On the use of learning curves for the estimation of construction productivity

To account for the skill of construction crew, an ad hoc coefficient is used as a multiplier of productivity estimations in most available methodologies today. However, little has been published on the determination of this coefficient and, moreover, on the way it is affected by repetition at the same working conditions. As such, research has been undertaken to associate the crew skill coefficient to the learning curve of a specific repetitive activity. Field measurements have been used as the basis to develop statistically sound datasets for subsequent use in simulation-based analysis. Field and simulation results have been compared through the use of an existing prototype simulation platform (CaissonSim) in order to determine the most appropriate crew skill coefficient. All developed models have been validated through the application of a strict statistical protocol. A sensitivity analysis has been undertaken in order to examine the cause of the productivity estimates' variation. The main conclusions are that the crew skill coefficient depends on the learning rate and the determination of the standard production point, which denotes the termination of the learning phenomenon.