A Stochastic Optimization Model for Sewer Rehabilitation Policy : Considering Uncertainty in Inflow/Infiltration Flow

The goal of the sewer rehabilitation planning is to find an optimal policy that would minimize the total expected cost within a specified budget with the least amount of Infiltration and Inflow (I/I). The planning is however, complicated by uncertain effects of actions such as repair/replace/rehabilitation. In order to consider the uncertainty of I/I flow in the sewer rehabilitation problem, we present a Markov Decision Process (MDP) model, allowing uncertain effects of actions, which is a stochastic optimization model. The components of the MDP model is defined as follows: i) state is identified as the amount of I/I flow, ii) actions are classified as 'keep' and 'replace', iii) cost is defined as the expected cost associated with actions, iv) transition probabilities are derived from transition frequency between each state. Using the MDP model, we can develop sewer rehabilitation strategy for a drainage area of Seoul Metropolitan City.