Multi-objective optimisation for biopharmaceutical manufacturing under uncertainty

This work addresses the multi-objective optimisation of manufacturing strategies of monoclonal antibodies under uncertainty. The chromatography sequencing and column sizing strategies, including resin at each chromatography step, number of columns, column diameters and bed heights, and number of cycles per batch, are optimised. The objective functions simultaneously minimise the cost of goods per gram and maximise the impurity reduction ability of the purification process. Three parameters are treated as uncertainties, including bioreactor titre, and chromatography yield and capability to remove impurities. Using chance constraint programming techniques, a multi-objective mixed integer optimisation model is proposed. Adapting both ε-constraint method and Dinkelbach's algorithm, an iterative solution approach is developed for Pareto-optimal solutions. The proposed model and approach are applied to an industrially-relevant example, demonstrating the benefits of the proposed model through Monte Carlo simulation. The sensitivity analysis of the confidence levels used in the chance constraints of the proposed model is also conducted.