Pathways Involved in Switching between Growth and Differentiation during Murine and Human Early Liver Organogenesis

The chemical and process industry has relied on computational simulations for decades. These simulations help companies better understand, troubleshoot and predict real-world behaviors of their equipment and processes. More recently, computational power and modeling technique advancements have enabled the simulation of complex multiscale systems. These have benefited the chemical, pharmaceutical, bioprocessing and other related industries.

With growing competition in the fast-moving chemical and process industry, engineers must quickly address operational problems to meet production timelines. Reducing computational times is one way to tackle this issue: Problem-adapted CFD simulations, parallelization and the use of fast hardware, e.g., computing on GPUs, helps here to find better answers faster.

But typically, CFD simulations are focusing on narrow aspects of a given problem, so that engineers may risk missing important aspects of the entire process. E.g., the performance of an apparatus may depend on the upstream and/or downstream or a specific assumption might no longer hold. So instead, issues must often be address holistically.

In this contribution we will address both aspects: We will demonstrate and focus on how running finite volume CFD simulations on GPU with Simcenter STAR-CCM+ by Siemens reduce the turn-around time for transient as well as steady-state mixing tank simulations by orders of magnitude. But even more important is to bring the insights from CFD simulation to operation. And this can only be achieved by using the digital twin approach for at least the next few years or maybe decades, no matter how far a CFD simulation can be accelerated. Therefor we will briefly discuss based on two use cases the use of Reduced Order Model (ROM) and it's use for the executable digital twin.