PQR Recorded variable averages?

By swnorris (****)

Date 10-16-2007 16:59

Look at the Electrical Parameters in Table 4.5

By rickc

Date 10-16-2007 17:15

Yes, but if my welder varied in travel speed from pass to pass which one do I choose for my range in accordance with Table 4.5? For example, if my welder was using FCAW and did 10 IPM on the first pass and 11 on the second and then hit something like 20 IPM on the cap which one do I base my +/-25% on? Or, say, his amps ranged from 270 to 290 while running the coupon what number do I base my +/-10% on?

I thought I had read something about using the mean value from qualification to base my acceptable ranges on but, I can't find where I might have read that in the code.

By swnorris (****)

Date 10-16-2007 17:34 Edited 10-16-2007 17:39

Just list the passes separately on your WPS. And add the notes:

Amps +/- 10%

WFS +/- 25%

That's just one way it can be done.

By rickc

Date 10-16-2007 18:03

Yes, that's how they did it in one of the examples in Annex N but, in another they showed 16 passes with 15 of them at 200 Amps and 1 at 180 Amps on their PQR (page 340) and on the WPS (Page 341) they showed a range of 180-220 Amps.

In that example from AWS, the average amps were 198.75 ((200*15+180)/16)) or 159 to 238.5 Amps qualified by for the WPS. However, their WPS shows a range of 180-220 Amps for all passes. Their Travel Speed ranged from 8 to 11 IPM with an average speed of 9.875 IPM qualifying 7.406 to 12.344 IPM and the WPS shows 8 to 12 IPM for all passes. Volts were 26-27 on the PQR with an average of 26.938 Volts allowing 25.052 to 28.823 volts on the WPS with theirs showing 25-26 Volts. All the numbers shown on the WPS are in the acceptable range based on the averages recorded on the PQR but, where in the code does it tell you that you can/should use the average? I can't remember where I saw it.

By 803056

Date 10-16-2007 18:59

I record the parameters of each pass welded. They are recorded in an Excel spread sheet. Then I use the analysis tools to find the median and standard deviations. I base my ranges on one standard deviation to throw out the oddballs.

Good luck - Al

By rickc

Date 10-16-2007 19:35

That seems like a great idea! I'll have to use that. Thank you very much.
...Still, does anybody know where in the code it says you can do that? I can't find it. Maybe I saw it in a different standard. Grumble.

By HgTX

Date 10-16-2007 19:55

I have a vague and currently unsubstantiatable recollection the next edition of D1.5 (or maybe the one after that) will be adding such language, along with a restriction on just how much variation those passes can have.

This is an issue that drives me nuts. I'm willing to throw out root and cap and min and max, but if the rest of the values are still all over the place, then what exactly are we testing? If you give me half your passes with heat input of 30 kJ/in and half 60 kJ/in, that's not the same as giving me a test weld with 45 kJ/in on each pass. You give me physical test results, but what parameters' physical properties exactly are being tested?

I'll choke down the rest of my rant now.

Hg

By swnorris (****)

Date 10-16-2007 20:05

rickc

I don't recall it being in the current code, but I found it in the 92 D1.1, in Part B Procedure Qualification. It states: "a change of more than 10% above or below the specified mean amperage for each electrode diameter....... a change of more than 7% above or below the specified mean arc voltage...... a change of more than 25% above or below the specified mean travel speed........

mean being something midway between two extremes.

That still applies today. I just think it got lost in translation as newer editions were released.

By rickc

Date 10-16-2007 23:42

Thank you Scott, that must be it!
I was reviewing a WPS/PQR combo to the 1992 standard semi-recently and using the mean is specified multiple times on pages 89-91 of Section 5.5 in AWS D1.1:1992. Ugh, so many clues.

Y'all are awesome!

By 803056

Date 10-16-2007 20:46

You make a very valid point. There are some people that make no distinction between mechanized welding systems and processes that are manual. For instance, it is not reasonable for one to expect that a welder, using a manual process, to weld with no variation in travel speed (that being the one parameter most likely to vary) while welding a groove in thick plate.

