DOE testing - query on settings to reach a specified target

[#1]

Hi all.

I am currently working on a DOE assignment. One part of the question relates to getting the settings for 4 factors to achieve a target distance within a range of x +/-5cm. I believe I should be able to achieve this through the profiler but I cannot figure out how to achieve this. Can you offer me any advice in relation to this? It would be greatly appreciated.

Thanking you in advance.

[#2]

In the profiler you can turn on the 'desirability functions', then you edit the Desirability graphs on the right to reflect the goals you are after and then you come back to the menu and do 'maximize desirability'

[#3]

Paul, thanks for your assistance with this query. Can I ask if you believe that the below gives me the factor settings, Pin evaluation 155.79965 Cup elevation 250.00 Bungee 170.00 Release Angle 157.2634, which would achieve the target distance of 230cm +/- 5cm?

[#4]

This recipe gives you a mean performance of 230 with a 95% CI of 225.36 - 234.36. If you want to know the distribution of indivudial trials, turn on the simulator under the hotspot, input some variability for each of the factors. Then turn on the spec limits under the siimulator hotspot and input LSL 225 USL 235 and hit simulate to check the distribution of distances.

[#5]

Paul,
My apologies for this again. Following your instructions I end up with the following screen - and when I "Simulate to Table" I get 827 variations from the 5000 runs that would give me an answer between LSL225 and USL235 - which in my mind makes sense. The question that was set seems to indicate that there should be only a small number of answers for the settings. I am just wondering am I missing one final step in the puzzle.

[#6 · Accepted solution]

No you got that right.

The mean (230.6) is prefectly on target so that is working.

The SD (24.7) is quite high.

The rate under the simulate button says that 83% of your runs are out of spec, this is because your SD on the simulated runs is so big. If you decrease the SD of your input variations you will see that rate go down. So you have a whole bunch of problems but lets talk about release angel as it is the biggest:

Release angel is a sensitive factor (the slope is steep in the range you have experimented). And your variability on release angel is big in that window as well, this means that you regularly get values on the low end of your range which results in low distance. So either you make your process less sensitive to release angel (design a tool/fixture or something more advanced like a self guided missile) or you have to make the variability of your release angel smaller (training, tool). If you can do this for all SD's on your inputs you should be able to get the output correct.

Welcome to you first capability analysis

[#7]

Paul thanks so much for your assistance with this. It is greatly appreciated.