Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Renewable and Sustainable Energy Reviews 174 (2023) 113124

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews


journal homepage: www.elsevier.com/locate/rser

A review of computational methods for studying oscillating water columns


– the Navier-Stokes based equation approach
F. Opoku *, M.N. Uddin, M. Atkinson
Department of Mechanical Engineering, North Carolina A&T State University, Greensboro, NC, 27411, USA

A R T I C L E I N F O A B S T R A C T

Keywords: This review evaluates the state-of-the-practice numerical tools used to predict the performance of Oscillating
Oscillating water column Water Column (OWC). The OWC is a widely studied Wave Energy Converter that provides a reliable form of
Wave energy converter renewable form of electricity that can potentially meet global energy needs. However, the fluid-flow phenomena
Marine renewable energy
affecting its hydrodynamic performance are not fully understood. While there has been the successful full-scale
Computational fluid dynamics
Reynolds Averaged Navier Stokes
deployment of OWCs, various computational methods are being explored to optimize this technology. Potential
Large-Eddy simulation flow theory is commonly used to evaluate the efficiency of OWCs; however, this assumption tends to over-predict
ANSYS Fluent the hydrodynamic performance. Recently, numerical studies using a diverse set of commercial, open-source, or
in-house Computational Fluid Dynamics (CFD) using Reynolds Averaged Navier Stokes (RANS) and Large-Eddy
Simulation codes show a better comparison to available experimental results but are computationally expensive.
ANSYS Fluent was found to be the most widely used CFD code applied to the study of the OWCs, with a high
degree of accuracy in terms of experimental validation of numerical results.

creating bi-direction airflow in and out of the chamber. The bi-


directional airflow is commonly referred to as inhalation and exhala­
1. Introduction tion. Inhalation occurs when the oscillating air inside the OWC is less
than the pressure of the atmosphere, while exhalation occurs when it is
Renewable energy can provide consistent eco-friendly electricity greater [5]. Pneumatic power is transferred to the air within the
that addresses climate change issues and helps preserve our planet [1]. chamber, which drives the PTO system (i.e., air turbine-generator). The
Wave energy is a renewable and sustainable energy form that results air turbine-generator then converts the pneumatic power into elec­
from the radiant energy from the sun. The earth’s surface is subjected to tricity. The OWC is likely the most studied WEC; however, it has not
varying energy intensities from the sun. That causes air to flow, which reached technological maturity since its hydrodynamics are not well
forms winds that blow over the surface of the oceans to create waves. understood [6]. The primary design challenge for an OWC is the
The wind blows with sufficient consistency and force, ensuring contin­ chamber and duct [7], as shown in Fig. 1, which shows efforts have been
uous wave motion. This wave energy has tremendous potential for en­ made to maximize the pneumatic power from these devices [8]. Planes
ergy generation. Ref. [2] cited 29,500 TWh/yr as the global total 1, 2, and 3 represent the inlet, midpoint, and exit sections where flow­
theoretical potential for wave energy, and Ref. [3] stated that 10–20% of field parameters are measured in the air-duct. There have been studies
this amount can be exploited for useful energy production. Wave energy carried out to optimize the OWC Chamber [9–12] and the air-duct [13,
devices like the OWC extract energy from the surface motion of these 14] to increase the efficiency of energy conversion (i.e., kinetic energy to
waves. The energy intensity of water waves is affected by the wave­ pneumatic power). To maximize wave energy extraction, Ref. [15]
length, wave frequency, and amplitude; consequently, ocean waves with noted that the resonance frequency of the OWC device should be
a high wave height and period possess the highest energy flux [4]. The approximately equal to that of the incident water wave. Furthermore,
energy flux of a traveling wave decreases as the waves move from the the air turbine commonly used, the Wells turbine, is shown in Fig. 2(i),
deep sea to the shores due to the shoaling effect. which is self-rectifying [16] by rotating in an uni-directional manner
The OWC comprises a chamber that harnesses power from the through a cyclical bi-directional airflow. Fig. 2 shows other form of air
incoming ocean waves transferred to the trapped air. The oscillating turbines used in OWC systems. A Modified Dennis-Auld turbine was
water waves cause the air column inside the chamber to act like a piston

* Corresponding author.
E-mail address: fopoku@aggies.ncat.edu (F. Opoku).

https://doi.org/10.1016/j.rser.2022.113124
Received 6 March 2022; Received in revised form 18 November 2022; Accepted 20 December 2022
Available online 30 December 2022
1364-0321/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

Abbreviations PCR Piecewise Constant Reconstruction


PLRC Piecewise Linear Reconstruction Calculation
BEM Boundary Element Method PTO Power Take-Off
BIEM Boundary Integral Equation Model RANS Reynold Averaged Navier Stokes
CFD Computational Fluid Dynamics REWEC1 Resonant Wave Energy Converter
CICSAM Compressive Interface Capturing Scheme for Arbitrary RSM Reynold Stress Model
Meshes SAS Scale-Adaptive Simulation
CV Control Volume SDR Specific Dissipation Rate
CWR Capture Width Ratio SMMC Surface Marker and Micro Cell
DES Detached Eddy Simulation SOR Second-Order Reconstruction
DNS Direct Numerical Simulation SPH Smoothed Particle Hydrodynamics
DOF Degree of Freedom SSR Stair-Stepped Reconstruction
FDM Finite Difference Method STACS Switching Technique for Advection and Capturing of
FVM Finite Volume Method Surfaces
HRIC High-Resolution Interface Capturing SWENSE Spectral Wave Explicit Navier-Stokes Equations
LBM Lattice Boltzmann Method TDR Turbulent Dissipation Rate
LBE Lattice Boltzmann Equation TEV Turbulent Eddy Viscosity
LES Large-Eddy Simulation TKE Turbulent Kinetic Energy
MAC Marker-and-Cell TPN Turbulent Prandtl Number
MEL Mixed Eulerian-Lagrangian UDF User Defined Function
NCAT North Carolina A&T State University URANS Unsteady Reynolds Averaged Navier-Stokes
NSE Navier-Stokes Equations VOF Volume of Fluid
NWT Numerical Wave Tank WEC Wave Energy Converter
OpenFOAM Open-source Field Operation and Manipulation WENO Weighted Essentially Non-Oscillatory
OWC Oscillating Water Column

Fig. 1. Schematic of a Novel OWC and Air-duct showing Sectional Plane.

2
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

Fig. 2. Air Turbines used in OWCs Applications (i). Wells Turbine (ii). Dennis-Auld Turbine (iii). Modified Dennis-Auld Turbine and (iv). Impulse Turbine.

system, CFD can have high computational costs but tend to be more
accurate than other simplified fluid models (e.g., potential flow) [6].
Most CFD studies of OWCs are carried out to perform parametric studies
to optimize the device [19]. Researchers have used experiments and CFD
to understand the flow topology and its interaction with OWC devices
[24,25] and wave tanks [26]. A schematic chart provided by Ref. [27]
compares the accuracy and computational cost for solving the NSE by
numerical methods. DNS solves the full NSE, but it is the most compu­
tationally expensive. In DNS, all scales and dynamics in the temporal
and spatial range are resolved. The other models on the chart are Linear
Cummins, efficient non-linear hydrodynamic, RANS, Hybrid RANS, and
LES, in increasing order of accuracy and computational requirements.
Ref. [27] noted that high computational costs restrain DNS from being
used to model the OWC system and other WEC devices. LES intends to
reduce the computational expense by modeling the smallest length
scales resolved using DNS. However, it remains too computationally
costly when used for parametric design studies of WECs. RANS modeling
of WECs has increased in popularity and has shown good comparison to
Fig. 3. Schematic of an OWC. experimental data. The present paper discusses common CFD ap­
proaches, and methodologies used to investigate OWCs under different
investigated by Ref. [17] and compared to the Wells turbine for per­ flow conditions.
formance. OWCs are broadly classified into (i) Fixed Structure OWC and
(ii) Floating OWC. The floating structure OWC, as shown in Fig. 3 is 2. The fundamental equations
designed for offshore applications. That configuration has the disad­
vantage of having electrical equipment and cables exposed to corrosive When the fluid is considered a continuum, the conservation laws (i.
saltwater; installation and maintenance are complex [18]. On the other e., Mass, Momentum, and Energy) are solved.
hand, the fixed structure OWC is designed for onshore or near-shore
applications where electrical equipment does not encounter ocean
2.1. Conservation of mass
water but requires electrical lines to transport power to the grid.
WECs have been studied using CFD [19], non-linear potential flow
The principle of conservation of mass is a statement that expresses
[20], or potential flow models such as (i). spectral-domain [21], (ii).
the fact that mass is conserved in a non-nuclear continuum. The sum of
time-domain [22], or iii). frequency-domain [23]. CFD models, which
masses flowing in and out of the continuum for a unit time is typically
are the focus of this review paper, involve solving the Navier-Stokes or
the same as the change in mass as a result of the change in density per
Euler equations numerically. When applied to the study of the OWC
unit time [28]. The continuity equation for an unsteady fluid flow in

3
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

vector notation in the general form is given by: 3. Computational modeling


Dρ →
+ ρ div U = 0 (1) Numerical studies using Potential Flow models such as frequency-
Dt
domain analysis [4,31–34], time-domain [35,36], spectral-domain
→ [37], and fully non-linear potential flow [38] in the inviscid flow
where ρ, t, and U represent fluid density, flow time, and the velocity
regime for the full or part of the flow domain have been carried out.
field, respectively. For an incompressible fluid flow, the density is
Potential flow theory tends to over-predict the efficiency of the device
assumed to be constant; therefore, the total derivative of mass is equal to
[39], which explains why researchers tend to use CFD and experiments
zero (i.e., Dρ/Dt = 0). Therefore, from equation (1), the equation of
to evaluate the performance of OWCs accurately. The authors found
continuity for an incompressible flow is expressed as [29]:
different methodologies to compute the flow field of an OWC. The

div U = 0 (2) following sections will describe various numerical studies describing
CFD approaches.
The continuity equation of the airflow inside the OWC can be
expressed in the general form as equation (1). In some analyses where
3.1. Inviscid simulations
the airflow is assumed incompressible due to negligible density changes,
equation (2) can be applied. Also, since the flow in the OWC is biphasic,
Inviscid flow refers to the flow of fluid where the influence of vis­
equation (2) is applied to the water flow.
cosity is zero or very minimal. When the viscosity is neglected, the shear
stress tensor drops out of the momentum equation (3), leading to the
2.2. Conservation of momentum Euler equation shown in equation (5). The Euler equation and the Po­
tential Flow models assume that the flow is inviscid. While potential
The momentum equation is a statement of Newton’s Second Law that flow assumes that the flow is irrotational and ideal, the Euler equation
expresses the force exerting on the fluid element to its rate of change of does not assume that the flow is irrotational and, therefore, can support
momentum. Considering a general Newtonian viscous fluid, the con­ vorticity [27]. A flow is said to be irrotational when the vorticity is zero.
servation of momentum equation, which is the so-called NSE, is Substituting μ = 0 in equation (3) yields the Euler equation in the vector
expressed in vector notation as [30]: notation as in equation (5) below:
[ ( ) →

