The unsafe behaviors of construction workers are often the immediate causes of construction accidents, but the underlying causation of such behaviors are not well understood. This research regards the management of construction safety as a system, and seeks to use system dynamics to demonstrate how the system influences construction workers in terms of unsafe behaviors. First, individual and environmental conditions that can lead to an unsafe behavior are identified through a holistic cognitive analysis and management conditions that affect such conditions are identified. Second, a system dynamics model for the causation of unsafe behaviors (SD-CUB) is developed to characterize the causal structure of the system. The SD-CUB model involves relationships among management, individuals, and environmental conditions that can eventually lead to workers' unsafe behaviors. Third, a variety of model tests are conducted to build the confidence of the SD-CUB model. A five-week survey and observation on a building construction project is conducted to demonstrate that the SD-CUB model generates correct patterns of behavior. The model tests also imply that safety and production can actually support each other, management conditions on supervisory level are effective on the improvement of workers' safety awareness, and preventive actions are more effective than reactive actions on the enhancement of safety performance. The SD-CUB model can also be used as a basis for simulation of various site scenarios to explore the best solution to prevent and correct unsafe behaviors, and by redesigning the causal structure, the leverage points and critical management strategies can be determined.

No abstract

No abstract

In order to gain deep insight into the underlying causation and identify correspondent management strategies in the prevention of construction workers' unsafe behaviors, a conceptual system dynamics model was proposed. The paper mainly focuses on the confidence building of the model through model testing, which basically contains tests of model structure, model behavior and model's policy implication. The conceptual model is quantified first before the formal process of model tests. Second, the paper applies semi-structured interviews for the structure-verification test to ensure the model's appropriateness in depicting the structure of the real system. Third, tests of model behavior such as extreme-condition tests and behavior-reproduction tests are conducted, so as to demonstrate the plausibility of the patterns and behaviors the model generates. Fourth, model's policy implication tests are presented to examine whether the model's responses to predesigned policies are consistent with the real system. As a result, the model shows confidence in revealing the underlying causation of unsafe behaviors, which can be used as reference to the routine safety management in practice. Further suggestions for the application of the model as a tool of simulation in real construction projects are also given. 80