TABLE OF CONTENTS The section of Chapter 1 entitled 'When Not to Model' drew attention to the difficulties inherent in modeling highly complex systems and the need in these cases for at least some experimentation to derive the desired information. Heat and mass transfer coefficients in turbulent flow were two cases in which our advice was to set aside modeling, at least in part, in favor of experimentation. It was also pointed out that this task can be considerably eased by combining the pertinent physical parameters of the system into dimensionless groups, thus reducing the number of variables to be dealt with.
Suppose, for example, that six physical parameters affect the behavior of a system. In the case of heat transfer in turbulent flow in a pipe, these might be flow velocity v, pipe diameter d, and the physical properties of the fluid (density, heat capacity, viscosity, and thermal conductivity). Suppose further that we wish to run tests at 10 values of each parameter. The total number of experiments required would then amount to 10 6. If, on the other hand, we were able to combine the parameters into only three dimensionless groups, the number of tests would drop to 10 3. An experienced graduate student, running one test a day, would thus require only 3 years to complete his Ph D. Without the use of dimensionless groups, this number would balloon to 3000 years. This process of expressing the behavior of a physical system in...
