betaFeedback

WorksPhysics and Astronomy

Surface wakes on ultra-soft solids

Abstract

We explore the dynamical response of the free surface of an ultra-soft solid driven by a localized moving pressure disturbance. Experiments reveal a steady V-shaped wake analogous to a surface Mach wedge. A simple geometric argument provides a qualitative explanation consistent with observations. A theoretical framework combining elastodynamic, capillary, and gravitational effects yields a generalized dispersion relation that smoothly interpolates between Kelvin's theory of liquid interface wakes and Rayleigh's theory of elastic surface waves. Together, our experiments and theory reveal the existence of a soft wake regime that bridges fluid and solid surface wave physics, offering new routes for probing the dynamics of soft surfaces.

Soft elastic solids are easily deformable when subjected to external loads, and are characterized by relatively small (zero-frequency) shear moduli. The interface of a soft elastic solid is even softer than the bulk solid because in addition to a weak material response, it is soft by virtue of geometry (i.e., it is free on one side). Thus, soft interfaces are susceptible to both body forces (e.g., gravity) and interfacial forces (e.g., surface tension). For a solid with density ρ\rho, elastic shear modulus μ\mu, a surface energy per unit area γ\gamma, subject to a gravitational body force per unit volume ρg\rho g, one can define two natural length scales: an elastocapillary length, ℓce=γ/μ\ell_{ce} = \gamma/\mu and an elastogravity length, ℓeg=μ/ρg\ell_{eg} = \mu/\rho g. For ultra-soft solids, elastic, capillary and gravitational effects all become simultaneously important when the elastocapillary length and the elastogravity length become comparable. Then, letting ℓeg∼ℓce\ell_{eg} \sim\ell_{ce} yields an expression for the shear modulus μ∗∼(ρgγ)1/2\mu^* \sim(\rho g\gamma)^{1/2}; substituting ρ=1000 kg/m3\rho= 1000\ \mathrm{kg/m^3}, g=10 m/s2g = 10\ \mathrm{m/s^2}, γ=10 mN/m\gamma= 10\ \mathrm{mN/m} yields μ∗∼10 Pa\mu^* \sim10\ \mathrm{Pa}. The recent ability to create very weak materials [3] with ultra-low moduli in this range, i.e., μ∼O(10) Pa\mu\sim O(10)\ \mathrm{Pa} makes this capillary-gravity-elasticity regime accessible via tabletop experiments [4].

However, little is still known about the dynamics of ultra soft solids, when there is a delicate interplay between inertia, elasticity, gravity and capillarity, e.g., when a soft interface is driven by a transient surface force [8], a topic that has implications for fields such as elastography [12] [11] and non-destructive imaging [14] [13]. Here, we study the 2-dimensional wake left by a localized pressure source moving on the surface of a 3-dimensional solid, using experiment and theory to understand the wake on an ultra-soft solid interface, a problem connecting two classical problems: Rayleigh waves on the surface of an elastic solid [15] and the Kelvin wake behind a ship moving in deep water [16].

To create a dynamic pressure source moving at a speed UU above a soft elastic solid (Figure 1a), we start with a large tank filled with about 10 litres of an ultrasoft polyacrylamide hydrogel with elastic shear modulus, μ≈6.5 Pa\mu\approx6.5\ \mathrm{Pa} (See SI for details of the protocol to create and measure the rheology of the gel) and density ρ=1000 kg/m3\rho= 1000\ \mathrm{kg/m^3}. The dimensions of the soft solid slab are 50 cm×25 cm50\ \mathrm{cm} \times25\ \mathrm{cm} with a depth H=8 cmH = 8\ \mathrm{cm}. The pressure source was created by forcing compressed air out through a thin nozzle (d=0.84 mmd = 0.84\ \mathrm{mm}), mounted on a rail that can slide horizontally above the gel surface at a constant height of ∼4 cm\sim4\ \mathrm{cm} above it. As the pressure source moves at a uniform speed (UU) above the gel surface, we observe a wake behind the moving disturbance (Figure 1b, SI movie S1,S2), with ridges of maximum surface elevation on the soft interface forming a linear wedge-like profile. The opening angle of the leading dominant ridge (2α2\alpha) depends on the speed of the pressure source UU; faster sources yield wakes with smaller angles. Indeed, plotting sin⁡α\sin\alpha versus the scaled inverse speed of the pressure source cs/Uc_s/U, where cs=μ/ρc_s = \sqrt{\mu/\rho} is the shear-wave speed in the soft solid, we see a linear relationship with sin⁡α≈4.5cs/U\sin\alpha\approx4.5c_s/U (Figure 1c).

