Optimal Project Portfolio Execution - Using analytical and simulation models with realistic project layouts and resource behavior

Sammanfattning: This project aims to increase the knowledge in the area of resource allocation in R&D portfolios, a topic interesting to both the industry and academic research. The thesis investigates how to optimally allocate resources to projects in a R&D portfolio, with focus on how many projects that are optimal to run in parallel. A complex mixed-integer nonlinear programming model with a realistic stage-gate project layout and advanced resource behaviour, including project resource learning and other efficiency losses, is developed and also proven unsolvable in realistic time. A simplified model handling learning is proposed and solved using a mixed integer solver for a small portfolio. A simulation framework implementing all complexities is developed, and used to find portfolio parameters affecting the optimal number of projects to run in parallel, using a Monte Carlo method. From the simplified mathematical model optimal allocations from small portfolios are presented, and from the simulation several results from large portfolios using different resource allocation strategies are presented. From these results it is argued that there exists an optimal number of projects to run in a portfolio, and that a portfolio run with either larger or smaller number of running projects produces a lower gain. This number is however found to be highly dependent on the size and specification of the portfolio, and how resources are allocated to projects.