severe consequences of plant or operator failure such as the offshore oil and gas
industry [7, 54, 55]. Older educational facilities for training in the oil and gas
industry were based on physical copies of the control room, which are expensive
and no longer needed [54]. Almost simultaneously with the first appearance of
Digital Twins in the chemical industry, they were used as a tool for control strategy
development [54]. In the beginning, these were relatively simple control engineering
tasks, but they became more complex with the advancing development of Digital
Twins [54, 55].
Dudley et al. (2008) [7] described the use of a Digital Twin of a pebble bed
modular reactor plant for the development and testing of control strategies before
using them on the real plant. He et al. (2019) [4] described the use of a Digital Twin
for the Tennessee Eastman benchmark process. Effectiveness and performance of
the Digital Twin in the development of control strategies were demonstrated in the
presence of realistic fault scenarios. Three types of process faults, i.e., sensor faults,
actuator faults and process disturbances were investigated and the corresponding
fault size and temporal behaviour were discussed. All simulation studies and numerical results indicated that the proposed configurations are valid for safe operations in
the event of a process fault. Zhang et al. (2019) [6] described the use of a Digital
Twin for carbon emission reduction in intelligent manufacturing. Here, the plants’
carbon emission is predicted by the Digital Twin model. A carbon emission control
strategy was then optimised utilising the Digital Twin, to minimise exhaust gas
emissions.
Compared to chemical processes, the application of Digital Twins for
bioprocesses is still in its infancy. Thoroughness is required for modelling
bioprocesses since a wide variety of parallel reactions take place at the same time.
Even small changes of key process variables, such as pH or temperature, may have
an immense influence on the kinetics [33].
Pörtner et al. (2011) used an “early-stage” Digital Twin for the optimisation of
process control strategies for mammalian cell cultivations [56]. The developed
bioprocess simulator is a digital replica of the cultivation of mammalian cell lines
in a small scale STR. The bioprocess simulator was used to simulate the impact of
various constant feed rates of glucose and glutamine during fed-batch on cell density
and antibody concentration of a mammalian cell line. The feed rates were determined by design of experiments (DoE) methods. By using the bioprocess simulator,
the cultivation process could be optimised in a considerably shorter time and fewer
experiments compared to process control optimisation on the real process.
In a contribution by Hass et al. [17] the utilisation of an industrial biotechnology
OTS was presented. Control strategies that were developed using a new bioethanol
plant OTS illustrated the potential for enhanced resource efficiency and reduced
energy consumption. According to the authors, the potential savings in raw materials
have a direct impact on the long-term profitability of the bioethanol plant and enable
a reduction of operating costs. By using the OTS, the time course and dynamics of
the entire plant could be analysed and subsequently optimised using new process
control strategies. Performing such a study on a real plant would have been overly
complex and expensive, if not impossible.
74
C. Appl et al.
industry [7, 54, 55]. Older educational facilities for training in the oil and gas
industry were based on physical copies of the control room, which are expensive
and no longer needed [54]. Almost simultaneously with the first appearance of
Digital Twins in the chemical industry, they were used as a tool for control strategy
development [54]. In the beginning, these were relatively simple control engineering
tasks, but they became more complex with the advancing development of Digital
Twins [54, 55].
Dudley et al. (2008) [7] described the use of a Digital Twin of a pebble bed
modular reactor plant for the development and testing of control strategies before
using them on the real plant. He et al. (2019) [4] described the use of a Digital Twin
for the Tennessee Eastman benchmark process. Effectiveness and performance of
the Digital Twin in the development of control strategies were demonstrated in the
presence of realistic fault scenarios. Three types of process faults, i.e., sensor faults,
actuator faults and process disturbances were investigated and the corresponding
fault size and temporal behaviour were discussed. All simulation studies and numerical results indicated that the proposed configurations are valid for safe operations in
the event of a process fault. Zhang et al. (2019) [6] described the use of a Digital
Twin for carbon emission reduction in intelligent manufacturing. Here, the plants’
carbon emission is predicted by the Digital Twin model. A carbon emission control
strategy was then optimised utilising the Digital Twin, to minimise exhaust gas
emissions.
Compared to chemical processes, the application of Digital Twins for
bioprocesses is still in its infancy. Thoroughness is required for modelling
bioprocesses since a wide variety of parallel reactions take place at the same time.
Even small changes of key process variables, such as pH or temperature, may have
an immense influence on the kinetics [33].
Pörtner et al. (2011) used an “early-stage” Digital Twin for the optimisation of
process control strategies for mammalian cell cultivations [56]. The developed
bioprocess simulator is a digital replica of the cultivation of mammalian cell lines
in a small scale STR. The bioprocess simulator was used to simulate the impact of
various constant feed rates of glucose and glutamine during fed-batch on cell density
and antibody concentration of a mammalian cell line. The feed rates were determined by design of experiments (DoE) methods. By using the bioprocess simulator,
the cultivation process could be optimised in a considerably shorter time and fewer
experiments compared to process control optimisation on the real process.
In a contribution by Hass et al. [17] the utilisation of an industrial biotechnology
OTS was presented. Control strategies that were developed using a new bioethanol
plant OTS illustrated the potential for enhanced resource efficiency and reduced
energy consumption. According to the authors, the potential savings in raw materials
have a direct impact on the long-term profitability of the bioethanol plant and enable
a reduction of operating costs. By using the OTS, the time course and dynamics of
the entire plant could be analysed and subsequently optimised using new process
control strategies. Performing such a study on a real plant would have been overly
complex and expensive, if not impossible.
74
C. Appl et al.