Abstract
We present two accurate and efficient algorithms for solving the incompressible, irrotational Euler equations with a free surface in two dimensions with background flow over a periodic, multiply connected fluid domain that includes stationary obstacles and variable bottom topography. One approach is formulated in terms of the surface velocity potential while the other evolves the vortex sheet strength. Both methods employ layer potentials in the form of periodized Cauchy integrals to compute the normal velocity of the free surface, are compatible with arbitrary parameterizations of the free surface and boundaries, and allow for circulation around each obstacle, which leads to multiple-valued velocity potentials but single-valued stream functions. We prove that the resulting second-kind Fredholm integral equations are invertible, possibly after a physically motivated finite-rank correction. In an angle-arclength setting, we show how to avoid curve reconstruction errors that are incompatible with spatial periodicity. We use the proposed methods to study gravity-capillary waves generated by flow around several elliptical obstacles above a flat or variable bottom boundary. In each case, the free surface eventually self-intersects in a splash singularity or collides with a boundary. We also show how to evaluate the velocity and pressure with spectral accuracy throughout the fluid, including near the free surface and solid boundaries. To assess the accuracy of the time evolution, we monitor energy conservation and the decay of Fourier modes and compare the numerical results of the two methods to each other. We implement several solvers for the discretized linear systems and compare their performance. The fastest approach employs a graphics processing unit (GPU) to construct the matrices and carry out iterations of the generalized minimal residual method (GMRES).
Article PDF
Similar content being viewed by others
Avoid common mistakes on your manuscript.
References
Ablowitz, M., Fokas, A.: Complex Variables: Introduction and Applications. Cambridge Texts in Applied Mathematics. Cambridge University Press, Cambridge (1997)
Akers, B., Ambrose, D., Wright, J.: Traveling waves from the arclength parameterization: vortex sheets with surface tension. Interfaces Free Bound. 15(3), 359–380 (2013)
Akers, B.F., Ambrose, D.M., Wright, J.D.: Gravity perturbed crapper waves. Proc. R. Soc. A 470(2161), 20130526 (2014)
Alben, S.: Wake-mediated synchronization and drafting in coupled flags. J. Wake-Mediated Fluid Mech. 641, 489–496 (2009)
Alben, S., Shelley, M.: Coherent locomotion as an attracting state for a free flapping body. Proc. Natl. Acad Sci. 102(32), 11163–11166 (2005)
Ambrose, D.: Well-posedness of vortex sheets with surface tension. SIAM J. Math. Anal. 35(1), 211–244 (2003)
Ambrose, D., Masmoudi, N.: The zero surface tension limit of two-dimensional water waves. Comm. Pure Appl. Math. 58(10), 1287–1315 (2005)
Ambrose, D., Masmoudi, N.: Well-posedness of 3d vortex sheets with surface tension. Commun. Math Sci. 5, 391–430 (2007)
Ambrose, D., Wilkening, J.: Computation of symmetric, time-periodic solutions of the vortex sheet with surface tension. Proc. Natl. Acad. Sci. 107(8), 3361–3366 (2010)
Ambrose, D., Siegel, M., Tlupova, S.: A small-scale decomposition for 3D, boundary integral computations with surface tension. J. Comput. Phys. 247, 168–191 (2013)
Andrade, D., Nachbin, A.: A three-dimensional Dirichlet-to-Neumann operator for water waves over topography. J. Fluid Mech. 845, 321–345 (2018)
Ascher, U., Ruuth, S., Wetton, B.: Implicit-explicit methods for time-dependent partial differential equations. SIAM J. Numer. Anal. 32 (3), 797–823 (1995)
Baker, G., Nachbin, A.: Stable methods for vortex sheet motion in the presence of surface tension. SIAM J. Sci. Comput. 19(5), 1737–1766 (1998)
Baker, G., Meiron, D., Orszag, S.: Generalized vortex methods for free-surface flow problems. J. Fluid Mech. 123, 477–501 (1982)
Baker, A. H., Dennis, J. M., Jessup, E. R.: On improving linear solver performance: a block variant of GMRES. SIAM J. Sci. Comput. 27(5), 1608–1626 (2006)
