2021 Volume 77 Issue 4 Pages I_467-I_478
In a large-scale disaster, comprehensive collection of damage information becomes difficult, therefore, the prospect of recovery process of lifelines is inevitably based on limited information. In this study, fundamental analysis was performed for decreasing processes of electric power outage in order to develop a sequential update model of recovery prediction. Firstly, the decreasing process of electric power supply associated with major disasters in Japan was compared using a mixed exponential distribution model which divides the whole process into short- and long-term effects. Next, the sequential update of the prospect of recovery was performed by applying an exponential function to the trend of recovery process at an arbitrary point of time. On this basis, time required for 80, 90 and 95% recovery of power outages was estimated and compared with the observed values. The results suggest that appropriate setting of time interval for curve fitting to the recovery trend is required for rapid and stable update of prediction.