The argument is advanced that test cells are outdated for most purpose
s in building thermal design. Comparisons of one energy conservation t
echnique with another are extremely dependent on the local weather. Be
cause very small test-boxes are irrelevant, and reasonably sized test
cells not transportable, simulation is seen as the vehicle to answer c
omparative questions. Computational fluid dynamic codes are in the pos
ition of thermal models 10 years ago. But they will gain preeminence o
ver test rooms for the design of airflow and, particularly, natural ve
ntilation systems. Component testing in scale models is seen as insuff
iciently sophisticated for most purposes. The vital areas where test c
ells remain essential, is for the improvement of model algorithms, the
calibration of simulation models and the reinforcement of users' conf
idence in such models.