bioreactor design may aim to create the least heterogeneous impact possible
[1]. Independent of the bioreactor type at hand, heterogeneities happen in large
scale, and as a result product loss seems inevitable while scaling up. Heterogeneities
expose cell to various stresses which will in turn translate to product loss. Trail
experiments in actual industrial bioreactors are not feasible at all, and that is one of
the main motivations of a “digital twin” of the bioreactor.
A digital twin allows us to investigate and locate the physical and biological
bottleneck. It enables the user to predict various scenarios in parallel saving significant resources. Once the digital twin of the bioreactor is validated, it can be
translated to a scale-down bioreactor that represents the imperfections of the larger
scale. This setup can be investigated to reveal the most significant biological
behaviors of the cell and ultimately providing the digital twin for the cell.
In theory once digital twins for the cell and for the bioreactor are available,
versatile optimization of bioprocesses would be carried out in only a fraction of the
time and resource. This offers the operator the choice between improving the
bioreactor, microorganism, and both.
2 Embedding Cells in Microenvironmental Heterogeneities
of Bioreactors
To predict cellular responses in large-scale bioreactors, two crucial prerequisites are
needed: (1) models simulating spatially resolved substrate availabilities, flows, and
mass transfer and (2) models translating microenvironmental heterogeneities into
proper cellular responses. Research of the last years enabled substantial improvements in both fields, computational fluid dynamics and thorough experimental
studies, thereby providing the ground for large-scale prediction of microbial performance a priori.
Large-scale bioreactor conditions need to be calculated, aiming at a spatial
resolution of mass, momentum, and energy balances via numerical simulations. In
particular, Navier-Stokes (NSE) and continuity equations representing the conservation of momentum and mass should be solved. Basically, NSEs describe the
motion of viscous fluid flows with the fluids considered as a continuum rather than
colliding particles. Under the typical mixing conditions given, the occurrence of
turbulent zones is likely. They are integrated via additional transport equations
typically demanding for two additional equations depending on the models applied.
Turbulence is provoked by eddies which affect molecular diffusion, heat transfer,
and the mixing behavior. For flows involving heat transfer or compressibility, an
additional equation for energy conservation should be solved. Furthermore, the
balancing of individual species (particles or cells) that undergo mixing and reactive
changes requires the implementation of proper conservation terms. Depending on
the available computing power, simulations can range from direct numerical simulation (DNS) via large eddy simulation (LES) to Reynolds-averaged Navier-Stokes
Euler-Lagrangian Simulations: A Proper Tool for Predicting Cellular Performance. . .
231
[1]. Independent of the bioreactor type at hand, heterogeneities happen in large
scale, and as a result product loss seems inevitable while scaling up. Heterogeneities
expose cell to various stresses which will in turn translate to product loss. Trail
experiments in actual industrial bioreactors are not feasible at all, and that is one of
the main motivations of a “digital twin” of the bioreactor.
A digital twin allows us to investigate and locate the physical and biological
bottleneck. It enables the user to predict various scenarios in parallel saving significant resources. Once the digital twin of the bioreactor is validated, it can be
translated to a scale-down bioreactor that represents the imperfections of the larger
scale. This setup can be investigated to reveal the most significant biological
behaviors of the cell and ultimately providing the digital twin for the cell.
In theory once digital twins for the cell and for the bioreactor are available,
versatile optimization of bioprocesses would be carried out in only a fraction of the
time and resource. This offers the operator the choice between improving the
bioreactor, microorganism, and both.
2 Embedding Cells in Microenvironmental Heterogeneities
of Bioreactors
To predict cellular responses in large-scale bioreactors, two crucial prerequisites are
needed: (1) models simulating spatially resolved substrate availabilities, flows, and
mass transfer and (2) models translating microenvironmental heterogeneities into
proper cellular responses. Research of the last years enabled substantial improvements in both fields, computational fluid dynamics and thorough experimental
studies, thereby providing the ground for large-scale prediction of microbial performance a priori.
Large-scale bioreactor conditions need to be calculated, aiming at a spatial
resolution of mass, momentum, and energy balances via numerical simulations. In
particular, Navier-Stokes (NSE) and continuity equations representing the conservation of momentum and mass should be solved. Basically, NSEs describe the
motion of viscous fluid flows with the fluids considered as a continuum rather than
colliding particles. Under the typical mixing conditions given, the occurrence of
turbulent zones is likely. They are integrated via additional transport equations
typically demanding for two additional equations depending on the models applied.
Turbulence is provoked by eddies which affect molecular diffusion, heat transfer,
and the mixing behavior. For flows involving heat transfer or compressibility, an
additional equation for energy conservation should be solved. Furthermore, the
balancing of individual species (particles or cells) that undergo mixing and reactive
changes requires the implementation of proper conservation terms. Depending on
the available computing power, simulations can range from direct numerical simulation (DNS) via large eddy simulation (LES) to Reynolds-averaged Navier-Stokes
Euler-Lagrangian Simulations: A Proper Tool for Predicting Cellular Performance. . .
231