ROBUSTNESS OF WORK SAMPLING FOR MEASURING TIME WASTE

Abstract

Construction can be considered a socio-technical system, which is challenging to model due to the many agents interacting either in a managed way or autonomously. Therefore, cause and effect models are hard to validate, and a traditional correlation approach is insufficient. In this study, the method of robustness testing was applied to test the effect stability when assumptions of a model are changed. The research objective is to apply robustness testing on WS data to assess the robustness and validity of the WS method. An actual refurbishment project was the case for this study, where data was acquired through nine days of continuous WS application. Time-series data were grouped into Direct Work (DW), Indirect Work, and Waste Work. Several different robustness tests were applied. It can be concluded that the WS method is robust, i.e., the effect (DW) is stable even if the assumptions are changed severely. Deleting 90% of the sample does, for instance, almost not change the effect. Likewise, if errors are infused into the sample, the effect is stable. Also, if certain structural parts are excluded from the sample, e.g., observations during morning startup, etc., the effect is still stable.