Fluid constraints
Fluid constraints act on velocity and the pressure gauge. No-slip enforces ; slip enforces with zero tangential traction; a pressure reference fixes the additive pressure nullspace without by itself defining an outlet flux law.
| Kind | Required fields | Meaning |
|---|---|---|
fluid.velocity | target, three-component value in m/s or vector velocity signal | Prescribe the full velocity vector |
fluid.moving_wall | target, three-component value in m/s or vector velocity signal | Prescribe the full wall-velocity vector |
fluid.no_slip_wall | target | Match velocity to a stationary wall |
fluid.slip_wall | target | Prevent normal flow while permitting tangential slip |
fluid.symmetry | target | Apply a symmetry-plane condition |
fluid.pressure_reference | target, value | Establish the pressure datum |
fluid.periodic | periodic pair/transform | Pair periodic flow boundaries Not yet releasedStatus |
signal inlet_speed : velocity over constant {
value = 1.5 m/s
}
excitation inlet_velocity {
inlet_velocity {
target = geometry.channel.region.inlet
signal = signal.inlet_speed
direction = global([1, 0, 0])
kinematics = fixed_frame
}
}
constraint channel_walls {
fluid.no_slip_wall { target = geometry.channel.region.walls }
}
constraint pressure_datum {
fluid.pressure_reference {
target = geometry.channel.region.outlet
value = 0 Pa
}
}How do I prescribe an inlet velocity profile?
fluid.velocity and fluid.moving_wall require exactly three signed velocity components.
Inline components are explicitly global coordinates, for example:
constraint inlet {
fluid.velocity {
target = geometry.channel.region.inlet
value = [-1.5 m/s, 0 m/s, 0 m/s]
}
}value = signal.<name> instead retains a declared three-component velocity signal, including
its frame and constant/table representation. It is not converted to a speed, maximum magnitude,
or guessed axis. The formulation must explicitly support realizing that signal at the boundary;
unsupported frames or histories are rejected. Scalar constants and scalar signals are not vector
constraint inputs. The reduced channel adapter cannot realize these exact vector constraints and
rejects them; this shared input contract alone does not qualify a numerical flow solver.
For a scalar speed with an explicitly declared direction, the shared excitation form is:
signal inlet_speed : velocity over constant {
value = 1.5 m/s
}
excitation uniform_inlet {
inlet_velocity {
target = geometry.channel.region.inlet
signal = signal.inlet_speed
direction = global([1, 0, 0])
kinematics = fixed_frame
}
}
load_case operating { include excitation.uniform_inlet }direction is a dimensionless global vector. The signal carries m/s and may instead be defined over
time when a transient procedure consumes it. Use mass_flow when the total flow is known but its
spatial profile is not, when the selected formulation supports conservative realization.
The runtime velocity-constraint constructors likewise require a cell vector of three typed
fea.Velocity values (global components), or a framed three-component fea.VelocitySignal
also retained in the model signal inventory. They do not reinterpret one scalar velocity value
as a vector. A pressure reference remains a gauge, not a physical outlet boundary; physical
outlet pressure uses the existing fluid.outlet_pressure boundary operator.
How do I prescribe only the normal inlet speed?
Use the signal-driven inlet_velocity form above when scalar speed and a global direction are the
controlled inputs. For a curved inlet, one global direction is not a uniform local-normal field.
The current source surface has no normal(...) direction or spatial-profile expression; use an
admitted topology field/generator when the selected formulation provides one. Otherwise that case
is not expressible by the current typed input and must be rejected, not approximated implicitly.
How do I apply a developed inlet profile?
The directed excitation is uniform. A parabolic, turbulent, measured, or mapped profile must enter through a typed topology field or function that preserves coordinates, units, and interpolation. RunMat does not infer “fully developed” from a region name. See fluid loads.
How do I model a moving wall in CFD?
constraint moving_lid {
fluid.moving_wall {
target = geometry.cavity.region.lid
value = 1 m/s
}
}This is the exact current constraint shape: it carries a scalar speed and no authorable direction
or frame. Use it only when the consuming formulation gives that scalar an unambiguous wall-motion
meaning. For directed or time-varying translation, use a wall_motion excitation with a scalar
velocity signal and direction = global([x, y, z]). For motion supplied by a structural solution,
use an FSI interface instead of duplicating the displacement history manually.
How do I model a rotating no-slip wall?
The wall field must satisfy . Neither
the scalar fluid.moving_wall constraint nor a single-direction wall_motion excitation expresses
that spatial law. A rotating-wall generator with axis, origin, angular speed, and frame is not in
the current closed source surface. Use a rotating-domain/formulation input that explicitly admits
the law; otherwise rotating-wall kinematics are unsupported.
