HÖGSKOLAN DALARNA School of Technology and Business Studies An application of gravity p-median model with different distance decay functions

The location-allocation problem has been studied over 50 years and recently a new method called “gravity p-median model” is introduced to the public. The key idea is that the probability of customers visit a facility is based on a distance decay function rather than directly choose the nearest one. An empirical test shows that the solution of gravity p-median model with exponential distance decay function is unstable and of limit use. This paper extends the research to apply gravity p-median model with three different distance decay functions in Dalecarlia, Sweden. The distance decay functions are estimated from a Swedish survey by maximum likelihood estimation. The models are optimized by simulated annealing. The result suggests that different distance decay functions dominate the solutions of gravity p-median model, and the log-normal decay function can provide stable solutions.