Image not extracted

Figure 1. (a) Schematic of the experimental setup, where a pressure source of compressed air moving at speed UU impinges on the surface of an elastic solid, generating a surface wake. (b) Top-view images of wakes for two representative speeds, U=0.55 m/sU = 0.55\ \mathrm{m/s} (left) and U=1.28 m/sU = 1.28\ \mathrm{m/s} (right) leads to different opening angles (2α2\alpha) on an ultra-soft solid (μ≈6.5 Pa\mu\approx6.5\ \mathrm{Pa}). The wake angle decreases with the increasing speed of the pressure source. A projected grid on the surface of the gel slab from the top enables visualization of the surface deformation. (c) Experimental data showing Mach-like scaling, where sin⁡α\sin\alpha varies linearly with cs/Uc_s/U, and cs=μ/ρc_s = \sqrt{\mu/\rho} is the Rayleigh (shear) wave speed. (d) Schematic of the geometry of wave fronts on a soft solid surface due to a moving pressure source.

A natural comparison of this observation is with the wake observed on liquid surfaces in the ‘deep-water’ limit [17, 18, 19, 20] [16]. However, the wake angle here deviates substantially from the Kelvin angle (i.e, 19.4°) characteristic of gravity wakes at low Froude numbers (Fr < 1) in a few ways. Firstly, the wake is sharper and more V-like without the feathered decorations seen in gravity waves. Secondly, the angle varies with the speed of the source and does not saturate to the Kelvin limit for classical gravity-capillary wakes [21]. Finally, in contrast to gravity–capillary wakes, the wake patterns we observe exhibit only faint fine features away from the source, indicating strong damping (See SI movie S1 for a comparison between soft gel and water wakes).

To get a qualitative understanding of our observations, we first consider a minimal model for the elastic wake, ignoring, in the first instance, the effects of gravity and capillarity. On the surface of an incompressible solid, freely propagating non-dispersive Rayleigh waves propagate with a speed proportional to the shear wave speed csc_s [15]. If a pressure source moves on the surface with a speed UU greater than the csc_s, we expect a Mach-like wake to emerge. A geometric argument for the wake half-angle, α\alpha, follows by considering the superposition of the circular wavefronts (Figure 1d) and leads to the simple relationship,

sin⁡α=Acs/U,(1)\sin\alpha= A c_s/U, \tag*{(1)}

which is qualitatively consistent with our experimental measurements where A∼4.5A \sim4.5, i.e., the wake angle increases with decreasing speed. To explain the weakly dispersive nature of the wake (see SI movie 1,2), and the prefactor requires a quantitative theory of surface waves including the contributions from gravity, surface tension and elastodynamics that we now turn to.

We assume that the free surface bounds a soft incompressible elastic half-space (z≤0z \le0) of uniform density ρ\rho and shear modulus μ\mu. Displacements of the solid are given by ξ(x,z,t)≡(u,w)\boldsymbol{\xi}(\boldsymbol{x},z,t) \equiv(u,w), where x=(x,y)\boldsymbol{x} = (x,y). In the linear elastodynamic regime relevant for our observations, the dynamical equations of motion read

ρ∂2ξ∂t2=−∇P+μ∇2ξ,(2)\rho\frac{\partial^2 \boldsymbol{\xi}}{\partial t^2} = -\nabla P + \mu\nabla^2 \boldsymbol{\xi}, \tag*{(2)}

subject to the condition of incompressibility ∇⋅ξ=0\nabla\cdot\boldsymbol{\xi} = 0, which upon substituting into eq. (2) gives

∇2P=0.(3)\nabla^2 P = 0. \tag*{(3)}

Since we are interested in the response to small-amplitude disturbances at the surface, we take PP to be the excess pressure from the pre-stressed equilibrium configuration, so that the boundary conditions deep in the interior are given by ξ→0\boldsymbol{\xi} \to0 and P→0P \to0 as z→−∞z \to-\infty. On the surface, capillary forces provide a stress proportional to the surface’s mean curvature, yielding the free-surface boundary conditions σiz,0=(μ∂w/∂xi+μ∂ui/∂z)z=0=0\sigma_{iz,0}=(\mu\partial w/\partial x_i+\mu\partial u_i/\partial z)_{z=0}=0, σzz,0=(−P+2μ∂w/∂z)z=0=(Pz+γ∇∥2w−ρgw)z=0\sigma_{zz,0}=(-P+2\mu\partial w/\partial z)_{z=0}=(P_z+\gamma\nabla_{\parallel}^2w-\rho g w)_{z=0}, where i=x,yi=x,y and ∇∥=(∂/∂x,∂/∂y)\nabla_{\parallel}=(\partial/\partial x,\partial/\partial y), and Pz(x,t;U,L)P_z(x,t;U,L) is the external moving pressure source applied onto the surface, moving in the xx direction with a constant speed UU. For simplicity, we assume an isotropic pressure source with a characteristic size LL.

