Use measured or physically justified turbulence quantities when available. If they are uncertain, quantify sensitivity rather than hiding the uncertainty behind a software default.
Problem: the solution is sensitive to inlet turbulence
Symptoms include too-fast or too-slow jet spreading, incorrect pressure recovery, different separation location, heat-transfer coefficients that disagree with test data, or strong changes when switching between intensity/length-scale defaults.
The effect decays downstream only if the flow has enough development length. Short domains, compact heat exchangers, turbomachinery and separated flows can remain sensitive.
Know what the inlet quantities mean
- Turbulence intensity.
Represents RMS velocity fluctuations relative to the mean velocity.
- Length scale.
Represents the size of energy-containing eddies and should relate to the upstream geometry or flow-generating mechanism.
- k.
Turbulent kinetic energy follows from velocity and turbulence intensity.
- ε or ω.
These set the dissipation timescale and depend on k plus a chosen length scale or turbulence timescale.
Convert inlet turbulence quantities consistently
Calculate k, ε and ω from velocity, intensity and length scale before applying them to the CFD model.
Choose physically defensible inlet values
Measurements are best when available. Otherwise infer turbulence from upstream hardware: pipe/duct development, screens, honeycombs, bends, valves, fans, compressors, contractions or environmental turbulence.
For internal flow, hydraulic diameter can help scale the length scale, but it is still an engineering assumption—not a universal law. For external aerodynamics, free-stream turbulence should reflect the test facility or atmospheric environment.
Diagnose whether inlet turbulence is driving the result
- Plot turbulence fields from inlet to region of interest.
Check whether inlet values persist or rapidly adjust.
- Run a bounded sensitivity study.
Vary intensity and length scale separately within plausible ranges.
- Track the engineering output.
Compare pressure drop, heat transfer, separation, mixing or force—not residuals alone.
- Check turbulence/wall-treatment consistency.
Inlet turbulence and near-wall resolution can interact through boundary-layer development.
- Review inlet location.
If uncertainty dominates, adding upstream development length may be more physical than tuning turbulence numbers.
Report turbulence uncertainty explicitly
If the inlet state is uncertain and materially affects the answer, document it as an input uncertainty. Validation against experiment should use the experimental inlet turbulence when possible rather than a generic software default.
For design studies, use one consistent turbulence convention across variants so geometry effects are not mixed with inlet-condition changes.
Common mistakes
- Using 5% intensity for every CFD problem because it is a familiar default.
- Specifying intensity but leaving an unrelated default length scale.
- Mixing k-ε and k-ω inlet conversions without consistent definitions.
- Tuning turbulence values until one result matches experiment without independent evidence.
- Ignoring the upstream devices that actually generate turbulence.
Related Abecator resources
Need help defining or challenging inlet turbulence?
Submit the upstream geometry, inlet velocity or flow rate, turbulence model and the result that appears sensitive to the inlet specification.