No abstract

This paper reports the optimization of a perfusion bioreactor system previously reported by us (Chouinard et al., 2009). The implementation of a proportional-integral (PI) controller algorithm to control oxygen concentration and pH is presented and discussed. P and I values used by the controller were first estimated using a First-Order-Plus-Dead-Time (FOPDT, Matlab Simulink) and then tuned manually. A new gas exchanger design compatible with the PI controller was introduced and validated to decrease interaction between the injected gases and overall inertia of the system. The gas exchanger was used to adjust both pH and dissolved oxygen (DO) concentration. This new bioreactor system allowed real-time PI control over pH and DO concentration at different flow rates (from 2 to 70 mL min(-1)). Cell viability and proliferation were investigated to validate the updated bioreactor design and performance.

This study reveals that residence time distribution (RTD) analysis with pH monitoring after acid bolus injection can be used to globally study the flow dynamics of a perfusion bioreactor, while fluorescence microscopy and magnetic resonance imaging (MRI) were used to locally investigate mass transport within a hydrogel scaffold seeded or not with cells. The bioreactor used in this study is a close-loop tubular reactor. A dispersion model in one dimension has been used to describe the non-ideal behavior of the reactor. From open-loop experiments (single-cycle analysis), the presence of stagnant zones and back mixing were observed. The impact of the flow rate, the compliance chamber volume and mixing were investigated. Intermediate flows (30, 45, 60, and 90 mL min(-1)) had no effect over RTD function expressed in reduced time (θ). Lower flow rates (5 and 15 mL min(-1)) were associated to smaller extent of dispersion. The compliance chamber volume greatly affected the dynamics of the RTD function, while the effects of mixing and flow were small to non-significant. An empirical equation has been proposed to localize minima of the RTD function and to predict Per . Finally, cells seeded in a gelatin gel at a density of 800,000 cells mL(-1) had no effect over the permeability and the apparent diffusion coefficient, as revealed by fluorescent microscopy and MRI experiments.