betaFeedback

WorksPhysics and Astronomy

Wave energy conversion by floating and submerged piezoelectric bimorph plates

Abstract

Gaining insight into the interaction between flexible piezoelectric structures and ocean waves can inform the development of compact, high-efficiency wave-energy converters that harvest renewable energy from the marine environment. In this paper, the problem of wave energy absorption by floating and submerged piezoelectric plates is investigated. The equations of motion for a plate consisting of two piezoelectric layers separated by an elastic substrate are derived in dimensional form from the full piezoelectric constitutive laws. A solution using a modal expansion method is proposed, in which the component radiation and diffraction problems are reduced to hypersingular integral equations and solved numerically using a constant panel method. The method is general and can solve the equations of motion for submerged rigid, flexible elastic or flexible piezoelectric plates. Extensive numerical results for the energy absorption and efficiency are given for a range of parameters, including different piezoelectric materials: polyvinylidene fluoride (PVDF) and lead zirconate titanate (PZT-5H). Importantly, greater energy absorption is obtained for submerged plates when compared to plates floating on the surface. Furthermore, clamped boundary conditions give slightly larger energy absorption compared to the simply supported case. Our open-source code is provided at https://github.com/zjwegert/SemiAnalyticWECs.jl.

Introduction

The efficient and cost-effective extraction of ocean wave energy is the subject of extensive current research [1, 2]. Numerous designs for wave energy converters (WECs) have been built or proposed [3], and one type that has been the subject of recent investigation is piezoelectric WECs (PWECs). Several recent works have given an overview of technologies that use piezoelectric materials to convert the mechanical motion of ocean waves into electrical energy and discussed different types of structures, piezoelectric materials, their operating principles, advantages, and limitations [4, 5, 6, 7, 8, 9]. In particular, Kargar and Hao [5] provide a comprehensive compilation of 85 piezoelectric energy harvester designs tailored for oceanic applications, which are categorised based on their configurations (e.g., cantilever beam, diaphragm, stacked, and cymbal), materials, coupling modes, operational locations, and power output ranges. In terms of mathematical modelling, the first papers to investigate submerged elastic plates were theoretical extensions of the work on floating elastic plates, and had no clear application [10, 11, 12]. Building on this work, Renzi [13] was one of the first authors to analytically couple the hydroelectromechanical behaviour of such systems. The paper presents a fully coupled theoretical model that integrates fluid dynamics, structural mechanics, and piezoelectric effects to describe how a flexible plate with embedded piezoelectric layers can convert ocean wave energy into electricity.

Based on Renzi’s work [13], PWECs have been extensively studied with particular attention to the placement of piezoelectric plates (floating versus submerged). Here, we first discuss recent studies of submerged piezoelectric plates. Zheng et al. [14, 15] analysed a WEC comprised of a submerged flexible plate with piezoelectric layers, mounted in front of a floating breakwater, using a coupled hydro-elastic and electromechanical model based on linear potential flow and eigenfunction matching. Based on their multi-parameter study, Zheng et al. [14] concluded that deeper submergence of the plate causes the main peaks in the frequency response of the absorption efficiency to become lower in magnitude and narrower in bandwidth. Additionally, Zheng et al. [15] found that increasing the width of the PWEC can lead to more peaks in the frequency response of the wave power absorption efficiency. Vipin and Koley [16] considered a WEC comprising a flexible submerged plate with piezoelectric layers attached on both faces, mounted over a seabed with variable (undulated) topography. Using potential-flow theory combined with Euler-Bernoulli plate theory and a boundary-element method to solve the boundary-value problem, they showed that seabed undulation and plate boundary conditions significantly influence the hydrodynamic excitation and therefore the predicted energy harvesting potential. The seabed topography and structural tuning in submerged piezoelectric WEC design is also topic of consideration in other recent work [17, 18, 19].

Recently, Trivedi et al. [20] extended the earlier work of Vipin and Koley [16] by considering two horizontal piezoelectric plates over an undulated seabed. The results demonstrate that dual piezoelectric plates are capable of generating more power than the single piezoelectric plate system. Trivedi and Koley [21] focus on optimising the performance of PWECs by employing the Taguchi method (a multi-factor optimisation process) to identify the most effective design parameters for the devices. The Taguchi method has also been used to study a submerged piezoelectric plate integrated with a breakwater over an undulated seabed [19] and a hybrid WEC combining a flexible piezoelectric plate and an oscillating water column device mounted over an undulated seabed [18]. Shoele [22] investigated a flexible piezoelectric plate submerged near the free surface subjected to both gravity waves and ocean current using an inviscid-fluid model to capture the hydrodynamic coupling between the plate, incident waves, and current. The study indicates that optimal performance occurs when the plate is submerged to a depth less than half its length. Using an artificial neural network model to optimise the design of a submerged PWEC, Vipin and Koley [23] found that maximum power can be extracted when the ratio of the plate half-length to the ocean depth is approximately 2.7–2.9 and the submergence depth is approximately 1.2% of the ocean depth. We note that instead of a rectangular plate, other geometries of submerged piezoelectric plate WECs have also been considered, such as a circular cylinder [24] and a U-shaped geometry [25]. It should be noted that although PWECs are a promising method for power generation, currently proposed PWECs lack broadband energy capture and are not highly efficient across large frequency ranges. It may be possible to resolve this by having multiple devices that are tuned to different frequency ranges as investigated by Westcott et al. [26] for buoy WECs. In the context of PWECs, tuning the frequency range could be achieved by adjusting parameters including the submergence depth, surface conductivity, and poling angle. We therefore consider varying these quantities in our numerical results.

Although much research has focused on submerged plates due to the potentially higher power generation, there are also a number of recent works that focus on floating WECs. Meylan et al. [27] investigated the wave scattering, motion and energy absorption of floating elastic plates with an imaginary component to their bending rigidity. The imaginary part was included because energy can be extracted from bending by leveraging the piezoelectric effect. More recently, the hydrodynamic component of that work [27] was extended to three-dimensional water waves, in particular considering rectangular anisotropic plates on the fluid surface, however the application to PWECs was not considered directly [28]. For the study of floating plates that incorporate the piezoelectric effect, Li et al. [29] investigated the hydroelastic behaviour of a piezoelectric plate placed on the wave surface using viscous flow theory and coupled it with a finite element approach for the elastic deformation of the plate and the calculation of the static electric field. They chose surface placement to increase the static pressure differential between the top and bottom of the plate and amplify the wave-induced loads. Although the above papers consider rectangular plates, there are other configurations of floating PWECs that are designed to improve efficiency in energy harvesting (see, for example, the work of Chen et al. [30, 31], Cai et al. [32], Liu et al. [33]). We also note that Zheng et al. [34] and Michele et al. [35] have considered wave energy conversion by circular flexible plates that are either submerged or surface mounted, respectively, however they considered power takeoff mechanisms unrelated to piezoelectricity. Piezoelectric effects have not yet been considered for circular plates.

Renzi’s early hydroelectromechanical analysis [13] presented non-dimensionalised governing equations for a submerged bimorph PWEC comprising two oppositely-poled piezoelectric layers separated by an elastic substrate. Our derivation in this paper follows the key steps of Renzi [13], however we derive the governing equations in dimensional form and begin with the full, three-dimensional piezoelectric constitutive laws. We provide this full derivation relevant to both surface and submerged PWECs and fill in additional details. The form of our derivation is particularly advantageous because it can readily be extended to three dimensions, and in this paper we incorporate an arbitrary rotation of the poling angle for the two oppositely-poled piezoelectric layers in the PWEC. We consider two types of piezoelectric materials, namely polyvinylidene fluoride (PVDF) and lead zirconate titanate (PZT-5H). Both PVDF and PZT-5H are commonly used in PWEC applications [5]. As discussed by Kargar and Hao [5], PZT-type materials can produce more electricity due to their large piezoelectric coupling, while PVDF possesses a better load-bearing capacity.

The objectives of this paper are to i) present a detailed derivation of the governing equations for PWECs, including an extension to an arbitrary poling angle; ii) develop a novel and efficient solution method based on a hypersingular boundary integral equation; and iii) evaluate and compare the power conversion performance of PVDF and PZT-5H as well as different parameter regimes. The outline of the paper is as follows. In §2 we outline the mathematical theory of a submerged piezoelectric bimorph. In §3 we present a novel solution method using a matrix presentation of the fluid operator, mimicking the standard method in general hydroelasticity. In §4 we give a detailed numerical study, focusing on realistic piezoelectric materials. We end with a brief conclusion in §5, followed by three appendices that give, in turn, the derivations of the material constants, the complex analytic method used to calculate the Green’s function numerically, and a validating calculation that can serve as a benchmark for submerged elastic plates.

Mathematical theory

Equations of motion for a piezoelectric bimorph

In the following, we derive the equations of motion for a piezoelectric bimorph arranged in series as shown in Figure 1. The configuration is similar to Renzi [13], with extension to include an arbitrary rotation of the poling angle for the piezoelectric layers. The derivation is also based on the approach of Renzi [13], however, we instead give the derivation in dimensional form. We also provide more details, including beginning the derivation with the full material tensors that describe a three-dimensional piezoelectric material.

Visualisation of a piezoelectric bimorph

Figure 1. Visualisation of a piezoelectric bimorph circuited in series comprising two oppositely poled piezoelectric layers with arbitrary poling direction θ\theta, resistance RR, and thicknesses dpd_p and d0d_0 for the piezoelectric layers (pink) and elastic substrate (blue), respectively. The black dashed lines represent the electrodes on the top and bottom of each piezoelectric plate.

Constitutive laws

As indicated in Figure 1, the two plates in the piezoelectric bimorph have opposite poling directions that relate to the alignment of electric dipoles within each piezoelectric layer. A poling angle of θ=0∘\theta= 0^\circ relates to the vertical poling considered by Renzi [13], and to consider arbitrary θ\theta we apply an appropriate coordinate transformation to the base linear piezoelectric material constants for each of the piezoelectric layers.

We introduce the following tensors describing the behaviour of the base piezoelectric material: a compliance tensor SˉijklE\bar{S}^{E}_{ijkl}, a piezoelectric coupling tensor dˉijk\bar{d}_{ijk}, and a dielectric tensor κˉijσ\bar{\kappa}^{\sigma}_{ij}. The superscripts EE and σ\sigma that appear on the compliance tensor and dielectric tensor denote that these tensors are specified for constant electric field and stress, respectively. In Voigt notation, the three piezoelectric tensors for an orthotropic piezoelectric material poled in the zz direction can be written as [36]

SˉE=(Sˉ11ESˉ12ESˉ13E000Sˉ12ESˉ22ESˉ23E000Sˉ13ESˉ23ESˉ33E000000Sˉ44E000000Sˉ55E000000Sˉ66E),(1)\bar{S}^{E} = \begin{pmatrix} \bar{S}^{E}_{11} & \bar{S}^{E}_{12} & \bar{S}^{E}_{13} & 0 & 0 & 0 \\ \bar{S}^{E}_{12} & \bar{S}^{E}_{22} & \bar{S}^{E}_{23} & 0 & 0 & 0 \\ \bar{S}^{E}_{13} & \bar{S}^{E}_{23} & \bar{S}^{E}_{33} & 0 & 0 & 0 \\ 0 & 0 & 0 & \bar{S}^{E}_{44} & 0 & 0 \\ 0 & 0 & 0 & 0 & \bar{S}^{E}_{55} & 0 \\ 0 & 0 & 0 & 0 & 0 & \bar{S}^{E}_{66} \end{pmatrix}, \tag*{(1)}
dˉ=(0000dˉ150000dˉ2400dˉ31dˉ32dˉ33000),(2)\bar{d} = \begin{pmatrix} 0 & 0 & 0 & 0 & \bar{d}_{15} & 0 \\ 0 & 0 & 0 & \bar{d}_{24} & 0 & 0 \\ \bar{d}_{31} & \bar{d}_{32} & \bar{d}_{33} & 0 & 0 & 0 \end{pmatrix}, \tag*{(2)}
κˉσ=(κˉ11σ000κˉ22σ000κˉ33σ),(3)\bar{\kappa}^{\sigma} = \begin{pmatrix} \bar{\kappa}^{\sigma}_{11} & 0 & 0 \\ 0 & \bar{\kappa}^{\sigma}_{22} & 0 \\ 0 & 0 & \bar{\kappa}^{\sigma}_{33} \end{pmatrix}, \tag*{(3)}

where the specific values of the nonzero components relate to the piezoelectric material being used.