Baker, A., Jessup, E., Kolev, T.: A simple strategy for varying the restart parameter in GMRES(m). J. Comput. Appl. Math. 230, 751–761 (2009)
Caflisch, R., Li, X. -F.: Lagrangian theory for 3D vortex sheets with axial or helical symmetry. Transp. Theory Stat. Phys. 21(4–6), 559–578 (1992)
Camassa, R., Wu, T. -T.: Stability of some stationary solutions for the forced kdv equation. Phys. D: Nonlinear Phenom. 51(1-3), 295–307 (1991)
Castro, A., Córdoba, D., Fefferman, C. L., Gancedo, F., Gómez-Serrano, J.: Splash singularity for water waves. Proc. Natl. Acad. Sci. 109(3), 733–738 (2012)
Castro, A., Córboda, D., Fefferman, C., Gancedo, F., Gómez-Serrano, J.: Finite time singularities for the free boundary incompressible euler equations. Ann. Math. 1061–1134 (2013)
Ceniceros, H., Hou, T.: An efficient dynamically adaptive mesh for potentially singular solutions. J. Comput. Phys. 172(2), 609–639 (2001)
Choi, W., Camassa, R.: Exact evolution equations for surface waves. J. Eng. Mech. 125(7), 756–760 (1999)
Chorin, A. J., Marsden, J. E.: A Mathematical Introduction to Fluid Mechanics, 3rd edn. Springer, New York (1993)
Colton, D., Kress, R.: Inverse Acoustic and Electromagnetic Scattering Theory, vol. 93. Springer Nature (2019)
Cox, S., Matthews, P.: Exponential time differencing for stiff systems. J. Comput. Phys. 176(2), 430–455 (2002)
Craig, W., Sulem, C.: Numerical simulation of gravity waves. J. Comput. Phys. 108, 73–83 (1993)
Crapper, D. G.: An exact solution for progressive capillary waves of arbitrary amplitude. J. Fluid Mech. 2, 532–540 (1957)
Crowdy, D.: Analytical solutions for uniform potential flow past multiple cylinders. Eur. J. Mech.-B/Fluids 25(4), 459–470 (2006)
Dagan, G., Tulin, M.: Two-dimensional free-surface gravity flow past blunt bodies. J. Fluid Mech. 51(3), 529–543 (1972)
Dyachenko, A. I., Kuznetsov, E. A., Spector, M., Zakharov, V. E.: Analytical description of the free surface dynamics of an ideal fluid (canonical formalism and conformal mapping). Phys. Lett. A 221(1–2), 73–79 (1996)
El, G., Grimshaw, R., Smyth, N.: Unsteady undular bores in fully nonlinear shallow-water theory. Phys. Fluids 18(2), 027104 (2006)
El, G., Grimshaw, R., Smyth, N.: Transcritical shallow-water flow past topography: finite-amplitude theory. J. Fluid Mech. 640, 187–214 (2009)
Folland, G.: Introduction to Partial Differential Equations, 2nd edn. Princeton University Press, Princeton (1995)
Forbes, L.: Free-surface flow over a semicircular obstruction, including the influence of gravity and surface tension. J. Fluid Mech. 127, 283–297 (1983)
Froehle, B., Persson, P.-O.: A high-order discontinuous galerkin method for fluid-structure interaction with efficient implicit-explicit time stepping. J. Comput. Phys. 272, 455–470 (2014)
Golub, G. H., Loan, C. F. V.: Matrix Computations. Johns Hopkins University Press, Baltimore (1996)
Grandison, S., Vanden-Broeck, J. -M.: Truncation approximations for gravity-capillary free-surface flows. J. Eng. Math. 54(1), 89 (2006)
Grimshaw, R., Smyth, N.: Resonant flow of a stratified fluid over topography. J. Fluid Mech. 169, 429–464 (1986)
Hairer, E., Norsett, S. P., Wanner, G.: Solving Ordinary Differential Equations I: Nonstiff Problems, 2nd edn. Springer, Berlin (2000)
Havelock, T.: The method of images in some problems of surface waves. Proc. R. Soc. Lond. Ser. A, Containing Papers of a Mathematical and Physical Character 115(771), 268–280 (1927)
Havelock, T.: The vertical force on a cylinder submerged in a uniform stream. Proc. R. Soc. Lond. Ser. A, Containing Papers of a Mathematical and Physical Character 122(790), 387–393 (1929)
Helsing, J., Ojala, R.: On the evaluation of layer potentials close to their sources. J. Comput. Phys. 227, 2899–2921 (2008)
Hirata, M., Okino, S., Hanazaki, H.: Numerical simulation of capillary gravity waves excited by an obstacle in shallow water. Proc. Estonian Acad.f Sci. 64(3), 278 (2015)
