Research On Project Portfolio Selection Problem With Learning Effect Under The Flexible Time Horizon

In today's business environment with increasingly fierce competition,all the enterprises face the contradiction between the limited resources and multiple objectives.How to allocate limited resource to a subset of projects,which can bring highest profit to enterprises,is a great challenge which should be considered by decision makers.Project portfolio selection problem is usually defined as the problem of selecting a subset of projects,which can maximize the enterprises' objective,without violating the constraints.This paper will focus on the project portfolio selection problem with learning effect.The results show that making full use of the learning effect when executing the projects can bring higher profit to enterprises.Meanwhile,decision makers can usually adjust the length of time horizon based on the realistic environment when executing the project.This strategy is usually defined as “flexible time horizon”.Thus,this paper will further discuss the project portfolio selection problem under flexible time horizon,which can make the research more theoretical and practical significance.In view of the shortcomings of the existing literatures,this paper develops a new project portfolio selection model under flexible time horizon.The comparison between the new model and traditional model without considering the flexible time horizon shows the capability and characteristics of the new model.Furthermore,the mathematical models of return and investment considering the leaning effect are developed and incorporated into the project portfolio selection model.Finally,considering that the parameters such as return and investment are difficultly to be estimated due to the information uncertainty in the real world,this paper builds and solves a sensitivity analysis model.The rank of sensitivity of uncertain parameters is then given out.Our research is able to provide decision makers with a solution which can handle the uncertainty of parameters.Finally,numerical examples are conducted to highlight the capability and characteristics of the proposed model.Some conclusions are then drawn as follows:(1)the greater the learning coefficient is,the more profit enterprises can obtain.(2)Under different learning conditions,the optimal portfolio and the schedule of selected project are different.(3)The greater the learning coefficient is,the more sensitive the individual project income factors are.The greater the learning coefficient is,the less sensitive to the cost of production preparation is,and the net present value and the project cost of the project are reversed.The greater the learning factor is,the less sensitive the cost of a single project is.Under the same learning coefficient,the sensitivity of each factor is ranked as follows: single project income,single project cost,production preparation cost.