Most studies on fire properties, including smoke control, have been based on simulations. However, the results of simulation experiments have an average error rate of approximately 36.7%, and data with low reliability are obtained, compared to theoretical values and actual-scale measurements. In addition, experimental studies using scale models exhibit an average error rate of 6.9%, significantly lower than those of simulation experiments. In this study, a method for decreasing the error rate of simulation experiments was investigated by constructing a database using scale-model smoke control experiments.