How do I apply a no-slip wall?
constraint viscous_walls {
fluid.no_slip_wall { target = geometry.channel.region.walls }
}No-slip imposes zero wall-relative velocity and is appropriate for ordinary viscous solid walls.
Check that inlet/outlet faces are excluded from a multi-face walls region; overlapping conditions
can overconstrain boundary DOFs.
How do I model a stationary solid wall in a moving reference frame?
The current scalar wall constraint cannot author a rotating or translating coordinate frame. “Stationary” in the laboratory frame can be nonzero in a rotating frame, so a moving-reference-frame procedure must own that transformation explicitly. Do not infer a frame from the CAD orientation or the region name.
How do I apply a slip wall?
constraint ideal_slip_wall {
fluid.slip_wall { target = geometry.channel.region.slip_faces }
}Slip blocks normal velocity while allowing tangential motion. Use it for a deliberate inviscid or symmetry-like wall approximation, not as a convergence workaround for a viscous boundary layer.
How do I model an inviscid wall or centerline?
Use fluid.slip_wall for an impermeable inviscid wall where tangential traction is zero. Use
fluid.symmetry for an actual symmetry plane. They can produce similar local velocity restrictions,
but symmetry also asserts mirrored geometry, loading, and solution.
How do I apply a fluid symmetry plane?
constraint centerplane {
fluid.symmetry { target = geometry.half_channel.region.symmetry_face }
}The geometry, inlet profile, other boundaries, and expected solution must all be symmetric. The reported flow/resultants correspond to the modeled fraction and must be scaled accordingly.
How do I set the pressure datum in incompressible flow?
fluid.pressure_reference sets the additive pressure gauge:
constraint pressure_reference {
fluid.pressure_reference {
target = geometry.channel.region.reference_face
value = 0 Pa
}
}It is not automatically an outlet law. Use fluid.outlet_pressure for a natural outlet boundary.
Do not add multiple incompatible references to suppress a singular matrix.
Where should I place the incompressible pressure reference?
Place it on a physically meaningful reference region admitted by the formulation, usually associated with a pressure outlet or known datum. The value fixes only the additive gauge. Results in pressure difference and velocity should be invariant to a consistent datum shift; absolute thermodynamic pressure for density/material laws remains a separate convention.
Why is pressure still singular after adding a reference?
Check disconnected fluid components, empty/unresolved targets, and periodic components not connected to the referenced region. Every independent pressure nullspace needs one admitted datum. Adding several overlapping references to one component can instead overconstrain it.
How do I model periodic flow through a repeating passage?
constraint repeating_passage {
fluid.periodic {
primary = geometry.cell.region.inlet_pair
secondary = geometry.cell.region.outlet_pair
translation = [50 mm, 0 mm, 0 mm]
tolerance = 1 um
}
}Plain periodicity maps velocity/pressure fluctuations across the pair. A macroscopic pressure drop or mass-flow driving condition remains a separate physical input; periodicity does not invent it.
How do I model rotationally periodic turbomachinery passages?
Use fluid.periodic with a complete rotational transform when the formulation admits sector
periodicity. The primary-to-secondary mapping needs origin, axis, angle, and compatible face
topology. Rotor/stator interfaces with different pitch or relative motion are not ordinary periodic
pairs and require a dedicated interface.
How do I model a free surface, porous wall, or outlet backflow?
These are not aliases for the essential constraints on this page. A free surface needs a multiphase/free-surface law, a porous wall needs normal resistance or mass transfer, and outlet backflow belongs to the outlet formulation with transported-state data. Use the corresponding typed physics contract when admitted rather than weakening a no-slip or velocity constraint.
Why is the velocity field overconstrained at an inlet edge?
An inlet face and adjacent no-slip wall share edge nodes. The formulation must reconcile the inlet profile with zero wall-relative velocity; a uniform nonzero profile has a discontinuity at the corner. Use a compatible developed profile or documented weak enforcement and inspect localized residuals. Do not assign duplicate kinematic owners to the same boundary entities.
What should I inspect after applying fluid constraints?
Inspect exact target faces, wall/inlet overlap, excitation direction, normal flux, pressure datum ownership, periodic pair coverage, mass conservation, wall-relative velocity, and constraint residuals. In the Desktop preview, verify motion vectors and periodic pair annotations before the solve.
An outlet-pressure natural boundary is documented under fluid loads. Choose one pressure datum consistent with the formulation and avoid simultaneously prescribing incompatible velocity and pressure data on the same boundary.
How do I combine an inlet, walls, and outlet in one RunMat fluid model?
