Decision-making Analyses For Research And Development Projects Under Uncertainty

In the increasingly competitive market, research and development (R&D) has become a source or an impetus to maintaining the competitive advantage for many industries. The decision-makings of R&D projects are usually made under the uncertain conditions with high risk because the R&D projects are affected by many uncertain factors, such as market, technique, capital and material. This dissertation is to investigate the problems of selection, scheduling and investment decision of R&D projects under uncertainty. The detailed works are described as follows.The multi-criteria selection method of R&D projects is presented based on the fuzzy simulation. The selection decision process of R&D projects is divided into the strategic decision period and the tactical decision period. The fuzzy preference model and fuzzy multi-criteria evaluation model are applied to evaluate the candidate projects in the two periods, respectively. Finally, the optimal R&D project has to be selected in accord with the strategy of the firm.The program evaluation and review technique (PERT) based on the fuzzy simulation is developed to find the possible critical path of the R&D project network in which the activities durations times are assumed to be fuzzy variables. On the other hand, the contingency model is presented to consider the fuzzy nature of change orders and their impact on the schedule of an R&D project. The fuzzy simulation is employed to estimate the expected value and pessimistic value of the total project delay. It may be helpful to the project scheduling problem of R&D projects at the early stages of project planning and development.The optimal stopping decision models of R&D projects are constructed based on the renewal reward processes in the random and random fuzzy environments, respectively. The stopping time and investment strategy are considered as decision variables and the interarrival times between discoveries (jumps) are assumed to be random variables and random fuzzy variables, respectively. The simultaneous perturbation stochastic approximation (SPSA) algorithms based on simulation techniques are designed to solve the models. Finally, the numerical examples are presented to illustrate the effectiveness of these methods.The multi-period decision model of an R&D project is considered based on dynamic programming in a fuzzy environment. At each period, the decision maker can take any one of three possible actions: continue, improve, or abandon. The question on how the variances (increasing uncertainty) of these uncertain factors influence the value of managerial flexibility of an R&D project is analysed in a fuzzy environment.