source of light is artificial, maintaining an angle with the light leads to a spiral flying path towards the source that causes the moth’s death [47]. In MFO algorithm, moths and flames are both solutions. The moths are actual search agents that fly in hyper-dimensional space by changing their position vectors, and the flames are considered pins that moths drop when searching the search space [48]. The problem’s variables are the position of moths in the space. Each moth searches around a flame and updates it in case of finding a better solution. The fitness value is the return value of each moth’s fitness (objective) function. The position vector of each moth is passed to the fitness function, and the output of the fitness function is assigned to the corresponding moth. With this mechanism, a moth never loses its best solution [49]. Some attributes of this algorithm are as follows:

  • •It takes different values to converge moth in any point around the flame.
  • •Distance to the flame is lowered to be eventually minimized.
  • •When the position gets closer to the flame, the updated positions around the flame become more frequent.

As another method, the Multi-Verse Optimizer is based on a multiverse theory which proposes there are other universes besides the one in which we all live. According to this theory, there are more than one big bang in the universe, and each big bang leads to the birth of a new universe [50]. Multi-Verse Optimizer (MVO) is mainly inspired by three phenomena in cosmology: white holes, black holes, and wormholes. A white hole has never been observed in our universe, but physicists believe the big bang could be considered a white hole [51]. Black holes, which behave completely in contrast to white holes, attract everything including light beams with their extremely high gravitational force [52]. In the multiverse theory, wormholes are time and space tunnels that allow objects to move instantly between any two corners of a universe (or even simultaneously from one universe to another) [53]. Based on these three concepts, mathematical models are designed to perform exploration, exploitation, and local search, respectively. The concept of white and black holes is implied as an exploration phase, while the concept of wormholes is considered as an exploitation phase by MVO. Additionally, each solution is analogous to a universe, and each variable in the solution represents an object in that universe. Furthermore, each solution is assigned an inflation rate, and the time is used instead of iterations. Following are the universe rules in MVO:

  • •The possibility of having white hole increases with the inflation rate.
  • •The possibility of having black hole decreases with the inflation rate.
  • •Objects tend to pass through black holes more frequently in universes with lower inflation rates.
  • •Regardless of inflation rate, wormholes may cause objects in universes to move randomly towards the best universe. [54]

Modeling the white/black hole tunnels and exchanging objects of universes mathematically was accomplished by using the roulette wheel mechanism. With every iteration, the universes are sorted according to their inflation rates, then, based on the roulette wheel, the one with the white hole is selected as the local extremum solution. This is accomplished through the following steps:

Assume that

(16)���=����1<��(��)����1≥��(��)

Where ��� represents the jth parameter of the ith universe, Ui indicates the ith universe, NI(Ui) is normalized inflation rate of the ith universe, r1 is a random number in [0,1], and j xk shows the jth parameter of the kth universe selected by a roulette wheel selection mechanism [54]. It is assumed that wormhole tunnels always exist between a universe and the best universe formed so far. This mechanism is as follows:(17)���=if�2<���:��+���×((���-���)×�4+���)�3<0.5��-���×((���-���)×�4+���)�3≥0.5����:���where Xj indicates the jth parameter of the best universe formed so far, TDR and WEP are coefficients, where Xj indicates the jth parameter of the best universelbjshows the lower bound of the jth variable, ubj is the upper bound of the jth variable, and r2, r3, and r4 are random numbers in [1][54].

Finally, one of the newest optimization algorithms is WOA. The WOA algorithm simulates the movement of prey and the whale’s discipline when looking for their prey. Among several species, Humpback whales have a specific method of hunting [55]. Humpback whales can recognize the location of prey and encircle it before hunting. The optimal design position in the search space is not known a priori, and the WOA algorithm assumes that the best candidate solution is either the target prey or close to the optimum. This foraging behavior is called the bubble-net feeding method. Two maneuvers are associated with bubbles: upward spirals and double loops. A unique behavior exhibited only by humpback whales is bubble-net feeding. In fact, The WOA algorithm starts with a set of random solutions. At each iteration, search agents update their positions for either a randomly chosen search agent or the best solution obtained so far [56][55]. When the best search agent is determined, the other search agents will attempt to update their positions toward that agent. It is important to note that humpback whales swim around their prey simultaneously in a circular, shrinking circle and along a spiral-shaped path. By using a mathematical model, the spiral bubble-net feeding maneuver is optimized. The following equation represents this behavior:(18)�→(�+1)=�′→·�bl·cos(2��)+�∗→(�)

Where:(19)�′→=|�∗→(�)-�→(�)|

X→(t+ 1) indicates the distance of the it h whale to the prey (best solution obtained so far),� is a constant for defining the shape of the logarithmic spiral, l is a random number in [−1, 1], and dot (.) is an element-by-element multiplication [55].

Comparing the four above-mentioned methods, simulations are run with 10 search agents for 400 iterations. In Figure 15, there are 20 plots the optimal values of different parameters in optimization algorithms. The five parameters of this study are freeboard, bottom elevations, number of ribs on the converter, rib thickness, and rib Height. The optimal value for each was found by optimization algorithms, naming WOA, MVO, MFO, and GWO. By looking through the first row, the freeboard parameter converges to its maximum possible value in the optimization process of GWO after 300 iterations. Similarly, MFO finds the same result as GWO. In contrast, the freeboard converges to its minimum possible value in MVO optimizing process, which indicates positioning the converter under the water. Furthermore, WOA found the optimal value of freeboard as around 0.02 after almost 200 iterations. In the second row, the bottom elevation is found at almost 0.11 (m) in all algorithms; however, the curves follow different trends in each algorithm. The third row shows the number of ribs, where results immediately reveal that it should be over 4. All algorithms coincide at 5 ribs as the optimal number in this process. The fourth row displays the trends of algorithms to find optimal rib thickness. MFO finds the optimal value early and sets it to around 0.022, while others find the same value in higher iterations. Finally, regarding the rib height, MVO, MFO, and GWO state that the optimal value is 0.06 meters, but WOA did not find a higher value than 0.039.

4.2. HCMVO Bi-level Approach

Despite several strong search characteristics of MVO and its high performance in various optimization problems, it suffers from a few deficiencies in local and global search mechanisms. For instance, it is trapped in the local optimum when wormholes stochastically generate many solutions near the best universe achieved throughout iterations, especially in solving complex multimodal problems with high dimensions [57]. Furthermore, MVO needs to be modified by an escaping strategy from the local optima to enhance the global search abilities. To address these shortages, we propose a fast and effective meta-algorithm (HCMVO) to combine MVO with a Random-restart hill-climbing local search. This meta-algorithm uses MVO on the upper level to develop global tracking and provide a range of feasible and proper solutions. The hill-climbing algorithm is designed to develop a comprehensive neighborhood search around the best-found solution proposed by the upper-level (MVO) when MVO is faced with a stagnation issue or falling into a local optimum. The performance threshold is formulated as follows.(20)Δ����THD=∑�=1�����TH��-����TH��-1�where BestTHDis the best-found solution per generation, andM is related to the domain of iterations to compute the average performance of MVO. If the proposed best solution by the local search is better than the initial one, the global best of MVO will be updated. HCMVO iteratively runs hill climbing when the performance of MVO goes down, each time with an initial condition to prepare for escaping such undesirable situations. In order to get a better balance between exploration and exploitation, the search step size linearly decreases as follows:(21)��=��-����Ma�iter��+1where iter and Maxiter are the current iteration and maximum number of evaluation, respectively. �� stands for the step size of the neighborhood search. Meanwhile, this strategy can improve the convergence rate of MVO compared with other algorithms.

Algorithm 1 shows the technical details of the proposed optimization method (HCMVO). The initial solution includes freeboard (�), bottom elevation (�), number of ribs (Nr), rib thickness (�), and rib height(�).

5. Conclusion

The high trend of diminishing worldwide energy resources has entailed a great crisis upon vulnerable societies. To withstand this effect, developing renewable energy technologies can open doors to a more reliable means, among which the wave energy converters will help the coastal residents and infrastructure. This paper set out to determine the optimized design for such devices that leads to the highest possible power output. The main goal of this research was to demonstrate the best design for an oscillating surge wave energy converter using a novel metaheuristic optimization algorithm. In this regard, the methodology was devised such that it argued the effects of influential parameters, including wave characteristics, WEC design, and interaction criteria.

To begin with, a numerical model was developed in Flow 3D software to simulate the response of the flap of a wave energy converter to incoming waves, followed by a validation study based upon a well-reputed experimental study to verify the accuracy of the model. Secondly, the hydrodynamics of the flap was investigated by incorporating the turbulence. The effect of depth, wave height, and wave period are also investigated in this part. The influence of two novel ideas on increasing the wave-converter interaction was then assessed: i) designing a flap with different widths in the upper and lower part, and ii) adding ribs on the surface of the flap. Finally, four trending single-objective metaheuristic optimization methods

Empty CellAlgorithm 1: Hill Climb Multiverse Optimization
01:procedure HCMVO
02:�=30,�=5▹���������������������������������
03:�=〈F1,B1,N,R,H1〉,…〈FN,B2,N,R,HN〉⇒lb1N⩽�⩽ubN
04:Initialize parameters�ER,�DR,�EP,Best�,���ite��▹Wormhole existence probability (WEP)
05:��=����(��)
06:��=Normalize the inflation rate��
07:for iter in[1,⋯,���iter]do
08:for�in[1,⋯,�]do
09:Update�EP,�DR,Black����Index=�
10:for���[1,⋯,�]��
11:�1=����()
12:if�1≤��(��)then
13:White HoleIndex=Roulette�heelSelection(-��)
14:�(Black HoleIndex,�)=��(White HoleIndex,�)
15:end if
16:�2=����([0,�])
17:if�2≤�EPthen
18:�3=����(),�4=����()
19:if�3<0.5then
20:�1=((��(�)-��(�))�4+��(�))
21:�(�,�)=Best�(�)+�DR�
22:else
23:�(�,�)=Best�(�)-�DR�
24:end if
25:end if
26:end for
27:end for
28:�HD=����([�1,�2,⋯,�Np])
29:Bes�TH�itr=����HD
30:ΔBestTHD=∑�=1�BestTII��-BestTII��-1�
31:ifΔBestTHD<��then▹Perform hill climbing local search
32:BestTHD=����-�lim��������THD
33:end if
34:end for
35:return�,BestTHD▹Final configuration
36:end procedure

The implementation details of the hill-climbing algorithm applied in HCMPA can be seen in Algorithm 2. One of the critical parameters isg, which denotes the resolution of the neighborhood search around the proposed global best by MVO. If we set a small step size for hill-climbing, the convergence speed will be decreased. On the other hand, a large step size reinforces the exploration ability. Still, it may reduce the exploitation ability and in return increase the act of jumping from a global optimum or surfaces with high-potential solutions. Per each decision variable, the neighborhood search evaluates two different direct searches, incremental or decremental. After assessing the generated solutions, the best candidate will be selected to iterate the search algorithm. It is noted that the hill-climbing algorithm should not be applied in the initial iteration of the optimization process due to the immense tendency for converging to local optima. Meanwhile, for optimizing largescale problems, hill-climbing is not an appropriate selection. In order to improve understanding of the proposed hybrid optimization algorithm’s steps, the flowchart of HCMVO is designed and can be seen in Figure 16.

Figure 17 shows the observed capture factor (which is the absorbed energy with respect to the available energy) by each optimization algorithm from iterations 1 to 400. The algorithms use ten search agents in their modified codes to find the optimal solutions. While GWO and MFO remain roughly constant after iterations 54 and 40, the other three algorithms keep improving the capture factor. In this case, HCMVO and MVO worked very well in the optimizing process with a capture factor obtained by the former as 0.594 and by the latter as 0.593. MFO almost found its highest value before the iteration 50, which means the exploration part of the algorithm works out well. Similarly, HCMVO does the same. However, it keeps finding the better solution during the optimization process until the last iteration, indicating the strong exploitation part of the algorithm. GWO reveals a weakness in exploration and exploitation because not only does it evoke the least capture factor value, but also the curve remains almost unchanged throughout 350 iterations.

Figure 18 illustrates complex interactions between the five optimization parameters and the capture factor for HCMVO (a), MPA (b), and MFO (c) algorithms. The first interesting observation is that there is a high level of nonlinear relationships among the setting parameters that can make a multi-modal search space. The dark blue lines represent the best-found configuration throughout the optimisation process. Based on both HCMVO (a) and MVO (b), we can infer that the dark blue lines concentrate in a specific range, showing the high convergence ability of both HCMVO and MVO. However, MFO (c) could not find the exact optimal range of the decision variables, and the best-found solutions per generation distribute mostly all around the search space.

Empty CellAlgorithm 1: Hill Climb Multiverse Optimization
01:procedure HCMVO
02:Initialization
03:Initialize the constraints��1�,��1�
04:�1�=Mi�1�+���1�/�▹Compute the step size,�is search resolution
05:So�1=〈�,�,�,�,�〉▹���������������
06:�������1=����So�1▹���������ℎ���������
07:Main loop
08:for iter≤���ita=do
09:���=���±��
10:while�≤���(Sol1)do
11:���=���+�,▹����ℎ���ℎ��������ℎ
12:fitness��iter=�������
13:t = t+1
14:end while
15:〈�����,������max〉=����������
16:���itev=���Inde�max▹�������ℎ�������������������������������ℎ�������
17:��=��-����Max��+1▹�����������������
18:end for
19:return���iter,����
20:end procedure

were utilized to illuminate the optimum values of the design parameters, and the best method was chosen to develop a new algorithm that performs both local and global search methods.

The correlation between hydrodynamic parameters and the capture factor of the converter was supported by the results. For any given water depth, the capture factor increases as the wave period increases, until a certain wave period value (6 seconds) is reached, after which the capture factor gradually decreases. It is expected since the flap cannot oscillate effectively when the wavelength is too short for a certain water depth. Conversely, when the wavelength is too long, the capture factor decreases. Furthermore, under a constant wave period, increasing the water depth does not affect the capture factor. Regarding the sensitivity analysis, the study found that increasing the flap bottom elevation causes turbulence flow behind the flap and limitation of rotation, which leads to less interaction with the incoming waves. Furthermore, while keeping the flap bottom elevation constant, increasing the freeboard improves the capture factor. Overtopping happens when the freeboard is negative and the flap is below the water surface, which has a detrimental influence on converter performance. Furthermore, raising the freeboard causes the wave impact to become more violent, which increases converter performance.

