Experimental fluid mechanics: the impact of modern instrumentation

Abstract

The importance and, in fact, the very existence of any experimental approach had to be defended in former days against the extreme theoretical epistemologists like Eddington. Presently experimental science has to come to grips with the computer and a new set of extremists; in every field of physics, except elementary paitides, the forces and governing equations are known and hence in principle all observed and applied physical phenomena are computable, given only sufficient intelligence and a sufficiently dimensioned computer. All problems in fluid mechanics as a largely classical discipline should therefore be computable from Navier-Stokes or similar equations or even from general molecular Boltzmann-like equations. This extreme point of view which has been put forward on occasions is clearly naive and, at least on the basis of today’s facts, easily contradicted. Nobody as yet has succeeded in computing the flow past a sphere over the full range of Reynolds numbers and the chance of a purely theoretical or computational prediction of the effect of the addition of minute quantities of polymers on the turbulent skin friction of water is rather small. One of the most important numerical computations of former days is the 1937 Taylor-Green work on the vorticity cascade; an order of magnitude progress with the same problem is even today anything but easy. It is not difficult to add further counter-examples to the list even without using “strange” fluids like plasmas, ferro fluids and superfluids. Consequently, I start with the assumption that experimental fluid mechanics is going to be a vital part of mechanics and physics for some time to come. However, the impact of the extremely rapid developments in instrumentation and computers has a lasting influence on the approach to and the scope of experiments, and the attitude of an experimentalist to a problem will change with it.