A Framework for Improving the Productivity of Operational Preventive Maintenance Activities for Wastewater Collection System

Abstract

Operational maintenance of the wastewater collection system is an important part of urban infrastructure management. It involves various activities, such as visual inspection, low-pressure flushing (LPF), high-pressure flushing (HPF), catch basin cleaning, and hydro-mechanized cleaning. Large cities require significant budget and resources to perform the necessary cleaning activities at various locations around the city at regular intervals. For instance, the collection system in Edmonton, Canada, comprises over 5,500 km of sewer pipes, and as of 2014 there are over 1,400 prescheduled HPF locations. However, planning and scheduling these activities can be challenging because of the wide variation of actual on-site flushing duration, which depends on a number of factors such as length and diameter of the pipes, frequency of flushing, structural condition, age, and season. Moreover, travelling between these locations results in a large amount of unproductive time. Reviews of the literature and of current industry practice reveal that the existing models and algorithms do not specifically address these issues. This research, therefore, develops a framework for improving the productivity of these activities by optimizing operational maintenance schedule. The research consists of two primary modules: (i) developing a forecasting model to estimate the on-site duration of activities, and (ii) developing an optimization algorithm to maximize productivity. The models are developed and tested using historical data of HPF activity from the Drainage Operations group at the City of Edmonton. The forecasting model captures the majority of the variations in on-site flushing duration and provides useful insight into the factors affecting on-site productivity. For optimization, this research formulates the drainage operations scheduling problem (DOSP) as a special class of the stochastic and capacitated vehicle routing problem (VRP), where the objective is to maximize value-added on-site flushing time while minimizing travel. Alongside existing algorithms (such as integer programming, genetic algorithm), a heuristic algorithm is developed to meet the specific needs of this complex combinatorial problem. The proposed optimization algorithm is tested and compared with other algorithms by simulating a monthly HPF schedule. The results show that accurate estimation of on-site duration, coupled with schedule optimization can improve daily productivity by a considerable margin. The outcome of this research makes significant academic, economic, and environmental contributions by proposing a systematic approach to planning and scheduling operational maintenance for wastewater collection systems.