A dynamic programming approach to optimization of road maintenance

Road conditions in Malaysia rely on the traffic intensity of the road. Based on this factor, the road condition assumes a stochastic process in which the state of the road for the following year is highly dependent on the current state. The problem faced by Jabatan Kerja Raya and Projek Lebuhraya Utara Selatan is in determining the best time to perform road maintenance task in Malaysia. This is due to the fact that too frequent repair and maintenance would increase maintenance cost whereas less or no maintenance would lead to an increase of road accidents resulting from unsatisfactory road condition. This paper proposes a new approach in road maintenance by analysing the state of the road as a Markov chain process. Through the transition matrix and reward matrix which is obtained from the number of road accidents resulted from specific road condition, an optimal solution is obtained as to when it is the best time road maintenance task should be exercised. The procedure calls for dynamic programming in determining the optimal solution. [Malay.]