Thermal Management in Practice: Sense, Nonsense, and Things that Go Wrong.

There must be an ideal world
A sort of mathematicians paradise
Where everything happens
As it does in textbooks
Bertrand Russell

For many designers, real-life is more like hell than paradise. There is always a big discrepancy between the simplified textbook examples and the complexity encountered in practice. This is especially true in the thermal management of electronic parts and systems. This short course will address several topics, widely used, that cause trouble when trying to solve thermal problems in industry. We start with the discussion of the widespread misuse of heat transfer correlations. Generations of designers have been trained to use the correlations that fill the textbooks without proper regard for their limitations. “For convenience of engineering applications, correlation equations are developed over the entire range of dimensions and for any Pr-number". The course will look at several correlations and address the question: How useful are these correlations in real-life ?

Another topic is the use and, often, misuse of thermal analysis tools, especially Computational Fluid Dynamics (CFD). We strongly recommend using CFD because it can provide a lot of insight and reduce the number of costly redesigns but, there seems to be a widespread misunderstanding of the accuracy that can be reached in practice. We will talk about these issues from both the computational and the practical viewpoints.

The final topic is related to experimental issues: What accuracy can we expect from a heat transfer experiment ? How can we estimate the uncertainty in an experiment and use that estimate to help “clean-up” the experiment. How do errors arise, how can they be prevented ? For example, it is not unusual to find two experiments whose results disagree by 20% even though each claims to be accurate within 5%. What’s going on ? Sometimes the situations are subtly different but sometimes there were measurement errors that were not recognized. For example, most people use thermocouples to measure temperature but very few people know how to ensure that the thermocouple is measuring the right temperature, and that the signal is being properly handled (without adding error to it).

The course will discuss in detail how to reduce the difference between what you are measuring and what you wanted to measure.
Clemens Lasance, Philips Research Laboratories, Eindhoven, The Netherlands
Robert Moffat, Stanford University, USA