Advances in Industry 4.0 - Development of a Smart Robotic Platform for Autonomous Synthesis of Downstream Processes

The perspective of Industry 4.0 is the development of smart and interconnected systems aiming at equipment automation and flexible operation [1]. For biochemical process design automating the upstream section has been focused on a lot in academia and industry (e.g., [2, 3]). Since a considerable amount of costs for biochemical products arises from purification steps, automating the design of downstream processes will enable efficient process development of complete production processes.

There is a shortcoming of smart platforms which can autonomously synthesize and carry out an entire downstream process with the aid of intelligent software. Especially in early stages of process development, many decisions based on incomplete data must be made by engineers. However, such decisions will have extensive consequences on performance and efficiency of the process. Up to now, robotic platforms are applied for experimental purposes only without intelligent decision-authorized software. Thus, our new approach focuses on the overall automation of downstream process design, including decision-making. For this purpose, experiments are carried out in a miniaturized form on a custom-built robotic platform, which executes several unit operations, e.g., extraction and adsorption in an automated and resource-efficient manner. Flexible interconnection of these purification steps leads to a complete downstream process generated by the robotic platform itself.

The presentation will demonstrate the development and validation of an automated design methodology for the separation of steroids out of a fermentation broth consisting of two liquid phases, soybean oil and water. The steroids are located mainly in the soybean oil phase. The decision-making process of different unit operations, e.g., extraction and adsorption, is transferred to the software of the robotic platform. For a liquid-liquid extraction, for example, extraction solvents are selected, miscibility gaps are investigated, and operating conditions are tested motivating each decision economically. A systematic concept has been developed for solvent selection applicable to every complex material system. Thereby, a broad range of solvents with different polarities is tested automatically and ranked by key performance indicators (KPIs) like the purification performance index (PPI) and the separation cost index (SCI) [4].

Furthermore, these solvents are tested automatically on miscibility gaps on the robotic platform. Pictures are taken via an integrated camera and MATLAB^® is used to detect a miscibility gap between feed solvent and extraction solvent, and the phase volumes (Figure 1). Suitable solvents are selected and ranked by KPIs. Then, the solvent with the most promising performance, indicated by the lowest SCI, is chosen for further experiments to find optimal operating conditions. For the soybean oil phase of the fermentation broth possible solvents for extraction are acetonitrile, methanol, ethanol and water. Water is neglected due to the low solubility of the steroids resulting in yields < 1 %. The lowest SCI of about 67 € g_steroid^-1 is reached for the extraction with acetonitrile compared to around 70 € g_steroid^-1 with methanol and about 85 € g_steroid^-1 with ethanol. So far, the extraction experiments are executed for a phase ratio of 4 g_solvent g_feed^-1 and thereby a constant amount of solvent. In subsequent experiments the influence of varying solvent quantities on the SCI will be investigated.

The presentation will end with evaluating the possibility of transferring the methodology developed to other unit operations, e.g., adsorption. Thereby, important decisions are the selection of a suitable adsorbent, the determination of operating conditions and the design of a desorption step as well. First experimental results will be presented and discussed.

References:

[1] B. Sun, S.-L. Jämsä-Jounela, Y. Todorov, L. E. Olivier, I. K. Craig, IFAC-PapersOnLine 2017, 50 (2), 65 – 70.

[2] S. Zimmermann, S. Gretzinger, M.-L. Schwab, C. Scheeder, P. K. Zimmermann, S. A. Oelmeier, E. Gottwald, A. Bogsnes, M. Hansson, A. Staby, J. Hubbuch, Journal of chromatography. A 2016, 1464, 1 – 11.

[3] P. Xu, C. Clark, T. Ryder, C. Sparks, J. Zhou, M. Wang, R. Russell, C. Scott, Biotechnology progress 2017, 33 (2), 478 – 489.

[4] T. Winkelnkemper, G. Schembecker, Separation and Purification Technology 2010, 72 (1), 34 – 39.

Acknowledgment: The research leading to these results has received funding from the Federal Ministry of Education and Research (Funding Code: 031B0620).