Calculating Error

I have a general question about calculating error.

Though it is an overestimate method that one of the technical professors mentioned where you can develop a range for error by simply using the most extreme values for equation inputs.

For example, if a = 1 +/- 0.1 and b = 1 +/- 0.1, then a x b would be 1, and would range from .9 x .9 to 1.1 x 1.1

This method is far less rigorous and could yield large overestimates, but it would be conservative (potentially a good thing) and would massively simply calculations. If we did the error calculations the proper way for a, w, PQ, B, Young's modulus, and yield strength for 3 different sections worth of data, it seems like we would be working on this report until December.

The problem is that doing the uncertainties that way will greatly exaggerate the uncertainties in Kq and your plot with error bars will look pretty bad.

The good news is that you don't really need to worry about the error from some of those quantities, because their uncertainties are so small and can probably be neglected (i.e. w and S). The dominant sources of uncertainty when finding Kq arise from the calculation of Pq and measurement of the crack length, a.

You definitely want to automate the uncertainty calculation process as much as you can by setting up nice formulae in your spreadsheet, Matlab script, etc. so you don't spend so much time on the uncertainty math. Really, once you have the post-processing for your section data set up, you should be able to chug the other two sections pretty quickly.