2.3 Combination of Simulation and Optimization Methods
31
sequential coupling case a). In the case of upstream optimization (Figure 2.9,
sequential coupling case b), the simulation serves to check the feasibility of the
proposed solution. Often, it is not possible to formulate the causal conditions of
a complex system completely analytical in form of conditions and restrictions.
To analyze and optimize this system, a simple optimization model is created,
which allows the calculation of an optimum, but which does not cover all facets
of reality [MK2011, p. 44; VD2016a, p. 3]. With the help of the subsequent
simulation, which depicts causal relationships by depicting relevant behavioral
rules, the optimization solution can be tested for feasibility.
Optimization
Simulation
Optimization
Simulation
Optimization
Simulation
Simulation
Optimization
Sequential Coupling
Hierarchical Coupling
a)
b)
c)
d)
Figure 2.9 Different cases of sequential and hierarchical coupling of simulation and
optimization [VD2016a, pp. 3–4]
While for the sequential coupling the results of one method are available before
the following one, there is a dominant method in the hierarchical architecture that
controls the other method. The dominant method calls the other method as a
subcomponent during its execution (Figure 2.9, scenarios c and d). In a simulation experiment, decision points may exist, at which a solution must be selected
from alternatives to determine the further course of the experiment. Optimization
methods are then used to make this selection of an alternative. The simulation
model transfers the current state as an input parameter to the optimization model,
which returns a value that serves as a decision parameter for the simulation model
to determine the further progress of the simulation run [VD2016a, p. 4]. If the
optimization is the dominant method, the simulation is started by the optimization.
The simulation then forms the basis for an assessment of the dynamic behavior
of the depicted system. The simulation is used to calculate the objective function value, the optimization procedure represents the alternative search [MK2011,
31
sequential coupling case a). In the case of upstream optimization (Figure 2.9,
sequential coupling case b), the simulation serves to check the feasibility of the
proposed solution. Often, it is not possible to formulate the causal conditions of
a complex system completely analytical in form of conditions and restrictions.
To analyze and optimize this system, a simple optimization model is created,
which allows the calculation of an optimum, but which does not cover all facets
of reality [MK2011, p. 44; VD2016a, p. 3]. With the help of the subsequent
simulation, which depicts causal relationships by depicting relevant behavioral
rules, the optimization solution can be tested for feasibility.
Optimization
Simulation
Optimization
Simulation
Optimization
Simulation
Simulation
Optimization
Sequential Coupling
Hierarchical Coupling
a)
b)
c)
d)
Figure 2.9 Different cases of sequential and hierarchical coupling of simulation and
optimization [VD2016a, pp. 3–4]
While for the sequential coupling the results of one method are available before
the following one, there is a dominant method in the hierarchical architecture that
controls the other method. The dominant method calls the other method as a
subcomponent during its execution (Figure 2.9, scenarios c and d). In a simulation experiment, decision points may exist, at which a solution must be selected
from alternatives to determine the further course of the experiment. Optimization
methods are then used to make this selection of an alternative. The simulation
model transfers the current state as an input parameter to the optimization model,
which returns a value that serves as a decision parameter for the simulation model
to determine the further progress of the simulation run [VD2016a, p. 4]. If the
optimization is the dominant method, the simulation is started by the optimization.
The simulation then forms the basis for an assessment of the dynamic behavior
of the depicted system. The simulation is used to calculate the objective function value, the optimization procedure represents the alternative search [MK2011,