→ ( →)T ( →) ] DU
ρ
DU
= ρ→
g − ∇p + ∇. μ ∇ U + ∇ U + λ ∇. U I (3) ρ = ρ→
g − ∇p (5)
Dt Dt
An NWT was used by Refs. [24,25] to study an offshore OWC using
where → g , ∇p, μ, I, and λ represent gravitational acceleration, pressure an inviscid model (Euler Equation) in ANSYS Fluent with the VOF
gradient, viscosity coefficient, identity tensor, and the bulk viscosity method. Ref. [24] studied six different OWC configurations to examine
coefficient (λ = - 23 μ), respectively. the flow topology. It concluded that the OWC configuration with a
waterside chamber angle, θ1 = 45o , shoreside chamber angle, θ2 = 90o ,
and immersion depth of d = 0.0 m resulted in the highest average power
2.3. Conservation of energy and efficiency. The study further found that the static pressure inside the
duct was the primary parameter that contributed to higher pneumatic
The conservation of energy relates the energy of the system as con­ power. Also, when added to the geometry, an immersion depth lowered
stant. It states that the sum of the heat added and the work done on the the total air pressure and velocity inside the duct. In Ref. [25], four cases
system results in a gain in total energy [30]. Mathematically, it is stated of the novel OWC device were investigated, and the optimal device was
as: attained. The resonant state of the device was found to occur when the
Dh Dp ( →)[ ( → ( →)T ) ( →) ] length of the chamber was half of the chamber water height. In another
ρ = + div(kc ∇T) + ∇ U μ ∇ U + ∇ U + λ ∇. U I study, CFD was applied to model a two-dimensional OWC using the
Dt Dt
incompressible, unsteady Euler equations coupled to the VOF method
(4)
employing a dynamic mesh [40]. A piston-type wavemaker was applied
where h, p, kc , ∇T represent the enthalpy, pressure, thermal conduc­ to generate the wave in the NWT using a UDF to prescribe the desired
tivity, and temperature gradient, respectively. wave amplitude and frequency. The study investigated the consequence
In some analyses, the fluid flow in the OWC system is assumed to be of changing wave height on the device’s efficiency. Varying the ampli­
incompressible, meaning, ρ = constant. This assumption allows the NSE tude from 0.002 m to 0.008 m in an accretion of 0.002 m showed that
to be solved numerically using a set of four dependent equations: the the device’s efficiency increased steadily between 0.002 m and 0.006 m,
momentum equation in the x, y, z spatial space, and the continuity reaching a maximum of 67%. The efficiency decreased between 0.006 m
equation. The unknown parameters (i.e., p, u, v, w) can be computed and 0.008 m, reaching a minimum of 50% at 0.008 m.
without including the energy equation. However, the incompressibility
assumption can lead to errors when applied to modeling the air dy­ 3.2. Viscous simulations
namics in a full-scale OWC chamber. Therefore, air compressibility must
be considered to model the full-scale device accurately, leading to Viscous CFD simulations include the effects of viscosity. If the Co­
thermodynamic consideration [27]. In considering air compressibility, efficient of viscosity is assumed to be constant, equation (3) reduces to:
the NSE and continuity equation will yield five unknown parameters (i. →
e., p, ρ, u, v, w); therefore, the conservation of energy introduces tem­ DU →
ρ = ρ→
g − ∇p + μ∇2 U (6)
perature (T) as an unknown quantity leading to five equations with six Dt
unknowns. The equation of state is introduced to provide six equations These simulations can be laminar, turbulent, or hybrid. Most nu­
with six unknowns for closure [27]. merical viscous simulations assume that the flow is fully turbulent;

4
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

however, in Ref. [41], an investigation was carried out to determine the derivation of the transport equation for k emanates from an exact
performance of an OWC chamber with and without a stepped bottom. equation representing physical considerations; on the other hand, the
Two NWTs were constructed, with NWT-1 assuming fully laminar flow transport equation for ε was arrived at through reasoning. This model
and NWT-2 assuming fully turbulent flow, to study flow topology using assumes the flow to be fully turbulent in most cases, and molecular
the k-ε turbulence model. The CFD results for both cases showed good viscosity is negligible [46]. The transport equation in vector notation for
agreement with linear theory for incident wave-free surface elevation k and ε are as follows:
for fully developed flow. However, there was disagreement on the free ( → ) [( ) ]
∂(ρk) μ
surface elevation found in the OWC device for the turbulent simulations + ∇. ρ U k = ∇. μ + t ∇k + Gk + Gb − ρε − YM + Sk (9)
∂t σk
when the flow has not attained a fully developed state. Additionally,
comparing the laminar results showed good agreement with a previous ( →) [( ) ]
∂(ρε) μt ε ε2
study using the BIEM and an analytical method. However, due to vis­ + ∇. ρ U ε = ∇. μ+ ∇ε + C1ε (Gk + C3ε Gb ) − C2ε ρ
∂t σε k k
cosity in the laminar and turbulent flow computations, the CFD results
showed a lower peak efficiency than the BIEM (i.e., potential flow the­ + Sε
ory) and the analytical work in Ref. [42]. (10)

3.2.1. Turbulent simulation where Gk , Gb , and YM represent the production of k due to mean velocity
There has not been a universally accepted definition of turbulent shear, the production of k due to buoyancy, and the fluctuating expan­
flow. The Reynolds number (Re) has been used to characterize flow sion in compressible turbulence compared to total TDR, respectively.
2
change from laminar to turbulent. If a flow is considered in the turbulent The TEV is modeled as μt = ρCμ kε , where Cμ is a constant value and C1ε ,
regime, it is expected to have a Re larger than the upper critical number C2ε , C3ε are constant terms known as the model coefficients. σε and σk
[29]. Also, turbulent flow is typically characterized by properties such as are TPNs for ε and k, respectively. The model coefficients terms and the
unsteadiness, three-dimensionality, high vorticity, rotational, irregular, TPN have default values of C1ε = 1.44, C2ε = 1.92, C3ε determines the
chaotic, increased diffusivity, large Re, dissipative, broad range of time degree to which the TDR (ε) is affected by buoyancy, Cμ = 0.09, σ ε = 1.3
scales, and length fluctuation. However, the flow is generally treated as and σk = 1.0 as stated in Ref. [46]. Sε and Sk are user-defined source
a continuum. terms for ε and k, respectively.
ANSYS Fluent simulations using the k-ε model has been successfully
3.2.1.1. Reynolds Averaged Navier-Stokes. The RANS equation is a time- used to model and analyze the OWC system. These studies were vali­
averaged equation used to model the turbulent motion of fluids. It is dated by experimental results, for example, Ref. [48] used ANSYS Fluent
applied by separating parameters into their time-average and fluctu­ to formulate an NWT to investigate the absorption characteristics for
ating components [43]. Since RANS models do not directly resolve both highly non-linear steep and linear waves of an OWC. The study’s main
the small and large turbulent structures, it reduces the computational objective was to investigate the air turbine for optimum pneumatic
requirements compared to LES; hence, larger time steps and mesh damping when non-linear waves were deployed. The results were vali­
spacing can be used [27]. RANS modeling is the most common method dated with analytical data for the pneumatic damping coefficients using
to resolve a turbulent flowfield [44]. The RANS model given in vector linear waves, and it was found that there was good agreement with the
notation is: CFD results. When considering non-linear waves, the results showed that
( →) the absorption efficiency of the device decreases as the wave height
[ ( )]
∂ ρU ( →→) → ( →)T 2 ( →) increases. Using linear wave theory showed that the maximum absorp­
+ ∇. ρ U U = − ∇p + ∇. μ ∇ U + ∇ U − ∇.U I tion efficiency of the wave energy for highly non-linear waves occurs
∂t 3
( →’ →’ ) when the pneumatic damping coefficient is higher than the CFD results.
− ∇. ρ U U + ρ→ g Ref. [49] used ANSYS Fluent to compare the numerical results to
(7) available analytical solutions for an OWC chamber. The numerical re­
sults compared favorably well with the analytical results after the initial
→′ →′ transient wave elevation had diminished, and there was a good agree­
where U and U represent the fluctuating and mean fluctuating velocity
ment when the wave profile at t = (23 × incident wave period) was
→′ →′
field. The term − ρ U U is referred to as the Reynolds stresses, which compared. In a different study in Ref. [50], ANSYS Fluent was used to
represent the shear stresses in the fluid layers accelerating and decel­ compare the experimental and numerical results of an integrated
erating due to the exchange of momentum [45]. The Reynolds stresses Caisson-breakwater OWC chamber. The incident wave period, still
terms must be computed to close the RANS equation. A popular method water depth, wave heights, and air duct diameters were observed to
for modeling the Reynolds stress is the Boussinesq hypothesis. It relates affect the relative height inside the integrated system. Ref. [51] used
it to the average velocity gradients as below [46]: ANSYS Fluent to construct a two-dimensional NWT to predict the device
( ) efficiency when subjected to linear and non-linear waves. The results
→’ →’ → ( →)T 2 2 ( →) showed a substantial increase in OWC hydrodynamic capture efficiency
− ρ U U = μt ∇ U + ∇ U − ρkI − ∇.U I (8)
3 3 with an increasing wave height of fully non-linear waves. Also, ANSYS
Fluent was used to investigate the power capture efficiency of a floating
where μt and k represent the eddy viscosity and TKE, respectively. These
heave-only OWC with an undamped mooring system. The results
additional terms introduced in the RANS equations give rise to the
showed that the power capture efficiency was affected by the incident
various turbulence models discussed in section 3.2.2.
wave, elasticity coefficient, pneumatic damping, wave frequency and
mooring of the turbine, and the mooring spring elasticity coefficient
3.2.2. The transport RANS equations
[52]. The effect of shape parameters and incident wave conditions on
OWC converting efficiency was investigated in Ref. [53]. This study
3.2.2.1. The k-ε turbulence model. The k-ε model developed in Ref. [47], concluded that the three-dimensional NWT is a better prediction tool for
introduced the transport equation for the TKE (k) and the TDR (ε). The analyzing the OWC air chamber than the two-dimensional flume.

