betaFeedback

WorksPhysics and Astronomy

Bulldozing an immersed granular material in a confined channel

Abstract

The motion of an immersed granular material in a channel is characterised by complex interactions among the grains, between the grains and the permeating liquid, and between the grains and the channel walls. Here, we develop a reduced-order continuum model for the bulldozing of an immersed, sedimented granular material by a piston in a channel. In our continuum approach, the granular pile and the overlying fluid layer evolve as a system of coupled thin films. We model the granular phase as a dense, porous, visco-plastic material that experiences Coulomb-like friction with the walls. Conservation of mass and momentum under a thin-film approximation leads to an elliptic equation for the velocity of the grains that is coupled with an evolution equation for the height of the granular pile. We solve our model numerically for a variety of different scenarios to explore the interactions between wall friction, internal viscous-like stresses, and fluid flow above and through the pile. We complement our numerical results with a series of experiments that provide insight into the validity and limitations of the model.

Banner appropriate to article type will appear here in typeset article

Introduction

Depending on the confining stress, granular materials can respond mechanically at the continuum scale as a solid, liquid, or gas [21]. Granular flows are relevant to a wide variety of natural processes, including avalanches [14, 41], terrestrial and submarine landslides [24, 35], blood flows through small vessels [22], and debris flows [27], as well as to a variety of industrial applications. The continuum rheology of granular materials has been studied extensively, primarily focusing on internal friction and thus on relative motion and sliding within the granular material. For example, many studies consider scenarios in which a granular material is confined and sheared between a fixed plate or cylinder and a moving one [3, 25, 37, 40], whereas classical bedload transport involves the shearing of an unconfined granular material by an overlying fluid flow [23]. Avalanche-type flows have also attracted substantial attention, in which an unconfined granular material flows downhill under gravity [10, 12, 14, 21, 41]. Finally, Sauret et al. [50] studied the unconfined pile-up of a granular material during bulldozing by a vertical plate on a rotating table, focusing on the initial growth of the granular pile and then on its long-time shape.

Additional challenges arise in confined granular-flow problems where the grains are able to slide along the confining walls. The flow of sand through an hourglass is a relatable example; similar interactions occur within subsurface faults and fractures [15] and in silos [32] (Janssen 1895). This class of problems has received less attention. Some authors have considered the impact of the confining vertical side walls on avalanche-type flows in channels [29, 51, 2]. Simulations and experiments show that these walls can have a significant impact on the flow, potentially changing the convexity of the flow profile if the channel is sufficiently narrow. Marks et al. [36] used the discrete-element method (DEM) to simulate the piston-driven bulldozing of dry grains between horizontal top and bottom plates in two spatial dimensions. They found that, once the grains have piled up to bridge the gap between the two plates, the bulldozing force grows exponentially with the downstream length of the bridged region due to sliding friction between the grains and the plates, which is consistent with the classical Janssen effect (Janssen 1895). Morgan et al. [39] performed experiments to study the motion of granular material along a confined, quasi-2D channel where the motion of the grains is driven by a net flow of liquid through the channel. They observed that the velocity profile of the grains was non-monotonic in the vertical direction.

Additional inspiration for the present study comes from a series of experiments involving the bulldozing of grains by capillary forces. This problem has been considered in both one- and two-dimensional geometries. In the one-dimensional case, air was injected into a capillary tube containing a wet granular suspension (Dumazez et al. 2016, 2020). Because the air is non-wetting to the grains, the liquid–air meniscus acts to bulldoze the grains like a piston. When the grains pile-up to bridge the gap, the granular plug becomes jammed and the gas phase eventually percolates through the static plug and emerges on the other side, resulting in a repeating pattern of granular plugs alternating with gas regions. In the two-dimensional case of a Hele–Shaw cell, the system has an additional degree of freedom [20, 49, 48, 56]. The gas always pushes the granular material in the direction of least resistance, resulting in the growth of a fingering pattern with a morphology that is distinct from the traditional viscous fingers that would form in the absence of the grains [42, 47]. It has been shown that ad hoc, rule-based simulations can reproduce these patterns (Eriksen et al. 2015b,a; Knudsen et al. 2008), but a true quasi-1D or quasi-2D continuum model derived from the fundamental mechanics of the problem is still lacking.

Here, we develop a continuum model to describe the bulldozing of an immersed granular material in a confined, horizontal geometry. We then use this model to explore the interactions between fluid flow, granular motion and pile-up, and sliding friction. To do so, we focus on the scenario illustrated in Figure 1, where a thin gap between two parallel plates is filled with solid grains and a fluid. We take the grains to be denser than the fluid, such that they sediment in a uniform layer on the bottom plate. A piston then bulldozes both grains and fluid from left to right at a constant velocity. This problem is a generalisation of the work by Marks et al. [36], who considered the limiting case where the fluid phase is ignored. We formulate a two-phase, two-layer model comprising a single Newtonian fluid in the top layer and an immersed granular material (grains and fluid) in the bottom layer, such that depth-averaging leads to a system of coupled thin-film flows. For the granular phase, we adopt the established μ(I)\mu(I)-type rheology (Jop et al. [29]; Pouliquen et al. [44]; Cassar et al. [12]), where the granular continuum behaves like a yield-stress fluid with a pressure-dependent yield stress and a rate-dependent friction coefficient μ\mu. We consider both the inertial μ(I)\mu(I) rheology (Midi [38]; Da Cruz et al. [13]; Jop et al. [29]) and the viscous μ(Iv)\mu(I_v) rheology (Cassar et al. [12]; Boyer et al. [11]), which are distinguished by whether the viscosity of the interstitial fluid is expected to play an unimportant role or a dominant role in the mechanics, respectively. Further, we impose simple Coulomb-like friction at the top and bottom walls, such that the grains will only slide along the walls if the local shear stress is sufficiently large relative to the local normal stress (Pennestrì et al. [43]).

Figure 1. A schematic of granular bulldozing in a confined, horizontal channel. (a) The system consists of two parallel plates separated by a gap bb. The gap is initially filled with a mixture of fluid and grains, with the grains forming a uniform, fluid-saturated layer of height h0h_0 on the bottom plate. (b) A piston moves from left to right with constant velocity δ˙\dot{\delta}, bulldozing both grains and fluid. The grains pile up near the piston and eventually bridge the gap between the plates. (c) As bulldozing continues, the domain comprises three regions: a bridged region where h=bh=b, a wedge-shaped transition region where h0<h<bh_0<h<b, and an undisturbed region where h=h0h=h_0 and the solid is stationary.

The remainder of this work is organised as follows. In § 2, we derive our model based on the approach described above. As previously mentioned, the interactions between the grains and the walls are governed by Coulomb friction. This model gives an expression for the shear stress at the walls only if the magnitude of the shear stress is sufficiently large, which is not known a priori. To address this difficulty, we introduce a novel regularisation that gives an expression for the shear stress over the entire computational domain. In §3 and §4, we present a numerical study of our model. In §3, we consider the inertial (‘dry’) μ(I)\mu(I) rheology where the viscosity of the fluid phase is negligible. We investigate the granular velocity profile in the region where the grains have bridged the gap, as well as in the wedge-shaped transition region (see Figure 1b,Figure 1c). In §4 and § 5, we consider the viscous (‘wet’) μ(Iv)\mu(I_v) rheology and focus on the influence of the fluid on the grains, comparing results from our model with laboratory experiments. Finally, in § 6, we summarise our findings, comment on the validity of our modelling assumptions, and suggest avenues for future work.

Model development

We consider the piston-driven bulldozing of a dense, fluid-saturated granular material in a confined horizontal channel. The channel consists of a gap of height bb between two parallel, horizontal plates. The gap is filled with an incompressible Newtonian fluid and a granular material with mean grain diameter dd. The solid grains are negatively buoyant and form a layer of height h(x,t)h(x,t) on the bottom plate. The granular layer has a uniform initial height h(x,0)=h0h(x,0)=h_0 [Figure 1a]. A rigid piston located at x=δ(t)x=\delta(t) travels from left to right (i.e., in the positive xx-direction) at a constant speed δ˙\dot{\delta}, bulldozing both grains and fluid [Figure 1(b) and (c)]. Note that we ignore motion in the yy-direction (i.e., into the page).

We denote the local volume fractions of the fluid and solid phases as ϕf\phi_f and ϕs\phi_s, respectively, such that ϕf+ϕs=1\phi_f+\phi_s=1. For simplicity, we take ϕs\phi_s to be uniform and equal to a constant ϕs∗\phi_s^* within the granular layer and zero outside the granular layer. Thus, we neglect elastic compression of the grains and shear-induced compaction and dilation of the granular layer in order to focus on the physical interactions of internal friction, pile-up, wall friction, and fluid flow. The solid-fraction distribution is then given by

ϕs(x,z,t)={ϕs∗,for 0≤z≤h,0,for h<z≤b.(1)\phi_s(x,z,t)= \begin{cases} \phi_s^*, & \text{for } 0\le z\le h,\\ 0, & \text{for } h<z\le b. \end{cases} \tag*{(1)}

We denote the fluid velocity by vf=uf(x,z,t)ex+wf(x,z,t)ez\mathbf{v}_f=u_f(x,z,t)\mathbf{e}_x+w_f(x,z,t)\mathbf{e}_z and the solid velocity by vs=us(x,z,t)ex+ws(x,z,t)ez\mathbf{v}_s=u_s(x,z,t)\mathbf{e}_x+w_s(x,z,t)\mathbf{e}_z, with ex\mathbf{e}_x and ez\mathbf{e}_z the unit vectors in the xx- and zz-directions, respectively. The volume-averaged velocity is then v=u(x,z,t)ex+w(x,z,t)ez=ϕfvf+ϕsvs\mathbf{v}=u(x,z,t)\mathbf{e}_x+w(x,z,t)\mathbf{e}_z=\phi_f\mathbf{v}_f+\phi_s\mathbf{v}_s. Conservation of mass for the solid phase can then be written in vertically integrated form as

∂h∂t+∂∂x(uˉsh)=0,(2)\frac{\partial h}{\partial t}+\frac{\partial}{\partial x}(\bar{u}_s h)=0, \tag*{(2)}

where

uˉs=1h∫0hus dz(3)\bar{u}_s=\frac{1}{h}\int_0^h u_s\,\mathrm{d}z \tag*{(3)}

is the average horizontal velocity of the solid. Similarly, conservation of total volume requires that the average horizontal velocity on any vertical slice must be equal to the piston speed,

1b∫0bu dz=δ˙.(4)\frac{1}{b}\int_0^b u\,\mathrm{d}z=\dot{\delta}. \tag*{(4)}

Before discussing the model in detail, we briefly consider the anticipated structure of the flow, as sketched in Figure 1. We expect the flow to be predominantly horizontal and quasi-static, provided, as discussed in § 2.4 below, that the macroscopic Froude number and fluid Reynolds number are sufficiently small that friction and viscosity dominate the momentum balance. As noted above, the granular layer has a frictional yield strength. Thus, sufficiently far ahead of the piston, the grains must be undisturbed at their original height h0h_0. Nonetheless, the motion of the piston dictates that there must be a net rightward flow of material at all values of xx [eq. (4)]; in the far field, we therefore expect a net fluid flow both through and above the grains. Near the piston we expect grains to pile up and “bridge” the gap [Figure 1b], leading to a translating ‘transition’ region that links the bridged near-piston region with the undisturbed far-field region [Figure 1c]. Once formed, the bridged region must grow steadily in length as grains pile up ahead of the piston. The transition region can then either grow continually in length (if its leading edge travels faster than its trailing edge, both of which must travel faster than the piston) or evolve toward a steady shape that translates like a travelling wave (if its leading and trailing edges travel at the same speed, again faster than the piston). In the latter scenario, conservation of solid volume dictates that the transition region must travel at a speed bδ˙/(b−h0)>δ˙b\dot{\delta}/(b-h_0)>\dot{\delta}. We observe both of these scenarios in the results below. An alternative scenario, also observed in some cases (§4), is that the grains fail to displace all of the viscous fluid from the overlying fluid layer, and thus never fully bridge the gap. In this scenario, there is no bridged region and the transition region must grow continually.

Mechanics

We focus on the limit in which the channel is sufficiently thin, or the piston sufficiently slow, that the inertia of the fluid–solid mixture is unimportant in the macroscopic momentum balance. In this limit, which we quantify more formally in § 2.4, conservation of linear momentum can be written

∇⋅σ=ρgez,(5)\nabla\cdot\boldsymbol{\sigma}=\rho g\mathbf{e}_z, \tag*{(5)}

where σ\boldsymbol{\sigma} is the total stress in the mixture and ρ=ϕfρf+ρsϕs\rho=\phi_f\rho_f+\rho_s\phi_s is the local volume-averaged density of the mixture, and where we have adopted the convention that tensile stresses are positive. Going forward, we assume that the ratio of typical vertical to horizontal length scales is sufficiently small that we can adopt a lubrication-type approximation, in which w≪uw\ll u and ∂x≪∂z\partial_x\ll\partial_z. We discuss the validity of these assumptions in § 5 and § 6.