In the last part, we discussed the search process of each algorithm and visualized their performance and convergence curves as they try to find the best values for decision variables. Among the four selected metaheuristic algorithms, the Multi-verse Optimizer proved to be the most effective in achieving the best answer in terms of the WEC capture factor. However, the MVO needed modifications regarding its escape approach from the local optima in order to improve its global search capabilities. To overcome these constraints, we presented a fast and efficient meta-algorithm (HCMVO) that combines MVO with a Random-restart hill-climbing local search. On a higher level, this meta-algorithm employed MVO to generate global tracking and present a range of possible and appropriate solutions. Taken together, the results demonstrated that there is a significant degree of nonlinearity among the setup parameters that might result in a multimodal search space. Since MVO was faced with a stagnation issue or fell into a local optimum, we constructed a complete neighborhood search around the best-found solution offered by the upper level. In sum, the newly-developed algorithm proved to be highly effective for the problem compared to other similar optimization methods. The strength of the current findings may encourage future investigation on design optimization of wave energy converters using developed geometry as well as the novel approach.

CRediT authorship contribution statement

Erfan Amini: Conceptualization, Methodology, Validation, Data curation, Writing – original draft, Writing – review & editing, Visualization. Mahdieh Nasiri: Conceptualization, Methodology, Validation, Data curation, Writing – original draft, Writing – review & editing, Visualization. Navid Salami Pargoo: Writing – original draft, Writing – review & editing. Zahra Mozhgani: Conceptualization, Methodology. Danial Golbaz: Writing – original draft. Mehrdad Baniesmaeil: Writing – original draft. Meysam Majidi Nezhad: . Mehdi Neshat: Supervision, Conceptualization, Writing – original draft, Writing – review & editing, Visualization. Davide Astiaso Garcia: Supervision. Georgios Sylaios: Supervision.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

This research has been carried out within ILIAD (Inte-grated Digital Framework for Comprehensive Maritime Data and Information Services) project that received funding from the European Union’s H2020 programme.

Data availability

Data will be made available on request.

References

Fig. 1. A typical Boiling Water Reactor (BWR) and selected segment of study for simulation

Understanding dry-out mechanism in rod bundles of boiling water reactor

끓는 물 원자로 봉 다발의 건조 메커니즘 이해

Liril D.SilviaDinesh K.ChandrakercSumanaGhoshaArup KDasb
aDepartment of Chemical Engineering, Indian Institute of Technology, Roorkee, India
bDepartment of Mechanical Engineering, Indian Institute of Technology, Roorkee, India
cReactor Engineering Division, Bhabha Atomic Research Centre, Mumbai, India

Abstract

Present work reports numerical understanding of interfacial dynamics during co-flow of vapor and liquid phases of water inside a typical Boiling Water Reactor (BWR), consisting of a nuclear fuel rod bundle assembly of 7 pins in a circular array. Two representative spacings between rods in a circular array are used to carry out the simulation. In literature, flow boiling in a nuclear reactor is dealt with mechanistic models or averaged equations. Hence, in the present study using the Volume of Fluid (VOF) based multiphase model, a detailed numerical understanding of breaking and making in interfaces during flow boiling in BWR is targeted. Our work will portray near realistic vapor bubble and liquid flow dynamics in rod bundle scenario. Constant wall heat flux for fuel rod and uniform velocity of the liquid at the inlet patch is applied as a boundary condition. The saturation properties of water are taken at 30 bar pressure. Flow boiling stages involving bubble nucleation, growth, merging, local dry-out, rewetting with liquid patches, and complete dry-out are illustrated. The dry-out phenomenon with no liquid presence is numerically observed with phase fraction contours at various axial cut-sections. The quantification of the liquid phase fraction at different axial planes is plotted over time, emphasizing the progressive dry-out mechanism. A comparison of liquid-vapor distribution for inner and outer rods reveals that the inner rod’s dry-out occurs sooner than that of the outer rod. The heat transfer coefficient to identify the heat dissipation capacity of each case is also reported.

현재 작업은 원형 배열에 있는 7개의 핀으로 구성된 핵연료봉 다발 어셈블리로 구성된 일반적인 끓는 물 원자로(BWR) 내부의 물의 증기 및 액체상의 동시 흐름 동안 계면 역학에 대한 수치적 이해를 보고합니다.

원형 배열의 막대 사이에 두 개의 대표적인 간격이 시뮬레이션을 수행하는 데 사용됩니다. 문헌에서 원자로의 유동 비등은 기계론적 모델 또는 평균 방정식으로 처리됩니다.

따라서 VOF(Volume of Fluid) 기반 다상 모델을 사용하는 본 연구에서는 BWR에서 유동 비등 동안 계면의 파괴 및 생성에 대한 자세한 수치적 이해를 목표로 합니다.

우리의 작업은 막대 번들 시나리오에서 거의 사실적인 증기 기포 및 액체 흐름 역학을 묘사합니다. 연료봉에 대한 일정한 벽 열유속과 입구 패치에서 액체의 균일한 속도가 경계 조건으로 적용됩니다. 물의 포화 특성은 30bar 압력에서 취합니다.

기포 핵 생성, 성장, 병합, 국소 건조, 액체 패치로 재습윤 및 완전한 건조를 포함하는 유동 비등 단계가 설명됩니다. 액체가 존재하지 않는 건조 현상은 다양한 축 단면에서 위상 분율 윤곽으로 수치적으로 관찰됩니다.

다른 축 평면에서 액상 분율의 정량화는 점진적인 건조 메커니즘을 강조하면서 시간이 지남에 따라 표시됩니다. 내부 막대와 외부 막대의 액-증기 분포를 비교하면 내부 막대의 건조가 외부 막대보다 더 빨리 발생함을 알 수 있습니다. 각 경우의 방열 용량을 식별하기 위한 열 전달 계수도 보고됩니다.

Fig. 1. A typical Boiling Water Reactor (BWR) and selected segment of study for simulation
Fig. 1. A typical Boiling Water Reactor (BWR) and selected segment of study for simulation
Fig. 2. (a-c) dimensions and mesh configuration for G = 6 mm; (d-f) dimensions and mesh configuration for G = 0.6 mm
Fig. 2. (a-c) dimensions and mesh configuration for G = 6 mm; (d-f) dimensions and mesh configuration for G = 0.6 mm
Fig. 3. Simulating the effect of spacer (a) Spacer configuration around rod bundle (b) Mesh structure in spacer zone (c) Distribution of vapor bubbles in a rod bundle with spacer (d) Liquid phase fraction comparison for geometry with and without spacer (e,f,g) Wall temperature comparison for geometry with and without spacer; WS: With Spacer, WOS: Without Spacer; Temperature in the y-axis is in (f) and (g) is same as (e).
Fig. 3. Simulating the effect of spacer (a) Spacer configuration around rod bundle (b) Mesh structure in spacer zone (c) Distribution of vapor bubbles in a rod bundle with spacer (d) Liquid phase fraction comparison for geometry with and without spacer (e,f,g) Wall temperature comparison for geometry with and without spacer; WS: With Spacer, WOS: Without Spacer; Temperature in the y-axis is in (f) and (g) is same as (e).
Fig. 4. Validation of the present numerical model with crossflow boiling over a heated cylindrical rod [40]
Fig. 4. Validation of the present numerical model with crossflow boiling over a heated cylindrical rod [40]
Fig. 5. Grid-Independent study in terms of vapor volume in 1/4th of computational domain
Fig. 5. Grid-Independent study in terms of vapor volume in 1/4th of computational domain
Fig. 6. Interface contour for G = 6 mm; ul = 1.2 m/s; q˙ w = 396 kW/m2; they are showing nucleation, growth, merging, and pseudo-steady-state condition.
Fig. 6. Interface contour for G = 6 mm; ul = 1.2 m/s; q˙ w = 396 kW/m2; they are showing nucleation, growth, merging, and pseudo-steady-state condition.
Fig. 7. Interface contours for G = 0.6 mm; ul = 1.2 m/s; q˙ w = 396 kW/m2; It shows dry-out at pseudo-steady-state near the exit
Fig. 7. Interface contours for G = 0.6 mm; ul = 1.2 m/s; q˙ w = 396 kW/m2; It shows dry-out at pseudo-steady-state near the exit
Fig. 8. Vapor-liquid distribution across various distant cross-sections (Black color indicates liquid; Gray color indicates vapor); Magnification factor: 1 × (for a and b), 1.5 × (for c and d)
Fig. 8. Vapor-liquid distribution across various distant cross-sections (Black color indicates liquid; Gray color indicates vapor); Magnification factor: 1 × (for a and b), 1.5 × (for c and d)
Fig. 21. Two-phase flow mixture velocity (u¯z); for G = 6 mm, r = 5 means location at inner heated wall and r = 25 means location at outer adiabatic wall; for G = 0.66 mm, r = 5 means location at inner heated wall and r = 16.6 mm means location at outer adiabatic wall.
Fig. 21. Two-phase flow mixture velocity (u¯z); for G = 6 mm, r = 5 means location at inner heated wall and r = 25 means location at outer adiabatic wall; for G = 0.66 mm, r = 5 means location at inner heated wall and r = 16.6 mm means location at outer adiabatic wall.

References

[1] J. Würtz, An Experimental and Theoretical Investigation of Annular Steam-Water Flow in Tubes and Annuli at 30 to 90 Bar, Risø National Laboratory,
Roskilde, 1978.
[2] W. Tian, A. Myint, Z. Li, S. Qiu, G.H. Su, D. Jia, Experimental study on dryout point in vertical narrow annulus under low flow conditions, in: International Conference on Nuclear Engineering, 4689, 2004, pp. 643–648. Jan
1Vol.
[3] K.M. Becker, C.H. Ling, S. Hedberg, G. Strand, An experimental investigation of
post dryout heat transfer, R. Inst. Technol. (1983).
[4] K.M. Becker, A Burnout Correlation for Flow of Boiling Water in Vertical Rod
Bundles, AB Atomenergi, 1967.
[5] Jr J.R. Barbosa, G.F. Hewitt, S.M. Richardson, High-speed visualisation of nucleate boiling in vertical annular flow, Int. J. Heat Mass Transf. 46 (26) (2003)
5153–5160 1, doi:10.1016/S0017-9310(03)00255-2.
[6] Y. Mizutani, A. Tomiyama, S. Hosokawa, A. Sou, Y. Kudo, K. Mishima, Twophase flow patterns in a four by four rod bundle, J. Nucl. Sci. Technol. 44 (6)
(2007) 894–901 1, doi:10.1080/18811248.2007.9711327.
[7] S.S. Paranjape, Two-Phase Flow Interfacial Structures in a Rod Bundle Geometry, Purdue University, 2009.
[8] D. Lavicka, J. Polansky, Model of the cooling of a nuclear reactor fuel rod, Multiph. Sci. Technol. 25 (2-4) (2013), doi:10.1615/MultScienTechn.v25.i2-4.90.
[9] M. Thurgood, J. Kelly, T. Guidotti, R. Kohrt, K. Crowell, Tech. rep., Pacific Northwest National Laboratory, 1983.
[10] S. Sugawara, Droplet deposition and entrainment modeling based on the
three-fluid model, Nucl. Eng. Des. 122 (1-3) (1990) 67–84, doi:10.1016/
0029-5493(90)90197-6.
[11] C. Adamsson, J.M. Le Corre, Modeling and validation of a mechanistic tool
(MEFISTO) for the prediction of critical power in BWR fuel assemblies, Nucl.
Eng. Des. 241 (8) (2011) 2843–2858, doi:10.1016/j.nucengdes.2011.01.033.
[12] S. Talebi, H. Kazeminejad, A mathematical approach to predict dryout in a rod
bundle, Nucl. Eng. Des. 249 (2012) 348–356, doi:10.1016/j.nucengdes.2012.04.
016.
[13] H. Anglart, O. Nylund, N. Kurul, M.Z. Podowski, CFD prediction of flow and
phase distribution in fuel assemblies with spacers, Nucl. Eng. Des. 177 (1-3)
(1997) 215–228, doi:10.1016/S0029-5493(97)00195-7.
[14] H. Li, H. Anglart, CFD model of diabatic annular two-phase flow using the
Eulerian–Lagrangian approach, Ann. Nucl. Energy 77 (2015) 415–424, doi:10.
1016/j.anucene.2014.12.002.
[15] G. Sorokin, A. Sorokin, Experimental and numerical investigation of liquid metal boiling in fuel subassemblies under natural circulation conditions, Prog. Nucl. Energy 47 (1-4) (2005) 656–663, doi:10.1016/j.pnucene.2005.
05.069.
[16] W.D. Pointer, A. Tentner, T. Sofu, D. Weber, S. Lo, A. Splawski, Eulerian
two-phase computational fluid dynamics for boiling water reactor core analysis, Joint International Topical Meeting on Mathematics and Computation and
Supercomputing in Nuclear Applications (M and C± SNA), 2007.
[17] K. Podila, Y. Rao, CFD modelling of supercritical water flow and heat transfer
in a 2 × 2 fuel rod bundle, Nucl. Eng. Des. 301 (2016) 279–289, doi:10.1016/j.
nucengdes.2016.03.019.
[18] H. Pothukuchi, S. Kelm, B.S. Patnaik, B.V. Prasad, H.J. Allelein, Numerical investigation of subcooled flow boiling in an annulus under the influence of eccentricity, Appl. Therm. Eng. 129 (2018) 1604–1617, doi:10.1016/j.applthermaleng.
2017.10.105.
[19] H. Pothukuchi, S. Kelm, B.S. Patnaik, B.V. Prasad, H.J. Allelein, CFD modeling of
critical heat flux in flow boiling: validation and assessment of closure models,
Appl. Therm. Eng. 150 (2019) 651–665, doi:10.1016/j.applthermaleng.2019.01.
030.
[20] W. Fan, H. Li, H. Anglart, A study of rewetting and conjugate heat transfer
influence on dryout and post-dryout phenomena with a multi-domain coupled CFD approach, Int. J. Heat Mass Transf. 163 (2020) 120503, doi:10.1016/j.
ijheatmasstransfer.2020.120503.
[21] R. Zhang, T. Cong, G. Su, J. Wang, S. Qiu, Investigation on the critical heat
flux in typical 5 by 5 rod bundle at conditions prototypical of PWR based
on CFD methodology, Appl. Therm. Eng. 179 (2020) 115582, doi:10.1016/j.
applthermaleng.2020.115582.