5
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

Similarly, Ref. [54], reached the same general conclusion. However, are as follows:
Ref. [55] investigated the interaction of airflow and wave elevation in ( → ) [( ) ]
∂(ρk) μt
the column of the OWC. The k-ε turbulence model was used to study the + ∇. ρ U k = ∇. μ+ ∇k + Gk + Gb − ρε − YM + Sk (11)
∂t σk
interaction between an onshore OWC geometry with regular and
irregular waves [55]. The study revealed that random waves provided ( → ) [( ) ]
∂(ρε) μt ε2 ε
better device performance assessment than regular waves. + ∇. ρ U ε = ∇. μ+ ∇ε + ρC1 Sε − ρC2 √̅̅̅̅̅ − C1ε C3ε Gb
∂t σε k + νε k
Like ANSYS Fluent, ANSYS CFX is commonly used to determine the
effect of chamber size, front wall orientation, and immersion depth on + Sε
the efficiency of an OWC [9]. The orientation of the shoreside wall at (12)
180 was found to be the best configuration for unit efficiency, sup­

porting the results of Ref. [7]. The latter study used ICEM-CFD and CFX where the model constants are C1 , C2 , C1ε , C3ε and Cμ , while σ ε = 1.2
2
with a regular Stokes second-order wave generated in an NWT to study and σk = 1.0. The TEV is modeled as μt = ρCμ kε but Cμ is not constant in
PTO sizes, wave conditions, air column size, chamber upper, and front the realizable k-ε model. In modeling Cμ in the realizable k-ε model, it
wall shape. The results showed that the best orientation of the front wall can reclaim the conventional value of 0.09 when dealing with an inertial
configuration is counter to the flow direction. Also, a conical-shaped subrange in a boundary layer that is in equilibrium [46]. Ref. [61]
upper chamber of the OWC leads to better hydrodynamic efficiency, applied the realizable k-ε model to the OWC. In this simulation, ANSYS
and the number of PTOs does not affect device performance. In another Fluent was used to evaluate the behavior of a spar buoy OWC; the
study, ANSYS CFX was used to determine the optimal operating point of outcome showed good agreement with the experiment for chamber
the OWC by the sequential optimization procedure [56]. The study pressure drop.
found that the OWC configuration, PTO damping, and wave conditions A blend of realizable k-ε and laminar CFD method with VOF method
affect device performance. The width of the OWC chamber and front to capture the interface between water and air, and the dynamic mesh
wall immersion height are also important parameters in OWC design method using layering was used by Ref. [62] to study a heaving OWC.
[9]. The study applied the laminar and realizable k-ε model to separate zones
Ref. [57] used a temporal series to represent the piston-like motion of in the computational domain. Specifying the initial part of the flow
the water in the OWC. It was noted that a new method that defines the domain as laminar was to avoid wave attenuation before entry into the
state-space system solving the Cummins equation provided good OWC. The other part of the flow domain was modeled as a realizable k-ε
agreement with the IH2VOF simulation for different regular and irreg­ turbulent model to sufficiently capture the flow physics while ensuring
ular wave trains and required less computational resources. Also, a that the wave decays since the right pressure-outlet boundary was
time-domain model was developed, and the results compared to the modeled as an open channel. It was noticed that the pure k-ε model
IH2VOF RANS model in Ref. [58], which concluded that the simplified underestimates the wave height compared to the hybrid method.
time-domain model was accurate enough and can be used as a first step However, the water height, oscillatory motion of the device, and the
in modeling a multi-chamber OWC. pressure difference of the offshore OWC agree very well with the
The FLUINCO code is based on the RANS equations discretized by the experiment. This methodology ensures that sufficient grid refinement is
Taylor-Galerkin method, while the ANSYS Fluent code is based on the placed in zones where the physics of the flow is essential to under­
FVM discretization formulation to solve the fundamental equations of standing the flow dynamics while making optimal use of the computa­
fluid flow. The FLUINCO and ANSYS Fluent codes were used to analyze tional resources.
the regular waves for an onshore OWC in Ref. [18]. The paper aimed to
perform a numerical analysis of an onshore OWC subjected to regular 3.2.2.3. The k-ω turbulence model. Different versions of the k-ω model
waves. The comparison of both codes showed similar results in flow have been developed and proposed to address the limiting features of the
behavior, surface elevation, sloshing, and amplification factor, H/ Ho , k-ε model; however, the model proposed by Wilcox is noted as the
where H is wave height in the chamber and Ho is the incident wave modern version [63]. The transport equation in vector notation for k and
height. Also, a comparison of ANSYS Fluent and FLUINCO was carried ω is noted as follows:
out in Ref. [59]. The FLUINCO model was used to investigate the turbine ( → ) [( ) ]
characteristics and an OWC geometry. The results obtained by both ∂(ρk) μ
+ ∇. ρ U k = ∇. μ + t ∇k + Gk − Yk + Sk (13)
codes were in remarkable agreement for the free surface elevation at ∂t σk
periods T = 8 and 10s inside the OWC chamber. It was concluded that ( → ) [( ) ]
the Taylor-Galerkin method of FLUINCO is promising as a prediction
∂(ρω) μt
+ ∇. ρ U ω = ∇. μ+ ∇ω + Gω − Yω + Sω (14)
tool to determine the aerodynamic and hydrodynamic performance of ∂t σω
an OWC.
where Gk and Gω represent the production of k and ω. Sk and Sω are user-
OpenFOAM code and the toolbox waves2Foam were used to validate
defined source terms for k and ω, respectively. Yk and Yω represent the
the RANS k-ε model for the OWC interaction problem [6]. The paper
dissipation of k and ω. The various terms are given by Gk = − ρui uj ∂xji ;
′ ′ ∂u
concluded that the computed OpenFOAM RANS results accurately pre­
dicted the experimental results when the OWC was exposed to regular Gω = α ωk Gk ; Yk = ρβ∗ f β∗ kω; Yω = ρβf β ω2 ; where α, β, β∗ , f β , f β∗ are pa­
and irregular waves. In a similar study, STAR-CCM+ was used to rameters and coefficients that are specified [46].
generate an NWT to test varying OWC damping conditions under regular REEF3D, an open-source CFD code, was used to investigate the hy­
and irregular waves [60]. It was determined that turbine damping af­ drodynamics of an OWC using a two-dimensional NWT by solving the
fects the capture factor. The wave conditions affect the damping coef­ incompressible k-ω RANS equations [64,65]. The parameters of the free
ficient, which at the optimum value maximizes the capture factor. Also, surface, that is, relative amplitude, pressure, and velocity inside the
it was noted that the wave period affects the damping coefficient OWC chamber, were studied for various wavelengths. The numerical
substantially. results of these parameters correlate well with the experiment [64].
The dependence of the device efficiency on incident wavelength was
3.2.2.2. The realizable k-ε turbulence model. This model adopts a new investigated in Ref. [65]. It was shown that the PTO damping influences
TEV expression that involves a variable Cμ and a new equation for ε; the hydrodynamics and the chamber of the OWC at varying incident
however, the transport equation for the TKE takes the same form as the wave conditions. The key finding is that the OWC’s hydrodynamic ef­
standard k-ε model. The transport equation in vector notation for k and ε ficiency depends on the wave height, incident wavelength, and PTO
damping. In Ref. [66], a study of a two-dimensional single- and

6
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

dual-chamber OWC device was carried out using the standard k-ω Table 1
transport model. The results showed that the dual-chamber device ex­ Advantages and disadvantages of the different turbulent models.
hibits better hydrodynamic performance than the single-chamber OWC, Turbulent Models Advantages Disadvantages
especially at small wavelengths. In another study, an OWC-type device
1 k-ε This model has been The model is formulated
termed REWEC1, a submerged caisson, was studied to compare exper­ widely applied to assuming the flow has a
imental and numerical results for various values of air-pocket heights simulate the fluid flow of high Reynolds number.
formed in its U-tube [67]. The investigation found that the k-ω model OWC devices due to its This assumption means
compares well to experiments. accuracy for flows at the the model cannot
far wall regions [80]. accurately predict flow
Since most investigations conditions at the near-
3.2.2.4. The k-ω SST model. The transport equation for k and ω is in a of the OWC do not wall and the boundary
similar form to the standard k-ω model: consider the near-wall layer phenomenon in the
[( ) ] region to influence OWC OWC.
∂(ρk) ( → ) μ performance strongly, it
+ ∇. ρ U k = ∇. μ + t ∇k + Gk − Yk + Sk (15) explains the broad
∂t σk
application of this model
( → ) [( ) ] to the study of OWCs.
∂(ρω) μt
+ ∇. ρ U ω = ∇. μ+ ∇ω + Gω − Yω + Dω + Sω (16) 2 Realizable k-ε Realizable k-ε model was In application to the OWC
∂t σω proposed to address some system, this model will
deficiencies for the TDR result in non-physical and
where Dω represents the cross-diffusion parameter. The terms are given calculation, which tends inaccurate turbulent
to overpredict the viscosities around the
by Gk = − ρui uj ∂xji ; Gω = ναt Gk ; Yk = ρβ∗ kω; Yω = ρβω2 ; where α, β, β∗ are
′ ′ ∂u

turbulent viscosity and rotating air turbine and


parameters and coefficients that are determined in Ref. [46]. provides inaccurate the stationary fluid areas,
STAR-CCM+ k-ω SST RANS modeling was used to study an OWC length scales [44,81]. such as the boundary
The realizable model has layers of the rotating
[68–74]. In one study, 130 CFD simulations and 220 experiments
shown promising results devices.
explored the effect of mooring line pre-tension, regular and irregular for flows with complex
waves, and PTO damping on device performance for both secondary features and
floating-moored and fixed offshore OWCs [68]. The paper found no separated flows [46].
significant effect on the device efficiency due to mooring line 3 k-ω k-ω model was proposed The main weakness of
to model the near-wall this model is its inability
pre-tension for both devices. The advantages of floating versus fixed
phenomenon for wall- to accurately model the
devices under regular wave conditions were also observed for irregular bounded flows freestream region due to
waves. It was also noted that the PTO damping did not affect the effi­ effectively. It excels at solution sensitivity to
ciencies of either device. A similar study compared the survivability of predicting flow both the initial TKE and
separation due to an the specific TDR (ω) at
damaged and undamaged mooring systems for floating-moored OWCs
adverse pressure the far wall [80].
[69]. It was observed that the largest wave height does not cause the gradient.
greatest surge and mooring line tension for an irregular wave at a given 4 k-ω SST k-ω SST model combines This model usually
sea state. An experimental and numerical study of the wave load was the accuracy of the k-ω produces considerable
carried out on a model of an offshore stationary OWC [70]. The vertical model at the near wall for turbulence in flow
flows bounded by a wall domains with high
wave load acting on the device was consistently smaller than the hori­
and that of the k-ε models normal strain, e.g., areas
zontal wave load. Ref. [71] simulated a three-dimensional fixed offshore for flows at the far wall with high flow
OWC subjected to regular waves under varying periods and wave regions to model the acceleration and
heights but with constant PTO damping. The conclusion was that entire flow domain [82]. stagnation regions.
5 Reynold Stress Eddy-viscosity-based Since this model directly
modeling a floating OWC in a two-dimensional domain overestimates
Model models (e.g., k-ω and k-ε) resolves the individual
the hydrodynamic efficiency of the device when the wave frequency is cannot accurately model Reynold stresses, which
greater than the resonant frequency of the chamber; however, modeling complex turbulent flows are then used to
a three-dimensional domain was in good agreement with experimental like rotating flow formulate closure for the
results. In another numerical study, Ref. [72] explored the effect of passages, swirling flows RANS equation, the
in engine combustors, model expends more
scaling and compressibility of air on device performance. The paper
cyclone flows, and stress- computational resources
observed that air compressibility is essential when the OWC device induced flows in pipes. than the lower-order
operates at full-scale. When simulating a scale-model device, neglecting The RSM was proposed to models like the eddy-
compressibility can result in a 12% reduction in optimum efficiency address these viscosity ones.
shortcomings. It is
under regular waves, optimum PTO damping, and resonant frequency
required when the fluid
conditions. Similarly, a two-dimensional RANS computation was used to flow features result from
conduct an energy balance analysis of a floating offshore OWC [73]. anisotropy in the Reynold
Different parameters, such as PTO damping, incident wave frequency, Stresses. It is an elaborate
and wave height were tested. The results indicated a good agreement model that solves
Reynolds stresses as a
with experimental data. Also, a two-dimensional RANS-VOF technique
transport equation and
was used to investigate the effect of underwater rear and front lips on the calculates the dissipation
performance of fixed offshore OWC [74]. The overall hydrodynamic rate by avoiding the
efficiency was improved over the given frequency range by having eddy-viscosity
suitable values for the lip thickness and the submergence ratio of hypothesis. It accounts
for the outcome of
non-symmetric lips. complex fluid
Laboratory experiments were carried out to validate the results of the interactions, rotation,
IHFOAM solver [75]. The numerical model agreed with the experi­ streamlines curvature,
mental results for the air velocity and pressure, free surface elevation at whirl, and fast changes in
the rate of strain of the
the front, behind, and inside the chamber. Therefore, IHFOAM was
flow [46].
proven to be a suitable CFD tool for OWC analysis. Similarly, Open­
(continued on next page)
FOAM was used to assess the hydrodynamic behavior of an OWC device

