AWS D1.5 Bridge Code: PQR question 4

[DrMetal]

My company and our customer have a difference of opinion on the meaning of AWS D1.5 Para 5.12.1 and 5.12.3 and other associated paragraphs in section 5.12.

BACKGROUND: We performed a PQR on HSLA steel plate (A709/A572-50) for a bridge construction job in San Diego. We used the "Heat Input Method, 5.12" for the PQR. This is a 1G postion (i.e. flat) PQR coupon. We used a 1"T x 29"L test plate, with a 1/4" backing plate, with a 1/4" root opening, and a 45 degree included angle (i.e 22.5 deg per side). We ran two stringer root passes at relatively slow speed to tie in the plate roots with the backing plate. The root passes were the highest heat input of about 52 KJ/in.

Then we ran about 8 layers of additional stringer weld beads, for a total of about 19-20 passes. Out of those 20 passes, we had only 3 passes (including the 2 root passes) that exceeded 50KJ/in heat input.

The balance of the stringers in the PQR varied from about 40KJ/in down to a low of 25KJ/in. The PQR was then PWHT'd (i.e. stressed relieved), and then completely tested per AWS fig 5.1 (i.e side bends, charpys, full section tensiles, and macros). All tests passed and were well above the minimum requirements. CVN's were almost double the minimums, and absolutely NO cracks were visible on the bend coupons. In other words, in our opinion ... a perfect PQR.

So here's the problem ... our customer's welding expert consultant is rejecting the PQR because of the large variation in the heat inputs between the passes, but gives no basis for the rejection except to state that, in his opinion, the heat inputs on the lowest heat input passes are not high enough, and not uniform enough (i.e. that there it too much difference between the highest and lowest heat inputs). He is interpreting AWS 5.12.1 as meaning that ALL passes MUST be run at the highest heat input possible ... which is nearly impossible to do. We can run slow on the flat passes, but have to speed up on the steep side wall passes to prevent bead rollover, undercut, etc.

AWS 5.12.1 states to use "...a WPS that produces the highest calculated heat input ...". Then AWS 5.12.3.2 states the the WPS welding parameters for production shall not exceed the maximum values recorded in the PQR (which implys that there will be variations in the heat inputs from pass to pass).

There is NO mention in the code about any tolerances on heat input variations from pass to pass when welding the PQR using the 5.12 method. We welded the PQR coupon, using the maximum heat inputs from pass to pass (i.e. slowest travel speeds) that allowed us to deposit a sound weld on individual passes. And we will write the WPS's to be below the highest heat input recorded.

The customer's expert is rejecting the PQR, saying that the PQR does not represent the high heat input because of the relatively few high heat passes relative to the 19 ro 20 passes, and the variatoins in heat input between the passes. In addition, they are telling us that if needed, we are to grind out portions of previous passes during the PQR to allow us to weld larger beads to keep the heat inputs higher (e.g. on the final cover passses). We have never heard of this approach, and these are the issues where we disagee.

We maintain that the proof is in the extensive testing, all of which passed with flying colors, and that EACH individual weld bead is completely represented in the test results, due to the way the tests are conducted. If a single weld bead were too brittle, it would show up in those tests, especially in the bend tests.

Does anyone have any experience with this? Or know of a source or expert or official interpretation on this issue? Any help would be greatly appreciated. Our customer's expert has the customer convinced that they are correct and that we need to do another PQR ... which will take a few weeks and cost several thousand dollars. We need to convince our customer that his expert is wrong, and is trying to rewrite the code to satisfy his opinion. But, maybe we are wrong. Thanks in advance for any help. Mike Perkins, AMECO, Cleveland, OH Cell: 216-272-9300

 

[DVWE]

I've dealt with a similar situation, but per ASME, not AWS. And I'm not familiar with the bridge code. But, this may or may not help. Per ASME, a WPS which has been qualified with everything except for impact testing does not require complete requalification. In fact all that has to be done is welding on a thick enough plate, following the WPS, to get some impact specimens. In our case, we welded 5/8" thick plates that were approximately 8" long if I recall correctly (just thick enough to get the required 10mm impact specimens). Our PQR had the same issue, which the customer did not like, which ranged from 30 to 65KJ/in. On the re-test, we were able to narrow the gap to about 50 - 65KJ/in, doing the same thing you are being requested to do (grind on the previous passes). The customer ended up buying off on this, but it took a lot of work to convince them why it is difficult to narrow the heat input gap - as you have indicated, other problems may arise such as overlap, undercut, etc.

