Impose the quantities known from the real system and let CFD solve the unknown response. If upstream and downstream pressures are known, the flow rate should generally emerge from the pressure difference and system resistance. If mass flow is prescribed physically, use a flow boundary and allow pressure to respond according to the solver formulation.
Start with the physical control variables
Before selecting a solver menu item, draw the physical system and mark which quantities are actually controlled or measured. A fan curve, reservoir pressure, imposed pump flow, atmospheric discharge and measured mass flow describe different problems.
Pressure-driven system
Known upstream/downstream pressure states drive a flow through the hydraulic resistance of the domain.
Flow-driven system
A pump, controller or test condition imposes mass or volume flow and the required pressure difference becomes an output.
Mixed system
One boundary may prescribe flow while another provides a pressure reference or environmental condition.
The solver-specific implementation varies, but this physical distinction is fundamental. Boundary conditions should provide enough information to close the equations without specifying mutually inconsistent responses.
What a pressure boundary does—and does not do
A pressure boundary specifies a pressure state according to the solver's definition. It does not automatically guarantee a particular mass flow. The resulting flow depends on the pressure difference, density, viscosity, geometry, losses, wall treatment, compressibility and any momentum sources or rotating components.
If the calculated flow rate is unexpected, first check whether the imposed pressure values represent the same physical quantities as the reference data. Then verify system resistance before changing solver numerics.
Build an analytical pressure-loss baseline
Estimate friction and local losses for a comparable internal-flow path.
Flow boundaries are appropriate when flow is the known input
Mass-flow or velocity boundaries are useful when the physical system controls flow directly or when a test condition specifies it. The pressure field then adjusts to satisfy momentum and continuity together with the remaining boundary conditions.
Be careful with velocity profiles. A uniform velocity profile can be inconsistent with a developed duct flow, swirl or upstream device. For turbulence modelling, the velocity condition and turbulence quantities should describe the same inlet state.
- Confirm whether the specified value is mass flow, volume flow or velocity.
- Check density assumptions when converting between mass and volume flow.
- Define the direction/profile consistently with the physical inlet.
- Provide compatible turbulence and thermal quantities.
Static pressure and total pressure are not interchangeable
Static pressure is the local thermodynamic pressure. Total or stagnation pressure includes the kinetic contribution associated with bringing the flow to rest isentropically under the assumptions used by the formulation. In low-speed incompressible reasoning, the familiar dynamic-pressure scale ½ρU² helps show why the two are different.
A comparison can be wrong even when the CFD is correct if one side uses total pressure and the other uses static pressure. Always define pressure stations and pressure type before comparing pressure drop.
Check the solver's reference/operating pressure convention. A gauge value without the correct reference can shift the intended absolute state, which becomes especially important for compressible flow, density-dependent properties or phase-change problems.
Avoid over-constraining the same physical response
Problems arise when the setup tries to prescribe both the cause and the response inconsistently—for example, imposing a flow rate and also imposing pressure values that would require a different flow through the modeled resistance.
The exact mathematical constraints depend on solver formulation and boundary type, but the engineering rule is simple: do not prescribe more independent information than the physical system actually provides.
If a pressure-pressure setup gives the “wrong” flow, do not add a velocity condition just to force the desired value. Diagnose why the modeled resistance or pressure definition differs from the reference system.
Diagnostic workflow for internal-flow boundaries
- Define the physical system.
Identify reservoirs, fans, pumps, ducts, openings and measured/control variables.
- Define pressure type and reference.
Static, total, absolute and gauge pressure must not be mixed silently.
- Check characteristic scales.
Use Reynolds number, hydraulic diameter and expected velocity to establish the flow regime.
- Estimate the pressure-loss scale independently.
Compare friction and local losses with the CFD path.
- Check mass conservation.
Unexpected flow can be a reporting or leakage/interface problem rather than a boundary-value problem.
- Test boundary placement sensitivity.
Move boundaries away from strong recirculation, jets or developing regions where practical.
- Only then review numerical controls.
Numerical tuning should not compensate for inconsistent physical constraints.
Connect the full internal-flow check
Use Reynolds number, hydraulic diameter, pressure loss and validation in one workflow.
Common pressure-boundary mistakes
- Comparing total-pressure loss with a static-pressure difference.
- Using gauge pressure without checking the operating/reference pressure convention.
- Expecting a pressure outlet to impose the desired mass flow.
- Adding a flow constraint to force a pressure-driven model to match a target.
- Ignoring roughness, local losses, screens, porous resistance or hardware missing from the CFD geometry.
- Placing an outlet inside a strong recirculation region and treating reverse flow as a solver failure.
- Using incompressible pressure reasoning when density changes materially with pressure, temperature or Mach number.