When the piezoelectric material is rotated by an angle of θ\theta in the xzxz-plane, the piezoelectric material tensors transform as

SijklE(θ)=Rim(θ)Rjl(θ)Rko(θ)Rlp(θ)SˉmlopE,(4)S^{E}_{ijkl}(\theta) = R_{im}(\theta)R_{jl}(\theta)R_{ko}(\theta)R_{lp}(\theta)\bar{S}^{E}_{mlop}, \tag*{(4)}
dijk(θ)=Rim(θ)Rjl(θ)Rko(θ)dˉmlo,(5)d_{ijk}(\theta) = R_{im}(\theta)R_{jl}(\theta)R_{ko}(\theta)\bar{d}_{mlo}, \tag*{(5)}
κijσ(θ)=Rim(θ)Rjl(θ)κˉmlσ,(6)\kappa^{\sigma}_{ij}(\theta) = R_{im}(\theta)R_{jl}(\theta)\bar{\kappa}^{\sigma}_{ml}, \tag*{(6)}

where Einstein’s summation convention is followed and Rij(θ)R_{ij}(\theta) are the components of a standard clockwise coordinate rotation given by

R(θ)=(cos⁡(θ)0sin⁡(θ)010−sin⁡(θ)0cos⁡(θ)).(7)R(\theta) = \begin{pmatrix} \cos(\theta) & 0 & \sin(\theta) \\ 0 & 1 & 0 \\ -\sin(\theta) & 0 & \cos(\theta) \end{pmatrix}. \tag*{(7)}

This yields the following non-zero components for the piezoelectric material tensors:

SE(θ)=(S11E(θ)S12E(θ)S13E(θ)0S15E(θ)0S12E(θ)S22E(θ)S23E(θ)0S25E(θ)0S13E(θ)S23E(θ)S33E(θ)0S35E(θ)0000S44E(θ)0S46E(θ)S15E(θ)S25E(θ)S35E(θ)0S55E(θ)0000S46E(θ)0S66E(θ)),(8)\mathbf{S}^{E}(\theta)=\begin{pmatrix} S^{E}_{11}(\theta) & S^{E}_{12}(\theta) & S^{E}_{13}(\theta) & 0 & S^{E}_{15}(\theta) & 0 \\ S^{E}_{12}(\theta) & S^{E}_{22}(\theta) & S^{E}_{23}(\theta) & 0 & S^{E}_{25}(\theta) & 0 \\ S^{E}_{13}(\theta) & S^{E}_{23}(\theta) & S^{E}_{33}(\theta) & 0 & S^{E}_{35}(\theta) & 0 \\ 0 & 0 & 0 & S^{E}_{44}(\theta) & 0 & S^{E}_{46}(\theta) \\ S^{E}_{15}(\theta) & S^{E}_{25}(\theta) & S^{E}_{35}(\theta) & 0 & S^{E}_{55}(\theta) & 0 \\ 0 & 0 & 0 & S^{E}_{46}(\theta) & 0 & S^{E}_{66}(\theta) \end{pmatrix}, \tag*{(8)}
d(θ)=(d11(θ)d12(θ)d13(θ)0d15(θ)0000d24(θ)0d26(θ)d31(θ)d32(θ)d33(θ)0d35(θ)0),(9)\mathbf{d}(\theta)=\begin{pmatrix} d_{11}(\theta) & d_{12}(\theta) & d_{13}(\theta) & 0 & d_{15}(\theta) & 0 \\ 0 & 0 & 0 & d_{24}(\theta) & 0 & d_{26}(\theta) \\ d_{31}(\theta) & d_{32}(\theta) & d_{33}(\theta) & 0 & d_{35}(\theta) & 0 \end{pmatrix}, \tag*{(9)}
κσ(θ)=(κ11σ(θ)0κ13σ(θ)0κ22σ(θ)0κ13σ(θ)0κ33σ(θ)).(10)\boldsymbol{\kappa}^{\sigma}(\theta)=\begin{pmatrix} \kappa^{\sigma}_{11}(\theta) & 0 & \kappa^{\sigma}_{13}(\theta) \\ 0 & \kappa^{\sigma}_{22}(\theta) & 0 \\ \kappa^{\sigma}_{13}(\theta) & 0 & \kappa^{\sigma}_{33}(\theta) \end{pmatrix}. \tag*{(10)}

We now can specify the constitutive law for the top piezoelectric layer shown in Figure 1 in Voigt notation as [36]

\begin{pmatrix} \boldsymbol{\varepsilon} \\ \mathbf{D} \end{pmatrix} = \begin{pmatrix} \mathbf{S}^{E}(\theta) & \mathbf{d}(\theta)^{\mathsf{T}} \\ \mathbf{d}(\theta) & \boldsymbol{\kappa}^{\sigma}(\theta) \end{pmatrix} \begin{pmatrix} \boldsymbol{\sigma} \\ \mathbf{E} \end{pmatrix}, \tag*{(11)}\end{pmatrix}

where

ε=(ε11,ε22,ε33,2ε23,2ε13,2ε12)(12)\boldsymbol{\varepsilon}=(\varepsilon_{11},\varepsilon_{22},\varepsilon_{33},2\varepsilon_{23},2\varepsilon_{13},2\varepsilon_{12}) \tag*{(12)}

and

σ=(σ11,σ22,σ33,σ23,σ13,σ12)(13)\boldsymbol{\sigma}=(\sigma_{11},\sigma_{22},\sigma_{33},\sigma_{23},\sigma_{13},\sigma_{12}) \tag*{(13)}

are the Voigt notation forms of the strain tensor εij\varepsilon_{ij} and the stress tensor σij\sigma_{ij}, respectively. In addition, E\mathbf{E} is the electric field vector and D\mathbf{D} is the electric displacement vector.

For the elastic substrate, we assume that the material is linear with a compliance tensor Sijkl0S^{0}_{ijkl} that has the same symmetries as SijklES^{E}_{ijkl}. The constitutive law for the elastic layer is given by

ε=S0σ,(14)\boldsymbol{\varepsilon}=\mathbf{S}^{0}\boldsymbol{\sigma}, \tag*{(14)}

where S0\mathbf{S}^{0} is the Voigt matrix form for Sijkl0S^{0}_{ijkl}.

We make several assumptions to simplify these three-dimensional constitutive laws. Following Erturk and Inman [37], if we assume that the plate is thin in the zz-direction and that the electrodes cover the faces perpendicular to the zz-direction, we have plane stress conditions

σ33=σ23=σ13=0,(15)\sigma_{33}=\sigma_{23}=\sigma_{13}=0, \tag*{(15)}

and electrical conditions

E1=E2=0.(16)E_{1}=E_{2}=0. \tag*{(16)}

As a result, Equations (11) and (14) can be written as

\begin{pmatrix} \varepsilon_{11} \\ \varepsilon_{22} \\ 2\varepsilon_{12} \\ D_{3} \end{pmatrix} = \begin{pmatrix} S^{E}_{11}(\theta) & S^{E}_{12}(\theta) & 0 & d_{31}(\theta) \\ S^{E}_{12}(\theta) & S^{E}_{22}(\theta) & 0 & d_{32}(\theta) \\ 0 & 0 & S^{E}_{66}(\theta) & 0 \\ d_{31}(\theta) & d_{32}(\theta) & 0 & \kappa^{\sigma}_{33}(\theta) \end{pmatrix} \begin{pmatrix} \sigma_{11} \\ \sigma_{22} \\ \sigma_{12} \\ E_{3} \end{pmatrix}, \tag*{(17)}\end{pmatrix}

and

\begin{pmatrix} \varepsilon_{11}\\ \varepsilon_{22}\\ 2\varepsilon_{12} \end{pmatrix} = \begin{pmatrix} S^{E}_{11} & S^{E}_{12} & 0\\ S^{E}_{12} & S^{E}_{22} & 0\\ 0 & 0 & S^{E}_{66} \end{pmatrix} \begin{pmatrix} \sigma_{11}\\ \sigma_{22}\\ \sigma_{12} \end{pmatrix}, \tag*{(18)}\end{pmatrix}

respectively. Rearranging Equation (11) to be in stress-electric displacement form and Equation (12) to be in stress form yields

\begin{pmatrix} \sigma_{11}\\ \sigma_{22}\\ \sigma_{12}\\ D_{3} \end{pmatrix} = \begin{pmatrix} S^{E}_{11}(\theta) & S^{E}_{12}(\theta) & 0 & 0\\ S^{E}_{12}(\theta) & S^{E}_{22}(\theta) & 0 & 0\\ 0 & 0 & S^{E}_{66}(\theta) & 0\\ -d_{31}(\theta) & -d_{32}(\theta) & 0 & 1 \end{pmatrix}^{-1} \begin{pmatrix} 1 & 0 & 0 & -d_{31}(\theta)\\ 0 & 1 & 0 & -d_{32}(\theta)\\ 0 & 0 & 1 & 0\\ 0 & 0 & 0 & \kappa^{\sigma}_{33}(\theta) \end{pmatrix} \begin{pmatrix} \varepsilon_{11}\\ \varepsilon_{22}\\ 2\varepsilon_{12}\\ E_{3} \end{pmatrix}, \tag*{(19)}\end{pmatrix}

and

\begin{pmatrix} \sigma_{11}\\ \sigma_{22}\\ \sigma_{12} \end{pmatrix} = \begin{pmatrix} S^{E}_{11} & S^{E}_{12} & 0\\ S^{E}_{12} & S^{E}_{22} & 0\\ 0 & 0 & S^{E}_{66} \end{pmatrix}^{-1} \begin{pmatrix} \varepsilon_{11}\\ \varepsilon_{22}\\ 2\varepsilon_{12} \end{pmatrix}, \tag*{(20)}\end{pmatrix}

respectively. Finally, we assume that both the forcing due to the fluid and the bending of the plate will be invariant in yy. We therefore can assume that we have plane strain in yy, as in [38], yielding the strain conditions

ε22=ε12=0.(21)\varepsilon_{22}=\varepsilon_{12}=0. \tag*{(21)}

The constitutive law for the top thin piezoelectric plate becomes

σ11=C^11E(θ)ε11−e^31(θ)E3,(22)\sigma_{11}=\hat{C}^{E}_{11}(\theta)\varepsilon_{11}-\hat{e}_{31}(\theta)E_{3}, \tag*{(22)}
D3=e^31(θ)ε11+κ^33ε(θ)E3,(23)D_{3}=\hat{e}_{31}(\theta)\varepsilon_{11}+\hat{\kappa}^{\varepsilon}_{33}(\theta)E_{3}, \tag*{(23)}

and the constitutive law for the elastic substrate is

σ11=C^110ε11,(24)\sigma_{11}=\hat{C}^{0}_{11}\varepsilon_{11}, \tag*{(24)}

where

C^11E(θ)=S22E(θ)S11E(θ)S22E(θ)−S12E(θ)2,(25)\hat{C}^{E}_{11}(\theta)=\frac{S^{E}_{22}(\theta)}{S^{E}_{11}(\theta)S^{E}_{22}(\theta)-S^{E}_{12}(\theta)^2}, \tag*{(25)}
e^31(θ)=d32(θ)S12E(θ)−d31(θ)S22E(θ)S11E(θ)S22E(θ)−S12E(θ)2,(26)\hat{e}_{31}(\theta)=\frac{d_{32}(\theta)S^{E}_{12}(\theta)-d_{31}(\theta)S^{E}_{22}(\theta)}{S^{E}_{11}(\theta)S^{E}_{22}(\theta)-S^{E}_{12}(\theta)^2}, \tag*{(26)}
κ^33ε(θ)=d32(θ)2S11E(θ)−2d31(θ)d32(θ)S12E(θ)+d31(θ)2S22E(θ)−κ33σ(θ)[S11E(θ)S22E(θ)−S12E(θ)2]S11E(θ)S22E(θ)−S12E(θ)2,(27)\hat{\kappa}^{\varepsilon}_{33}(\theta)=\frac{d_{32}(\theta)^2S^{E}_{11}(\theta)-2d_{31}(\theta)d_{32}(\theta)S^{E}_{12}(\theta)+d_{31}(\theta)^2S^{E}_{22}(\theta)-\kappa^{\sigma}_{33}(\theta)\left[S^{E}_{11}(\theta)S^{E}_{22}(\theta)-S^{E}_{12}(\theta)^2\right]}{S^{E}_{11}(\theta)S^{E}_{22}(\theta)-S^{E}_{12}(\theta)^2}, \tag*{(27)}
C^110=S220S110S220−(S120)2.(28)\hat{C}^{0}_{11}=\frac{S^{0}_{22}}{S^{0}_{11}S^{0}_{22}-(S^{0}_{12})^2}. \tag*{(28)}

In the above, we use a hat to denote coefficients that have been reduced from three dimensions using plane stress conditions. The expressions for these in terms of components of the piezoelectric material tensors in the original reference frame are given in Appendix A.

Note that for the series bimorph in Figure 1, the poling direction of the bottom plate is opposite to the top plate. This corresponds to a phase shift of 180∘180^\circ in the poling angle for the piezoelectric material constants of the bottom plate. As a result, one can show that on the bottom plate, Equations (16a) and (16b) become

σ11=C^11E(θ)ε11+e^31(θ)E3,(29)\sigma_{11}=\hat{C}^{E}_{11}(\theta)\varepsilon_{11}+\hat{e}_{31}(\theta)E_{3}, \tag*{(29)}
D3=−e^31(θ)ε11+κ^33(θ)E3.(30)D_{3}=-\hat{e}_{31}(\theta)\varepsilon_{11}+\hat{\kappa}_{33}(\theta)E_{3}. \tag*{(30)}

Circuiting

Before we consider the equations for the bending of the plate and the resulting charge, we must first derive the equations that relate the voltage to the electric field. Connecting piezoelectric devices in series or in parallel controls whether voltage or charge is accumulated, respectively [39, 40, 41]. The piezoelectric bimorph shown in Figure 1 is connected in series through an elastic substrate. The electrodes are assumed to cover the entire top and bottom of each piezoelectric layer. With the layers connected in series, the voltage across the resistor is the sum of the voltages across the plates. Furthermore, the way the two piezoelectric layers are oppositely poled results in the same voltage across the top and bottom piezoelectric layers. That is,

V(x,t)=2Vp(x,t),(31)V(x,t)=2V_{p}(x,t), \tag*{(31)}

where V(x,t)V(x,t) is the voltage across the resistor and Vp(x,t)V_{p}(x,t) is the voltage across the top piezoelectric plate. Furthermore, Vp(x,t)V_{p}(x,t) is given by

Vp(x,t)=−∫γE(x,t)⋅dr,(32)V_{p}(x,t)=-\int_{\gamma}\mathbf{E}(x,t)\cdot\mathrm{d}\mathbf{r}, \tag*{(32)}

where E(x,t)\mathbf{E}(x,t) is the electric field and r\mathbf{r} parameterises the curve γ\gamma that connects the bottom and top electrodes on either side of the piezoelectric layer [13]. The electric field does not vary with zz due to the assumption that the electrodes cover the top and bottom of the piezoelectric layers [42]. The two components E1E_{1} and E2E_{2} are zero (see Equation (16)), and the integral can readily be computed as a one-dimensional integral to give

Vp(x,t)=−E3(x,t)dp,(33)V_{p}(x,t)=-E_{3}(x,t)d_{p}, \tag*{(33)}

and therefore

V(x,t)=−2E3(x,t)dp.(34)V(x,t)=-2E_{3}(x,t)d_{p}. \tag*{(34)}

Bending

We can now derive the equation for plate bending in terms of displacement, voltage, and surface load. Assuming that the layers are perfectly bonded, applying a Kirchhoff thin plate assumption, and neglecting xx and yy axial displacements, the equation of motion of the plate is given by [38]

∂x2M11−Ib∂t2W=−P,(35)\partial_{x}^{2}M_{11}-I_{b}\partial_{t}^{2}W=-P, \tag*{(35)}

where W(x,t)W(x,t) is the out-of-plane displacement, P(x,t)P(x,t) is the surface load perpendicular to the plate, and M11M_{11} is the bending moment given by

M11=∫−db/2db/2zσ11 dz.(36)M_{11}=\int_{-d_{b}/2}^{d_{b}/2}z\sigma_{11}\,\mathrm{d}z. \tag*{(36)}

where db=d0+2dpd_b = d_0 + 2d_p is the total height of the plate (refer to Figure 1). Furthermore, IbI_b is the first mass moment of inertia given by

Ib=∫−db/2db/2ρ(z) dz=d0ρ0+2dpρp,(37)I_b = \int_{-d_b/2}^{d_b/2} \rho(z)\,\mathrm{d}z = d_0\rho_0 + 2d_p\rho_p, \tag*{(37)}

where ρ0\rho_0 and ρp\rho_p are the constant densities of the elastic and piezoelectric materials, respectively.

