YOU - the user

In considering YOU as the user of TEXSTAN, we present a passage from the Preface of our graduate-level textbook Convective Heat and Mass Transfer (4th ed., McGraw-Hill, 2005, by William M. Kays, Michael E. Crawford, and Bernhard Weigand) to help reaffirm how important it is to study the benchmark datasets described in the external flows and internal flows sections of this website.

There are several challenges facing you, the graduate-level heat-transfer engineer, if you choose a numerical computer code as the tool appropriate for solving a given problem. You need to understand how the code works and you need to know what assumptions are embodied in the code. To some degree you need to know the trend of the answer before you begin to search for it (Professor Kays admonition to all of his graduate students). And as you choose a numerical code for solution you have to keep in mind the time and monetary cost of the choice. Often a simple time-saving and money-economical choice leads to an answer that is nearly the same as a more time- or money-cost intensive solution (the 10-minute analysis versus the 10-hour analysis).

There exists a strong analogy between numerical experimentation and physical experimentation, and the study of one enhances the effectiveness of performing the other. With either class of experimentation there are two essential requirements: the numerical (or) physical tool must be calibrated and the user of the tool must be "calibrated". And with either numerical experimentation and physical experimentation, the answers must be checked for plausibility, and an uncertainty analysis must be carried out to accompany the numerical or physical data. Calibrating the user can usually be carried out by having them perform benchmark experiments that yield data which can be compared against benchmark results. Ideally the class of benchmark results will be drawn from approximate or analytical methods as well as "accepted" numerical or experimental data. Convective examples might include reproducing fully-developed heat transfer data or Blasius/Falkner-Skan data, or their turbulent counterparts. The end result is that the user learns how to not misuse the tool. Calibrating the tool requires the user to understand the limitations of the tool, and in the case of a numerical tool, to understand what physics are incorporated into the code containing the numerics. This is an especially important check because for commercial codes, experience shows that what is advertised is not a guarantee of what has been correctly implemented into the code.

The end result of all your personal effort is that the you gain confidence about how the limits of the tool match the assumptions that underlie the model.

One of the grand masters of experimentation is Professor Peter K. Stein, a world-class expert on mechanical instrumentation. During his long career, he developed a unified approach to the engineering of measurement systems, and his belief is that one can and one must develop the knowledge to obtain valid data on purpose. We extend his philosophy to apply to the world of numerical convective heat and mass transfer, and it is in alignment with the ideas of calibration of the tool and the user (of the tool) before the tool is applied to convective problems. To this extent we believe the analogy between numerical experimentation and physical experimentation converges.

We encourage you to read the numerical accuracy section to help keep in mind that computer solutions (or numerical simulations) can be quite sensitive to grid and integration stepsize. This is why we have developed a set of TEXSTAN-generated initial conditions that help to guarantee grid independent solutions, and we encourage you to use them. For stepsize, the user can follow the guidelines laid out in the three example datasets for the laminar boundary layer, laminar pipe flow, and the turbine blade.