A demonstration of the effectiveness of inter-program comparative testing for diagnosing and repairing solution and coding errors in building simulation programs

The validation of a building simulation program or model is a daunting task, and one that should receive as much attention as algorithm and code development. Previous research in this field has led to a well-accepted approach composed of analytical verification, empirical validation and inter-program comparative testing to diagnose model deficiencies, mathematical solution errors and coding errors. Through a case study using a model for predicting the thermal and electrical performance of fuel cell micro-cogeneration devices, this article demonstrates the utility of the inter-program comparative testing validation construct. It shows that by comparing program-to-program results, solution problems, coding errors and deficiencies in mathematical model descriptions can be efficiently identified, diagnosed and subsequently repaired.