7
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

Table 1 (continued )
Turbulent Models Advantages Disadvantages

6 Large Eddy LES resolves the flow However, finer mesh


Simulation field by modeling small sizes are necessary for
eddies at the sub-grid LES computations than
scale while large eddy’s RANS simulations [46],
turbulent structure is especially for flows
directly resolved [27], bounded by a wall in the
allowing for larger mesh moderate to high
spacing and time steps Reynolds number.
compared to DNS.
7 Scale-Adaptive SAS model was proposed When the SAS model
Simulation by Ref. [83] as a method results in a steady
used to simulate flows solution, it has a
with strong unsteadiness limitation where
and instabilities. It unsteadiness cannot be
displays LES imposed on the flow.
characteristics in Also, it does not ensure
unsteady regions and enough damping for the
RANS behavior in steady tiniest scales at the
regions. It is also discretization limit.
considered an alternative
method for hybrid
RANS-LES models. Fig. 4. Percentage distribution of RANS model applied to the study of OWC.
8 Detached Eddy These hybrid models can The main weakness of the
Simulation couple Spalart-Allmaras, DES is its inability to
Realizable k-ε, and k-ω respond well to ( ) ( )
SST, among others. The ambiguous discretization ∂ ρu′i u′j ∂ ρuk u′i u′j ∂ [ ′ ′ ′ ( )]
DES model gives better where the parallel grid + =− ρui uj uk + p′ δkj u′i + δik u′j
∂t ∂xk ∂xk
accuracy than the spacing of the wall is [ ] ( )
standard RANS and proportional to the order ∂ ∂ ( ′ ′) ′ ′ ∂uj ′ ′ ∂ui
+ μ ui uj − ρ ui uk + uj uk
URANS models. of the boundary layer ∂xk ∂xk ∂xk ∂xk
thickness. ( ) ( ′ ′)

9 Direct Numerical Ref. [84] stated that the The computational ′ ′ ∂ui ∂uj
− ρβ gi uj θ + gj ui θ + p′ +
Simulations DNS is the simplest expense needed for a DNS ∂xj ∂xi
method but unrivaled in simulation increases ′ ′
( )
terms of its temporal and rapidly with the Reynolds ∂ui ∂uj ′ ′
− 2μ − 2ρΩk uj u′m εikm + ui u′m εjkm + Suser
spatial resolution. number, so the method is ∂xk ∂xk
limited to fluid flow of (17)
low to moderate
Reynolds numbers [84].
where ui , uj , uk , ui , uj , and uk represent the fluctuating and mean fluc­
′ ′ ′ ′ ′ ′
Since most practical
applications involve tuating velocities in their respective spatial coordinates, and Suser is the
turbulent flows, the
user-defined source term. Furthermore, θ, β, p , and Ωk are the temper­

computational resources
needed are beyond the ature gradient, coefficient of thermal expansion, fluctuating pressure,
most powerful computer and rotation vector, respectively. Moreover, εikm = 1, 0, or − 1 if the
currently in existence. indices are different and in cyclic order; any two of the indices are the
Also, the minuscule time same; or the indices are not in cyclic order, respectively.
steps used in the
simulation of WECs, like
Ref. [78] used STAR-CCM+ for a two-dimensional RANS CFD model
OWCs, prohibit the DNS to investigate the energy balance analysis of a fixed onshore OWC. The
approach from being k-ω SST model agreed reasonably well with the RSM, while the k-ε
applied [27]. model could not capture the flow physics even with a highly refined
10 Smooth Particle SPH is ideal for modeling There is difficulty in
mesh. That shows that the k-ω SST can better predict the flowfield of
Hydrodynamics problems with large selecting the appropriate
deformations and Kernel function due to OWC systems than other traditional RANS models while being compu­
complex boundary the complex nature of the tationally cheaper. The hydrodynamic wave forces exerting on a
conditions, such as in the boundary conditions in fixed-floating offshore OWC were investigated in Ref. [79] by a two and
OWC. OWC problems. three-dimensional CFD model based on the RANS-VOF technique. The
vertical wave force was found to be smaller than the horizontal force.
with a stepped sea bottom [76]. The wave period and height influenced Choosing which turbulent model to apply in a particular OWC
the device’s performance in regular waves. Also, to determine the TKE, investigation is a complex decision that will depend on a thorough
k, and the SDR, ω, of a dual-chamber OWC, Ref. [77] used the k-ω SST literature search and preliminary test to demonstrate how numerical
model. The interaction between regular waves and the OWC device was results compare with the experiment. Table 1 reviews the advantages
studied. The impact of the CWR of the front chamber was observed to and disadvantages of using the different turbulent models in investi­
have a more considerable influence on device efficiency than that of the gating the OWC.
rear chamber. The distribution of the RANS model applied to the study of the OWC
system reviewed in this paper presented in Fig. 4 shows that the k-ε
model is the widely used model.
3.2.2.5. The Reynold stress model. In the RSM, the individual Reynolds
stresses, ρui uj , must be computed by differential transport equations.
′ ′

3.2.3. Large eddy simulation


These Reynold stresses are applied to get closure of the RANS equation. The early studies using LES were carried out mostly in the atmo­
The exact transport equation is as follows: spheric disciplines [85–87]. This model is now being applied to flows
with complex geometries in engineering applications [84].

8
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

Fig. 5. Wave flume at the NCAT fluid laboratory.

Decomposing the velocity and pressure filters into their resolvable scale the experiment.
and subgrid-scale part ui = ui + ui and p = p + p , respectively. Where ui ,
′ ′

p represents the resolvable scale parts and ui , p represents the sub-grid 3.3. Other high-resolution models
′ ′

part. The filtered equations developed from the incompressible


Navier-Stokes equation is as follows: 3.3.1. Scale-adaptive simulation
( ) The transport equations for k and ω for the SST-SAS model is
∂ui ∂ui uj 1 ∂p ∂ ∂ui ∂ui established on Rotta’s method to the k-ω SST model, which is as follows:
+ =− +ν + (18)
∂t ∂xj ρ ∂xI ∂xj ∂xj ∂xj [( ) ]
∂(ρk) ( → ) μ
Substituting the velocity and pressure filters decomposition into + ∇. ρ U k = ∇. μ + t ∇k + Gk − ρcμ kω (20)
∂t σk
equation (18) and filtering the resulting equations leads to equation (19)
as the resolved field: ∂(ρω) ( → ) [( )
μt
]
ω
( ) + ∇. ρ U ω = ∇. μ+ ∇ω + α Gk − ρβω2 + QSAS + (1
∂ui ∂ui uj 1 ∂p ∂ ∂ui ∂uj ∂τij ∂t σω k
+ =− +ν + − (19) 2ρ 1
∂t ∂xj ρ ∂xi ∂xj ∂xj ∂xi ∂xj − F1 ) ∇k∇ω
σω,2 ω
where the filtered advection term τij = ui uj − ui uj . The equation (19) can (21)
be derived for the residual field, i.e., the subgrid-scale field.
In Ref. [88], the sea wall type OWC device of low reflection was where QSAS and F1 represents an additional source term and blending
investigated for optimal shape using a new CFD solver known as 2PM3D function, respectively. The model constants are α, β, σω,2 . Applying this
using LES. The reflection coefficients result from this new solver were model to the study of the OWC system will benefit from a lower
validated with the experimental result and were quite promising. It was computational cost than the LES.
further found that three-dimensional modeling is the most appropriate
model to study the optimal shape of the air duct. Ref. [89] used an LBM 3.3.2. Detached eddy simulation
solver to compare the results of a two-dimensional wave flume and a DES was initially developed by Ref. [95] as a hybrid approach
three-dimensional LBM-LES numerical wave flume. The paper combining the capabilities of the LES and the Spalart-Allmaras model. In
concluded that the LBM-LES model could not accurately model the this model, the unsteady RANS methods are applied to the near-wall
dissipation of waves. region, and on the other hand, the LES is applied to a flow field’s
Several studies using OpenFOAM with the LES turbulence model far-wall (freestream) region where large turbulence scales dominate.
were carried out [90–94]: Ref. [90] studied the error induced due to The DES model was designed to address wall-bounded flows with a high
neglecting the air compressibility effect in modeling the OWC device. It Re since the LES model’s computational expense to resolve the near-wall
was found that the consequence of air compressibility was reduced for a problems can be very high. The DES method has been expanded where it
1:5 scale model or larger when modeling an OWC. When at a scale of can be applied to solve a problem with different RANS models
1:50, compressibility affects the results. A CFD investigation of the depending on the flow field [63].
combined effect of turbine damping, fixed OWC geometry, and wave
conditions was carried out by Ref. [91]. The observation was that there 3.3.3. Direct numerical simulations
was a coupled interaction between all the parameters considered in the The DNS approach is a CFD method that solves the NSE numerically
study. Also, the study noted that the optimization of the device could by resolving all scales of motion and applying the initial and boundary
only be achieved by considering all the design parameters involved with conditions to the flow under consideration. The number of operations
the OWC. Optimization of an OWC device was carried out by Ref. [92] needed to complete a DNS simulation is directly proportional to the
for a site-specific area in the Mediterranean by device testing and CFD number of time-steps and mesh points and grows as Re3 .
simulations. The paper indicated that the length of the back wall of the
OWC device equal to approximately one-third of the wavelength is 3.4. Smooth particle hydrodynamics
needed to give a maximum wave reflection at the front of the OWC,
resulting in more significant wave oscillation thus higher performance. SPH is a Lagrangian method initially developed as a probabilistic
In Ref. [93], the results for the LES and k-ω SST models were compared meshless particle-tracking approach to solving astrophysical problems
for the water surface oscillations, air chamber pressure, and vertical air [96,97]. However, it has been modified as a deterministic mesh-less
velocity. The LES model was in very good agreement with experimental particle-tracking method [98,99] with increasing application to fluid
results but at a higher computational expense. Also, a multi-phase LES dynamics problems [100–104] that have complex boundary conditions
modeling of the OWC was carried out in Ref. [94]. It was determined and free surface flows [105], like the flow topology of the OWC. When
that the numerical results for surface elevation and chamber pressure applied to the OWC, the individual fluid particles are tracked, and the
using a monochromatic wave were in relatively good agreement with flowfield is formulated using the SPH methodology. This method is

