32
2 Simulation-Based Optimization
p. 43]. In literature, most authors refer to the coupling form shown in case d) as
the classical simulation-based optimization or simulation optimization. Therefore,
this work is as well based on this type of coupling of simulation and optimization
when talking about simulation-based optimization.
In addition to sequential and hierarchical coupling, another way of categorizing the mode of coupling is to look at how the interaction between optimization
and simulation is implemented [VD2016a, p. 4]. The method’s interaction mode
can be totally independent, meaning the two software tools run independently
and the data exchange between them is done manually. Alternatively, the interaction mode can be based on a common data exchange format. The tools still act
independently without any communication during runtime but the data exchange
after every completed run is executed automatically. A third option of interaction
is the inclusion of the other method respectively by a software-based coordination mechanism, for example using a communication or control platform to
parameterize, control, and synchronize bidirectional interactions simulation and
optimization. The fourth implementation level is the direct communication of
simulation and optimization tools as well as the mutual coordination through builtin interfaces. [VD2016a, p. 4]. “Researchers have proposed and developed many
different methods that attempt to optimize a simulation by searching through the
space of possible input-factor combinations […] with the results from simulating
earlier configurations being used to suggest promising new directions to […] better system performance” [La2007, p. 658]. The commercial and freely available
software systems already provide optimization components that can efficiently
support this search process. However, the concrete implementation of the interfaces between simulation and optimization as well as the application of tools and
the interpretation of the results remain mainly the tasks of the user [VD2016a,
p. 5].
2.3.2 Objectives and Challenges of Simulation-based
Optimization
The simulation-based optimization has the objective to improve a process by
checking whether a change in values of variables has an influence on the process
and which process inputs are most influential on the process outputs of interest.
The basic idea behind this approach is to find an optimal solution by using multiple replications simulating different system configurations [Fu2013, p. 1418].
The simulation is started by the optimization, produces the result data, and forms
2 Simulation-Based Optimization
p. 43]. In literature, most authors refer to the coupling form shown in case d) as
the classical simulation-based optimization or simulation optimization. Therefore,
this work is as well based on this type of coupling of simulation and optimization
when talking about simulation-based optimization.
In addition to sequential and hierarchical coupling, another way of categorizing the mode of coupling is to look at how the interaction between optimization
and simulation is implemented [VD2016a, p. 4]. The method’s interaction mode
can be totally independent, meaning the two software tools run independently
and the data exchange between them is done manually. Alternatively, the interaction mode can be based on a common data exchange format. The tools still act
independently without any communication during runtime but the data exchange
after every completed run is executed automatically. A third option of interaction
is the inclusion of the other method respectively by a software-based coordination mechanism, for example using a communication or control platform to
parameterize, control, and synchronize bidirectional interactions simulation and
optimization. The fourth implementation level is the direct communication of
simulation and optimization tools as well as the mutual coordination through builtin interfaces. [VD2016a, p. 4]. “Researchers have proposed and developed many
different methods that attempt to optimize a simulation by searching through the
space of possible input-factor combinations […] with the results from simulating
earlier configurations being used to suggest promising new directions to […] better system performance” [La2007, p. 658]. The commercial and freely available
software systems already provide optimization components that can efficiently
support this search process. However, the concrete implementation of the interfaces between simulation and optimization as well as the application of tools and
the interpretation of the results remain mainly the tasks of the user [VD2016a,
p. 5].
2.3.2 Objectives and Challenges of Simulation-based
Optimization
The simulation-based optimization has the objective to improve a process by
checking whether a change in values of variables has an influence on the process
and which process inputs are most influential on the process outputs of interest.
The basic idea behind this approach is to find an optimal solution by using multiple replications simulating different system configurations [Fu2013, p. 1418].
The simulation is started by the optimization, produces the result data, and forms