6.3 Optimization Experiments
185
replaced by machine condition-related mean values in an extra model (mean value
scenario) for the 1-min resolution datasets.
The reference scenarios with the different data resolutions of one data record
per minute and one data record per second were tested, both with the same production task (13.728 pieces/week) using the same time frame for the simulation
experiment (May 20 th starting at 06:00 am until May 24 th finishing at 11:59 pm).
Comparing both simulation runs, the deviation of the total energy consumption
was only 0,2%, even though the energy profiles of the different machine states in
the two data resolutions differed between 0,1 and 2,8% depending on the machine
looked at. The comparison of the total energy consumption of the optimization
scenarios with the 1-min and the 1-sec resolution showed a deviation of 0,7%. At
this point it becomes clear that the deviations in the energy consumption profiles
of the machine states caused by the resolution affect the results. If one compares
the ideal parameter configurations determined for the idle and standby optimizer
parameters in the two optimization scenarios, one also notices deviations here
(Table 6.8).
Table 6.8 Comparison of the idle and standby optimizer parameter configurations in the
optimization scenarios at different data resolutions
Optimization Scenario
(1-second resolution)
Optimization Scenario
(1-minute resolution)
IdleOptimizer_Line1
1.872 sec (31,2 min)
44,8 min
IdleOptimizer_Line2
1.874 sec (31,2 min)
19 min
StandbyOptimizer_Line1
1.140 sec (19 min)
19 min
StandbyOptimizer_Line2
1.140 sec (19 min)
19 min
IdleOptimizer_L1wash
3.600 sec (120 min)
120 min
IdleOptimizer_L2wash
3.600 sec (120 min)
120 min
The deviation of the single machine state energy consumption profiles in the
two data resolutions resulted in different optimal parameter configurations to
reach the minimal total energy consumption value. The data-resolution-dependent
optimal parameter configurations led to the fact that the single machine states
have different shares of the total simulation time varying in a range of 0,1% to
3,3% (Figure 6.24).
The data resolution comparison shows that impurities and deviations in the
database affect the entire simulation and its different experiment runs and have a
significant impact on the results. The use of different measuring instruments to
185
replaced by machine condition-related mean values in an extra model (mean value
scenario) for the 1-min resolution datasets.
The reference scenarios with the different data resolutions of one data record
per minute and one data record per second were tested, both with the same production task (13.728 pieces/week) using the same time frame for the simulation
experiment (May 20 th starting at 06:00 am until May 24 th finishing at 11:59 pm).
Comparing both simulation runs, the deviation of the total energy consumption
was only 0,2%, even though the energy profiles of the different machine states in
the two data resolutions differed between 0,1 and 2,8% depending on the machine
looked at. The comparison of the total energy consumption of the optimization
scenarios with the 1-min and the 1-sec resolution showed a deviation of 0,7%. At
this point it becomes clear that the deviations in the energy consumption profiles
of the machine states caused by the resolution affect the results. If one compares
the ideal parameter configurations determined for the idle and standby optimizer
parameters in the two optimization scenarios, one also notices deviations here
(Table 6.8).
Table 6.8 Comparison of the idle and standby optimizer parameter configurations in the
optimization scenarios at different data resolutions
Optimization Scenario
(1-second resolution)
Optimization Scenario
(1-minute resolution)
IdleOptimizer_Line1
1.872 sec (31,2 min)
44,8 min
IdleOptimizer_Line2
1.874 sec (31,2 min)
19 min
StandbyOptimizer_Line1
1.140 sec (19 min)
19 min
StandbyOptimizer_Line2
1.140 sec (19 min)
19 min
IdleOptimizer_L1wash
3.600 sec (120 min)
120 min
IdleOptimizer_L2wash
3.600 sec (120 min)
120 min
The deviation of the single machine state energy consumption profiles in the
two data resolutions resulted in different optimal parameter configurations to
reach the minimal total energy consumption value. The data-resolution-dependent
optimal parameter configurations led to the fact that the single machine states
have different shares of the total simulation time varying in a range of 0,1% to
3,3% (Figure 6.24).
The data resolution comparison shows that impurities and deviations in the
database affect the entire simulation and its different experiment runs and have a
significant impact on the results. The use of different measuring instruments to