In the absence of in-plane displacements, ε11\varepsilon_{11} is related to the out-of-plane displacement via [38]

ε11=−z∂x2W.(38)\varepsilon_{11} = -z\partial_x^2 W. \tag*{(38)}

The constitutive relations then give the stress for the top piezoelectric plate and elastic substrate as

σ11=−zC^11E(θ)∂x2W−e^31(θ)E3,(39)\sigma_{11} = -z\widehat{C}_{11}^{E}(\theta)\partial_x^2 W - \widehat{e}_{31}(\theta)E_3, \tag*{(39)}

and

σ11=−zC^110∂x2W,(40)\sigma_{11} = -z\widehat{C}_{11}^{0}\partial_x^2 W, \tag*{(40)}

respectively, which have been obtained simply by putting Equation (27) into the two constitutive relations Equation (16a) and Equation (17). Using Equation (23) that relates the electric field component E3E_3 to the voltage across the resistor, Equation (28) then simplifies to

σ11=−zC^11E(θ)∂x2W+e^31(θ)2dpV.(41)\sigma_{11} = -z\widehat{C}_{11}^{E}(\theta)\partial_x^2 W + \frac{\widehat{e}_{31}(\theta)}{2d_p}V. \tag*{(41)}

Substituting Equation (29) and (30) into Equation (25) and performing the integration gives the bending moment as

M11(x,t)=−B(θ)∂x2W+η(θ)V,(42)M_{11}(x,t) = -B(\theta)\partial_x^2 W + \eta(\theta)V, \tag*{(42)}

where B(θ)B(\theta) and η(θ)\eta(\theta) are given by

B(θ)=d0312C^110+2dp(dp23+d0dp2+d024)C^11E(θ),(43)B(\theta) = \frac{d_0^3}{12}\widehat{C}_{11}^{0} + 2d_p\left(\frac{d_p^2}{3} + \frac{d_0d_p}{2} + \frac{d_0^2}{4}\right)\widehat{C}_{11}^{E}(\theta), \tag*{(43)}
η(θ)=12(d0+dp)e^31(θ).(44)\eta(\theta) = \frac{1}{2}(d_0+d_p)\widehat{e}_{31}(\theta). \tag*{(44)}

Substituting this expression for the bending moment (Eqn (31)) into Equation (24) yields an equation relating the displacement WW, voltage on the resistor VV, and surface load PP:

−B(θ)∂x4W+η(θ)∂x2V−Ib∂t2W=−P.(45)-B(\theta)\partial_x^4 W + \eta(\theta)\partial_x^2 V - I_b\partial_t^2 W = -P. \tag*{(45)}

Charge

In order to derive the electrical equations for the system, we shall consider the electric displacement for the top plate. This is given by the constitutive relation (Eqn (16b)) as

D3=e^31(θ)ε11+κ^33(θ)E3,(46)D_3 = \widehat{e}_{31}(\theta)\varepsilon_{11} + \widehat{\kappa}_{33}(\theta)E_3, \tag*{(46)}

or

D3=−ze^31(θ)∂x2W−κ^33e(θ)2dpV(47)D_3=-z\hat{e}_{31}(\theta)\partial_x^2W-\frac{\hat{\kappa}^{e}_{33}(\theta)}{2d_p}V \tag*{(47)}

using Equation (38) and Equation (34) that relate the strain and electric field components to the out-of-plane displacement and voltage, respectively. In the literature [e.g., 13, 42, 43], it is common to take an expansion in zz about the midpoint z=d0+dp2z=\frac{d_0+d_p}{2} of the top piezoelectric layer so that D3D_3 can be rewritten as

D3=−12(d0+dp)e^31(θ)∂x2W−κ^33e(θ)2dpV+O(z−d0+dp2).(48)D_3=-\frac{1}{2}(d_0+d_p)\hat{e}_{31}(\theta)\partial_x^2W-\frac{\hat{\kappa}^{e}_{33}(\theta)}{2d_p}V+O\left(z-\frac{d_0+d_p}{2}\right). \tag*{(48)}

This is equivalent to replacing the bending strain with the average bending strain in the piezoelectric layer. Because we assume that the plate is thin (d0+2dp≪Ld_0+2d_p\ll L, where LL is the half-length of the plate), the higher-order terms can be neglected.

The average charge accumulated over each piezoelectric layer is given by [13]

Q^=∫AD⋅n dA,(49)\hat{Q}=\int_A \mathbf{D}\cdot\mathbf{n}\,\mathrm{d}A, \tag*{(49)}

where AA is the electrode area. Denoting the average electric charge per unit area as Q(x,t)=dQ^dA=D3Q(x,t)=\frac{\mathrm{d}\hat{Q}}{\mathrm{d}A}=D_3, we have

Q(x,t)=−η(θ)∂x2W−C(θ)V,(50)Q(x,t)=-\eta(\theta)\partial_x^2W-C(\theta)V, \tag*{(50)}

where we have used Equation (32b) and denoted C(θ)=κ^33e(θ)2dpC(\theta)=\frac{\hat{\kappa}^{e}_{33}(\theta)}{2d_p} to be the electrical surface capacitance. Finally, Ohm’s law gives

∂tQ=GV,(51)\partial_tQ=GV, \tag*{(51)}

where GG is the surface conductance, which is the reciprocal of the electrical resistance per unit area [13].

Equations (33), (38) and (39) fully describe the electrical and elastic dynamics of the bimorph in response to the pressure load PP of the fluid.

Time harmonic formulation

Suppose that the motion of the plate, the charge, the voltage, and the surface load are harmonic with frequency ω\omega. Specifically, we take

W(x,t)=Re⁡{w(x)e−iωt},(52)W(x,t)=\operatorname{Re}\{w(x)e^{-i\omega t}\}, \tag*{(52)}
Q(x,t)=Re⁡{q(x)e−iωt},(53)Q(x,t)=\operatorname{Re}\{q(x)e^{-i\omega t}\}, \tag*{(53)}
V(x,t)=Re⁡{v(x)e−iωt},(54)V(x,t)=\operatorname{Re}\{v(x)e^{-i\omega t}\}, \tag*{(54)}
P(x,t)=Re⁡{p(x)e−iωt}.(55)P(x,t)=\operatorname{Re}\{p(x)e^{-i\omega t}\}. \tag*{(55)}

where Re⁡\operatorname{Re} denotes the real part, tt is time, and the scalar functions ww, qq, vv, and pp are complex-valued displacement, charge, voltage, and surface load. In the following, we omit Re⁡\operatorname{Re} for brevity. Substituting the above into Equation (45), (38), and (39) gives

−B(θ)∂x4w+η(θ)∂x2v+Ibω2w=−p,(56)-B(\theta)\partial_x^4w+\eta(\theta)\partial_x^2v+I_b\omega^2w=-p, \tag*{(56)}
q(x)=−η(θ)∂x2w−C(θ)v,(57)q(x) = -\eta(\theta)\partial_x^2 w - C(\theta)v, \tag*{(57)}

and

−iωq=Gv,(58)-\mathrm{i}\omega q = Gv, \tag*{(58)}

respectively. Substituting Equation (41c) into Equation (41b) and rearranging for v(x)v(x) gives

v(x)=iωη(θ)G−iωC∂x2w.(59)v(x) = \frac{\mathrm{i}\omega\eta(\theta)}{G-\mathrm{i}\omega C}\partial_x^2 w. \tag*{(59)}

Using this, Equation (41a) can then be written as

[B(θ)−iωη(θ)2G−iωC]∂x4w−Ibω2w=p,(60)\left[B(\theta)-\frac{\mathrm{i}\omega\eta(\theta)^2}{G-\mathrm{i}\omega C}\right]\partial_x^4 w-I_b\omega^2w=p, \tag*{(60)}

or equivalently

D∂x4w−Ibω2w=p,(61)D\partial_x^4w-I_b\omega^2w=p, \tag*{(61)}

where DD is given by

D=B(θ)+ωη(θ)2ωC+Gi.(62)D=B(\theta)+\frac{\omega\eta(\theta)^2}{\omega C+Gi}. \tag*{(62)}

This result is similar to the one given by Renzi [13], except Equation (62) is in dimensional form (see Equations (37), (32a) and (32b) for relevant expressions) and allows a rotation θ\theta of the poling angle of the piezoelectric layers clockwise from the vertical. It should be noted that there is a typo in the last exponent of the expression for BB in Renzi [13 Eq. (2.8)].

Equations of motion for the fluid

For completeness, we present here the standard equations of linear water wave theory. The reader is referred to Linton and McIver [44] for further details. The problem is initially posed in a three-dimensional coordinate system with the (x,y)(x,y)-plane coinciding with the undisturbed free surface of the fluid and the zz-axis pointing vertically upwards. A flat seabed is situated at the plane z=−Hz=-H, meaning the fluid occupies the region Ω~=R2×(−H,0)\Γ~\tilde{\Omega}=\mathbb{R}^2\times(-H,0)\backslash\tilde{\Gamma}, where Γ~=(−L,L)×R×{−h}\tilde{\Gamma}=(-L,L)\times\mathbb{R}\times\{-h\} is the region occupied by the infinitely thin plate, in which 2L2L is the extent of the plate in the xx direction. Note that the plate is parallel to the seabed, and can either be on the surface (h=0h=0) or submerged (−H<−h<0-H<-h<0).

Assuming an inviscid, incompressible fluid with irrotational motion implies the flow field u\mathbf{u} is of the form

u(x,y,z,t)=∇Φ(x,y,z,t)for (x,y,z)∈Ω~,(63)\mathbf{u}(x,y,z,t)=\nabla\Phi(x,y,z,t)\quad\text{for }(x,y,z)\in\tilde{\Omega}, \tag*{(63)}

where Φ\Phi, the velocity potential, satisfies

∇2Φ=0for (x,y,z)∈Ω~,(64)\nabla^2\Phi=0\quad\text{for }(x,y,z)\in\tilde{\Omega}, \tag*{(64)}

in which ∇=[∂x ∂y ∂z]T\nabla=[\partial_x\ \partial_y\ \partial_z]^{\mathsf{T}} and ∇2=∇⋅∇\nabla^2=\nabla\cdot\nabla. The linearised kinematic and dynamic free surface conditions are

∂tζ=∂zΦz=0,∂tΦ+gζ=0z=0,(65)\begin{aligned} \partial_t\zeta&=\partial_z\Phi&& z=0,\\ \partial_t\Phi+g\zeta&=0 && z=0, \tag*{(65)} \end{aligned}
∂tΦ+gζ=0z=0,(66)\partial_t\Phi+g\zeta=0\quad z=0, \tag*{(66)}

respectively, where ζ(x,y,t)\zeta(x,y,t) describes the free surface elevation. Equations (48) hold for x∈R∖(−L,L)x \in\mathbb{R}\setminus(-L,L) if the plate is on the surface or x∈Rx \in\mathbb{R} if the plate is submerged. The constant gg appearing in Equation (48b) is acceleration due to gravity. The boundary condition on the seabed is

∂zΦ=0,(67)\partial_z\Phi= 0, \tag*{(67)}

for z=−Hz=-H. We assume motion is time harmonic with angular frequency ω\omega and invariant in the yy direction, implying

Φ(x,y,z,t)=Re⁡{ϕ(x,z)e−iωt},(68)\Phi(x,y,z,t) = \operatorname{Re}\left\{\phi(x,z)e^{-\mathrm{i}\omega t}\right\}, \tag*{(68)}
ζ(x,y,t)=Re⁡{ζ^(x)e−iωt},(69)\zeta(x,y,t) = \operatorname{Re}\left\{\hat{\zeta}(x)e^{-\mathrm{i}\omega t}\right\}, \tag*{(69)}

where ϕ\phi is a complex-valued velocity potential satisfying

Δϕ=0for (x,z)∈Ω,(70)\Delta\phi= 0 \qquad\text{for } (x,z)\in\Omega, \tag*{(70)}
∂zϕ=αϕfor z=0,(71)\partial_z\phi= \alpha\phi\qquad\text{for } z=0, \tag*{(71)}
∂zϕ=0for z=−H,(72)\partial_z\phi= 0 \qquad\text{for } z=-H, \tag*{(72)}

where Δ≡∂x2+∂z2\Delta\equiv\partial_x^2+\partial_z^2 is the two-dimensional Laplacian operator and α=ω2/g\alpha=\omega^2/g. The two-dimensional fluid domain is given by Ω=(R×(−H,0))∖Γ\Omega=(\mathbb{R}\times(-H,0))\setminus\Gamma, where the region occupied by the plate’s cross section is Γ=(−L,L)×{−h}\Gamma=(-L,L)\times\{-h\}. Note that following from Equations (48), Equation (51b) holds for x∈R∖(−L,L)x\in\mathbb{R}\setminus(-L,L) if the plate is on the surface or x∈Rx\in\mathbb{R} if the plate is submerged. In the far-field the solution is subject to the Sommerfeld radiation condition. This is given by

{∂xϕsc∓ikϕsc}→0as x→±∞,(73)\left\{\partial_x\phi^{\mathrm{sc}}\mp\mathrm{i}k\phi^{\mathrm{sc}}\right\}\to0 \qquad\text{as } x\to\pm\infty, \tag*{(73)}

where kk is the positive real solution of the dispersion equation

ktanh⁡(kH)=ω2g,(74)k\tanh(kH)=\frac{\omega^2}{g}, \tag*{(74)}

ϕsc=ϕ−ϕin\phi^{\mathrm{sc}}=\phi-\phi^{\mathrm{in}} is the scattered wave potential, and

ϕin(x,z)=−igAωcosh⁡kHeikxcosh⁡k(z+H),(75)\phi^{\mathrm{in}}(x,z)=\frac{-\mathrm{i}gA}{\omega\cosh kH}e^{\mathrm{i}kx}\cosh k(z+H), \tag*{(75)}

is the velocity potential of a plane incident wave with amplitude AA. We take A=1A=1 m throughout this paper. To complete the system of equations (51), it remains to specify the boundary conditions on the plate.

