CFD simulation

CFD simulation (computational fluid dynamics) is the numerical calculation of liquid and gas flow and the heat transfer that goes with it. It solves the conservation laws of mass, momentum and energy (the Navier-Stokes equations) in a space divided into small cells. The results are velocity, pressure and temperature fields and the quantities derived from them, such as pressure drop, flow rate, drag force or component temperatures.

The workflow: a computational domain is created from the CAD model, meaning the space the fluid flows through (the inside of a pipe or enclosure, or the surroundings of a body). The domain is meshed, usually with refined layers of cells near the walls, where the velocity changes sharply. Then the fluid properties and boundary conditions are defined: velocity or flow rate at the inlet, pressure at the outlet, walls, heat sources and fan curves. Commercial codes mostly use the finite volume method.

Most engineering flows are turbulent, so turbulence is represented by a model. Industry relies mainly on RANS models (such as k-ε or k-ω SST), which deliver a time-averaged result in reasonable time. LES (large eddy simulation) is more accurate but an order of magnitude or more expensive to compute. Problems are solved as steady-state, or as transient when vortex shedding, filling or pulsation matter. Solving heat transfer in the fluid and in the solid parts together is called conjugate heat transfer (CHT).

Tools range from CAD-integrated modules (such as SolidWorks Flow Simulation) to dedicated software like ANSYS Fluent, Simcenter STAR-CCM+ or the open-source OpenFOAM.

CFD simulation pays off when flow or heat decides whether a product works and measuring a prototype would be expensive or slow: cooling of electronics and control cabinets, design of vents and ducts, pressure drop across valves, manifolds and filters, HVAC, mixing, aerodynamic drag or airflow around buildings. It lets you compare several shape variants and pick the best one before a prototype is built.

It also helps you understand problems on existing equipment, for example why one area overheats or where uneven flow and noise originate.

Colorful flow plots look convincing even when the simulation is wrong. Check convergence (falling residuals and stable monitored quantities such as pressure drop or temperature), mesh independence of the result and a domain large enough that the boundary conditions do not distort the solution. The turbulence model and the near-wall mesh resolution must suit the problem; otherwise pressure drop and heat transfer in particular tend to be wrong.

With natural convection cooling, do not forget radiation, which can carry a substantial share of the heat. Compare the results with a hand estimate or a measurement at one point at least. Ask the simulation provider for a description of the boundary conditions, turbulence model, mesh and convergence criteria, and for specific values, not just pictures.