APPLYING GENETIC ALGORITHMS ON THE COMPUTER SIMULATION FOR IMPROVING CONSTRUCTION PRODUCTIVITY.

Computer simulation has been applied on construction operation for many decades. Simulation usually shows its advantages on finding out how to change operation processes or choosing which resource combination would improve productivity. Through what-if mechanism, simulation can achieve the latter tasks. In other words, through a series of computer runs, data can be obtained about the performance of system using various design alternatives that include different resource combinations. On the other hand, however, for obtaining best resources combination, all possible alternatives of resource combination should be run in simulation. Therefore, simulation is not considered as an optimization technique. Since Holland proposed genetic Algorithms (GA) in 1975, GA has been widely used for solving optimization problems in different research areas and gaining good performance. This paper aims at proposing new approach that applies GA as a pre-processor for efficiently screening out the resource combination that has good influence on construction productivity. A CYCLONE-based simulation program named COST (Construction Operation Simulation Tool) is adopted for adding GA function and then is used for running case to demonstrate the advantage of applying GA on construction operation simulation. Introduction Different construction processes usually form a construction operation; therefore the flows between the processes and the resource utilizations of the processes could directly affect the performance of the construction operation. To better understand the performance of the construction operation, construction project planners can use computer simulation to predict the performance of the construction operation in terms of process flows and resources utilization. However, to find the best resource utilization of the construction operation, all possible resource combinations should be tested within the simulation process. That is, an exhaustive enumeration of resource combinations should be conducted, which is not economic if the possible resource combinations increase explosively. Therefore, simulation is not considered as an optimization technique (Anderson et al. 1997). Genetic Algorithms (GA) was developed by John Holland in 1975 and has been widely applied to the different disciplines of research for solving the optimization problems because of its robustness structure. GA is also known for its flexibility in hybridizing with other methodologies to obtain better solutions. This paper presents a simulation mechanism that integrates simulation with GA to find the best resource combination of the construction operations for improving their productivity. Related Research Recently, researchers began to integrate simulation techniques with GA to obtain the better solutions of the stochastic system. Parmar et al. (1996) developed a simulation model to determine the net return of the machinery cost with a given machinery set. The optimal machinery set was determined by using GA as an intelligent search scheme. They concluded that by integrating with GA the total time required to search for the near-optimal solution is 10% of the total time required by the exhaustive enumeration search scheme. Hart et al. (1998) optimized the management variables of the pastoral dairy farm problem by introducing a hybrid mechanism that combines hill-climbing and genetic algorithms. They claimed the hybrid mechanism could quickly locate the good regions of the search space and discover near-optimal solutions. Lin and Malasri (1998) used CBU-EXORESS computer simulation system with GA to solve the routes problem of collecting solid waste. They found that the local optimal solution could be achieved after several trials. In addition, less time consuming compared to other searching scheme is concluded in their study. McHaney (1999) discussed the method of integrating GA with computer simulation. According to his computational experiences, he suggested that GA is relatively adaptable to the discrete-event computer simulation environment. Azadivar and Tompkins (1999) presented a simulation-optimization process. They used GA to guide the simulation system toward better solutions. Zhao and Souza (2000) proposed a GA-based heuristic search that integrated computer simulation and intelligent production line balancing scheme to optimize the production of hard disk drive. They also concluded that by integrating with GA, the optimum could be achieved. Many hybrid mechanisms that integrate simulation techniques with GA have been widely applied to different disciplines of research and are proven to be efficient to find the optimal solution. However, in the field of construction engineering and management such simulation mechanism has not been well adopted to study construction operations. Therefore, judging from the state of research, there is a need to develop a new mechanism that is designed especially for analyzing and optimizing construction operation simultaneously. Genetic Algorithms Genetic Algorithms (GA) is the search algorithm developed by Holland (1975), which is based on the mechanics of natural selection and genetics to search through decision space for optimal solutions (Gen and Cheng 1999). The metaphor underlying GA is natural selection. In evolution, the problem each species faces is to search for beneficial adaptations to the complicated and changing environment. In other words, each species has to change its chromosome combination to survive in the living world. In GA, a string represents a set of decisions (chromosome combination), a potential solution to a problem. Each string is evaluated on its performance with respect to the fitness function (objective function). The ones with better performance (fitness value) are more likely to survive than the ones with worse performance. Then the genetic information is exchanged between strings by crossover and perturbed by mutation. The result is a new generation with (usually) better survival abilities. This process is repeated until the strings in the new generation are identical, or certain termination conditions are met. Mechanism for Applying GA on Simulation As a simulation methodology, CYCLONE has been widely used in the design and analysis of the construction operation over the last two decades (Gonzalez-Quevedo et al. 1993). Therefore, A CYCLONE-based simulation program named COST (Construction Operation Simulation Tool) is adopted for adding GA function and therefore form a new simulation system named GACOST (Cheng et al. 2000). The flows of the simulation mechanism that integrate COST with GA to search for the optimal solution are shown in Fig. 1. This simulation mechanism utilizes GA as a selection engine to screen out the resource combinations that produce bad system performances of the construction operation. The system performance of the construction operation, also defined as the fitness value of the string, is evaluated by CYCLONE module. The flow of this simulation mechanism is described as follows: