8.6 Preliminary Experimental Analysis
Before proceeding with a full scale investigation, it will be useful to do two things. First, an assessment for the number of required sample holders is needed. This will provide a range of values to consider during the design alternative search. The second issue is to examine the lower bounds on the required number testers at each test cell. The desire is to prevent a woefully under capacitated system from being examined (either in the Process Analyzer or in an OptQuest run). The execution of a woefully under capacitated system may cause an execution error that may require the experiments to have to be re-done. The analysis files associated with this and subsequent sections are available in the chapter files.
Based on Table 8.13, in order to keep the testers busy in the high resource case, the system would need to have at least (2 + 2 + 3 + 4 + 5 = 16) sample holders, one sample holder to keep each unit of each test cell busy. To investigate the effect of sample holders on the system, a set of experiments was run using the Process Analyzer on the high resource case with the number of sample holders steadily decreasing down to 16 sample holders. The results from this initial analysis are shown in Table 8.15. As can be seen in the table, the performance of the system degrades significantly if the number of sample holders goes below 16. In addition, it should be clear that for the high resource settings the system is easily capable of meeting the design requirements. Also, it is clear that too many sample holders can be detrimental to the system time. Thus, a good range for the number of sample holders appears to be from 20 to 36.
The low resource case was also explored for the cases of 16 and 48 sample holders. The system appears to be highly under capacitated at the low resource setting. Since the models were able to execute for the entire run length of 10 days, this provides evidence that any scenarios that have more resources will also be able to execute without any problems.
| Cell 1 | Cell 2 | Cell 3 | Cell 4 | Cell 5 | #Holders | Non-Rush | Rush | (minutes) |
|---|---|---|---|---|---|---|---|---|
| 2 | 2 | 3 | 4 | 5 | 48 | 0.998 | 0.960 | 17.811 |
| 2 | 2 | 3 | 4 | 5 | 44 | 0.997 | 0.971 | 17.182 |
| 2 | 2 | 3 | 4 | 5 | 40 | 0.994 | 0.976 | 17.046 |
| 2 | 2 | 3 | 4 | 5 | 36 | 0.994 | 0.983 | 16.588 |
| 2 | 2 | 3 | 4 | 5 | 32 | 0.996 | 0.982 | 16.577 |
| 2 | 2 | 3 | 4 | 5 | 28 | 0.995 | 0.986 | 16.345 |
| 2 | 2 | 3 | 4 | 5 | 24 | 0.988 | 0.990 | 16.355 |
| 2 | 2 | 3 | 4 | 5 | 20 | 0.961 | 0.991 | 21.431 |
| 2 | 2 | 3 | 4 | 5 | 18 | 0.822 | 0.987 | 32.839 |
| 2 | 2 | 3 | 4 | 5 | 17 | 0.721 | 0.990 | 39.989 |
| 2 | 2 | 3 | 4 | 5 | 16 | 0.494 | 0.989 | 57.565 |
| 2 | 2 | 3 | 4 | 5 | 12 | 0.000 | 0.988 | 1309.683 |
| 1 | 1 | 2 | 2 | 3 | 16 | 0.000 | 0.985 | 2525.434 |
| 1 | 1 | 2 | 2 | 3 | 48 | 0.000 | 0.749 | 2234.676 |
Based on the preliminary results, you should feel confident enough to design experiments and proceed with an analysis of the problem in order to make a recommendation. These experiments are described in the next section.