The problem is, however, that these tools and concepts, namely scale-down
bioreactors, high-throughput mini bioreactors, and model-based tools, have mainly
developed in parallel, with little or no interaction among them. In fact, scale-down
bioreactors are still operated as standalone, low-throughput devices [3]; and the
benefits of mathematical models are not fully exploited in both scale-down and highthroughput systems [29]. Therefore, the actual challenge is to combine these very
special techniques such that they can work together efficiently.
In this work, we discuss the current state of process development focusing on
scale-down, the typically underestimated milestone. We discuss existing experimental tools, sensor technologies, and latest advances in computational methods for
the design of scale-down investigations. Next, we demonstrate the issues related to
the current decoupled efforts to address process development. Finally, we describe
the required steps to reach a proper integration of all tools to create a digital twin of
the bioprocess development procedure together with its potential and future
applications.
2 The Digital Twin in Bioprocess Development
The answer to the current challenges in advanced bioprocess development (see
Fig. 1) is a digital twin that covers all developmental stages and allows an efficient
and effortless transfer of knowledge and information throughout the complete
process [30]. Thus, the term “digital twin” covers more than just the mathematical
model of a single component of the process. With regard to industrial bioprocesses,
“digital twins” can describe the biological system itself or parts of the system,
e.g. the three-dimensional structure of the protein product. They can also describe
phases of process development, such as strain screening, different scales of the
fermentation process and downstream operations, and in final production they can
be used for the design or installation of a production plan, as well as for the control of
the actual manufacturing process including its optimization.
The required advances in automation, process analytical technologies (PAT), and
computer-aided tools for bioprocess monitoring and control are available [31]. The
main challenge in building a functional digital twin is the difficulty in harmonizing
these existing technologies through standardized communication protocols and data
management systems. Such a digital framework tightly embedded into the highly
automated experimental systems and production facilities through PAT and
advanced mathematical modeling tools can build the path for knowledge transfer
between the whole bioprocess development workflow.
Scale-up and scale-down present arguably the most descriptive examples for the
challenges of knowledge transfer as well as its relevance in bioprocess development
[32, 33]. Scaling is basically an effort to transfer the information generated in one
stage to another aiming to maximize the generation of relevant knowledge for the
industrial process [17].
Potential of Integrating Model-Based Design of Experiments Approaches and. . .
5
bioreactors, high-throughput mini bioreactors, and model-based tools, have mainly
developed in parallel, with little or no interaction among them. In fact, scale-down
bioreactors are still operated as standalone, low-throughput devices [3]; and the
benefits of mathematical models are not fully exploited in both scale-down and highthroughput systems [29]. Therefore, the actual challenge is to combine these very
special techniques such that they can work together efficiently.
In this work, we discuss the current state of process development focusing on
scale-down, the typically underestimated milestone. We discuss existing experimental tools, sensor technologies, and latest advances in computational methods for
the design of scale-down investigations. Next, we demonstrate the issues related to
the current decoupled efforts to address process development. Finally, we describe
the required steps to reach a proper integration of all tools to create a digital twin of
the bioprocess development procedure together with its potential and future
applications.
2 The Digital Twin in Bioprocess Development
The answer to the current challenges in advanced bioprocess development (see
Fig. 1) is a digital twin that covers all developmental stages and allows an efficient
and effortless transfer of knowledge and information throughout the complete
process [30]. Thus, the term “digital twin” covers more than just the mathematical
model of a single component of the process. With regard to industrial bioprocesses,
“digital twins” can describe the biological system itself or parts of the system,
e.g. the three-dimensional structure of the protein product. They can also describe
phases of process development, such as strain screening, different scales of the
fermentation process and downstream operations, and in final production they can
be used for the design or installation of a production plan, as well as for the control of
the actual manufacturing process including its optimization.
The required advances in automation, process analytical technologies (PAT), and
computer-aided tools for bioprocess monitoring and control are available [31]. The
main challenge in building a functional digital twin is the difficulty in harmonizing
these existing technologies through standardized communication protocols and data
management systems. Such a digital framework tightly embedded into the highly
automated experimental systems and production facilities through PAT and
advanced mathematical modeling tools can build the path for knowledge transfer
between the whole bioprocess development workflow.
Scale-up and scale-down present arguably the most descriptive examples for the
challenges of knowledge transfer as well as its relevance in bioprocess development
[32, 33]. Scaling is basically an effort to transfer the information generated in one
stage to another aiming to maximize the generation of relevant knowledge for the
industrial process [17].
Potential of Integrating Model-Based Design of Experiments Approaches and. . .
5