Hoogedoorn, G. J. E., Beyene, A.: Aero-elastic behavior of a flexible blade for wind turbine application: a 2D computational study. Energy 35, 778–785 (2010)
Hou, T., Li, R.: Computing nearly singular solutions using pseudo-spectral methods. J. Comput. Phys. 226(1), 379–397 (2007)
Hou, T., Lowengrub, J., Shelley, M.: Removing the stiffness from interfacial flows with surface tension. J. Comput. Phys. 114(2), 312–338 (1994)
Hou, T., Lowengrub, J., Shelley, M.: The long-time motion of vortex sheets with surface tension. Phys. Fluids 9(7), 1933–1954 (1997)
Huang, Y.: A boundary integral method for modeling axisymmetric flow around a rising bubble in a vertical tube and accurate numerical evaluation of orthogonal polynomials. Phd Thesis, University of California, Berkeley (2020)
Jackson, J.: Electrodynamics. The Optics encyclopedia: Basic Foundations and Practical Applications (2007)
Kennedy, C. A., Carpenter, M. H.: Additive Runge-Kutta schemes for convection-diffusion-reaction equations. Appl. Numer. Math. 44(1–2), 139–181 (2003)
Koga, K.: A new approach to boundary integral simulations of axisymmetric droplet dynamics. PhD thesis, Kyoto University (2020)
Lamb, H.: Hydrodynamics. Cambridge University Press, Cambridge (1932)
Li, Y. A., Hyman, J. M., Choi, W.: A numerical study of the exact evolution equations for surface waves in water of finite depth. Stud. Appl. Math. 113(3), 303–324 (2004)
Mavroyiakoumou, C., Alben, S.: Large-ampitude membrane flutter in inviscid flow. J. Fluid Mech. 891, A23-1–24 (2020)
Milewski, P., Vanden-Broeck, J. -M.: Time dependent gravity-capillary flows past an obstacle. Wave Motion 29(1), 63–79 (1999)
Milewski, P. A., Vanden-Broeck, J. -M., Wang, Z.: Dynamics of steep two-dimensional gravity–capillary solitary waves. J Fluid Mech. 664, 466–477 (2010)
Miloh, T., Tyvand, P.: Nonlinear transient free-surface flow and dip formation due to a point sink. Phys. Fluids A: Fluid Dyn. 5(6), 1368–1375 (1993)
Moreira, R., Peregrine, D.: Interactions between water waves and singularities. In: IUTAM Symposium on Free Surface Flows, pp 205–212. Springer (2001)
Moreira, R., Peregrine, D.: Nonlinear interactions between deep-water waves and currents. J. Fluid Mech. 691, 1–25 (2012)
Muskhelishvili, N.: Singular Integral Equations, 2nd edn. Dover Publications, Inc, New York (1992)
Nie, Q.: The nonlinear evolution of vortex sheets with surface tension in axisymmetric flows. J. Comput. Phys. 174(1), 438–459 (2001)
Nishimura, N.: Fast multipole accelerated boundary integral equation methods. Appl. Mech. Rev. 55(4), 299–324 (2002)
Peregrine, D.: Interaction of water waves and currents. Adv. Appl. Mech. 16, 9–117 (1976)
Persson, P. -O., Strang, G.: A simple mesh generator in MATLAB. SIAM Rev. 46(2), 329–345 (2004)
Pucci, P., Serrin, J.: The maximum principle, vol. 73. Springer Science & Business Media (2007)
Pullin, D. I.: Numerical studies of surface-tension effects in nonlinear Kelvin–Helmholtz and Rayleyigh–Taylor instabilities. J. Fluid Mech. 119, 507–532 (1982)
Robinson, Q.: Theory and experiment for water waves over obstacles. UNC Ph.D Thesis (2018)
Rokhlin, V.: Rapid solution of integral equations of classical potential theory. J. Comput. Phys. 60, 187–207 (1983)
Scullen, D., Tuck, E.: Nonlinear free-surface flow computations for submerged cylinders. J. Ship Res. 39(3), 185–193 (1995)
Tornberg, A. -K., Shelley, M.: Simulating the dynamics and interactions of flexible fibers in Stokes flows. J. Comput. Phys. 196, 8–40 (2004)
Tuck, E.: The effect of non-linearity at the free surface on flow past a submerged cylinder. J. Fluid Mech. 22(2), 401–414 (1965)
Turner, M. R., Bridges, T. J.: Time-dependent conformal mapping of doubly-connected regions. Adv Comput. Math. 42, 947–972 (2016)
Van de Vooren, A. I.: A numerical investigation of the rolling up of vortex sheets. Proc. R. Soc. Lond. Ser. A 373, 67–91 (1980)
Viotti, C., Dutykh, D., Dias, F.: The conformal-mapping method for surface gravity waves in the presence of variable bathymetry and mean current. Procedia IUTAM 11, 110–118 (2014)
Whitham, G.: Linear and Nonlinear Waves, vol. 42. Wiley, New York (2011)
