How does an engineer with thirty years of experience in fluid dynamics, heat transfer and CFD arrive at calling his website ViscousLimit?

For me, the answer begins at the boundary layer.

Throughout my career, I have repeatedly found that some of the most consequential engineering physics occurs in that remarkably thin region where a flowing fluid meets a solid surface. It determines friction and heat transfer, but also separation — and through separation can determine the behaviour of an entire flow field.

My interest in that interface began long before I was using 3-D CFD. It started in 1996, as a young aerothermal engineer at Rolls-Royce in Bristol, trying to understand how turbine components could survive in a gas stream considerably hotter than the metal from which they were made.

Over the years, the application has changed — gas turbine cooling, diesel-engine coolant flow, intake-port aerodynamics, diesel spray-combustion and piston-bowl flame-wall Interaction (FWI) — and the computational tools have become more powerful. But I repeatedly found myself returning to that same interface.

What began as an interest in predicting what happens at the wall became an interest in understanding how what happens at the wall can determine the behaviour of the entire flow field.

That is part of what I mean by ViscousLimit.

I don’t use the term as a formally defined fluid-dynamic quantity. I use it to describe a recurring modelling boundary: the point at which an engineering approximation that has made a calculation practical begins to suppress physics important enough to determine the answer.

Every CFD calculation is a model. Decisions about turbulence modelling, near-wall treatment, numerical schemes, spatial and temporal resolution and boundary conditions are necessary to make real engineering calculations tractable. Those approximations are not inherently undesirable — engineering has always depended upon them. The important question is whether we understand what has been approximated, and whether that approximation remains appropriate for the problem we are trying to solve.

Modern CFD has become extraordinarily capable. But greater computational throughput does not necessarily mean greater physical fidelity. Many assumptions originally adopted because of computational limitations remain attractive because they produce robust calculations and rapid engineering turnaround.

What physics did we throw away to make the calculation affordable — and is that still the right bargain?

Where does the convenient model meet the physics it can no longer safely neglect?

Those are the questions I want to explore here.