9
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

Fig. 6. Wave generation in the wave flume.

commonly implemented in the Navier-stokes-based equations using the


Lagrange framework, and the boundary condition is specified to solve
OWC problems. The SPH depends on the Kernel function as they are
closely related to its computational accuracy and stability. Therefore,
researchers have consistently been trying to improve the accuracy of the
Kernel [105]. For example, in Ref. [106] the hydrodynamics of an OWC
device without a PTO, where the effect of air motion inside the OWC,
when ignored, was studied using an improved kernel in an SPH turbu­
lent model (SPHysics). The results of the free surface elevation, wave
force, and torque of the improved SPH model were validated with Fig. 7. The volume of fluid method.
analytical solutions of the second-order stokes wave equation, and it was
found that the improved SPH agrees very well with the analytical results 4.1. Wave generation and absorption
compared to the original SPH model. The complex nature of the solid
boundary conditions when modeling the OWC using SPH is due to the There are several wavemakers, but the commonly used ones are
difficulty in selecting the appropriate Kernel truncation, Ref. [107] used classified into (i) paddles [110] and (ii) plungers [111]. Examples of the
the regional ghost particle method to model the thin front wall using a paddle-type are the piston- and flap-type wavemakers. Wave absorption
two-dimensional NWT instead of the multi-node fixed ghost method, is necessary to prevent the disturbance of incoming waves, which in­
thereby providing a more efficient option and saving computational terferes with the flowfield. Sloped bottom and materials that absorb
resources. In other studies, the SPH method was applied in the study of wave energy, like wire mesh or gravel, are often positioned at the end of
the OWC, Ref. [108] investigated the onshore OWC in terms of its hy­ the wave tank [112]. Various numerical methods are used to generate
drodynamics conditions, the amplification factor, and the forces exerted waves in an NWT investigating WEC devices. These methods are cate­
on the vertical wall of the OWC in a rough sea state under sophisticated gorized as the relaxation method, mass source function, impulse source,
wave breaking, overtopping wave conditions, and impact load using the static, and dynamic boundary method. On the other hand, numerical
SPHyCE, which is a code based on SPH. The phase lag and amplification wave absorption methods are categorized as relaxation, numerical
factor results compared very well with the standard k-ε turbulence beach, sloped beach, passive absorption, sloped bathymetry, mesh
model. Also, the SPHyCE model that was applied accurately predicted stretching, static and dynamic boundary methods. These methods are
the resonance condition in the OWC device. Ref [109] applied another described in detail in Refs. [44,80,112].
SPH-based code (DualSPHysics) to investigate sea wave interaction and
the effect of mooring devices on an offshore OWC. The model replicated
4.2. Numerical wave tank
the water surface accurately in the OWC when the air pressure was
ignored. The authors project the SPH approach to increase in popularity
In the design of marine structures, wave loads are important forces
in modeling the OWC mainly due to its suitability in simulating complex
that must be considered; however, carrying out site tests at sea can be
boundary dynamics as opposed to the Eulerian approach, which uses a
risky, expensive, and sometimes impossible to perform. That led to the
grid-based methodology with difficulties in solving problems with large
development of experimental wave tanks and flumes to generate waves;
deformations. That makes the meshless SPH approach accurate for
however, wave generation in the laboratories is still relatively expen­
highly nonlinear flows, such as flows in the OWC with complex
sive, leading to the development of NWTs in recent years [113]. An NWT
boundary conditions [106].
is a computational tool that simulates the conditions of a wave tank
using CFD codes. They can model motions of floating bodies, calculate
4. Numerical wave generation
hydrodynamic forces, and account for non-linear free surface conditions
[44].
In the laboratory, a wave tank or flume shown in Fig. 5 is used for
Ref. [114] used two numerical codes, Flow-3D and Fluent, to create
wave generation (shown in Fig. 6) to test and investigate wave energy
regular waves in an NWT. The results of the two numerical codes were
devices and other marine structures such as the OWC. Numerical codes
similar, but there was a slight phase difference when comparing the
to generate a numerical wave tank can be formulated to describe this
Fluent results with the theoretical results. Moreover, Ref. [113] used a
laboratory testing.
two-dimensional numerical wave flume to investigate the damping and
generation of waves by the BEM proposed by Ref. [115]. The boundary

10
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

condition of the NWT was fully non-linear at the free surface, with the
other appropriate boundary conditions specified to describe the inlet,
wall, and artificial damping zone. The MEL proposed by Ref. [116],
based on the potential flow theory, was used to develop the NWT. It was
found that the NWT, a BEM solver, was in good agreement with theo­
retical results. Ref. [117] developed a three-dimensional multidirec­
tional non-linear NWT based on the NSE using the FVM. That was
achieved using the flow solver naoe-FOAM-SJTU, based on OpenFOAM.
The solver was shown to be capable of generating different unidirec­
tional and multidirectional non-linear waves accurately, can solve
wave-structure interaction problems, and is used to design offshore
structures. In Ref. [118], a three-dimensional NWT REEF3D was
developed to estimate wave propagation and hydrodynamics by
numerically solving the NSE. The free surface was captured using the
level-set method with high-order discretization in time and space to
prevent the unphysical damping of the water waves generated. The
fifth-order WENO scheme was applied for convection discretization due
to employing a Cartesian grid which provided stable and accurate re­
sults. There was a good correlation with the experimental wave force
data, velocity, and free surface. It also accurately predicted wave Fig. 8. Percentage distribution of CFD codes applied to study the OWC.
shoaling and the accompanying wave transformation, such as wave
breaking. In Ref. [119], the CFD code – Fluent was used to model a
The volume fraction, q, is specified in equation (23). V1 is the volume
three-dimensional NWT with a piston-type wavemaker. The RANS-VOF
of liquid in a cell, and Vcell is the volume of the cell.
method was applied, and the model was validated against theoretical
results. Also, a numerical wave-structure interaction with a vertical V1
q= (23)
cylinder was studied for different wave periods and heights and then Vcell
verified with experimental results. The NWT was capable of modeling
The advection of volume fraction, q for the b-th fluid in a system of, n
actual ocean wave conditions and accurately predicting wave trans­
number of fluids is shown in equation (24):
formation and load.
∂qb →
+ v .∇qb = 0 (24)
5. Volume of fluid model ∂t
In each numerical cell of the interface of the flow domain, the vol­
The VOF method shown in Fig. 7 and proposed by Ref. [120], models ume fraction must sum up to 1 [122].
the interface between immiscible fluids. The interface is treated as
discontinuous in the flow field [121]. The techniques used to track, ∑
n
qs = 1 (25)
capture, and locate fluid interfaces are classified into interface tracking s=1
and interface capturing methods [122]. The interface tracking methods
include Height Functions, Line Segments, and Marker Particles [120]. An artificial compression term is included in equation (24) to reduce
While, the interface capturing methods are: the CICSAM Scheme the smearing of the air-water interface due to numerical diffusion. It also
[123–125], the HRIC Scheme [123,125], the Level-set Method [122], ensures the conservation of q and the boundedness of equation (24).
the MAC scheme, the STACS scheme [123], the SMMC method [126], Ref. [120] noted that the amount of q moving from a donor to an
and the VOF [120,122,127,128]. acceptor cell in a time step, δt is δq × r, and r = u δt, where r is the total
The VOF method is not considered to be computationally flux of fluid volume moving through the cells and u is the cell-face
demanding. That is one reason for its wide implementation [129,130]. normal velocity. The δq is given by Ref. [120]:
The VOF is a simplified Eulerian multi-phase modeling approach used to δq = min[qAD |r| + CF, qD δxD ] (26)
model continuous-continuous phase interactions of distinct interface
and non-miscible fluids. The mesh consists of moving and fixed mesh where,
parts; the moving mesh traces the curvature of the interface, then it is
CF = max [(1.0 − qAD )|r| − (1.0 − qD )δxD , 0.0 ] (27)
applied to the VOF computation, where the equation of continuity,
momentum, and energy are solved as in a single-phase flow. However, The subscripts A and D refer to the acceptor, and donor, respectively.
the properties of the various phases are different. These properties are Moreover, AD represent either A or D. Also, CF and δx represent the
calculated using a volume fraction equation. The VOF model provides an additional fluid flux and width of cells, respectively. Equations (26) and
economical and straightforward way to track free boundaries in two- or (27) refer to the artificial compression terms needed to contain nu­
three-dimensional meshes [120]. merical diffusion. Ref. [120] provides detailed information on the vol­
ume being advected due to the flux of the donor and acceptor cell.
5.1. Volume fraction equation

A single set of turbulent transport equations are used to compute the 5.2. Material properties
entire flowfield [122] for both phases [131], and the volume fraction is
traced for the individual fluid throughout the flowfield [121]. A volume The viscosity and density stated as a function of q are averaged as
fraction, q, is instituted to differentiate the various fluids, which is follows [133–135]:
defined as [26,132]:
μ(q) = qμliq + (1 − q)μair (28)

⎨ 0 air
q = 0 < q < 1 liquid − air interface (22) ρ(q) = qρliq + (1 − q)ρair (29)

1 liquid
The viscosity and density in the fundamental equations are average

11
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

Table 2 Table 2 (continued )