Wilkening, J., Yu, J.: Overdetermined shooting methods for computing standing water waves with spectral accuracy. Comput. Sci Disc. 5, 014017:1–38 (2012)
Wilkening, J., Zhao, X.: Quasi-periodic travelling gravity-capillary waves. J. Fluid Mech. 915, A7:1–35 (2021)
Wilkening, J., Zhao, X.: Spatially quasi-periodic water waves of infinite depth. J Nonlin. Sci. 31(52), 1–43 (2021)
Wu, T. Y.-T.: Cavity and wake flows. Annu. Rev. Fluid Mech. 4, 243–284 (1972)
Zahr, M., Persson, P. -O., Wilkening, J.: An adjoint method for a high-order discretization of deforming domain conservation laws for optimization of flow problems. J. Comput. Phys. 326, 516–543 (2016)
Zakharov, V.: Stability of periodic waves of finite amplitude on the surface of a deep fluid. J. Appl. Mech. Tech. Phys. 9, 190–194 (1968)
Zakharov, V. E., Dyachenko, A. I., Vasilyev, O. A.: New method for numerical simulation of a nonstationary potential flow of incompressible fluid with a free surface. Eur. J. Mech.-B/Fluids 21(3), 283–291 (2002)
Funding
This work was supported in part by the National Science Foundation under award numbers DMS-1907684 (DMA), NSF DMS-1352353 & DMS-1909035 (JLM), DMS-1716560 (JW) and DMS-1910824 (RC & RM); by the Office of Naval Research under award number ONR N00014-18-1-2490 (RC & RM); and by the Department of Energy, Office of Science, Applied Scientific Computing Research, under award number DE-AC02-05CH11231 (JW). JLM wishes to thank the Mathematical Sciences Research Institute for hosting him while a portion of this work was completed.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare no competing interests.
Additional information
Communicated by: Michael O’Neil
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This article belongs to the Topical Collection: Advances in Computational Integral Equations Guest Editors: Stephanie Chaillat, Adrianna Gillman, Per-Gunnar Martinsson, Michael O’Neil, Mary-Catherine Kropinski, Timo Betcke, Alex Barnett
Appendices
Appendix A. Verification of the HLS equations
In Section 2.2, we proposed evolving only P𝜃 via (2.17) and constructing P0𝜃, sα and ζ(α) from P𝜃 via (2.13) and (2.16). Here we show that both equations of (2.10) hold even though P0𝜃 and sα are computed algebraically rather than by solving ODEs, and that these equations, in turn, imply that the curve kinematics are correct, i.e., (ξt,ηt) = Un + V t.
From (2.13), we have \(S_{t} = P_{0}\left [({\cos \limits } P\theta )(P\theta )_{t}\right ]\), \(C_{t} = -P_{0}\left [({\sin \limits } P\theta )(P\theta )_{t}\right ]\), and
In the last step, we used (2.15) and the fact that P is self-adjoint. Combining (2.17) and (A.1), we obtain
We must show that the second term is zero. This follows from Vα = P(𝜃αU) in (2.11). Indeed,
where the integrals are from 0 to 2π and we used (2.15). Similarly, we have
Using Vα = P(𝜃αU) again, we find that
Combining this with (A.3), we obtain sαt = −P0[𝜃αU], as claimed.
As for the second assertion that (ξt,ηt) = Un + V t, note that the equations of (2.10) are equivalent to
By equality of mixed partials, the left-hand side equals \(\partial _{\alpha }\left [\zeta _{t}\right ]\), so we have ζt = (V + iU)ei𝜃 up to a constant that could depend on t but not α. However, to enforce ξt(0) = ∂t0 = 0 in (2.16), we choose V (0) in (5.1) so that the real part of (V + iU)ei𝜃 is zero at α = 0. We conclude that ζt − (V + iU)ei𝜃 = ia, where a is real and could depend on time but not α. We need to show that a = 0. Note that
The divergence theorem implies that \({\int \limits }_{\Gamma } U ds=0\). This is because ∇ϕ is single-valued and divergence free in Ω; \(U=\nabla \phi \cdot \hat {\mathbf n}\) on Γ; \(\nabla \phi \cdot \hat {\mathbf n}=0\) on the solid boundaries; and \((\nabla \phi \cdot \hat {\mathbf n})\vert _{x=2\pi }=-(\nabla \phi \cdot \hat {\mathbf n})\vert _{x=0}\) since ∇ϕ is periodic while \(\hat {\mathbf n}\) changes sign. From (2.16), \({\int \limits }_{0}^{2\pi }\eta \xi _{\alpha } d\alpha =0\) for all time. Differentiating, we obtain
Thus a = 0 and ζt = (V + iU)ei𝜃, as claimed.