Upper layer (fluid)

The upper layer, h<z<bh<z<b, comprises a single incompressible Newtonian fluid of viscosity ηf\eta_f, for which the lubrication approximation suggests that the fluid pressure pfp_f is hydrostatic,

pf(x,z,t)=Pf(x,t)+ρfg(b−z),p_f(x,z,t)=P_f(x,t)+\rho_f g(b-z),

such that

∂σxz∂z=∂Pf∂xwithσxz=ηf∂uf∂z,(6)\frac{\partial\sigma_{xz}}{\partial z}=\frac{\partial P_f}{\partial x}\quad\text{with}\quad\sigma_{xz}=\eta_f\frac{\partial u_f}{\partial z}, \tag*{(6)}

where PfP_f is the unknown fluid pressure at z=bz=b. We impose a no-slip boundary condition on the flow at the top wall, uf=0u_f=0 on z=bz=b, as is standard. The appropriate boundary condition at the interface z=hz=h is less obvious, but, for simplicity, we follow previous models of bedload transport and impose no tangential slip between the overlying fluid and the granular solid by requiring that uf=usu_f=u_s at z=h+z=h^+ ([e.g. 3]). Note that this choice does not constrain the fluid velocity on the granular side of the interface, where in general we expect a nonzero Darcy flow (i.e., fluid flow relative to the solid grains), and neither does it prevent fluid flow across the interface. Note also that, in reality, some degree of tangential slip is expected at the boundary between an external fluid flow and a porous medium ([e.g. 34]). Many associated slip boundary conditions have been proposed, most famously the so-called Beavers–Joseph condition ([9, 34]). We revisit this point briefly below, at the end of § 2.2.

Imposing these two no-slip conditions, we arrive at an expression for the velocity of the fluid in the upper layer:

uf(x,z,t)=(b−z)(h−z)2ηf∂Pf∂x+b−zb−hus(z=h).(7)u_f(x,z,t)=\frac{(b-z)(h-z)}{2\eta_f}\frac{\partial P_f}{\partial x}+\frac{b-z}{b-h}u_s(z=h). \tag*{(7)}

The shear stress exerted on the granular layer by this flow is then

σxz(z=h)=ηf[h−b2ηf∂Pf∂x−us(z=h)b−h].(8)\sigma_{xz}(z=h)=\eta_f\left[\frac{h-b}{2\eta_f}\frac{\partial P_f}{\partial x}-\frac{u_s(z=h)}{b-h}\right]. \tag*{(8)}

Equation (7) suggests that the flow in the upper layer is a Poiseuille–Couette flow, combining a parabolic profile from the background pressure gradient with a linear shear profile from the motion of the granular layer relative to the top wall.

Lower layer (fluid–solid mixture)

For the lower layer, we adopt Terzaghi’s principle [53] by decomposing the total stress into the fluid pressure and the effective solid stress σ′\boldsymbol{\sigma}',

σ=σ′−pfI.(9)\boldsymbol{\sigma} = \boldsymbol{\sigma}' - p_f \boldsymbol{I}. \tag*{(9)}

Following Jop et al. (2006), we then decompose σ′\boldsymbol{\sigma}' into its isotropic and deviatoric components,

σ′=−psI+τ,(10)\boldsymbol{\sigma}' = -p_s \boldsymbol{I} + \boldsymbol{\tau}, \tag*{(10)}

where ps=−(1/3)tr⁡(σ′)p_s = -(1/3)\operatorname{tr}(\boldsymbol{\sigma}') and tr⁡(τ)=0\operatorname{tr}(\boldsymbol{\tau}) = 0. Further, we impose that the fluid flows through the porous, granular material according to Darcy’s law,

(1−ϕs∗)(vf−vs)=−kηf(∇pf−ρfgez).(11)(1-\phi_s^*) (\boldsymbol{v}_f-\boldsymbol{v}_s) = -\frac{k}{\eta_f}\left(\nabla p_f-\rho_f g\boldsymbol{e}_z\right). \tag*{(11)}

where kk is the permeability of the granular skeleton, which we expect to scale with the square of the grain size: k∼d2k \sim d^2.

Combining equations (5), (9), and (10) under the lubrication approximation, we arrive at

∂∂z(ps+pf)=−ρgand∂σxz′∂z=∂∂x(ps+pf),(12)\frac{\partial}{\partial z}(p_s+p_f)=-\rho g \qquad\text{and} \qquad\frac{\partial\sigma'_{xz}}{\partial z}=\frac{\partial}{\partial x}(p_s+p_f), \tag*{(12)}

along with σxx′=σzz′=−ps\sigma'_{xx}=\sigma'_{zz}=-p_s. The vertical component of Equation (11) suggests that the fluid pressure is hydrostatic; further requiring that it be continuous across z=hz=h leads to pf=Pf+ρfg(b−z)p_f=P_f+\rho_f g(b-z). It then follows that ps=Ps(x,t)+ϕs∗Δρg(h−z)p_s=P_s(x,t)+\phi_s^*\Delta\rho g(h-z), where Δρ=ρs−ρf\Delta\rho=\rho_s-\rho_f and PsP_s is the as-yet-unspecified solid pressure at the interface z=hz=h. As such,

σxz′=σxz=σxz(z=h)−(h−z)(∂P∂x+ϕs∗Δρg∂h∂x),(13)\sigma'_{xz}=\sigma_{xz}=\sigma_{xz}(z=h)-(h-z)\left(\frac{\partial P}{\partial x}+\phi_s^*\Delta\rho g\frac{\partial h}{\partial x}\right), \tag*{(13)}

where P=Pf+PsP=P_f+P_s. Further, the horizontal component of equation (11) becomes

(1−ϕs∗)(uf−us)=−kηf∂Pf∂x.(14)(1-\phi_s^*)(u_f-u_s)=-\frac{k}{\eta_f}\frac{\partial P_f}{\partial x}. \tag*{(14)}

Note that continuity of normal stress requires that PsP_s must vanish when the solid does not bridge the gap (i.e., when h<bh<b), such that Ps≠0P_s\ne0 only in the bridged region (i.e., where h≡bh\equiv b). Correspondingly, conservation of mass requires that the fluid and solid in the bridged region move at the same speed, and thus ∂xPf=0\partial_xP_f=0 in the bridged region.

Imposing no tangential slip between the granular solid and the fluid in the upper layer, as we did above, implies a jump in tangential fluid velocity between the free flow in the upper layer and the Darcy flow in the lower layer (i.e., across the interface). In reality, we anticipate a narrow transition region at the interface (on the scale of one grain/pore, or ∼d∼k\sim d\sim\sqrt{k}) over which the fluid velocity smoothly adjusts (see, e.g., Beavers & Joseph 1967; Le Bars & Worster 2006). We expect the difference between applying a no-slip condition and a more elaborate slip condition to scale with k\sqrt{k}, which is typically extremely small. As noted above, this boundary condition does not constrain flow across the interface; rather, the normal component of the fluid velocity is dictated by conservation of mass, as is standard in lubrication problems.

Note that the incomplete model above already provides some insight into the flow in the far field, where the grains are undisturbed at uniform depth h0h_0 and the fluid flows both above them and through them. For stationary grains (us=0u_s = 0), the flow in the upper layer is a simple Poiseuille flow (eq. (7)) and the uniform Darcy flow through the lower layer follows from eq. (14). The total flow rate is δ˙b\dot{\delta}b, as determined by the piston, with a fraction (b−h0)3/[(b−h0)3+12h0k](b-h_0)^3/[(b-h_0)^3+12h_0k] flowing in the upper layer and a fraction 12h0k/[(b−h0)3+12h0k]12h_0k/[(b-h_0)^3+12h_0k] flowing through the grains in the lower layer.

Granular friction and wall friction

As noted above, we adopt a μ(I)\mu(I)-type rheology for the internal mechanics of the granular material [29, 44, 12]. In these model constructions, the internal mechanical interactions between grains are taken to obey a Coulomb-like friction model,

∣σxz′∣=μpsif∣σxz′∣≥μsps,(15)\lvert\sigma'_{xz}\rvert= \mu p_s \quad\text{if} \quad\lvert\sigma'_{xz}\rvert\ge\mu_s p_s, \tag*{(15)}
∂us∂z=0otherwise,(16)\frac{\partial u_s}{\partial z}=0 \quad\text{otherwise}, \tag*{(16)}

where μ\mu is a dimensionless, rate-dependent coefficient of kinetic friction, as discussed further in § 2.3, and μs\mu_s is a dimensionless constant coefficient of static friction. If the magnitude of the shear stress is sufficiently small, equation (16) indicates that grains will resist internal rearrangement and the solid phase will act locally as a rigid body. If the shear stress is sufficiently large, the solid phase will yield such that ∣∂zus∣>0\lvert\partial_z u_s\rvert> 0 and the solid velocity will satisfy equation (15), with the sign of σxz′\sigma'_{xz} given by the sign of the shear rate ∂zus\partial_z u_s. As a result, the vertical profile us(z)u_s(z) may feature multiple coupled yielded and un-yielded regions separated by free boundaries. We consider the various possible flow scenarios in § 2.2.2 below. We specify the granular rheology in more detail in § 2.3.

At the bottom surface of the granular layer, we assume that the mechanical interactions between the grains and the wall obey a Coulomb-like friction model:

σxz′(z=0)≤μwps(z=0),(17)\sigma'_{xz}(z=0) \le\mu_w p_s(z=0), \tag*{(17)}

such that

us(z=0)=0ifσxz′(z=0)<μwps(z=0),(18)u_s(z=0)=0 \quad\text{if} \quad\sigma'_{xz}(z=0)<\mu_w p_s(z=0), \tag*{(18)}
us(z=0)>0ifσxz′(z=0)=μwps(z=0),(19)u_s(z=0)>0 \quad\text{if} \quad\sigma'_{xz}(z=0)=\mu_w p_s(z=0), \tag*{(19)}

where μw\mu_w is the dimensionless constant coefficient of friction between the grains and the wall. Note that we assume that us≥0u_s \ge0, such that σxz′(z=0)≥0\sigma'_{xz}(z=0)\ge0 and, thus, that friction resists relative motion. To generate a single, continuous problem in xx, we unify these conditions in a regularised form in § 2.4.

The appropriate boundary condition at the top surface of the granular layer depends on whether or not the grains are in contact with the top wall. If the grains have bridged the gap (i.e., if h=bh=b), then we similarly apply Coulomb friction at the top wall:

σxz′(z=b)≥−μwps(z=b)ifh=b,(20)\sigma'_{xz}(z=b)\ge-\mu_w p_s(z=b) \quad\text{if} \quad h=b, \tag*{(20)}

such that

us(z=h)=0ifσxz′(z=b)≥−μwps(z=b)andh=b,(21)u_s(z=h)=0 \quad\text{if} \quad\sigma'_{xz}(z=b)\ge-\mu_w p_s(z=b) \quad\text{and} \quad h=b, \tag*{(21)}
us(z=h)>0ifσxz′(z=b)=−μwps(z=b)andh=b,(22)u_s(z=h)>0 \quad\text{if} \quad\sigma'_{xz}(z=b)=-\mu_w p_s(z=b) \quad\text{and} \quad h=b, \tag*{(22)}

where the differences in sign relative to equations (17)–(19) reflect the different direction of the normal vector, again ensuring that friction resists relative motion for us≥0u_s \ge0. If the grains have not bridged the gap between the plates (i.e., if h<bh < b), then we instead impose that the shear stress is continuous with the upper fluid layer via equation (8). To generate a single, continuous problem in xx, we also unify these conditions in a regularised form in §2.3.

Velocity profile

Before giving specific details of the granular rheology, we first present in Figure 2 and Figure 3 a selection of example velocity profiles that can occur in the transition zone (h<bh < b) and in the bridged region (h≡bh \equiv b), respectively. In both cases, the qualitative nature of the profile depends centrally on whether the coefficient of wall friction, μw\mu_w, is larger or smaller than the static coefficient of internal friction, μs\mu_s.

Figure 2. Examples of the different possible stress and velocity configurations in the unbridged state (h<bh < b and Ps≡0P_s \equiv0), neglecting the fluid phase (∂xpf=0\partial_x p_f = 0 and σxz′(z=h)=0\sigma'_{xz}(z = h) = 0). (a) Possible shear-stress profiles σxz′(z)\sigma'_{xz}(z) (red and blue lines), which must be linear from eq. (13), together with the effective yield stress of the granular layer ±μsps\pm\mu_s p_s, where ps=ϕsgΔρ(h−z)p_s = \phi_s g \Delta\rho(h-z) (solid yellow); the shear stress lies below the yield stress in the shaded region. Different colours correspond to whether the wall friction is less than the internal static friction, μw=μw1<μs\mu_w = \mu_{w1} < \mu_s (blue), or greater, μw=μw2>μs\mu_w = \mu_{w2} > \mu_s (red), with corresponding velocity profiles shown in (b,c) and (d,e), respectively. In each case, there are two possibilities: the shear stress on the bottom wall does not reach the slip stress, ∣σxz′(z=0)∣<μwps(0)|\sigma'_{xz}(z=0)| < \mu_w p_s(0), and the grains do not slip there (b,d), or it does and they do (c,e). Where the shear stress is below the internal yield stress, ∣σxz′∣<μsps|\sigma'_{xz}| < \mu_s p_s, the granular layer behaves as a rigid plug (b,c).

Figure 3. Examples of the different possible stress and velocity configurations in the bridged state (h≡bh \equiv b, Ps>0P_s > 0, and σxz′(z=h)<0\sigma'_{xz}(z=h) < 0), neglecting the fluid phase (∂xpf=0\partial_x p_f = 0). (a) Possible shear-stress profiles σxz′(z)\sigma'_{xz}(z) (red and blue lines), together with the effective yield stress of the granular layer ±μsps\pm\mu_s p_s, where ps=Ps+ϕsgΔρ(b−z)p_s = P_s + \phi_s g\Delta\rho(b-z) (solid yellow); the shear stress lies below the yield stress in the shaded region. If μw=μw1<μs\mu_w = \mu_{w1} < \mu_s (blue), the granular layer does not yield anywhere and must move as a rigid plug (b). If μw=μw1>μs\mu_w = \mu_{w1} > \mu_s (red), different flow profiles are possible depending on whether the shear stress attains the slip stress on either the upper or lower wall, as sketched in (c,d,e), each corresponding to the equivalent line style in (a).