We solve the linear eqs. (2) and (3) subject to the boundary conditions using 2-dimensional Fourier methods, assuming sufficient far-field decay, using the standard definition f(x,z,t)=∫dω∫d2k(2π)2f~(k,z,ω)ei(k⋅x−ωt)f(x,z,t)=\int d\omega\int\frac{d^2k}{(2\pi)^2}\tilde f(k,z,\omega)e^{i(k\cdot x-\omega t)}, where f~\tilde f is its corresponding transform, and the arbitrary function ff can be PP, uu, and ww. This yields the complete solution (see section B1 in SI for details)

P~=ω2(ω2−2cs2k2)ekzD(k,ω)P~z(k,ω),(4)\tilde P = \frac{\omega^2(\omega^2-2c_s^2k^2)e^{kz}}{D(k,\omega)}\tilde P_z(k,\omega), \tag*{(4)}
u~=(ik^)(ω2−2cs2k2)kekz+2(k2cs2)qeqzρD(k,ω)P~z(k,ω),(5)\tilde u = \frac{(i\hat{k})\left(\omega^2-2c_s^2k^2\right)ke^{kz}+2(k^2c_s^2)q e^{qz}}{\rho D(k,\omega)}\tilde P_z(k,\omega), \tag*{(5)}
w~=(ω2−2cs2k2)kekz+2(k2cs2)keqzρD(k,ω)P~z(k,ω),(6)\tilde w = \frac{(\omega^2-2c_s^2k^2)ke^{kz}+2(k^2c_s^2)k e^{qz}}{\rho D(k,\omega)}\tilde P_z(k,\omega), \tag*{(6)}

where k=∣k∣k=|\boldsymbol{k}|, q=k2−(ω/cs)2q=\sqrt{k^2-(\omega/c_s)^2}, and the denominator:

D(k,ω)=ω2(gk+γρk3)−(2cs2k2−ω2)2+4cs3k3cs2k2−ω2,(7)\begin{aligned} D(k,\omega)=\omega^2\left(gk+\frac{\gamma}{\rho}k^3\right)-\left(2c_s^2k^2-\omega^2\right)^2 \\ &+4c_s^3k^3\sqrt{c_s^2k^2-\omega^2}, \tag*{(7)} \end{aligned}

which is consistent with the characteristic equation derived in [10] for the 1-dimensional case.

For a disturbance pressure source moving with uniform velocity, as in our experiment, the Doppler shift sets ω=U⋅k\omega= \mathbf{U} \cdot\mathbf{k}. The stationary surface displacement field is then obtained from the Fourier transform of eq. (6) evaluated at z=0z = 0 as follows [22],

w(x)=1(2π)2ρ∬d2k eik⋅x(U⋅k)2D(k,U⋅k)P~z(k;L),(8)w(\mathbf{x}) = \frac{1}{(2\pi)^2\rho} \iint\mathrm{d}^2 k \, \mathrm{e}^{i\mathbf{k}\cdot\mathbf{x}} \frac{(\mathbf{U}\cdot\mathbf{k})^2}{D(\mathbf{k},\mathbf{U}\cdot\mathbf{k})} \widetilde{P}_z(\mathbf{k};L), \tag*{(8)}

where x−Ut→x\mathbf{x}-\mathbf{U}t \to\mathbf{x}, with U=Uex\mathbf{U}=U\mathbf{e}_x, is the new coordinate vector in the moving frame. The stationarity condition,

ω=U⋅k=Ukcos⁡ϕ,(9)\omega= \mathbf{U}\cdot\mathbf{k} = Uk\cos\phi, \tag*{(9)}

where ϕ\phi is the angle between the wave vector k\mathbf{k} and the direction of moving source (see Figure 5c), restricts the surface waves that interfere constructively to those with phase speed cp(k):=ω/k<Uc_p(k):=\omega/k<U. Furthermore, the dispersion relation for surface waves is given by D(k,ω)=0D(k,\omega)=0. We now examine the expression for DD to understand the effects of elasticity.

