Engineering Approach Using ANN to Improve and Predict Construction Labor Productivity under Different Influences

Abstract

Construction labor productivity is a fundamental piece of information for estimating, budgeting, and scheduling a construction project. The current practice to estimate construction labor productivity relies primarily on the traditional method, which uses the published productivity data and/or the estimator’s own experience, with an apparent lack of systematic approach to measuring, estimating, and predicting it. An investment for the sole purpose of productivity data collection and modeling is not likely to be successful for a construction company. To make the investment economically feasible, the productivity system should be integrated into the overall database and information system of the construction company. To achieve these advantages and overcome the current practice weaknesses, this paper introduces an engineering concept to document, control, predict, and improve the contractor’s labor productivity. A wide range of influencing factors on the micro level (project management and administration) and the micro/micro level (activity level at construction site) has been considered. The proposed engineering approach was applied to model construction labor productivity of two construction crafts, carpentry and fixing reinforcing steel bars of different types of concrete foundations, using the artificial neural network (ANN) technique and utilizing the transfer function of the hyperbolic tan function (tanh). The results showed an adequate convergence with reasonable generalization capabilities, and more accurate and credible results compared with not only the traditional method, but also the existing approaches in the literature. This study contributes to the construction engineering and management body of knowledge by providing insight into using different ANN activation and transfer functions along with a wide range of influencing factors to benchmark the contractor’s construction labor productivity. Moreover, the utilized engineering approach shows how a readily available practical database can help optimize several objectives. It supports two main pillars of sustainable construction: the economic dimension and the social dimension.