Dissimilar Cr-Mo Weld Joint Failure (continued) 4

[Guest102023]

Has anyone encountered DMW failures in P91-T23 joints welded with B9 filler? Because I think that is what I am seeing in my 1.5Ø x 1/4" T23 SH tube joined to a 2" thick P91 header ('joined' indicating the past tense). The hardness drop is 300 to 200 HV500. The tube is at an 'extreme' location at the end of the header, nearest the sidewall, in the first SH harp facing the CT.

Four metallographic sections were examined. Fracture in the tube HAZ follows the fusion line very closely in the outer 50~75% of the thickness, despite a very low fusion line angle, typically 45° but approaching 0° at a few points. Fracture is square in the inner 25~50%, which I believe is terminal fatigue. Unfortunately the failure is old and oxidation has obliterated the surface details.

 

[metengr]

Yup, along the T23 side of the weld joint. 9Cr filler metal is a sink for carbon diffusion (aka carbon migration) because of the affinity for chromium carbide.

 

[stanweld]

In the HAZ, near the fusion line on the T23 side of the joint decarburization will occur especially if the PWHT used was on the high side - carbon migration being a function of temerature and time. We make this weld using a near matching T23 welding filler metal from either of two manufacturers and PWHT at 1350 F.

 

[Guest102023]

Thanks to both for confirming my thinking on this.

Tube HAZ hardness is low, and remains fairly flat all the way out into unaffected base metal, so it appears PWHT was indeed on the high side. Hardness is below 200 everywhere in the tube, with 170 just 0.5mm from the fusion line, with several results in the 170s in the base (the lowest 167). Grain boundaries are well decorated with carbides.

(1) I think I have the evidence, but what detailed examination to confirm carbon migration would you recommend?
(2) Tell-tale orange carbonitride particles are evident in the matrix; is this expected or another possible indicator of incorrect processing?

p.s., the tube exhibits surface decarb, so it did not have a happy start in life. You might even say it was not born into a good family (like V&M). :-(

 

[metengr]

(1) I think I have the evidence, but what detailed examination to confirm carbon migration would you recommend?

Chemical etching and microhardness testing. The local region of decarburization caused by carbon migration is obvious in this DMW.

(2) Tell-tale orange carbonitride particles are evident in the matrix; is this expected or another possible indicator of incorrect processing?

No. In my experience optical metallography alone is not a reliable predictor for service performance with creep strength enhance ferritic steels.

 

[Guest102023]

metengr,

(1) That work is complete; my question is more of an academic nature. The hardness results certainly seem consistent with HAZ decarb.

(2) Not sure I was clear enough. I am asking whether the presence of these particles is normal or whether it indicates possible processing deficiencies. I would not attempt to draw any conclusions about creep strength from the microstructure. This was more of a 'BTW' question.

Further to a previous discussion re: etchants, I switched to Vilella's for this job, with far better results. It revealed more detail in the T23 and weld, whereas nital barely touched the B9 weld.

Thanks again.

 

[metengr]

brimstoner;
I would highly recommend you obtain the publication below from EPRI regarding Grade 91 steels. It was made available to the public domain because of safety concerns;

ERPI Report 1023199. This is one of the best reference documents to date to keep owner/user/manufacturers out of trouble with Grade 91 steels.

In this report, the desired microstructure for Grade 91 consists of M₂₃C₆ carbides and MX-type carbonitrides that precipitate on lath boundaries and defect sites. Typical photomicrographs will not display the precipitates. So, if you are seeing large MX-type carbo-nitrides, there is a processing problem.

 

[Guest102023]

Excellent document, a lot of the physical metallurgy will be transferable to other CSEF alloys. Perhaps a similar document is needed for grade 23; if anything, the issues are more numerous.

My carbonitrides are clearly visible at medium magnification, ~400x, although they would not be as obvious in a micrograph at that that magnification. At 1000x, no problem.

 

[Guest102023]

Now to drop the other shoe: what is the recommended NDE method for detecting future DMW failures in other, similar tube welds?
I expect x-ray would be difficult; perhaps reflected shear wave UT or advanced UT method. Is there an accepted industry practice?

 

[metengr]

Linear Phased array UT would be my choice.

 

[Guest102023]

After some more metallography focused on the fusion line, a light-etching zone about 8-10 micron wide is seen. Typical for what you guys have seen?

Does HAZ decarburization happen strictly during PWHT, or can it continue at service temperatures? I would think not very much, but I'd like it confirmed.

 

[metengr]

After some more metallography focused on the fusion line, a light-etching zone about 8-10 micron wide is seen. Typical for what you guys have seen?

Yes.

Does HAZ decarburization happen strictly during PWHT, or can it continue at service temperatures? I would think not very much, but I'd like it confirmed.

No. It is elevated temperature exposure in service for many thousands of hours.

 

[Guest102023]

Thanks again, metengr.
What is the actual fracture mechanism in this narrow zone?

 

[stanweld]

The creep rupture strength in the denuded zone is lowered. Generally speaking, the failures have been due to a combination of thermal fatigue and creep due to cyclic operation of combined cycle plants or straight creep in base load plants.

I'll be interested to learn how similar joints behave in some recent coal fired plants. While I know how we (as the Assembler) made the welds in the field, I'm not so well aware of how the welds were made in the boiler manufacturer's shop since I did not have any contractual controls over the manufacturer.

 

[Guest102023]

thanks stanweld,

As much as cycling plants are abused by operators, most of the failures I have investigated had pre-existing issues from the factory. With old-fashioned boilers this is rarely the case. The gas plant building boom happened with great haste, designing on the fly, and using new and under-appreciated exotic alloys. Among many other factors.

Much of the component fabrication is done out of sight and far away, and on-site surveillance (something I grew up with) is a favourite place for EPCs to slash their costs. Surveillance these days seems to be just an exercise in faxing. We've all seen what happens when regulators fail.

 

[stanweld]

Agreed re shop surviellance. "We don't need no stinkin shop surveillance" being the current mantra. Besides, the've got an ISO 9001 type Quality System (not worth the electrons retained); and the [captive] AI will find anything that was wrong.

One would think that after the millions of dollars that we have had to spend to rectify shop screw-ups, there might be some lessons learned. Oh well!!!

 

[Guest102023]

MBAs provide lots of material for me to help owners learn lessons, keeping me gainfully employed. So its not all bad. And you can buy ISO certification at the aptly named convenience store.

We could consume multiple lagers together discussing the problems of modern manufacturing...