[22] L.D. Silvi, A. Saha, D.K. Chandraker, S. Ghosh, A.K. Das, Numerical analysis of
pre-dryout sequences through the route of interfacial evolution in annular gasliquid two-phase flow with phase change, Chem. Eng. Sci. 212 (2020) 115356,
doi:10.1016/j.ces.2019.115356.
[23] L.D. Silvi, D.K. Chandraker, S. Ghosh, A.K. Das, On-route to dryout through sequential interfacial dynamics in annular flow boiling around temperature and
heat flux controlled heater rod, Chem. Eng. Sci. 229 (2021) 116014, doi:10.1016/
j.ces.2020.116014.
[24] J.U. Brackbill, D.B. Kothe, C. Zemach, A continuum method for modeling surface
tension, J. Comput. Phys. 100 (2) (1992) 335–354, doi:10.1016/0021-9991(92)
90240-Y.
[25] B. Lafaurie, C. Nardone, R. Scardovelli, S. Zaleski, G. Zanetti, Modelling merging
and fragmentation in multiphase flows with SURFER, J. Comput. Phys. 113 (1)
(1994) 134–147, doi:10.1006/jcph.1994.1123.
[26] I. Tanasawa, Advances in condensation heat transfer, Ad. Heat Transf. 21 (1991)
55–139 Vol, doi:10.1016/S0065-2717(08)70334-4.
[27] V.H. Del Valle, D.B. Kenning, Subcooled flow boiling at high heat flux, Int.
J. Heat Mass Transf. 28 (10) (1985) 1907–1920, doi:10.1016/0017-9310(85)
90213-3.
[28] B. Matzner, G.M. Latter, Reduced pressure drop space for boiling water reactor
fuel bundles, US Patent US5375154A, (1993)
[29] C. Unal, O. Badr, K. Tuzla, J.C. Chen, S. Neti, Pressure drop at rod-bundle spacers
in the post-CHF dispersed flow regime, Int. J. Multiphase Flow 20 (3) (1994)
515–522, doi:10.1016/0301-9322(94)90025-6.
[30] D.K. Chandraker, A.K. Nayak, V.P. Krishnan, Effect of spacer on the dryout of
BWR fuel rod assemblies, Nucl. Eng. Des. 294 (2015), doi:10.1016/j.nucengdes.
2015.09.004.
[31] S.K Verma, S.L. Sinha, D.K. Chandraker, A comprehensive review of the spacer
effect on performance of nuclear fuel bundle using computational fluid dynamics methodology, Mater. Today: Proc. 4 (2017) 100030–110034, doi:10.
1016/j.matpr.2017.06.315.
[32] S.K Verma, S.L. Sinha, D.K. Chandraker, Experimental investigation on the effect
of space on the turbulent mixing in vertical pressure tube-type boiling water
reactor, Nucl. Sci. Eng. 190 (2) (2018), doi:10.1080/00295639.2017.1413874.
[33] T. Zhang, Y. Liu, Numerical investigation of flow and heat transfer characteristics of subcooled boiling in a single rod channel with/without spacer grid,
Case Stud. Therm. Eng. 20 (2020) 100644, doi:10.1016/j.csite.2020.100644.
[34] K.M. Becker, G. Hernborg, M. Bode, O. Eriksson, Burnout data for flow of boiling water in vertical round ducts, annuli and rod clusters, AB Atomenergi
(1965).
[35] A. Saha, A.K. Das, Numerical study of boiling around wires and influence of
active or passive neighbours on vapour film dynamics, Int. J. Heat Mass Transf.
130 (2019) 440–454, doi:10.1016/j.ijheatmasstransfer.2018.10.117.
[36] M. Reimann, U. Grigull, Heat transfer with free convection and film boiling in
the critical area of water and carbon dioxide, Heat Mass Transf. 8 (1975) 229–
239, doi:10.1007/BF01002151.
[37] M.S. Plesset, S.A. Zwick, The growth of vapor bubbles in superheated liquids, J.
Appl. Phys. 25 (4) (1954) 493–500, doi:10.1063/1.1721668.
[38] N. Samkhaniani, M.R. Ansari, Numerical simulation of superheated vapor bubble rising in stagnant liquid, Heat Mass Transf. 53 (9) (2017) 2885–2899,
doi:10.1007/S00231-017-2031-6.
[39] N. Samkhaniani, M.R. Ansari, The evaluation of the diffuse interface method
for phase change simulations using OpenFOAM, Heat Transf. Asian Res. 46 (8)
(2017) 1173–1203, doi:10.1002/htj.21268.
[40] P. Goel, A.K. Nayak, M.K. Das, J.B. Joshi, Bubble departure characteristics in a
horizontal tube bundle under cross flow conditions, Int. J. Multiph. Flow 100
(2018) 143–154, doi:10.1016/j.ijmultiphaseflow.2017.12.013.
[41] K.M. Becker, J. Engstorm, B.Scholin Nylund, B. Sodequist, Analysis of the dryout
incident in the Oskarshamn 2 boiling water reactor, Int. J. Multiph. Flow 16 (6)
(1990) 959–974, doi:10.1016/0301-9322(90)90101-N.
[42] H.G. Weller, A New Approach to VOF-Based Interface Capturing Methods
for Incompressible and Compressible Flow, A New Approach to VOF-Based
Interface Capturing Methods for Incompressible and Compressible Flow, 4,
OpenCFD Ltd., 2008 Report TR/HGW.
[43] G. Boeing, Visual analysis of nonlinear dynamical systems: chaos, fractals, selfsimilarity and the limits of prediction, Systems 4 (4) (2016) 37, doi:10.3390/
systems4040037.

Figure 7. The simulated velocity (a) and simulated pressure pattern (b) across the Parshall flume. The patterns match the physical behavior of actual Parshall flumes [7].

Application of Numerical and Experimental Modeling to Improve the Efficiency of Parshall Flumes: A Review of the State-of-the-Art

Parshall Flumes의 효율성 향상을 위한 수치 및 실험 모델링의 적용: 최신 기술 검토

Mehdi Heyrani 1,* , Abdolmajid Mohammadian 1, Ioan Nistor 1 and Omerul Faruk Dursun 2

Abstract

열린 채널에서 흐름을 관리하는 기본 단계 중 하나는 속성을 결정하는 것입니다. 개방 수로의 흐름에 관한 추가 정보를 제공하기 위해 경험적 방정식이 개발되었습니다. 이러한 실험 방정식을 얻는 것은 비용과 시간이 많이 소요됩니다. 따라서 대체 솔루션이 모색되었습니다.

지난 세기 동안 움직이는 부분이 없는 정적 측정 장치인 Parshall 수로가 개방 수로의 흐름을 측정하는 데 중요한 역할을 했습니다. 많은 연구자들이 관개 및 폐수 관리와 같은 다양한 분야에서 Parshall 수로의 적용을 연구하는 데 관심을 집중해 왔습니다.

여러 학자들이 실험 결과를 사용하여 Parshall 수로의 등급 방정식을 향상시켰지만 다른 학자들은 수치 시뮬레이션을 사용하여 높이-방전 관계 방정식을 재보정하기 위해 대체 데이터 소스를 사용했습니다. 컴퓨팅 하드웨어가 지난 수십 년 동안 크게 발전하여 과거에 경험했던 제한된 해상도를 뛰어넘는 것이 가능해짐에 따라 CFD(Computational Fluid Dynamic) 소프트웨어가 오늘날 대중화되고 있습니다.

여러 CFD 모델은 가용성에 따라 오픈 소스 또는 상업적으로 허가되어 수위 결과를 생성하기 위해 다양한 구성의 수로, 특히 Parshall 수로에 대한 수치 시뮬레이션을 수행하는 데 사용되었습니다.

FLOW-3D, Ansys Fluent, OpenFOAM 등 지금까지 사용되어 온 다양한 CFD 도구에 대해 실험 데이터로 정밀 교정한 결과, 출력이 안정적이고 실제 시나리오에 구현할 수 있음이 확인되었습니다.

결과를 생성하기 위해 이 기술을 사용하는 이점은 필요한 경우 유속 또는 구조적 형상과 같은 초기 조건을 조정하는 CFD 접근 방식의 능력입니다. 수로 크기와 수로가 위치한 부지의 조건과 관련하여 상황에 적합한 특정 Parshall 수로로 선택이 좁혀집니다.

표준 Parshall 수로를 선택하는 것이 항상 가능한 것은 아닙니다. 따라서 엔지니어는 가장 가까운 수로 크기에 약간의 수정을 제공하고 정확한 유량을 생성하기 위해 새로운 등급 곡선을 제공합니다.

이 검토는 기존 등급 방정식을 향상시키거나 구조의 기하학에 대한 추가 수정을 제안하기 위해 Parshall 수로에서 수치 시뮬레이션 및 물리적 실험 데이터의 적용을 목표로 하는 여러 학자의 작업에 대해 수행되었습니다.

One of the primary steps in managing the flow in an open channel is determining its properties. Empirical equations are developed to provide further information regarding the flow in open channels. Obtaining such experimental equations is expensive and time consuming; therefore, alternative solutions have been sought. Over the last century, the Parshall flume, a static measuring device with no moving parts, has played a significant role in measuring the flow in open channels. Many researchers have focused their interest on studying the application of Parshall flumes in various fields like irrigation and wastewater management. Although various scholars used experimental results to enhance the rating equation of the Parshall flume, others used an alternative source of data to recalibrate the height–discharge relation equation using numerical simulation. Computational Fluid Dynamic (CFD) software is becoming popular nowadays as computing hardware has advanced significantly within the last few decades, making it possible to go beyond the limited resolution that was experienced in the past. Multiple CFD models, depending on their availability, either open-source or commercially licensed, have been used to perform numerical simulations on different configurations of flumes, especially Parshall flumes, to produce water level results. Regarding various CFD tools that have been used, i.e., FLOW-3D, Ansys Fluent, or OpenFOAM, after precise calibration with experimental data, it has been determined that the output is reliable and can be implemented to the actual scenarios. The benefit of using this technique to produce results is the ability of the CFD approach to adjust the initial conditions, like flow velocity or structural geometry, where necessary. With respect to channel size and the condition of the site where the flume is located, the choices are narrowed to the specific Parshall flume suitable to the situation. It is not always possible to select the standard Parshall flume; therefore, engineers provide some modification to the closest flume size and provide a new rating curve to produce accurate flowrates. This review has been performed on the works of a number of scholars who targeted the application of numerical simulation and physical experimental data in Parshall flumes to either enhance the existing rating equation or propose further modification to the structure’s geometry.

Keywords

Parshall flume; CFD; OpenFOAM; FLOW-3D; numerical simulation; turbulence model

Figure 1. Parshall flume measuring structure, installed [2].
Figure 1. Parshall flume measuring structure, installed [2].
Figure 2. Parshall flume measuring structure, uninstalled [3]
Figure 2. Parshall flume measuring structure, uninstalled [3]
Figure 4. Mesh sensitivity analysis: top view and side view of the Parshall flume: (a) contains 27,000 cells; (b) 52,000 cells; (c) 75,000 cells; (d) 270,000 cells. The C setup was used in their simulation [7].
Figure 4. Mesh sensitivity analysis: top view and side view of the Parshall flume: (a) contains 27,000 cells; (b) 52,000 cells; (c) 75,000 cells; (d) 270,000 cells. The C setup was used in their simulation [7].
Figure 7. The simulated velocity (a) and simulated pressure pattern (b) across the Parshall flume. The patterns match the physical behavior of actual Parshall flumes [7].
Figure 7. The simulated velocity (a) and simulated pressure pattern (b) across the Parshall flume. The patterns match the physical behavior of actual Parshall flumes [7].
Figure 8. Computational grid system in the Side A flume. (a) contains a triangular grid system (b) demonstrates the rectangular grid system. (c) and (d) are three-dimensional schematics showing the superimposed grid system. (e) magnifies the dashed section in (b). (Reprinted with permission from Ref. [11]. 2020 ELSEVIER). ).
Figure 8. Computational grid system in the Side A flume. (a) contains a triangular grid system (b) demonstrates the rectangular grid system. (c) and (d) are three-dimensional schematics showing the superimposed grid system. (e) magnifies the dashed section in (b). (Reprinted with permission from Ref. [11]. 2020 ELSEVIER). ).
Figure 10. The results of flow patterns in different flumes; (a) Cutthroat flume, (b) airfoil-shaped flume, (c) airfoil pillar-shaped flume, (d) optimized airfoil-shaped flume [23]
Figure 10. The results of flow patterns in different flumes; (a) Cutthroat flume, (b) airfoil-shaped flume, (c) airfoil pillar-shaped flume, (d) optimized airfoil-shaped flume [23]
Figure 11. Experimental setup: contraction ratio used on each flume [23].
Figure 11. Experimental setup: contraction ratio used on each flume [23].
Figure 12. Entire flume geometry [25]
Figure 12. Entire flume geometry [25]