Image not extracted

Figure 5. (a) Dimensionless phase and group speeds as functions of wavenumber from gravity-capillary elastodynamics (Cs>0C_s>0) following (14) and (16), showing the elasto-gravity (Cg<CpC_g<C_p) and elasto-capillary branches (Cg>CpC_g>C_p). (b) Schematic of the stationary wake pattern from interference of monochromatic waves generated by a source moving from S−tS_{-t} to S0S_0, adapted from [27] [21]. (c) Radiation angle β\beta for allowed wavenumbers at fixed Bo, Fr, with varying shear-wave speeds CsC_s, following (17). The dashed curves in (b) and (c) correspond to Cs=0C_s=0. (d) Schematic of the resultant wake pattern formed by superposition of broadband waves, defining the characteristic angles.

To work with dimensionless quantities, we use the pressure source size LL as the reference length and U/LU/L as the reference frequency, and define K:=kLK:=kL and Ω:=ωL/U\Omega:=\omega L/U. This induces the transformation (L3/U2g)D(K/L,ΩU/L)→D(K,Ω)(L^3/U^2g)D(K/L,\Omega U/L)\to D(K,\Omega), so that the dimensionless counterpart of eq. (7) is:

D(K,Ω)=Ω2Ωcg2−Fr2(2Cs2K2−Ω2)2+4Fr2Cs3K3Cs2K2−Ω2,(10)D(K,\Omega)=\Omega^2\Omega_{\rm cg}^2-\mathrm{Fr}^2(2C_s^2K^2-\Omega^2)^2+4\mathrm{Fr}^2C_s^3K^3\sqrt{C_s^2K^2-\Omega^2}, \tag*{(10)}

where Ωcg2:=K+Bo−2K3\Omega_{\rm cg}^2:=K+\mathrm{Bo}^{-2}K^3 is the dimensionless dispersion relation for gravity-capillary waves, where Bo:=ρgL2/γ=L/(ℓceℓeg)1/2\mathrm{Bo}:=\sqrt{\rho gL^2/\gamma}=L/(\ell_{\rm ce}\ell_{\rm eg})^{1/2} is the Bond number, Fr=U/gL\mathrm{Fr}=U/\sqrt{gL} is the Froude number, and Cs:=cs/UC_s:=c_s/U the scaled shear-wave speed in the soft solid.

The dispersion relation for the soft surface, given by D(K,Ω)=0D(K,\Omega)=0, interpolates between two venerable classical problems involving surface waves. When Cs=0C_s=0, we recover the well-known dispersion relation for deep water gravity-capillary waves [18, 23], with Ω=Ωcg/Fr\Omega=\Omega_{\rm cg}/\mathrm{Fr}. In the other limit, when Ωcg=0\Omega_{\rm cg}=0 (Fr→∞\mathrm{Fr}\to\infty), the dispersion relation satisfies

Ω2=2Cs2K2+2Cs3K3Cs2K2−Ω2\Omega^2=2C_s^2K^2+2\sqrt{C_s^3K^3\sqrt{C_s^2K^2-\Omega^2}}

[24] leading to the well-known Rayleigh wave speed on the surface of an incompressible half-space with vR=Ω/K≈0.955Csv_R=\Omega/K\approx0.955C_s [15].

In our experiments, with cs≈0.08 m/sc_s\approx0.08\ \mathrm{m/s} and U∈(0.5,1.4) m/sU\in(0.5,1.4)\ \mathrm{m/s}, we have Cs<0.2C_s<0.2. Assuming that the large-wavenumber contributions are suppressed by the finite bandwidth of the pressure source (CsK≪1C_sK\ll1), the effects of elasticity may be treated perturbatively. This allows us to derive a dispersion relation that minimally captures the leading order effects due to elasticity, but unlike in the 1-dimensional case [9, 10], the 2-dimensional problem warrants a distinction between modes as a function of the phase speeds (Cp:=cp/UC_p:=c_p/U) relative to the shear wave speed (CsC_s).

To facilitate discussion of these two different modes, we define a parameter λ(K):=Cs/Cp(K)\lambda(K):=C_s/C_p(K). Then, for modes propagating slower than the shear-wave (λ>1\lambda>1), i.e., Cp<Cs≪1C_p<C_s\ll1, the function DD (10) is real-valued and reduces to:

D(K,Ω)=Ω2(Ωcg2+4φ1(λ2)Fr2Cs2K2−Fr2Ω2),(11)D(K,\Omega)=\Omega^2\left(\Omega_{\rm cg}^2+4\varphi_1(\lambda^2)\mathrm{Fr}^2C_s^2K^2-\mathrm{Fr}^2\Omega^2\right), \tag*{(11)}

where φ1:[1,∞)→[0,1/2)\varphi_1:[1,\infty)\to[0,1/2) is a smooth bounded function (see SI B.2). We see that the softness of the elastic half-space introduces a quadratic correction to the dispersion relation for classical gravity-capillary waves on a liquid half-space. In contrast, for modes travelling faster than the shear-wave (λ<1\lambda<1), i.e., Cs<Cp<1C_s<C_p<1, the function DD (10) becomes complex-valued and reduces to:

D(K,Ω)=Ω2(Ωcg2+4φ2(λ2)Fr2Cs2K2−Fr2Ω2)+4iΩFr2Cs3K3φ2(λ2).(12)\begin{aligned} D(K,\Omega)={}&\Omega^2\left(\Omega_{\rm cg}^2+4\varphi_2(\lambda^2)\mathrm{Fr}^2C_s^2K^2-\mathrm{Fr}^2\Omega^2\right)\\ &+4i\Omega\mathrm{Fr}^2C_s^3K^3\sqrt{\varphi_2(\lambda^2)}. \tag*{(12)} \end{aligned}

where φ2:[0,1]→[0,1]\varphi_2:[0,1]\to[0,1] is yet another smooth bounded function (see SI B.2). Here, the softness introduces an additional leading-order cubic contribution to the imaginary part of DD, and the solution to D(K,Ω)=0D(K,\Omega)=0 lies in a small neighborhood of the solution to Re⁡(D)=0\operatorname{Re}(D)=0 in the complex KK-plane. Thus, elasticity introduces an effective damping of the propagating surface waves even in the absence of any viscous effects in our energy-conserving theory – a consequence of bulk radiation that leads to leaky waves [9].

For our experimental conditions, where the quantity μ/ρgγ=ℓeg/ℓce∼0.2\mu/\sqrt{\rho g\gamma}=\sqrt{\ell_{\rm eg}/\ell_{\rm ce}}\sim0.2 is small, there are no real solutions to D(K,Ω)=0D(K,\Omega)=0 at the critical condition Ω=CsK\Omega=C_sK, which precludes the simultaneous existence of fast and slow modes. Furthermore, we find that only the ‘leaky’ fast modes are supported, owing to the very low shear modulus of the ultra-soft gel (see section B.2 in SI) [25]. Henceforth, we consider only the reduced (real-part of the) dispersion relation for fast modes from eq. (12) to describe the wake pattern:

Fr Ω=(Ωcg2+4φ2(λ2)Fr2Cs2K2)1/2,(13)\mathrm{Fr}\,\Omega=\left(\Omega_{\rm cg}^2+4\varphi_2(\lambda^2)\mathrm{Fr}^2C_s^2K^2\right)^{1/2}, \tag*{(13)}

which yields the dimensionless phase speed:

Fr2Cp2=1/K+K/Bo2+4φ2(λ2)Fr2Cs2.(14)\mathrm{Fr}^2C_p^2=1/K+K/\mathrm{Bo}^2+4\varphi_2(\lambda^2)\mathrm{Fr}^2C_s^2. \tag*{(14)}

This curve, plotted in Figure 2, shows that the phase speed has a minimum at K∗=BoK^*=\mathrm{Bo}, with the corresponding value Cp,min⁡=2/(Fr2Bo)+4φ2(λ2)Cs2)1/2C_{p,\min}=2/(\mathrm{Fr}^2\mathrm{Bo})+4\varphi_2(\lambda^2)C_s^2)^{1/2}. We see that finite elasticity associated with non-zero CsC_s introduces a correction to the well-known minimum gravity-capillary phase speed (2/Fr2Bo)1/2(2/\mathrm{Fr}^2\mathrm{Bo})^{1/2} [17] [21], and leads to a vertical shift of the curve relative to the classical capillary-gravity case, shown in Figure 2. However, the minimum still occurs at the same scaled gravity-capillary wavenumber K∗=(ρgL2/γ)1/2K^*=(\rho gL^2/\gamma)^{1/2}.

Image not extracted

Figure 2. Dimensionless phase speed CpC_p of free waves on the surface of an infinitely deep ultra-soft solid with weak elasticity (Cs≪1C_s \ll1), shown as a function of the scaled wavenumber KK following (14). The minimum occurs at K∗=(ρgL2/γ)1/2K^*=(\rho gL^2/\gamma)^{1/2}, with a value Cp,min⁡=(2/(Fr2Bo)+4φ2(λ2)Cs2)1/2C_{p,\min}=(2/(\mathrm{Fr}^2\mathrm{Bo})+4\varphi^2(\lambda^2)C_s^2)^{1/2}. The dashed line corresponds to the capillary-gravity waves in infinitely deep inviscid liquid without elasticity (Cs=0C_s=0).