We first consider the unbridged portions of the granular layer (i.e., the transition region and the far field) in Figure 2, for which h<bh < b and Ps≡0P_s \equiv0. For simplicity, we neglect the role of the fluid by taking σxz′(z=h)=0\sigma'_{xz}(z = h) = 0 and ∂xpf=0\partial_x p_f = 0. Recalling that the shear stress at any horizontal location increases linearly with depth (eq. (13)), we find that four types of qualitative profile arise. When μw<μs\mu_w < \mu_s (blue curves), the shear stress can never grow large enough for the granular layer to yield internally: ∣σxz′∣<μsps|\sigma'_{xz}| < \mu_s p_s and ∣∂zus∣≡0|\partial_z u_s| \equiv0 for all zz. As a result, the grains do not deform and either remain stationary (Figure 2b) or move as a rigid body (Figure 2c), depending on whether σxz′(z=0)\sigma'_{xz}(z = 0) is less than, or equal to, μwps(z=0)\mu_w p_s(z = 0), respectively. By contrast, when μw>μs\mu_w > \mu_s (red curves), the granular layer yields over its entire thickness, again with either no slip (Figure 2d) or slip (Figure 2e) on the base, depending on the size of the basal stress. Note that analogous profiles have been observed in the case of viscoplastic fluids with wall slip [33].

We next consider the bridged portions of the granular layer, in Figure 3, for which h≡0h \equiv0, Ps>0P_s > 0, and σxz′(z=h=b)\sigma'_{xz}(z = h = b) is dictated by friction with the top wall. Note that we again neglect the role of the fluid. As in the unbridged portions (Figure 2), there are different cases depending on whether μw<μs\mu_w < \mu_s or μw>μs\mu_w > \mu_s. In the former case (blue, Figure 3b), the grains remain everywhere unyielded and slide over both boundaries as a rigid plug (because of the moving piston, the bridged region cannot remain stationary). If μw>μs\mu_w > \mu_s (red, Figure 3c,Figure 3d,Figure 3e), there are more possibilities. The grains may be able to slip, or not, on both the upper and lower walls depending on the local shear and normal stresses, and in all cases there may be some unyielded interior region where the shear stress falls below the effective yield stress (Figure 3a). Across the entire parameter space, we expect the solid velocity to be smaller near the bottom wall than near the top wall because psp_s increases from top to bottom due to gravity.

To accommodate the scenarios shown in Figure 2 and Figure 3, we seek in general a piecewise solid velocity of the form

us={us1(x,z,t),if 0≤z≤hy1,us2(x,t),if hy1<z<hy2,us3(x,z,t),if hy2≤z≤h,(23)u_s = \begin{cases} u_{s1}(x,z,t), & \text{if } 0 \le z \le h_{y1}, \\ u_{s2}(x,t), & \text{if } h_{y1} < z < h_{y2}, \\ u_{s3}(x,z,t), & \text{if } h_{y2} \le z \le h, \end{cases} \tag*{(23)}

such that the granular layer is unyielded (∂zus=0\partial_z u_s = 0) in an interior region hy1<z<hy2h_{y1} < z < h_{y2} and yielded (∣∂zus∣>0|\partial_z u_s| > 0) outside of that region, with continuity of velocity requiring that us1=us2u_{s1} = u_{s2} at z=hy1z = h_{y1} and us2=us3u_{s2} = u_{s3} at z=hy2z = h_{y2}. The locations hy1h_{y1} and hy2h_{y2} are determined by solving σxz′=±μsps\sigma'_{xz} = \pm\mu_s p_s to give

hy1=h−σxz′(z=h)−μsPs∂xP+ϕs∗Δρg(∂xh+μs)andhy2=h−σxz′(z=h)+μsPs∂xP+ϕs∗Δρg(∂xh−μs).(24)h_{y1} = h - \frac{\sigma'_{xz}(z = h) - \mu_s P_s}{\partial_x P + \phi_s^* \Delta\rho g(\partial_x h + \mu_s)} \quad\text{and}\quad h_{y2} = h - \frac{\sigma'_{xz}(z = h) + \mu_s P_s}{\partial_x P + \phi_s^* \Delta\rho g(\partial_x h - \mu_s)}. \tag*{(24)}

In the yielded regions (0<z<hy10 < z < h_{y1} and hy2<z<hh_{y2} < z < h), we combine equations (2.13) and (2.15) with the appropriate granular rheology from § 2.3 to obtain ordinary differential equations (ODEs) for us1u_{s1} and us3u_{s3}.

Granular Rheology

In the region where the granular layer has yielded, ∣σxz′∣>μsps|\sigma'_{xz}| > \mu_s p_s, we adopt a μ(I)\mu(I)-type rheology to describe its evolution [29, 44, 12]. These rheologies relate the effective viscosity of the solid phase, μ\mu, to either the inertial number, II, or the viscous number, IvI_v, both of which are defined below. The specific choice depends on the magnitude of the Stokes number, St∼I/Iv∼ρsd2∣∂zus∣/ηf\mathrm{St} \sim I/I_v \sim\rho_s d^2|\partial_z u_s|/\eta_f, which compares inertial forces to viscous forces on the scale of the grains. Viscous drag from the fluid is expected to dominate inertia at the grain scale for St≪1\mathrm{St} \ll1, whereas grain-scale inertia is expected to be dominant for St≫1\mathrm{St} \gg1. As such, different rheologies are appropriate in these two regimes. Some effort has been made to combine the two regimes into a single rheology [11, 8, 46, 52], but we treat them separately for simplicity.

In §3, we consider the dry case where St≫1\mathrm{St} \gg1, in which limit the fluid phase is negligible. In this case, we follow Iordanoff & Khonsari [26] and Da Cruz et al. [13] by taking

μ(I)=μs+(μ2−μsI+I0)I,(25)\mu(I) = \mu_s + \left(\frac{\mu_2-\mu_s}{I+I_0}\right)I, \tag*{(25)}

where I0I_0 and μ2\mu_2 are constants. The inertial number II is

I=d2ρsps∣∂us∂z∣,(26)I = \sqrt{\frac{d^2\rho_s}{p_s}\left|\frac{\partial u_s}{\partial z}\right|}, \tag*{(26)}

where, under the lubrication approximation, ∣∂us/∂z∣\left|\partial u_s/\partial z\right| is the leading-order contribution to the magnitude of the rate-of-strain tensor. Equation (25) is an empirical fit to experiments and numerical simulations, and is defined such that μ→μs\mu\to\mu_s for I≪I0I \ll I_0 and μ→μ2\mu\to\mu_2 for I≫I0I \gg I_0.

In §4, we consider the wet case where St≪1\mathrm{St} \ll1, in which limit fluid viscosity dominates granular inertia. In this case, we follow Boyer et al. [11] by taking

μ(Iv)=μs+(μ2−μsIv+I0)Iv+Iv+52ϕs∗Iv,(27)\mu(I_v) = \mu_s + \left(\frac{\mu_2-\mu_s}{I_v+I_0}\right)I_v + I_v + \frac{5}{2}\phi_s^*\sqrt{I_v}, \tag*{(27)}

where I0I_0 and μ2\mu_2 are again constants, the appropriate values of which are not necessarily the same as in equation (25). The viscous number IvI_v is

Iv=ηfps∣∂us∂z∣,(28)I_v = \frac{\eta_f}{p_s}\left|\frac{\partial u_s}{\partial z}\right|, \tag*{(28)}

having again taken ∣∂zus∣\left|\partial_z u_s\right| to be the leading-order contribution to the magnitude of the rate-of-strain tensor. The last two terms on the right-hand side of equation (27) mean that μ(Iv)\mu(I_v) will increase without bound as IvI_v increases.

By substituting either equations (25) and (26) or equations (27) and (28) into the frictional-rheology model in equation (15), we arrive at a non-linear ODE for usu_s with boundary conditions described in more detail below.

Note that both the μ(I)\mu(I) and the μ(Iv)\mu(I_v) models presented here have been shown to be mathematically ill-posed in their general, time-dependent, tensorial forms [6, 5]; specifically, they can be unstable to arbitrarily small-wavelength perturbations for some ranges of II and IvI_v. This problem can be resolved by carefully accounting for compressibility in each case [7, 5]. In the present lubrication framework, the momentum balance (eq. 2.12) is quasi-steady, which avoids any direct issues of ill-posedness. In the future, it could be instructive to consider a compressible version of the present model that remains formally well-posed even away from the lubrication limit.

Non-dimensionalisation and regularisation

To non-dimensionalise our model, we use bb, b/δ˙b/\dot{\delta}, δ˙\dot{\delta}, and ϕs∗Δρgb\phi_s^*\Delta\rho g b as characteristic spatial, temporal, velocity, and pressure scales. The key non-dimensional parameters are then

θ=ηfδ˙b2ϕs∗Δρg,K=kb2,andFr=δ˙dbbg,(29)\theta= \frac{\eta_f\dot{\delta}}{b^2\phi_s^*\Delta\rho g}, \qquad K = \frac{k}{b^2}, \qquad\text{and} \qquad\mathrm{Fr} = \frac{\dot{\delta}d}{b\sqrt{bg}}, \tag*{(29)}

as well as μw\mu_w, μs\mu_s, and the other constants in the μ(I)\mu(I) and μ(Iv)\mu(I_v) constitutive laws. The parameter KK is the non-dimensional permeability. Noting that ηfδ˙/b\eta_f\dot{\delta}/b is the characteristic shear stress from the upper fluid layer, the parameter θ\theta can be viewed as a modified Shields parameter that measures the strength of fluid shear relative to the weight of the granular bed. The modified Froude number Fr\mathrm{Fr} measures the strength of inertial forces relative to gravitational forces, scaled by d/bd/b. The ratio Fr2/θ≡St\mathrm{Fr}^{2}/\theta\equiv\mathrm{St} recovers the Stokes number, which, as noted above, compares the relative importance of inertial and viscous forces on the scale of the grains. In the ‘wet’ granular limit (St≪1\mathrm{St} \ll1), we find that Fr\mathrm{Fr} drops out of the problem; in the ‘dry’ granular limit (St≫1\mathrm{St} \gg1), Fr\mathrm{Fr} becomes the main control parameter.

In § 2.1 above, we neglected the inertia of the fluid–solid mixture in the macroscopic momentum balance (eq. (5)). The present scalings suggest that the relative importance of macroscopic inertia is governed by the square of the macroscopic Froude number, F2≡δ˙2/(bg)=(Fr b/d)2=St θ(b/d)2\mathcal{F}^{2} \equiv\dot{\delta}^{2}/(bg) = (\mathrm{Fr}\,b/d)^{2} = \mathrm{St}\,\theta(b/d)^{2}. Since the grains are much smaller than the gap, F\mathcal{F} is always larger than Fr\mathrm{Fr}, meaning that neglecting macroscopic inertia is reasonable only when Fr2≪1\mathrm{Fr}^{2} \ll1. Nonetheless, grain-scale inertia will control the granular rheology when St=Fr2/θ≫1\mathrm{St} = \mathrm{Fr}^{2}/\theta\gg1 (i.e., in the ‘dry’ limit). Together, these criteria suggest that macroscopic inertia can be neglected, even when the local dynamics are controlled by grain-scale inertia, when θ≪Fr2≪1\theta\ll\mathrm{Fr}^{2} \ll1. In the ‘wet’ limit (i.e., St≪1\mathrm{St} \ll1), Fr\mathrm{Fr} drops out and can be freely chosen to ensure that inertial effects are negligible. Note that we have also neglected the inertia of the overlying fluid layer, which is reasonable when the fluid Reynolds number Ref=ρfδ˙b/ηf\mathrm{Re}_{f} = \rho_{f}\dot{\delta}b/\eta_{f} is sufficiently small.

From this point on, we retain our original notation but all quantities are dimensionless unless explicitly noted otherwise. The evolution equation for the height of the granular layer, equation (2.2), becomes

∂h∂t+∂∂x(uˉsh)=0withuˉsh=∫0hy1us1 dz+(hy2−hy1)us2+∫hy2hus3 dz(30)\frac{\partial h}{\partial t}+\frac{\partial}{\partial x}(\bar{u}_{s}h)=0 \quad\text{with} \quad\bar{u}_{s}h=\int_{0}^{h_{y1}}u_{s1}\,\mathrm{d}z+(h_{y2}-h_{y1})u_{s2}+\int_{h_{y2}}^{h}u_{s3}\,\mathrm{d}z \tag*{(30)}

and

hy1=h−σxz′(z=h)−μsPs∂∂x(P+h)+μsandhy2=h−σxz′(z=h)+μsPs∂∂x(P+h)−μs,(31)h_{y1}=h-\frac{\sigma'_{xz}(z=h)-\mu_{s}P_{s}}{\frac{\partial}{\partial x}(P+h)+\mu_{s}} \quad\text{and} \quad h_{y2}=h-\frac{\sigma'_{xz}(z=h)+\mu_{s}P_{s}}{\frac{\partial}{\partial x}(P+h)-\mu_{s}}, \tag*{(31)}

recalling that P=Pf+PsP=P_{f}+P_{s}, where PfP_{f} and PsP_{s} are the dimensionless fluid pressure at z=1z=1 and the dimensionless solid pressure at z=hz=h, respectively. To solve for the velocity profile us(z)u_{s}(z), we equate equation (2.13) for the shear stress with equation (2.15) for the rheology to obtain

