Toward an understanding of the impact of production pressure on safety performance in construction operations

It is not unusual to observe that actual schedule and quality performances are different from planned performances (e.g., schedule delay and rework) during a construction project. Such differences often result in production pressure (e.g., being pressed to work faster). Previous studies demonstrated that such production pressure negatively affects safety performance. However, the process by which production pressure influences safety performance, and to what extent, has not been fully investigated. As a result, the impact of production pressure has not been incorporated much into safety management in practice. In an effort to address this issue, this paper examines how production pressure relates to safety performance over time by identifying their feedback processes. A conceptual causal loop diagram is created to identify the relationship between schedule and quality performances (e.g., schedule delays and rework) and the components related to a safety program (e.g., workers’ perceptions of safety, safety training, safety supervision, and crew size). A case study is then experimentally undertaken to investigate this relationship with accident occurrence with the use of data collected from a construction site; the case study is used to build a System Dynamics (SD) model. The SD model, then, is validated through inequality statistics analysis. Sensitivity analysis and statistical screening techniques further permit an evaluation of the impact of the managerial components on accident occurrence. The results of the case study indicate that schedule delays and rework are the critical factors affecting accident occurrence for the monitored project.We would like to thank our industry partners for their helpin data collection, as well as the anonymous participants whocompleted the surveys and reported the data. We would alsolike to thank Dr. Timothy R.B. Taylor for his help in statisticalscreening analysis. The work presented in this paper was supportedfinancially with two National Science Foundation Awards (No.CMMI-1161123 and CMMI-1200120) and a Collaborative Researchand Development Grant (CRDPJ 335345-05)