WPS for tri-metal weld F91-800H-316 ?

[davefitz]

There have been several failures of the dissimialr metal weld at F91- 316 piping components at combined cycle plants in the last few years, and we are proposing to be pro-active and upgrading these joints. The new joint to be proposed is 5" OD x 0.7"mw SA182 F91 to alloy 800H to SA182 F316, using inconel weld filler ENiCrMo-3 .

2 questions:
-can anyone advise an otpimized weld procedure to minimize weld reheat of the intial incolnel weld tothe F91 component?
-can anyone advise the best short term source for a short spool piece of alloy 800H?

thanks

 

[SRPJohn]

Davefitz,
We had two flow elements crack and leak a few years ago. We made temporary weld repairs then replaced them with an all P91 design. Besides, the 625 filler will embrittle at PWHT temps according to EPRI. Good luck.

 

[davefitz]

SRPJOHN:

I am familiar withthe flow element issue, but this is different. On Alstom GT26/ GT 24 plants, the once thru cooler OTC has a 316 outlet ring header, and its 4" NPS connection to the P91 transfer pipe is a forging with a dissimilar metal weld. Failures at that weld and also on the 1" NPS elvel taps are now being addressed.

The question is, what weld procedure would minimize the embrittlement?

Other tech papers propise Inco 82 weld, as applied to proposed liquid metal breeder reacotr piping.

 

[davefitz]

I have just heard that the embritllment is caused by service temperature migration of elements across the HAZ, so reqardless of weld procedure, the embrittlement is inescapable.

Is there availalbe a time-temperature LM paramter style correlation that canpredict the onset of embrittlement? Although the design temp is 1112F, the actual service temp is closer to 850F. A correlation might allow scheduling of replacment based on archived operating temperature.

 

[weldtek]

You might want to discuss this with an expert like Bill Newell. You can reach him thru Metrode in the US.

 

[stanweld]

I would recommend a P91-Alloy 600-316 joint. buttering the P91 alloy with NiCrFe or NiCrMo weld filler metal without niobium addidions and post weld heating the butterer P91. The filler metal used to make both dissemilar metals butt welds would be the same.

 

[davefitz]

stanweld + weldtek:

Is there any advantage to butter the F91 with the new Metrode P87 electrode ( developed by EPRI and in use at the Turk plant)? The C and Cr nearly matches P91 parent metal, so mgration of elements during service temperature might be minimized.

 

[metengr]

Absolutely, the Metrode P87 filler metal was designed for this DMW application. It was also assigned an F-No at the current ASME Code meeting in St. Louis. This should should work its way through committee.

 

[stanweld]

It wouldn't hurt to use the P87 electrode for the buttering and to make the alloy 600 to P91 weld recommended. It's probably not needed based on your stated "actual" operating temperature, unless the higher design temp. is periodically realized. To mitigate carbon migration, I would PWHT the buttered P91 at 1325 F.

 

[metengr]

I would not even use the alloy 600 in this combination.

 

[davefitz]

The P87 tech info from Metrode states it was specifically developed to avoid carbide formation at the F91- inconel eld HAZ.

The 800 H short spool piece would reduce the shear stress due to differential expansion by a factor of 2, compared to direct weld to 316.

The thicker wall to be used would reduce stresses caused by external bending moments and piping loads.

Will likely need to contact B+W or Metrode to determine othre welding details- apparently an elevated temperature test of the qualifying weld piece is needed before accepting the weld procedure.

 

[metengr]

davefitz;
You can optimize the DMW weld joint geometry to reduce thermal/mechanical stresses. You still have the CSEF steel as being the weak line in your proposed DMW. Either make it all Grade 91 as SRPJohn suggests or use T23 in conjunction with Grade 91.

 

[davefitz]

Meteng:

The upgraded weld is to applied to 14 duplicate plants , each already in service for 10+ years. Each OTC has been installed with 316 outlet risers and a common 316 ring header- replacing these with T23 risers and P23 / P91 outlet ring header might be today's fabrication choice, but I do not think it can be rationalized based on a pro-active upgrade of a single problematic weld.

 

[stanweld]

Alloy 800(H) is preferred as the intermediary material. I had a "Perry" moment and confused alloy 800H with austenitic stainless. Evidence of 'early onset" I suppose.

 

[davefitz]

As an upddate:
- TIG wire for P87 and INCoWeld A are also available, but you have to work at finding it

-a full summary of the development project is available free online from EPRI as report number 1018991 ( July 2009) at <

http://www.epri.com>

-a good presentation is " Opportunities for use of P87 weld Metal" by J Shingledecker at the 2nd Int'l Epric conf of 22 Jun11

-see also "Development of EPRI P87 solid wire " by J. A. Siefert et al , Mat'l at high temperatrue 27(3) pp 243-252

suppliers claim a sudden surge in demand for the P87 electrode in the last week.