Main advantages and disadvantages of the different CFD Solvers. CFD Solvers Advantages Disadvantages
CFD Solvers Advantages Disadvantages
The LBM solver enables The solver is not popular
1 ANSYS Fluent This solver uses the cell- It has a considerable complex fluid physics by in marine
centered FVM to number of DOF for every indirectly solving the hydrodynamic studies,
calculate fluxes in non- flowfield parameter NSE from discrete LBE and there is a lack of a
conforming cell resulting in high to resolve complicated consistent scheme that
interfaces, making it computational demand OWC geometrical addresses thermo-
more flexible in grid and memory usage for problems with the air hydrodynamics
motion, adaption, and application in three- turbine modeled as a problems in OWC.
generation. dimensional OWC porous media.
simulation. 9 In-house In-house codes allow the In-house codes typically
2. ANSYS CFX This solver uses the For the FVM used for flexibility of designing take much time to go
vertex-centered FVM, discretization, the dual- one to solve a particular through the verification
allowing it to combine meridian mesh used in problem specific to and validation process
smaller sub-CV around this solver can produce a OWC. to ensure that the
the vertex. Also, it CV of bad quality. And, research communities
requires less number of it takes a longer time for widely accept the results
iterations to attain each iteration to be of the code.
solution convergence. completed when using 10 SPH Based Codes Introducing solid The accuracy of the SPH
this solver. (In-house, geometries inside code depends highly on
3 IH2VOF Specifically, it is It is limited to solving SPHysics, moving fluid is easy due the Kernel function.
designed to solve wave two-dimensional DualSPHysics and to induced forces
problems for hybrid problems and the RANS SPHyCE) because of waves [109].
domains in a coupled k-ε turbulence model.
system of NSE, making it
one of the most quantities. Therefore, equations (28) and (29) are solved for their
advanced solvers for respective average values input into the fundamental equations. When
hydrodynamics and
the bulk of the fluid is water, q = 1, equations (28) and (29) will yield μliq
coastal structures
problems. It provides and ρliq respectively, and when a significant portion of the fluid is air, q
realistic and active wave = 0; equations (28) and (29) produce μair and ρair . At the interface, when
generation and the q value is intermediate, the density and viscosity will be a function of
absorption.
q.
4 OpenFOAM It has a toolbox named OpenFOAM has a steep
IHFOAM, specially learning curve and is not
designed to solve three- user-friendly. 5.3. Interface reconstruction
dimensional bi-phasic
flow problems in
Advection of material distribution in a multi-phase flow leads to non-
offshore, coastal, and
hydraulic engineering.
sharp interfaces because of the repeated interpolation over time from
IHFOAM has an the numerical diffusion of the material properties [136]. In modeling the
extensive collection of OWC using an NWT, the location of the interface between the air and
boundary conditions for water is of interest. Several techniques have been proposed for interface
three-dimensional wave
reconstruction, including PLRC, PCR, SOR, and SSR [137]. However, the
generation and active
absorption in NWT for widely applied techniques are PLRC and PCR. Refs. [130,137] devel­
OWC problems. oped the PLRC technique, an approach commonly used to keep the
5 Star-CCMþ Star-CCM+ has a It has not been widely interface sharp in a multi-phase flow. In that approach, the averaged
varying range of models used as ANSYS Fluent in values of conserved quantities are interpolated to the faces of the
to simulate the OWC. investigating the OWC.
Also, it produces far Fig. 8 shows that 19% of
computational cell from the cell center. Also, the PCR method has been
higher meshing quality the studies considered in widely referenced in the work of Refs. [120,138]. Ref. [139] noted that
compared to other CFD this review article [140]’s piecewise linear technique is the most accurate choice [141],
codes. applied Star-CCM+ but it is perceived to be difficult and computationally expensive, leading
compared to 37% for
many researchers and authors to prefer the less accurate PCR
ANSYS Fluent.
6 REEF3D REEF3D is an open- The turbulent models approximation.
source code with four are limited to a few
models designed RANS models (i.e., k-ε 6. CFD solvers
explicitly for wave and and k-ω), URANS
hydrodynamic (coupled with k-ε and k-
problems. That makes ω), and LES.
Commercial software that computes the flowfield of WECs includes
its application to but is not restricted to ANSYS™ Fluent, CFX, Star-CD/CCM+, and
studying OWC problems FLOW-3D [19]. In-house and open-source software such as OpenFoam,
for different wave Code-Saturne, AMAZON, and ComFLOW have also been used [19].
theories easier
Other codes like IH2VOF, REEF3D, 2PM3D have been used to investi­
compared to other CFD
models. gate OWCs. In this paper, the percentage distribution of the reviewed
7 2PM3D It can enable wave- 2PM3D uses the FDM for papers that have applied the various CFD codes to study OWC systems
breaking studies in OWC its spatial discretization are indicated in Fig. 8. ANSYS Fluent, Star-CCM+, and OpenFOAM are
and accounts for the leading to structured the most used codes that have been applied to the study of the OWC.
nonlinearities due to the meshing making its
wave-breaking application to
Table 2 list the important advantages and disadvantages of the various
phenomenon. complicated OWC solvers.
geometry challenging.
8 LBM Solvers

12
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

7. Conclusions [5] Sheng W, Alcorn R, Lewis A. On thermodynamics in the primary power


conversion of oscillating water column wave energy converters. J Renew Sustain
Energy 2013;5.
This review investigates the numerical modeling of OWCs. The aim is [6] Vyzikas T, Deshoulières S, Giroux O, Barton M, Greaves D. Numerical study of
to provide insight into the accuracy of CFD using NWTs for different flow fixed Oscillating Water Column with RANS-type two-phase CFD model. Renew
regimes. Based on the discussions in this review, we draw the following Energy 2017;102:294–305.
[7] Zaoui L, Bouali B, Larbi S, Benchattp A. Performance analysis of a 3D
conclusions: axisymmetric oscillating water column. Energy Proc 2014;50:246–54.
[8] Opoku F. Numerical investigation of a novel oscillating water column. Doctoral
• Most investigations of the OWC in the inviscid flow regime have been Dissertation, Dept. of Mech. Eng., North Carolina Agricultural and Technical
State University. ProQuest Diss. Publ; 2021.
conducted using the potential flow model, which tends to over­ [9] Bouali B, Larbi S. Contribution to the geometry optimization of an oscillating
estimate the efficiency of these devices. water column wave energy converter. Energy Proc 2013;36:565–73.
• The RANS transport models that have been used to model the OWC [10] Gomes MN, Nascimento CD, Bonafini BL, Santos ED, Isoldi LA, Rocha LAO. Two-
dimensional geometric optimization of an oscillating water column converter in
system include the k-ε, realizable k-ε, k-ω, k-ω SST, and the Reynolds laboratory scale. Rev Eng Térmica 2012;11:30–6.
Stress Model. These models have demonstrated that they can [11] Lopez I, Iglesias G, Lopez M, Castro F, Rodriguez MA. Turbine-chamber coupling
adequately reproduce experimental results with an acceptable level in an OWC wave energy converter. In: Proceedings of the 33rd conference on
coastal engineering; 2012.
of accuracy. Also, the standard k-ε model was established to be the [12] Zhang Y, Zou Q, Greaves D. Air-water two-phase flow modelling of hydrodynamic
most widely used among the research articles that were reviewed. performance of an oscillating water column device. Renew Energy 2012;41:
• No research articles investigating the OWC that were reviewed used 159–70.
[13] Halder P, Rhee SH, Samad A. Numerical optimization of wells turbine for wave
the following turbulence models: Spalart-Allmaras, RNG k-ε, SAS,
energy extraction. Int J Nav Archit Ocean Eng 2017;9:11–24.
DES, and DNS. However, there is no indication in the literature to [14] Halder P, Kumar PM. Coupled CAD-CFD automated optimization for leading and
suggest that these models cannot be applied. trailing edge of an axial impulse turbine blade. Ocean Eng 2020;213:1–15.
• While DNS is too computationally expensive to model a full-scale [15] Morrison IG, Greated CA. Oscillating water column modelling. Proceedings of the
23rd Int. Conf. Coast. Eng. 1992;502–11.
OWC, other high-fidelity models such as LES capture detailed flow [16] Chenari B, Saadatian SS, Ferreira A. Wave energy systems: an overview of
phenomena and show very good agreement with experimental different wave energy converters and recommendation for future improvements.
results. In: Proceedings of the 8th international technology. Education and Development
Conference; 2014.
• The NWT has been demonstrated to be an effective numerical tool to [17] Uddin MN, Atkinson M, Opoku F. Numerical investigation of a modified denniss-
investigate the OWC and other WECs, therefore helping to supple­ auld turbine. In: Proceedings of AIAA aviation forum; 2022.
ment and guide expensive and risky experimental and site tests. [18] Davyt D, Teixeira PRF, Ramalhais R, Didier E. Numerical analysis of regular
waves over an onshore oscillating water column. In: Proceedings of the 13th
• The VOF method is the most popular method for capturing the free Brazilian congress of thermal sciences and engineering; 2010.
surface in the study of the OWC since it is less computationally [19] Mingham C, Qian L, Causon D. Computational fluid dynamics (CFD) models. In:
demanding than other multi-phase simulations. Folley M, editor. Numerical modelling of wave energy converters: state-of-the-art
techniques for single devices arrays. London: Elsevier Inc; 2016. p. 105–22.
• The study of the OWC has used a variety of different CFD codes with [20] Fitzgerald CJ. Nonlinear potential flow models. In: Folley M, editor. Numerical
varying levels of numerical fidelity. This review found that the modelling of wave energy converters: state-of-the-art techniques for single
ANSYS Fluent using RANS modeling is the most used code with devices arrays. London: Elsevier Inc; 2016. p. 83–104.
[21] Folley M. Spectral-domain models. In: Folley M, editor. Numerical modelling of
diverse usage by researchers to study OWC systems.
wave energy converters: state-of-the-art techniques for single devices arrays.
London: Elsevier Inc; 2016. p. 67–80.
[22] Ricci P. Time-domain models. In: Folley M, editor. Numerical modelling of wave
Declaration of competing interest energy converters: state-of-the-art techniques for single devices arrays. London:
Elsevier Inc; 2016. p. 31–66.
[23] Alves M. Frequency-domain models. In: Folley M, editor. Numerical modelling of
The authors declare that they have no known competing financial
wave energy converters: state-of-the-art techniques for single devices arrays.
interests or personal relationships that could have appeared to influence London: Elsevier Inc; 2016. p. 11–30.
the work reported in this paper. [24] Opoku F, Atkinson M, Uddin MN. Numerical investigation of an offshore
oscillating water column. Am J Mech Eng 2020;8:88–105.
[25] Opoku F, Atkinson M, Uddin MN. In: Computational fluid dynamics investigation
Data availability of a novel oscillating water column. Proceedings of AIAA SciTech Forum; 2022.
[26] Uddin MN, Atkinson M, Opoku F. A computational fluid dynamics investigation
We have shared data as a supplementary material. of a numerically simulated wave tank. Am J Mech Eng 2020;8:40–9.
[27] Davidson J, Costello R. Efficient nonlinear hydrodynamic models for wave energy
converter design-A scoping study. J Mar Sci Eng 2020;8:1–65.
Acknowledgments [28] Schlichting H, Gersten K. Boundary-layer theory. ninth ed. Berlin, Germany:
Spinger-Verlag; 2017.
[29] Sheng W, Alcorn R, Lewis T. Physical modelling of wave energy converters.
The authors would like to thank the North Carolina Renewable Ocean Eng 2014;84:29–36.
Ocean Energy Program (NCROEP) and the United States Department of [30] White FM. Viscous fluid flow. third ed. Chennai, India: McGraw Hill Education
Energy (DOE) for the partial support that helped conduct this review (India) Private Limited; 2006. p. 59–73.
[31] Koirala P, Nagata S, Imai Y, Murakami T, Setoguchi T. A numerical study on
exercise. multi-chamber oscillating water columns. J JSCE 2015;3:93–104.
[32] Bull D. An improved understanding of the natural resonances of moonpools
contained within floating rigid-bodies: theory and application to oscillating water
Appendix A. Supplementary data
column devices. Ocean Eng 2015;108:799–812.
[33] Chang C, Chou FN-F, Chen Y, Hsieh Y. Analytical and experimental investigation
Supplementary data to this article can be found online at https://doi. of hydrodynamic performance and chamber optimization of oscillating water
column system. Energy 2016;113:597–614.
org/10.1016/j.rser.2022.113124.
[34] Rezanejad K, Bhattacharjee J, Guedes Soares C. Analytical and numerical study of
dual-chamber oscillating water columns on stepped bottom. Renew Energy 2015;
References 75:272–82.
[35] Josset C, Clément AH. A time-domain numerical simulator for oscillating water
column wave power plants. Renew Energy 2007;32:1379–402.
[1] Opoku F. Feasibility study of solar thermal electric-power generation in northern
[36] Chowdhury SD, Manasseh R. Behaviour of eigenmodes of an array of oscillating
Ghana. Master thesis. Kumasi, Ghana: Dept. of Mech. Eng., K.N.U.S.T.; 2010.
water column devices. Wave Motion 2017;74:56–72.
[2] Aderinto T, Li H. Ocean Wave energy converters: status and challenges. Energies
[37] Folley M, Whittaker T. Validating a spectral-domain model of an OWC using
2018;11:1–26.
physical model data. Int J Mar Energy 2013;2:1–11.
[3] Cruz J. Ocean Wave Energy: Current Status and Future Perspective 2008:93–132.
[38] Clément AH. Dynamic nonlinear response of OWC wave energy devices. Int J
[4] Morris-Thomas MT, Irvin RJ, Thiagarajan KP. An investigation into the
Offshore Polar Eng 1997;7:154–9.
hydrodynamic efficiency of an oscillating water column. J Offshore Mech Arctic
Eng 2007;129:273–8.

