Application of Fuzzy Logic and Hybrid SD-DES Approach in Predicting Construction Labor Productivity

Abstract

Labour productivity is an important indicator of the performance of the construction industry. Due to the lack of a standard productivity measurement system and ignoring various influential factors, labour productivity is constantly decreasing. Although different construction labour productivity modelling techniques have been developed, they focused on specific subjects, and their implementation is limited to particular countries. Therefore, this study aims to investigate the application of Fuzzy logic and hybrid System Dynamic (SD)-Discrete Event Simulation (DES) to predict construction labour productivity. To meet the objective of this study, Construction labor productivity influencing factors were extracted through Literature review, and a questionnaire survey was carried out. A relative importance index was calculated to identify the most significant factors used for modeling purposes. The production rate values of slab concreting and their influencing factors have been collected through cause study on real building project. A Fuzzy expert system in fuzzy logic concepts was used to predict the productivity rate by incorporating the influence of qualitative and quantitative factors. The performance prediction of the Fuzzy expert system was evaluated via Mean Absolute Percentage Error and Root Mean Square Error with their error rate of 9.33% and 0.08 respectively. The intricate effect of relationships among the contextual factors that influence labour productivity was modelled using the System Dynamic approach. Stock and flow diagram of system dynamic was developed to model complex structure of context variables. In addition, Discrete Event Simulation was applied to model the operational influencing factors and their effects on labour productivity. The performance measurement results show that the Discrete Event Simulation in this study gave high numerical accuracy with 5.7% of MAPE value. The productivity result predicted by a Fuzzy expert system, Discrete Event Simulation, and hybrid SD-DES in an 8-hour workday is 162, 168, and 176 m3 respectively, while real case productivity was 181 m3 . The predicted result by the models is slightly deviated by 168.67 ± 7.02. Therefore, the finding shows that the value obtained by the hybrid SD-DES modelling technique is more realistic. The proposed models can be useful in the planning stage and allow for a more precise prediction of concreting productivity.