The conceptual process design of novel bioprocesses in biorefinery setups is an important task,which remains yet challenging due to several limitations.We propose a novel framework incorporating superstructure optimiz...The conceptual process design of novel bioprocesses in biorefinery setups is an important task,which remains yet challenging due to several limitations.We propose a novel framework incorporating superstructure optimization and simulation-based optimization synergistically.In this context,several approaches for superstructure optimization based on different surrogate models can be deployed.By means of a case study,the framework is introduced and validated,and the different superstructure optimization approaches are benchmarked.The results indicate that even though surrogate-based optimization approaches alleviate the underlying computational issues,there remains a potential issue regarding their validation.The development of appropriate surrogate models,comprising the selection of surrogate type,sampling type,and size for training and cross-validation sets,are essential factors.Regarding this aspect,satisfactory validation metrics do not ensure a successful outcome from its embedded use in an optimization problem.Furthermore,the framework’s synergistic effects by sequentially performing superstructure optimization to determine candidate process topologies and simulationbased optimization to consolidate the process design under uncertainty offer an alternative and promising approach.These findings invite for a critical assessment of surrogatebased optimization approaches and point out the necessity of benchmarking to ensure consistency and quality of optimized solutions.展开更多
The domain of industrial biomanufacturing is enthusiastically embracing the concept of Digital Twin,owing to its promises of increased process efficiency and resource utilisation.However,Digital Twin in biomanufacturi...The domain of industrial biomanufacturing is enthusiastically embracing the concept of Digital Twin,owing to its promises of increased process efficiency and resource utilisation.However,Digital Twin in biomanufacturing is not yet clearly defined and this sector of the industry is falling behind the others in terms of its implementation.On the other hand,some of the benefits of Digital Twin seem to overlap with the more established practices of process control and optimization,and the term is vaguely used in different scenarios.In an attempt to clarify this issue,we investigate this overlap for the specific case of fermentation operation,a central step in many biomanufacturing processes.Based on this investigation,a framework built upon a five-step pathway starting from a basic steady-state process model is proposed to develop a fully-fledged Digital Twin.For demonstration purposes,the framework is applied to a bench-scale second-generation ethanol fermentation process as a case study.It is proposed that the success or failure of a fully-fledged Digital Twin implementation is determined by key factors that comprise the role of modelling,human operator actions,and other propositions of economic value.展开更多
基金Supported by the Mexican National Council for Science and Technology (CONACyT# 118903)the Danish Research Council for Technology and Production Sciences (FTP# 274-07-0339)
基金The authors would like to express their gratitude to the Novo Nordisk Foundation(Grant No.NNF17SA0031362)for funding the Fermentation-Based Biomanufacturing Initiative of which this project is a part.
文摘The conceptual process design of novel bioprocesses in biorefinery setups is an important task,which remains yet challenging due to several limitations.We propose a novel framework incorporating superstructure optimization and simulation-based optimization synergistically.In this context,several approaches for superstructure optimization based on different surrogate models can be deployed.By means of a case study,the framework is introduced and validated,and the different superstructure optimization approaches are benchmarked.The results indicate that even though surrogate-based optimization approaches alleviate the underlying computational issues,there remains a potential issue regarding their validation.The development of appropriate surrogate models,comprising the selection of surrogate type,sampling type,and size for training and cross-validation sets,are essential factors.Regarding this aspect,satisfactory validation metrics do not ensure a successful outcome from its embedded use in an optimization problem.Furthermore,the framework’s synergistic effects by sequentially performing superstructure optimization to determine candidate process topologies and simulationbased optimization to consolidate the process design under uncertainty offer an alternative and promising approach.These findings invite for a critical assessment of surrogatebased optimization approaches and point out the necessity of benchmarking to ensure consistency and quality of optimized solutions.
基金The work is funded by the Novo Nordisk Foundation in the frame of the‘Accelerated Innovation in Manufacturing Biologics’(AIMBio)project(Grant number NNF19SA0035474).
文摘The domain of industrial biomanufacturing is enthusiastically embracing the concept of Digital Twin,owing to its promises of increased process efficiency and resource utilisation.However,Digital Twin in biomanufacturing is not yet clearly defined and this sector of the industry is falling behind the others in terms of its implementation.On the other hand,some of the benefits of Digital Twin seem to overlap with the more established practices of process control and optimization,and the term is vaguely used in different scenarios.In an attempt to clarify this issue,we investigate this overlap for the specific case of fermentation operation,a central step in many biomanufacturing processes.Based on this investigation,a framework built upon a five-step pathway starting from a basic steady-state process model is proposed to develop a fully-fledged Digital Twin.For demonstration purposes,the framework is applied to a bench-scale second-generation ethanol fermentation process as a case study.It is proposed that the success or failure of a fully-fledged Digital Twin implementation is determined by key factors that comprise the role of modelling,human operator actions,and other propositions of economic value.