The coupled system

The complex velocity potential, ϕ\phi, is coupled to the complex plate displacement function, ww, through linearised kinematic and dynamic conditions, respectively. The kinematic condition is given by

∂zϕ=−iωwfor (x,z)∈Γ,(76)\partial_z\phi=-\mathrm{i}\omega w \qquad\text{for } (x,z)\in\Gamma, \tag*{(76)}

which enforces that the fluid and plate velocities coincide at their interface. The dynamic condition (restated here from Equation (61)), is

D∂x4w−Ibω2w=pfor (x,z)∈Γ,D\partial_x^4w-I_b\omega^2w=p \qquad\text{for } (x,z)\in\Gamma,

where the loading pp is now determined from the fluid pressure as

p={iωρwϕ−ρwgw,h=0,−iωρw⟦ϕ⟧,0<h<H,(77)p = \begin{cases} i\omega\rho_w\phi-\rho_w g w, & h=0,\\ -i\omega\rho_w\llbracket\phi\rrbracket, & 0<h<H, \end{cases} \tag*{(77)}

where ρw\rho_w is the density of the fluid and ⟦ϕ⟧=ϕ+−ϕ−\llbracket\phi\rrbracket=\phi_+-\phi_- denotes the jump of ϕ\phi across the plate. The notation ϕ+\phi_+ and ϕ−\phi_- denotes the limit of ϕ\phi as z→−hz\to-h from above and below, respectively. Note that for the plate on the surface (h=0h=0), the first and second summands contributing to pp are the hydrodynamic and hydrostatic pressure, respectively. Finally, we have that the jump in the normal derivative of the potential across the submerged plate is zero, namely,

⟦∂zϕ⟧=0for (x,z)∈Γ.(78)\llbracket\partial_z\phi\rrbracket=0 \quad\text{for }(x,z)\in\Gamma. \tag*{(78)}

For the boundary conditions on the plate, we assume that the edges are either clamped:

w=∂xw=0for x=±L,(79)w=\partial_xw=0 \quad\text{for }x=\pm L, \tag*{(79)}

or simply supported:

w=∂x2w=0for x=±L.(80)w=\partial_x^2w=0 \quad\text{for }x=\pm L. \tag*{(80)}

In summary, we have the following coupled system:

Δϕ=0(x,z)∈Ω,∂zϕ=0(x,z)∈R×{H},∂zϕ=ω2gϕ(x,z)∈(R×{0})∖Γ,∂zϕ=−iωw(x,z)∈Γ,D∂x4w−Ibω2w=p(x,z)∈Γ,⟦∂zϕ⟧=0(x,z)∈Γ,{∂xϕsc∓ikϕsc}→0x→±∞.(81)\begin{aligned} \Delta\phi&=0 &&(x,z)\in\Omega,\\ \partial_z\phi&=0 &&(x,z)\in\mathbb{R}\times\{H\},\\ \partial_z\phi&=\frac{\omega^2}{g}\phi&&(x,z)\in(\mathbb{R}\times\{0\})\setminus\Gamma,\\ \partial_z\phi&=-i\omega w &&(x,z)\in\Gamma,\\ D\partial_x^4w-I_b\omega^2w&=p &&(x,z)\in\Gamma,\\ \llbracket\partial_z\phi\rrbracket&=0 &&(x,z)\in\Gamma,\\ \{\partial_x\phi^{\mathrm{sc}}\mp ik\phi^{\mathrm{sc}}\}&\to0 &&x\to\pm\infty. \tag*{(81)} \end{aligned}

with either clamped (Eqn (57)) or simply supported (Eqn (58)) boundary conditions. Figure 2 shows a schematic of the above equations.

Solution methods

Surface problem

For the plate floating on the ocean’s surface we use the approach described by Meylan et al. [27], in which an expansion in modes and a Green’s function are used to analyse the behaviour. We refer the reader to the referenced work for further details.

Submerged problem

In the following, we describe our new approach for solving the problem of a submerged plate by decomposing into a diffraction problem (in which the incident wave scatters off a fixed plate) and radiation problems (in which the fluid motion is forced by plate vibration in a given mode). We then obtain the solution of the coupled problem (81) using the modal expansion method of Newman [45]. The component diffraction and radiation problems are solved by formulating them as hypersingular boundary integral equations and solving numerically using a constant panel method. Our method extends earlier work of Meylan et al. [27] for submerged plates. Our method is advantageous compared to the eigenfunction expansion method used by Renzi [13] because it does not require the solution of a complicated dispersion relation. Furthermore, the method extends to non-horizontal plates and to two-dimensional plates of arbitrary shape, however it becomes inefficient as the plate becomes larger.

Figure 2. Visualisation of the problem setup. For the plate on the surface (h=0h = 0), the complex-valued surface load is given by p=iωρwϕ−wρwg\mathrm{p} = \mathrm{i}\omega\rho_w\phi- w\rho_w g. When the plate is submerged (−H<h<0-H < h < 0), the surface load is instead p=−iωρw⟦ϕ⟧\mathrm{p} = -\mathrm{i}\omega\rho_w\llbracket\phi\rrbracket.

Diffraction problem

The velocity potential associated with diffraction is of the form ϕ=ϕin+ϕdi\phi= \phi^{\mathrm{in}} + \phi^{\mathrm{di}}, where ϕdi\phi^{\mathrm{di}} is the as yet unknown diffraction potential. Assuming that the plate is held fixed, we have ∂z(ϕin+ϕdi)=0\partial_z(\phi^{\mathrm{in}} + \phi^{\mathrm{di}}) = 0 for (x,z)∈Γ(x,z) \in\Gamma. Since ϕ\phi also satisfies Equations (51) and ϕdi\phi^{\mathrm{di}} satisfies the Sommerfeld radiation condition (Eqn (51d)), we obtain the following system of equations for ϕdi\phi^{\mathrm{di}}

Δϕdi=0(x,z)∈Ω(82)\Delta\phi^{\mathrm{di}} = 0 \qquad(x,z) \in\Omega \tag*{(82)}
∂zϕdi=αϕdiz=0(83)\partial_z\phi^{\mathrm{di}} = \alpha\phi^{\mathrm{di}} \qquad z = 0 \tag*{(83)}
∂zϕdi=−∂zϕin(x,z)∈Γ(84)\partial_z\phi^{\mathrm{di}} = -\partial_z\phi^{\mathrm{in}} \qquad(x,z) \in\Gamma \tag*{(84)}
∂zϕdi=0z=−H(85)\partial_z\phi^{\mathrm{di}} = 0 \qquad z = -H \tag*{(85)}
(∂∣x∣−ik)ϕdi→0as ∣x∣→∞(86)(\partial_{\lvert x\rvert} - \mathrm{i}k)\phi^{\mathrm{di}} \to0 \qquad\text{as } \lvert x\rvert\to\infty \tag*{(86)}

The Green’s function for a two dimensional fluid of finite depth is [44]

2πG(x,z;x′,z′)=ln⁡r−ln⁡r~−2∫0∞cos⁡μXcosh⁡μh[cosh⁡μ(z+H)cosh⁡μ(z′+H)μsinh⁡μH−αcosh⁡μH+e−μHμsinh⁡μzsinh⁡μz′]dμ(87)2\pi G(x,z;x',z') = \ln r - \ln\tilde r - 2\int_{0}^{\infty}\frac{\cos\mu X}{\cosh\mu h}\left[\frac{\cosh\mu(z+H)\cosh\mu(z'+H)}{\mu\sinh\mu H-\alpha\cosh\mu H}+\frac{\mathrm{e}^{-\mu H}}{\mu}\sinh\mu z\sinh\mu z'\right]\mathrm{d}\mu \tag*{(87)}

in which X=x−x′X=x-x', r=(x−x′)2+(z−z′)2r=\sqrt{(x-x')^2+(z-z')^2} and r~=(x−x′)2+(z+z′)2\tilde{r}=\sqrt{(x-x')^2+(z+z')^2}, which satisfies

ΔG=δ(x−x′)δ(z−z′)(x,z)∈Ω∪Γ,∂zG=αGz=0,∂zG=0z=−H,(∂∣x∣−ik)G→0as ∣x∣→∞.(88)\begin{aligned} \Delta G&=\delta(x-x')\delta(z-z') && (x,z)\in\Omega\cup\Gamma,\\ \partial_zG&=\alpha G && z=0,\\ \partial_zG&=0 && z=-H,\\ (\partial_{\lvert x\rvert}-\mathrm{i}k)G&\to0 && \text{as } \lvert x\rvert\to\infty. \tag*{(88)} \end{aligned}

The method to numerically calculate the integral in the Green’s function expression is discussed in Appendix B. Note that GG can be represented as the sum of a singular part and a regular part as

2πG(x,y;x′,y′)=ln⁡r+M(x,y;x′,y′).(89)2\pi G(x,y;x',y')=\ln r+M(x,y;x',y'). \tag*{(89)}

Applying Green’s third identity to ϕdi\phi^{\mathrm{di}} gives

ϕdi(x,z)=−∫−LL∂z′G(x,z;x′,−h)[ ⁣[ϕdi(x′)] ⁣] dx′.(90)\phi^{\mathrm{di}}(x,z)=-\int_{-L}^{L}\partial_{z'}G(x,z;x',-h)[\![\phi^{\mathrm{di}}(x')]\!]\,\mathrm{d}x'. \tag*{(90)}

The incident forcing condition (Eqn (60c)) yields

−∂zϕin(x,−h)=−∂z∫−LL∂z′G(x,−h;x′,−h)[ ⁣[ϕdi(x′)] ⁣] dx′.(91)-\partial_z\phi^{\mathrm{in}}(x,-h)=-\partial_z\int_{-L}^{L}\partial_{z'}G(x,-h;x',-h)[\![\phi^{\mathrm{di}}(x')]\!]\,\mathrm{d}x'. \tag*{(91)}

Interchanging the derivative and the integral in the above expression is permitted provided the integral is interpreted as a Hadamard finite part integral [here denoted \fint, cf. [46]]. The Hadamard finite part integral is defined for a continuous function f:[a,b]→Cf:[a,b]\to\mathbb{C} as

\fint_a^b\frac{f(t)}{(t-s)^2}\,\mathrm{d}t=\lim_{\varepsilon\to0^+}\left\{\int_a^{s-\varepsilon}\frac{f(t)}{(t-s)^2}\,\mathrm{d}t+\int_{s+\varepsilon}^{b}\frac{f(t)}{(t-s)^2}\,\mathrm{d}t-\frac{2f(s)}{\varepsilon}\right\}. \tag*{(92)}

Equation (65) becomes

-\partial_z\phi^{\mathrm{in}}(x,-h)=-\fint_{-L}^{L}\partial_z\partial_{z'}G(x,-h;x',-h)[\![\phi^{\mathrm{di}}(x')]\!]\,\mathrm{d}x'. \tag*{(93)}

Our goal is to discretise the integral equation above using a constant panel method. We divide the interval [−L,L][-L,L] into NN subintervals of equal length with midpoints x1,…,xNx_1,\ldots,x_N and spacing xj+1−xj=Δxx_{j+1}-x_j=\Delta x. Equation (93) becomes

-\partial_z\phi^{\mathrm{in}}(x,-h)=-\sum_{j=1}^{N}\fint_{x_j-\Delta x/2}^{x_j+\Delta x/2}\partial_z\partial_{z'}G(x,-h;x',-h)[\![\phi^{\mathrm{di}}(x')]\!]\,\mathrm{d}x' \tag*{(94)}
\approx-\sum_{j=1}^{N}[\![\phi^{\mathrm{di}}(x_j)]\!]\fint_{x_j-\Delta x/2}^{x_j+\Delta x/2}\partial_z\partial_{z'}G(x,-h;x',-h)\,\mathrm{d}x', \tag*{(95)}

where we have assumed [ ⁣[ϕdi(x)] ⁣][\![\phi^{\mathrm{di}}(x)]\!] is slowly varying with respect to Δx\Delta x in order to apply the constant panel approximation. Evaluating both sides of Equation (68b) at the collocation points xix_i yields the matrix equation

−∑j=1NKij[ ⁣[ϕdi(xj)] ⁣]≈−∂zϕin(xi,−h),(96)-\sum_{j=1}^{N}K_{ij}[\![\phi^{\mathrm{di}}(x_j)]\!]\approx-\partial_z\phi^{\mathrm{in}}(x_i,-h), \tag*{(96)}

where the terms KijK_{ij} are introduced as approximations of the integrals in Equation (68b), i.e.

