Development and evaluation of cross-training policies for manufacturing teams.

1. Introduction

A manufacturing team can be defined as a group of workers responsible for manufacturing a family of various part types. The team makes use of a set of different machines required to manufacture the various part types. Each part-type must visit one or more machines according to its routing structure. In many manufacturing systems, the number of machines is larger than the number of workers in a team. Such systems are known as "Dual Resource Constrained" (DRC) systems, referring to the need for both workers and machines (Treleven, 1989). Labor flexibility is a major research topic related to DRC systems. Cross-training, which is a particular specification of labor flexibility, concerns the training of multiple workers for certain tasks or machines. This paper is devoted to the question of which workers should be trained for which machines. In more general terms, it asks: "what is an effective cross-training policy?" This is an important question, since the specific way in which workers and machines are connected determines the agility of the workforce with respect to changes in the demand for and the supply of labor. In order to answer this question, we must first consider which aspects are important in developing a cross-training policy. Second, once alternative cross-training policies have been developed, the effectiveness of the various cross-training configurations resulting from the application of these policies should be evaluated. Figure 1 presents an overview of the theoretical model upon which this paper is based.

A cross-training policy can be regarded as a set of rules for determining the distribution of workers' skills. These rules specify what decisions are made concerning aspects that are considered important in the development of a cross-training policy. Developing cross-training policies thus involves deciding which aspects are important to include, and also defining decision rules to specify how these aspects will be addressed. Applying a cross-training policy to a manufacturing team results in a skill matrix indicating which workers should be cross-trained for which machine. We call this resulting skill matrix a "cross-training configuration". In this paper, we use simple aggregated data from a generic manufacturing team applying cross-training policies to create cross-training configurations. We use information concerning the workloads of various machines and the current skill matrix of workers as a starting point. To evaluate alternative cross-training policies, we use simulation to study the performance of the resulting cross-training configurations in more detail. Our focus is on minimizing mean flow time from an Operations Management (OM) viewpoint, and on minimizing the standard deviation of the distribution of workload among workers (S[D.sub.workload]) from a Human Resource Management (HRM) viewpoint. More detailed contextual information, including details on the routing of jobs and other modeling assumptions, is required for the simulation. Routing structure is studied as a contextual variable.