The condition of stationarity that derives from (9) determines the allowable wavenumbers for pattern formation, so that setting Cp=Ω/K=1C_p=\Omega/K=1 in eq. (14) and solving for KK gives the critical wavenumbers K1,2(Cs)K_{1,2}(C_s). Consequently, wavenumbers in the range K∈[K1(Cs),K2(Cs)]K\in[K_1(C_s),K_2(C_s)] contribute to constructive interference, as shown in Figure 2. For a given pressure source speed, the range of allowable wavenumbers becomes narrower with increasing elasticity, owing to the vertical shift of the curve shown in Figure 2.

To compare our theory with observations, we perform numerical simulations to compute the far-field surface wave amplitude that develops in response to an applied pressure field using eq. (8), with the complete dispersion expression corresponding to eq. (10). We model the pressure source as a radially-symmetric Gaussian field of size LL moving with uniform velocity UU in the xx-direction: Pz(x,t;U,L)=P0exp⁡[−π(∣x−Ut∣)2/L2]P_z(\boldsymbol{x},t;U,L)=P_0\exp[-\pi(|\boldsymbol{x}-U t|)^2/L^2]. Introducing dimensionless variables X:=x/LX:=x/L, W:=w/LW:=w/L, and recalling the earlier scaling K=kLK=kL, Ω=ωL/U\Omega=\omega L/U, the scaled version of the surface elevation in the co-moving frame in eq. (8) is:

W(X)=lim⁡ϵ→01(2π)2Re⁡∬d2K eiK⋅XΩϵ2P~z∗(K)D(K,Ωϵ)(15)W(\boldsymbol{X})=\lim_{\epsilon\to0}\frac{1}{(2\pi)^2}\operatorname{Re}\iint\mathrm{d}^2K\,e^{i\boldsymbol{K}\cdot\boldsymbol{X}}\frac{\Omega_\epsilon^2\widetilde{P}_z^*(K)}{D(K,\Omega_\epsilon)} \tag*{(15)}

where DD follows from eq. (10), Ωϵ:=Ω+iϵ=U^⋅K+iϵ\Omega_\epsilon:=\Omega+i\epsilon=\widehat{\boldsymbol{U}}\cdot\boldsymbol{K}+i\epsilon, and P~z∗(K):=P~z(K)/(ρgL3)∝exp⁡(−K2/4π)\widetilde{P}_z^*(K):=\widetilde{P}_z(K)/(\rho gL^3)\propto\exp(-K^2/4\pi). For the assumed pressure field, we numerically compute W(X)W(\boldsymbol{X}) using a 2-dimensional inverse fast Fourier transform on a fixed square domain (spanning 6 wavelengths of gravity wave), discretized on an N×N=40962N\times N=4096^2 grid, with the regularization Ω→Ωϵ\Omega\to\Omega_\epsilon, choosing ϵ≈10−2\epsilon\approx10^{-2} as a compromise between artificial oscillations introduced by small ϵ\epsilon and exponential decay at large ϵ\epsilon, respectively (see SI for details).

Sweeping over the scaled parameter space (Fr,Cs)(\mathrm{Fr},C_s), keeping Bo fixed allows us to compare the results with our experiments.

experiments. In Figure 3, we show the wake patterns characterized by the opening semi-angle α\alpha of the leading ridge [26] (see Figure 6 in End Matter for definitions of α\alpha and standard wake angle), from two different sets of simulations performed with Bo≈3.7\mathrm{Bo}\approx3.7. In Figure 3a, the scaled shear-wave speed CsC_s is varied while keeping the Froude number constant. The opening semi-angle α\alpha increases with increasing elasticity, and the wake distinctly becomes more spatially damped, consistent with both the theoretical predictions and the experimental trend as shown in Figure 1b, c. In Figure 3b, we show the effects of the Froude number Fr on wake angle, that can be achieved by changing either the disturbance size LL or speed UU. The opening angle decreases at large Fr, in qualitative agreement with our experimental observations in Figure 1b.

Image not extracted

Figure 3. Wake pattern on ultra-soft solids showing the normalized surface elevation, computed using (15) at Bo≈3.7\mathrm{Bo}\approx3.7, for different scaled shear-wave speeds Cs=cs/UC_s=c_s/U and Froude numbers Fr as functions of coordinates X/λgX/\lambda_g and Y/λgY/\lambda_g, where λg=2πFr2\lambda_g=2\pi\mathrm{Fr}^2 is the dimensionless wavelength of the gravity wave corresponding the source size LL. The opening half-angle α\alpha of the leading dominant ridge (dashed black line) is shown for increasing (a) CsC_s at Fr=1.1\mathrm{Fr}=1.1 and (b) Fr at Cs=0.02C_s=0.02. The computational domain is approximately 0.75Fr20.75\mathrm{Fr}^2 times the length of the experimental tank.

