A hybrid fuzzy discrete event simulation framework for analysis of stochastic and subjective uncertainties in construction projects

Discrete event simulation (DES) has proven to be an indispensable tool for planning and analyzing construction projects. Appropriate consideration of uncertainties of the inputs of DES results in more realistic outputs. Uncertainty in general can be categorized as stochastic and subjective. Stochastic uncertainty is a system property and represents the uncertainty associated with variation of a variable. Stochastic uncertainty can be represented by a probability distribution. On the other hand, subjective uncertainty represents the lack of knowledge of the system modeller regarding the actual value of a variable. Subjective uncertainty, for example, can be a result of lack of data or linguistic expression. Subjective uncertainty is often encountered in construction simulation due to the linguistic expression, and use of expert judgment in estimating activity durations. However, traditional DES is only able to consider stochastic uncertainty using probability distributions; and cannot handle subjective uncertainty. Fuzzy set theory provides a methodology for mathematical modelling of subjective uncertainty. Recently, fuzzy discrete event simulation (FDES) has been proposed for considering subjective uncertainty in construction simulation models. However, the fundamental differences between fuzzy numbers and probability distributions introduce new challenges to FDES frameworks. Furthermore, subjective and stochastic uncertainties may simultaneously exist in a simulation model. However, no framework is available that is able to consider both types of uncertainties in a discrete event simulation model. Firstly, this research, proposes a methodology for considering subjective uncertainty in estimating the activity durations or productivity of construction projects. Secondly, a FDES framework is proposed for dealing with subjective uncertainty of activity durations. The proposed framework advances the previous FDES frameworks by: (1) solving the problem of time paradox (overestimation or underestimation of the simulation time) (2) proposing a methodology for analyzing queues in FDES. Furthermore, this research proposes a novel hybrid discrete event simulation (HDES) framework that can simultaneously deal with both stochastic and subjective uncertainties. FDES framework is integrated within the proposed HDES framework for processing fuzzy uncertainty. Sampling from probability distributions are used to process stochastic uncertainty. The proposed framework is validated against analytically solved queuing examples containing both fuzzy and probabilistic uncertainty. The practicality of this framework is demonstrated using a case study of a module assembly yard. The results of this case study are compared with the results of FDES and traditional DES to demonstrate the advantages of the proposed HDES framework.