The impact of dynamic learning and training on the personnel staffing decision

Abstract

This paper studies the personnel staffing decision accounting for the planning of workforce training and the inclusion of learning and forgetting, which dynamically change the workforce skill level, influencing the required staffing budget. This contrasts staffing models presented in literature, which determine the required number of workers under the assumption of a fixed and/or predefined skill mix of the workforce. We consider different types of on-the-job and off-the-job training activities, traditional learning associated with execution of regular tasks and forgetting according to which a worker’s efficiency is adapted dynamically throughout the time horizon. The acquired worker’s efficiency level determines the speed at which assigned tasks can be processed, which is modelled via the definition of multiple discrete modes per task. To improve the accuracy of the staffing decision and account for the effect of gradually changing the workers’ skill levels, a baseline schedule is composed that comprises the solving of an integrated days-off scheduling and task assignment problem. We propose a dedicated solution procedure that relies on a constructive heuristic to find an initial solution and subsequently a branch-and-price algorithm to derive a (near-)optimal solution. The branch-and-price algorithm is featured by a multi-stage pricing procedure to find promising columns, alternating between mathematical mixed-integer programming, a single-pass heuristic and a genetic algorithm. The computational experiments validate the performance of the proposed procedure and demonstrate the benefits of different training and learning methods under different conditions.