Computer-Aided Civil and Infrastructure Engineering 24 (2009) 401–415 Integrating Messy Genetic Algorithms and Simulation to Optimize Resource Utilization Tao-ming Cheng ∗ & Rong-zheng Yan Department of Construction Engineering, Chaoyang University of Technology, Taiwan Abstract: This article presents a mechanism for in- tegrating messy genetic algorithms (MGAs) and a dis- crete event simulation technique to facilitate the simu- lation of optimal resource utilization to enhance system performance, such as in relation to the production rate or unit cost. Various resource distribution modeling sce- narios were tested in simulation to determine their sys- tem performances. MGA operations were then applied in the selection of the best resource utilization schemes based on those performances. A case study showed that this new modeling mechanism, along with the imple- mented computer program, could not only ease the pro- cess of developing optimal resource utilization, but could also improve the system performance of the simulation model. 1 INTRODUCTION Although the construction industry is still using com- mercial software packages primarily based on decades- old critical path method (CPM) significantly more pow- erful methods have been developed for construction scheduling and planning in the past decade or so. A major breakthrough in this area was CONSOM (CON- struction Scheduling, Cost Optimization, and Change Order Management) (Adeli and Karim, 1997; Karim and Adeli, 1999a) based on an object-oriented informa- tion model (Karim and Adeli, 1999b) and the patented neural dynamics optimization model of Adeli and Park (1995, 1997) that provides cost minimization of the con- struction plan, time–cost trade-off analyses, and change order management. ∗ To whom correspondence should be addressed. E-mail: tmcheng@ cyut.edu.tw. Although great promotion in scheduling technique was seen, while performing the activities of a project, construction planners still have to select construction methods or technologies and associated resources, in- cluding crew sizes and equipment, from different al- ternatives. Those alternatives create various durations and costs for that particular activity and the project as a whole. The more complicated the project, the more tasks are needed for networking it. As a result, planners need an effective technique to analyze and de- termine the most efficient resource assignments to com- plete a project with the least cost and within a desirable duration. Discrete event simulation can be used to as- sist construction engineers in such an analysis. Exam- ples include modeling concrete batching operations to study the impact of using different resource combina- tions (Zayed and Halpin, 2001), analyzing earthmov- ing operations for the selection of fleet configurations (Marzouk and Moselhi, 2004), and utilizing a simulation model to study the residential construction inspection process (Sawhney et al., 2005). Despite the success of applying simulation tech- niques to the optimization of resource utilization, the traditional way to determine an optimal solution in simulation is to exhaustively examine all of the re- source combinations. Therefore, Riggs (1979) proposed a computer-based program called sensitivity analysis for facilitating such enumerations. However, if the number of possible resource combinations increases explosively, sensitivity analysis could be extremely time-consuming. As a result, AbouRizk and Shi (1994) proposed a heuristic algorithm (HA) that efficiently lo- cates the most appropriate resource allocation in the simulation system. However, the solution of such an approach is usually the local optimum and the perfor- mance of HA is problem dependent. Hence, Cheng C  2009 Computer-Aided Civil and Infrastructure Engineering. DOI: 10.1111/j.1467-8667.2008.00588.x