Skip to content

Supersonic boundary layers: boundaries and shock interaction ​

The compressible inflow and boundary toolchain that a shock/boundary-layer interaction needs, after the STREAmS solver (Bernardini et al., CPC 2021, 2023): Sutherland's viscosity law, a characteristic non-reflecting outflow and far field, an oblique-shock generator on a y face, and the wall-modelled interaction itself. The boundaries are verified against the kinds they replace and against the oblique-shock relations and the exact regular reflection; the interaction against the STREAmS DNS and the IUSTI experiment. It continues the compressible boundary layer of the turbulence group.

CaseTypeWhat it verifies
Non-reflecting boundaryP[bc_*] type = non-reflecting against the kinds it would replace, extrapolation and inflow: what an outgoing packet leaves in the box, the relaxation of a uniform overpressure, bit-for-bit equality with extrapolation where it must reduce to it
Shock generator and regular reflectionPthe shock of type = shock-generator against the oblique-shock relations, and its regular reflection off a slip wall against the exact reflection
Shock/boundary-layer interactionPthe wall-modelled interaction at Mach 2.28: the interaction length, the separation from Cf and the wall pressure against the STREAmS DNS and the IUSTI experiment; a 2:1 box under the interaction, the matching height halved: the dip of Cf halves, no mean separation; the pressure-gradient term: a bubble of 0.6δ0 against the DNS's 2.2, the compression still half an interaction length late
Sutherland viscosityP+Cthe law against the formula, both wall models under it, its input contract, the cold-wall boundary layer rerun under it

The non-reflecting and shock-generator legs are in src/tests/flume/verification/boundary/check.sh; their sg legs are distinct from the sg legs of the subgrid-scale models.

Sutherland viscosity ​

The law μ=μref(T/Tref)3/2(Tref+S)/(T+S) is in models and its keys in input. It has no page of its own; it is verified where it is used:

WhereWhat it verifies
unit tests, test_wall_model (leg wm2)the law against the formula (6.7⋅10−16); both wall models under Sutherland against their independent fine-grid references
input contract of the dissipative terms (DC)the Sutherland and kelvin refusals, S logged exactly from the kelvin anchor, two conversions through the [reference] layer
boundary layer, Mach 5.86the cold-wall layer rerun under Sutherland's law with the DNS's Reynolds number: Cf −3.4 % against −3.8 %, Bq −0.0190
shock/boundary-layer interactionthe interaction run uses Sutherland's law, S/T∞=0.776

The power and constant laws are unchanged bit for bit: with S=0 the law takes the operations of the power law.

How it was built ​

The toolchain is #121, which continues the compressible boundary layer of #119 T3.7.

PhaseWhatLegsPage
S1Sutherland viscosity ([physics] viscosity_law = sutherland), the kelvin anchor temperature_unit_kelvintest_wall_model (wm2); dissipation contract DCSutherland viscosity
S2Characteristic non-reflecting outflow and far field ([bc_*] type = non-reflecting)nr1–nr4Non-reflecting boundary
S3Oblique-shock generator on a y face (type = shock-generator)sg1–sg3 (boundary)Shock generator
S4aInviscid regular reflection off a slip wallsg4 (boundary)Shock generator
S4bWall-modelled interaction, Mach 2.28, IUSTIcampaign, no legShock/boundary-layer interaction
S4cS4b with a 2:1 box under the interactioncampaign (t 550, done)Shock/boundary-layer interaction
S4dThe equilibrium wall model with the pressure gradient (wall_model_pressure_gradient, wall_model_pressure_gradient_filter)wm12 (turbulence), test_wall_model (wm2); a run restarted from S4bShock/boundary-layer interaction
S4eS4b without the wall model's eddy viscosity below the matching height: the near-wall layer moves the onset, not the separationa run restarted from S4bShock/boundary-layer interaction
S4fS4d without the eddy viscosity below the matching height: the bubble shrinks, the onset movesa run restarted from S4bShock/boundary-layer interaction
S4gThe Bernoulli-consistent pressure gradient (wall_model_pressure_gradient_ypg): no separationwm2 (test_wall_model); a run restarted from S4bShock/boundary-layer interaction
S4hThe model fed with the resolved convection (wall_model_convection): the onset and Lint move toward the DNS, the zero-gradient Cf by +5 to +12 %wm2, wm13 (turbulence); a run restarted from S4bShock/boundary-layer interaction
S4iAnd the resolved stress removed, the velocity scale from τw (wall_model_resolved_stress): passes the zero-gradient layer; in the interaction τw<0 under a forward first cell, Lint 2.94wm13 (turbulence); a run restarted from S4bShock/boundary-layer interaction
S4jThe slow forcing (wall_model_velocity_scale = sum, the resolved stress removed, the forcing filtered over 100): the incoming layer within 2 % of S4b's, the interaction S4b's (Lint 3.12, Lsep 0.14)wm13 (turbulence); a run restarted from S4bShock/boundary-layer interaction
S4kThe 2:1 box of S4c with the matching height fixed at S4b's, 5/24δi (wall_model_height): the zero-gradient layer within 1.7 % of S4b's, Lint 3.26, a mean reversed first cell over 1.10δ0 under a positive τwa run restarted from S4cShock/boundary-layer interaction
S4lS4k with the viscosity coupling (wall_model_coupling = viscosity, #126): the wall stress separates in the mean over 1.66δ0, Lint 3.64 (10.3 % above the DNS's), the zero-gradient stretch 3.3 % above S4b'swm14 (turbulence); a run restarted from S4kShock/boundary-layer interaction
S5Recycling–rescaling turbulent inflow—not started
PagePhasesLegsType
Non-reflecting boundaryS2nr1–nr4P
Shock generatorS3, S4asg1–sg4 (boundary)P
Shock/boundary-layer interactionS4b–S4lcampaignP