Your question seems to be apples and oranges. How is incorporating non-data going to improve accuracy? Is the actual phenomenon that random? Rather than spending time trying to account for this W not being there at all, shouldn't you be spending time re-vamping the experiment to be more reliable?

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! https://www.youtube.com/watch?v=BKorP55Aqvg
FAQ731-376: Eng-Tips.com Forum Policies forum1529: Translation Assistance for Engineers Entire Forum list http://www.eng-tips.com/forumlist.cfm

You don't know in advance whether W occurs? If W "does not occur" does variable W = 0, or some other random value, and if "W does occur", does variable W have a finite value or a range?

In the more complex case of running cars over roads, where there is a finite probability that we will hit a particular bump at a particular speed, we get round that sort of problem by running a monte carlo analysis where the variables are randomly set for a whole bunch of runs. We then linearise the resulting responses. For instance we might be worried about the effect of jounce bumper engagement travel on the fatigue life of the spring tower, the output would be some sort of load factor vs travel. We don't know (or even care about) the exact combination of bumps that would maximise the loads, for a given travel.

Is that what you were thinking of?

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?