(303h) A New Algorithm for Bioprocess Feasibility Index under Uncertainty

To comply FDA strict regulatory requirements, it is necessary to design the bioprocess so that the process performance satisfies the specifications not only at set points but also in a wide range of set points, and limit for the process failure occurs needs to be defined as well. It is crucial to understand the bioprocess feasibility under uncertainty in order to design safe and robust manufacturing process for the new generation therapeutic products arising from advanced life science discovery.

Motivated by the need for feasibility studies of bioprocess operational variables, a new feasibility index to indicate the robust operating ranges for the process has been defined in the sense of a worst case scenario, i.e., the operating point is robustly feasible if it is feasible over the range of variations of the variables. Largest hyperrectangle centred at the operating point is sought to define the process feasibility and its upper and lower bounds for the variations of operating variables. Up to date, the process feasibility under uncertainty is often formulated as a complicated max-min-max problem e.g. Swaney and Grossmann's work, which posts a great computational challenge because of the non-smoothness of the objective functions and non-convexity of the feasible space. A new problem formulation for the problem is presented in this paper. Instead of searching for the largest hyperrectangle, a largest hypersphere inscribed in the feasible space is sought so that the difficulty of non-convexity of feasible space can be overcome. Hence, a hyperrectangule can be defined within the hypersphere. The solution obtained is a lower bound approximation of the true solution. By discretisation, the problem can be formulated as the minimum distance between a point and a point set so that the complexity of the algorithm has been reduced significantly. Such an algorithm can be utilised to find the feasibility index for a given operating point and the optimal operating point with largest feasibility index in the feasible space.