References

  1. Cone, V.M. The Venturi Flume; U.S. Government Printing Office: Washington, DC, USA, 1917.
  2. 20-Foot Concrete Parshall Flume with Radius Wing Walls. Available online: https://www.openchannelflow.com/assets/uploads/
    media/_large/20-foot-parshall-flume-curved-wing-walls.jpg (accessed on 12 January 2021).
  3. Fiberglass 6-Inch Parshall Flume with Gauge. Available online: https://www.openchannelflow.com/assets/uploads/media/
    _large/flume-parshall-6-inch-fiberglass.png (accessed on 12 January 2021).
  4. Parshall, R.L. The Parshall Measuring Flume; Colorado State College, Colorado Experiment Station: Fort Collins, CO, USA, 1936.
  5. Selecting Between a Weir and a Flume. 2022. Available online: https://www.openchannelflow.com/blog/selecting-a-primarydevice-part-1-choosing-between-a-weir-and-a-flume (accessed on 29 December 2021).
  6. Parshall, R.L. The Improved Venturi Flume. Trans. Am. Soc. Civ. Eng. 1928, 89, 841–851. [CrossRef]
  7. Heyrani, M.; Mohammadian, A.; Nistor, I. Numerical Simulation of Flow in Parshall Flume Using Selected Nonlinear Turbulence
    Models. Hydrology 2021, 8, 151. [CrossRef]
  8. Heyrani, M.; Mohammadian, A.; Nistor, I.; Dursun, O.F. Numerical Modeling of Venturi Flume. Hydrology 2021, 8, 27. [CrossRef]
  9. Alfonsi, G. Reynolds-Averaged Navier–Stokes Equations for Turbulence Modeling. Appl. Mech. Rev. 2009, 62, 040802. [CrossRef]
  10. Imanian, H.; Mohammadian, A. Numerical Simulation of Flow over Ogee Crested Spillways under High Hydraulic Head Ratio.
    Eng. Appl. Comput. Fluid Mech. 2019, 13, 983–1000. [CrossRef]
  11. Khosronejad, A.; Herb, W.; Sotiropoulos, F.; Kang, S.; Yang, X. Assessment of Parshall Flumes for Discharge Measurement of
    Open-Channel Flows: A Comparative Numerical and Field Case Study. Measurement 2020, 167, 108292. [CrossRef]
  12. Dursun, O.F. An Experimental Investigation of the Aeration Performance of Parshall Flume and Venturi Flumes. KSCE J. Civ. Eng.
    2016, 20, 943–950. [CrossRef]
  13. Shih, T.-H.; Liu, N.-S.; Chen, K.-H. A Non-Linear k-Epsilon Model for Turbulent Shear Flows. In Proceedings of the 34th
    AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, Cleveland, OH, USA, 13 July 1998; p. 3983.
  14. Lien, F.S. Low-Reynolds-Number Eddy-Viscosity Modelling Based on Non-Linear Stress-Strain/Vorticity Relations. In Proceedings of the 3rd Symposium on Engineering Turbulence Modelling and Measurement, Heraklion, Greece, 27 May 1996.
  15. Davis, R.W.; Deutsch, S. A Numerical-Experimental Study of Parhall Flumes. J. Hydraul. Res. 1980, 18, 135–152. [CrossRef]
  16. Xiao, Y.; Wang, W.; Hu, X.; Zhou, Y. Experimental and Numerical Research on Portable Short-Throat Flume in the Field. Flow
    Meas. Instrum. 2016, 47, 54–61. [CrossRef]
  17. Wright, S.J.; Tullis, B.P.; Long, T.M. Recalibration of Parshall Flumes at Low Discharges. J. Irrig. Drain. Eng. 1994, 120, 348–362.
    [CrossRef]
  18. Heiner, B.; Barfuss, S.L. Parshall Flume Discharge Corrections: Wall Staff Gauge and Centerline Measurements. J. Irrig. Drain.
    Eng. 2011, 137, 779–792. [CrossRef]
  19. Savage, B.M.; Heiner, B.; Barfuss, S. Parshall Flume Discharge Correction Coefficients through Modelling. Proc. ICE Water Manag.
    2013, 167, 279–287. [CrossRef]
  20. Zerihun, Y.T. A Numerical Study on Curvilinear Free Surface Flows in Venturi Flumes. Fluids 2016, 1, 21. [CrossRef]
  21. Sun, B.; Zhu, S.; Yang, L.; Liu, Q.; Zhang, C.; Zhang, J. ping Experimental and Numerical Investigation of Flow Measurement
    Mechanism and Hydraulic Performance on Curved Flume in Rectangular Channel. Arab. J. Sci. Eng. 2020. [CrossRef]
  22. Hu, H.; Huang, J.; Qian, Z.; Huai, W.; Yu, G. Hydraulic Analysis of Parabolic Flume for Flow Measurement. Flow Meas. Instrum.
    2014, 37, 54–64. [CrossRef]
  23. Sun, B.; Yang, L.; Zhu, S.; Liu, Q.; Wang, C.; Zhang, C. Study on the Applicability of Four Flumes in Small Rectangular Channels.
    Flow Meas. Instrum. 2021, 80, 101967. [CrossRef]
  24. Willeitner, R.P.; Barfuss, S.L.; Johnson, M.C. Using Numerical Modeling to Correct Flow Rates for Submerged Montana Flumes. J.
    Irrig. Drain. Eng. 2013, 139, 586–592. [CrossRef]
  25. Ran, D.; Wang, W.; Hu, X. Three-Dimensional Numerical Simulation of Flow in Trapezoidal Cutthroat Flumes Based on FLOW-3D.
    Front. Agric. Sci. Eng. 2018, 5, 168–176. [CrossRef]
  26. Kim, S.-Y.; Lee, J.-H.; Hong, N.-K.; Lee, S.-O. Numerical Simulation for Determining Scale of Parshall Flume. Proc. Korea Water
    Resour. Assoc. Conf. 2010, 719–723.
  27. Tekade, S.A.; Vasudeo, A.D.; Ghare, A.D.; Ingle, R.N. Measurement of Flow in Supercritical Flow Regime Using Cutthroat Flumes.
    Sadhana 2016, 41, 265–272. [CrossRef]
  28. Wahl, T.L.; Replogle, J.A.; Wahlin, B.T.; Higgs, J.A. New Developments in Design and Application of Long-Throated Flumes. In
    Proceedings of the Joint Conference on Water Resource Engineering and Water Resources Planning and Management, Minneapolis,
    MN, USA, 30 July–2 August 2000.
  29. Howes, D.J.; Burt, C.M.; Sanders, B.F. Subcritical Contraction for Improved Open-Channel Flow Measurement Accuracy with an
    Upward-Looking ADVM. J. Irrig. Drain. Eng. 2010, 136, 617–626. [CrossRef]
  30. Tiwari, N.K.; Sihag, P. Prediction of Oxygen Transfer at Modified Parshall Flumes Using Regression Models. ISH J. Hydraul. Eng.
    2020, 26, 209–220. [CrossRef]
  31. Thornton, C.I.; Smith, B.A.; Abt, S.R.; Robeson, M.D. Supercritical Flow Measurement Using a Small Parshall Flume. J. Irrig.
    Drain. Eng. 2009, 135, 683–692. [CrossRef]
  32. Cox, A.L.; Thornton, C.I.; Abt, S.R. Supercritical Flow Measurement Using a Large Parshall Flume. J. Irrig. Drain. Eng. 2013, 139,
    655–662. [CrossRef]
  1. Ribeiro, Á.S.; Sousa, J.A.; Simões, C.; Martins, L.L.; Dias, L.; Mendes, R.; Martins, C. Parshall Flumes Flow Rate Uncertainty
    Including Contributions of the Model Parameters and Correlation Effects. Meas. Sens. 2021, 18, 100108. [CrossRef]
  2. Singh, J.; Mittal, S.K.; Tiwari, H.L. Discharge Relation for Small Parshall Flume in Free Flow Condition. Int. J. Res. Eng. Technol.
    2014, 3, 317–321.
  3. Kim, S.-D.; Lee, H.-J.; Oh, B.-D. Investigation on Application of Parshall Flume for Flow Measurement of Low-Flow Season in
    Korea. Meas. Sci. Rev. 2010, 10, 111. [CrossRef]
  4. Willeitner, R.P.; Barfuss, S.L.; Johnson, M.C. Montana Flume Flow Corrections under Submerged Flow. J. Irrig. Drain. Eng. 2012,
    138, 685–689. [CrossRef]
  5. Dufresne, M.; Vazquez, J. Head–Discharge Relationship of Venturi Flumes: From Long to Short Throats. J. Hydraul. Res. 2013, 51,
    465–468. [CrossRef]
Figure 3 Simulation PTC pipes enhanced with copper foam and nanoparticles in FLOW-3D software.

다공성 미디어 및 나노유체에 의해 강화된 수집기로 태양광 CCHP 시스템의 최적화

Optimization of Solar CCHP Systems with Collector Enhanced by Porous Media and Nanofluid


Navid Tonekaboni,1Mahdi Feizbahr,2 Nima Tonekaboni,1Guang-Jun Jiang,3,4 and Hong-Xia Chen3,4

Abstract

태양열 집열기의 낮은 효율은 CCHP(Solar Combined Cooling, Heating, and Power) 사이클의 문제점 중 하나로 언급될 수 있습니다. 태양계를 개선하기 위해 나노유체와 다공성 매체가 태양열 집열기에 사용됩니다.

다공성 매질과 나노입자를 사용하는 장점 중 하나는 동일한 조건에서 더 많은 에너지를 흡수할 수 있다는 것입니다. 이 연구에서는 평균 일사량이 1b인 따뜻하고 건조한 지역의 600 m2 건물의 전기, 냉방 및 난방을 생성하기 위해 다공성 매질과 나노유체를 사용하여 태양열 냉난방 복합 발전(SCCHP) 시스템을 최적화했습니다.

본 논문에서는 침전물이 형성되지 않는 lb = 820 w/m2(이란) 정도까지 다공성 물질에서 나노유체의 최적량을 계산하였다. 이 연구에서 태양열 집열기는 구리 다공성 매체(95% 다공성)와 CuO 및 Al2O3 나노 유체로 향상되었습니다.

나노유체의 0.1%-0.6%가 작동 유체로 물에 추가되었습니다. 나노유체의 0.5%가 태양열 집열기 및 SCCHP 시스템에서 가장 높은 에너지 및 엑서지 효율 향상으로 이어지는 것으로 밝혀졌습니다.

본 연구에서 포물선형 집열기(PTC)의 최대 에너지 및 엑서지 효율은 각각 74.19% 및 32.6%입니다. 그림 1은 태양 CCHP의 주기를 정확하게 설명하기 위한 그래픽 초록으로 언급될 수 있습니다.

The low efficiency of solar collectors can be mentioned as one of the problems in solar combined cooling, heating, and power (CCHP) cycles. For improving solar systems, nanofluid and porous media are used in solar collectors. One of the advantages of using porous media and nanoparticles is to absorb more energy under the same conditions. In this research, a solar combined cooling, heating, and power (SCCHP) system has been optimized by porous media and nanofluid for generating electricity, cooling, and heating of a 600 m2 building in a warm and dry region with average solar radiation of Ib = 820 w/m2 in Iran. In this paper, the optimal amount of nanofluid in porous materials has been calculated to the extent that no sediment is formed. In this study, solar collectors were enhanced with copper porous media (95% porosity) and CuO and Al2O3 nanofluids. 0.1%–0.6% of the nanofluids were added to water as working fluids; it is found that 0.5% of the nanofluids lead to the highest energy and exergy efficiency enhancement in solar collectors and SCCHP systems. Maximum energy and exergy efficiency of parabolic thermal collector (PTC) riches in this study are 74.19% and 32.6%, respectively. Figure 1 can be mentioned as a graphical abstract for accurately describing the cycle of solar CCHP.

1. Introduction

Due to the increase in energy consumption, the use of clean energy is one of the important goals of human societies. In the last four decades, the use of cogeneration cycles has increased significantly due to high efficiency. Among clean energy, the use of solar energy has become more popular due to its greater availability [1]. Low efficiency of energy production, transmission, and distribution system makes a new system to generate simultaneously electricity, heating, and cooling as an essential solution to be widely used. The low efficiency of the electricity generation, transmission, and distribution system makes the CCHP system a basic solution to eliminate waste of energy. CCHP system consists of a prime mover (PM), a power generator, a heat recovery system (produce extra heating/cooling/power), and thermal energy storage (TES) [2]. Solar combined cooling, heating, and power (SCCHP) has been started three decades ago. SCCHP is a system that receives its propulsive force from solar energy; in this cycle, solar collectors play the role of propulsive for generating power in this system [3].

Increasing the rate of energy consumption in the whole world because of the low efficiency of energy production, transmission, and distribution system causes a new cogeneration system to generate electricity, heating, and cooling energy as an essential solution to be widely used. Building energy utilization fundamentally includes power required for lighting, home electrical appliances, warming and cooling of building inside, and boiling water. Domestic usage contributes to an average of 35% of the world’s total energy consumption [4].

Due to the availability of solar energy in all areas, solar collectors can be used to obtain the propulsive power required for the CCHP cycle. Solar energy is the main source of energy in renewable applications. For selecting a suitable area to use solar collectors, annual sunshine hours, the number of sunny days, minus temperature and frosty days, and the windy status of the region are essentially considered [5]. Iran, with an average of more than 300 sunny days, is one of the suitable countries to use solar energy. Due to the fact that most of the solar radiation is in the southern regions of Iran, also the concentration of cities is low in these areas, and transmission lines are far apart, one of the best options is to use CCHP cycles based on solar collectors [6]. One of the major problems of solar collectors is their low efficiency [7]. Low efficiency increases the area of collectors, which increases the initial cost of solar systems and of course increases the initial payback period. To increase the efficiency of solar collectors and improve their performance, porous materials and nanofluids are used to increase their workability.

There are two ways to increase the efficiency of solar collectors and mechanical and fluid improvement. In the first method, using porous materials or helical filaments inside the collector pipes causes turbulence of the flow and increases heat transfer. In the second method, using nanofluids or salt and other materials increases the heat transfer of water. The use of porous materials has grown up immensely over the past twenty years. Porous materials, especially copper porous foam, are widely used in solar collectors. Due to the high contact surface area, porous media are appropriate candidates for solar collectors [8]. A number of researchers investigated Solar System performance in accordance with energy and exergy analyses. Zhai et al. [9] reviewed the performance of a small solar-powered system in which the energy efficiency was 44.7% and the electrical efficiency was 16.9%.

Abbasi et al. [10] proposed an innovative multiobjective optimization to optimize the design of a cogeneration system. Results showed the CCHP system based on an internal diesel combustion engine was the applicable alternative at all regions with different climates. The diesel engine can supply the electrical requirement of 31.0% and heating demand of 3.8% for building.

Jiang et al. [11] combined the experiment and simulation together to analyze the performance of a cogeneration system. Moreover, some research focused on CCHP systems using solar energy. It integrated sustainable and renewable technologies in the CCHP, like PV, Stirling engine, and parabolic trough collector (PTC) [21215].

Wang et al. [16] optimized a cogeneration solar cooling system with a Rankine cycle and ejector to reach the maximum total system efficiency of 55.9%. Jing et al. analyzed a big-scale building with the SCCHP system and auxiliary heaters to produced electrical, cooling, and heating power. The maximum energy efficiency reported in their work is 46.6% [17]. Various optimization methods have been used to improve the cogeneration system, minimum system size, and performance, such as genetic algorithm [1819].

Hirasawa et al. [20] investigated the effect of using porous media to reduce thermal waste in solar systems. They used the high-porosity metal foam on top of the flat plate solar collector and observed that thermal waste decreased by 7% due to natural heat transfer. Many researchers study the efficiency improvement of the solar collector by changing the collector’s shapes or working fluids. However, the most effective method is the use of nanofluids in the solar collector as working fluid [21]. In the experimental study done by Jouybari et al. [22], the efficiency enhancement up to 8.1% was achieved by adding nanofluid in a flat plate collector. In this research, by adding porous materials to the solar collector, collector efficiency increased up to 92% in a low flow regime. Subramani et al. [23] analyzed the thermal performance of the parabolic solar collector with Al2O3 nanofluid. They conducted their experiments with Reynolds number range 2401 to 7202 and mass flow rate 0.0083 to 0.05 kg/s. The maximum efficiency improvement in this experiment was 56% at 0.05 kg/s mass flow rate.

Shojaeizadeh et al. [24] investigated the analysis of the second law of thermodynamic on the flat plate solar collector using Al2O3/water nanofluid. Their research showed that energy efficiency rose up to 1.9% and the exergy efficiency increased by a maximum of 0.72% compared to pure water. Tiwari et al. [25] researched on the thermal performance of solar flat plate collectors for working fluid water with different nanofluids. The result showed that using 1.5% (optimum) particle volume fraction of Al2O3 nanofluid as an absorbing medium causes the thermal efficiency to enhance up to 31.64%.

The effect of porous media and nanofluids on solar collectors has already been investigated in the literature but the SCCHP system with a collector embedded by both porous media and nanofluid for enhancing the ratio of nanoparticle in nanofluid for preventing sedimentation was not discussed. In this research, the amount of energy and exergy of the solar CCHP cycles with parabolic solar collectors in both base and improved modes with a porous material (copper foam with 95% porosity) and nanofluid with different ratios of nanoparticles was calculated. In the first step, it is planned to design a CCHP system based on the required load, and, in the next step, it will analyze the energy and exergy of the system in a basic and optimize mode. In the optimize mode, enhanced solar collectors with porous material and nanofluid in different ratios (0.1%–0.7%) were used to optimize the ratio of nanofluids to prevent sedimentation.

