Project portfolio selection under uncertainty with outsourcing opportunities

The paper presents a stochastic optimization model for project portfolio selection under uncertainty about the real efforts required for the execution of the work packages contained in the projects. As a subproblem, the assignment of the work to human resources and the distribution of work over time is addressed. The available workforce is assumed as multi-skilled. Required efforts are modeled as random variables. The recourse action for the case where the available capacities of the internal human resources do not suffice to cover the actual work times consists in delegating parts of the work to external human resources. The staffing-and-scheduling subproblem is solved by means of a Frank-Wolfe type algorithm. To solve the upper-level problem of project portfolio determination, a modification of the Variable Neighborhood Search (VNS) algorithm is applied. Experimental results for a benchmark of synthetically generated test instances and for an illustrative example from the E-Commerce Competence Center Austria are provided.