Appendix B. Variant specifying the stream function on the solid boundaries
The integral equations of Section 3.2 are tailored to the case where V1, a2, …, aN in the representation (3.1) for Φ are given and the constant values ψ|k are unknown. If instead ψ is completely specified on Γk for 1 ≤ k ≤ N, then we would have to solve for a2, …, aN along with the ωj. In this scenario, \(\varphi =\phi \vert _{{\Gamma }_{0}^{-}}\) is given on the free surface, from which we can extract V1 as the change in φ over a period divided by 2π. So we can write
where \(a_{j}[\omega ] = \langle \mathbf {1}_{j},\omega \rangle = \frac 1{2\pi }{\int \limits }_{0}^{2\pi }\omega _{j} d\alpha \) are now functionals that extract the mean from ω2, …, ωN. Instead of (3.13), we would define
The right-hand side b in (3.15) would become \(b_{0}(\alpha )=\left [\varphi (\alpha )-V_{1}\xi (\alpha )\right ]\) and bk(α) = [ψ(ζk(α)) − V1ηk(α)], where φ and \(\psi \vert _{{\Gamma }_{k}}\) are given. The latter would usually be constant functions, though a nonzero flux through the cylinder boundaries can be specified by allowing \(\psi \vert _{{\Gamma }_{k}}\) to depend on α. However, we still require \(\psi \vert _{{\Gamma }_{k}}\) to be periodic (since the stream function is single-valued in our formulation), so the net flux out of each cylinder must be zero.
We now prove invertibility of this version of \(\mathbb {A}\), which maps ω to the restriction of the real or imaginary parts of \(\check {\Phi }(z)\) to the boundary. We refer to these real or imaginary parts as the “boundary values” of \(\check {\Phi }\). In the same way, \(\mathbb {B}\) maps ω to the boundary values of \(\tilde {\Phi }\) in (3.10). Note that \(\mathbb {A}\) differs from \(\mathbb {B}\) by a rank N − 1 correction in which a basis for \(\mathcal {V}=\ker \mathbb {B}\) is mapped to a basis for the space \(\mathcal {R}_{\text {cyl}}\) of boundary values of \(\operatorname {span}\{{\Phi }_{\text {cyl}}(z-z_{j})\}_{j=2}^{N}\). From Section 6.1, we know that \(\dim \left (\operatorname {coker}(\mathbb {B})\right )=N-1\), so we just have to show that \(\mathcal {R}_{\text {cyl}}\cap \operatorname {ran}(\mathbb {B})=\{0\}\). Suppose the boundary values of \({\Phi }_{c}(z)={\sum }_{j=2}^{N} a_{j}{\Phi }_{\text {cyl}}(z-z_{j})\) belong to \(\operatorname {ran}(\mathbb {B})\). Then there are dipole densities ωj such that the corresponding sum of Cauchy integrals \(\tilde {\Phi }(z)={\sum }_{j=0}^{N}{\Phi }_{j}(z)\) has these same boundary values. The imaginary part, \(\tilde \psi \), satisfies the Laplace equation in Ω, has the same Dirichlet data as ψc on Γ1, …, ΓN, and the same Neumann data as ψc on Γ0 (due to ∂nψ = ∂sϕ). Since solutions are unique, \(\tilde \psi =\psi _{c}\). But the conjugate harmonic function to \(\tilde \psi \) is single-valued while that of ψc is multiple-valued unless all the aj = 0. We conclude that \(\mathcal {R}_{\text {cyl}}\cap \operatorname {ran}(\mathbb {B})=\{0\}\), as claimed.
