More Efficient Drug Production Using Cloud-Based CFD Simulation of Bioreactors
Aerated stirred reactors, the most common type of both small- and large-scale bioreactors, are used for performing microbial fermentation or mammalian cell culture unit operations for the production of biological therapeutics such as vaccines, hormones, proteins and antibodies. Usually, basic design criteria have been adapted in such a way as to meet the requirements of cells. In particular, the shear sensitivity requires consideration in impeller design, aspect ratio and aeration. Sufficient oxygen transfer and carbon dioxide removal are very important criteria in selecting a bioreactor system. Taking into account the process criteria, the scale-up process of bioreactors still presents a challenge and requires detailed knowledge about diverse fields such as the mixing processes, agitation, aeration, heat and mass transfer, etc.
Computational Fluid Dynamics (CFD) is a simulation approach that can be successfully used for the characterization of bioreactors by evaluating process parameters. Useful processes can be obtained using CFD already in an early development stage of the devices without the need for building-up a prototype. Furthermore, CFD tools can be successfully used in the scale up/down process, in order to reduce the number of prototypes and therewith production costs.
Goals of the experiment
The main challenge in the calculation is the treatment of multiphase systems and long process time of several hours which leads to long calculation time, which is not suitable for industrial application. In order to overcome these limitations and to apply CFD simulations in the development process, a highly optimized workflow and huge computational resources are required. For instance, an estimation of the oxygen mass transfer coefficient for only 1 variant takes about 1 week of computational time (using a single computer with 12 CPUs). This is mainly caused by the need to run transient simulations up to 20 seconds. The time must be completely simulated and in addition the simulation has to run with very small time steps caused by multiphase simulations (e.g. 0.01s).
Simulation is becoming more and more interesting for the industry in the case of comparing several simulation variants but only if the calculation takes no more than 1 week. There are several advantages of high performance computing, such as the possibility of parallel calculations using “supercomputers” and simultaneously full cost control (costs per hour of use). Furthermore it offers an opportunity to use high-end simulation technology in the development process without additional fixed costs (such as licences and hardware costs). An optimized simulation workflow for the HPC/Cloud infrastructure offers SMEs a possibility to exploit the advantage of this technology directly for their own products, without huge investment costs and a long period of vocational adjustment.
The aim of the presented project was to adapt the virtual process for cloud-based multiphase simulations of a bioreactor in order to perform a DoE analysis (Design of Experiments). SES-Tec, as the end-user in the experiment, is planning to offer the obtained know-how and services to their customers with a significantly faster, automated and proven simulation workflow using the CloudFlow infrastructure. Obtained results during the project time will be used for demonstration, where DoE calculations will be shown as a state-of-the-art method for the bioreactor analysis already available for the complex multiphase simulations in pharmaceutical industry.
Using the Cloud-based simulation technology it is possible to carry out analysis of bioreactors with clearly reduced costs due to reduced calculation time of parallel simulation variants and full cost control. Through the parallel calculation of a huge number of variants in a very short calculation time it enables the opportunity to get new customers. Addressing a customer segment defined by the pharma, biopharma, automotive and general engineering industries, in a three-year horizon SES-Tec estimates to face a market size of around 4.7 billion dollars, with a potential share reaching 20,000-50,000 euros/year and leading to an increase of 15 percent of actual turnover due to only Cloud computing. Furthermore, SES-Tec is planning to hire a new employee, which deals exclusively with cloud computation. AVL-List as independent software vendor in the experiment benefits by new customers or more sold licences. Moreover, a new Cloud-based business model was created as well. In the three-year perspective, AVL is expecting an increase of the sold Cloud-based licences by 5-10 percent (about 150,000 euros in three-year perspective). Only for the Cloud computing, AVL is expecting to offer between 2 and 5 new job positions. Furthermore, AVL also expects to gain new clients and to increase the number of AVL-FIRE users up to 15 percent in three-year perspective.
The computational time for design of experiments (DoE) analysis was decreased from 5 weeks to 1 week. Due to huge computational resources in the cloud, all 25 simulations variants can be run in parallel and not one after the other. This number of variants are typical for DoE analysis, but are not limited anymore thanks to Cloud-based technology. Furthermore, the number of simulation variants is no longer related to the in-house hardware resources and therefore no investments are needed. Finally, each bioreactor manufacturer can benefit from proven and validated simulation technology and workflows for this kind of application.