Complexity Science mini-project: Integrated Capacity Planning and Portfolio Selection for Biopharmaceutical Production Richard Allmendinger (UCL), Juergen Branke (Warwick Business School), Suzy Farid (UCL) Background: Development and production of biopharmaceutical drugs is expensive and risky. Efficient capacity planning and portfolio selection are crucial, but at the same time become ever more complex as the number of potential drug candidates increases and their manufacturing processes vary. Furthermore, decisions are interdependent, i.e., the decision about which drugs to develop depends on the available capacities and resource needs of the other drugs, and vice versa. And because of the uncertainties involved in drug development and testing, plans have to be continually updated as new test results become available [1]. In a research project funded by EPSRC and in collaboration between UCL and Warwick, we develop novel decision-support tools for such scenarios. Mini-project: The mini-project will develop an evolutionary algorithm and/or approximate dynamic programming in combination with simulation to optimize the selection of drugs to be developed as well as the schedule. The project thus consists of three parts: Modelling (which aspects/constraints of the problem are most relevant and need to be modelled accurately), methodology development (designing and implementing the optimization algorithm) and testing the developed algorithms on a given dataset. PhD prospect: The mini-project is only a first step in developing a tool to support biopharmaceutical companies in making selection and planning decisions in this complex environment. Promising techniques to be used are evolutionary algorithms or approximate dynamic programming, and a sound comparison of these techniques would have relevance also for other applications. Notable extensions of the problem scenario include the consideration of multiple objectives [2,3], the alternative of outsourcing certain production steps to third parties [4], or the consideration of changeover times [1]. Because a simulation-based approach to evaluate solutions may become computationally very expensive as the problem size grows, one may have to integrate surrogate models [6], which opens an entire new field of interesting research questions. EPSRC funding for a continuation of the project as PhD student may be available. Deliverables: Literature survey on pharmaceutical project selection Simulation model for a simple biopharmaceutical drug development and production process Either an evolutionary algorithm or an approximate dynamic programming model to optimize the combined drug selection and production planning problem Student’s requirements: Programming skills Exemplary bibliography [1] J. Yaochu and J. Branke. Evolutionary optimization in uncertain environments-a survey. IEEE Transactions on Evolutionary Computation 9(3): 303-317 (2005) [2] J. Branke, K. Deb, K. Miettinen, and R. Slowinski (Eds.). Multiobjective optimization: Interactive and evolutionary approaches (Vol. 5252), Springer (2008) [3] E. George, N. J. Titchener-Hooker, and S. S. Farid. A multi-criteria decision-making framework for the selection of strategies for acquiring biopharmaceutical manufacturing capacity. Computers & Chemical Engineering 31(8): 889-901 (2007) [4] C. Siganporia, S. Ghosh, T. Daszkowski, L. G. Papageorgiou, and S. S. Farid. Production planning of batch and semi-continuous bioprocesses across multiple biopharmaceutical facilities. Computer Aided Chemical Engineering 30:377-381 (2012) [5] E. George and S. S. Farid. Stochastic Combinatorial Optimization Approach to Biopharmaceutical Portfolio Management. Industrial and Engineering Chemistry Research 47(22): 8762-8774 (2008) [6] J. Knowles and H. Nakayama. Meta-modeling in multiobjective optimization. Multiobjective Optimization. Springer Berlin Heidelberg, pp 245-284 (2008).