σxz′(z=h)−(h−z)(∂P∂x+∂h∂x)=μps,0≤z<hy1,(32)\sigma'_{xz}(z=h)-(h-z)\left(\frac{\partial P}{\partial x}+\frac{\partial h}{\partial x}\right)=\mu p_{s}, \qquad0\leq z<h_{y1}, \tag*{(32)}
∂us∂z=0,hy1≤z<hy2,(33)\frac{\partial u_{s}}{\partial z}=0, \qquad h_{y1}\leq z<h_{y2}, \tag*{(33)}
σxz′(z=h)−(h−z)(∂P∂x+∂h∂x)=−μps,hy2≤z<h,(34)\sigma'_{xz}(z=h)-(h-z)\left(\frac{\partial P}{\partial x}+\frac{\partial h}{\partial x}\right)=-\mu p_{s}, \qquad h_{y2}\leq z<h, \tag*{(34)}

where ps=Ps+(h−z)p_{s}=P_{s}+(h-z) is the dimensionless solid pressure and noting that the velocity profile us(z)u_{s}(z) must be continuous in zz. The shear stress at the top of the granular layer, σxz′(z=h)\sigma'_{xz}(z=h), is given below. As discussed in § 2.3, we use a frictional rheology model with μ=μ(I)\mu=\mu(I) for St≫1\mathrm{St}\gg1 (2.25) and μ=μ(Iv)\mu=\mu(I_{v}) for St≪1\mathrm{St}\ll1 (2.27), where the inertial and viscous numbers are

I=Frps∣∂us∂z∣andIv=θps∣∂us∂z∣,(35)I=\frac{\mathrm{Fr}}{\sqrt{p_{s}}}\left|\frac{\partial u_{s}}{\partial z}\right| \quad\text{and} \quad I_{v}=\frac{\theta}{p_{s}}\left|\frac{\partial u_{s}}{\partial z}\right|, \tag*{(35)}

respectively, and the associated expressions for μ\mu are given in equations (2.25) and (2.27). The shear stress σxz′(z=0)\sigma'_{xz}(z=0) is determined by wall friction via equations (2.18) and (2.19). Similarly to internal granular friction, a challenge with wall friction is that the horizontal profile us(x,z=0)u_s(x, z = 0) may feature multiple coupled sliding and non-sliding regions. To avoid the need to track these regions explicitly, we unify equations (18) and (19) in a dimensionless, regularised form as

σxz′(z=0)=μw(Ps+h)tanh⁡[us(z=0)ε](36)\sigma'_{xz}(z = 0) = \mu_w(P_s + h)\tanh\left[\frac{u_s(z = 0)}{\varepsilon}\right] \tag*{(36)}

where ε≪1\varepsilon\ll1 is a dimensionless regularisation parameter. Equation (36) provides a single continuous expression for σxz′(z=0)\sigma'_{xz}(z = 0) over the entire xx-domain, ensuring that ∣σxz′(z=0)∣<μw(Ps+h)\lvert\sigma'_{xz}(z = 0)\rvert< \mu_w(P_s + h) when us(z=0)≲ε≪1u_s(z = 0) \lesssim\varepsilon\ll1 while converging rapidly to equation (19) for larger values of us(z=0)u_s(z = 0). This type of regularisation is common in the numerical simulation of viscoplastic fluids, because it avoids the need to introduce internal free boundaries between yielded and un-yielded regions (e.g., [4]). Adopting it here for wall friction allows us to solve a single, continuous problem in xx, which is convenient for the quasi-1D flow studied here and even more-so for complex, quasi-2D frictional-flow patterns (i.e., [48]). Some care should be taken with regularisations of this form to avoid the potential misidentification of yielded regions (e.g., [45]); we illustrate the convergence of our solutions with decreasing ε\varepsilon in Appendix A.

The shear stress σxz′(z=h)\sigma'_{xz}(z = h) is determined locally by wall friction if the grains are in contact with the top wall (i.e., eqs. 2.21 and 2.21 for h≡1h \equiv1), or by fluid shear otherwise (i.e., eq. (8) for h<1h < 1). These scenarios can be captured concisely by introducing an indicator function χ(h)\chi(h) that transitions from χ(h<1)=0\chi(h < 1) = 0 to χ(h≡1)=1\chi(h \equiv1) = 1, such that

σxz′(z=h)=−χμwPstanh⁡[us(z=h)ε]+(1−χ)[h−12∂Pf∂x−θus(z=h)1−h](37)\sigma'_{xz}(z = h) = -\chi\mu_w P_s\tanh\left[\frac{u_s(z = h)}{\varepsilon}\right] + (1-\chi)\left[\frac{h-1}{2}\frac{\partial P_f}{\partial x} - \frac{\theta u_s(z = h)}{1-h}\right] \tag*{(37)}

where the first term on the right is regularised wall friction and the second is viscous shear.

In equation (37), the term proportional to θ\theta is inversely proportional to 1−h1-h, which illustrates an awkward feature of the model. As discussed in §2.1.1 above, the upper fluid layer satisfies no-slip conditions on both the stationary top wall and the moving granular layer. However, the thickness of this fluid layer vanishes as the granular layer approaches and then touches the top wall (i.e., as h→1h \to1; the bridging transition). As a result, the shear stress required to drive relative motion between the granular layer and the wall — which requires shearing the vanishing fluid layer — can become arbitrarily large as h→1h \to1, driving the system toward a no-slip condition between the granular layer and the wall. For a system of two fluid layers, this smooth approach toward no-slip between the lower layer and the wall is physically appropriate. In the present scenario, however, the lower layer should instead transition a frictional condition when it touches the upper wall; this frictional condition features an upper limit on the shear stress, allowing for sliding (i.e., slip) when that limit is reached. Hence, the lower layer could experience a jump in velocity and shear stress upon contacting the top wall unless this transition is regularised. In reality, the top of the granular layer is a permeable surface with grain-scale roughness. As a result, we expect fluid flow in the upper layer to transition smoothly into flow through the granular porous medium, and the shear stress at the top surface of the granular layer to transition smoothly from fluid shear to wall friction, as the irregular granular surface gradually makes contact with the top wall.

To reflect this smooth transition, we regularise the implementation of equation (37) in two ways. First, we place a non-zero minimum on the thickness of the fluid layer (i.e., on 1−h1-h) as it appears in the final term of equation (37), which reflects the fact that this construction breaks once the gap becomes comparable to the grain size. Specifically, we use the dimensionless permeability KK as a floor on 1−h1-h in the last term in equation (37) in order to link the regularisation to the grain size (K\sqrt{K} would scale more directly with the grain diameter, but would also regularise more aggressively). Second, we replace the sharp indicator function χ(h)\chi(h) with a smoothed step function of the form χ=hn\chi= h^n with n≫1n \gg1, thus regularising the bridging transition. The latter regularisation has the added benefit of preserving a continuous problem in xx, avoiding the need to track the free boundary between bridged and unbridged regions. We show in Appendix A, and specifically Figure 17c, that our results are insensitive to the value of nn provided that it is large enough.

Figure 17. (a,b) Illustration of the convergence of numerical solutions with improved regularisation parameter for the wall friction, ε\varepsilon, showing (a) the solid velocity at the base, and (b) the height of the layer as ε\varepsilon is reduced in the direction of the arrows (light to dark blue: ε=1,0.5,0.1\varepsilon= 1, 0.5, 0.1 and 0.010.01). (c) Convergence of the height of the layer with varying choose of smoothing function χ=hn\chi= h^n, for n=10,25,50,100,200,n = 10, 25, 50, 100, 200, and 400400 (light to dark blue, arrow in the direction of increasing nn). Simulations are performed μs=0.32\mu_s = 0.32, μ2=0.7\mu_2 = 0.7, μw=0.55\mu_w = 0.55, K=10−5K = 10^{-5} θ=10−5\theta= 10^{-5}

Equations (32)–(34) lead to a nonlinear first-order ODE for us(z)u_s(z) that also depends on PsP_s and ∂xPf\partial_x P_f. Expressions for the latter two quantities follow from evaluating equation (13) at z=0z = 0,

σxz′(z=h)−σxz′(z=0)=h(∂P∂x+∂h∂x).(38)\sigma'_{xz}(z = h) - \sigma'_{xz}(z = 0) = h\left(\frac{\partial P}{\partial x} + \frac{\partial h}{\partial x}\right). \tag*{(38)}

Combined with equations (36) and (37), this relationship determines ∂xPf\partial_x P_f in the unbridged region (where Ps=0P_s = 0) and PsP_s in the bridged region (where ∂xPf=0\partial_x P_f = 0), recalling that P=Ps+PfP = P_s + P_f.

Finally, the ODE for usu_s requires a boundary condition. The appropriate constraint follows from global conservation of mass, equation (4), which becomes

1=huˉs−kˉθ∂Pf∂x+1−h2us(z=h),(39)1 = h\bar{u}_s - \frac{\bar{k}}{\theta}\frac{\partial P_f}{\partial x} + \frac{1-h}{2}u_s(z = h), \tag*{(39)}

where kˉ=Kh+(1−h)3/12\bar{k} = Kh + (1-h)^3/12 is the dimensionless gap-averaged permeability.

Numerical details; boundary and initial conditions

To solve the model presented in § 2.4, we simplify computation by shifting to a frame X=x−tX = x-t that moves with the piston. With this change of frame, equation (30) becomes

∂h∂t+∂∂X[(uˉs−1)h]=0.(40)\frac{\partial h}{\partial t} + \frac{\partial}{\partial X}\left[(\bar{u}_s - 1)h\right] = 0. \tag*{(40)}

We solve the problem from initial condition h(X,0)=h0h(X,0) = h_0 and subject to boundary conditions us=uf=0u_s = u_f = 0 at X=0X = 0 (grains and fluid travelling with the impermeable piston) and ∂h/∂X=0\partial h/\partial X = 0 at X=LXX = L_X (grains undisturbed in the far field), where the domain size LXL_X is chosen to be sufficiently far ahead of the transition zone that its value has no impact on the results. We discretize the numerical domain 0≤X≤LX0 \le X \le L_X into 1+LX/ΔX1 + L_X/\Delta X nodes of equal spacing ΔX=5×10−3\Delta X = 5 \times10^{-3}. We then solve equation (40) for hh by approximating the spatial derivative using a second-order, essentially non-oscillatory scheme and integrating in time using the forward-Euler method with a fixed time step Δt=10−5\Delta t = 10^{-5}.

Evolving equation (40) in time requires that we compute uˉs\bar{u}_s at every discrete point XjX_j via equations (31)–(39) at each time step. To do so, we discretise the vertical domain at each point XjX_j, 0≤z≤h(Xj)0 \le z \le h(X_j), into Nz=250N_z = 250 points of equal spacing Δzj=h(Xj)/(Nz−1)\Delta z_j = h(X_j)/(N_z - 1). We then approximate ∂zus\partial_z u_s and ∂XPs\partial_X P_s at each point XjX_j as (∂zus)j≈(us(Xj,zi+1)−us(Xj,zi))/Δzj(\partial_z u_s)_j \approx(u_s(X_j,z_{i+1}) - u_s(X_j,z_i))/\Delta z_j and (∂XPs)j≈(Ps(Xj+1)−Ps(Xj))/ΔX(\partial_X P_s)_j \approx(P_s(X_{j+1}) - P_s(X_j))/\Delta X, respectively, noting that Ps=0P_s = 0 in the unbridged region. Thus, equations (31)–(39) become a coupled, nonlinear system of equations in Nz+6N_z + 6 unknowns: Ps(Xj)P_s(X_j), (∂XPf)j(\partial_X P_f)_j, σxz′(Xj,0)\sigma'_{xz}(X_j,0), σxz′(Xj,h(Xj))\sigma'_{xz}(X_j,h(X_j)), hy1(Xj)h_{y1}(X_j), hy2(Xj)h_{y2}(X_j), and us(Xj,zi)u_s(X_j,z_i) for i=1i = 1 to NzN_z. We solve this system of equations via Newton’s method using the solution at the previous time step as the initial guess. We then compute uˉs(Xj)\bar{u}_s(X_j) by averaging usu_s over appropriate regions in zz.

For the regularisation parameters in equations (36) and (37), we fix ε=10−3\varepsilon= 10^{-3} and n=200n = 200; our results are insensitive to these specific values (see Appendix A).

Dry model (μ(I)\mu(I) rheology, St≫1\mathrm{St} \gg1)

We first consider the ‘dry’ limit in which the fluid is unimportant, implying that St≫1\mathrm{St} \gg1. As discussed in § 2.3, we therefore impose the inertial μ(I)\mu(I) rheology given by equations (25) and (35) in the region where ∣∂zus∣>0\lvert\partial_z u_s\rvert> 0. The dry limit also implies that θ≪1\theta\ll1, such that equation (39) indicates ∂xPf∼0\partial_x P_f \sim0. For the results that follow, we adopt the illustrative parameter values I0=0.3I_0 = 0.3, μs=0.4\mu_s = 0.4, and μ2=0.7\mu_2 = 0.7, which are typical of monodisperse glass beads [30, 44], and we explore the roles of μw\mu_w, Fr\mathrm{Fr}, and h0h_0. The parameters θ\theta and KK drop out of the model in this limit.

To demonstrate the behaviour of the model, we present in Figure 4(a) an example simulation performed with Fr=0.1\mathrm{Fr} = 0.1 and μw=0.85\mu_w = 0.85. Our simulation qualitatively reproduces the behaviour illustrated in Figure 1. The grains undergo a period of initial pile-up until they bridge the gap. Thereafter, the length of the bridged region, LL, continues to grow and the transition region evolves toward a steady travelling-wave with constant speed L˙=1/(1−h0)\dot{L} = 1/(1-h_0) (Figure 4b), which follows from conservation of solid volume. As such, for visualisation purposes, it is convenient to introduce a travelling-wave coordinate x~=x−L\tilde{x} = x - L, such that the bridged region occupies x~<0\tilde{x} < 0 while the transition and undisturbed regions occupy x~>0\tilde{x} > 0. We consider the model to have reached a steady state once the shape of the transition region and the solid velocity within the transition region cease to vary in the travelling frame, which generally occurs quickly after the grains have bridged the gap.