2. Cycle Description

CCHP is one of the methods to enhance energy efficiency and reduce energy loss and costs. The SCCHP system used a solar collector as a prime mover of the cogeneration system and assisted the boiler to generate vapor for the turbine. Hot water flows from the expander to the absorption chiller in summer or to the radiator or fan coil in winter. Finally, before the hot water wants to flow back to the storage tank, it flows inside a heat exchanger for generating domestic hot water [26].

For designing of solar cogeneration system and its analysis, it is necessary to calculate the electrical, heating (heating load is the load required for the production of warm water and space heating), and cooling load required for the case study considered in a residential building with an area of 600 m2 in the warm region of Iran (Zahedan). In Table 1, the average of the required loads is shown for the different months of a year (average of electrical, heating, and cooling load calculated with CARRIER software).Table 1 The average amount of electric charges, heating load, and cooling load used in the different months of the year in the city of Zahedan for a residential building with 600 m2.

According to Table 1, the maximum magnitude of heating, cooling, and electrical loads is used to calculate the cogeneration system. The maximum electric load is 96 kW, the maximum amount of heating load is 62 kW, and the maximum cooling load is 118 kW. Since the calculated loads are average, all loads increased up to 10% for the confidence coefficient. With the obtained values, the solar collector area and other cogeneration system components are calculated. The cogeneration cycle is capable of producing 105 kW electric power, 140 kW cooling capacity, and 100 kW heating power.

2.1. System Analysis Equations

An analysis is done by considering the following assumptions:(1)The system operates under steady-state conditions(2)The system is designed for the warm region of Iran (Zahedan) with average solar radiation Ib = 820 w/m2(3)The pressure drops in heat exchangers, separators, storage tanks, and pipes are ignored(4)The pressure drop is negligible in all processes and no expectable chemical reactions occurred in the processes(5)Potential, kinetic, and chemical exergy are not considered due to their insignificance(6)Pumps have been discontinued due to insignificance throughout the process(7)All components are assumed adiabatic

Schematic shape of the cogeneration cycle is shown in Figure 1 and all data are given in Table 2.

Figure 1 Schematic shape of the cogeneration cycle.Table 2 Temperature and humidity of different points of system.

Based on the first law of thermodynamic, energy analysis is based on the following steps.

First of all, the estimated solar radiation energy on collector has been calculated:where α is the heat transfer enhancement coefficient based on porous materials added to the collector’s pipes. The coefficient α is increased by the porosity percentage, the type of porous material (in this case, copper with a porosity percentage of 95), and the flow of fluid to the collector equation.

Collector efficiency is going to be calculated by the following equation [9]:

Total energy received by the collector is given by [9]

Also, the auxiliary boiler heat load is [2]

Energy consumed from vapor to expander is calculated by [2]

The power output form by the screw expander [9]:

The efficiency of the expander is 80% in this case [11].

In this step, cooling and heating loads were calculated and then, the required heating load to reach sanitary hot water will be calculated as follows:

First step: calculating the cooling load with the following equation [9]:

Second step: calculating heating loads [9]:

Then, calculating the required loud for sanitary hot water will be [9]

According to the above-mentioned equations, efficiency is [9]

In the third step, calculated exergy analysis as follows.

First, the received exergy collector from the sun is calculated [9]:

In the previous equation, f is the constant of air dilution.

The received exergy from the collector is [9]

In the case of using natural gas in an auxiliary heater, the gas exergy is calculated from the following equation [12]:

Delivering exergy from vapor to expander is calculated with the following equation [9]:

In the fourth step, the exergy in cooling and heating is calculated by the following equation:

Cooling exergy in summer is calculated [9]:

Heating exergy in winter is calculated [9]:

In the last step based on thermodynamic second law, exergy efficiency has been calculated from the following equation and the above-mentioned calculated loads [9]:

3. Porous Media

The porous medium that filled the test section is copper foam with a porosity of 95%. The foams are determined in Figure 2 and also detailed thermophysical parameters and dimensions are shown in Table 3.

Figure 2 Copper foam with a porosity of 95%.Table 3 Thermophysical parameters and dimensions of copper foam.

In solar collectors, copper porous materials are suitable for use at low temperatures and have an easier and faster manufacturing process than ceramic porous materials. Due to the high coefficient conductivity of copper, the use of copper metallic foam to increase heat transfer is certainly more efficient in solar collectors.

Porous media and nanofluid in solar collector’s pipes were simulated in FLOW-3D software using the finite-difference method [27]. Nanoparticles Al2O3 and CUO are mostly used in solar collector enhancement. In this research, different concentrations of nanofluid are added to the parabolic solar collectors with porous materials (copper foam with porosity of 95%) to achieve maximum heat transfer in the porous materials before sedimentation. After analyzing PTC pipes with the nanofluid flow in FLOW-3D software, for energy and exergy efficiency analysis, Carrier software results were used as EES software input. Simulation PTC with porous media inside collector pipe and nanofluids sedimentation is shown in Figure 3.

Figure 3 Simulation PTC pipes enhanced with copper foam and nanoparticles in FLOW-3D software.

3.1. Nano Fluid

In this research, copper and silver nanofluids (Al2O3, CuO) have been added with percentages of 0.1%–0.7% as the working fluids. The nanoparticle properties are given in Table 4. Also, system constant parameters are presented in Table 4, which are available as default input in the EES software.Table 4 Properties of the nanoparticles [9].

System constant parameters for input in the software are shown in Table 5.Table 5 System constant parameters.

The thermal properties of the nanofluid can be obtained from equations (18)–(21). The basic fluid properties are indicated by the index (bf) and the properties of the nanoparticle silver with the index (np).

The density of the mixture is shown in the following equation [28]:where ρ is density and ϕ is the nanoparticles volume fraction.

The specific heat capacity is calculated from the following equation [29]:

The thermal conductivity of the nanofluid is calculated from the following equation [29]:

The parameter β is the ratio of the nanolayer thickness to the original particle radius and, usually, this parameter is taken equal to 0.1 for the calculated thermal conductivity of the nanofluids.

The mixture viscosity is calculated as follows [30]:

In all equations, instead of water properties, working fluids with nanofluid are used. All of the above equations and parameters are entered in the EES software for calculating the energy and exergy of solar collectors and the SCCHP cycle. All calculation repeats for both nanofluids with different concentrations of nanofluid in the solar collector’s pipe.

4. Results and Discussion

In the present study, relations were written according to Wang et al. [16] and the system analysis was performed to ensure the correctness of the code. The energy and exergy charts are plotted based on the main values of the paper and are shown in Figures 4 and 5. The error rate in this simulation is 1.07%.

Figure 4 Verification charts of energy analysis results.

Figure 5 Verification charts of exergy analysis results.

We may also investigate the application of machine learning paradigms [3141] and various hybrid, advanced optimization approaches that are enhanced in terms of exploration and intensification [4255], and intelligent model studies [5661] as well, for example, methods such as particle swarm optimizer (PSO) [6062], differential search (DS) [63], ant colony optimizer (ACO) [616465], Harris hawks optimizer (HHO) [66], grey wolf optimizer (GWO) [5367], differential evolution (DE) [6869], and other fusion and boosted systems [4146485054557071].

At the first step, the collector is modified with porous copper foam material. 14 cases have been considered for the analysis of the SCCHP system (Table 6). It should be noted that the adding of porous media causes an additional pressure drop inside the collector [922263072]. All fourteen cases use copper foam with a porosity of 95 percent. To simulate the effect of porous materials and nanofluids, the first solar PTC pipes have been simulated in the FLOW-3D software and then porous media (copper foam with porosity of 95%) and fluid flow with nanoparticles (AL2O3 and CUO) are generated in the software. After analyzing PTC pipes in FLOW-3D software, for analyzing energy and exergy efficiency, software outputs were used as EES software input for optimization ratio of sedimentation and calculating energy and exergy analyses.Table 6 Collectors with different percentages of nanofluids and porous media.

In this research, an enhanced solar collector with both porous media and Nanofluid is investigated. In the present study, 0.1–0.5% CuO and Al2O3 concentration were added to the collector fully filled by porous media to achieve maximum energy and exergy efficiencies of solar CCHP systems. All steps of the investigation are shown in Table 6.

Energy and exergy analyses of parabolic solar collectors and SCCHP systems are shown in Figures 6 and 7.

Figure 6 Energy and exergy efficiencies of the PTC with porous media and nanofluid.

Figure 7 Energy and exergy efficiency of the SCCHP.

Results show that the highest energy and exergy efficiencies are 74.19% and 32.6%, respectively, that is achieved in Step 12 (parabolic collectors with filled porous media and 0.5% Al2O3). In the second step, the maximum energy efficiency of SCCHP systems with fourteen steps of simulation are shown in Figure 7.

In the second step, where 0.1, −0.6% of the nanofluids were added, it is found that 0.5% leads to the highest energy and exergy efficiency enhancement in solar collectors and SCCHP systems. Using concentrations more than 0.5% leads to sediment in the solar collector’s pipe and a decrease of porosity in the pipe [73]. According to Figure 7, maximum energy and exergy efficiencies of SCCHP are achieved in Step 12. In this step energy efficiency is 54.49% and exergy efficiency is 18.29%. In steps 13 and 14, with increasing concentration of CUO and Al2O3 nanofluid solution in porous materials, decreasing of energy and exergy efficiency of PTC and SCCHP system at the same time happened. This decrease in efficiency is due to the formation of sediment in the porous material. Calculations and simulations have shown that porous materials more than 0.5% nanofluids inside the collector pipe cause sediment and disturb the porosity of porous materials and pressure drop and reduce the coefficient of performance of the cogeneration system. Most experience showed that CUO and AL2O3 nanofluids with less than 0.6% percent solution are used in the investigation on the solar collectors at low temperatures and discharges [74]. One of the important points of this research is that the best ratio of nanofluids in the solar collector with a low temperature is 0.5% (AL2O3 and CUO); with this replacement, the cost of solar collectors and SCCHP cycle is reduced.

5. Conclusion and Future Directions

In the present study, ways for increasing the efficiency of solar collectors in order to enhance the efficiency of the SCCHP cycle are examined. The research is aimed at adding both porous materials and nanofluids for estimating the best ratio of nanofluid for enhanced solar collector and protecting sedimentation in porous media. By adding porous materials (copper foam with porosity of 95%) and 0.5% nanofluids together, high efficiency in solar parabolic collectors can be achieved. The novelty in this research is the addition of both nanofluids and porous materials and calculating the best ratio for preventing sedimentation and pressure drop in solar collector’s pipe. In this study, it was observed that, by adding 0.5% of AL2O3 nanofluid in working fluids, the energy efficiency of PTC rises to 74.19% and exergy efficiency is grown up to 32.6%. In SCCHP cycle, energy efficiency is 54.49% and exergy efficiency is 18.29%.

In this research, parabolic solar collectors fully filled by porous media (copper foam with a porosity of 95) are investigated. In the next step, parabolic solar collectors in the SCCHP cycle were simultaneously filled by porous media and different percentages of Al2O3 and CuO nanofluid. At this step, values of 0.1% to 0.6% of each nanofluid were added to the working fluid, and the efficiency of the energy and exergy of the collectors and the SCCHP cycle were determined. In this case, nanofluid and the porous media were used together in the solar collector and maximum efficiency achieved. 0.5% of both nanofluids were used to achieve the biggest efficiency enhancement.

In the present study, as expected, the highest efficiency is for the parabolic solar collector fully filled by porous material (copper foam with a porosity of 95%) and 0.5% Al2O3. Results of the present study are as follows:(1)The average enhancement of collectors’ efficiency using porous media and nanofluids is 28%.(2)Solutions with 0.1 to 0.5% of nanofluids (CuO and Al2O3) are used to prevent collectors from sediment occurrence in porous media.(3)Collector of solar cogeneration cycles that is enhanced by both porous media and nanofluid has higher efficiency, and the stability of output temperature is more as well.(4)By using 0.6% of the nanofluids in the enhanced parabolic solar collectors with copper porous materials, sedimentation occurs and makes a high-pressure drop in the solar collector’s pipe which causes decrease in energy efficiency.(5)Average enhancement of SCCHP cycle efficiency is enhanced by both porous media and nanofluid 13%.

Nomenclature

:Solar radiation
a:Heat transfer augmentation coefficient
A:Solar collector area
Bf:Basic fluid
:Specific heat capacity of the nanofluid
F:Constant of air dilution
:Thermal conductivity of the nanofluid
:Thermal conductivity of the basic fluid
:Viscosity of the nanofluid
:Viscosity of the basic fluid
:Collector efficiency
:Collector energy receives
:Auxiliary boiler heat
:Expander energy
:Gas energy
:Screw expander work
:Cooling load, in kilowatts
:Heating load, in kilowatts
:Solar radiation energy on collector, in Joule
:Sanitary hot water load
Np:Nanoparticle
:Energy efficiency
:Heat exchanger efficiency
:Sun exergy
:Collector exergy
:Natural gas exergy
:Expander exergy
:Cooling exergy
:Heating exergy
:Exergy efficiency
:Steam mass flow rate
:Hot water mass flow rate
:Specific heat capacity of water
:Power output form by the screw expander
Tam:Average ambient temperature
:Density of the mixture.

Greek symbols

ρ:Density
ϕ:Nanoparticles volume fraction
β:Ratio of the nanolayer thickness.

Abbreviations

CCHP:Combined cooling, heating, and power
EES:Engineering equation solver.

Data Availability

For this study, data were generated by CARRIER software for the average electrical, heating, and cooling load of a residential building with 600 m2 in the city of Zahedan, Iran.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

This work was partially supported by the National Natural Science Foundation of China under Contract no. 71761030 and Natural Science Foundation of Inner Mongolia under Contract no. 2019LH07003.

