Much of today’s finite elements-based software for electromagnetic field computation has been developed on an ad hoc basis, often as an outgrowth of ongoing research whether at universities or agencies like NASA. “Design” is seen as the execution of computational algorithms yielding the computed design, rather than the holistic process that begins with the design of the software itself and ends in the computations using that software. As a result finite element software product development has not been informed by the benefits of starting with a rigorous requirements analysis and going through the normally mandated design process in any rational software engineering development process. This paper examines such a development and identifies the various benefits that arise from that process. Of particular note is the ability to have a list of software components from which we pick the most appropriate to the problem being addressed – thereby giving users choice to suit their particular computational environment, be it in methods or hardware; and the ability to implement a model of the design using UML and then transform that into source code using modern facilities. The latter makes porting to new programming languages easy and ensures that the finite element code is proper rather than simply working. Reengineering legacy software by transforming existing code to UML for analysis becomes possible. And the planning document leads to all goals being realised and is easily transformed into a user manual.

Keywords: Electromagnetic field, Finite element, UML