Empowerment Scheduling: A Multi-objective Optimization Approach Using Guided Local Search

Field Workforce Scheduling (FWS) is a very important and practical problem in the service industries. It concerns the scheduling of multi-skilled employees to geographically dispersed tasks. In FWS, employee efficiency is highly important, and thus they have to be managed in an effective way. Employee empowerment is a relatively new and flexible management concept. It promises to benefit both organizations and employees by enhancing employee morale, satisfaction and productivity. This motivates the incorporation of empowerment when designing FWS models, which has not been thoroughly investigated. This thesis describes the development of a new efficient empowerment scheduling model, called EmS, for FWS. The key feature of EmS is that it is strongly linked to the management literature on empowerment from which the requirements are derived. EmS provides employees with a simple, yet flexible and fair means of involvement in the scheduling decision, through which they can suggest their own schedules. This is formulated using a multi-objective optimization (MOO) approach where the task is to find a balance between employee empowerment and the employer’s interest. To evaluate EmS, a series of empirical experiments are conducted, presenting the first extensive and in-depth study of the feasibility of empowerment in the FWS context, as well as the efficiency of an empowerment scheduling model. To tackle the empowerment scheduling problem, a new MOO method based on Guided Local Search (GLS) is developed. The new method (called GPLS) retains the main characteristic of GLS, that is simplicity with few parameters to tune. GPLS enables Pareto Local Search, which is a simple yet effective MOO method, to overcome the problem of being trapped at Pareto local optima in an intelligent manner. In addition, a number of GPLS-based frameworks are proposed, which prove the potential of GPLS to be a central part of more advanced frameworks. GPLS and its frameworks are extensively tested on two standard MOO benchmarks, and EmS. Computational results confirm that GPLS is an effective scheduling technique, and it is competitive with state-of-the-art MOO metaheuristics.