Partner Selection Problem For Project Considering VaR And Its Ant Colony Algorithm

With the rapid development of Chinese economy, China is growing strong.The government are increasingly focused on infrastructure. At present, a large number of projects are flourished together such as water conservancy, transportation, telecommunications and real estate. However, the projects are large-scale, many-personnel and time-consuming. An enterprise cannot complete a project alone, and it must call for bids to select partners so that they work together. Due to construction time uncertainty, a project faces tardiness risk. Therefore, it is essential to manage project risk.For project partner selection and tardiness risk problem, the following work is done in this paper:First, this paper reviews literature of project risk management and partner selection and describes some related concepts. With the help of the theory of VaR (Value at Risk), an uncertainty model about partner selection with the objective of minimizing VaR. When partners’ construction time follows normal distribution, the uncertainty model is then converted into certainty model according to the relevant concepts of normal distribution.Second, for the certainty model, this paper designed four algorithms: enumeration algorithm, the basic ant colony algorithm, the ant colony algorithm with lower and upper limit of pheromone, and the proposed ant colony algorithm with time-varying parameters and critical path information guide according to the characteristics of the problem.By three different-scale examples, parameters and comparison are analyzed among these four algorithms. The results show that the proposed ant colony algorithm with time-varying parameters and critical path information guide is effective. Meanwhile, how the confidence level and coefficient of variation affect decision is discussed.Third, for the uncertainty model, this paper designed two strategies:Monte Carlo simulation for all jobs and Monte Carlo simulation for jobs on critical path. Correspondingly, two algorithms were proposed:one is Monte Carlo simulation embedded ant colony algorithm for all jobs; the other is Monte Carlo simulation embedded ant colony algorithm for jobs on critical path. Experiments show that Monte Carlo simulation for all jobs is more effective.