The benefits of continuous biologics manufacturing are generally accepted, but transitioning to continuous processing from batch processing is hindered by the lack of modeling and simulation tools. Noticeably absent are the comprehensive, open-source, facility-wide processing simulation platforms needed for decision support.
Despite the availability of individual process models, “the existing toolkit for process synthesis, design, control, and optimization heavily relies on experiential and recipe-driven knowledge, and employs rule-of-thumb techniques,” Seyed Soheil Mansouri, PhD, associate professor, department of chemical and biochemical engineering, Technical University of Denmark (DTU), tells GEN. Mansouri is also co-founder of Sqale ApS, a biologics manufacturing company spun out of DTU.
An alternative, which Mansouri and colleagues outlined in a recent paper, is a comprehensive, plant-wide model for continuous biomanufacturing benchmarking.
Dubbed KTB1, this decision support tool consists of an upstream element to help maximize productivity, a control system to monitor and control unit operations and synchronization, and a downstream structure to isolate products.
It helps biomanufacturers evaluate “monitoring strategies, process design, process optimization, and control for biomanufacturing,” the paper points out. It features tailored strategies to control dynamic simulations as well as noise and lag time from process sensors and actuators for greater accuracy.
Model melds up- and downstream
KTB1, the scientists report, “integrates upstream and downstream processes based on industrial guidance, literature review, and thermodynamic principles.” It factors in initial values, constant variables, parameters, and assumptions so users can more accurately determine their effects on the final product. Significantly, the model allows new research to be added and host organisms to be changed.
The model’s control strategy is based upon a combination of industrial perspectives, literature reports, and mathematical indices (i.e., the Niederlinski index), as well as sensor control models to improve realism and applicability in bioprocessing. KTB1 also includes a systematic framework structure that guides biomanufacturers to perform their own experimentation and analyses as they optimize processes.
At-scale validation needed
This model does have some limitations, Mansouri and colleagues admit. “The model must be used with care since it is not validated with manufacturing data at scale,” he says, and has not amassed a comprehensive data library.
“This is a theoretical framework, rather than a validated representation of an existing process, since obtaining manufacturing data in biologics manufacturing is near to impossible because of various technical and legal dimensions,” Mansouri elaborates.
“However,” he continues, “the model structure, configuration, and solution strategy—from a model-based viewpoint—is valid, as it captures an entire plant’s dynamics.” It is a tool “to study and develop strategies in biomanufacturing…based on expert judgment and heuristics.”
The developers modeled the process using the data, equipment, and production parameters used to produce lovastatin, an active pharmaceutical ingredient produced in a fermentation-based manufacturing context. The KTB1 model, however, can be adapted to other types of reactions, and applies to biopharmaceuticals as well as to bioprocesses for other applications, such as specialty chemical manufacturing or food processing.
The platform is available open-source via GitHub, along with a comprehensive user manual. It is implemented in MATLAB.
