How Do You Calculate Reject Rate?

Date 03-09-2011 11:17

There isn't two trains of thought in piping. There is straight percentage and there are cheaters, the intent always being to find a way to reduce the reported reject rate.
I can the possibility of an inch calc in some applications but not pipe.

By 46.00

Date 03-16-2011 15:05

Hi!

What about large diameter pipework, I'm thinking 36" and over, would it be fair to use "straight percentage" or would an "inch calc" give a more realistic view of the repair rate?

Interesting.

By js55

Date 03-17-2011 11:37

When you say 36" you are essentially establishing an arbitrary limit. There is no justifying logic. But if you wish to use such criteria keep in mind it is not ASME code compliant. But if Code compliance is not an issue then its between you and your customer.

By Jim Hughes (***)

Date 03-23-2011 12:17

"But if you wish to use such criteria keep in mind it is not ASME code compliant."

This is what ASME B31.1 says in 136.4.5 (A) Acceptance Standard. "Welds that are shown by radiography to have any of the following types of discontinuities are unacceptable" Inferring that if the weld is 1" in diameter or 60" in diameter the weld is unacceptable.

With that being said, ASME primary concern is safety. They are giving us a criteria and guidance for making sure that a weld that is found to be unacceptable is removed or repaired according to the code. They are not giving any direction on how you track the rejects. We have had clients on pipeline projects (30" to 42" pipe) that wanted us to track by inch and discontinuity not by weld, rejects.

Jim

By welderbrent (*****)

Date 03-16-2011 16:21

To my way of thinking this depends first of all on the Code being worked to and secondly on the contract from the customer or agency in control and thirdly who you are working for (fabricator or customer).

Reason, there are differences between 'repairs' and 'rejections'. There are different specs depending upon the Code. There are different responsibilities between TPI and in house QC.

In house QC may be responsible to management to document all discontinuities regardless of rather it is a simple repair by grinding or welding or rather it is indeed 'rejectable' to the point of scrapping and starting over. They want to know where their weaknesses are and who may need additional training.

TPI on the other hand is making sure QC is being properly handled by the fabricator and that the product being shipped is ACCEPTABLE. How much of the job was originally 'rejected' is not neccessarily an issue for the TPI unless it starts effecting the quality of the entire project.

One Code may call X amount of undercut rejectable while another Code calls out a different amount. Same with porosity, undersize welds, etc.

So the bottom line is that it will depend totally upon who I represent and their expectations for the job at hand.

Have a Great Day, Brent

By 46.00

Date 03-16-2011 20:13

Hi!

From my point of view, repair rates have always been a KPI on every job I have been on. Brent is correct in that rejection (not repair) rates can be contract specific and also, as stated different codes = different acceptance criteria.

As a TPI, I would be interested in how much of a job was originally 'rejected'! This would be an indicator of potential missed problems and 'trends'. (New catch word is 'Trends').

To get back to the original question, what if two, or more welders (half each), worked on a large diameter pipe weld joint which later was shown to have a unacceptable defect, discovered by radiography at around either the 12 oclock or the 6 oclock position, would it be fair to allocate both welders (or all the welders) with a repair?

Regards

By js55

Date 03-17-2011 11:38

My experience has been that it is not that difficult to determine who is responsible.

By Shane Feder (****)

Date 03-17-2011 21:35

Jeff,
In certain situations it can be nigh on impossible to differentiate.
I remember years ago welding a 54" pipe in tandem with another welder using the "stovepipe" method.
When stoving around the bottom of the pipe one welder always continues to about 7.30/8 o'clock on the pipe, never stopping at 6 o'clock.
The other welder then comes down his side and overlaps and continues across to approx 7.30 on the other welders side.
This is reproduced over and over dependent on wall thickness.
We had a repair called at approx 6.30 on my side for an unacceptable slag inclusion but how do you decide which welder is responsible ?
In the end we were given 1/2 a repair each.
Regards,
Shane

By js55

Date 03-18-2011 13:01

Shane,
You're right. It happens. I've had it too, but rarely. You find a wya to work it out.

By OBEWAN

Date 03-16-2011 21:00

In six sigma, the reject rate is defined by "quality opportunities". 100" inches of weld in one pipe job would offer only one opportunity to leak. 100" inches of weld in two seperate pipes would offer two opportunities to leak. It seems to me that that line of reasoning should play into it. So you could revise your calculation to include a new length calculation for every new leak opportunity or every new pipe job.

