For designers of bioreactor one of the key parameters is how efficiently oxygen is transferred to the cell culture. Designers often rely on empirical formulas to calculate the kLa value. However, when is deviated from the standard design or when upscaling these empirical formulas become inaccurate. Together with Applikon Biotechnology, a leading supplier of bioreactors, we have validated our CFD calculations with measurements and the empirical formulas and aim to better predict bioreactor performance. A good agreement between CFD simulations and measurements/analytical results is found.
To validate the simulations two scenario’s are simulated. The first scenario considers dissolving oxygen solely through ventilating the headspace. This will be a good test scenario for the advanced mass transfer models present in ANSYS Fluent. The second investigated scenario is a more common setup where air is provided to the bioreactor by means of a sparger.
The considered geometry is an autoclavable bioreactor with a marine impeller rotating at a speed of 250 RPM (tip speed 0.59 m/s). It has diameter of 13 cm and a working volume of 2.4 L. The CFD simulations are performed at a temperature of 25 °C.
Note that in the above animation the impeller seems to remain stationary, however the impeller motion is simulated by means of a rotating reference frame. At the start of this simulation the liquid medium is initialized without any oxygen dissolved. The headspace initially contains only nitrogen. At the start of the simulation air (containing oxygen) is supplied at the top of the bioreactor. In the first minute the amount of oxygen increases to standard air conditions (mass fraction of 0.23). After this oxygen is dissolved through the liquid-gas interface. Finally, the kLa value calculated by monitoring the amount of oxygen dissolved in time. For this scenario a value of 0.45 1/h is calculated. This corresponds well to the analytical value which in this case is 0.42 1/h.
As demonstrated by the previous scenario dissolving oxygen through only ventilating the headspace goes quite slowly, which is more suitable for cell line cultures. The second investigated scenario is more suited for microbial cultures, here air is provided to the bioreactor by means of a sparger. The air bubbles are sparged just underneath the impeller and the oxygen contained in these bubbles are dissolved in the liquid medium. For this scenario a kLa value of 7.6 1/h is calculated and oxygen is transferred faster to the cells as compared to the headspace aeration scenario. Based on measurements performed by Applikon Biotechnology a kLa value of 7.3 1/h is reported.
The above simulations have shown that an accurate prediction of the kLa value can be obtained by means of CFD simulations. An advantage of CFD over empirical formula’s or measurements is that the same analysis can be performed for any bioreactor size, presence of baffles, shape of the impeller or other obstructions to the flow without having to worry about empirical formulas still being valid or having to rebuilt an entire measurement setup. Other design parameters such as temperature or bubble size can easily be changed in the CFD simulations. Another benefit is that the CFD analysis can easily be extended to identify potential dead zones as the entire velocity field is calculated.
Acknowledgement: Applikon Biotechnology BV is thanked for providing the bioreactor data to set up these simulations.