The Validationof EM Modelling Codes. A User Viewpoint

Abstract

Computer codes for solving radiation and scattering problems have becomed powerful and widely available. The user of such a code must initially convince himself that his copy of the code executes correctly on his specific computer with his particular compiler. He must then build up expertise in constructing models for solution by the code which obey the restrictions imposed by the "modelling guildelines" accompanying the code in a user's manual. Users often want to solve "real" problems that do not conform to the problem geometry envisaged by the code writer. The user replaces the "real" structure with a computer model solvable by the code, and which obeys the modelling guidelines. The user must then carry out a "model validation" in which the computer model is tested against full-scale or scale model measurements. A successful "model validation" contributes to the user community's "experience base" and lends confidence to both the computer code and to the modelling process. Sometimes an unsuccessful attempt at modeling exposes a genuine limitation of the code. Then a new "modelling guideline" can be formulated to aid other users in avoiding the same difficulties. This paper reviews code development to highlight the origin of "modelling guidelines", and how they are extended by the user community. From the user's point of view, the "experience base" is augmented whenever a successful "model validation" is carried out and reported. Several examples are presented of the difficulties that may be encountered in computer modelling, and how such difficulties lead to further "modelling guidelines" aimed at aiding others in solving similar problems.