Regardless of whether it's a pressure containing pipe or a structural pipe, I don't see how you would justify using anything other than the minimum measured thickness, minus some factor of ignorance to account for the probability of having not searched for, found and measured the actual minimum that exists anywhere on the pipe.

Mike Halloran
Pembroke Pines, FL, USA

I agree - You don't say how long the "pipe" is, but if we assume 12m, this is a very low density of readings to do anything else with.

As usual it depends on the details - if your variances are very small you could use the mean, but the number of your data points is too low for any serious analysis.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

Thanks MikeHalloran and LittleInch for your quick input.
This is a structural member 18"OD and 60ft long that is part of a 4-legged jacket platform. I should have said that my intent just was to model the member (not designing it) so that the variation of thickness will be represented in the stiffness of an equivalent element with uniform thickness. I think that using the minimum value will be overconservative to represent the stiffness of the member.
I agree with you that the minimum thickness should be used to check the structural capacity of the element.

Insufficient data to answer the question.

I hate Windowz 8!!!!

Agree with BigInch. What are you trying to check?
1. The radial external pressure capability?
2. The structural capability?
3. What?

If you use the minimum thickness then fine for (1) but will be totally wrong for anything else as you will be under-estimating the stiffness of the system.

Actually, using T_min, the minimum As-Found thickness is pretty standard. There is absolutely no assurance that there is not a lower thickness near one of the thickness measurement locations. If you need to get dataset good enough to do 'close' calculations, vs. good but conservative approximations, the testing has to be redone.

You need an ultrasound "B-Scan", a continious line of data [actually several hundred measurements per inch] electronically stored. At 18" diameter, I would be comfortable with 4ea. 'lines' along the pipe. Now you can determine the 'real' internal profile of the pipe, and make some stiffness calculations based on nominal as-corroded thickness [rigidity] in various areas of interest. Don't forget to project the Corrosion Rate out at least 5-years [ask client]. Nobody really wants to know about their structure 'today'. They want to know how far into the future that it will be acceptable, and when repairs may be required.