Kij≈∫xj−Δx/2xj+Δx/2∂z∂z′G(x,−h;x′,−h) dx′.(97)K_{ij} \approx\int_{x_j-\Delta x/2}^{x_j+\Delta x/2} \partial_z\partial_{z'}G(x,-h;x',-h)\,\mathrm{d}x'. \tag*{(97)}

Our strategy to approximate these terms is to integrate the singular part exactly (in the Hadamard finite part sense) and approximate the integral of the regular part using the midpoint rule. We first compute

2π∂z∂z′G(x,y;x′,y′)∣z=z′=−1(x−x′)2+∂z∂z′M(x,y;x′,y′).(98)\left.2\pi\partial_z\partial_{z'}G(x,y;x',y')\right|_{z=z'}=\frac{-1}{(x-x')^2}+\partial_z\partial_{z'}M(x,y;x',y'). \tag*{(98)}

Then the terms KijK_{ij} are computed as

2πKij={4Δx+∂z∂z′M(xi,−h;xi,−h)Δxi=j,1xi−xj+Δx/2−1xi−xj−Δx/2+∂z∂z′M(xi,−h;xj,−h)Δxi≠j.(99)2\pi K_{ij}= \begin{cases} \frac{4}{\Delta x}+\partial_z\partial_{z'}M(x_i,-h;x_i,-h)\Delta x & i=j,\\ \frac{1}{x_i-x_j+\Delta x/2}-\frac{1}{x_i-x_j-\Delta x/2}+\partial_z\partial_{z'}M(x_i,-h;x_j,-h)\Delta x & i\ne j. \end{cases} \tag*{(99)}

Modes of vibration of an elastic beam

The equations of motion of a uniform elastic beam satisfy the eigenvalue problem

D∂x4wn=Ibωn2wn,(100)D\partial_x^4w_n=I_b\omega_n^2w_n, \tag*{(100)}

for x∈(−L,L)x\in(-L,L), where ωn\omega_n and wnw_n are the angular frequencies and modes of vibration, respectively. We further introduce the quantities κn=Ibωn2/D4\kappa_n=\sqrt[4]{I_b\omega_n^2/D}. The eigenvalue problem Equation (100) is solved in tandem with boundary conditions at x=±Lx=\pm L. The modes of vibration satisfy the prescribed boundary conditions at x=±Lx=\pm L, namely, those for either clamped (Eqn (79)) or simply supported (Eqn (80)) plates. Note that the eigenmodes are scaled to be orthonormal, namely

∫−LLwn(x)wm(x) dx=δmn,(101)\int_{-L}^{L}w_n(x)w_m(x)\,\mathrm{d}x=\delta_{mn}, \tag*{(101)}

where δmn\delta_{mn} is the Kronecker delta. The expressions for the eigenmodes and eigenvalues can be found in Reddy [Chapt. 7.3., 38].

Radiation problem

We consider the radiation problems, whose solutions are the complex velocity potentials ϕn\phi_n resulting from excitation by forced motion of the plate in a single mode of vibration wnw_n, in the absence of an incident wave. The boundary value problem to be solved is

Δϕn=0(x,z)∈Ω,∂zϕn=αϕnz=0,∂zϕn=−iωwn(x,z)∈Γ,∂zϕn=0z=−H,(∂∣x∣−ik)ϕn→0as ∣x∣→∞.(102)\begin{aligned} \Delta\phi_n &= 0 && (x,z)\in\Omega,\\ \partial_z\phi_n &= \alpha\phi_n && z=0,\\ \partial_z\phi_n &= -\mathrm{i}\omega w_n && (x,z)\in\Gamma,\\ \partial_z\phi_n &= 0 && z=-H,\\ (\partial_{\lvert x\rvert}-\mathrm{i}k)\phi_n &\to0 && \text{as }\lvert x\rvert\to\infty. \tag*{(102)} \end{aligned}

Following the steps outlined in §3.2.1, we find

ϕn(x,z)=−∫−LL∂z′G(x,z;x′,−h)⟦ϕn(x′)⟧ dx′.(103)\phi_n(x,z)=-\int_{-L}^{L}\partial_{z'}G(x,z;x',-h)\llbracket\phi_n(x')\rrbracket\,\mathrm{d}x'. \tag*{(103)}

Accordingly, the pressure difference ⟦ϕn⟧\llbracket\phi_n\rrbracket can be computed numerically at the collocation points xjx_j by solving the linear system of equations

−∑j=1NKij⟦ϕn(xj)⟧≈−iωwn(xi).(104)-\sum_{j=1}^{N}K_{ij}\llbracket\phi_n(x_j)\rrbracket\approx-\mathrm{i}\omega w_n(x_i). \tag*{(104)}

Scattering by a piezoelectric plate

The coupled problem Equations (59) is solved using a dry modes expansion [45] of the form

ϕ=ϕin+ϕdi+∑ncnϕn,(105)\phi=\phi^{\mathrm{in}}+\phi^{\mathrm{di}}+\sum_n c_n\phi_n, \tag*{(105)}
w=∑ncnwn,(106)w=\sum_n c_nw_n, \tag*{(106)}

where the coefficients cnc_n must be determined. Substituting these into the dynamic condition of the plate (Eqn (44)) with hydrodynamic loading given by Equation (55) gives

D∑n=1∞cnκn4wn−ω2Ib∑n=1∞cnwn+iωρ∑n=1∞cn⟦ϕn⟧=−iωρ⟦ϕdi⟧,(107)D\sum_{n=1}^{\infty}c_n\kappa_n^4w_n-\omega^2I_b\sum_{n=1}^{\infty}c_nw_n+\mathrm{i}\omega\rho\sum_{n=1}^{\infty}c_n\llbracket\phi_n\rrbracket=-\mathrm{i}\omega\rho\llbracket\phi^{\mathrm{di}}\rrbracket, \tag*{(107)}

where we have used the fact that ⟦ϕin⟧=0\llbracket\phi^{\mathrm{in}}\rrbracket=0. A linear system of equations for the unknown coefficients cnc_n is obtained by truncating the sum in Equation (79) after NmN_m terms, then taking the L2(−L,L)L^2(-L,L) inner product of both sides of the resulting equation with wmw_m for m=1,…,Nmm=1,\ldots,N_m. We obtain

Dκm4cm−ω2Ibcm+iωρw∑n=1Nm(∫−LL⟦ϕn(x)⟧wm(x)‾ dx)cn=−iωρw∫−LL⟦ϕdi(x)⟧wm(x)‾ dx,(108)D\kappa_m^4c_m-\omega^2I_bc_m+\mathrm{i}\omega\rho_w\sum_{n=1}^{N_m}\left(\int_{-L}^{L}\llbracket\phi_n(x)\rrbracket\overline{w_m(x)}\,\mathrm{d}x\right)c_n=-\mathrm{i}\omega\rho_w\int_{-L}^{L}\llbracket\phi^{\mathrm{di}(x)}\rrbracket\overline{w_m(x)}\,\mathrm{d}x, \tag*{(108)}

where we have used Equation (74). In matrix form, we have

(Kstiffness−ω2Mmass−ω2Amass−iωBdamping)c=f,(109)\left(\mathbf{K}_{\mathrm{stiffness}}-\omega^2\mathbf{M}_{\mathrm{mass}}-\omega^2\mathbf{A}_{\mathrm{mass}}-\mathrm{i}\omega\mathbf{B}_{\mathrm{damping}}\right)\mathbf{c}=\mathbf{f}, \tag*{(109)}

where Kstiffness\mathbf{K}_{\mathrm{stiffness}}, Mmass\mathbf{M}_{\mathrm{mass}}, Amass\mathbf{A}_{\mathrm{mass}} and Bdamping\mathbf{B}_{\mathrm{damping}} are the Nm×NmN_m\times N_m stiffness, mass, added mass and damping matrices given by

(Kstiffness)mn=Dκm4δmn,(110)(\mathbf{K}_{\mathrm{stiffness}})_{mn}=D\kappa_m^4\delta_{mn}, \tag*{(110)}
(Mmass)mn=Ibδmn,(111)(\mathbf{M}_{\mathrm{mass}})_{mn}=I_b\delta_{mn}, \tag*{(111)}
(Amass)mn=−1ω2Re⁡{iωρ∫−LL⟦ϕn(x)⟧wm(x)‾ dx},(112)(\mathbf{A}_{\mathrm{mass}})_{mn}=-\frac{1}{\omega^2}\operatorname{Re}\left\{\mathrm{i}\omega\rho\int_{-L}^{L}\llbracket\phi_n(x)\rrbracket\overline{w_m(x)}\,\mathrm{d}x\right\}, \tag*{(112)}
(Bdamping)mn=−1ωIm⁡{iωρ∫−LL⟦ϕn(x)⟧wm(x)‾ dx},(113)(\mathbf{B}_{\mathrm{damping}})_{mn}=-\frac{1}{\omega}\operatorname{Im}\left\{\mathrm{i}\omega\rho\int_{-L}^{L}\llbracket\phi_n(x)\rrbracket\overline{w_m(x)}\,\mathrm{d}x\right\}, \tag*{(113)}

the vector of unknown coefficients is given by c=[c1,…,cNm]T\mathbf{c}=[c_1,\ldots,c_{N_m}]^{\mathsf{T}} and the forcing vector is given by

(f)m=−iωρw∫−LL⟦ϕdi(x)⟧wm(x)‾ dx.(114)(\mathbf{f})_m=-\mathrm{i}\omega\rho_w\int_{-L}^{L}\llbracket\phi^{\mathrm{di}}(x)\rrbracket\overline{w_m(x)}\,\mathrm{d}x. \tag*{(114)}

Reflection and transmission

The velocity potential of the hydrodynamic coupled problem is

ϕ=ϕin+ϕdi+∑ncnϕnra,(115)\phi= \phi^{\mathrm{in}} + \phi^{\mathrm{di}} + \sum_n c_n\phi_n^{\mathrm{ra}}, \tag*{(115)}
=ϕin+ϕsc,(116)= \phi^{\mathrm{in}} + \phi^{\mathrm{sc}}, \tag*{(116)}

where

ϕsc(x,z)=−∫−LL∂zG(x,z;x′,−h)⟦ϕ(x′)⟧dx′.(117)\phi^{\mathrm{sc}}(x,z) = -\int_{-L}^{L}\partial_zG(x,z;x',-h)\llbracket\phi(x')\rrbracket\mathrm{d}x'. \tag*{(117)}

Asymptotically, we have

ϕ(x,z)∼{−igAωcosh⁡kHcosh⁡(k(z+H))(eikx+Re−ikx)as x→−∞,−igAωcosh⁡kHcosh⁡(k(z+H))Teikxas x→∞,(118)\phi(x,z) \sim \begin{cases} -\dfrac{\mathrm{i}gA}{\omega\cosh kH}\cosh(k(z+H))(e^{\mathrm{i}kx}+Re^{-\mathrm{i}kx}) & \text{as } x\to-\infty,\\ -\dfrac{\mathrm{i}gA}{\omega\cosh kH}\cosh(k(z+H))Te^{\mathrm{i}kx} & \text{as } x\to\infty, \end{cases} \tag*{(118)}

where RR and TT are the reflection and transmission coefficients, respectively. The asymptotic behavior of the Green’s function is given by [44]

∂zG(x,z;x′,−h)∼−iH(1+sinh⁡(2kH)/2kH)cosh⁡(k(z+H))sinh⁡(k(H−h))eik∣x−x′∣,(119)\partial_zG(x,z;x',-h)\sim\frac{-\mathrm{i}}{H(1+\sinh(2kH)/2kH)}\cosh(k(z+H))\sinh(k(H-h))e^{\mathrm{i}k|x-x'|}, \tag*{(119)}

as ∣x−x′∣→∞|x-x'|\to\infty. Equating Equation (117) with Equation (118) using the above leads to

R=−ωcosh⁡kHAgH(1+sinh⁡(2kH)/2kH)sinh⁡(k(H−h))∫−LLeikx′⟦ϕ(x′)⟧dx′,(120)R=-\frac{\omega\cosh kH}{AgH(1+\sinh(2kH)/2kH)}\sinh(k(H-h))\int_{-L}^{L}e^{\mathrm{i}kx'}\llbracket\phi(x')\rrbracket\mathrm{d}x', \tag*{(120)}
T=1−ωcosh⁡kHAgH(1+sinh⁡(2kH)/2kH)sinh⁡(k(H−h))∫−LLe−ikx′⟦ϕ(x′)⟧dx′.(121)T=1-\frac{\omega\cosh kH}{AgH(1+\sinh(2kH)/2kH)}\sinh(k(H-h))\int_{-L}^{L}e^{-\mathrm{i}kx'}\llbracket\phi(x')\rrbracket\mathrm{d}x'. \tag*{(121)}

The proportion of energy absorbed can be calculated using the reflection and transmission coefficients as

1−∣R∣2−∣T∣2.(122)1-|R|^2-|T|^2. \tag*{(122)}

Calculation of energy absorption efficiency

We need to consider a realistic wave spectrum to talk about efficiency for realistic conditions. We assume here the Pierson–Moskowitz spectrum although other spectra could be considered. The energy absorption efficiency is

Eeff=EpE=∫0∞S(ω)(1−∣R∣2−∣T∣2) dω∫0∞S(ω) dω,(123)E_{\mathrm{eff}}=\frac{E_p}{E}=\frac{\int_0^\infty S(\omega)(1-|R|^2-|T|^2)\,\mathrm{d}\omega}{\int_0^\infty S(\omega)\,\mathrm{d}\omega}, \tag*{(123)}

where S(ω)S(\omega) is given by the Pierson–Moskowitz spectrum [47, 48]

S(ω)=ag2ω−5exp⁡(−54(2πTpω)4)(124)S(\omega)=ag^2\omega^{-5}\exp\left(-\frac{5}{4}\left(\frac{2\pi}{T_p\omega}\right)^4\right) \tag*{(124)}

with a=8.1×10−3a=8.1\times10^{-3}. This spectrum describes the distribution of energy as a function of frequency for a wave with peak period TpT_p, and EpE_p is the energy absorbed by the PWEC for the given peak period. To approximate Equation (123) we truncate the integrals so that they are between the minimum and maximum frequencies considered in the results. Note that Equation (123) can be calculated using either the transmission and reflection coefficients (Eqn. (122)) or the nearfield power (see Appendix § 9).