I don't know if the bridge code will allow this or not, but you may want to suggest it. Because it can be done at a fraction of the cost of a full blown PQR. To me, when all was said and done, it was sort of a "meet in the middle" solution, of course, after all tempers had subsided!

 

[DrMetal]

Thanks DVWE ... but I won't show your info to our customer. Then they'll say ...see, we're not the only ones who do this!

What I am really after is an interpretation of the wording in 5.12.1. We are unclear as to what exactly it means when it says the qualification WPS must "produce the highest calculated heat input". Does this mean that you use the highest heat input of the highest single weld bead, or that ALL weld beads need to be at that high heat ... which you and I know from a practical standpoint, would be impossible to do ... especially on a horizontal or vertical up PQR ... but maybe a little easier to obtain in a flat position. We say that the absence of any further guidance in 5.12.1 is allowing our single bead theory, because there are NO tolerances or limits set on weld parameters specified in that section.

I have sent an official inquiry to AWS today ... but no telling how long that will take for a reply. So we are probably stuck doing another complete PQR. I don't think AWS (or our customer) will let us do only the CVN coupon as you suggest. BTW ... the bridge code is quite a bit more restrictive than ASME in many areas ... surprising to me at first, but considering what bridges do ... I believe that makes sense.

 

[GRoberts]

Caltrans has special provisions regarding this exact issue. They restrict the amps, volts, travel speed on the PQR to be within the qualification window of table 5.3. For instance, voltage can only be +/- 7% on the PQR and the WPS. So your customer is right in that he might not be the only one requiring it, but the fact that Caltrans has this specifically spelled out in their special provisions indicates that they do not think it is a requriement of D1.5 and must impose it seperately. Typically, even for Caltrans work, a common way to calculate heat input is to use the average heat input for the PQR as the maximum heat input for production. That way the differences in heat input on the PQR are somewhat compensated for.

 

[DrMetal]

Thanks for the info GR ... We know about the Caltrans Special Provisions, as they are a part of our contract. However, those provisions you refer to specifically apply to qualification to AWS 5.13 method, and-or to joints that do NOT meet figs 2.4 or 2.5. Our problem is that the customer is trying to apply them to us anyway, and after the fact.

These provisions do not apply when you qualify using the AWS 5.12 High Heat method with joints that DO comply with figs 12.4 and 12.5. With the 5.12 method, you must use the highest calculated heat input of the PQR to determine your WPS production weld parameters, which have to be within 60% to 100% of the high heat pass on the WPS. The code is quite clear on this.

We then dropped down 10% less than the high heat, and wrote the WPS to the +/-10%A and +/-7%V rules you refer to. Then our travel speed has to be restricted to within the 60% heat input limit. The only Special Provision rule that applies to this method is to average the volts and amps (but NOT travel speed, not heat input) in calculating the heat inputs for the production WPS. But that is not an issue, since we ran constant Volts & Amps throughout the PQR process.

This really boils down to a single issue, and that is the interpretation of the wording in 5.12.1 that the customer is interpreting as meaning to run ALL passes at MAX heat input ... but they cannot give us any tolerances to stay within when we ask them. They are just not comfortable with the range of heat inputs in our PQR. We say, there are NO tolerances or limits, and that is what the entire qualification process is all about ... albeit to prove the procedure works by all the required PQR testing. Any other thoughts?

 

[HgTX]

D1.5 is written in a fantasy world in which all the PQR passes are the same. It talks about "the value" on "the WPS". There is a change coming up in a future edition (um, 2012?) clarifying the limits of PQR pass variability, along the lines of what California did (I think AWS did it by % rather than Table 5.3).