Figure 4. (a) Height profiles of the granular layer at various times from numerical solutions to equations (30)–(39) with h0=0.5h_0 = 0.5, Fr=0.1\mathrm{Fr} = 0.1, and μw=0.85\mu_w = 0.85. The vertical, dotted black lines denote the location of the piston and the black arrow indicates direction of increasing time, with solutions being plotted at time intervals of 0.25 for 0≤t≤2.250 \le t \le2.25. (b) The translation speed of the transition region (i.e., the rate of change of the length of the bridged region, dL/dt\mathrm{d}L/\mathrm{d}t) in the steady travelling-wave limit from transient numerical simulations (red dots) and from conservation of solid volume, L˙=1/(1−h0)\dot{L} = 1/(1-h_0) (blue curve).

We next compute the solid velocity within the steady travelling wave over a range of values of Fr\mathrm{Fr} for μs<μw<μ2\mu_s < \mu_w < \mu_2 and for μs<μ2<μw\mu_s < \mu_2 < \mu_w, as shown in Figure 5. The horizontal red lines indicate where the grains are sliding along the walls (i.e., the regions where us>0u_s > 0 at z=0z = 0 and 11). Recall that μ(I)→μ2\mu(I) \to\mu_2 for I≫I0I \gg I_0, such that μ2\mu_2 is the limiting value of intra-granular friction at large shear rates. We do not illustrate the case μw<μs\mu_w < \mu_s because it leads to the trivial velocity profile us=1u_s = 1 throughout the bridged and transition regions for all Fr\mathrm{Fr} (Figure 2c and Figure 3b). Next, we investigate in detail the motion of the grains in the bridged region (§ 3.1), and in the transition (§ 3.2).

Figure 5. The velocity profile of the solid phase at steady state computed with μw=0.55\mu_w = 0.55 (left) and 0.85 (right). The Froude number (top to bottom) is Fr=0.01\mathrm{Fr} = 0.01, 0.05, and 0.35. The dotted black lines denote the location of hy1h_{y1} and hy2h_{y2} defined by equations (31), recalling that ∂zus=0\partial_z u_s = 0 for hy1<z<hy2h_{y1} < z < h_{y2}. The solid black vertical lines denote the end/start of the bridged/transition regions. The red lines denote the regions where the grains slide over the walls, i.e., us>0u_s > 0 at z=0z = 0 and 1.

Bridged region

Since h=1h = 1 in the bridged region by definition, equation (30) becomes unnecessary and equation (39) simplifies to uˉs=1\bar{u}_s = 1. However, as discussed in § 2.2, the solid velocity is nontrivial due to intra-granular friction and wall friction. As illustrated in Figure 5 row 1, when Fr is sufficiently small, the grains will yield over the entire gap around the start of the bridged region (hy1=hy2=1h_{y1} = h_{y2} = 1), and the grains do not slide over the bottom plate. As the length of the bridged region grows, the grains stop sliding over the top plate, resulting in the velocity profile illustrated in Figure 3(e). When μw<μ2\mu_w < \mu_2 (column 1), as Fr increases, we find that the grains begin to slide over both the top and bottom plates. Further, for large Fr, the grains appear to resist internal rearrangement such that ∣∂zus∣≪1|\partial_z u_s| \ll1 across the entire gap. In comparison, when μw>μ2\mu_w > \mu_2 (column 2), we find that the grains do not slide over the bottom or top plate for all Fr, and the velocity profile does not appear to change significantly as Fr is varied.

To better understand the influence of the Froude number and the wall friction parameter, we compute the velocity profile of the grains at z=1z = 1 and 0 over a range of values of Fr with μs<μw=0.55<μ2\mu_s < \mu_w = 0.55 < \mu_2, shown in Figure 6(a) and (Figure 6b). The velocity of the grains in contact with the top plate [Figure 6(a)] is fastest at the start of the bridged region and monotonically decreases to 0 as the length of the bridged region grows, after which the grains are stationary. Further, the length of the region where grains slide over the top plate increases with Fr. For the grains in contact with the bottom plate [Figure 6(b)], Fr must be sufficiently large or the grains will be stationary over the entire bridged region. When Fr becomes sufficiently large the grains will begin to slide, which corresponds to the point where μ(I)=μw\mu(I) = \mu_w shown in Figure 6(c). However, unlike grains in contact with the top plate, grains will be stationary at the start of the bridged zone, resulting in a region where grains slide sandwiched between regions where the grains are stationary. As Fr increases further, this region where us(z=0)>0u_s(z = 0) > 0 will ‘spread’ and the grains at the start of the bridged zone will begin to move (red curves). In comparison, when μw=0.85\mu_w = 0.85, grains do not slide over either plate for all Fr. Figure 6(c) indicates that while μ(I)\mu(I) (denoted by dashed lines for μw=0.85\mu_w = 0.85) increases with Fr as it does for μw=0.55\mu_w = 0.55, it will always be smaller than μw\mu_w, in this case, because it is bounded above by μ2=0.7<μw\mu_2 = 0.7 < \mu_w.

Figure 6. Velocity of the grains on the (a) top and (b) bottom plates with μw=0.55\mu_w = 0.55 and (light to dark blue) Fr = 0.01, 0.04, 0.06, 0.1, and 0.15. Plot (c) shows the friction factor μ(I)\mu(I) evaluated at z=0z = 0 for μw=0.55\mu_w = 0.55 (solid) and 0.85 (dashed). The dotted black lines denote μs=0.4\mu_s = 0.4, and the two values of μw\mu_w being considered.

As the length of the bridged region grows and PsP_s becomes sufficiently large, for all Fr and μw>μs\mu_w > \mu_s, the grains will not be able to slide over both the top and bottom plates. Further, as indicated by Figure 7(a), hy1h_{y1} and hy2h_{y2} evolve from 0 to 1, respectively, toward 1. As such, the velocity profile of the grains will tend to us≈1u_s \approx1, except for the increasingly thin regions around z=0z = 0 and 1. Such a profile is not observed in experiments, discussed further in section 5, and we anticipate that our continuum approximation will break down when hy1h_{y1} and 1−hy21-h_{y2} become sufficiently small. When the velocity of the grains in contact with the walls are of order ε\varepsilon, Figure 7(b) shows that tanh⁡(us/ε)\tanh(u_s/\varepsilon) at z=0z = 0 and 1 will tend to μs/μw\mu_s/\mu_w, which still satisfies us≪1u_s \ll1. Thus from equation (38), it follows that ∂xPs∼−2μsPs\partial_x P_s \sim-2\mu_s P_s, indicating that pressure builds up exponentially through the bridged region, as illustrated in Figure 7(c). By comparison, when μw<μs\mu_w < \mu_s where we have the trivial velocity profile us=1u_s = 1, tanh⁡(us/ε)∼1\tanh(u_s/\varepsilon) \sim1 and ∂xPs∼−2μwPs\partial_x P_s \sim-2\mu_w P_s resulting in the same exponential behaviour, but not controlled by the wall friction factor. This exponential relationship between the length of the bridged region and the normal stress at the piston is consistent with previous DEM simulations [36].

Yield layers, regularised shear stress, and pressure gradient

Figure 7. (a) The location of the yield layers hy1h_{y1} (dashed curve) and hy2h_{y2} (solid curve) determined by equations (31). (b) The hyperbolic tangent of the solid velocity at z=1z = 1 (solid curve) and z=0z = 0 (dashed line), which are used to compute the regularised shear stress according to equations (36) and (37), recalling that ε=10−3\varepsilon= 10^{-3}. (c) Natural log of ∣∂xPs∣|\partial_x P_s| computed from equation (38). Solutions are computed with Fr = 0.1 and μw=0.55\mu_w = 0.55 (blue) and 0.85 (red).

Transition zone

We now turn our attention to the behaviour of the transition zone, which, as illustrated in Figure 1(c), is the wedged-shape region with h<1h < 1 and Ps=0P_s = 0. In this region, given that the impact of the fluid on the grains is negligible, it follows that σxz(z=h)∼0\sigma_{xz}(z = h) \sim0, and the (non-dimensional) shear stress, equation (13), reduces to

σxz′=(z−h)∂h∂x.(41)\sigma'_{xz}=(z-h)\frac{\partial h}{\partial x}. \tag*{(41)}

As illustrated in Figure 2, when σxz(z=h)=0\sigma_{xz}(z=h)=0 the entire solid layer will be plugged (hy1=0h_{y1}=0 and hy2=hh_{y2}=h) if μw<μs\mu_w<\mu_s. If μw>μs\mu_w>\mu_s, the solid layer will yield (hy1=hy2=hh_{y1}=h_{y2}=h) if ∣∂xh∣>μs\lvert\partial_x h\rvert>\mu_s or be plugged otherwise. As such, we can reduce equation (23) to

us={us(z=0)if 0≤z<hy,usy(x,z,t)if hy≤z≪h,(42)u_s= \begin{cases} u_s(z=0) & \text{if }0\le z<h_y,\\ u_{sy}(x,z,t) & \text{if }h_y\le z\ll h, \end{cases} \tag*{(42)}

where

hy={0if ∣∂xh∣>μs,hotherwise.(43)h_y= \begin{cases} 0 & \text{if }\lvert\partial_x h\rvert>\mu_s,\\ h & \text{otherwise}. \end{cases} \tag*{(43)}

As such, we arrive at the so-called Bagnold profile ([1]) in the yielded region

usy=us(z=0)+2I03Fr∂xh+μs∂xh+μ2[(h−z)3/2−(h−hy)3/2],(44)u_{sy}=u_s(z=0)+\frac{2I_0}{3\mathrm{Fr}}\frac{\partial_x h+\mu_s}{\partial_x h+\mu_2}\left[(h-z)^{3/2}-(h-h_y)^{3/2}\right], \tag*{(44)}

where

0=∂h∂x+μwtanh⁡(us(z=0)ε).(45)0=\frac{\partial h}{\partial x}+\mu_w\tanh\left(\frac{u_s(z=0)}{\varepsilon}\right). \tag*{(45)}

Further, by substituting the inertial number, I=Fr⋅∂zus/h−zI=\mathrm{Fr}\cdot\partial_z u_s/\sqrt{h-z}, into equation (25), we arrive at

I=−I0(∂xh+μs)∂xh+μ2,(46)I=-\frac{I_0(\partial_x h+\mu_s)}{\partial_x h+\mu_2}, \tag*{(46)}

which combined with equation (25) reduces to μ(I)=−∂xh\mu(I)=-\partial_x h.

Recall that representative solutions were shown in Figure 5 with μw<μ2\mu_w<\mu_2 in column 1 and μ2<μw\mu_2<\mu_w in column 2. We find that as Fr→0\mathrm{Fr}\to0 (row 1), the height of the granular pile tends to a line of slope −μs-\mu_s. Further, the grains on the bottom plate are stationary such that the velocity profile takes the form illustrated in Figure 2(b). When μw<μ2\mu_w<\mu_2 (column 1), as Fr\mathrm{Fr} increases, the length of the transition zone decreases and the grains begin to slide along the bottom plate. For sufficiently large Fr\mathrm{Fr} (row 3), the grains appear to resist internal rearrangement such that ∣∂zus∣≪1\lvert\partial_z u_s\rvert\ll1. When μ2<μw\mu_2<\mu_w (column 2), again the slope of the transition zone decreases with increasing Fr\mathrm{Fr}. However, unlike in column 1, the grains remain stationary on the bottom plate for all Fr\mathrm{Fr}. Analogous behaviour was observed in the bridged region discussed previously. For the case where μw<μs\mu_w<\mu_s (not shown) the grains do not yield; as such, solutions are independent of Fr\mathrm{Fr} and the granular pile takes the shape of a wedge with ∂xh=−μw\partial_x h=-\mu_w.

We compute the maximum magnitude of the slope of hh as a function of Fr\mathrm{Fr} over a range of values of μw\mu_w [Figure 8(a)], as well as the corresponding length λ\lambda of the transition zone [Figure 8(b)], where we define the latter as the distance between the last bridged point and the first undisturbed point. We determine these values by integrating our full time-dependent model for a sufficiently long time such that the shape and velocity of the transition zone become constant. We find that for all μw>μs\mu_w>\mu_s, max⁡∣∂xh∣\max\lvert\partial_x h\rvert — and in turn max⁡∣μ(I)∣\max\lvert\mu(I)\rvert — will tend to μs\mu_s as Fr→0\mathrm{Fr}\to0. This behaviour can be explained by taking Fr→0\mathrm{Fr}\to0 with us1∼O(1)u_{s1}\sim O(1) in equation (44), which implies that ∂xh∼−μs\partial_x h\sim-\mu_s. When μs<μw<μ2\mu_s<\mu_w<\mu_2, max⁡∣∂xh∣\max\lvert\partial_x h\rvert ( max⁡∣μ(I)∣\max\lvert\mu(I)\rvert will monotonically increase with FrFr from μs\mu_s and plateau at μw\mu_w. Once max⁡∣∂xh∣\max\lvert\partial_x h\rvert has plateaued, the shear stress is sufficiently large on the base such that the grains will begin sliding, us(z=0)>0u_s(z=0)>0. Given that the inertial number, I∼Fr⋅∂zusI\sim Fr\cdot\partial_z u_s, does not increase above μw\mu_w once grains begin sliding along the base, it follows that ∂zus→0\partial_z u_s\to0 as FrFr becomes large, resulting in the plug-like flow observed in row 3 of Figure 5. Further, taking the limit that FrFr is large, equations (44) and (45) imply that us∼us(z=0)u_s\sim u_s(z=0) and ∂xh∼−μw\partial_x h\sim-\mu_w. When μw>μ2\mu_w>\mu_2, we find that max⁡∣∂xh∣\max\lvert\partial_x h\rvert will monotonically increase from μs\mu_s to μ2\mu_2 with increasing FrFr, while λ\lambda will decrease from λ→1/(2μs)\lambda\to1/(2\mu_s) to 1/(2μ2)1/(2\mu_2). As max⁡∣∂xh∣<μw\max\lvert\partial_x h\rvert<\mu_w for all FrFr, the shear stress at the base is never sufficiently large for the grains to slide over the bottom plate and remain stationary. As such, solutions are independent of μw\mu_w for all μw>μ2\mu_w>\mu_2. In the case where μw<μs\mu_w<\mu_s, as discussed above, we find that max⁡∣∂xh∣=μw\max\lvert\partial_x h\rvert=\mu_w and λ=1/(2μw)\lambda=1/(2\mu_w) for all FrFr. Note that it is no longer reasonable to neglect the macroscopic inertia of fluid–solid mixture once Fr≳O(1)Fr\gtrsim O(1), as discussed in § 2.4, so inertial effects should play a role for the solutions with the largest FrFr here.