Validation using conservation of energy

We check that the far-field power takeoff given by hydrodynamics for both the submerged plate and floating plate matches the near-field power takeoff that can be calculated using the equations for piezoelectricity [13]. The far-field power takeoff is given by

Pfarfield=12ρwgA2Cg(1−∣R∣2−∣T∣2),(125)P_{\mathrm{farfield}}=\frac{1}{2}\rho_w g A^2 C_g\left(1-\lvert R\rvert^2-\lvert T\rvert^2\right), \tag*{(125)}

where the group velocity CgC_g is given by

Cg=ω2k(1+2kHsinh⁡(2kH)).(126)C_g=\frac{\omega}{2k}\left(1+\frac{2kH}{\sinh(2kH)}\right). \tag*{(126)}

The near-field power takeoff is given by

Pnearfield=G2∫−LL∣v∣2 dx=ω2η2G2(G2+ω2C2)∫−LL∣∂x2w∣2 dx.(127)P_{\mathrm{nearfield}}=\frac{G}{2}\int_{-L}^{L}\lvert v\rvert^2\,\mathrm{d}x=\frac{\omega^2\eta^2G}{2(G^2+\omega^2C^2)}\int_{-L}^{L}\lvert\partial^2_x w\rvert^2\,\mathrm{d}x. \tag*{(127)}

In our numerical experiments, we found that these two expressions for the power takeoff generally matched to a relative precision of 10−1310^{-13}. Appendix D shows the validation for a simply supported bimorph made of PZT-5H.

Validation using other numerical methods

Our method has been validated using both boundary element code [16] and eigenfunction matching code [11]. A comparison with the solution by eigenfunction matching is given in Appendix C for a purely elastic plate. This also serves as a benchmark calculation.

Numerical studies

Parameter values and single-frequency results

In our numerical experiments, we take the depth of the fluid to be H=10H=10 m, the total length of the plate to be 2L=202L=20 m, and the depth of submergence to be h=2h=2 m for the submerged plate. For the elastic substrate, we use the material values for silicone rubber with a Young’s modulus of E0=3.2×106E_0=3.2\times10^6 Pa, a Poisson’s ratio of ν0=0.49\nu_0=0.49, and a density of ρ0=1250\rho_0=1250 kg m−3^{-3} [49]. Under plane stress, the resulting stiffness of the elastic substrate is then C^11=E01−ν02≈4.21×106\hat{C}_{11}=\frac{E_0}{1-\nu_0^2}\approx4.21\times10^6 Pa. We take the thickness of the substrate to be d0=0.01d_0=0.01 m. For the top and bottom piezoelectric plates, we use the material properties given in Table 1 which are based on known engineering values for PVDF and PZT-5H. We initially fix the ratio of G/C=1 s−1G/C = 1\,\mathrm{s}^{-1}, where we recall GG is the surface conductance and CC is the electrical surface capacitance. It is important to note that the surface conductance is a tunable parameter in the model related to the resistivity, and we will vary this ratio of G/CG/C in later numerical experiments. We take the thickness of the piezoelectric plates to be dp=1.1×10−4 md_p = 1.1 \times10^{-4}\,\mathrm{m}. Finally, unless otherwise stated, we take Δx=0.05 m\Delta x = 0.05\,\mathrm{m}. We note that the geometric quantities chosen in this work are the same as those used by Renzi [13] and are similar to those that appear in Zheng et al. [14]. We believe that these are representative of realistic engineering scales; however, further experimental work is needed to understand the engineering practicality.

Materialρp\rho_p (kg m−3^{-3})C^11\hat{C}_{11} (Pa)e^31\hat{e}_{31} (C m−2^{-2})κ^33ε\hat{\kappa}_{33}^{\varepsilon} (F m−1^{-1})Source
PVDF17803.54×1093.54 \times10^{9}0.1131.13×10−101.13 \times10^{-10}[50, 51]
PZT-5H75006.55×10106.55 \times10^{10}-23.21.77×10−81.77 \times10^{-8}[36]

Table 1. Piezoelectric material values for vertically poled PVDF and PZT-5H. For PVDF, we compute the above quantities using C^11=Ep/(1−νp2)\hat{C}_{11}=E_p/(1-\nu_p^2) and e^31=d31Ep/(1−νp)\hat{e}_{31}=d_{31}E_p/(1-\nu_p) where Ep=3×109E_p=3\times10^9 Pa and d31=2.3×10−12d_{31}=2.3\times10^{-12} mV−1^{-1} [50], and νp=0.392\nu_p=0.392 [51]. For PZT-5H, we use the expressions in Appendix A for θ=0\theta=0.

Before considering the power conversion across a range of frequencies, we briefly consider single-frequency results for the real and imaginary parts of the free-surface displacement, plate displacement, moment, pressure, voltage, and charge for the submerged simply-supported bimorph. Figure 3b and Figure 3a show these quantities for a wave period of 4 seconds for submerged bimorphs made of PZT-5H and PVDF, respectively. The results for the voltage and charge demonstrate Ohm’s law in the frequency domain (58), in particular that the charge is proportional to a 90∘90^\circ phase shift of the voltage (cf. Re⁡(v)\operatorname{Re}(v) to Im⁡(q)\operatorname{Im}(q)). In addition, the increase in the number of oscillations in the plate displacement for PVDF compared to PZT-5H demonstrates the increased stiffness of the latter material. Finally, it should be noted that the plate displacement, moment, and pressure are in-phase, while the displacement is out of phase with the voltage and charge. The latter is because these quantities are related to the displacement through higher order derivatives, while the former is because of the relationship between displacement, moment, and pressure given by Equation (35) and Equation (42).

Surface plate versus submerged plate

We now compare PVDF and PZT-5H piezoelectric bimorphs when they are submerged or on the fluid surface for a range of incident wave frequencies and with either clamped or simply supported boundary conditions. Figure 4 shows the results for the energy absorption and power take-off over wave periods of 3–9 seconds for the parameter values described above. Table 2 shows the energy absorption efficiency computed for a peak period of Tp=5 sT_p = 5\,\mathrm{s}. These results show that submerged PWECs are able to absorb much more energy than equivalent PWECs placed on the ocean surface. This is likely caused by the lack of hydrostatic restoring force for the submerged plate, which means that the wave length of the oscillation of the plate is decreased compared to that of the incident wave; hence the second derivative of the out-of-plane displacement is greater than if the plate was on the surface and following the waves. The shorter wave length of the plate’s oscillation can be seen in the displacement results in Figure 3.

Image not extracted

(a)

Image not extracted

(b)

Figure 3. Single-frequency real and imaginary results for the free-surface displacement, plate displacement, moment, pressure, voltage, and charge for a submerged simply-supported bimorph made of (a) PVDF and (b) PZT-5H for a wave period of 4 seconds. The blue and yellow lines correspond to the coloured y-axis labels.

Results for the energy absorption and power take-off for periods of 3–9 seconds for the piezoelectric bimorph constructed with PVDF or PZT-5H

Figure 4. Results for the energy absorption and power take-off for periods of 3–9 seconds for the piezoelectric bimorph constructed with (left) PVDF or (right) PZT-5H.

ProblemBoundary conditionEfficiencyProblemBoundary conditionEfficiency
SubmergedSimply supported0.65%SubmergedSimply supported2.52%
SubmergedClamped1.64%SubmergedClamped4.68%
SurfaceSimply supported0.039%SurfaceSimply supported1.01%
SurfaceClamped0.075%SurfaceClamped1.85%
(a) PVDF(b) PZT-5H

Table 2. Energy absorption efficiency of the bimorphs constructed with (a) PVDF and (b) PZT-5H corresponding to the results in Figure 4. The peak period is taken to be Tp=5T_p = 5 s.

Figure 4 and Table 2 also shows that clamped boundary conditions slightly increase the energy absorbed and efficiency, presumably due to the large strains that appear near the end points of the plate, although the effect is not large. Finally, as noted in other works (see Li et al. [7]), PZT-5H is able to absorb more energy than PVDF because it has larger piezoelectric coupling coefficients. This comes with the trade-off that PZT is more brittle because it is a piezoceramic. It remains to be seen whether this material can be used for practical wave energy applications.

Submerged plate depth

Having established that submerged PWECs generate more power than surface plates, we now consider varying the depth of submergence hh. In these experiments, we consider simply supported bimorphs constructed with PVDF or PZT-5H. We use the same parameters as in Sections 4.1 and 4.2 above, except for the submergence depth. It should be noted that to ensure convergence for small depth (h<0.5h < 0.5), Δx\Delta x should be reduced so that it is smaller than the depth. We take Δx=h/10\Delta x = h/10 for these cases.

Figure 5 and Table 3 shows that as the depth of submergence is decreased, the energy absorption and efficiency increase, respectively. For large depths, this is expected because the energy in the wave decreases with depth. For very small depths (e.g., h=0.05h = 0.05 m), the energy absorption is several orders of magnitude greater than the surface plate results (indicated by the dashed lines in Figure 5), however these appear in a narrow frequency band. This increase in energy absorption has also been shown by other authors [e.g., 14, 22, 23]. Mathematically, it is not possible to take this limit using linear theory. This is demonstrated by the occurrence of an increasing number of resonant peaks as the depth decreases. In practice, once the upper fluid layer becomes similar in size to the wave amplitude, the linear submergence theory breaks down. In particular, we expect that for h<1h < 1 non-linear effects will start to dominate and the results in Figure 5 for these depths may not be physically meaningful. However, only a non-linear study could quantify this difference. Experiments have shown that the thin layer of water becomes better modelled by the shallow water equations and that the global motion is well approximated by the floating plate equations (since the thin layer of fluid can be neglected) [52, 53]. This phenomenon should be investigated in future work.

Image not extracted

Figure 5. Numerical results for the energy absorption for a simply supported bimorph constructed with (left) PVDF or (right) PZT-5H at various submergence depths. We show the result for the surface case with the dashed line. All values in the legend have units m.

(a) PVDF
Submergence depth (hh)0.000.050.101.002.005.008.00
Efficiency0.039%2.91%2.18%1.14%0.651%0.142%0.017%
(b) PZT-5H
Submergence depth (hh)0.000.050.101.002.005.008.00
Efficiency1.01%12.7%11.1%5.07%2.52%0.477%0.053%

Table 3. Energy absorption efficiency for simply supported bimorphs constructed of (a) PVDF and (b) PZT-5H at various submergence depths, corresponding to the results in Figure 5. The result for the surface case is given by h=0h = 0. The peak period is taken to be Tp=5T_p = 5 s.

Bimorph surface conductance

Next, we consider the effect on energy absorption of varying the surface conductance GG. We consider a simply supported bimorph constructed with PVDF or PZT-5H and submerged at a depth of 2 m. Note that in the limits G→0G \to0 and G→∞G \to\infty, the energy absorption is zero because the imaginary part of DD in Equation (62) vanishes. Physically, these limits are equivalent to an open circuit or short circuit, respectively. Figure 6 and Table 4 shows the energy absorption and efficiency for various surface conductivities. The results demonstrate that taking G/C=1G/C = 1 s−1^{-1} (green curve in Figure 6) gives reasonable energy absorption, however the surface conductance could be tuned to increase the energy absorption for waves of certain periods. For example, for incoming waves of peak period Tp=5T_p = 5 s, Table 4 shows that G/C=2G/C = 2 s−2^{-2} and G/C=5G/C = 5 s−2^{-2} give the best energy absorption efficiency for PVDF and PZT-5H, respectively.

Image not extracted

Figure 6. Numerical results for the energy absorption of a simply supported bimorph with (left) PVDF or (right) PZT-5H for varied surface conductance GG. All values in the legend have units s−1s^{-1}.

(a) PVDF
G/CG/C0.20.51.02.05.0
Efficiency0.599%0.638%0.651%0.653%0.635%
(b) PZT-5H
G/CG/C0.20.51.02.05.0
Efficiency1.84%2.20%2.52%2.77%3.05%

Table 4. Energy absorption efficiency for simply supported bimorphs constructed of (a) PVDF and (b) PZT-5H for varied surface conductance, corresponding to the results in Figure 6. The peak period is taken to be Tp=5T_p = 5 s.

Poling angle

In this section, we vary the poling angle of the piezoelectric layers in the bimorph, and instead of the two layers being oppositely poled with poling in the vertical direction, the poling direction is rotated clockwise by angle θ\theta. We consider a simply supported bimorph constructed with PZT-5H submerged at a depth of 2 m. We do not consider PVDF in this case because the change in poling angle requires the full piezoelectric material tensors. For these experiments we fix the ratio G/C=1G/C = 1 s−1s^{-1}. It should be noted that a poling angle of 90∘90^\circ corresponds to poling in the xx-direction (i.e., parallel to the plate). In this case, the imaginary part of DD again vanishes and thus no energy is absorbed. Figure 7 and Table 5 shows the numerical results for the energy absorption and efficiency for various poling angles. The results show that while peaks appear in energy absorption spectrum, the energy absorbed typically decreases as the rotation of the poling direction increases towards 90∘90^\circ. Table 5 demonstrates that for incoming waves of peak period Tp=5T_p = 5 s, adjusting the poling angle to 60∘60^\circ yields nearly a 25% improvement in the energy absorption efficiency. This suggests that varying the poling angle may provide practical benefit to energy absorption. More generally, we expect that varying the piezoelectric material orientation could improve energy absorption for three-dimensional anisotropic plates. Adjusting the poling angle and material orientation should continue to be investigated in future studies.

