Stokes's First Problem (Impulsively Started Wall)

verification

Diffusing shear-layer velocity field
Streamwise-velocity field at the final time t=0.2 s, showing the shear layer that has diffused upward from the impulsively started wall into the fluid at rest above.

Overview

Stokes's first problem, also called the Rayleigh problem, is the textbook unsteady viscous flow that develops when an infinite plane wall bounding a fluid at rest is impulsively set into steady motion in its own plane. Momentum diffuses away from the wall into the fluid, and a shear layer of thickness δ4νt grows in time. Because there is no pressure gradient and the flow is independent of the streamwise coordinate, the problem has a closed-form self-similar solution: the velocity profile is the complementary error function of a single similarity variable, so profiles taken at different instants collapse onto one curve. The case exercises the unsteady (implicit) time integration coupled to the wall-normal diffusion operator, with no convective or pressure-gradient terms. The reference solution is the one given by Schlichting & Gersten.

Problem Setup

The domain is a tall, thin slab above the moving wall: a streamwise length L=1 m and a wall-normal height H=2.5 m, meshed as a single spanwise layer of cells so the simulation is quasi-two-dimensional. The fluid is initially at rest everywhere. At t=0 the lower wall begins translating in its own plane at the constant speed U0=1 m/s, imposed through the solver's moving-frame mechanism so the no-slip wall carries the fluid with it without any normal motion. The streamwise faces are a translational periodic pair, which removes any dependence on x; the top boundary is a symmetry plane placed far above the wall so the diffusing shear layer never reaches it over the simulated interval, matching the quiescent far field; and the spanwise faces are symmetry planes. The fluid has constant density ρ=1 kg/m³ and constant viscosity μ=1 Pa·s, giving a kinematic viscosity ν=1 m²/s. The second-order implicit time scheme advances a fixed step dt=103 s to a final time tmax=0.2 s, at which the shear-layer thickness δ1.8 m is still well inside the domain.

Quantities of Interest

The verification quantity is the streamwise velocity profile u(y,t). With no pressure gradient the streamwise momentum balance reduces to the one-dimensional diffusion equation ut=ν2uy2, subject to u=U0 at the wall and u0 far from it, starting from rest. Its exact self-similar solution is u(y,t)U0=erfc(η),η=y2νt, where erfc is the complementary error function. The computed profile is sampled at three physical times (t=0.05,0.1,0.2 s) and plotted as u/U0 against the similarity variable η=y/(2νt), collapsing onto the single erfc(η) curve. The computed u/U0 at the similarity stations η=0.5 and η=1 is reported beside its exact erfc(η) values; the two agree within 0.1% of the analytic erfc solution.

Sources

H. Schlichting and K. Gersten, Boundary-Layer Theory, 8th ed., Springer-Verlag, Berlin Heidelberg, 2000.: https://doi.org/10.1007/978-3-642-85829-1
Stokes's first problem (the Rayleigh problem): the impulsively started wall has the exact similarity solution u(y,t)/U0=erfc(y/(2νt)).

Results

Similarity collapse of the velocity profile

Similarity collapse of the velocity profile
Computed streamwise velocity u/U0 at three physical times, each plotted against its own similarity variable η=y/(2νt); the profiles collapse onto the single analytic erfc(η) curve.

Reference: H. Schlichting and K. Gersten, Boundary-Layer Theory, 8th ed., Springer-Verlag, Berlin Heidelberg, 2000. (similarity solution u/U0=erfc(η)): https://doi.org/10.1007/978-3-642-85829-1

Similarity-solution comparison

Sourceu/U0 at η=0.5u/U0 at η=1
Luminary 0.479510.15732
erfc(η) (Schlichting & Gersten) 0.47950.1573

The computed u/U0 at two similarity stations at the final time beside their exact erfc(η) values.

Solver configuration

features exercised
Verification2DLaminarConstant densityUnsteady