This can be read as a piece written by a firm against its own interest. It is still the truth [1] [2].
When CFD is not needed
When a standard correlation exists. Pressure loss in a straight pipe, single phase, fully developed — Darcy-Weisbach and the Moody chart give the answer in seconds, and more reliably than CFD, because they rest on experiment.
When the equipment is catalogued and the duty point is inside the catalogue range. The manufacturer's curve has already been measured.
When the decision is not sensitive to the result. If you would make the same decision even with a 20% error, a more precise number has no value.
Where CFD genuinely earns its cost
When the geometry is non-standard. Correlations do not hold outside the geometry family they were derived from.
When the field distribution matters. If what matters is not the average but where things happen — hot spot, separation region, dead volume — CFD has no substitute.
When measurement is impossible or very expensive. Flight conditions, high temperatures, inaccessible internal volumes.
When many alternatives must be compared. Comparing ten designs in a model rather than building and testing them.
When you are looking for the cause of a problem. Measurement tells you the result; CFD shows you the reason.
The grey zone between
Most problems sit between the two. The right approach there is to do a hand calculation first to establish the order of magnitude, then use CFD to confirm that order and reveal the detail. If hand calculation and CFD differ by an order of magnitude, that must be understood first — and the error is usually in the CFD.
References
- Ferziger, J. H., Perić, M. & Street, R. L. (2020). Computational Methods for Fluid Dynamics, 4th ed. Springer. DOI: 10.1007/978-3-319-99693-6
- Shah, R. K. & Sekulić, D. P. (2003). Fundamentals of Heat Exchanger Design. John Wiley & Sons. DOI: 10.1002/9780470172605