Use a directed inlet_velocity excitation on the inlet, fluid.no_slip_wall on stationary viscous
walls, one physical pressure reference when required, and fluid.outlet_pressure as the outlet
natural boundary. Give each role a nonoverlapping CAD region:
signal bulk_speed : velocity over constant { value = 2 m/s }
excitation pipe_inlet {
inlet_velocity {
target = geometry.pipe.region.inlet
signal = signal.bulk_speed
direction = global([1, 0, 0])
kinematics = fixed_frame
}
}
constraint pipe_wall {
fluid.no_slip_wall { target = geometry.pipe.region.wall }
}
constraint outlet_gauge {
fluid.pressure_reference {
target = geometry.pipe.region.outlet
value = 0 Pa
}
}
load outlet_traction {
fluid.outlet_pressure {
target = geometry.pipe.region.outlet
value = 0 Pa
}
}
load_case operating {
include excitation.pipe_inlet
include load.outlet_traction
}The pressure reference removes the algebraic gauge; the outlet load supplies the physical natural boundary. Whether both are needed on the same outlet is formulation-dependent and must be admitted explicitly. Before solving, require zero unclassified boundary area and no overlap among inlet, wall, and outlet regions. After solving, check against the study tolerance, report the pressure drop, and compare wall shear/resultant with a mesh-refined run.
Should I use a slip wall or a symmetry plane?
Both suppress normal flow under common formulations, but symmetry also constrains mirrored field
behavior and is justified only when geometry, loading, and expected solution are symmetric. A slip
wall represents an impermeable boundary without tangential no-slip shear. Use fluid.symmetry for a
cut model and fluid.slip_wall for a physical inviscid or slip boundary; inspect tangential traction
and normal flux.
How do I model wall motion from a prescribed mechanical motion?
Use scalar fluid.moving_wall only when the consuming formulation defines its wall-speed direction
unambiguously; use directed wall_motion for supported prescribed translation. Use
fluid_structure.no_slip when structural displacement and velocity are solved and must feed back
with fluid traction. Applying both to the same face creates two owners for wall kinematics unless the
coupling procedure explicitly defines one as an override.
How do I constrain a fluid domain with several disconnected cavities?
Each disconnected incompressible pressure component needs its own datum or a boundary relation that establishes pressure. One reference in cavity A cannot remove the constant-pressure nullspace in cavity B. Inspect connected components after interfaces and periodic pairs resolve; do not scatter arbitrary point references until the linear system happens to factor.
What fluid constraint evidence should be archived?
Retain resolved boundary areas and normals, excitation directions, wall-relative velocity, pressure-reference ownership by connected component, periodic mapping coverage, constraint residuals, and signed mass flux. For moving walls, also retain the motion signal or producing FSI state identity.
How do I choose CFD boundary conditions that are mathematically well posed?
For each connected incompressible domain, prescribe enough velocity/flow information to drive the problem and exactly one independent pressure datum or outlet relation to remove constant-pressure freedom. Classify every remaining face as wall, symmetry, periodic, opening, or coupled interface. Do not prescribe all velocity components and pressure on the same boundary unless the formulation explicitly defines that mixed condition.
Run the boundary audit before meshing: report face coverage, overlaps, unclassified faces, connected fluid components, and pressure-reference ownership. After solving, check signed mass conservation.
How do I model a stationary pipe or duct wall?
Use fluid.no_slip_wall for viscous stationary walls. Use fluid.slip_wall only for an inviscid or
intentional zero-normal-flow approximation. A symmetry plane also imposes symmetry on transported
fields and gradients; it is not merely a low-friction wall.
Near-wall mesh and turbulence treatment remain formulation choices. A no-slip declaration does not prove that wall shear or heat transfer is resolved; retain or the applicable near-wall metric, wall force, and mesh-convergence evidence.
How do I constrain a rotating or translating reference-frame model?
Use a rotating-domain formulation whose frame owns Coriolis/centrifugal terms. The current scalar
fluid.moving_wall syntax has no frame, axis, or origin and therefore cannot by itself establish an
inertial-frame rotating wall. Do not combine a rotating frame and an independently prescribed
absolute wall motion without an explicit supported transformation. At an interface between frames,
use the admitted sliding/mixing relation rather than a generic periodic pair.
Archive frame identity, axis, origin, angular speed, relative wall velocity, and torque/power. Check that a wall intended to be stationary relative to the rotor has zero relative velocity.
How do I constrain an internal-flow periodic cell?
Pair inlet and outlet geometry with fluid.periodic, declare the exact translation/rotation, and
supply the driving pressure-gradient, body-force, or mass-flow contract separately. Periodicity only
relates fields; it does not create pressure drop or flow.
Validate pair coverage and transformed normals, then compare mass flux and transported quantities on both faces. The pressure decomposition (periodic fluctuation plus imposed mean gradient) must belong to the selected formulation and be visible in evidence.