References

  1. A. Fudholi and K. Sopian, “Review on solar collector for agricultural produce,” International Journal of Power Electronics and Drive Systems (IJPEDS), vol. 9, no. 1, p. 414, 2018.View at: Publisher Site | Google Scholar
  2. G. Yang and X. Zhai, “Optimization and performance analysis of solar hybrid CCHP systems under different operation strategies,” Applied Thermal Engineering, vol. 133, pp. 327–340, 2018.View at: Publisher Site | Google Scholar
  3. J. Wang, Z. Han, and Z. Guan, “Hybrid solar-assisted combined cooling, heating, and power systems: a review,” Renewable and Sustainable Energy Reviews, vol. 133, p. 110256, 2020.View at: Publisher Site | Google Scholar
  4. Y. Tian and C. Y. Zhao, “A review of solar collectors and thermal energy storage in solar thermal applications,” Applied Energy, vol. 104, pp. 538–553, 2013.View at: Publisher Site | Google Scholar
  5. J. M. Hassan, Q. J. Abdul-Ghafour, and M. F. Mohammed, “CFD simulation of enhancement techniques in flat plate solar water collectors,” Al-Nahrain Journal for Engineering Sciences, vol. 20, no. 3, pp. 751–761, 2017.View at: Google Scholar
  6. M. Jahangiri, O. Nematollahi, A. Haghani, H. A. Raiesi, and A. Alidadi Shamsabadi, “An optimization of energy cost of clean hybrid solar-wind power plants in Iran,” International Journal of Green Energy, vol. 16, no. 15, pp. 1422–1435, 2019.View at: Publisher Site | Google Scholar
  7. I. H. Yılmaz and A. Mwesigye, “Modeling, simulation and performance analysis of parabolic trough solar collectors: a comprehensive review,” Applied Energy, vol. 225, pp. 135–174, 2018.View at: Google Scholar
  8. F. Wang, J. Tan, and Z. Wang, “Heat transfer analysis of porous media receiver with different transport and thermophysical models using mixture as feeding gas,” Energy Conversion and Management, vol. 83, pp. 159–166, 2014.View at: Publisher Site | Google Scholar
  9. H. Zhai, Y. J. Dai, J. Y. Wu, and R. Z. Wang, “Energy and exergy analyses on a novel hybrid solar heating, cooling and power generation system for remote areas,” Applied Energy, vol. 86, no. 9, pp. 1395–1404, 2009.View at: Publisher Site | Google Scholar
  10. M. H. Abbasi, H. Sayyaadi, and M. Tahmasbzadebaie, “A methodology to obtain the foremost type and optimal size of the prime mover of a CCHP system for a large-scale residential application,” Applied Thermal Engineering, vol. 135, pp. 389–405, 2018.View at: Google Scholar
  11. R. Jiang, F. G. F. Qin, X. Yang, S. Huang, and B. Chen, “Performance analysis of a liquid absorption dehumidifier driven by jacket-cooling water of a diesel engine in a CCHP system,” Energy and Buildings, vol. 163, pp. 70–78, 2018.View at: Publisher Site | Google Scholar
  12. F. A. Boyaghchi and M. Chavoshi, “Monthly assessments of exergetic, economic and environmental criteria and optimization of a solar micro-CCHP based on DORC,” Solar Energy, vol. 166, pp. 351–370, 2018.View at: Publisher Site | Google Scholar
  13. F. A. Boyaghchi and M. Chavoshi, “Multi-criteria optimization of a micro solar-geothermal CCHP system applying water/CuO nanofluid based on exergy, exergoeconomic and exergoenvironmental concepts,” Applied Thermal Engineering, vol. 112, pp. 660–675, 2017.View at: Publisher Site | Google Scholar
  14. B. Su, W. Han, Y. Chen, Z. Wang, W. Qu, and H. Jin, “Performance optimization of a solar assisted CCHP based on biogas reforming,” Energy Conversion and Management, vol. 171, pp. 604–617, 2018.View at: Publisher Site | Google Scholar
  15. F. A. Al-Sulaiman, F. Hamdullahpur, and I. Dincer, “Performance assessment of a novel system using parabolic trough solar collectors for combined cooling, heating, and power production,” Renewable Energy, vol. 48, pp. 161–172, 2012.View at: Publisher Site | Google Scholar
  16. J. Wang, Y. Dai, L. Gao, and S. Ma, “A new combined cooling, heating and power system driven by solar energy,” Renewable Energy, vol. 34, no. 12, pp. 2780–2788, 2009.View at: Publisher Site | Google Scholar
  17. Y.-Y. Jing, H. Bai, J.-J. Wang, and L. Liu, “Life cycle assessment of a solar combined cooling heating and power system in different operation strategies,” Applied Energy, vol. 92, pp. 843–853, 2012.View at: Publisher Site | Google Scholar
  18. J.-J. Wang, Y.-Y. Jing, and C.-F. Zhang, “Optimization of capacity and operation for CCHP system by genetic algorithm,” Applied Energy, vol. 87, no. 4, pp. 1325–1335, 2010.View at: Publisher Site | Google Scholar
  19. L. Ali, “LDA–GA–SVM: improved hepatocellular carcinoma prediction through dimensionality reduction and genetically optimized support vector machine,” Neural Computing and Applications, vol. 87, pp. 1–10, 2020.View at: Google Scholar
  20. S. Hirasawa, R. Tsubota, T. Kawanami, and K. Shirai, “Reduction of heat loss from solar thermal collector by diminishing natural convection with high-porosity porous medium,” Solar Energy, vol. 97, pp. 305–313, 2013.View at: Publisher Site | Google Scholar
  21. E. Bellos, C. Tzivanidis, and Z. Said, “A systematic parametric thermal analysis of nanofluid-based parabolic trough solar collectors,” Sustainable Energy Technologies and Assessments, vol. 39, p. 100714, 2020.View at: Publisher Site | Google Scholar
  22. H. J. Jouybari, S. Saedodin, A. Zamzamian, M. E. Nimvari, and S. Wongwises, “Effects of porous material and nanoparticles on the thermal performance of a flat plate solar collector: an experimental study,” Renewable Energy, vol. 114, pp. 1407–1418, 2017.View at: Publisher Site | Google Scholar
  23. J. Subramani, P. K. Nagarajan, S. Wongwises, S. A. El-Agouz, and R. Sathyamurthy, “Experimental study on the thermal performance and heat transfer characteristics of solar parabolic trough collector using Al2O3 nanofluids,” Environmental Progress & Sustainable Energy, vol. 37, no. 3, pp. 1149–1159, 2018.View at: Publisher Site | Google Scholar
  24. E. Shojaeizadeh, F. Veysi, and A. Kamandi, “Exergy efficiency investigation and optimization of an Al2O3-water nanofluid based Flat-plate solar collector,” Energy and Buildings, vol. 101, pp. 12–23, 2015.View at: Publisher Site | Google Scholar
  25. A. K. Tiwari, P. Ghosh, and J. Sarkar, “Solar water heating using nanofluids–a comprehensive overview and environmental impact analysis,” International Journal of Emerging Technology and Advanced Engineering, vol. 3, no. 3, pp. 221–224, 2013.View at: Google Scholar
  26. D. R. Rajendran, E. Ganapathy Sundaram, P. Jawahar, V. Sivakumar, O. Mahian, and E. Bellos, “Review on influencing parameters in the performance of concentrated solar power collector based on materials, heat transfer fluids and design,” Journal of Thermal Analysis and Calorimetry, vol. 140, no. 1, pp. 33–51, 2020.View at: Publisher Site | Google Scholar
  27. M. Feizbahr, C. Kok Keong, F. Rostami, and M. Shahrokhi, “Wave energy dissipation using perforated and non perforated piles,” International Journal of Engineering, vol. 31, no. 2, pp. 212–219, 2018.View at: Google Scholar
  28. K. Khanafer and K. Vafai, “A critical synthesis of thermophysical characteristics of nanofluids,” International Journal of Heat and Mass Transfer, vol. 54, no. 19-20, pp. 4410–4428, 2011.View at: Publisher Site | Google Scholar
  29. K. Farhana, K. Kadirgama, M. M. Rahman et al., “Improvement in the performance of solar collectors with nanofluids – a state-of-the-art review,” Nano-Structures & Nano-Objects, vol. 18, p. 100276, 2019.View at: Publisher Site | Google Scholar
  30. M. Turkyilmazoglu, “Condensation of laminar film over curved vertical walls using single and two-phase nanofluid models,” European Journal of Mechanics-B/Fluids, vol. 65, pp. 184–191, 2017.View at: Publisher Site | Google Scholar
  31. X. Zhang, J. Wang, T. Wang, R. Jiang, J. Xu, and L. Zhao, “Robust feature learning for adversarial defense via hierarchical feature alignment,” Information Sciences, vol. 2020, 2020.View at: Google Scholar
  32. X. Zhang, T. Wang, W. Luo, and P. Huang, “Multi-level fusion and attention-guided CNN for image dehazing,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 1, 2020.View at: Publisher Site | Google Scholar
  33. X. Zhang, M. Fan, D. Wang, P. Zhou, and D. Tao, “Top-k feature selection framework using robust 0-1 integer programming,” IEEE Transactions on Neural Networks and Learning Systems, vol. 1, pp. 1–15, 2020.View at: Publisher Site | Google Scholar
  34. X. Zhang, D. Wang, Z. Zhou, and Y. Ma, “Robust low-rank tensor recovery with rectification and alignment,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 1, pp. 238–255, 2019.View at: Google Scholar
  35. X. Zhang, R. Jiang, T. Wang, and J. Wang, “Recursive neural network for video deblurring,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 1, 2020.View at: Publisher Site | Google Scholar
  36. X. Zhang, T. Wang, J. Wang, G. Tang, and L. Zhao, “Pyramid channel-based feature attention network for image dehazing,” Computer Vision and Image Understanding, vol. 1, 2020.View at: Google Scholar
  37. M. Mirmozaffari, “Machine learning algorithms based on an optimization model,” 2020.View at: Google Scholar
  38. M. Mirmozaffari, M. Yazdani, A. Boskabadi, H. Ahady Dolatsara, K. Kabirifar, and N. Amiri Golilarz, “A novel machine learning approach combined with optimization models for eco-efficiency evaluation,” Applied Sciences, vol. 10, no. 15, p. 5210, 2020.View at: Publisher Site | Google Scholar
  39. M. Vosoogha and A. Addeh, “An intelligent power prediction method for wind energy generation based on optimized fuzzy system,” Computational Research Progress in Applied Science & Engineering (CRPASE), vol. 5, pp. 34–43, 2019.View at: Google Scholar
  40. A. Javadi, N. Mikaeilvand, and H. Hosseinzdeh, “Presenting a new method to solve partial differential equations using a group search optimizer method (GSO),” Computational Research Progress in Applied Science and Engineering, vol. 4, no. 1, pp. 22–26, 2018.View at: Google Scholar
  41. F. J. Golrokh, Gohar Azeem, and A. Hasan, “Eco-efficiency evaluation in cement industries: DEA malmquist productivity index using optimization models,” ENG Transactions, vol. 1, pp. 1–8, 2020.View at: Google Scholar
  42. H. Yu, “Dynamic Gaussian bare-bones fruit fly optimizers with abandonment mechanism: method and analysis,” Engineering with Computers, vol. 1, pp. 1–29, 2020.View at: Google Scholar
  43. C. Yu, “SGOA: annealing-behaved grasshopper optimizer for global tasks,” Engineering with Computers, vol. 1, pp. 1–28, 2021.View at: Google Scholar
  44. W. Shan, Z. Qiao, A. A. Heidari, H. Chen, H. Turabieh, and Y. Teng, “Double adaptive weights for stabilization of moth flame optimizer: balance analysis, engineering cases, and medical diagnosis,” Knowledge-Based Systems, vol. 1, p. 106728, 2020.View at: Google Scholar
  45. J. Tu, H. Chen, J. Liu et al., “Evolutionary biogeography-based whale optimization methods with communication structure: towards measuring the balance,” Knowledge-Based Systems, vol. 212, p. 106642, 2021.View at: Publisher Site | Google Scholar
  46. Y. Zhang, “Towards augmented kernel extreme learning models for bankruptcy prediction: algorithmic behavior and comprehensive analysis,” Neurocomputing, vol. 1, 2020.View at: Google Scholar
  47. Y. Zhang, R. Liu, X. Wang, H. Chen, and C. Li, “Boosted binary Harris hawks optimizer and feature selection,” Engineering with Computers, vol. 1, pp. 1–30, 2020.View at: Google Scholar
  48. H.-L. Chen, G. Wang, C. Ma, Z.-N. Cai, W.-B. Liu, and S.-J. Wang, “An efficient hybrid kernel extreme learning machine approach for early diagnosis of Parkinson’s disease,” Neurocomputing, vol. 184, pp. 131–144, 2016.View at: Publisher Site | Google Scholar
  49. L. Hu, G. Hong, J. Ma, X. Wang, and H. Chen, “An efficient machine learning approach for diagnosis of paraquat-poisoned patients,” Computers in Biology and Medicine, vol. 59, pp. 116–124, 2015.View at: Publisher Site | Google Scholar
  50. L. Shen, H. Chen, Z. Yu et al., “Evolving support vector machines using fruit fly optimization for medical data classification,” Knowledge-Based Systems, vol. 96, pp. 61–75, 2016.View at: Publisher Site | Google Scholar
  51. J. Xia, H. Chen, Q. Li et al., “Ultrasound-based differentiation of malignant and benign thyroid Nodules: an extreme learning machine approach,” Computer Methods and Programs in Biomedicine, vol. 147, pp. 37–49, 2017.View at: Publisher Site | Google Scholar
  52. C. Li, L. Hou, B. Y. Sharma et al., “Developing a new intelligent system for the diagnosis of tuberculous pleural effusion,” Computer Methods and Programs in Biomedicine, vol. 153, pp. 211–225, 2018.View at: Publisher Site | Google Scholar
  53. X. Zhao, X. Zhang, Z. Cai et al., “Chaos enhanced grey wolf optimization wrapped ELM for diagnosis of paraquat-poisoned patients,” Computational Biology and Chemistry, vol. 78, pp. 481–490, 2019.View at: Publisher Site | Google Scholar
  54. M. Wang and H. Chen, “Chaotic multi-swarm whale optimizer boosted support vector machine for medical diagnosis,” Applied Soft Computing Journal, vol. 88, 2020.View at: Publisher Site | Google Scholar
  55. X. Xu and H.-L. Chen, “Adaptive computational chemotaxis based on field in bacterial foraging optimization,” Soft Computing, vol. 18, no. 4, pp. 797–807, 2014.View at: Publisher Site | Google Scholar
  56. R. U. Khan, X. Zhang, R. Kumar, A. Sharif, N. A. Golilarz, and M. Alazab, “An adaptive multi-layer botnet detection technique using machine learning classifiers,” Applied Sciences, vol. 9, no. 11, p. 2375, 2019.View at: Publisher Site | Google Scholar
  57. A. Addeh, A. Khormali, and N. A. Golilarz, “Control chart pattern recognition using RBF neural network with new training algorithm and practical features,” ISA Transactions, vol. 79, pp. 202–216, 2018.View at: Publisher Site | Google Scholar
  58. N. Amiri Golilarz, H. Gao, R. Kumar, L. Ali, Y. Fu, and C. Li, “Adaptive wavelet based MRI brain image de-noising,” Frontiers in Neuroscience, vol. 14, p. 728, 2020.View at: Publisher Site | Google Scholar
  59. N. A. Golilarz, H. Gao, and H. Demirel, “Satellite image de-noising with Harris hawks meta heuristic optimization algorithm and improved adaptive generalized Gaussian distribution threshold function,” IEEE Access, vol. 7, pp. 57459–57468, 2019.View at: Publisher Site | Google Scholar
  60. M. Eisazadeh and J. Rezapour, “Multi-objective optimization of the composite sheets using PSO algorithm,” 2017.View at: Google Scholar
  61. I. Bargegol, M. Nikookar, R. V. Nezafat, E. J. Lashkami, and A. M. Roshandeh, “Timing optimization of signalized intersections using shockwave theory by genetic algorithm,” Computational Research Progress in Applied Science & Engineering, vol. 1, pp. 160–167, 2015.View at: Google Scholar
  62. B. Bai, Z. Guo, C. Zhou, W. Zhang, and J. Zhang, “Application of adaptive reliability importance sampling-based extended domain PSO on single mode failure in reliability engineering,” Information Sciences, vol. 546, pp. 42–59, 2021.View at: Publisher Site | Google Scholar
  63. J. Liu, C. Wu, G. Wu, and X. Wang, “A novel differential search algorithm and applications for structure design,” Applied Mathematics and Computation, vol. 268, pp. 246–269, 2015.View at: Publisher Site | Google Scholar
  64. X. Zhao, D. Li, B. Yang, C. Ma, Y. Zhu, and H. Chen, “Feature selection based on improved ant colony optimization for online detection of foreign fiber in cotton,” Applied Soft Computing, vol. 24, pp. 585–596, 2014.View at: Publisher Site | Google Scholar
  65. D. Zhao, “Chaotic random spare ant colony optimization for multi-threshold image segmentation of 2D Kapur entropy,” Knowledge-Based Systems, vol. 24, p. 106510, 2020.View at: Google Scholar
  66. H. Chen, A. A. Heidari, H. Chen, M. Wang, Z. Pan, and A. H. Gandomi, “Multi-population differential evolution-assisted Harris hawks optimization: framework and case studies,” Future Generation Computer Systems, vol. 111, pp. 175–198, 2020.View at: Publisher Site | Google Scholar
  67. J. Hu, H. Chen, A. A. Heidari et al., “Orthogonal learning covariance matrix for defects of grey wolf optimizer: insights, balance, diversity, and feature selection,” Knowledge-Based Systems, vol. 213, p. 106684, 2021.View at: Publisher Site | Google Scholar
  68. G. Sun, B. Yang, Z. Yang, and G. Xu, “An adaptive differential evolution with combined strategy for global numerical optimization,” Soft Computing, vol. 24, pp. 1–20, 2019.View at: Google Scholar
  69. G. Sun, C. Li, and L. Deng, “An adaptive regeneration framework based on search space adjustment for differential evolution,” Neural Computing and Applications, vol. 24, pp. 1–17, 2021.View at: Google Scholar
  70. A. Addeh and M. Iri, “Brain tumor type classification using deep features of MRI images and optimized RBFNN,” ENG Transactions, vol. 2, pp. 1–7, 2021.View at: Google Scholar
  71. F. J. Golrokh and A. Hasan, “A comparison of machine learning clustering algorithms based on the DEA optimization approach for pharmaceutical companies in developing countries,” Soft Computing, vol. 1, pp. 1–8, 2020.View at: Google Scholar
  72. H. Tyagi, P. Phelan, and R. Prasher, “Predicted efficiency of a low-temperature nanofluid-based direct absorption solar collector,” Journal of Solar Energy Engineering, vol. 131, no. 4, 2009.View at: Publisher Site | Google Scholar
  73. S. Rashidi, M. Bovand, and J. A. Esfahani, “Heat transfer enhancement and pressure drop penalty in porous solar heat exchangers: a sensitivity analysis,” Energy Conversion and Management, vol. 103, pp. 726–738, 2015.View at: Publisher Site | Google Scholar
  74. N. Akram, R. Sadri, S. N. Kazi et al., “A comprehensive review on nanofluid operated solar flat plate collectors,” Journal of Thermal Analysis and Calorimetry, vol. 139, no. 2, pp. 1309–1343, 2020.View at: Publisher Site | Google Scholar
Energy and exergy analysis of an enhanced solar CCHP system with a collector embedded by porous media and nano fluid