Here's what I do when checking PQRs to make sure the passes aren't too scattered:

I discount the root and cap passes if they're different. They're not really entering into the physical tests, and they throw off the average for no significant reason. I also throw out the minimum and maximum heat input passes, just to be generous, and because one outlier doesn't make much of a difference. Everything else, I look at the minimum and the maximum and the average for the amps, volts, and heat input, and also travel speed if it's a 5.13 qualification.

I care about all this because if those passes are too variable, then the test doesn't mean anything. If you give me something where half the passes are at 40 kJ/in and half are at 100 kJ/in, that's not the same as if you'd run the whole test at 70 kJ/in. The physical test results are supposed to say something about what you'd expect if you ran the test at a particular set of parameters, and if you're all over the place, the results can't be tied back to anything in particular. (I've had some fabricators run the passes all over the place and then try to use the average, like the example I gave above; I've had others try to use a single outlier as the official minimum or maxium. Neither of those will I accept.)

I use Table 5.3 as a guideline but not an absolute, since there's nothing in the code.

Then the logic goes like this.

Suppose you have a WPS that was followed during the welding of the test plate. Let's pretend that test plate isn't a PQR plate but some production piece. The WPS, if it was to have been followed during the welding of the plate, needs to cover the full range of what was done during that welding. That WPS then needs to be qualified by a (hypothetical) PQR. If you were to set up the parameters for a PQR, what values would you choose for your amps & volts & travel speed to qualify that WPS? If you can't pick one set of parameters that you'd use because that WPS is spread too wide, then the welding on that test plate can't *be* qualified, let alone be used as the basis of qualification for something else.

So sometimes I use Table 5.3, and sometimes I look to see how low the lowest value in the range is compared to the average (a la 5.12.1). There's nothing official in the code, so it comes down to common sense and how the test results can be related to a set of parameters to be used in welding, and varied from by the specified percentages in the code.

As far as I'm concerned, it shouldn't be that much of a hardship to run consistent passes. Maybe not identical, but consistent. For an automatic process, no problem at all. For manual, one still needs to be able to be consistent. If one can't control one's travel speed, one can't control one's fillet weld size, and one needs to be able to control one's fillet weld size or one won't be able to follow a simple fillet weld procedure or set of plans.

Looking at the OP's example, I don't know whether I would have accepted your results or not; I'd have to see all the data. But you'd be cutting it real close, at best. I do agree with you that the cap passes shouldn't come into it.

I also don't think you're correct in saying that a single bad bead would be shown up in the test. That single bad bead is partially remelted by subsequent "good" passes, and besides, steel is good at redistributing stresses. It's not the same as if you'd done the whole weld with those same parameters.

Hg

Eng-Tips policies: faq731-376

 

[DrMetal]

Hg ... thanks for the info. We think we have a solution to satisfy the customer w/o re-qualifying. It involves some of what you suggest, but modified to allow us more freedom of movement in the travel speeds.

I get what you are saying in regards to the uniformity of the weld passes, and that is the sticking point with my customer too. But, as a MetE, and with lots of experience in this field, I personally believe that the PQR variations are fully represented from root to cover on a bead by bead basis.

The tensiles are FULL section, as are the side bends. On a side note, I don't think too many people appreciate how severe that side bend tests are. I had an ASME vessel overlay job a few years back where the Inconel overlay had to pass the side bends ... and we played hell getting them to pass with less than any single 1/16" crack in one or two passes on each test coupon. We finally figured this out, but it was quite illustrative to look at the side bends. When you have even one single small brittle pass, we would see those small cracks in the individual passes.

As you know, high heat stringer passes are slower, and thus thicker (assuming no weaving). If you have a single "brittle" pass in the layers, I believe it will show up in the tensiles and bends.