Appendix C. Cauchy integrals, layer potentials and sums over periodic images
In this section we consider the connection between Cauchy integrals and layer potentials and the effect of summing over periodic images and renormalization. As is well-known [60], Cauchy integrals are closely related to single- and double-layer potentials through the identity
where ζ − z = rei𝜃. We adopt the sign convention of electrostatics [24, 49] and define the Newtonian potential as \(N(\zeta ,z)=-(2\pi )^{-1}\log |\zeta -z|\). The double-layer potential (with normal nζ pointing left from the curve ζ, as in Section 3 above) has the geometric interpretation
For a closed contour in the complex plane, we have
so, if ω is real-valued, the real part of a Cauchy integral is a double-layer potential with dipole density ω while the imaginary part is a single-layer potential with charge density − dω/ds. In the spatially periodic setting, the real part of the two formulas in (3.3) may be written
and
Equation (C.5) follows from Euler’s product formula \({\sin \limits } w \! =\! w{\prod }_{m=1}^{\infty }(1-(w/m \pi )^{2})\), which gives
where \(c_{0}=-\frac 1{2\pi }\log 2\) and \(c_{m}=-\frac 1{2\pi }\log |2\pi m|\) if m≠ 0. It was possible to drop the terms cm in (C.5) and (C.6) since \(\omega _{j}^{\prime }(\alpha )\) is integrated over a period of ωj(α). However, these terms have to be retained to express (C.6) as a principal value integral,
Through (C.7), we can regard \(\log |\sin \limits (w/2)|\) as a renormalization of the divergent sum of the Newtonian potential over periodic images in 2D. Setting aside these technical issues, it is conceptually helpful to be able to interpret ϕ0(z) and ϕj(z) from (3.3) as double and single layer potentials with dipole and charge densities ω0(α) and \(\omega _{j}^{\prime }(\alpha )/s_{\alpha }\), respectively, over the real line or over the periodic array of obstacles. Of course, it is more practical in 2D to work directly with the formulas involving complex cotangents over a single period, but (C.4) and (C.5) are a useful starting point for generalization to 3D.
Appendix D. Alternative derivation of the vortex sheet strength equation
In this appendix, we present an alternative derivation of (4.14) that makes contact with results reported elsewhere [7, 14] in the absence of solid boundaries. As in Section 3, the velocity potential is decomposed into \(\phi (z)=\tilde \phi (z) + \phi _{\text {mv}}(z)\) where \(\tilde \phi (z)\) is the sum of layer potentials and ϕmv(z) is the multi-valued part. We also define W as in (4.10), where the component Birkhoff-Rott integrals W0j are given in complex form by
The Plemelj formulas (4.3) imply that when the interface is approached from the fluid region,
Recall that φ(α,t) = ϕ(ζ(α,t),t) is the restriction of the velocity potential to the free surface as it evolves in time, and note that \(\varphi _{\alpha }=s_{\alpha }\nabla \phi \cdot \hat {\mathbf {t}}\). Solving for γ0, then, we have
Differentiating with respect to time, we get
In Section 4.2, we avoided directly taking time derivatives of γ0(α,t), W(α,t) and φ(α,t), which lead to more involved calculations here due to the moving boundary. We know that \(\hat {\mathbf {t}}_{t}=\theta _{t}\hat {\mathbf {n}},\) and that 𝜃t = (Uα + V 𝜃α)/sα. We substitute these to obtain
We now work on the equation for φαt. As was done in [7], the convective derivative (3.25) together with the Bernoulli equation gives
We write \(\mathbf {W} =U\hat {\mathbf {n}}+(\mathbf {W} \cdot \hat {\mathbf {t}})\hat {\mathbf {t}}\), substitute (D.2) into (D.5), and use \(\mathbf {W} \cdot \mathbf {W} =U^{2}+(\mathbf {W} \cdot \hat {\mathbf {t}})^{2}\):
We differentiate with respect to α:
We substitute (D.6) into (D.4), noticing that the UUα terms cancel:
We group this as follows:
The quantity in square brackets simplifies considerably using the equations Vα = sαt + 𝜃αU, \(U=\mathbf {W} \cdot \hat {\mathbf {n}},\) and \(\hat {\mathbf {t}}_{\alpha }=\theta _{\alpha }\hat {\mathbf {n}}\). Together with the boundary condition for the pressure (the Laplace-Young condition), we obtain
This agrees with the equation for γ0,t as found in [7] if one assumes (sα)α = 0. The calculation of [7] has no solid boundaries and a second fluid above the first, which we take to have zero density when comparing to (D.7).