Energy and exergy analysis of an enhanced solar CCHP system with a collector embedded by porous media and nano fluid

Year 2021, Volume 7, Issue 6, 1489 – 1505, 02.09.2021

N. TONEKABONI  H. SALARIAN  M. Eshagh NIMVARI  J. KHALEGHINIA https://doi.org/10.18186/thermal.990897

Abstract

The low efficiency of Collectors that absorb energy can be mentioned as one of the drawbacks in solar cogeneration cycles. In the present study, solar systems have been improved by adding porous media and Nanofluid to collectors. One advantage of using porous media and nanomaterials is to absorb more energy while the surface area is reduced. In this study, first, solar collectors are enhanced using 90% porosity copper in solar combined cooling, heating and power systems (SCCHP). Second, different percentages of CuO and Al2O3 nano-fluids are added to a flat plate and parabolic collectors to enhance thermal properties. Simulations are performed in different modes (simple parabolic collectors, simple flat plate collectors, improved flat plate collectors, parabolic collectors with porous media, and flat plate and parabolic collectors with different density of CuO and Al2O3 nanofluids). A case study is investigated for warm and dry regions with mean solar radiation Ib = 820 w / m2 in Iran. The maximum energy and exergy efficiencies are 60.12% and 18.84%, respectively, that is related to enhanced parabolic solar collectors with porous media and nanofluids. Adding porous media and nano-fluids increases an average 14.4% collector energy efficiency and 8.08% collector exergy efficiency.

Keywords

Exergy analysisSolar cogeneration systemPorous mediaNanofluid

References

  • [1] Center TU. Annual report on China building energy efficiency. China Construction Industry Press (In Chinese). 2016.
  • [2] Tonekaboni N, Salarian H, Fatahian E, Fatahian H. Energy and exergy economic analysis of cogeneration cycle of homemade CCHP with PVT collector. Canadian Journal of Basic and Applied Sciences 2015;3:224-233.
  • [3] Hassan JM, Abdul-Ghafour QJ, Mohammed MF. CFD simulation of enhancement techniques in flat plate solar water collectors. Al-Nahrain Journal for Engineering Sciences 2017;20:751-761.
  • [4] Sopian K, Daud WR, Othman MY, Yatim B. Thermal performance of the double-pass solar collector with and without porous media. Renewable Energy 1999;18:557-564. https://doi.org/10.1016/S0960-1481(99)00007-5
  • [5] Feizbahr M, Kok Keong C, Rostami F, Shahrokhi M. Wave energy dissipation using perforated and non perforated piles. International Journal of Engineering 2018;31:212-219. https://doi.org/10.5829/ije.2018.31.02b.04
  • [6] Tian Y, Zhao CY. A review of solar collectors and thermal energy storage in solar thermal applications. Applied Energy 2013;104:538-553. https://doi.org/10.1016/j.apenergy.2012.11.051
  • [7] Wang F, Tan J, Wang Z. Heat transfer analysis of porous media receiver with different transport and thermophysical models using mixture as feeding gas. Energy Conversion and Management 2014;83:159-166. https://doi.org/10.1016/j.enconman.2014.03.068
  • [8] Korti AI. Numerical 3-D heat flow simulations on double-pass solar collector with and without porous media. Journal of Thermal Engineering 2015;1:10-23. https://doi.org/10.18186/jte.86295
  • [9] Sharma N, Diaz G. Performance model of a novel evacuated-tube solar collector based on minichannels. Solar Energy 2011;85:881-890. https://doi.org/10.1016/j.solener.2011.02.001
  • [10] Tyagi VV, Kaushik SC, Tyagi SK. Advancement in solar photovoltaic/thermal (PV/T) hybrid collector technology. Renewable and Sustainable Energy Reviews 2012;16:1383-1398. https://doi.org/10.1016/j.rser.2011.12.013
  • [11] Zhai H, Dai YJ, Wu JY, Wang RZ. Energy and exergy analyses on a novel hybrid solar heating, cooling and power generation system for remote areas. Applied Energy 2009;86:1395-1404. https://doi.org/10.1016/j.apenergy.2008.11.020
  • [12] Wang J, Dai Y, Gao L, Ma S. A new combined cooling, heating and power system driven by solar energy. Renewable Energy 2009;34:2780-2788. https://doi.org/10.1016/j.renene.2009.06.010
  • [13] Jing YY, Bai H, Wang JJ, Liu L. Life cycle assessment of a solar combined cooling heating and power system in different operation strategies. Applied Energy 2012;92:843-853. https://doi.org/10.1016/j.apenergy.2011.08.046
  • [14] Temir G, Bilge D. Thermoeconomic analysis of a trigeneration system. applied thermal engineering. Applied Thermal Engineering 2004;24:2689-2699. https://doi.org/10.1016/j.applthermaleng.2004.03.014
  • [15] Wang JJ, Jing YY, Zhang CF. Optimization of capacity and operation for CCHP system by genetic algorithm. Applied Energy 2010;87:1325-1335. https://doi.org/10.1016/j.apenergy.2009.08.005
  • [16] Kleinstreuer C, Chiang H. Analysis of a porous-medium solar collector. Heat Transfer Engineering 1990;11:45-55. https://doi.org/10.1080/01457639008939728
  • [17] Mbaye M, Bilgen E. Natural convection and conduction in porous wall, solar collector systems without vents. Jornal of Solar Energy Engineering 1992;114:40-46. https://doi.org/10.1115/1.2929980
  • [18] Hirasawa S, Tsubota R, Kawanami T, Shirai K. Reduction of heat loss from solar thermal collector by diminishing natural convection with high-porosity porous medium. Solar Energy 2013;97:305-313. https://doi.org/10.1016/j.solener.2013.08.035
  • [19] Jouybari HJ, Saedodin S, Zamzamian A, Nimvari ME, Wongwises S. Effects of porous material and nanoparticles on the thermal performance of a flat plate solar collector: an experimental study. Renewable Energy 2017;114:1407-1418. https://doi.org/10.1016/j.renene.2017.07.008
  • [20] Subramani J, Nagarajan PK, Wongwises S, El‐Agouz SA, Sathyamurthy R. Experimental study on the thermal performance and heat transfer characteristics of solar parabolic trough collector using Al2O3 nanofluids. Environmental Progress & Sustainable Energy 2018;37:1149-1159. https://doi.org/10.1002/ep.12767
  • [21] Yousefi T, Veysi F, Shojaeizadeh E, Zinadini S. An experimental investigation on the effect of Al2O3–H2O nanofluid on the efficiency of flat-plate solar collectors. Renewable Energy 2012;39:293-298. https://doi.org/10.1016/j.renene.2011.08.056
  • [22] Tyagi H, Phelan P, Prasher R. Predicted efficiency of a low-temperature nanofluid-based direct absorption solar collector. Journal of Solar Energy Engineering 2009;131:041004. https://doi.org/10.1115/1.3197562
  • [23] Shojaeizadeh E, Veysi F, Kamandi A. Exergy efficiency investigation and optimization of an Al2O3–water nanofluid based Flat-plate solar collector. Energy and Buildings 2015;101:12-23. https://doi.org/10.1016/j.enbuild.2015.04.048
  • [24] Tiwari AK, Ghosh P, Sarkar J. Solar water heating using nanofluids–a comprehensive overview and environmental impact analysis. International Journal of Emerging Technology and Advanced Engineering 2013;3:221-224. [25] Akram N, Sadri R, Kazi SN, Zubir MN, Ridha M, Ahmed W, et al. A comprehensive review on nanofluid operated solar flat plate collectors. Journal of Thermal Analysis and Calorimetry 2020;139:1309-1343. https://doi.org/10.1007/s10973-019-08514-z
  • [26] Lemington N. Study of solar driven adsorption cooling potential in Indonesia. Journal of Thermal Engineering 2017;3:1044-1051. https://doi.org/10.18186/thermal.290257
  • [27] Tong Y, Lee H, Kang W, Cho H. Energy and exergy comparison of a flat-plate solar collector using water, Al2O3 nanofluid, and CuO nanofluid. Applied Thermal Engineering 2019;159:113959. https://doi.org/10.1016/j.applthermaleng.2019.113959
  • [28] Khanafer K, Vafai K. A critical synthesis of thermophysical characteristics of nanofluids. International Journal of Heat And Mass Transfer 2011;54:4410-4428. https://doi.org/10.1016/j.ijheatmasstransfer.2011.04.048
  • [29] Farhana K, Kadirgama K, Rahman MM, Ramasamy D, Noor MM, Najafi G, et al. Improvement in the performance of solar collectors with nanofluids—A state-of-the-art review. Nano-Structures & Nano-Objects 2019;18:100276. https://doi.org/10.1016/j.nanoso.2019.100276
  • [30] Turkyilmazoglu M. Condensation of laminar film over curved vertical walls using single and two-phase nanofluid models. European Journal of Mechanics-B/Fluids 2017;65:184-91. https://doi.org/10.1016/j.euromechflu.2017.04.007
  • [31] Chen CC, Huang PC. Numerical study of heat transfer enhancement for a novel flat-plate solar water collector using metal-foam blocks. International Journal of Heat And Mass Transfer 2012;55:6734-6756. https://doi.org/10.1016/j.ijheatmasstransfer.2012.06.082
  • [32] Huang PC, Chen CC, Hwang HY. Thermal enhancement in a flat-plate solar water collector by flow pulsation and metal-foam blocks. International Journal of Heat and Mass Transfer 2013;61:696-720. https://doi.org/10.1016/j.ijheatmasstransfer.2013.02.037
  • [33] Hajipour M, Dehkordi AM. Mixed-convection flow of Al2O3–H O nanofluid in a channel partially filled with porous metal foam: experimental and numerical study. Experimental Thermal and Fluid Science 2014;53:49-56. https://doi.org/10.1016/j.expthermflusci.2013.11.002
  • [34] Rashidi S, Bovand M, Esfahani JA. Heat transfer enhancement and pressure drop penalty in porous solar heat exchangers: a sensitivity analysis. Energy Conversion and Management 2015;103:726-738. https://doi.org/10.1016/j.enconman.2015.07.019
  • [35] Manikandan GK, Iniyan S, Goic R. Enhancing the optical and thermal efficiency of a parabolic trough collector–A review. Applied Energy 2019;235:1524-1540. https://doi.org/10.1016/j.apenergy.2018.11.048

Details

Primary LanguageEnglish
SubjectsEngineering
Journal SectionArticles
AuthorsN. TONEKABONI  This is me
Islamic Azad University Nour Branch
0000-0002-1563-4407
IranH. SALARIAN  This is me (Primary Author)
Islamic Azad University Nour Branch
0000-0002-2161-0276
IranM. Eshagh NIMVARI  This is me
Amol University of Special Modern Technologies
0000-0002-7401-315X
IranJ. KHALEGHINIA  This is me
Islamic Azad University Nour Branch
0000-0001-5357-193X
Iran
Publication DateSeptember 2, 2021
Application DateDecember 28, 2020
Acceptance DateMay 9, 2020
Published in IssueYear 2021, Volume 7, Issue 6
Figure 15. Localized deformations on revetment due to run-down and sliding of armor from body laboratory model (left) and numerical modeling (right).