By 46.00

Date 03-16-2011 21:23

Exactly what I'm after!

By js55

Date 03-17-2011 12:06 Edited 03-17-2011 12:09

Keep in mind ASME couldn't care less about six sigma and six sigma people don't know shyt about piping.
But, if we wanna open the 'opportunities for leak' criteria what about significant porosity in the cap as opposed to minimal insufficient fusion at the root.
Clearly one would have greater opportunity for leak than the other and not necessarily related to its extent. Which would count more?
Or how about 6" of IF in a carbon steel air line at room temp and 110psi, as opposed to 1" of IF in service with a pH of 4 or 12 at 500degF and 250psi? There is clearly some variance in 'opportunities for leak' here. Though I doubt even six sigma wants to address this 'can of worms'.
I think the bottom line for me is that efforts to represent with greater accuracy what is going on are too often transparent attempts to reduce reject rates.
And one other problem. Who's six sigma program would we be talking about? How does a fabricator assess 'opportunities for leak' when in general he is totally ignorant of the design criteria(knows nothing of specific stresses or corrosion). The only answer could be that the 'opportunity for leak' criteria would be entirely arbritrary. And if the point is to represent with greater accuracy what is happening I do not see how reinserting more arbitrariness in the back door does this.
For me its keep it simple. You bust a weld, you bust a weld. Deal with it. Fix it. Its just a number. Quit trying to pretty it up.

By welderbrent (*****)

Date 03-17-2011 16:13

Good point and application.

Have a Great Day, Brent

By Shane Feder (****)

Date 03-18-2011 00:21

Guys,
It is an interesting topic but IMHO repair / reject rates are pure B/S.
They are only statistics and as js55 has stated they can be easily manipulated.
A client / owner / customer / TPI may be interested in seeing the reports but how true are they ?.
If you are the QC manager of a workshop you should hope that any problems are resolved by whoever is responsible for the welders (in Australia / NZ it is generally a qualified Welding Supervisor) well before they become damning statistics.
If a problem is noted then the problem is assessed, the problem is discussed with the welder and advice may be given to the welder or he may be removed from production welding and retrained.
I once had a very good pipewelder who turned into a "trainwreck" in one week. His repair rate went through the roof.
I took him for a beer after work and discovered his wife had just left him and he was trying to care for his two children as well as work full-time.
A bit of compassionate leave and a bit of assistance from the company and we had ourselves a very good welder back again.
If it had been up to the bean-counters or those who state things like - The employee shall be terminated if his repair rate is over a certain percentage would have had a good man looking for another job.

To get back to my statement about repair / reject rates being B/S.

Straight Percentage
Welder 1 welds 10 x 2" pipe welds (74" of weld) and has 1 x repair
Welder 2 welds 10 x 16" (495" of weld) pipe welds and has 1 x repair
They both have a 10% repair rate

Welder 1 welds 1 x 12" pipe weld and has 1 x repair (1 x 6" LOP / LOF)
Welder 2 welds 1 x 12" pipe weld and has 1 x repair (1 x Gas Pore)
They both have a 10% repair rate

Linear Percentage
Welder 1 welds 100" of weld on thin wall pipe or thin plate
Welder 2 welds 100" of weld on heavy wall pipe or thick plate
They both have a 2" slag inclusion in their welds.
As acceptance criterias are generally based on parent metal thicknesses Welder 1's weld fails while Welder 2's passes.
Exactly the same defect but one welder has the black mark while the other doesn't.

There are too many variables to ever be able to accurately gauge someones performance based on NDT - especially if it is random NDT.
The best I have heard of with RT is number of films rejected divided by the total number of films.
If that was the case in my first scenario
Welder 1 welds 10 x 2" pipe welds (74" of weld) and has 1 x repair (2 films per butt x 10 = 20 films / 20 films divided by 1 reject = 5% repair rate
Welder 2 welds 10 x 16" (495" of weld) pipe welds and has 1 x repair (4 films per butt x 10 = 40 films / 40 films divided by one reject = 2.5% repair rate
Much fairer way of recording but it is still open to debate (small films used on 2" versus large films used on 16")

Regards,
Shane

By js55

Date 03-18-2011 13:07

Its two problems. Internally you can do it any way you want. And no method is particularly better.
However, when it comes to representing your comapny to customers the rules change. If you are using a non industry standard method that improves your numbers you are essentially lying and you know it. You will represent your company as having a better reject rate when in reality you don't.