Image not extracted

Figure 7. Numerical results for the energy absorption of a simply supported bimorph made of PZT-5H with varied poling angle θ\theta.

Poling angle θ\theta0∘0^\circ30∘30^\circ45∘45^\circ60∘60^\circ75∘75^\circ
Efficiency2.52%2.42%2.77%3.14%2.52%

Table 5. Energy absorption efficiency of a simply supported bimorph made of PZT-5H with varied poling angle θ\theta, corresponding to the results in Figure 7. The peak period is taken to be Tp=5T_p = 5 s.

Conclusions

In this paper we have numerically investigated wave energy absorption by piezoelectric bimorphs when submerged or placed on the surface of the ocean. In the first part of the paper we carefully derived the equations of motion in dimensional form starting from the constitutive relations for piezoelectricity. This allowed us to consider realistic piezoelectric material properties and extend the model to arbitrary poling angle for the two oppositely-poled piezoelectric layers. Our derivation verifies the non-dimensional model developed by Renzi [13] and gives additional details. Although we do not describe it here, the derivation can also be extended to three-dimensional problems.

To solve the coupled system in Equations (81) for the submerged plate we implemented a modal expansion method. The component diffraction and radiation problems were reduced to hypersingular integral equations and solved numerically using a constant panel method. A key step in this approach is the numerical integration of the derivative of the Green’s function, which is achieved using residue calculus and numerical integration over the resulting deformed contour. The method is general and can be extended to problems involving plates submerged vertically or at an angle, curved plates, and three-dimensional problems. We note that the models are limited by the assumption that the plate thickness can be neglected and that the wave amplitude and plate motion are small.

We considered several numerical experiments using the above approach. In the first numerical study, we computed the energy absorption and efficiency for a peak period of 5 s for piezoelectric bimorphs, with the piezoelectric layer consisting of PVDF or PZT-5H. We compared energy absorption and efficiency when the plate is submerged or on the surface and showed that much greater energy absorption is obtained using submerged plates, particularly for bimorphs constructed with PZT-5H. We considered clamped and simply supported boundary conditions. We found that these give qualitatively similar results with a slight increase in energy absorption for clamped plates because of the large strains that appear near the end points of the plate.

In the remaining numerical experiments, we considered varying the plate submergence depth, surface conductance, and poling angle. The results for varied plate submergence depth showed that for small depths, the energy absorption is several orders of magnitude larger than the surface plate results. We know that as the plate approaches the surface there is no stable limit, and the response becomes dominated by resonant waves on the top of the plate. Moreover, no matter how close to the surface, the hydrodynamic restoring force is missing from the equations of motion of the submerged plate. There is significant scope for further investigation of this phenomena and it is not clear whether the submerged plate works better in absorbing energy for all parameter values. The results for varying the surface conductance show that this is a tunable parameter for increasing energy absorption and efficiency at certain wave periods. Finally, we found that varying the poling angle can provide practical benefit by increasing the energy absorption efficiency, with a near 25% improvement for waves with a peak period of 5 s.

Data availability

The source code and data for this work is available at https://github.com/zjwegert/SemiAnalyticWECs.jl.

Acknowledgment

This research supported by the Australian Government through the Australian Research Council (ARC), Discovery Grant DP240102104.

Expressions for material constants

In the following, we give the expressions for C^11E(θ)\hat{C}^{E}_{11}(\theta), e^31(θ)\hat{e}_{31}(\theta) and κ^33σ(θ)\hat{\kappa}^{\sigma}_{33}(\theta) in terms of components of the piezoelectric material tensors SˉpqE\bar{S}^{E}_{pq}, dˉip\bar{d}_{ip}, and κˉijσ\bar{\kappa}^{\sigma}_{ij} in the original reference frame. To shorten the expressions, we denote s=sin⁡θs=\sin\theta and c=cos⁡θc=\cos\theta. These expressions are given by

C^11E(θ)=8Sˉ22ESˉ22ESˉ55E−s4(8[Sˉ23E]2−8Sˉ22ESˉ33E+Sˉ22ESˉ55E)+2c2s2(8Sˉ13ESˉ22E−8Sˉ12ESˉ23E+3Sˉ22ESˉ55E)−c4(8[Sˉ12E]2+Sˉ22E(−8Sˉ11E+Sˉ55E))(128)\hat{C}^{E}_{11}(\theta)=\frac{8\bar{S}^{E}_{22}}{\bar{S}^{E}_{22}\bar{S}^{E}_{55}-s^{4}\left(8[\bar{S}^{E}_{23}]^{2}-8\bar{S}^{E}_{22}\bar{S}^{E}_{33}+\bar{S}^{E}_{22}\bar{S}^{E}_{55}\right)+2c^{2}s^{2}\left(8\bar{S}^{E}_{13}\bar{S}^{E}_{22}-8\bar{S}^{E}_{12}\bar{S}^{E}_{23}+3\bar{S}^{E}_{22}\bar{S}^{E}_{55}\right)-c^{4}\left(8[\bar{S}^{E}_{12}]^{2}+\bar{S}^{E}_{22}(-8\bar{S}^{E}_{11}+\bar{S}^{E}_{55})\right)} \tag*{(128)}
e^31(θ)=8(c3(dˉ32Sˉ12E−dˉ31Sˉ22E)+cs2(dˉ15Sˉ22E−dˉ33Sˉ22E+dˉ32Sˉ23E))−Sˉ22ESˉ55E+s4(8[Sˉ23E]2−8Sˉ22ESˉ33E+Sˉ22ESˉ55E)−2c2s2(8Sˉ13ESˉ22E−8Sˉ12ESˉ23E+3Sˉ22ESˉ55E)+c4(8[Sˉ12E]2+Sˉ22E(−8Sˉ11E+Sˉ55E))(129)\hat{e}_{31}(\theta)=\frac{8\left(c^{3}(\bar{d}_{32}\bar{S}^{E}_{12}-\bar{d}_{31}\bar{S}^{E}_{22})+cs^{2}(\bar{d}_{15}\bar{S}^{E}_{22}-\bar{d}_{33}\bar{S}^{E}_{22}+\bar{d}_{32}\bar{S}^{E}_{23})\right)}{-\bar{S}^{E}_{22}\bar{S}^{E}_{55}+s^{4}\left(8[\bar{S}^{E}_{23}]^{2}-8\bar{S}^{E}_{22}\bar{S}^{E}_{33}+\bar{S}^{E}_{22}\bar{S}^{E}_{55}\right)-2c^{2}s^{2}\left(8\bar{S}^{E}_{13}\bar{S}^{E}_{22}-8\bar{S}^{E}_{12}\bar{S}^{E}_{23}+3\bar{S}^{E}_{22}\bar{S}^{E}_{55}\right)+c^{4}\left(8[\bar{S}^{E}_{12}]^{2}+\bar{S}^{E}_{22}(-8\bar{S}^{E}_{11}+\bar{S}^{E}_{55})\right)} \tag*{(129)}
κ^33σ(θ)=1Sˉ22E−c2[dˉ32]2+c2κˉ33σSˉ22E+κˉ11σs2Sˉ22E+{8(c3(dˉ32Sˉ12E−dˉ31Sˉ22E)+cs2(dˉ15Sˉ22E−dˉ33Sˉ22E+dˉ32Sˉ23E))}2−Sˉ22ESˉ55E+s4(8[Sˉ23E]2−8Sˉ22ESˉ33E+Sˉ22ESˉ55E)−2c2s2(8Sˉ13ESˉ22E−8Sˉ12ESˉ23E+3Sˉ22ESˉ55E)+c4(8[Sˉ12E]2+Sˉ22E(−8Sˉ11E+Sˉ55E))(130)\hat{\kappa}^{\sigma}_{33}(\theta)=\frac{1}{\bar{S}^{E}_{22}}\frac{-c^{2}[\bar{d}_{32}]^{2}+c^{2}\bar{\kappa}^{\sigma}_{33}\bar{S}^{E}_{22}+\bar{\kappa}^{\sigma}_{11}s^{2}\bar{S}^{E}_{22}+\left\{8\left(c^{3}(\bar{d}_{32}\bar{S}^{E}_{12}-\bar{d}_{31}\bar{S}^{E}_{22})+cs^{2}(\bar{d}_{15}\bar{S}^{E}_{22}-\bar{d}_{33}\bar{S}^{E}_{22}+\bar{d}_{32}\bar{S}^{E}_{23})\right)\right\}^{2}}{-\bar{S}^{E}_{22}\bar{S}^{E}_{55}+s^{4}\left(8[\bar{S}^{E}_{23}]^{2}-8\bar{S}^{E}_{22}\bar{S}^{E}_{33}+\bar{S}^{E}_{22}\bar{S}^{E}_{55}\right)-2c^{2}s^{2}\left(8\bar{S}^{E}_{13}\bar{S}^{E}_{22}-8\bar{S}^{E}_{12}\bar{S}^{E}_{23}+3\bar{S}^{E}_{22}\bar{S}^{E}_{55}\right)+c^{4}\left(8[\bar{S}^{E}_{12}]^{2}+\bar{S}^{E}_{22}(-8\bar{S}^{E}_{11}+\bar{S}^{E}_{55})\right)} \tag*{(130)}

For PZT-5H, the piezoelectric material tensors in the reference frame with poling in the zz-direction are [36]

SˉPZT5HE=(16.7−4.84−8.50.00.00.0−4.8416.7−8.50.00.00.0−8.5−8.520.80.00.00.00.00.00.043.50.00.00.00.00.00.043.50.00.00.00.00.00.042.9)×10−12 m2N−1,(131)\bar{S}^{E}_{\mathrm{PZT5H}} = \begin{pmatrix} 16.7 & -4.84 & -8.5 & 0.0 & 0.0 & 0.0 \\ -4.84 & 16.7 & -8.5 & 0.0 & 0.0 & 0.0 \\ -8.5 & -8.5 & 20.8 & 0.0 & 0.0 & 0.0 \\ 0.0 & 0.0 & 0.0 & 43.5 & 0.0 & 0.0 \\ 0.0 & 0.0 & 0.0 & 0.0 & 43.5 & 0.0 \\ 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 42.9 \end{pmatrix} \times10^{-12}\ \mathrm{m}^{2}\mathrm{N}^{-1}, \tag*{(131)}
dˉPZT5H=(0.00.00.00.07.390.00.00.00.07.390.00.0−2.75−2.755.940.00.00.0)×10−10 mV−1,(132)\bar{d}_{\mathrm{PZT5H}} = \begin{pmatrix} 0.0 & 0.0 & 0.0 & 0.0 & 7.39 & 0.0 \\ 0.0 & 0.0 & 0.0 & 7.39 & 0.0 & 0.0 \\ -2.75 & -2.75 & 5.94 & 0.0 & 0.0 & 0.0 \end{pmatrix} \times10^{-10}\ \mathrm{m}\mathrm{V}^{-1}, \tag*{(132)}
κˉPZT5Hσ=(2.760002.760003.04)×10−8 Fm−1.(133)\bar{\kappa}^{\sigma}_{\mathrm{PZT5H}} = \begin{pmatrix} 2.76 & 0 & 0 \\ 0 & 2.76 & 0 \\ 0 & 0 & 3.04 \end{pmatrix} \times10^{-8}\ \mathrm{F}\mathrm{m}^{-1}. \tag*{(133)}

Calculation of the Green's function

The Green’s function in Equation (87) requires the evaluation of the integral

I(X)=−2∫0∞cos⁡(μX)F(μ) dμ,(134)I(X) = -2 \int_{0}^{\infty} \cos(\mu X)F(\mu)\,\mathrm{d}\mu, \tag*{(134)}

with

F(μ)=1cosh⁡μH[cosh⁡μ(z+H)cosh⁡μ(z′+H)μsinh⁡μH−αcosh⁡μH+e−μHμsinh⁡μzsinh⁡μz′].(135)F(\mu) = \frac{1}{\cosh\mu H}\left[\frac{\cosh\mu(z+H)\cosh\mu(z^{\prime}+H)}{\mu\sinh\mu H-\alpha\cosh\mu H}+\frac{\mathrm{e}^{-\mu H}}{\mu}\sinh\mu z\sinh\mu z^{\prime}\right]. \tag*{(135)}

This is not suitable for numerical calculation due to the pole at μ=k\mu=k, where we recall kk satisfies ksinh⁡kH−αcosh⁡kH=0k\sinh kH-\alpha\cosh kH=0. We first write Equation (B.1) as I(X)=I+(X)+I−(X)I(X)=I_{+}(X)+I_{-}(X) where

I±(X)=−∫0∞e±iμXF(μ) dμ.(136)I_{\pm}(X)=-\int_{0}^{\infty}\mathrm{e}^{\pm i\mu X}F(\mu)\,\mathrm{d}\mu. \tag*{(136)}

We assume that X>0X>0, noting that the case X<0X<0 can be recovered from the relationship I(−X)=I(X)I(-X)=I(X). The assumption X>0X>0 allows us to deform the contour of integration for I±(X)I_{\pm}(X) to be over C±=CR±∪C2±C^{\pm}=C_{R}^{\pm}\cup C_{2}^{\pm} as R→∞R\to\infty. We parameterise CR±C_{R}^{\pm} as γ(R)=Re±πi/4\gamma(R)=R\mathrm{e}^{\pm\pi i/4} for R∈[0,∞)R\in[0,\infty). Figure B.8 shows a visualisation of this. When closing the contour for I+(X)I_{+}(X) in the upper-half plane, we pick up a contribution from the pole at μ=k\mu=k. It can be shown that as R→∞R\to\infty, the contributions from CR±C_{R}^{\pm} to I±(X)I_{\pm}(X) vanish. As R→∞R\to\infty, we are left with

I+(X)=−∫C2+eiμXF(μ) dμ−2πiRes⁡μ=k{eiμXF(μ)},(137)I_{+}(X)=-\int_{C_{2}^{+}}\mathrm{e}^{i\mu X}F(\mu)\,\mathrm{d}\mu-2\pi i\underset{\mu=k}{\operatorname{Res}}\left\{\mathrm{e}^{i\mu X}F(\mu)\right\}, \tag*{(137)}

and

I−(X)=−∫C2−e−iμXF(μ) dμ.(138)I_{-}(X)=-\int_{C_{2}^{-}}\mathrm{e}^{-i\mu X}F(\mu)\,\mathrm{d}\mu. \tag*{(138)}

Figure B.8. Deformed integration contour

where the residue is given by [44]

2πiRes⁡μ=k{eiμXF(μ)}=πikH(1+sinh⁡(2kH)/2kH)cosh⁡(k(z+H))cosh⁡(k(z′+H))eikX.(139)2\pi i \operatorname{Res}_{\mu=k}\left\{e^{i\mu X}F(\mu)\right\}=\frac{\pi i}{kH\left(1+\sinh(2kH)/2kH\right)}\cosh\left(k(z+H)\right)\cosh\left(k(z^{\prime}+H)\right)e^{ikX}. \tag*{(139)}

Finally, we observe that the integrals appearing in the expressions for I+(X)I_{+}(X) and I−(X)I_{-}(X) are complex conjugates, so we only require one to be calculated and this can be done with numerical quadrature. This completes the calculation of the Green’s function.