Figure 8. (a) The maximum magnitude of the slope of hh, recalling that I=−∂xhI=-\partial_x h, computed with h0=0.5h_0=0.5, μs=0.4\mu_s=0.4, and μw=0.3,0.4,…,0.7\mu_w=0.3, 0.4, \ldots, 0.7. (b) The corresponding length λ\lambda of the transition region.

Although hh will tend to a straight line in the limits such that FrFr becomes sufficiently small or large, the shape of the granular pile is not trivial for intermediate values of FrFr. To better understand how the shape of the transition zone varies, we solve for hh over a range of values of FrFr [Figure 9(a)], and extract the corresponding slope [Figure 9(b)]. For small FrFr, the magnitude of the slope of hh tends to μs\mu_s as expected. As FrFr increases, hh becomes slightly more non-linear and ∣∂xh∣\lvert\partial_x h\rvert non-uniformly increases, where the slope is steeper at the start of the transition zone and decreases to 0 by the end. The magnitude of the slope increases with FrFr until it reaches μw\mu_w, which corresponds to the point at which the grains will slide along the bottom plate. We note that the shape of the transition zone is in good qualitative agreement with previous results. Specifically, Marks et al. [36], who performed DEM simulations of bulldozing in a confined geometry, showed that the shape of the transition zone is approximately linear with a gradient that steepens as the equivalent of the Froude number increases. Similarly, Sauret et al. [50], who investigated a continuum model of dry bulldozing in an unconfined geometry, showed that profiles are primarily linear in shape but can become slightly ‘curvy’, particularly around the leading edge, with increasing FrFr.

Figure 9. (a) The height of the granular pile for h0=0.5h_0 = 0.5, μw=0.6\mu_w = 0.6, and Fr=0.001\mathrm{Fr} = 0.001, 0.017, 0.036, 0.06, 0.1, 0.17, and 1. (b) Corresponding magnitude of the slope of hh. Dotted black lines in (b) denote values of μs\mu_s, μw\mu_w, and μ2\mu_2.

We repeat this process for a fixed FrFr across a range of values of h0h_0 [Figure 10]. For larger values of h0h_0 (darker blue curves), ∣∂xh∣\lvert\partial_x h\rvert monotonically decreases over the transition zone. For smaller values of h0h_0 (blue curves), ∣∂xh∣\lvert\partial_x h\rvert is non-monotonic such that it initially increases and reaches a peak and then decreases to 0. For sufficiently small h0h_0, max⁡∣∂xh∣\max\lvert\partial_x h\rvert rises above μw\mu_w, suggesting that there will be a region where the grains slide along the base (∣σzx∣>μwps\lvert\sigma^x_z\rvert>\mu_w p_s at z=0z=0) sandwiched between regions where the grains are stationary, us(z=0)=0u_s(z=0)=0.

Figure 10. (a) The height of the granular pile for Fr=0.06\mathrm{Fr} = 0.06, μs=0.4\mu_s = 0.4, μw=0.6\mu_w = 0.6, and h0=0.1h_0 = 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, and 0.7. (b) Corresponding magnitude of the slope of hh. Dashed black lines denote the values of μs\mu_s, μw\mu_w, and μ2\mu_2.

Viscous model (μ(Iv)\mu(I_v) rheology)

We turn our attention to the influence of the ambient fluid on the grains. In this case, the Stokes number, recalling St∼1/ηf\mathrm{St} \sim1/\eta_f, can be small, and as such we impose μ(Iv)\mu(I_v) rheology in the region where the grains have yielded, as discussed in §2.2. As illustrative values for the following results, we take K=10−4K = 10^{-4}, I0=0.005I_0 = 0.005, μs=0.32\mu_s = 0.32 and μ2=0.7\mu_2 = 0.7, which were measured by Boyer et al. [11] in the case of plastic beads, and we fix h0=0.5h_0 = 0.5. Recall that the model becomes independent of Fr\mathrm{Fr} in this limit, provided that F2=(Fr b/d)2≪1\mathcal{F}^2 = (\mathrm{Fr}\,b/d)^2 \ll1 to ensure that macroscopic inertia plays no role. Following Figure 5, we compute the steady state velocity profiles over a range of values of θ\theta, shown in Figure 11 (recall, θ\theta provides a measure of the relative importance of fluid shear forces, and increases with both the piston speed and the fluid viscosity; see (29)). In the bridged region to the left of the vertical black line, equations (39) and (11) lead to 1=ϕs∗uˉs+(1−ϕs∗)uˉf1 = \phi_s^*\bar{u}_s + (1 - \phi_s^*)\bar{u}_f, which implies that uˉs=uˉf\bar{u}_s = \bar{u}_f and ∂xPf=0\partial_x P_f = 0. As such, the ambient fluid does not have a direct impact on the behaviour of the grains in the bridged region. While imposing μ(Iv)\mu(I_v) rheology results in a different velocity profile to μ(I)\mu(I), we observe the same general behaviours with both models. That is, we find a non-monotonic profile with a plugged region sandwiched between two regions where the grains slide over one another. However, we find the behaviour of the grains to be noticeably different in the transition region. In particular, Figure 11 shows that the length of the transition region varies non-monotonically with increasing θ\theta, initially decreasing (although this is hard to see by eye) before sharply increasing. In fact, we are unable to generate steady velocity profiles at all above a certain value of θ\theta, as discussed below.

Figure 11. The horizontal velocity profile of the solid phase at steady state computed with μw=0.55\mu_w = 0.55 for different values of θ\theta. The dotted black lines denote the location of hy1h_{y1} and hy2h_{y2} defined by equations (31), recalling that ∂zus=0\partial_z u_s = 0 for hy1<z<hy2h_{y1} < z < h_{y2}.

A useful metric for quantifying the role of the fluid is the (non-dimensional) horizontal force exerted by the fluid on the grains at the interface between the upper and lower layers,

Ff=∣∂Pf∂x∣+σxz′(z=h)h,(47)F_f = \left|\frac{\partial P_f}{\partial x}\right|+\frac{\sigma'_{xz}(z=h)}{h}, \tag*{(47)}

where max⁡∣μ(Iv)∣≈max⁡∣Ff−∂xh∣\max|\mu(I_v)| \approx\max|F_f-\partial_x h|. As a point of comparison, we also computed solutions in which the overlying fluid was ignored, and thus Ff=0F_f=0, by solving equations (30)–(39) with ∂xPf=σxz′(z=h)=0\partial_xP_f=\sigma'_{xz}(z=h)=0.

Figure 12 shows details of the steady travelling wave from an example solution for moderate θ\theta. The length of the transition zone (Figure 12a) is larger when the fluid is included (solid lines) compared to when the fluid is ignored (dashed lines). There is a dip in the associated surface slope near the back of the transition zone (Figure 12b), which corresponds to a peak in force from the overlying fluid FfF_f (Figure 12c). Physically, increasing viscous forces in the fluid phase makes it harder for it to be displaced by the solid as the grains pile up. This effect is exacerbated as θ\theta increases, which causes the transition zone to be elongated.

Figure 12. (a) The height of the granular pile computed with θ=3.6×10−4\theta= 3.6 \times10^{-4} and μw=0.6\mu_w = 0.6. Plots (b) and (c) are the corresponding magnitude of dh/dx\mathrm{d}h/\mathrm{d}x and fluid force, equation (47). The dashed line denotes the case where FfF_f is ignored by setting ∂xP=σxz′(z=h)=0\partial_x P = \sigma'_{xz}(z = h) = 0. Note that this is distinct from the dry rheology because viscosity is still important for the internal rearrangement of the grains.

This behaviour can be clearly seen in Figure 13(a), which shows the steady-state length of the transition zone λ\lambda as a function of θ\theta. The transition zone length gently decreases for small θ\theta in line with the fluid-free case (dashed lines), but increases sharply for larger θ\theta as the resistance from the fluid layer starts to have an impact. These lines terminate at a critical value of θ\theta, beyond which computations never evolve to a steady travelling-wave solution. The transition point varies with the wall friction μw\mu_w, but is also strongly influenced by the permeability of the grains (inset of Figure 13a), with higher permeability increasing the value of θ\theta beyond which no steady solution is attained. Physically, the ability for the upper fluid to flow through the grain pack aides the displacement (via drainage) of that layer, and so grains with lower resistance to drainage (higher permeability) have a higher critical θ\theta.

Figure 13. (a) The length of the transition zone, (b) the maximum magnitude of the slope of hh, (c) the maximum of the fluid force determined by equation (47), and (d) the maximum value of hh over the duration of a simulation. The dots at which the various lines terminate denote the largest value of θ\theta at which the numerical solution converged to a steady travelling wave. The inset in (a) indicates how the impact of permeability KK on the value θ=θ∗\theta= \theta^* at which there ceases to be a steady travelling wave for μw=0.5\mu_w = 0.5. The dashed lines denote the case where the fluid phase is ignored by solving equations (30)–(39) with ∂xPf=σxz′(z=h)=0\partial_x P_f = \sigma'_{xz}(z = h) = 0.

The corresponding maximum slope of the free surface (Figure 13b) similarly peels off from the fluid-free case as θ\theta is increased, which also reflects the elongation of the transition region. Figure 13(c) shows the maximum value of the fluid force in the steady travelling-wave solution, which gives a clear indication of the impact of the fluid layer above the grains: the force is negligible when θ\theta is sufficiently small, but increases significantly when the fluid-free and full results start to deviate in the first two panels. Note that the force-free (i.e., fluid-free) results in Figure 13 follow much the same trend as the ‘dry’ results in Figure 8, although there are some small quantitative differences from imposing μ(Iv)\mu(I_v) instead of μ(I)\mu(I).

Figure 13(d) shows the maximum height ever reached by any of the grains, which is 1 (i.e. the height of the channel) if bridging occurs. Evidently the grains never bridge the gap at all if θ\theta is sufficiently large. However, comparison with the other panels shows that this transition from bridging to no bridging occurs at a notably larger value of θ\theta than the point at which a steady transition zone is no longer observed: there is evidently a range of θ\theta for which the grains do bridge the gap but the transition zone does not attain a steady travelling wave.

To further explore this behaviour, we show in figure 14 details of how hh evolves for three different values of θ\theta. The first (figure 14a) attains a steadily propagating transition zone; the

Evolution of the thickness of the granular layer and associated transition-region speeds and lengthVector (PDF)

caption: (a)–(c) Evolution of the thickness of the granular layer from the viscous model with μw=0.5\mu_w = 0.5 for increasing θ\theta (top to bottom) at equally spaces times in 0≤t<40 \leq t < 4. (d) The associated speed of the leading and trailing edges of the transition region, where these edges exist. The leading and trailing edges are the maximum values of XX where h=h0h = h_0 (solid lines) and where h=1h = 1 (dotted lines), respectively. Both speeds exceed the speed of the piston (horizontal black line) in all cases. The two edges converge to the same speed for θ=10−4\theta= 10^{-4} (teal) and to different speeds for θ=7.74×10−4\theta= 7.74 \times10^{-4} (light purple). The transition region has no trailing edge for the largest value of θ\theta, which never develops a bridged region (dark purple). (e) Time evolution of the length λ\lambda of the transition region, defined as the distance between the leading and trailing edges of the transition region. Colours in (d,e) correspond to the different cases in (a–c); the case in (c) never bridges the gap, such that λ\lambda is not defined.

:::

second (figure 14b) bridges the gap, but leaves a transition zone that grows steadily longer; and the third (figure 14c) never bridges the gap. Panels (d) and (e) show measurements of the speed of the front and back of the transition zone, and its length, which confirm this behaviour: beyond a certain value of θ\theta the back of the transition region moves more slowly than the front and the transition region inexorably lengthens over time; beyond a larger value of θ\theta, the grains are unable to bridge the gap and the upper fluid layer reaches all the way back to the piston.

Comparison with Experiments

We present a number of laboratory experiments conducted in a horizontal glass tube with a square cross-section of internal side height b=3b = 3 mm and width W=3W = 3 mm, and a piston consisting of a steel rod with a moulded gasket (Vytaflex 60). For each experiment, the tube is initially filled with a liquid before the partial loading of wet soda lime glass beads (Honite Blast Media) with ρs=2480\rho_s = 2480 kg⋅\cdotm−3^{-3} sieved to size range 150<d<200150 < d < 200 μ\mum. The granular material was fed via a syringe into the tube, angled at approximately 45 degrees, and allowed to fall due to gravity until forming a layer of height h0=0.75±0.15h_0 = 0.75 \pm0.15 mm. The piston was driven into the stationary tube by a syringe pump (Harvard PHD Ultra) at speeds ranging from 1.85×10−21.85 \times10^{-2} to 2.78 mm/s. Two interstitial liquids were used: water and a solution of 70wt% glycerol (Fisher Scientific) in water (ρf=1181\rho_f = 1181 kg⋅\cdotm−3^{-3} and ηf=2.25×10−2\eta_f = 2.25 \times10^{-2} Pa⋅\cdots). Experiments were captured with a Nikon Z6II camera and 105 mm macro lens. Following previous models of dense granular transport [3, 41], we estimate the permeability of the grains via the Kozeny–Carman formula to be k=4.5×10−5 mm2k = 4.5 \times10^{-5}\ \mathrm{mm}^2. Further, μs\mu_s and μw\mu_w are taken to be 0.48 and 0.53, which we estimated from the angle of repose and the angle at which the beads slide over a flat surface, respectively. The parameters μ2\mu_2 and I0I_0 are more difficult to determine without more advanced rheometry, so we use the values reported by Boyer et al. [11] (μ2=0.7\mu_2 = 0.7, I0=0.005I_0 = 0.005). Over the range of these experiments, the square of the macroscopic Froude number F2\mathcal{F}^2 is never greater than O(10−3)O(10^{-3}), suggesting that it is reasonable to neglect the inertia of the grain–fluid mixture. The maximum Reynolds number in the fluid layer Ref\mathrm{Re}_f is O(1)O(1), suggesting that fluid inertia should start to play a role in the fastest experiments.

