This paper describes the techniques for validating dynamic thermal models devised by collaborating institutions in the United Kingdom. Following a review of past work on Imodel validation, the United States Solar Energy Research Institute (SERI) methodology was used as a starting point. Approximations and errors can arise at all stages of development, revision and use of a program. Emphasis was placed on thorough theoretical reviews of basic physical processes treated by programs and on the actual techniques adopted n some widely used programs. These were used to test the validation methodology. The main processes reviewed were internal longwave radiation, external convection, conduction and shortwave solar radiation. A new set of analytical tests for conduction was devised. These provide a sensitive tool for detecting errors and for examining the consequences of different numerical solution methods. A review of the treatment of internal longwave radiation uncovered major errors in some existing programs and a set of analytical tests was devised. These were applied to the algorithms in seven current programs and conclusons drawn as to their adequacy. The adequacy of common approximations in convection and solar radiation algorithms was examined and areas where the worst errors could arise identified. Past validation efforts have concentrated on inter-model comparisons and empirical validation studies. The experience gained from such work was reviewed and several new studies were conducted. In order to avoid past problems a great deal of attention was paid to the definiton of the specification of the building and its operation. One study was conducted by a single user with seven commonly used programs. Another study using a very simple structure was conducted as part of an International Energy Agency collaborative project (IEA Solar Task VIII). The latter led to a set of tests for evaluating and selecting programs. A detailed review of experimental data sets was conducted in order to assess their suitability for validating thermal simulation programs. Attempts were also made to devise improved statistical techniques for comparing measured and predicted quantities. Many of the existing datasets were not adequate for ustablishing the existence of program errors. The best dataset located was used to develop an empirical validation tool and this was tested using three programs. This paper highlights the main conclusions and describes relevant new activities such as the formation of a research/industry Club in the UK and a new IEA Annex 'The calculation of environmental performance of buildings1 (Buildinga & Community Systems Annex 21).
Evaluation procedures for building thermal simulation programs
Building Simulation, Vancouver, Canada, 1989, p. 217-222