Activity: Talk or presentation › Talk at conference or symposium › Science to science
Computer simulation has become a vital tool in learning to understand the processes inside multiphase reactors such as aerated stirred tanks used in biopharmaceutical manufacturing. This in-silico design approach allows for the identification of bottlenecks that limit the overall efficiency of large scale bioreactors, without relying on empirical knowledge gained on a pilot-plant.
In order to simulate the three dimensional flow field inside multiphase reactors in an acceptable wall time, a highly-parallelized implementation of the lattice Boltzmann method is performed on state-of-the-art graphics processing units. A Lagrangian model is used to describe the movement of air bubbles inside the reactor, including coalescence and break-up effects. The phases are two-way coupled. Advection of substrate, oxygen and carbon dioxide is modelled by a novel species transport model that takes advantage of the high degree of parallelization.
An additional carbon source in the form of emulsified oil drops is modelled by locally solving a population balance equation including a coalescence and break-up kernel. The volume fraction of oil droplets affects the local fluid mixture viscosity. Local parameters such as oil, oxygen and substrate concentration as well as shear rate are used in a biological model that predicts the productivity of individual biological cells. These cells are introduced as massless particles which follow the fluid flow field. The trajectory and age of each cell inside the reactor affect the biological model.
Simulations are made for several bioreactor sizes ranging from lab scale up to the full industrial production scale. The results are compared to literature data. Additionally, the gas phase phase movement, holdup and gas absorption is validated with measurements made on a 150l pilot-plant reactor.