Do not compare only one inlet and one outlet until every mass source, sink and accumulation term is accounted for. In a steady single-phase closed system, the net boundary mass flux should approach zero. In a transient or source-driven system, storage or source terms belong in the balance.
Problem: the continuity balance looks wrong
Engineers often discover mass imbalance while validating a pressure drop, flow split or thermal result. The inlet report might differ from the sum of outlet reports, or a solver continuity residual may stay high even when other monitors appear stable.
The magnitude alone does not identify the cause. The balance must be constructed for the actual physics and all relevant boundaries.
Write the conservation statement your model should satisfy
- Steady, single-phase, no volumetric mass source.
The algebraic sum of all boundary mass fluxes should tend toward zero.
- Transient model.
A temporary difference between inlet and outlet can be balanced by mass accumulation or depletion inside the domain.
- Compressible flow.
Density can change spatially and temporally; use mass flow rather than comparing volume flow blindly.
- Species, evaporation or reacting systems.
Total mass can still conserve while individual phase/species fluxes change through transfer or reaction.
- Porous, source-term or injection models.
Include any explicit mass source or sink in the control-volume balance.
Likely cause groups
Incomplete convergence
Pressure and velocity fields have not reached a state that satisfies continuity closely enough for the engineering purpose.
Missing boundary
A leak, vent, opening, porous jump, interface or unintended region boundary carries flux that is absent from the hand balance.
Reporting convention
Signs, normals, mass-flow averages or region selection make a correct balance appear inconsistent.
Transient accumulation
The system is filling, draining, compressing or expanding, so boundary flows should not yet match instantaneously.
Numerical instability
Oscillatory pressure-velocity coupling can create a wandering mass balance even if mean values look plausible.
Mesh/connectivity issue
Disconnected regions, unintended interfaces or poor cells can disrupt the expected control volume and convergence.
How to diagnose the imbalance
- Report every boundary mass flux with a consistent sign convention.
Build a table that includes all inlets, outlets, openings, interfaces that cross the control volume and source terms.
- Normalize the imbalance.
Use a ratio such as |Σṁ| divided by a representative inlet mass flow. An absolute error means little without system scale.
- Plot the normalized balance versus iteration or time.
A steadily decreasing error indicates convergence; periodic behavior may indicate physical or numerical unsteadiness.
- Check pressure and flow monitors at the same time.
If mass balance improves while the quantity of interest still drifts, the model is not ready just because continuity is small.
- Inspect topology and boundaries for hidden flux paths.
Unexpected open faces or region connections are common causes of unexplained flow loss.
- For transient cases, evaluate domain mass storage.
Compare the change in total domain mass over the timestep with the net boundary flux.
Fix order
- Correct the control-volume accounting.
Include every physical mass path and use consistent signs.
- Correct geometry, interfaces and boundary types.
Remove unintended leaks or disconnected flow regions.
- Improve convergence or temporal resolution.
For steady cases, continue or stabilize pressure-velocity coupling. For transients, ensure the timestep resolves the accumulation process.
- Repair local mesh problems in high-flux regions.
Bad cells at jets, outlets, narrow passages or interfaces can prevent a clean balance.
- Then assess whether the remaining imbalance is acceptable for the decision.
Conservation tolerance should be tied to the accuracy required in flow split, pressure, force or heat transfer.
Do not use one universal mass-imbalance threshold
A very small balance error is desirable, but the practical acceptance criterion depends on the quantity being reported. If two outlet branches differ by only a few percent, a comparable global mass imbalance can invalidate the flow-split conclusion even if the solver technically runs.
Use conservation as one convergence criterion together with residuals, stable engineering monitors and mesh/time sensitivity.
Related Abecator resources
Mass balance still unclear?
Submit the boundary list, mass-flow reports and convergence history without confidential geometry.