We compare the height of the granular pile determined from experiments and simulations in Figure 15 for a relatively small [(a) and (c)] and large [(b) and (d)] value of θ\theta, corresponding to relatively slow and fast piston velocity, respectively. There is broad qualitative agreement between the experiments and simulations. That is, experiments indicate that both the length and shape of the transition region is θ\theta-dependent. For low θ\theta, the granular pile appears to have a roughly linear profile, while for a larger value of θ\theta, hh appears more non-linear. In Figure 15(e), we compare the length of the transition zone, which is computed manually from experimental images using the processing software ImageJ, with steady-state measurements of at least three experiments being averaged. Again, both experiments and simulations reproduce the same key behaviours. Specifically, λ\lambda is non-monotonic in θ\theta, where λ\lambda decreases due to the balance of frictional forces between grains and walls, before sharply increasing due to the increasing viscous forces imposed on the grains. We do note however that the experiments measure a lower value of λ\lambda compared to the model. One possible reason for this discrepancy is that the influence of the sides of the tube is not included in the model. As such, we extend our model to account for frictional forces due to the sides of the tube, with details being provided in Appendix B. Note that this extended model includes side-wall friction only under the assumption that there is negligible variation in the flow in the horizontal (yy) direction, and this assumption must break down if the side-wall friction is sufficiently strong. Figure 15 shows that the inclusion of side-wall friction (dashed line) into the model improves the agreement with experiments, suggesting that such friction has a significant effect. A further possible reason for the discrepancy between theory and experiments here could be inaccuracy. in the prediction of the permeability of the grains. The results in the inset of Figure 13(a) showed that a higher permeability leads to a higher value of θ\theta beyond which no steady λ\lambda exists, and so we may also be slightly underestimating the permeability of the grain pack here.

Figure 15. Panels (a) and (b) show experimental snapshots of the travelling-wave state for (a) θ=4.83×10−7\theta= 4.83 \times10^{-7} (slow pushing) and (b) 1.23×10−41.23 \times10^{-4} (fast pushing). Panels (c) and (d) show the thickness of the granular layer from the experiments (red) and from corresponding simulations (without side walls; blue). For the experiments, dotted curves are the mean of three repeats and the light red bands denote one standard deviation above and below the mean. (e) The non-dimensional length of the transition region as a function of θ\theta from experiments (symbols) and simulations (curves). Blue and red dots denote experiments performed with water and with a solution of 70 wt% glycerol in water, respectively. Solid and dashed black curves denote simulations performed without and with the effect of side walls, respectively. Simulations are computed with K=5×10−6K = 5 \times10^{-6}, μw=0.53\mu_w = 0.53, μs=0.48\mu_s = 0.48, I0=0.005I_0 = 0.005, and h0=0.25h_0 = 0.25.

We next compare the solid velocity profiles from experiments with simulations for a ‘moderate’ value of θ\theta, shown in Figure 16. We measure the solid velocity in the experiments from from side-wall observations, noting that observations from above suggest that the flow field is fairly uniform across the cross-section. We position-correct videos to track the start of the transition region and then compute the steady velocity field using particle image velocimetry (PIV), time-averaging over several frames [54]. We find that simulations and experiments are in reasonable agreement with each other. In the transition region, x~≥0\tilde{x} \ge0, experiments show that the velocity profile is monotonically increasing in the vertical direction, and the shear stress at z=0z = 0 is sufficient for the grains to slide over the base, which is consistent with the model. Further, for a smaller θ\theta (not shown), grains in contact with the bottom of the tube are stationary, again consistent with simulations. We do note, however, that experiments appear to produce profiles where ∂z2us>0\partial_z^2 u_s > 0, while the model produces profiles where ∂z2us<0\partial_z^2 u_s < 0 for all values of θ\theta. Incorporating side walls into the model (dashed lines) can produce profiles where ∂z2us<0\partial_z^2 u_s < 0, suggesting again that the presence of the sides of the tube has a significant influence on the dynamics.

Slices of the non-dimensional solid velocity profiles in the lab frame

Figure 16. Slices of the non-dimensional solid velocity profile in the lab frame with θ=1.23×10−5\theta= 1.23 \times10^{-5} from the model (blue) and experiments (red dots) taken at (left to right) x~=−1,−0.5,…,1.5\tilde{x} = -1, -0.5, \ldots, 1.5. The dotted and solid lines denote profiles determined from the model with and without the effects of the side walls, respectively.

Regarding the velocity profile in the bridged region, recall that our model exhibits a piecewise velocity profile of the form of eq. (23) illustrated in Figure 3. Figure 16 shows that qualitatively, experiments and simulations produce similar behaviour, where grains move slower near the top and bottom walls and faster in the interior, away from the walls. One notable difference between the two is that, unlike the simulations, the experiments do not exhibit a yielded profile in the interior. Unlike in the transition region, accounting for side walls does not appear to have a qualitatively or quantitatively important impact on the velocity profile, and the two scenarios (side walls and no side walls) appear to converge moving to the left, away from the transition region. However, side walls do have an important impact on the bulldozing force. As discussed in § 3.1, we find that without side walls ∣∂xPs∣∼2μwPs|\partial_x P_s| \sim2\mu_w P_s in the bridged region; with side walls, the growth rate of the normal stress in the bridged region effectively doubles, such that ∣∂xPs∣∼4μwPs|\partial_x P_s| \sim4\mu_w P_s. We further note that the vertical velocity of the grains (not shown) from experiments is several orders of magnitude smaller than the horizontal component across the entire domain, supporting our assumption of a lubrication-type flow (ws≪usw_s \ll u_s; § 2.2).

Conclusion

We have presented a continuum model for the bulldozing of a fluid-saturated granular material in a confined geometry. We derived this model by framing the problem as a depth-averaged system of coupled thin films: an incompressible Newtonian fluid overlying a deformable, porous granular layer. To model intra-granular friction, we imposed either the inertial μ(I)\mu(I) rheology (‘dry’ case) or the viscous μ(Iv)\mu(I_v) rheology (‘wet’ case), depending on the relative importance of fluid viscosity — as quantified by the value of the Stokes number. We modelled contact with the bottom and top walls as simple Coulomb friction, such that the shear stress at the wall must be sufficiently large for the grains to slide. We neglect macroscopic inertial forces in both the fluid layer and the granular suspension. By solving our model numerically, we investigated the motion of the grains in the region where the granular layer has bridged the gap between the plates and in the transition zone between the bridged region and the undisturbed far-field. We also provide a comparison of our model with experiments in § 5.

In §3, we considered the ‘dry’ case in which the fluid is negligible. We found that, in the bridged region (h=bh=b), the grains exhibit a non-monotonic velocity profile comprising a plugged region (∂zus=0\partial_z u_s=0) sandwiched between two yielded regions (∣∂zus∣>0|\partial_z u_s|>0). This profile comes from a balance between the rate-dependent frictional intra-granular stresses with sliding friction at the walls. Further, we find that the normal stress at the piston grows exponentially with the length of the bridged region, as expected from previous work. In § 3.2, we considered the behaviour of the grains in the wedge-shaped transition zone. We found that the shape of the transition zone depends on the Froude number, generally decreasing in length with increasing Fr due to the increasing resistance to internal rearrangement as the shear rate increases. As in the bridged region, the specific velocity profile and height of the granular layer are determined by a balance between intra-granular friction and wall friction.

In §4, we considered the ‘wet’ case in which fluid viscosity plays an increasingly important role at faster rates. We found that, for a sufficiently small Shields number θ\theta, the impact of the fluid is negligible and solutions exhibit the same qualitative behaviour as the ‘dry’ model. Specifically, the transition zone tends toward a triangular wedge with a slope μs\mu_s (i.e., the static angle of repose) for θ≪1\theta\ll1. This wedge shortens as θ\theta increases, with the slope steepening toward μw\mu_w (provided that μw>μs\mu_w>\mu_s). As θ\theta increases further, the transition zone instead eventually begins to lengthen and becomes less triangular due to the increasingly important role of fluid viscosity, unlike in the dry case. For sufficiently large θ\theta, our model predicts that the transition region has no steady shape; rather, it lengthens continually as viscous shear from the overlying fluid acts to flatten the granular free surface. For even larger θ\theta, viscous shear is sufficiently strong that the granular layer is unable to bridge the gap.

This non-monotonic variation of the length of the transition region with θ\theta in the ‘wet’ case is consistent with experiments presented in § 5, which also show a gentle decrease and then a steep increase in λ\lambda as θ\theta increases. For low θ\theta, both model and experiments show that the grains in the transition zone will exhibit no-slip on the bottom wall, with a horizontal velocity that increases monotonically in zz. For larger θ\theta, the shear stress at the base becomes sufficiently large such that the grains can slide along the bottom wall.

Although the model presented here produces reasonable qualitative agreement with experiments, several improvements are possible. For example, the model provides a somewhat better quantitative match with experiments when friction with the side walls is included. In § 2, we assumed that the solid fraction of granular layer is constant; however, in reality, granular materials are known to compact or dilate under shear. This behaviour can be incorporated by introducing additional constitutive laws for ϕ(I)\phi(I) for the inertial regime ([21]) or ϕ(Iv)\phi(I_v) for the viscous regime ([11]). Similarly, our model ignores inertial effects at the continuum scale, which must become more important at higher shear rates and could, for example, enable the granular layer to overcome viscous resistance from the fluid and bridge the gap for large θ\theta. Our experiments at the highest values of θ\theta had F2∼10−3\mathcal{F}^{2} \sim10^{-3} but a fluid Reynolds number Ref∼1Re_f \sim1, suggesting that inertia in the fluid layer is likely to have more of an effect on the dynamics here than inertia in the granular mixture. Finally, although our model is based on a lubrication approximation, the flow may not be predominantly horizontal in all cases; as a result, secondary recirculating flow or non-trivial normal stresses could become important and are not captured in this modelling framework.

As discussed in § 1, one motivation for this work was to provide insight into the behaviour of a granular material that is being bulldozed by a fluid-fluid interface, such as during the injection of a gas into a hydrophilic, water-saturated packing. As long as capillary forces are sufficient to prevent the gas from invading the pore space of the packing, this scenario can be thought of as the bulldozing of grains by a deformable piston. Experiments in both capillary tubes (Dumazer et al. [16]; Thorens et al. [55]) and Hele-Shaw cells (Sandnes et al. [48]) show the formation of a variety of patterns resulting from the balance between frictional and capillary forces. With this in mind, an avenue for future work is to extend the two-phase model presented here to incorporate an additional fluid that is non-wetting to the grains.

Funding. This work was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 Programme [Grant No. 805469] and by the UK Engineering and Physical Sciences Research Council (EPSRC) [Grant No. EP/S034587/1 and EP/X028771/1].

Declaration of interests. The authors report no conflict of interest.

Appendix A. Validation of regularisation

The model presented here contains regularisation of the shear stress on the bottom and top (when the gap is bridged) of the channel, as introduced in §2.1.3. Figure 17(a,Figure 17b) illustrates the convergence of the numerical solution of our model with the regularisation parameter ε\varepsilon. Panel (b) shows that as ε\varepsilon decreases the difference in the predicted height profile becomes indistinguishable between the two lowest values of ε\varepsilon shown here, which are still at least an order of magnitude larger than the value we use throughout this paper (ε=10−3\varepsilon= 10^{-3}).

We also use a regularisation method to describe the transition between bridged and unbridged grains, as discussed after (39). Here we cap the potentially diverging shear stress coming from the upper fluid, and we also smooth out the jump between bridged and unbridged fluid using a smoothed indicator function χ=hn\chi= h^n. We find that changing the value of the cap on the shear stress has no discernible impact on the results. For the smoothed indicator function, Figure 17(c) shows results for different choices of the parameter nn. At this scale, differences in hh are negligible for nn larger than 100; we use n=200n = 200 in all our simulations here.