How many years has A709 / A572-50 steel been around and welded with everything from 0.032" GMAW/FCAW to 3/16" dia SUBARC processes? I'd have to guess as many as 30 years or more. Heavy SUBARC will dump in much higher heat inputs (maybe as high as 70"Kj/in or more) than the 0.063"FCAW wire we are qualifying (at 25 to 52 Kj/in), and still pass the physical tests. We believe the AWS code committee recognizes this long term history, and therefore does not restrict the heat input variability in the 5.12 method because they already know that if you pass the extensive testing required in 5.1, that the PQR is a good solid procedure, regardless of the variation in the heat input (within some reasonable limits of course). I realize this is counter to typical Engineering thinking about controlling an Engineering process to the closest possible parameters to prove the process works. But in this case, the process will vary in production, so a PQR with variable heat inputs qualifies a similar production process. And lacking any further guidance by the code, we believe we have an proven, acceptable PQR. To second guess the code's intent, or to interpret it to suit a personal opinion ... well ... that's what Technical Inquiries to AWS are for.

We feel there is no tolerance specified in the 5.12 method, because, the system, as specified, has been proven to work many times over. If AWS is going to revise this section, then that's great, if it clears up this confusion.

I tend to disagree with your assumption that a 40 to 100Kj PQR (which I would consider extreme variations) doesn't mean anything, and that it does not "represent" one run at a constant 70Kj. Do you have any data or studies to substantiate that statement? I'd really like to see them. It may change my thinking on this subject completely.

However, I'd bet that if you could actually get a PQR to pass with 40 to 100Kj, that a similar PQR at 70Kj would certainly pass as well. I'd love it if someone could provide data to disprove this. In your example, you are testing at extreme ends of a scale, and if successful, there is every reason to believe that anything in between will pass as well, assuming all other things equivalent (i.e same base and filler metal, preheat, interpass limits, ect.)

Simply put, table 5.3 DOES NOT apply to 5.12. Our customer keeps doing the same thing. But the code is very clear about the differences between 5.12 and 5.13. I don't see any basis to cross apply them, except to give someone a comfort zone not otherwise provided by 5.12.

I am not quite sure I follow your logic on your next example. It looks like you are working in reverse. In our case, the full range that was used in the PRQ/WPS was "qualified". Then, any production WPS written from the PQR that uses weld parameters anywhere in the range of the PQR is acceptable to use. If we choose to use +10%A and +7%V, then we have to increase our travel speed to remain below the max heat input ... just like the code requires. In reality, our production WPS is written with nominal amps and volts at -10%A and -7%V below the max values used in the PQR so that, in practice, we will never exceed the MAX PQR heat input.

As far as running consistent passes ... not too difficult on a 1G plate. But can you do it for a 2G position in a 1"T plate, with a 45 degree included angle (all on one side), and the bottom edge of the joint flat? It's a "little" more difficult to maintain consistent heat input.

I think I addressed the single bead pass theory prior, but I still believe all those individual beads are fully represented in the PQR testing. None of them are fully melted thru by succeeding passes, and all are fully distinguishable in the macro etches and bend tests (after etching). Again, if anyone has data to prove otherwise, I'd love to review it. On the other hand, we have a entire battery of PQR testing for 3 similar PQR's using trhe same base metal, and filler metal, and all without a single physical test failure.

I think what our entire process has proved thus far, is that, within the range or heat inputs we are running, this particular filler metal is relatively insensitive to heat input.

In any event, I am anxious to see what the code revisions say. Thanks so much again for you help on this. These postings have been quite helpful in raising new issues and thought processes.

 

[HgTX]

Trust me, the lack of tolerances given in D1.5 is not because they assume you can be all over the place and everything will be fine; it's because they assumed you'd be perfect and have everything identical. Not a reasonable assumption, but there it is.

The intent was not to put your full range of qualifiable passes into a single test weld. If that were the intent, there would not be min-max qualification (with two tests); there'd be an explicit provision for multiple parameters within a single weld.

I didn't say Table 5.3 applied to 5.12. I said I use it as a guideline. (It's also not to be used as tolerances applied on the shop floor to production WPSs; if one wants to work within a range, put that range on the WPS. But I digress.)

The discussion here doesn't come down to what will or won't work from a metallurgical standpoint; it comes down to what the code does or doesn't say. In truth, you could probably run most welding processes at most normal heat inputs and not have a problem. Code requires testing anyway, and in a particular way with a particular set of assumptions. Where the code is silent, I have to go with the code's assumptions when drawing my own conclusions about how to handle uncovered situations. (Of course, different people with different levels of exposure to the code development process and history will have different results when attempting to channel the code's "hive mind".)

Rumor now has it that the new code will be out in 2010.

Hg

Eng-Tips policies: faq731-376

 

[DrMetal]

Thanks Again Hg ... I agree with all you've said here. Since my last posting, our customer's Engineer has contacted several other people, includng someone at AWS on this issue. Verbally, the AWS guy restated the average heat input theory ... but also stated that there are NO limits in AWS using 5.12.

Here is what the Engineer told us ...

We withdraw our statement rejecting the PQR samples. The contractor is correct that there is no part of the code that limits the welding speed or Heat Input for the sample.

However, for determining the max heat input to be used in the production WPS, the heat input of the entire sample must be used as the max heat input for the WPS. That effectively means the average for the entire sample. There is no part of the code that allows the use of a single pass of weld in the sample. The calculated heat input for the entire sample must be used. We feel that this position is supported by the code, and has been supported by AWS Engineer. The AWS contact is also seeking a second opinion of the council on the issue, who meet next week.

If the contractor feels that the Heat Input of the PQRs done to date will not allow him the flexibility he needs during production, then they may choose to do another sample that results in a higher total Heat Input.

So there we have it ... at least until the Code committee convenes on this.

In the meantime, we have come up with what we feel is a good solution to this issue.

1) We averaged the heat input of all passes.

2) We made up a small excel chart that calculates the total travel speed range allowed for all possible allowed combinations of WPS Amp and Volt settings.

3) We attached the chart to the WPS, and reference the chart on the WPS fields for amps, volts, and travel speed.

Using this method allows for much greater range of travel speed for any given Amp/Volt setting. Typically we are getting 6-7 in/min range vs. only 1-2 in/min (if we use all possible setting ranges grouped together). When taken as a whole, the travel speed range gets very narrow so there is no chance of getting outside the heat input limits. But with individual calculations for each A/V combo, the range opens up quite a bit. I hope I explained this ok.

 

[DVWE]

DrMetal,

Please keep us posted. Very interesting scenario indeed.

Thanks.

 

[HgTX]

I use the average too; I just don't accept excessive variability among passes, but you and I are not going to convince each other on that.

I've seen other fabricators use charts like yours that give heat input range for each volt-current combo. I'm quite fond of those charts.

Who did you talk to at AWS?

Hg

Eng-Tips policies: faq731-376

 

[DrMetal]

Hg ... No, I agree with you. Within a single weld pass, we would not except hardly any variation in the PQR or in production. And within a production WPS weld joint, we would also expect the welder to choose his power settings within the WPS allowances, that allow the process to run the best in that particular weld joint and position, and then maintain a good consistancy in travel speed throughout. Maybe the only thing we might disagree on is the variation between passes in the PQR.

My customer (not I) spoke w/ Brian McGrath @ AWS. I don't know him.

 

[1oldwelder]

How was this issue finally resolved ??

 

[DrMetal]

- Still waiting for AWS to reply. They have acknowledged the inquiry, but no official response as of Aug-31-2009. From a practical standpoint, we resolved the "restrictive" weld parameter situation by adding an excel chart to the WPS showing Volts vs. Amps and calculating the min and max travel speeeds for each possible setting. This solved our immediate welding parameter (i.e. travel speed) dilemma for this project. If anyone is interested in the excel set-up, I would be happy to post it.

 

[DABwilldo]

DrMetal,

I'd be interested in a copy of your excel spreadsheet. Would you please post a copy of it?

Thanks,

 

[MalinQC]

I also would like a copy of that spreadsheet. My company is just getting into AISC SBR certification, and I am having to build this thing from the ground up. I have lots of ASME experience, but no AWS experience. As I read through the D1.5 I notice it is way more restrictive. SO any help is great.

 

[DrMetal]

Gents,
- Sorry for the delay ... been up to my beltline on this CALTRANS project. BTW ... still no official reply from AWS on this issue ... this project may be complete before we hear from them. I uploaded the excel spreadsheet for the WPS attachment. It is fairly self explanatory, but let me know if you have any questions. I will try to respond as soon as I can.