13
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

[39] Singh U, Abdussamie N, Hore J. Hydrodynamic performance of a floating offshore [71] Elhanafi A, Macfarlane G, Fleming A, Leong Z. Investigations on 3D effects and
OWC wave energy converter: an experimental study. Renew Sustain Energy Rev correlation between wave height and lip submergence of an offshore stationary
2020:117. OWC wave energy converter. Appl Ocean Res 2017;64:203–16.
[40] Ozdamar G, Pekbey Y, Ozdamar A. A numerical study of the effect of wave [72] Elhanafi A, Macfarlane G, Fleming A, Leong Z. Scaling and air compressibility
amplitude on the efficiency of a wave power system. In: Proceedings of the 7th effects on a three-dimensional offshore stationary OWC wave energy converter.
international advances in applied physics and materials science; 2018. p. 153–5. Appl Energy 2016;189:1–20.
[41] Mohapatra P, Sahoo T. Hydrodynamic performance analysis of a shore fixed [73] Elhanafi A, Fleming A, Macfarlane G, Leong Z. Numerical hydrodynamic analysis
oscillating water column wave energy converter in the presence of bottom of an offshore stationary–floating oscillating water column–wave energy
variations. In: Proceedings of the institution of mechanical engineers. Part M: converter using CFD. Int J Nav Archit Ocean Eng 2016;9:77–99.
Journal of Engineering for the Maritime Environment; 2019. p. 1–11. [74] Elhanafi A, Fleming A, Macfarlane G, Leong Z. Underwater geometrical impact on
[42] Rezanejad K, Bhattacharjee J, Guedes Soares C. Stepped sea bottom effects on the the hydrodynamic performance of an offshore oscillating water column–wave
efficiency of nearshore oscillating water column device. Ocean Eng 2013;70: energy converter. Renew Energy 2017;105:209–31.
25–38. [75] Iturrioz A, Guanche R, Lara JL, Vidal C, Losada IJ. Validation of OpenFOAM® for
[43] Alfonsi G. Reynolds-averaged Navier-Stokes equations for turbulence modeling. oscillating water column three-dimensional modeling. Ocean Eng 2015;107:
Appl Mech Rev 2009;62:1–20. 222–36.
[44] Windt C, Davidson J, Ringwood JV. High-fidelity numerical modelling of ocean [76] Rezanejad K, Gadelho JFM, Guedes Soares C. Hydrodynamic analysis of an
wave energy systems: a review of computational fluid dynamics-based numerical oscillating water column wave energy converter in the stepped bottom condition
wave tanks. Renew Sustain Energy Rev 2018;93:610–30. using CFD. Renew Energy 2019;135:1241–59.
[45] Versteeg HK, Malalasekera W. An introduction to computational fluid mechanics - [77] Wang C, Zhang Y. Numerical investigation on the wave power extraction for a 3D
the Finite Volume Method. second ed. Essex, England: Pearson Prentice Hall; dual-chamber oscillating water column system composed of two closely
2007. connected circular. Appl Energy 2021;295:1–19.
[46] ANSYS Inc. ANSYS fluent theory guide. WY, USA: SAS IP, Inc.; 2013. [78] Elhanafi A, Fleming A, Macfarlane G, Leong Z. Numerical energy balance analysis
[47] Launder BE, Spalding DB. The numerical computation of turbulent flows. Comput for an onshore oscillating water column–wave energy converter. Energy 2016;
Methods Appl Mech Eng 1974;3:269–89. 116:539–57.
[48] Anbarsooz M, Faramarzi A, Ghasemi A. A numerical study on the performance of [79] Elhanafi A. Prediction of regular wave loads on a fixed offshore oscillating water
fixed oscillating water column wave energy converter at steep waves. In: column - wave energy converter using CFD. J Ocean Eng Sc 2016;1:268–83.
Proceedings of the ASME 2016 conference collocated with the ASME 2016 10th [80] Windt C, Davidson J, Schmitt P, Ringwood JV. On the assessment of numerical
international conference on energy sustainability and the ASME 2016 14th wave makers in CFD simulations. J Mar Sci Eng 2019;7:1–36.
international conference on fuel cell science. Engineering and Technology; 2016. [81] Shih T-H, Liou WW, Shabbir A, Yang Z, Zhu J. A new k-ε eddy viscosity model for
[49] Liu Z, Hyun B, Jin J. Numerical analysis of wave field in OWC chamber using VOF high Reynolds number turbulent flows, vol. 24. Elsevier Sci Ltd; 1995. p. 227–38.
model. J Ocean Eng Technol 2008;22:1–6. [82] Menter FR. Zonal two equation k-ω turbulence models for aerodynamic flows. In:
[50] Liu Z, Shi H, Hyun B. Practical design and investigation of the breakwater OWC Proceedings of the 23rd fluid dynamics conference; 1992.
facility in China. Energy Convers 2009;304–8. [83] Menter F, Egorov Y. The scale-adaptive simulation method for unsteady turbulent
[51] Luo Y, Nader J-R, Cooper P, Zhu S-P. Nonlinear 2D analysis of the efficiency of flow predictions. Part 1: theory and Model Description. J Flow Turbul Combust
fixed Oscillating Water Column wave energy converters. Renew Energy 2014;64: 2010;85:113–38.
255–65. [84] Pope SB. Turbulent flows. Cambridge, England: Cambridge University Press;
[52] Luo Y, Wang Z, Peng G, Xiao Y, Zhai L, Liu X, et al. Numerical simulation of a 2000.
heave-only floating OWC (oscillating water column) device. Energy 2014;76: [85] Smagorinsky J. General circulation experiments with the primitive equations.
799–806. Mon Weather Rev 1963;91:99–164.
[53] Liu Z, Hyun B-S, Hong K. Numerical study of air chamber for oscillating water [86] Lilly DK. The representation of small-scale turbulence in numerical simulation
column wave energy convertor. China Ocean Eng 2011;25:169–78. experiments. In: Goldstine HH, editor. Proceedings of IBM scientific computing
[54] Liu Z, Jin J, Hyun B, Hong K. Review of application of VOF-based NWT on symposium on environmental sciences; 1967.
integrated OWC system. J Korean Soc Mar Environ Eng 2012;15:111–7. [87] Deardoff JW. Three-dimensional numerical study of the height and mean
[55] Teixeira PRF, Didier E. Numerical analysis of the response of an onshore structure of a heated planetary boundary layer. Boundary-Layer Meteorol 1974:
oscillating water column wave energy converter to random waves. Energy 2021; 81–106.
220:1–12. [88] Shin S, Lee K-H, Kim D-S, Kim K-H, Hong K. A study on the optimal shape of wave
[56] Bouali B, Larbi S. Sequential optimization and performance prediction of an energy conversion system using an oscillating water column. J Coast Res 2013;65:
oscillating water column wave energy converter. Ocean Eng 2017;131:162–73. 1663–8.
[57] Armesto JA, Guanche R, Iturrioz A, Vidal C, Losada IJ. Identification of state- [89] Thorimbert Y, Latt J, Cappietti L, Chopard B. Virtual wave flume and Oscillating
space coefficients for oscillating water columns using temporal series. Ocean Eng Water Column modeled by lattice Boltzmann method and comparison with
2014;79:43–9. experimental data. Int J Mar Energy 2016;14:41–51.
[58] Iturrioz A, Guanche R, Armesto JA, Alves MA, Vidal C, Losada IJ. Time-domain [90] Simonetti I, Cappietti L, Elsafti H, Oumeraci H. Evaluation of air compressibility
modeling of a fixed detached oscillating water column towards a floating multi- effects on the performance of fixed OWC wave energy converters using CFD
chamber device. Ocean Eng 2014;76:65–74. modelling. Renew Energy 2018;119:741–53.
[59] Teixeira PRF, Davyt DP, Didier E, Ramalhais R. Numerical simulation of an [91] Simonetti I, Cappietti L, Elsafti H, Oumeraci H. Optimization of the geometry and
oscillating water column device using a code based on Navier-Stokes equations. the turbine induced damping for fixed detached and asymmetric OWC devices: a
Energy 2013;61:513–30. numerical study. Energy 2017;139:1197–209.
[60] López I, Pereiras B, Castro F, Iglesias G. Optimisation of turbine-induced damping [92] Simonetti I, Crema I, Cappietti L, El Safti H, Oumeraci H. Site-specific
for an OWC wave energy converter using a RANS-VOF numerical model. Appl optimization of an OWC wave energy converter in a Mediterranean area. In:
Energy 2014;127:105–14. Progress in renewable energies offshore - guedes soares. Proceedings of the 2nd
[61] Connell KO, Thiebaut F, Kelly G, Cashman A. Development of a free heaving OWC international conference on renewable energies offshore (RENEW2016); 2016.
model with non-linear PTO interaction. Renew Energy 2018;117:108–15. [93] Simonetti I, Cappietti L, Safti HE, Manfrida G, Matthies H, Oumeraci H. The use of
[62] Zhan J-M, Fan Q, Hu W-Q, Gong Y-J. Hybrid realizable k− ε/laminar method in OpenFOAM as a virtual laboratory to simulate oscillating water column wave
the application of 3D heaving OWCs. Renew Energy 2020;155:691–702. energy converters. In: Proceedings of the 6th international conference on
[63] Wilcox DC. Turbulence modeling for CFD. third ed. La Canada, California: DCW computational methods in marine engineering; 2015.
Industries, Inc.; 2006. [94] Simonetti I, Cappietti L, Safti HE, Oumeraci H. 3D numerical modelling of
[64] Kamath A, Bihs H, Arntsen ØA. Numerical investigations of the hydrodynamics of oscillating water column wave energy conversion devices: current knowledge and
an oscillating water column device. Ocean Eng 2015;102:40–50. OpenFOAM® implementation. In: Progress in renewable energies offshore -
[65] Kamath A, Bihs H, Arntsen ØA. Numerical modeling of power take-off damping in guedes soares. Proceedings of the 1st international conference on renewable
an oscillating water column device. Int J Mar Energy 2015;10:1–16. energies offshore (RENEW2014); 2014.
[66] Haghighi AT, Nikseresht AH, Hayati M. Numerical analysis of hydrodynamic [95] Spalart PR, Jou WH, Strelets MK, Allmaras SR. Comments on the feasibility of LES
performance of a dual-chamber Oscillating Water Column. Energy 2021;221:1–9. for wings, and on a hybrid RANS/LES approach. In: Advances in DNS/LES,
[67] Scarpetta F, Torresi M, Camporeale SM, Filianoti PF. CFD simulation of the proceedings of the 1st AFOSR international conference on DNS/LES; 1997.
unsteady flow in an Oscillating Water Column: comparison between numerical [96] Gingold RA, Monaghan JJ. Smoothed particle hydrodynamics: theory and
and experimental results for a small scale experimental device. In: Proceedings of application to non-spherical stars. Mon Not Astron Soc 1977;181:375–89.
the 12th European wave and tidal energy conference; 2017. [97] Lucy LB. A numerical approach to the testing of the fission hypothesis. Astron J
[68] Elhanafi A, Macfarlane G, Fleming A, Leong Z. Experimental and numerical 1977;82:1013–24.
investigations on the hydrodynamic performance of a floating–moored oscillating [98] Monaghan JJ. Smoothed particle hydrodynamics. Annu Rev Astron Astrophys
water column wave energy converter. Appl Energy 2017;205:369–90. 1992;30:543–74.
[69] Elhanafi A, Macfarlane G, Fleming A, Leong Z. Experimental and numerical [99] Monaghan JJ. Simulating free surface flows with SPH. J Comput Phys 1994;110:
investigations on the intact and damage survivability of a floating–moored 399–406.
oscillating water column device. Appl Ocean Res 2017;68:276–92. [100] Dalrymple RA, Rogers BD. Numerical modeling of water waves with the SPH
[70] Elhanafi A, Macfarlane G, Fleming A, Leong Z. Experimental and numerical method. Coast Eng 2006;53:141–7.
measurements of wave forces on a 3D offshore stationary OWC wave energy [101] Gomez-Gesteira M, Dalrymple RA. Using a three-dimensional smoothed particle
converter. Ocean Eng 2017;144:98–117. hydrodynamics method for wave impact on a tall structure. J Waterw Port, Coast
Ocean Eng 2004;130:63–9.