That is why I like to use the Excel analysis tools to determine the "normalized" values for the welding parameters. The mean of each variable has its value, but more meaningful is the range that is encompassed by the standard deviation, one SD above and one SD below the mean. Nothing beats the "ol bell curve". One SD above and below the mean accounts for 68% of the welding conditions used to make the weld. The "very high" and "very low" values will be discounted. They are the conditions that you don't want to include in the usable range of welding parameters to make the weld. They might include the one or two passes where the welder used high travel speeds to correct the weld profile, or the one pass that utilized a slow travel speed, but doesn't represent the normal welding conditions.

The same approach can be used of the mechanized processes, but the standard deviation will be smaller since there is less variation, thus the average weld parameter will be closer to the median.

It's a method of rationalizing the means of establishing the norm.

Best regards - Al

By HgTX

Date 10-18-2007 15:40

But there's only so far the "but it's manual" excuse should carry. I get PQRs with 100% variation or more from pass to pass, and I'm sorry, but if they can't get a handle on their travel speed, how can they get a handle on their fillet weld size? If 32% percent of your passes are more than a standard dev out, what set of properties exactly are you testing when you pull and bend the metal extracted from that test plate?

By 803056

Date 10-18-2007 19:24

What does 100% variation mean?

If the welder varies the amperage from 100 amps to 50 amps, that's a 50% difference.

Travel speed that fluctuate between 8 and 10 ipm, is a 20% variation.

Making a consistent weld size is not to say each weld bead has to be identical when making a groove weld. It is reasonable for the welder to use different travel speed for the root bead and additional weld beads in the case of a multiple pass fillet weld or a groove weld. Consistency in the travel speed of a single pass is not the same as varying the travel speed from one bead to the next.

Likewise, what is the actual welding amperage used for a single pass bead? Anyone who has usedd an ammeter or voltmeter can attest to the fluctuations of both a digital and analog meter while the welder is welding. Do you record the highest value, the lowest value, the "average" value?

What is reasonable? I guess that's were experience comes into to play and statistics helps to separate the meaningful information from the chaff. To expect a manual process to have no variation isn't reasonable to anyone but the academics.

Al

By CHGuilford (****)

Date 10-18-2007 21:21

When we conduct PQR testing, I record amps, volts, weld length and time of welding, and preheat/interpass temperature for each pass. I shoot for all passes to remain within the ranges given in the essential variables, using the average of the passes as the baseline.

We run the 5/16" single pass fillet test first (D1.5) , to determine what we will use for the groove test parameters. The difficult part is that welders invarialbly try to run "screamer" passes on the cap to avoid excessive reinforcement - which would be outside the range.

I wish AWS or someone would standardize/define how testing should be done (maybe it exists somewhere but I missed it). I learned what I know through a few years of trial and error in interpreting the code requirements and trying to translate them to what the customer wants. Strangely enough, many different people have many different opinions on right and wrong (go figure!).

But seriously, some folks insist on using the highest and lowest recorded passes for the limits; some are so befuddled by the code that they will "stamp" anything; some want to ignore root and cap passes; some want to include all passes; some insist on seeing data for all passes; some want only the averge reported for the PQR - you name it and there is an advocate for it.
Even on this forum, some very knowledgeable and experienced people are not in full agreement with how testing should be conducted/evaluated. All of the opinions have some merit.

Does anyone know if this has been addressed anywhere? If not, is there anything on the radar?

By HgTX

Date 10-18-2007 21:49

I think there's a +/-10% heat input restriction on individual test weld passes coming to D1.5 (can't seem to verify at the moment). Me, I'd still throw out root and cap passes and maybe even an outlier or two. But I'm sweet and accommodating that way.

It comes down to making that biggol' weld nugget in the test machine be in some way representative of the weld that will be in my final part. (Let's not even get into the validity of using a groove weld to qualify fillet weld procedures. I already know all the complaints, and there's no solution in the immediate offing.)

Hg

By CHGuilford (****)

Date 10-19-2007 16:24

You're right, there is an essential variable range for heat input in D1.5 and it varies according to the type of test being conducted. I didn't mention it for the sake of simplicity.

By HgTX

Date 10-19-2007 19:56

I'm not talking about the essential variable range though. Those might be used as rules of thumb, but they're intended to control how far away the WPS is from the PQR, not how much individual passes can vary within the qualification test. This is a new provision I'm thinking of.