Validation for the case of a perfectly elastic plate

We present here a validation of the hypersingular boundary integral equation method for the case of a submerged perfectly elastic plate. This provides confidence that the code is correct and also provides a benchmark solution that may help with the development of other codes in the future. The presented method is compared with the solution by eigenfunction matching developed by Hassan et al. [11]. For this calculation we take the water depth as H=2.5H=2.5 m, the depth of submergence as h=0.25h=0.25 m, the water density as 10001000 kg m−3^{-3}, the flexural rigidity as D=104D=10^{4} N m2^{2}, and the mass moment of inertia to be Ib=10I_b=10 kg m−2^{-2}. Figure C.9 shows the results of this comparison, where we plot the absolute value of the reflection and transmission coefficients for wave periods between 3 and 7 seconds.

Validation of energy conservation

We present here a validation of energy conservation for the case of a simply supported bimorph made of PZT-5H. The parameter values used are as in § 4.1. Figure D.10 shows the proportion of energy absorbed calculated using either: i) the transmission and reflection coefficients (Eqn. (122)); or ii) the nearfield power (Eqn. (127)) divided by the wave power, Pwave=12ρwgA2CgP_{\mathrm{wave}}=\frac{1}{2}\rho_{w}gA^{2}C_{g}.

Figure C.9. Validation of the reflection and transmission coefficients resulting from the hypersingular boundary integral equation method (solid line) presented in this paper and eigenfunction matching (stars) using the method in [11]

Proportion of energy absorbed by a simply supported bimorph

Figure D.10. Proportion of energy absorbed by a simply supported bimorph made of PZT-5H computed from the transmission and reflection coefficients (Eqn. (122)) or the nearfield power (Eqn. (127)) divided by the wave power. The power in the wave is given by PwaveP_{\mathrm{wave}}.

References

  1. [1]A. F. de O. Falcão, Wave energy utilization: A review of the technologies, Renewable and Sustainable Energy Reviews 14 (2020) 899–918.
  2. [2]J. Falnes, A review of wave-energy extraction, Marine Structures 20 (2007) 185–201.DOI
  3. [3]E. Renzi, S. Michele, S. Zheng, S. Jin, D. Greaves, Niche applications and flexible devices for wave energy conversion: A review, Energies 14 (2021) 6537.DOI
  4. [4]M. R. Kiran, O. Farrok, M. Abdullah-Al-Mamun, M. R. Islam, W. Xu, Progress in piezoelectric material based oceanic wave energy conversion technology, IEEE Access 8 (2020) 146428–146449.
  5. [5]S. M. Kargar, G. Hao, An atlas of piezoelectric energy harvesters in oceanic applications, Sensors 22 (2022) 1949.DOI
  6. [6]R. Liu, L. He, X. Liu, S. Wang, L. Zhang, G. Cheng, A review of collecting ocean wave energy based on piezoelectric energy harvester, Sustainable Energy Technologies and Assessments 59 (2023) 103417.DOI
  7. [7]Z. Li, J. Roscow, H. Khanbareh, G. Haswell, C. Bowen, Energy harvesting from water flow by using piezoelectric materials, Adv. Energy Sustainability Res. 5 (2024) 2300235.
  8. [8]S. Wang, L. He, H. Wang, X. Li, B. Sun, J. Lin, Energy harvesting from water impact using piezoelectric energy harvester, Rev. Sci. Instrum. 95 (2024) 021501.DOI
  9. [9]Y. Xiao, Q. Han, N. Wu, Piezoelectric energy harvesting: a review of energy sources, structures, and working mechanisms in high-frequency excitations and operations, Smart Mater. Struct. 34 (2025) 023001.DOI
  10. [10]M. H. Meylan, A flexible vertical sheet in waves, Int. J. Offshore Polar 5 (1995) 105–110.
  11. [11]M. Hassan, M. H. Meylan, M. A. Peter, Water-wave scattering by submerged elastic plates, Q. J. Mech. Appl. Math. 62 (2009) 321–344.DOI
  12. [12]T. W. Williams, M. H. Meylan, The Wiener-Hopf and residue calculus soltions for a submerged semi-infinite elastic plate, J. Engng. Maths. 71 (2012) 1–26.DOI
  13. [13]E. Renzi, Hydroelectromechanical modelling of a piezoelectric wave energy converter, Proc. Roy. Soc. A: 472 (2016) 20160715.DOI
  14. [14]S. Zheng, M. Meylan, X. Zhang, G. Iglesias, D. Greaves, Performance of a plate-wave energy converter integrated in a floating breakwater, IET Renew. Power Gener. 15 (2021) 3206–3219.DOI
  15. [15]S. Zheng, M. H. Meylan, X. Zhang, G. Iglesias, D. Greaves, Wave power extraction from a piezoelectric wave energy converter integrated in a pile-supported breakwater, Proceedings of the 14th European Wave and Tidal Energy Conference 5-9th September 2021 (2021).
  16. [16]V. Vipin, S. Koley, Mathematical modeling of a submerged piezoelectric wave energy converter device installed over an undulated seabed, Renewable Energy 200 (2022) 1382–1392.DOI
  17. [17]V. Vipin, K. Trivedi, S. Koley, Performance of a submerged piezoelectric wave energy converter device floating over an undulated seabed, Energy Reports 8 (2022) 182–188.DOI
  18. [18]K. Trivedi, S. Koley, Performance of a hybrid wave energy converter device consisting of a piezoelectric plate and oscillating water column device placed over an undulated seabed, Applied Energy 333 (2023) 120627.DOI
  19. [19]V. Vipin, S. Koley, Power generation prediction and optimization of parameters of piezo-electric wave energy converter device, Fourth International Conference on Advances in Electrical, Computing, Communication and Sustainable Technologies (2024).
  20. [20]K. Trivedi, K. C. Swami, S. Koley, B. Wilks, M. H. Meylan, Wave power extraction by dual horizontal piezoelectric plate wave energy converters placed over undulated seabed, Results in Engineering (2025). URL: https://doi.org/10.1016/j.rineng.2025.108713.
  21. [21]K. Trivedi, S. Koley, Optimization of the Piezoelectric Wave Energy Converter Device, Springer, Singapore, 2023, pp. 463–474.DOI
  22. [22]K. Shoele, Hybrid wave/current energy harvesting with a flexible piezoelectric plate, J. Fluid Mech. 968 (2023) A31.arxiv.org/abs/2201.08751
  23. [23]V. Vipin, S. Koley, Design optimization of a submerged piezoelectric wave energy converter device using an artificial neural network model, Energy Reports 9 (2023) 322–326.DOI
  24. [24]Y. Shen, J. Wang, J. Wang, H. Zheng, K. Lin, H. Xu, Effect of length and attack angle of the splitter plates on circular cylinder piezoelectric water energy harvester, Phys. Fluids 36 (2024) 045131.DOI
  25. [25]I. n. Aramendia, A. Saenz-Aguirre, A. Boyano, U. Fernandez-Gamiz, E. Zulueta, Oscillating U-shaped body for underwater piezoelectric energy harvester power optimization, Micromachines 10 (2019) 737.DOI
  26. [26]A.-R. Westcott, L. G. Bennetts, N. Y. Sergiienko, B. S. Cazzolato, Broadband near-perfect capture of water wave energy by an array of heaving buoy wave energy converters, Journal of Fluid Mechanics 998 (2024) A5. doi:10.1017/jfm.2024.819.DOI
  27. [27]M. H. Meylan, V. J. Challis, N. Thamwattana, Z. J. Wegert, B. Wilks, Wave power absorption by floating plates with application to piezoelectric or other bending-based absorption, Ocean Engineering 328 (2025) 120931.DOI
  28. [28]B. Wilks, M. H. Meylan, Z. J. Wegert, V. J. Challis, N. Thamwattana, Water wave scattering by a surface-mounted rectangular anisotropic elastic plate, 2025. URL: https://arxiv.org/abs/2510.17872. arXiv:2510.17872.
  29. [29]S. Li, S. Sun, J. Cui, Fluid–structure interaction analysis of a piezoelectric wave energy converter positioned at the wave surface, Phys. Fluids 37 (2025) 013121.DOI
  30. [30]S.-E. Chen, R.-Y. Yang, Z.-H. Qiu, C.-C. Wu, A piezoelectric wave energy harvester using plucking-driven and frequency up-conversion mechanism, Energies 14 (2021) 8441.DOI
  31. [31]S.-E. Chen, R.-Y. Yang, G.-K. Wu, C.-C. Wu, A piezoelectric wave-energy converter equipped with a geared-linkage-based frequency up-conversion mechanism, Sensors 21 (2021) 204.DOI
  32. [32]W. Cai, V. Roussinova, V. Stoilo, Piezoelectric wave energy harvester, Renewable Energy 196 (2022) 973–982.DOI
  33. [33]R. Liu, C. Pan, H. Zhou, H. Xia, L. He, Enhancement ultra-low-frequency wave energy harvesting through a piezoelectric energy harvester based on c-shaped cantilever beams, Renewable Energy 256 (2026) 124220.DOI
  34. [34]S. Zheng, M. Meylan, D. Greaves, G. Iglesias, Water-wave interaction with submerged porous elastic disks, Physics of Fluids 32 (2020).DOI
  35. [35]S. Michele, S. Zheng, D. Greaves, Wave energy extraction from a floating flexible circular plate, Ocean Engineering 245 (2022) 110275.DOI
  36. [36]J. Yang, An Introduction to the Theory of Piezoelectricity, Advances in Mechanics and Mathematics, 9, 2nd ed. 2018. ed., Springer International Publishing, Cham, 2018.
  37. [37]A. Erturk, D. J. Inman, Piezoelectric energy harvesting, Wiley, 2011.
  38. [38]J. N. Reddy, Theory and Analysis of Elastic Plates and Shells, 2nd edition ed., CRC Press, 2006.DOI
  39. [39]H. Liu, J. Zhong, C. Lee, S.-W. Lee, L. Lin, A comprehensive review on piezoelectric energy harvesting technology: Materials, mechanisms, and applications, Applied Physics Reviews 5 (2018) 041306.DOI
  40. [40]Q.-M. Wang, X.-h. Du, B. Xu, L. E. Cross, Theoretical analysis of the sensor effect of cantilever piezoelectric benders, Journal of Applied Physics 85 (1999) 1702–1712.DOI
  41. [41]A. Erturk, D. J. Inman, John Wiley & Sons, Ltd, 2011, p. 49–96.
  42. [42]A. Erturk, D. J. Inman, A distributed parameter electromechanical model for cantilevered piezoelectric energy harvesters, Journal of Vibration and Acoustics 130 (2008) 041002.DOI
  43. [44]C. M. Linton, P. McIver, Handbook of mathematical techniques for wave/structure interactions, Chapman & Hall, 2001.DOI
  44. [45]J. N. Newman, Wave effects on deformable bodies, Appl. Ocean Res. 16 (1994) 47–59.DOI
  45. [46]N. Parsons, P. Martin, Scattering of water waves by submerged plates using hypersingular integral equations, Applied Ocean Research 14 (1992) 313–321.DOI
  46. [47]W. J. Pierson Jr, L. Moskowitz, A proposed spectral form for fully developed wind seas based on the similarity theory of sa kitaigorodskii, Journal of Geophysical Research 69 (1964) 5181–5190.
  47. [48]R. G. Dean, R. A. Dalrymple, Water wave mechanics for engineers and scientists, volume 2, world scientific publishing company, 1991.
  48. [49]Y. Tanaka, T. Oko, H. Mutsuda, R. Patel, S. McWilliam, A. A. Popov, An experimental study of wave power generation using a flexible piezoelectric device, Journal of Ocean and Wind Energy 2 (2015) 28–36.
  49. [50]M. S. Inc., Piezo Film Sensors Technical Manual, Hampton, VA, 2008.
  50. [51]H. S. Nalwa, Ferroelectric, Pyroelectric, and Piezoelectric Properties of Poly(vinylidene Fluoride), CRC Press, 1995, p. 183–232.DOI
  51. [52]M. Meylan, L. Bennetts, C. Cavaliere, A. Alberello, A. Toffoli, Experimental and theoretical models of wave-induced flexure of a sea ice floe, Physics of Fluids 27 (2015).DOI
  52. [53]D. Skene, L. Bennetts, M. Meylan, A. Toffoli, Modelling water wave overwash of a thin floating plate, Journal of Fluid Mechanics 777 (2015) R3.DOI

Paper details

vlnc:W_ftpf3km6
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents