Estimation of the workload boundary in socio-technical infrastructure management systems: The case of Belgian railroads

Infrastructure systems are large-scale complex sodo-technical systems that rely on humans for their safety critical decision-making activities. In the case of railroad networks, hierarchical organizations denoted as traffic control centers (TCCs) operate 24/7 in order to maintain successful network operations. Interacting social and technical factors influence TCC operational environments and thus the overall performance of the railroad system. This research presents a novel data envelopment analysis (DEA) application along with its implementation and validation by investigating the workload boundary of human performance through a case study built for the Belgian railway (INFRABEL) TCCs. We pursue two research foci. The first is to identify organizational, socio-economic, and technical factors that describe the performance environments in which TCC personnel operate. We use these factors to determine relatively homogeneous performance environments using multivariate statistical methods. The second focus is to design and implement on-site a socio-technical performance measurement framework, based on a new and unique dataset at the workstation level that is capable of considering sodo-technical heterogeneity. Our approach consists of three steps. First, we apply a two-stage clustering approach to generate statistically relatively homogeneous groups. Second, we calculate meta - and in-cluster efficiency scores. Finally, we assess the validity of our results with INFRABEL. Results reveal three insights: (i) efficiency improvement strategies require further investigation based on temporal trends; (ii) disregarding performance environment heterogeneity leads to over estimation in target setting; and (iii) socio-technical system design could be informed by applying DEA, provided that, domain specific expertise is used in the model formulation. (C) 2019 Elsevier B.V. All rights reserved.