Research On The Decision-making Model Of The Concession Period Of PPP Projects

PPP project has been widely applied in the field of infrastructure in recent years. Generally, PPP projects have a huge investment and a long period which can reach decades or even more years, which contains uncertainty factor. The government departments, private sector entities and the public which involved in the PPP project should well balanced, so the franchise operation period will became the key factor for the PPP project. The current franchise operation period models is too subjective on the measurement of the uncertainty factors, so it is meaningful to find a reasonably method to overcome the shortcomings. This study find20kind of uncertain factors through the literature questionnaire statistical, and identify the corresponding quantitative method for the measurement for the franchise operation, interest rate and cash flow three key factors. The study design evaluation matrix to calculate12relative important coefficient which come from the quantitative method. Considering the existence of "noise data ", we use fuzzy analytic hierarchy process(FAHP) that of consistency20kinds of factors on13relative important influence factor evaluation matrix. Aimed at the evaluation matrix noise data, the use of double S type evaluation matrix adaptive model and the matrix multi-aspect a ladder degree optimization, combined with the iterative termination criterion redistribution weight value. Finally take a highway as an example for empirical research, caparison before and after optimization weight of single, double weight method and traditional NPV, IRR method came to the concession period value, then analyze19scale NPV sensitivity. This model is adaptive, and contains the traditional method of the calculation process, and can eliminate the noise data of concession period decision-making produce adverse influence; easy to programming realization, avoid the BP network a ladder degree iterations is overmuch, step length value is set difficult defects, and the direction of more optimization calculation to avoid the local optimum; the optimized model has a moderate sensitivity and can be applied to the actual.