지속 가능한 해안 보호 구조로서 굴절식 콘크리트 블록 매트리스의 손상 메커니즘의 수치적 모델링

Numerical Modeling of Failure Mechanisms in Articulated Concrete Block Mattress as a Sustainable Coastal Protection Structure

Author

Ramin Safari Ghaleh(Department of Civil Engineering, K. N. Toosi University of Technology, Tehran 19967-15433, Iran)

Omid Aminoroayaie Yamini(Department of Civil Engineering, K. N. Toosi University of Technology, Tehran 19967-15433, Iran)

S. Hooman Mousavi(Department of Civil Engineering, K. N. Toosi University of Technology, Tehran 19967-15433, Iran)

Mohammad Reza Kavianpour(Department of Civil Engineering, K. N. Toosi University of Technology, Tehran 19967-15433, Iran)

Abstract

해안선 보호는 전 세계적인 우선 순위로 남아 있습니다. 일반적으로 해안 지역은 석회암과 같은 단단하고 비자연적이며 지속 불가능한 재료로 보호됩니다. 시공 속도와 환경 친화성을 높이고 개별 콘크리트 블록 및 보강재의 중량을 줄이기 위해 콘크리트 블록을 ACB 매트(Articulated Concrete Block Mattress)로 설계 및 구현할 수 있습니다. 이 구조물은 필수적인 부분으로 작용하며 방파제 또는 해안선 보호의 둑으로 사용할 수 있습니다. 물리적 모델은 해안 구조물의 현상을 추정하고 조사하는 핵심 도구 중 하나입니다. 그러나 한계와 장애물이 있습니다. 결과적으로, 본 연구에서는 이러한 구조물에 대한 파도의 수치 모델링을 활용하여 방파제에서의 파도 전파를 시뮬레이션하고, VOF가 있는 Flow-3D 소프트웨어를 통해 ACB Mat의 불안정성에 영향을 미치는 요인으로는 파괴파동, 옹벽의 흔들림, 파손으로 인한 인양력으로 인한 장갑의 변위 등이 있다. 본 연구의 가장 중요한 목적은 수치 Flow-3D 모델이 연안 호안의 유체역학적 매개변수를 모사하는 능력을 조사하는 것입니다. 콘크리트 블록 장갑에 대한 파동의 상승 값은 파단 매개변수( 0.5 < ξ m – 1 , 0 < 3.3 )가 증가할 때까지(R u 2 % H m 0 = 1.6) ) 최대값에 도달합니다. 따라서 차단파라미터를 증가시키고 파괴파(ξ m − 1 , 0 > 3.3 ) 유형을 붕괴파/해일파로 변경함으로써 콘크리트 블록 호안의 상대파 상승 변화 경향이 점차 증가합니다. 파동(0.5 < ξ m − 1 , 0 < 3.3 )의 경우 차단기 지수(표면 유사성 매개변수)를 높이면 상대파 런다운의 낮은 값이 크게 감소합니다. 또한, 천이영역에서는 파단파동이 쇄도파에서 붕괴/서징으로의 변화( 3.3 < ξ m – 1 , 0 < 5.0 )에서 상대적 런다운 과정이 더 적은 강도로 발생합니다.

Shoreline protection remains a global priority. Typically, coastal areas are protected by armoring them with hard, non-native, and non-sustainable materials such as limestone. To increase the execution speed and environmental friendliness and reduce the weight of individual concrete blocks and reinforcements, concrete blocks can be designed and implemented as Articulated Concrete Block Mattress (ACB Mat). These structures act as an integral part and can be used as a revetment on the breakwater body or shoreline protection. Physical models are one of the key tools for estimating and investigating the phenomena in coastal structures. However, it does have limitations and obstacles; consequently, in this study, numerical modeling of waves on these structures has been utilized to simulate wave propagation on the breakwater, via Flow-3D software with VOF. Among the factors affecting the instability of ACB Mat are breaking waves as well as the shaking of the revetment and the displacement of the armor due to the uplift force resulting from the failure. The most important purpose of the present study is to investigate the ability of numerical Flow-3D model to simulate hydrodynamic parameters in coastal revetment. The run-up values of the waves on the concrete block armoring will multiply with increasing break parameter ( 0.5 < ξ m − 1 , 0 < 3.3 ) due to the existence of plunging waves until it ( R u 2 % H m 0 = 1.6 ) reaches maximum. Hence, by increasing the breaker parameter and changing breaking waves ( ξ m − 1 , 0 > 3.3 ) type to collapsing waves/surging waves, the trend of relative wave run-up changes on concrete block revetment increases gradually. By increasing the breaker index (surf similarity parameter) in the case of plunging waves ( 0.5 < ξ m − 1 , 0 < 3.3 ), the low values on the relative wave run-down are greatly reduced. Additionally, in the transition region, the change of breaking waves from plunging waves to collapsing/surging ( 3.3 < ξ m − 1 , 0 < 5.0 ), the relative run-down process occurs with less intensity.

Figure 1.  Armor  geometric  characteristics  and  drawing  three-dimensional  geometry  of  a  breakwater section  in SolidWorks software.
Figure 1. Armor geometric characteristics and drawing three-dimensional geometry of a breakwater section in SolidWorks software.
Figure  5.  Wave  overtopping on  concrete block  mattress in (a)  laboratory  and (b)  numerical  model.
Figure 5. Wave overtopping on concrete block mattress in (a) laboratory and (b) numerical model.
Figure  7.  Mesh  block  for  calibrated  numerical  model  with  686,625  cells  and  utilization  of  FAVOR  tab to assess figure geometry.
Figure 7. Mesh block for calibrated numerical model with 686,625 cells and utilization of FAVOR tab to assess figure geometry.
Figure  10.  How to place different layers  (core, filter,  and revetment)  of the structure on slope.
Figure 10. How to place different layers (core, filter, and revetment) of the structure on slope.

Suggested Citation

Figure 11. Wave run-up on ACB Mat blocks in (a) laboratory model and (b) numerical modeling.
Figure 11. Wave run-up on ACB Mat blocks in (a) laboratory model and (b) numerical modeling.
Figure  15.  Localized  deformations  on  revetment  due  to  run-down  and  sliding  of  armor  from  body  laboratory  model  (left) and  numerical  modeling (right).
Figure 15. Localized deformations on revetment due to run-down and sliding of armor from body laboratory model (left) and numerical modeling (right).

References

  1. Capobianco, V.; Robinson, K.; Kalsnes, B.; Ekeheien, C.; Høydal, Ø. Hydro-Mechanical Effects of Several Riparian Vegetation Combinations on the Streambank Stability—A Benchmark Case in Southeastern Norway. Sustainability 2021, 13, 4046. [CrossRef]
  2. MarCom Working Group 113. PIANC Report No 113: The Application of Geosynthetics in Waterfront Areas; PIANC: Brussels, Belgium, 2011; p. 113, ISBN 978-2-87223-188-1.
  3. Hunt, W.F.; Collins, K.A.; Hathaway, J.M. Hydrologic and Water Quality Evaluation of Four Permeable Pavements in North Carolina, USA. In Proceedings of the 9th International Conference on Concrete Block Paving, Buenos Aires, Argentina, 18–21 October 2009.
  4. Kirkpatrick, R.; Campbell, R.; Smyth, J.; Murtagh, J.; Knapton, J. Improvement of Water Quality by Coarse Graded Aggregates in Permeable Pavements. In Proceedings of the 9th International Conference on Concrete Block Paving, Buenos Aires, Argentina, 18–21 October 2009.
  5. Chinowsky, P.; Helman, J. Protecting Infrastructure and Public Buildings against Sea Level Rise and Storm Surge. Sustainability 2021, 13, 10538. [CrossRef]
  6. Breteler, M.K.; Pilarczyk, K.W.; Stoutjesdijk, T. Design of alternative revetments. Coast. Eng. 1998 1999, 1587–1600. [CrossRef]
  7. Pilarczyk, K.W. Design of Revetments; Dutch Public Works Department (Rws), Hydraulic Engineering Division: Delft, The Netherlands, 2003.
  8. Hughes, S.A. Combined Wave and Surge Overtopping of Levees: Flow Hydrodynamics and Articulated Concrete Mat Stability; Engineer Research and Development Center Vicksburg Ms Coastal and Hydraulics Lab: Vicksburg, MS, USA, 2008.
  9. Gier, F.; Schüttrumpf, H.; Mönnich, J.; Van Der Meer, J.; Kudella, M.; Rubin, H. Stability of Interlocked Pattern Placed Block Revetments. Coast. Eng. Proc. 2012, 1, Structures-46. [CrossRef]
  10. Najafi, J.A.; Monshizadeh, M. Laboratory Investigations on Wave Run-up and Transmission over Breakwaters Covered by Antifer Units; Scientia Iranica: Tehran, Iran, 2010.
  11. Oumeraci, H.; Staal, T.; Pförtner, S.; Ludwigs, G.; Kudella, M. Hydraulic Performance, Wave Loading and Response of Elastocoast Revetments and their Foundation—A Large Scale Model Study; Leichtweiß Institut für Wasserbau: Braunschweig, Germany, 2010.
  12. Tripathy, S.K. Significance of Traditional and Advanced Morphometry to Fishery Science. J. Hum. Earth Future 2020, 1, 153–166. [CrossRef]
  13. Nut, N.; Mihara, M.; Jeong, J.; Ngo, B.; Sigua, G.; Prasad, P.V.V.; Reyes, M.R. Land Use and Land Cover Changes and Its Impact on Soil Erosion in Stung Sangkae Catchment of Cambodia. Sustainability 2021, 13, 9276. [CrossRef]
  14. Xu, C.; Pu, L.; Kong, F.; Li, B. Spatio-Temporal Change of Land Use in a Coastal Reclamation Area: A Complex Network Approach. Sustainability 2021, 13, 8690. [CrossRef]
  15. Mousavi, S.; Kavianpour, H.M.R.; Yamini, O.A. Experimental analysis of breakwater stability with antifer concrete block. Mar. Georesour. Geotechnol. 2017, 35, 426–434. [CrossRef]
  16. Yamini, O.; Aminoroayaie, S.; Mousavi, H.; Kavianpour, M.R. Experimental Investigation of Using Geo-Textile Filter Layer In Articulated Concrete Block Mattress Revetment On Coastal Embankment. J. Ocean Eng. Mar. Energy 2019, 5, 119–133. [CrossRef]
  17. Ghasemi, A.; Far, M.S.; Panahi, R. Numerical Simulation of Wave Overtopping From Armour Breakwater by Considering Porous Effect. J. Mar. Eng. 2015, 11, 51–60. Available online: http://dorl.net/dor/20.1001.1.17357608.1394.11.22.8.4 (accessed on 21 October 2021).
  18. Nourani, O.; Askar, M.B. Comparison of the Effect of Tetrapod Block and Armor X block on Reducing Wave Overtopping in Breakwaters. Open J. Mar. Sci. 2017, 7, 472–484. [CrossRef]
  19. Aminoroaya, A.O.; Kavianpour, M.R.; Movahedi, A. Performance of Hydrodynamics Flow on Flip Buckets Spillway for Flood Control in Large Dam Reservoirs. J. Hum. Earth Future 2020, 1, 39–47.
  20. Milanian, F.; Niri, M.Z.; Najafi-Jilani, A. Effect of hydraulic and structural parameters on the wave run-up over the berm breakwaters. Int. J. Nav. Archit. Ocean Eng. 2017, 9, 282–291. [CrossRef]
  21. Yamini, O.A.; Kavianpour, M.R.; Mousavi, S.H. Experimental investigation of parameters affecting the stability of articulated concrete block mattress under wave attack. Appl. Ocean Res. 2017, 64, 184–202. [CrossRef]
  22. Yakhot, V.; Orszag, S.A.; Thangam, S.; Gatski, T.B.; Speziale, C.G. Development of turbulence models for shear flows by a double expansion technique. Phys. Fluids 1992, 4, 1510–1520. [CrossRef]
  23. Bayon, A.; Valero, D.; García-Bartual, R.; López-Jiménez, P.A. Performance assessment of OpenFOAM and FLOW-3D in the numerical modeling of a low Reynolds number hydraulic jump. Environ. Model. Softw. 2016, 80, 322–335. [CrossRef]
  24. Jin, J.; Meng, B. Computation of wave loads on the superstructures of coastal highway bridges. Ocean Eng. 2011, 38, 2185–2200. [CrossRef]
  25. Yang, S.; Yang, W.; Qin, S.; Li, Q.; Yang, B. Numerical study on characteristics of dam-break wave. Ocean Eng. 2018, 159, 358–371. [CrossRef]
  26. Ersoy, H.; Karahan, M.; Geli¸sli, K.; Akgün, A.; Anılan, T.; Sünnetci, M.O.; Yah¸si, B.K. Modelling of the landslide-induced impulse waves in the Artvin Dam reservoir by empirical approach and 3D numerical simulation. Eng. Geol. 2019, 249, 112–128. [CrossRef]
  27. Zhan, J.M.; Dong, Z.; Jiang, W.; Li, Y.S. Numerical simulation of wave transformation and runup incorporating porous media wave absorber and turbulence models. Ocean Eng. 2010, 37, 1261–1272. [CrossRef]
  28. Owen, M.W. The Hydroulic Design of Seawall Profiles, Proceedings Conference on Shoreline Protection; ICE: London, UK, 1980; pp. 185–192.
  29. Pilarczyk, K.W. Geosythetics and Geosystems in Hydraulic and Coastal Engineering; CRC Press: Balkema, FL, USA, 2000; p. 913, ISBN 90.5809.302.6.
  30. Van der Meer, J.W.; Allsop, N.W.H.; Bruce, T.; De Rouck, J.; Kortenhaus, A.; Pullen, T.; Schüttrumpf, H.; Troch, P.; Zanuttigh, B. (Eds.) Manual on Wave Overtopping of Sea Defences and Related Structures–Assessment Manual; EurOtop.: London, UK, 2016; Available online: www.Overtopping-manual.com (accessed on 21 October 2021).
  31. Battjes, J.A. Computation of Set-up, Longshore Currents, Run-up and Overtopping Due to Wind-Generated Waves; TU Delft Library: Delft, The Netherlands, 1974.
  32. Van der Meer, J.W. Rock Slopes and Gravel Beaches under Wave Attack; Delft Hydraulics: Delft, The Netherlands, 1988.
  33. Ten Oever, E. Theoretical and Experimental Study on the Placement of Xbloc