Image not extracted

Figure 6. (a) Top and (b) 3D views of the surface wake pattern for a representative case (Fr=1.5\mathrm{Fr}=1.5, Bo=3.7\mathrm{Bo}=3.7, Cs=0.01C_s=0.01), showing the difference between the leading dominant ridge (dashed), characterized by the opening angle α\alpha, and the wake arm (solid), characterized by the wake angle βΣ\beta_{\Sigma}. The markers (×) in (a) indicate the local extremum of the surface displacement field in each ridge, and the connecting line passes through these extrema across the successive ridges.

To quantitatively compare the theory with our experiments, we measure wake opening angles α\alpha from simulations for Fr∈[1,3]\mathrm{Fr}\in[1,3] and Cs∈[0.01,0.2]C_s\in[0.01,0.2]. In Figure 4 we see that the linear relation obtained using geometric consid-

Image not extracted

Figure 4. Plot of sin⁡α\sin\alpha as a function of cs/Uc_s/U for different Froude numbers Fr\mathrm{Fr}, obtained from simulations following (15). Blue curves: iso-Fr\mathrm{Fr} contours from simulations, with Fr\mathrm{Fr} indicated above each curve. Yellow squares: simulations run at the experimental conditions. Red circles: experimental observations for comparison. The experimental data points that correspond to different pulse speeds UU move across iso-Fr\mathrm{Fr} curves, such that Fr CsBo=UgLμρU2(ρgL2γ)1/4=[μ2ρgγ]1/4\mathrm{Fr}\,C_s\sqrt{\mathrm{Bo}} = \frac{U}{\sqrt{gL}}\sqrt{\frac{\mu}{\rho U^2}\left(\frac{\rho gL^2}{\gamma}\right)^{1/4}} = \left[\frac{\mu^2}{\rho g\gamma}\right]^{1/4} is constant. The inset shows the wake angle scales with the inverse of Fr\mathrm{Fr}, for both experiments and simulations.

erations persists, with sin⁡α=Acs/U\sin\alpha= Ac_s/U; simulations yield A≈3.5A \approx3.5, while experiments yield A≈4.5A \approx4.5. Similarly, the dependence on Fr\mathrm{Fr} follows a Mach-like law, sin⁡α=B/Fr\sin\alpha= B/\mathrm{Fr}, with B≈0.91B \approx0.91 from simulations and B≈1.1B \approx1.1 from experiments, showing reasonable agreement. We believe that the primary reason for the difference is finite size effects associated with not being strictly in the deep-water limit.

Our study of surface wakes on ultra-soft elastic surfaces uses experiments and theory to probe a previously unexplored regime where gravity, capillarity, and elastodynamics act together, while bringing together the classical subjects of Rayleigh waves and Kelvin wakes. Our experiments allow us to observe wakes behind a moving source as a function of its speed and our theory shows how the geometric simplicity of Mach-like wakes combines with the dispersive character of gravity–capillary waves. Furthermore, we see that elasticity leads to effective damping through radiative leakage, even in the absence of viscosity. All together, our approach also provides a quantitative foundation for probing the dynamics of ultra-compliant solid surfaces.

Acknowledgments. This work was supported in part by the Simons Foundation and the Henri Seydoux Fund. The computations in this paper were run on the FASRC FASSE cluster supported by the FAS Division of Science Research Computing Group at Harvard University.

References

End Matter

Radiation and Wake Angles

To characterize the nature and form of the wake, we examine the group speed of the wave packets generated by the dispersive system (13), since it determines where the energy of the generated waves focuses spatially. The scaled group speed, Cg=cg/UC_g=c_g/U, corresponding to eq. (13) is:

Cg(K):=∂Ω∂K=1+3K2/Bo2+8φ2(λ2)Fr2Cs2K2Fr2Ω.(16)C_g(K) := \frac{\partial\Omega}{\partial K} = \frac{1+3K^2/\mathrm{Bo}^2+8\varphi_2(\lambda^2)\mathrm{Fr}^2C_s^2K}{2\mathrm{Fr}^2\Omega}. \tag*{(16)}

In Figure 5a, we plot CgC_g and the phase speed Cp=Ω/KC_p=\Omega/K as a function of KK. Just as for classical gravity-capillary waves, there are two distinct branches: waves with K∈[K1(Cs),K∗]K \in[K_1(C_s),K^*] lie on the elasto-gravity branch, where Cg<CpC_g<C_p, characteristic of gravity waves, and those with K∈[K∗,K2(Cs)]K \in[K^*,K_2(C_s)] lie on the elasto-capillary branch, where Cg>CpC_g>C_p analogous to capillary waves.

