Structural Safety, 12 (1993) 1-19 1 Elsevier Earthquake damage estimation and decision analysis for emergency shut-off of city gas networks using fuzzy set theory * Laurence Cret, Fumio Yamazaki, Shigeru Nagata and Tsuneo Katayama Institute of Industrial Science, The University of Tokyo, Roppongi 7- 22-1, Minato- ku, Tokyo 106, Japan Abstract. Earthquake damage estimation for lifelines can be used for many purposes, for example, planning repair works or simulating network recovery. For gas networks, however, the issue is more crucial, as secondary disasters due to leaks are possible. If necessary, the gas supply should be interrupted in heavily-damaged areas. Therefore, emergency shut-off decision is vital and earthquake damage estimation can be effectively used for a more reliable and swift decision-making. The system proposed in this paper uses fuzzy set theory to formalize knowledge acquired from experience and assess earthquake damage from ground conditions and ground motion characteristics. The results are obtained as fuzzy damage indices. Fuzzy decision analysis is then utilized to transform this imprecise information into a clear-cut decision on whether to cut or maintain the gas supply in a given area of the network. Key words: earthquake damage; gas network; fuzzy reasoning; fuzzy decision analysis. 1. Introduction When an earthquake occurs in a large city, extensive lifeline networks are especially vulnerable. For gas networks, there is an additional danger since secondary disasters such as fires or explosions may occur if pipelines or customers' houses are damaged. Hence, it may be necessary to interrupt the gas supply in the heavily-damaged areas as soon as possible. However, it should be kept in mind that an untimely shut-off would create confusion as the network recovery might take time. In the present control system of major gas utilities in Japan, the supply area of the network is monitored by sensors measuring the intensity of ground shaking. However, the actual decision to cut or to maintain the gas supply in a given area is made based on field survey reports. Since a gas network covers a vast area, it takes a long time (up to a few hours) to reach a rough damage figure and to make a decision. To reduce the time of decision-making, the monitored earthquake intensity information should be utilized more effectively. Since the relationship between ground shaking intensity and structural damage contains several kinds of uncertainty, the use of fuzzy inference is considered in this paper to estimate damage. In doing so, the * Discussion is open until October 1993 (please submit your discussion paper to the Editor, Ross B. Corotis). 0167-4730/93/$06.00 © 1993 - Elsevier Science Publishers B.V. All rights reserved