In performance-based fire safety design of buildings, design fires are assumed based on relatively coarse information on potential combustible objects described by the object name, weight, size and main constituent materials. Then the fire effect is calculated by using assumed design heat release rate (HRR) to judge appropriateness of fire safety provisions. For this purpose, a simple estimation method of design HRR curve (whole time history of HRR) was developed by statistical correlations. The method is based on collection of available experimental datasets of various combustibles, which were categorized into two groups, wooden-combustibles and plastic-combustibles. To describe characteristics of HRR curves, parameters were correlated with properties of combustibles. Fire growth rate was correlated with bulk density of object. Maximum HRR and total heat release (THR) were correlated with surface area and weight, respectively. Fire decay rate was averaged over datasets in each group. By using the proposed methods, it is possible to establish a simple formula to represent full HRR diagrams of each category group. The accuracy of estimation was examined against whole datasets, which is in the order of two. By increasing the parameter values with one standard deviation, it is possible to make conservative estimation to include more than about 80 percent of experimental datasets.