Our final task is to compute \(s_{\alpha }\mathbf {W}_{t}\cdot \hat {\mathbf {t}} = (\mathbf {W}_{00,t} + {\cdots } + \mathbf {W}_{0N,t})\cdot (s_{\alpha } \hat {\mathbf t})\) in the right-hand side of (D.7). Differentiating (D.1) with respect to time for 1 ≤ j ≤ N gives
Here, as above, a prime denotes ∂α and we note that the solid boundaries remain stationary in time. Suppressing t in the arguments of functions again, we conclude that for 1 ≤ j ≤ N,
where ζt is treated as the vector (ξt,ηt) in the dot product. When j = 0, we regularize the integral
and then differentiate both sides with respect to time
Observing that \(\zeta _{t}^{\prime }W_{00}^{*}=\left ([\zeta _{t}W_{00}^{*}]_{\alpha }-\zeta _{t}W_{00,\alpha }^{*}\right )\), we find that
Finally, setting Wmv = ∇ϕmv(ζ(α,t)), we compute
When (D.8), (D.9) and (D.10) are combined and substituted into (D.7), several of the terms cancel:
Also, in (D.9), ζt ⋅W00 cancels the ζt(α) term in the integrand, leaving behind a principal value integral. Including the other terms of (D.7), moving the unknowns to the left-hand side, and dividing by 2, we obtain (4.14).
Appendix E. Treating the bottom boundary as an obstacle
The conformal map w = e−iz maps the infinite, 2π-periodic region \({\Omega }_{1}^{\prime }\) below the bottom boundary to a finite domain, with \(-i\infty \) mapped to zero. Let \(w_{j}=e^{-iz_{j}}\) denote the images of the points zj in (3.2), which are used to represent flow around the obstacles via multi-valued velocity potentials. We also define the curves
which traverse closed loops in the w-plane, parameterized clockwise. The image of the fluid region lies to the right of Υ0(α) and to the left of Υj(α) for 1 ≤ j ≤ N. The terms V1z and ajΦcyl(z − zj) appearing in (3.2) all have a similar form in the new variables,
We can think of V1z as a multiple-valued complex potential on the 2π-periodic domain of logarithmic type with center at \(z_{1}=-i\infty \). It maps to \(V_{1}i\log (w-w_{1})\) in the w-plane, where w1 = 0. From (3.5), we see that the n th sheet of the Riemann surface for Φcyl(z(w) − zj) is given by − i Log(1 − wj/w) + 2πn, which has a branch cut from the origin to wj. When traversing the curve w = Υk(α) with α increasing, the function Φcyl(z(w) − zj) decreases by 2π if k = j, increases by 2π if k = 1, and returns to its starting value for the other boundaries, including the image of the free surface (k = 0). This is done so that only the V1z term has a multiple-valued real part on Γ0, which simplifies the linear systems analyzed in Section 6.1–6.2 above.
The cotangent-based Cauchy integrals Φj(z) in (3.3) transform to (1/w)-based Cauchy integrals in the new variables, aside from an additive constant in the kernels [24]. In more detail,
For 1 ≤ j ≤ N, we then have
with a similar formula for Φ0(z(w)), replacing iωj(α) by ω0(α). The second term is a constant function of w that prevents 11 from being annihilated by \(\mathbb {B}\) in Section 3.2. This is the primary way in which the bottom boundary differs from the other obstacles in the analysis of Sections 6.1–6.2.
We note that \(\tilde {\Phi }(z(w))={\sum }_{j=0}^{N}{\Phi }_{j}(z(w))\) is analytic at w = 0, which allows us to conclude that if its real or imaginary part satisfies Dirichlet conditions on \({\Gamma }_{1}^{-}\), it is zero in \({\Omega }_{1}^{\prime }\). A similar argument using w = eiz works for the region \({\Omega }_{0}^{\prime }\) above the free surface, which was needed in Section 6.2 above.
Appendix F. Evaluation of Cauchy integrals near boundaries
In this section we describe an idea of Helsing and Ojala [42] to evaluate Cauchy integrals with spectral accuracy even if the evaluation point is close to the boundary. We modify the derivation to the case of a 2π-periodic domain, which means the \(\frac 1z\) Cauchy kernels in [42] are replaced by \(\frac 12\cot \frac z2\) kernels here. The key idea is to first compute the boundary values of the desired Cauchy integral f(z). The interior values are expressed in terms of these boundary values. From the residue theorem, we have
where \(\partial {\Omega }=\cup _{k=0}^{N} {\Gamma }_{k}\). Multiplying the second equation by f(z) and subtracting from the first, we obtain
The integrand is a product of two analytic functions of z and ζ, namely \(\frac {\zeta -z}2\cot \frac {\zeta -z}2\) and the divided difference \(f[\zeta ,z]=\left (f(\zeta )-f(z)\right )/(\zeta -z)= {{\int \limits }_{0}^{1}} f^{\prime }(z+(\zeta -z)\alpha ) d\alpha \). In particular, \(f[\zeta ,\zeta ]=f^{\prime }(\zeta )\) is finite, and the k th partial derivative of f[ζ,z] with respect to ζ is bounded, uniformly in z, by \(\max \limits _{w\in {\Omega }}|f^{{(k+1)}}(w)|/(n+1)\). Thus, the integrand is smooth and the integral can be approximated with spectral accuracy using the trapezoidal rule,
Solving for f(z) gives
In (5.5), we interpret this as a quadrature rule for evaluating the first integral of (F.1) that maintains spectral accuracy even if z approaches or coincides with a boundary point ζk(αkm).