Appendix B. Influence of side walls

Here we consider the effect of the side walls of the tube, which are located at y=±W/2y = \pm W/2, by developing a heuristic model to account for some side-wall friction. We still consider the velocity in the yy and zz directions to be negligible such that vs=usex\mathbf{v}_{s} = u_{s}\mathbf{e}_{x}. As such, equation (12) becomes

∂σxy′∂y+∂σxz′∂z=∂∂x(ps+pf)=∂Pf∂x+∂Ps∂x+ϕsΔΔρg∂h∂x,(48)\frac{\partial\sigma'_{xy}}{\partial y} + \frac{\partial\sigma'_{xz}}{\partial z} = \frac{\partial}{\partial x}(p_s + p_f) = \frac{\partial P_f}{\partial x} + \frac{\partial P_s}{\partial x} + \phi_s^{\Delta}\Delta\rho g \frac{\partial h}{\partial x}, \tag*{(48)}

recalling ps=Ps+Δρgϕs∗(h−z)p_s = P_s + \Delta\rho g\phi_s^*(h-z) and pf=Pf+ρfg(b−z)p_f = P_f + \rho_f g(b-z) as before. Taking the horizontal (yy-direction) average of (B 1) and assuming that any variation in the vertical shear stress and the pressures in the yy direction is sufficiently weak that the averages can be approximated by their local values, we arrive at

∂σxz′∂z≈∂Pf∂x+∂Ps∂x+ϕs∗Δρg∂h∂x+σxy′(y=−b/2)−σxy′(y=b/2)W(49)\frac{\partial\sigma'_{xz}}{\partial z} \approx\frac{\partial P_f}{\partial x} + \frac{\partial P_s}{\partial x} + \phi_s^* \Delta\rho g \frac{\partial h}{\partial x} + \frac{\sigma'_{xy}(y=-b/2)-\sigma'_{xy}(y=b/2)}{W} \tag*{(49)}

Analogous to the friction imposed on the top and bottom plates as described in § 2, we impose a Coulomb friction condition on the side walls

us=0if σxy′(y=±W/2)<μwps,(50)u_s = 0 \qquad\text{if } \sigma'_{xy}(y=\pm W/2)<\mu_w p_s, \tag*{(50)}
us>0if σxy′(y=±W/2)=μwps,(51)u_s > 0 \qquad\text{if } \sigma'_{xy}(y=\pm W/2)=\mu_w p_s, \tag*{(51)}

We regularise equations (B 3) and (B 4) analogous to the bottom and top plates via equations (36) and (37) such that

σxy′(y=W/2)−σxy′(y=−W/2)=2μw(Ps+ϕs∗Δρg(h−z))tanh⁡(usε).(52)\sigma'_{xy}(y=W/2)-\sigma'_{xy}(y=-W/2)=2\mu_w\left(P_s+\phi_s^*\Delta\rho g(h-z)\right)\tanh\left(\frac{u_s}{\varepsilon}\right). \tag*{(52)}

Integrating equation (B 2) with respect to zz gives the expression for the shear stress

σxz′=σxz′(z=h)−(h−z)(∂Pf∂x+∂Ps∂x+ϕs∗Δρg∂h∂x+2μw(Ps+ϕs∗Δρg(h−z))Wtanh⁡(usε)).(53)\sigma'_{xz}=\sigma'_{xz}(z=h)-(h-z)\left(\frac{\partial P_f}{\partial x}+\frac{\partial P_s}{\partial x}+\phi_s^*\Delta\rho g\frac{\partial h}{\partial x}+\frac{2\mu_w\left(P_s+\phi_s^*\Delta\rho g(h-z)\right)}{W}\tanh\left(\frac{u_s}{\varepsilon}\right)\right). \tag*{(53)}

The effect of the side walls can be neglected if WW is sufficiently large. Conversely, if WW is very small we would expect the assumption that quantities depend only weakly on yy, made before (B 2), to break down. As before, equations (15) and (16) still apply, where μ(Iv)\mu(I_v) rheology (or μ(I)\mu(I)) is applied in the region where ∣σxz∣>μsps|\sigma_{xz}|>\mu_s p_s and ∂zus=0\partial_z u_s=0 otherwise.

References

  1. [1]Andreotti, B., Forterre, Y. & Pouliquen, O. 2013 Granular media: between fluid and solid. Cambridge University Press.
  2. [2]Artoni, R. & Richard, P. 2015 Effective wall friction in wall-bounded 3D dense granular flows. Phys. Rev. Lett. 115, 158001.DOI
  3. [3]Aussillous, P., Chauchat, J., Pailha, M., Médale, M. & Guazzelli, É. 2013 Investigation of the mobile granular layer in bedload transport by laminar shearing flows. J. Fluid Mech. 736, 594–615.DOI
  4. [4]Balmforth, Neil J., Frigaard, Ian A. & Ovarlez, Guillaume 2014 Yielding to stress: Recent developments in viscoplastic fluid mechanics. Annual Review of Fluid Mechanics 46 (Volume 46, 2014), 121–146.DOI
  5. [5]Barker, T., Gray, J. M. N. T., Schaeffer, D. G. & Shearer, M. 2023 Well-posedness and ill-posedness of single-phase models for suspensions. Journal of Fluid Mechanics 954, A17.DOI
  6. [6]Barker, T., Schaeffer, D. G., Bohorquez, P. & Gray, J. M. N. T. 2015 Well-posed and ill-posed behaviour of the μ(i) -rheology for granular flow. Journal of Fluid Mechanics 779, 794–818.DOI
  7. [7]Barker, T., Schaeffer, D. G., Shearer, M. & Gray, J. M. N. T. 2017 Well-posed continuum equations for granular flow with compressibility and μ(i)-rheology. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 473, 20160846, arXiv: https://royalsocietypublishing.org/rspa/article-pdf/doi/10.1098/rspa.2016.0846/364225/rspa.2016.0846.pdf.DOI
  8. [8]Baumgarten, A. S. & Kamrin, K. 2019 A general fluid–sediment mixture model and constitutive theory validated in many flow regimes. J. Fluid Mech. 861, 721–764.DOI
  9. [9]Beavers, G. S. & Joseph, D. D. 1967 Boundary conditions at a naturally permeable wall. J. Fluid Mech. 30, 197–207.DOI
  10. [10]Bouchut, F., Fernández-Nieto, E. D., Mangeney, A. & Narbona-Reina, G. 2016 A two-phase two-layer model for fluidized granular flows with dilatancy effects. J. Fluid Mech. 801, 166–221.DOI
  11. [11]Boyer, F., Guazzelli, É. & Pouliquen, O. 2011 Unifying suspension and granular rheology. Phys. Rev. Lett. 107, 188301.pubmed.ncbi.nlm.nih.gov/22107679
  12. [12]Cassar, C., Nicolas, M. & Pouliquen, O. 2005 Submarine granular flows down inclined planes. Phys. Fluids 17, 103301.DOI
  13. [13]Da Cruz, F., Emam, S. M., Prochnow, Roux, J. N. & Chevoir, F. 2005 Rheophysics of dense granular materials: Discrete simulation of plane shear flows. Phys. Rev. E 16, 021309.DOI
  14. [14]Doppler, D., Gondret, P., Loiseleux, T., Meyer, S. & Rabaud, M. 2007 Relaxation dynamics of water-immersed granular avalanches. J. Fluid Mech. 577, 161–181.DOI
  15. [15]Dorostkar, O., Guyer, R. A., Johnson, P. A., Marone, C. & Carmelit, J. 2017 On the micromechanics of slip events in sheared, fluid-saturated fault gouge. Geophys. Res. Lett. 44, 6101–6108.DOI
  16. [16]Dumazer, G., Sandnes, B., Ayaz, M., Måløy, K. J. & Flekkøy, E. G. 2016 Frictional fluid dynamics and plug formation in multiphase millifluidic flow. Phys. Rev. Lett. 117, 028002.DOI
  17. [20]Eriksen, J. A., Toussaint, R., Måløy, K. J., Flekkøy, E., Galland, O. & Sandnes, B. 2018 Pattern formation of frictional fingers in a gravitational potential. Phys. Rev. Fluids 3, 013801.
  18. [21]Forterre, Y. & Pouliquen, O. 2008 Flows of dense granular media. Annu. Rev. Fluid Mech. 40, 1–24.DOI
  19. [22]Freund, J. B. & Orescanin, M. M. 2011 Cellular flow in a small blood vessel. J. Fluid Mech. 671, 466–490.
  20. [23]Frey, P. & Church, M. 2011 Bedload: a granular phenomenon. ESPL 36, 58–69.DOI
  21. [24]Hampton, M. A., Homa, J. & Locat, J. 1996 Submarine landslides. Rev. Geophys. 34, 33–59.DOI
  22. [25]Houssais, M., Ortiz, C. P., Durian, D. J. & Jerolmack, D. J. 2016 Rheology of sediment transported by a laminar flow. Phys. Rev. E 94, 062609.DOI
  23. [26]Iordanoff, I. & Khonsari, M. M. 2004 Granular lubrication: toward an understanding between kinetic and fluid regime. ASME J. Tribol. 126, 137–145.
  24. [27]Iverson, R. M. 1997 The physics of debris flows. Rev. Geophys. 35, 245–296.DOI
  25. [29]Jop, P., Forterre, Y. & Pouliquen, O. 2005 Crucial role of sidewalls in granular surface flows: consequences for the rheology. J. Fluid Mech. 541, 167–192.DOI
  26. [30]Jop, P., Forterre, Y. & Pouliquen, O. 2006 A constitutive law for dense granular flows. Nature 441, 727–730.DOI
  27. [32]Landry, L. W., Grest, G. S., Silbert, L. E. & Plimpton, S. J. 2003 Confined granular packings: structure, stress, and forces. Phys. Rev. E 67, 041303.pubmed.ncbi.nlm.nih.gov/12786358
  28. [33]Lawal, A, Kalyon, D M & Yilmazer, U 1993 Extrusion and lubrication flows of viscoplastic fluids with wall slip. Chem. Eng. Commun. 122, 127–150.DOI
  29. [34]Le Bars, M. & Worster, M.G. 2006 Interfacial conditions between a pure fluid and a porous medium: implications for binary alloy solidification. J. Fluid Mech. 550, 149–173.DOI
  30. [35]Legros, F. 2002 The mobility of long-runout landslides. Eng. Geol. 63, 301–331.DOI
  31. [36]Marks, B., Sandnes, B., Dumazer, G., Eriksen, J. A. & Måløy, K. J. 2015 Compaction of granular material inside confined geometries. Front. Phys. 3, 41.DOI
  32. [37]Maurin, R., Chauchat, J. & Frey, P. 2016 Dense granular flow rheology in turbulent bedload transport. J. Fluid Mech. 804, 490–512.
  33. [38]Midi, G. D. R. 2004 On dense granular flows. Eur. Phys. J. E 14, 341–365.
  34. [39]Morgan, M. L., James, D. W., Barron, A. R. & Sandnes, B. 2020 Self-similar velocity profiles and mass transport of grains carried by fluid through a confined channel. Phys. Fluids 32, 113309.
  35. [40]Ouriemi, M., Aussillous, P. & Guazzelli, É. 2009 Sediment dynamics. part 1. bed-load transport by laminar shearing flows. J. Fluid Mech. 636, 295–319.DOI
  36. [41]Pailha, M. & Pouliquen, O. 2009 A two-phase flow description of the initiation of underwater granular avalanches. J. Fluid Mech. 633, 115–135.DOI
  37. [42]Paterson, L. 1981 Radial fingering in a Hele Shaw cell. J. Fluid Mech. 113, 513–529.DOI
  38. [43]Pennestrì, E., Rossi, V., Salvini, P. & Valentini, P. P. 2016 Review and comparison of dry friction force models. Nonlinear Dyn. 83, 1785–1801.DOI
  39. [44]Pouliquen, O., Cassar, C., Jop, P., Forterre, Y. & Nicolas, M. 2006 Flow of dense granular material: towards simple constitutive laws. JSTAT 2006, P07020.DOI
  40. [45]Putz, A., Frigaard, I.A. & Martinez, D.M. 2009 On the lubrication paradox and the use of regularisation methods for lubrication flows. Journal of Non-Newtonian Fluid Mechanics 163 (1), 62–77.DOI
  41. [46]Rettinger, C., Eibl, S., Rüde, U. & Vowinckel, B. 2022 Rheology of mobile sediment beds in laminar shear flow: effects of creep and polydispersity. J. Fluid Mech. 932, A1.DOI
  42. [47]Saffman, P. G. & Taylor, G. I. 1958 The penetration of a fluid into a porous medium or Hele-Shaw cell containing a more viscous liquid. Proc. Math. Phys. Eng. Sci. 245, 312–329.
  43. [48]Sandnes, B., Flekkøy, E. G., Knudsen, H. A., Måløy, K. J. & See, H. 2011 Patterns and flow in frictional fluid dynamics. Nat. Commun. 2, 288.DOI
  44. [49]Sandnes, B., Knudsen, H. A., Måløy, K. J. & Flekkøy, E. G. 2007 Labyrinth patterns in confined granular-fluid systems. Phys. Rev. Lett. 99, 038001.pubmed.ncbi.nlm.nih.gov/17678328
  45. [50]Sauret, A., Balmforth, N. J., Caulfield, C. P. & McElwaine, J. N. 2014 Bulldozing of granular material. J. Fluid Mech. 748, 143–174.
  46. [51]Taberlet, N., Richard, P. & Dellanay, R. 2008 The effect of sidewall friction on dense granular flows. Comput. Math. Appl. 55, 230–234.DOI
  47. [52]Tapia, F., Ichihara, M., Pouliquen, O. & Guazzelli, E 2022 Viscous to inertial transition in dense granular suspension. Physical Review Letters 129, 078001.DOI
  48. [53]Terzaghi, K. 1936 The shearing resistance of saturated soils and the angle between the planes of shear. In First international conference on soil mechanics, 1936, , vol. 1, pp. 54–59.
  49. [54]Thielicke, W. & Stamhuis, E. J. 2014 PIVLab – Towards User-friendly, Affordable and Accurate Digital Particle Image Velocimetry in MATLAB. Journal of Open Research Software 2.DOI
  50. [55]Thoren, L., Måløy, K. J., Flekkøy, E. G., Sandnes, B., Bourgoin, M. & Santucci, S. 2023 Capillary washboarding during slow drainage of a frictional fluid. Soft Matter 19, 9369–9378.
  51. [56]Zhang, D., Campbell, J. M., Eriksen, J. A., Flekkøy, E. G., Måløy, K. J., MacMinn, C. W. & Sandnes, B. 2023 Frictional fluid instabilities shaped by viscous forces. Nat. Commun. 14, 3044.

Paper details

vlnc:W_w38yec5a
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