A decision support system for technician routing with time windows

Purpose The purpose of this paper is to present a decision support system (DSS) for a Colombian public utility company in order to aid decision-making at the operational level regarding route planning and travel time. The aim is to provide a tool to assist technicians that perform interruption and reconnection of domiciliary services for about 2,000 customers a day. Design/methodology/approach The real-life problem is modeled as a Single Depot Vehicle Routing Problem with Time Windows (SDVRP-TW), which is a well-known optimization problem in Operations Research/Management Science. A two-stage approach integrated into decision-making software is provided. The first stage considers the clustering of customers generated by a combination of the sweep and the k-means algorithms, while the second phase plans the routing of technicians using the nearest-neighbor and the Or-opt heuristics. The proposed approach is tested using real data sets. Findings In comparison with the current route planning approach, the proposed method is able to obtain savings in total travel times, improving operational productivity by 22.2 percent. Research limitations/implications Since the analysis is carried out based on mathematical modeling, assumptions about the relationships between variables and elements of the actual complex problem might be simplified. Although the proposed approach aids the route planning, decision makers make the final decisions. Practical implications The proposed DSS has a critical impact on actual operational practices at the company. Productivity and service level are improved, while reducing operational costs. The decision-making process itself will be improved so technicians and higher decision makers can focus on performing other tasks. Originality/value The real-life problem is modeled using mathematical programming and efficiently solved through a two-stage approach based on simple, quite intuitive, solution procedures that have not been implemented for such services. In addition, as actual data from the company is employed for experimental purposes, the solution approach is tested and its efficiency and efficacy are both validated in a realistic setting, hence providing realistic behavior for decision makers at the company.