By OBEWAN

Date 03-18-2011 14:09

The only reason I bring up Six Sigma is that it forces the person making the calculation to clearly define what they call CTQ's or criteria that are critical to quality. Of course the defect criteria must be clearly defined to have a useful calculation without cheating or generating useless fluff. More importantly though the "quality opportunity" must be clearly defined and understood by the customer. If it is reducing repairs that is one thing and if it is preventing leaks or explosions that is another matter. I vote for the latter.

If it were me, I would compromise and create some kind of weighting factor for each new pipe. Each pipe could be a new opportunity but it could have a ratio or weighting factor based upon length of weld per pipe. Defects in each pipe would be divided by the weighting factor and then averaged. The weighting factor would be based upon an average length of weld. Say, 100". A 100" inch pipe would have a weighting factor of 1.0, and 80" pipe would have a weighting factor of .80, a 150" pipe would have a weighting factor of 1.5"...

The calculation for two pipe welds of 100" and 150" might go like this: reject rate = average(repairs pipe 1/1.0, repairs pipe 2/1.5, yada yada). It might be hard on weldors but would be honest for the customer.

By welderbrent (*****)

Date 03-18-2011 15:07

Having been on more than one side of this since I have been a production welder, a repair welder, a shop owner, and now an inspector I hope I see the positive parts of the positions and comments that all three of you are making. And there definitely needs to be a compassion factor for employers dealing with personnel.

It will always be a battle between the bean counters, QC, and management trying to come up with the BEST system to ACCURATELY monitor production, reward good work, correct bad work, offer incentive to promote desire to produce more good work without being overly punitive for bad work so that workers will not see additional training as belittling but as a positive direction for their future as well as the company, while all the while doing the absolute best possible job to promote the company's quality standards to impress future and repeat customers.

I don't think we will ever find a PERFECT reporting system. And I agree that overall it will probably not be an equally applied system. But companies do need some method of figuring time spent on repairs vs production and then see how further training can reduce repairs and increase production which will obviously increase profits.

The main point of application will always be different between customer, management, workers, and QC. Each is looking for something different. Each must be able to keep their own personal responsibilities separated from the area of responsibility of others. It is not QC's job to fire, only to state problem areas so the customer gets a high quality job. Then management must decide how to deal with the problem areas. Hopefully management will not seeing firing as the only means to correct a problem. But that is their area of responsibility.

I appreciate Shane's examples of comparison between percentages of rejections. Someone who has been on more than one side of the issue can see the various aspects of the problem. Those who have always been only in management, only in accounting, only in QC, and even those who have always been ONLY (??) in production have tunnel vision when it comes to truly dealing with the problem to correct it.

A very good discussion indeed. Helps all of us to keep perspective.

Have a Great Day, Brent

By js55

Date 03-18-2011 15:15

I will readly admit, my knowledge of Six Sigma can fit on the head of a pin with space left over, but my limited experience with these relatively new quality assurance systems (and some not so new such as ISO though I do not know the actual history and time frames to be sure) is that they all seem to have been generated by people who are involved in 'widgid world' and do not lend themselves very well to other industry types, despite their protestations to the contrary. You can make them fit but it isn't easy. Pipoe fab, job shops, etc. are not what the inventors of these systems had in mind.
I wouldn't disagree with your post, however I think the idea of being honest with the customer for ASME/API/AWS, etc., applications would require an overlaying of these standards way of calcualting in which case, as I stated in the previous post, you can set up a system that works for you but for 'apples to apples' still must deal with ASME world, as an example, or 'honest for the customer' gets a little fuzzy.

Customer:
"What is your current reject rate?"
Fabricator:
"Well, to be totally honest, using ASME standard industry practice for pipe fabrication it is 7.5%. But before you decide this may be too high keep in mind that utilizing our newly developed Wazoo Six Sigma Kung Fu Feng Shui Fuzzy Logic Phases of the Moon calculation its actually only 2.3%. We have every confidence the 2.3% is more valid."
Customer:
In developing your new system did you have a preconceived intuition that the resulting number would be lower?"
Fabricator:
"What?"

By bozaktwo1 (***)

Date 03-22-2011 17:02

My reject rate calculation is based solely upon defects per inches welded. It is a collection-heavy data set, as someone must count the number of inches of weld, taking into account multi-passes, tacks, etc. I use the data only for internal process monitoring, and apply it equally in each discipline or process. The data set is solid and gives me a benchmark for the welders to meet. It works well for me. I wouldn't necessarily recommend it for any other operation.