Cross-trained workforce planning models

Abstract

Cross-training has emerged as an effective method for increasing workforce flexibility in the face of uncertain demand. Despite recently receiving substantial attention in workforce planning literature, a number of challenges towards making the best use of cross-training remain. Most notably, approaches to automating the allocation of workers to their skills are typically not scalable to industrial sized problems. Secondly, insights into the nature of valuable cross-training actions are restricted to a small set of predefined structures. This thesis develops a multi-period cross-trained workforce planning model with temporal demand flexibility. Temporal demand flexibility enables the flow of incomplete work (or carryover ) across the planning horizon to be modelled, as well as an the option to utilise spare capacity by completing some work early. Set in a proposed Aggregate Planning stage, the model permits the planning of large and complex workforces over a horizon of many months and provides a bridge between the traditional Tactical and Operational stages of workforce planning. The performance of the different levels of planning flexibility the model offers is evaluated in an industry motivated case study. An extensive numerical study, under various supply and demand characteristics, leads to an evaluation of the value of cross-training as a supply strategy in this domain. The problem of effectively staffing a pre-fixed training structure (such as the modified chain or block) is an aspect of cross-training which has been extensively studied in the literature. In this thesis, we attempt to address the more frequently faced problem of ‘how should we train our existing workforce to improve demand coverage?’. We propose a two-stage stochastic programming model which extends existing literature by allowing the structure of cross-training to vary freely. The benefit of the resulting targeted training solutions are shown in application using a case study provided by BT. A wider numerical study highlights ‘rules-of-thumb’ for effective training solutions under a variety of characteristics for uncertain demand.