Not logged inAmerican Welding Society Forum
Forum AWS Website Help Search Login
Up Topic American Welding Society Services / Certifications / Significant Digits & Accuracy of PQR Data
- - By tom cooper (**) Date 06-11-2008 12:42
This is a simple but serious question. Approval of PQR data in my organization is an engineering function and the reviewers insist on the consistency of significant digits between the actual data recorded and the threhold data required. In other words, if tensile tests must meet the minimum UTS requirements of ASTM A36 and ASTM A36 cites a minimum UTS of 58ksi, then actual tensile test results must be reported as, say, 67 ksi and not 67.3 ksi and not 66.8ksi.  Rounding to two significant digits is required by engineering standard. I would like to know if AWS requirements discuss the accuracy or significant digit requirement of data. I cannot find any mention of this.

Why am I taking up valuable Forum space for such a trivial question? Because I have tensile data back from the lab for 6061-T6 specimens which were reported as: 23.7, 24.6, 25.9 & 26.2 (ksi).

AWS D1.2 Table 3.2 Material group 23 requires Minimum Tensile Strength of 6061-T6 to meet "24ksi".   Notice the AWS does not require the speciman to meet a minimum tensile strength of 24.0 ksi; therfore, by our engineering standard I should report my results as: 24, 25, 26 & 26 (ksi).   Doing so would push a possibly borderline tensile result over into acceptable category.  Is this legitimate from an AWS Code point of view? The help I am asking for here is to get confirmation of the above practice in rounding to 2 significant digits OR find out where in AWS does it direct how data is to be recorded. 

Thanks for your insight.
  
Parent - - By js55 (*****) Date 06-11-2008 13:39
Codes couldn't care less about your companies engineering standards and a desire to round up. If you don't meet the min you don't have a viable qualification.
Parent - - By js55 (*****) Date 06-11-2008 13:44
Tom,
Let me rephrase that. I understand you are not the one dictating the rounding procedure. Codes are only concerned with mins. If rounding doesn't effect the min reuqirement your good and the code essentially doesn't care. If rounding up is generated from under the requirement its unnacceptable.
Parent - By tom cooper (**) Date 06-11-2008 16:04
js55-
Nicely stated. I will forward that as a recommended rule for this set of data and all future data.

Joe Kane-
Thanks for the ASTM A370 suggestion. That has a reference to ASTM E29 which is also useful for exactly my question.  Thanks all!
Parent - By Joseph P. Kane (****) Date 06-11-2008 14:02
There is a "Rounding" section in ASTM A370 that should be applicable.
Parent - - By Kix (****) Date 06-12-2008 13:04
I have never heard of the whole rounding thing and am interested to know if you can because I could have done it myself a few times.:-)  I go by if the tensile pulls at 24,000psi it passes and if it pulls at 23,999, it fails.  Well, that's how I interpret the code anyway.  The facility that does you testing will have this on record and there name is on your PQR.  So if anyhting were to ever happen, someone would see the testing facilities name on your PQR and double check you.  23.7ksi fails by how I read D1.2.  When you record your data on your PQR, why don't you put your actual results in PSI?  Hope this helps a little bit.
Parent - - By GRoberts (***) Date 06-12-2008 19:36
As an example of rounding, you say that if a tensile fails at 23,999 psi, that it does not meet the criteria.  However, the lab reports the cross section of the tensile typically to 3 significant digits.  Such as .749" wide x .488" thick.  So with that data, the maximum amount of significant digits you can have is 3.  The cross sectional area is .365512, which rounds off to .366 sq in.  The lab might also report the force requried to break the specimen as 8770 (8771.9 is required to get to 23,999lb) since thier machine is only accurate to the nearest 10 lb.  So if you use .366 area, and 8770lb, your tensile comes in at 23,961 lb.  Does that pass or fail?  If the lab happened to report to 4 significant digits, they could have reported .7485" wide x .4875" thick and 8772 lb.  In this case, the tensile comes in at 24040lb.  Does this pass or fail?  It was the same test, just different significant digits!! 

You shouldn't be at the mercy of how the lab reports the data either.  Your lab should follow standard industry practices based on how thier machines are calibrated.   Your tensile data can only be as accurate as the least accurate measurement in the system.  Standard rounding practices should be used since you don't actually know if the test was 23,999lb or not since measurements were not that accurate. 
Parent - By js55 (*****) Date 06-12-2008 19:49
Wow Greg. Thats really good.
It just never occured to me to look that deep at it.
I suppose its because my quals always passed with greater margins so it was never an issue. Er, uh, or failed too miserably.   :)
Actually, in light of this most enlightening of threads I am certianly going to take look at the rounding practices of my lab.
Parent - - By Kix (****) Date 06-12-2008 20:32
Very good info!!! So would you fail 23.7ksi?
Parent - - By GRoberts (***) Date 06-12-2008 23:16
Without a copy of the ASTM (E29) rounding specification handy, and the lab information, it is hard to tell.  More than likely, you could report 3 significant digits accurately.  I don't remember if E29 had anything to say about the amount of significant digits in the requirement.  I.E. 24 ksi vs 24.0 ksi.  I haven't had the opportunity to be part of an AWS committee yet, but I have a feeling that they use 2 significant digits on purpose.  Can anyone shed any light on that?
Parent - By Kix (****) Date 06-13-2008 12:01
Yeah, in the code it only says 24ksi so I wonder also about the 2 sig nifs.  Interesting topic!
Up Topic American Welding Society Services / Certifications / Significant Digits & Accuracy of PQR Data

Powered by mwForum 2.29.2 © 1999-2013 Markus Wichitill