Prediction of the Future Condition of a Water Distribution Network Using a Markov Based Approach: A Case Study of Kampala Water

The increasing costs of managing urban water distribution systems coupled with limited budgets and new regulatory requirements has compelled water utilities to ensure that asset maintenance decisions move from a reactive to a proactive approach. This means that asset renewal decisions must shift from being made only when failures happen to well-planned priority- based replacement, repair and rehabilitation strategies. However the question of prediction of the future condition of the pipe network continues to trouble water utility managers because of the complexities in determining pipe conditions amidst poor data situations. Conventional approaches for prediction of pipe condition are skewed towards statistical analysis and do not consider failure history. Until now, there have been no approaches for prediction of future state based on pipe condition. This paper applies a Markov based approach as a decision support system to predict the future condition of a water distribution network. The approach is illustrated on a case study in Kampala Water, Uganda as a proof of concept. Data on pipe condition history per block is first checked to ensure it follows the Markovian process. Pipe condition has been based on a composite index that combines pipe age and break history. The goodness of fit has been evaluated using the X2-inference test. The Poisson distribution has been used to develop transition probability matrices required to forecast the state future condition of the water distribution network. The approach will help water utility managers optimize maintenance and repair decisions amidst budget limitations whilst taking into consideration both current and future states of the pipe network.