In Figure 5b, we show the geometry of monochromatic waves associated with a single wavenumber generated by the moving source. As it moves from S−tS_{-t} to S0S_0, where tt denotes the elapsed time, circular wave fronts emanate from successive source locations on the horizontal axis. Waves generated earliest at −t-t travel the farthest at the phase speed cpc_p, and their constructive interference with the waves generated subsequently produces focusing of wave amplitudes along a particular line. For each admissible wavenumber in the range [K1(Cs),K2(Cs)][K_1(C_s),K_2(C_s)], this line is characterized by the radiation angle β(K)\beta(K) between the source velocity vector (−U)(-U) and the group velocity vector of the wave packet in the moving frame (cgk^−U)(c_g\hat{\mathbf{k}}-U). Following the geometric constructions in [27] [21], and considering the magnitudes and directions of the stationary phase [16], the radiation angle of the wave group of a given KK is:

tan⁡β(K)=Cg(K)1−Cp2(K)1−Cg(K)Cp(K).(17)\tan\beta(K)=\frac{C_g(K)\sqrt{1-C_p^2(K)}}{1-C_g(K)C_p(K)}. \tag*{(17)}

We plot β(K)\beta(K) in Figure 5c for different shear-wave speeds by varying CsC_s. Unlike pure gravity waves in deep water where the wake arm is bounded at the classical Kelvin angle, the radiation angle of free waves on ultra-soft solid spans all directions, similar to gravity-capillary waves. Wavenumbers on the elasto-gravity branch radiate behind source (β<90∘\beta<90^\circ), whereas those on the elasto-capillary branch radiate in arbitrary directions including ahead of the source (β>90∘\beta>90^\circ). Increasing elasticity increases the radiation angle of every allowable wavenumber. However, the larger wavenumbers, which radiate at larger angles, are strongly damped due to elasticity, and thus appear less prominently in the wake pattern.

The constructive interference of the individual wave packets of different wavenumbers, each radiating in different angles β(K)\beta(K), generates the wake pattern with multiple ridges behind the source (see Figure 5d). The elevation peaks along these ridges determine the observed wake angle βΣ\beta_{\Sigma}, which is different from the opening angle of the leading dominant ridge α\alpha. This distinction is shown for a representative set of parameters in Figure 6. We note that spatial damping due to elasticity makes the observation and measurement of βΣ\beta_{\Sigma} in the experiments increasingly difficult for larger CsC_s, thus driving our choice to measure α\alpha rather than the entire wake pattern.

References

  1. [4][4–7]
  2. [8][8–10]
  3. [9]Y. Onodera and P.-K. Choi, The Journal of the Acoustical Society of America 104, 3358 (1998).
  4. [10]X. Shao, J. R. Saylor, and J. B. Bostwick, Soft Matter 14, 7347 (2018).
  5. [12]X. Jacob, S. Catheline, J.-L. Gennisson, C. Barrière, D. Royer, and M. Fink, The Journal of the Acoustical Society of America 122, 1917 (2007).
  6. [14]J. Harden, H. Pleiner, and P. Pincus, The Journal of Chemical Physics 94, 5208 (1991).
  7. [17]Lamb, H. (1932). Hydrodynamics, sixth ed.. In Cambridge University Press.
  8. [18]Lighthill, J. (1978). Waves in fluids. In Cambridge univ.
  9. [19]M. Rabaud and F. Moisy, Physical Review Letters 110, 214503 (2013).
  10. [20]R. Pethiyagoda, T. J. Moroney, C. J. Lustri, and S. W. McCue, Journal of Fluid Mechanics 915, A126 (2021).
  11. [22]T. H. Havelock, Proceedings of the Royal Society of London. Series A 81, 398 (1908).
  12. [23]Landau, L. D., & Lifshitz, E. M. (1987). Fluid Mechanics, Course of Theoretical Physics, Vol. 6. In Butterworth-Heinemann.
  13. [24]Setting $y = \Omega^2/K^2C_s^2$ we can rewrite this relation as $y(y^3 - 8y^2 + 24y - 16) = 0$, with a single non-trivial real root $y \approx 0.912$, i.e. $\Omega/KC_s \approx 0.955$.
  14. [25]For comparison, a solid with much larger modulus, say $\mu \gg 10^2 Pa$, can support both fast and slow modes.
  15. [26]The half-angle $\alpha$ defines the leading dominant ridge, which was visible and measured in the experiments; this is distinct from the wake angle based on $\beta$ given by {xref}`(17)`, but that was hard to observe.
  16. [27]F. S. Crawford, American Journal of Physics 52, 782 (1984).

Paper details

vlnc:W_jc55fw49
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