How do I set alert/action levels for areas not subject to normal Gaussian distributions?

I’m having troubles setting alert/action(especially the alert!) levels for Environmental monitoring program, thus I’ve gone through the Technical Report No.13 of PDA for assistance. They’re recommending a Normal Distribution Approach and to set mean plus two or three times the standard deviation for a alert level.

Then here comes my question. For environmental monitoring data that is not normally distributed, e.g Microbial count(I’ve read a previous thread similar to this and saw a user named Boomer_Chemist saying that Microbial counts are not subject to normal Gaussian distribution), then how am I supposed to set the alert/action level? In here in the report says that I should use Non-parametric Tolerance Limits Approach, but sincerely I have no idea of this and this does not articulate exact formula to calculate the exact number for alert/action levels.

In addition, i’m now putting environmental monitoring data(TOC, conductivity, airbourne particle count, microbial count) into the minitab and run test for normality, to tell that which data is normally distributed or not. Is there any guidance that tells which one is usually normally distributed and which is not?

My plant now applies mean+two times SD for every monitoring areas including Microbial limit for water and air, and it sets the alert level to be way low, causing problems. Please help.

Determine the distribution for the data, use the standard deviation formula for that distribution.

Source: LinkedIn Group