Abstract: Vulnerability to sudden service disruption due to deliberate sabotage and terrorist attacks is one of the major threats for network system.Thus,facility location strategy in network should concern the operational cost in peacetime and emergency.This article focuses on how to locate P facilities so as to minimize expected cost including the regular operational cost as well as the emergent operational cost of a worst-case attack with the interdiction of R facilities.We have cast this problem as a bi-level binary programming model where the top level problem involves the decisions about where to locate facilities and the lower level problementails the interdictor response on which facilities to at tack.We solve the bi-level problem through hybrid genetic algorithm based on lagrange relax method.Results of this problem and traditional P-median location problem applied to European 150 data set are presented.Several solutions derived from these models are presented in greater detail and demonst rate the degree to which the loss of R facilities affects network system efficiencies with different location strategy decided by these models.Then we analyze the results to changes in key parameters including the weight of regular condition,the number of facilities and the number of facilities interdicted.

Key words: facility location, interdiction, bi-level programming, hybrid algorithm