## SensorPx Example 3
Fewer runs are required for optimizations when the
operational alternatives have larger effects on results. We repeat the
workflow used in Example 2 to determine the probability that option c is
better than option a (which is shown in Example 1 to be the worst possible
choice). Use the same Example2 work directory
as in Example 2. Execute the following 5 commands (or perform the
equivalent in Windows Explorer, given in blue and in parenthesis following
each command) to run the same 1000 realizations of the unknowns for
operational option a: 1.
cd Example2\example2_1k (in
Windows Explorer (WE), navigate to the Example 2 example2_1k folder)
2.
ospe1a.mspx
(WE
- double click on ospe1a.mspx)
3. runs.bat
(WE
- double click on runs.bat)
4. move results results_spe1a
(WE
- right click on the results folder, and rename to
results_spe1a) 5.
move sensor.stat results_spe1a\a_sensor.stat
(WE
- copy and paste or drag sensor.stat to the results_spe1a folder, and rename
it to a_sensor.stat) The results file we
obtained is a_sensor.stat.
Comparison of cumulative oil production for the first 100 runs with that
reported in c_sensor.stat
shows that option c significantly outperforms option a in all of the first
100 cases. **We estimate that the probability that option c will
outperform option a is 100%, based on those first 100 cases.**. That
result can be correctly obtained with a minimum of 1 case, and confidence in
that 100% estimate may be sufficiently assured by the first 50 or 100
realizations that give it. |