Engineering & Technology · Fluid Mechanics, Flow Systems & Machines

Turbulence, compressibility and model limits

A display showing constant average flow can hide rapid local changes. A gas flowing through a duct can also change volume flow while conserving mass. “Steady,” “laminar” and “incompressible” describe different aspects of a model and should be checked separately.

Review force vectors, energy and power, control volumes and steady-flow thermodynamics, SI units and measurement uncertainty. Be comfortable with ratios, areas, square roots and reading graphs. Symbols are defined where used: ρ is density, μ dynamic viscosity, Q volume flow and ṁ mass flow. For the constructed liquid cases, use g = 9.81 m/s² where stated. Prior physics study of fluids is helpful but not required.

Back to the chapter

After reading, you can

Distinguish turbulence from general unsteadiness and recognize when density changes require a compressible-flow model.

A mean flow is not every instantaneous flow

Laminar flow has organized motion; turbulent flow contains irregular three-dimensional fluctuations and a range of interacting scales. Laminar flow can vary with time, and a turbulent flow can have a statistically steady mean. A deliberately oscillating laminar flow is therefore not made turbulent simply by being unsteady.

Represent a measured velocity as u(t) = U + u′(t), where U is a mean defined over a stated interval and u′ is its fluctuation. Fluctuations can transport momentum and heat even when their mean is zero. Averaging an equation containing products of velocities introduces correlations that a mean-velocity field alone cannot determine. An adequate sampling interval and rate are part of the measurement definition; a short average may miss slower changes.

Turbulence models represent unresolved effects

Reynolds-averaged Navier–Stokes models, abbreviated RANS, add modeled effects of turbulent fluctuations to equations for averaged quantities. Large-eddy simulation, LES, resolves larger changing structures and models unresolved smaller scales. Direct numerical simulation, DNS, aims to resolve all relevant turbulent length and time scales without a turbulence closure. Each still needs suitable initial and boundary conditions and numerical accuracy.

These approaches make different demands and answer different questions. A visually detailed flow animation does not prove that its stresses or pressure losses are accurate. Refining a mesh reduces some numerical errors when the method behaves properly, but does not automatically fix an inappropriate turbulence model. Validation must address the intended flow regime, geometry and quantity of interest.

Compressibility is a separate decision

All real fluids respond to pressure, but an incompressible approximation neglects density changes where their effects are sufficiently small. For a gas, Mach number M = U/a, using local sound speed a, helps assess changes associated with motion. Subsonic means M < 1; it does not mean density must be constant. Temperature change, strong pressure variation or composition change can matter even at low speed.

When density varies, steady continuity preserves ṁ = ρAU, not necessarily Q = AU. A compressible analysis couples mass, momentum and energy with fluid properties. Near sonic conditions, wave propagation and possible choking need additional treatment. Shock waves can create abrupt changes and entropy generation. A liquid system can also require compressibility when studying fast pressure waves, despite nearly constant density in a slow-flow calculation.

Horizontal axis: sample time, 0 to 1.9 s in 0.1 s steps. Vertical axis: local velocity, 7 to 13 m/s. The blue line connects 20 invented samples between 8 and 12 m/s. Their arithmetic mean is exactly 10 m/s, shown by a dashed orange line. A constant mean does not imply a constant instantaneous velocity. The connected samples are illustrative, not measurements of turbulence.

Constructed density-change check, without claiming a complete duct solution

Take two hypothetical uniform gas-flow sections with the same area 0.001 m² and steady mass flow 0.12 kg/s. If density at section 1 is 1.2 kg/m³, its speed is 0.12/(1.2 × 0.001) = 100 m/s and volume flow is 0.10 m³/s. If section 2 density is 0.60 kg/m³, speed must be 200 m/s and volume flow 0.20 m³/s. Equal volume flow would violate mass conservation.

For an illustrative equal local sound speed of 340 m/s at both sections, the Mach numbers are approximately 0.294 and 0.588. These are continuity and ratio checks, not proof that a particular adiabatic duct can realize the states. A complete solution must also satisfy momentum, energy, an equation of state and the stated thermal boundaries. The separate invented trace illustrates how instantaneous values can vary despite a constant sample mean.

Try the reasoning

Does a zero mean fluctuation imply that fluctuations have no mechanical effect?
Show the answer and reasoning
No. Positive and negative fluctuations can cancel in the mean while squared terms and correlations remain nonzero. Those terms affect transport and mean-flow balances.
Would doubling grid density establish that the turbulence model is valid?
Show the answer and reasoning
No. It can help assess numerical sensitivity, but physical model error and uncertain boundaries remain. The intended predictions need independent comparison and uncertainty assessment.
Must an M = 0.2 heated gas have constant density?
Show the answer and reasoning
No. Low Mach number may reduce motion-related compressibility effects, while heating can still cause substantial density change. The approximation must match the actual conditions.

What to carry forward

Check time dependence, turbulent fluctuations and density variation as separate modeling choices.

References

Teaching resources and research supporting this lesson’s explanations and examples.