Predictive modeling of single pass tangential flow filtration for continuous biomanufacturing

Opportunities for process intensification have made continuous biomanufacturing an area of active research. While tangential flow filtration (TFF) is typically employed within the biologics purification train to increase drug substance concentration, single-pass TFF (SPTFF) modifies its format by enabling continuity of this process and achieving a multifold concentration factor through a single-pass over the filtration membranes. In continuous processes feed concentration and flow rate are determined by the preceding unit operations. Therefore, tight control of SPTFF output concentration must be achieved through precise design of the membrane configuration, unlike TFF. However, predictive modeling can be utilized to identify configurations that achieve a desired target concentration across ranges of possible feed conditions with minimal experimental data, hence enabling accelerated process development and design flexibility. We hereby describe the development of a mechanistic model predicting SPTFF performance across a wide design space using the well-established stagnant film model, which we demonstrate is more accurate at higher feed flow rates. The flux excursion dataset was generated within time constraints and with minimal material consumption, showing the method's ability to be quickly adapted. While this approach eliminates characterizing complex physicochemical model variables or the need for users with specialized training, the model and its assumptions become inaccurate at low flow rates, below 25 L/m²/h, and high conversions, above 0.9. As this low flow rate, high conversion operating regime is relevant for continuous biomanufacturing, we explore the assumptions and challenges involved in predicting and modeling SPTFF processes, while suggesting added characterization to gain further process insight.