Modeling of Simultaneously Continuous and Stochastic Construction Activities for Simulation

Abstract

Construction operations have been modeled and simulated using discrete-event simulations with effectiveness and fidelity that have continuously increased over the past four decades. Today, it is possible to model construction operations so that its discrete components are represented very faithfully. However, construction operations often include activities that are simultaneously continuous and stochastic in nature. These continuous activities have typically been modeled by discretization and in some cases using combined discrete-continuous simulation. The current literature does not discuss the series of important statistical issues, such as divisibility of fitted distributions and a significant possibility of the discretized activities not being independent and identically distributed, which are brought along by discretization and can significantly impact the results of the simulation study. Combined discrete-continuous simulations have treated the continuous portion dynamically (i.e., based on differential equations of rate change) but either deterministically or disrespecting statistical properties of the underlying continuous stochastic process and its interactions with the discrete portion. These approaches can seriously affect the accuracy of the model in representing the system behavior, grossly impact the simulation results, and can be detrimental to the acceptance of the model due to difficulties in verification and validation. A clear understanding of the underlying problems is thus required when modeling simultaneously continuous and stochastic activities. This paper investigates the major issues associated with modeling continuous stochastic construction activities that have been traditionally modeled using discrete-event methods. This paper contributes to the body of knowledge by identifying the issues pertaining to and impacts of discretization of continuous activities. The paper improves awareness of these issues that have been ignored or not understood by previous studies and help future researchers in building valid and credible simulation models. The paper also demonstrates the effect of discretization on the behavior of the model and the simulation results using a case study of a road-paving operation.