Appendix G. Remarks on generalization to three dimensions
We anticipate that both methods of this paper generalize to 3D with some modifications. One aspect of the problem becomes easier in 3D, namely that the velocity potential is single-valued. However, one loses complex analysis tools such as summing over periodic images in closed form with the cotangent kernel and making use of the residue theorem to accurately evaluate layer potentials near the boundary.
The velocity potential method can be adapted to 3D by replacing constant boundary conditions for the stream function on the solid boundaries with homogeneous Neumann conditions for the velocity potential. This entails using a double-layer potential on the free surface and single-layer potentials on the remaining boundaries. In her recent PhD thesis [48], Huang shows how to do this in an axisymmetric HLS framework. She implemented the method to study the dynamics of an axisymmetric bubble rising in an infinite cylindrical tube. One of the biggest challenges was finding an analog of the Hilbert transform to regularize the hypersingular integral that arises for the normal velocity. Huang introduces a three-parameter family of harmonic functions involving spherical harmonics for this purpose. This method can handle background flow along the axis of symmetry, but many technical challenges remain for the non-axisymmetric case, e.g., for doubly periodic boundary conditions in the horizontal directions.
Analogues of the vortex sheet method in three dimensions have been developed previously in various contexts. Caflisch and Li [17] work out the evolution equations in a Lagrangian formulation of a density-matched vortex sheet with surface tension in an axisymmetric setting. Nie [61] shows how to incorporate the HLS method to study axisymmetric, density-matched vortex sheets. In his recent PhD thesis, Koga [51] studies the dynamics of axisymmetric vortex sheets separating a “droplet” from a density-matched ambient fluid. He develops a mesh refinement scheme based on signal processing and shows how to regularize singular axisymmetric Biot-Savart integrals with new quadrature rules. Koga implements these ideas using graphics processing units (GPUs) to accelerate the computations.
The non-axisymmetric problem with doubly periodic boundary conditions has been undertaken by Ambrose et al. [10]. They propose a generalized isothermal parameterization of the free surface, building on work of Ambrose and Masmoudi [8], which possesses several of the advantages of the HLS angle-arclength parameterization in 2D. The context of [10] is interfacial Darcy flow in porous media, which also involves Birkhoff-Rott integrals in 3D:
Here \(\vec {\alpha }=(\alpha ,\beta ),\) and the surface is given by \(\mathbf {X}(\vec {\alpha })=(\xi (\vec \alpha ), \eta (\vec \alpha ), \zeta (\vec \alpha ))\) with ζ now the z-coordinate instead of the complexified surface. In the integrand, the subscripts α and β represent derivatives with respect to these variables, and quantities without a prime are evaluated at \(\vec {\alpha }\) while quantities with a prime are evaluated at \(\vec {\alpha }^{\prime }\). The domain of integration is \(\mathbb {R}^{2}\). The quantity ω is, as in the 2D problem, the source strength in the double-layer potential.
The lack of a closed formula for the sum over periodic images in (G.1) contributes to the computational challenge of implementing the method in 3D. In [10], a fast method for calculation of this integral is introduced, based on Ewald summation. This involves splitting the calculation of the integral into a local component in physical coordinates and a complementary calculation in Fourier space; the method is optimized so that the two sums take similar amounts of work. We expect that the single layer potentials that occur at solid boundaries in the multiply connected case of the present paper could be computed similarly in 3D.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Ambrose, D.M., Camassa, R., Marzuola, J.L. et al. Numerical algorithms for water waves with background flow over obstacles and topography. Adv Comput Math 48, 46 (2022). https://doi.org/10.1007/s10444-022-09957-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s10444-022-09957-z
Keywords
- Water waves
- Multiply connected domain
- Layer potentials
- Cauchy integrals
- Overturning waves
- Splash singularity
- GPU acceleration