ASME B31.3 Progressive Sampling 1

[jm2804]

My understanding that progressive sampling in B31.3 work the following way:
1 Reject - 2 tracers
1 reject out of those tracer - 2 more tracers
1 reject out of those tracer - 100% NDE

what if in the first batch of sampling a welder has 4 rejects? The code does not exactly explain what to do in this situation.

 

[stanweld]

Fire the welder.

 

[metengr]

The Code does not and will not deal with welder productivity issues, only requirements for welders. Productivity is left up to the owner/user/contractor. I second the recommendation by stanweld.

 

[vpl]

My understanding would be that you would double the number rejected and do eight additional samples the second time around.

Or you could go directly to 100%, depending on the percent of rejects to passes in the batch.

Patricia Lougheed

******

Please see FAQ731-376: Eng-Tips.com Forum Policies for tips on how to make the best use of the Eng-Tips Forums.

 

[stanweld]

Per 341.3.4 (e) "if any of the items examined as required by (c) above reveals a defect, all items represented by the progressive sampling shall be either...". You must first define the items in the progressive sampling. If, for example, the first 4 welds "of the same kind" made by a welder are defective, terminate the welder and repair or replace all 4 of the defective welds. There is no requirement to test the welds made by other welders.

 

[REGRUMBLE]

You have not indicated how may the total batch is. If the welder only did 10 welds and four were rejects i would certainly insist that all of them are inspected. If his rate was 4 out of the first thousand then the person who waited until so many welds were complete should also be fired.

 

[Duwe6]

REGRUMBLE hit it on the head. Roll some heads.

As for B31.3 sample plan math: The 'batch' is the your production to date on the "Project". Any single RT reject requires a resampling of 2 additional 'tracer' RT'd welds. If EITHER of those fails, you must re-resample an additional 2 welds. If either of those second pair of 'tracers' fails, that welder's production work requires 100%.

Thus the intent of this Code is that if three samplings fail - 1st the random RT, 2nd either of the 2 tracers, 3rd either of the 2nd set of tracers - 100% RT or 100% removal of these welds [it's usually cheaper to just cut out the rest of this welders worek. RT's are expensive] So it only takes 3 rejected random RT's to engender the 100% 'death penalty'. You have 4. No further analysis required. Time to start cutting out some welds, or send a big $$$ P.O. to an RT firm.