14
F. Opoku et al. Renewable and Sustainable Energy Reviews 174 (2023) 113124

[102] Shao S, Lo EYM. Incompressible SPH method for simulating Newtonian and non- [122] Rhee SH, Makarov BP, Krishinan H, Ivanov V. Assessment of the volume of fluid
Newtonian flows with a free surface. Adv Water Resour 2003;26:787–800. method for free-surface wave flow. J Mar Sci Technol 2005;10:173–80.
[103] Shi H, Yu X, Dalrymple RA. Development of a two-phase SPH model for sediment [123] Darwish M, Moukalled F. Convective schemes for capturing interfaces of free-
laden flows. Comput Phys Commun 2017;221:259–72. surface flows on unstructured grids. Numer Heat Tran, Part B; Fund, An Intern J
[104] Wen H, Ren B, Dong P, Wang Y. A SPH numerical wave basin for modeling wave- of Comp Method 2007;49:19–42.
structure interactions. Appl Ocean Res 2016;59:366–77. [124] Ubbink O, Issa RI. A method for capturing sharp fluid interfaces on arbitrary
[105] Yang XF, Peng SL, Liu MB. A new kernel function for SPH with applications to free meshes. J Comput Phys 1999;153:26–50.
surface flows. Appl Math Model 2014;38:3822–33. [125] Waclawczyk T, Koronowicz T. Comparison of CICSAM and HRIC high-resolution
[106] Wen H, Ren B, Yu X. An improved SPH model for turbulent hydrodynamics of a schemes for interface capturing. J Theor Appl Mech 2008;46:325–45.
2D oscillating water chamber. Ocean Eng 2018;150:152–66. [126] Chen S, Johnson DB, Raad PE, Fadda D. The surface marker and micro cell
[107] Zhu G, Graham D, Zheng S, Hughes J, Greaves D. Hydrodynamics of onshore method. Int J Numer Methods Fluid 1997;25:749–78.
oscillating water column devices: a numerical study using smoothed particle [127] Scardovelli R, Zaleski S. Direct numerical simulation of free-surface and
hydrodynamics. Ocean Eng 2020;218:1–14. interfacial flow. Annu Rev Fluid Mech 1999;31:567–603.
[108] Didier E, Neves DRCB, Teixeira PRF, Días J, Neves MG. Smoothed particle [128] Weymouth GD, Yue DK-P. Conservative Volume-of-Fluid method for free-surface
hydrodynamics numerical model for modeling an oscillating water chamber. simulations on cartesian grids. J Comput Phys 2009;229:2853–65.
Ocean Eng 2016;123:397–410. [129] Gueyffier D, Li J, Nadim A, Scardovelli R, Zaleski S. Volume-of-Fluid interface
[109] Crespo AJC, Altomare C, Dominquez JM, Gonzalez-Cao J, Gomez-Gesteira M. tracking with smoothed surface stress methods for three-dimensional flows.
Towards simulating floating offshore oscillating water column converters with J Comput Phys 1999;152:423–56.
Smoothed Particle Hydrodynamics. Coast Eng 2017;126:11–26. [130] Youngs DL. Time-dependent multi-material flow with large fluid distortion.
[110] Sepri M. Application of wave generator theory to the development of a Wave Numer Methods Fluid Dyn 1982:273–85.
Energy Converter. Melbourne, Florida, USA, September: Master of Science, [131] Beygi NS, Hakimzadeh H, Chenaglou MR. Simulation of free surface flows using
Florida Institute of Technology; 2008. volume of fluid method and genetic algorithm. J Hydroinf 2014;16:1110–24.
[111] Wang S. Plunger-type wavemakers: theory and experiment. J Hydraul Res 1974; [132] Kleefsman KMT, Fekken G, Veldman AEP, Iwanowski B, Buchner B. A Volume-of-
12:357–88. Fluid based simulation method for wave impact problems. J Comput Phys 2005;
[112] Schmitt P, Elsaesser B. In: A review of wave makers for 3D numerical simulations. 206:363–93.
Proceedings of the VI International Conference on Computational Methods in [133] Chen J, Wen H, Wang Y, Wang G. A correlation study of optimal chamber width
Marine Engineering; 2015. MARINE 2015. with the relative front wall draught of onshore OWC device. Energy 2021;225:2.
[113] Aliabadi FH, Ghadimi P, Djeddi SR, Dashtimanesh A. 2-D numerical wave tank by [134] Wang C, Zhang Y. Hydrodynamic performance of an offshore Oscillating Water
boundary element method using different numerical techniques. Global J Math Column device mounted over an immersed horizontal plate: a numerical study.
Anal 2013;1:11–21. Energy 2021;222:1–15.
[114] Kh MZK, Mazaheri S, Mazyak AR. Wave generation in a numerical wave tank. Int [135] Huang Z, Huang S. Two-phase flow simulations of fixed 3D oscillating water
J Coast Offshore Eng 2017;5:33–43. columns using OpenFOAM: a comparison of two methods for modeling quadratic
[115] Grilli ST, Skourup J, Svendsen IA. An efficient boundary element method for power takeoff. Ocean Eng 2021;232:3.
nonlinear water waves. Eng Anal Bound Elem 1989;6:97–107. [136] Karch GK, Sadlo F, Meister C, Rauschenberger P, Eisenschmidt K, Weigand B,
[116] Longuet-Higgins MS, Cokelet ED. The deformation of steep surface waves on Ertl T. In: Visualization of piecewise linear interface calculation. Proceedings in
water - I. A numerical method of computation. In: Proceedings of the royal society IEEE Pacific Visualization Symposium; 2013.
A,. Math. Phys. Sci.; 1976. [137] Youngs DL. An interface tracking method for a 3D Eulerian Hydrodynamics code.
[117] Cao H-J, Wan D-C. Development of multidirectional nonlinear numerical wave AWRE; 1984. p. 3–20. Technical Report 44/92/35.
tank by naoe-FOAM-SJTU solver. Int J Ocean Syst Eng 2014;4:49–56. [138] Noh WF, Woodward P. In: Proceedings of the fifth international conference on
[118] Bihs H, Kamath A, Chella MA, Aggarwal A, Arntsen ØA. A new level set numerical numerical methods in fluid dynamics. Simple line interface calculation; 1976.
wave tank with improved density interpolation for complex wave hydrodynamics. [139] Kothe DB, Rider WJ. Comments on modeling interfacial flows with Volume-of-
Comput Fluids 2016;140:191–208. Fluid methods. CiteSeerX 1995:3–10.
[119] Tian X, Wang Q, Liu G, Deng W, Gao Z. Numerical and experimental studies on a [140] DeBar RB. Fundamentals of the KRAKEN code [Eulerian hydrodynamics code for
three-dimensional numerical wave tank. IEEE Access 2018;6:6585–93. compressible nonviscous flow of several fluids in two-dimensional (axially
[120] Hirt CW, Nichols BD. Volume of fluid (VOF) method for the dynamics of free symmetric) region], United States 1974.
boundaries. J Comput Phys 1981;39:201–25. [141] Pilliod JE, Puckett EG. Second-order accurate volume-of-fluid algorithms for
[121] François M. A study of the volume of fluid method for moving boundary tracking material interfaces. J Comput Phys 2004;199:465–502.
problems. Master of science. Daytona Beach: Embry-Riddle Aeronautical
University; 1998.

15

You might also like