P91 soft areas 2

[kossu]

I have heard of P91 piping as having soft areas at random in pipe as well as fittings. Assuming all heat treatment operations have been done correctly by original component mfg as well as assembly fabricator but somehow the chemical distribution was not uniform to begin with especially re the trace elements like V for example, could this uneven distribution result in the soft areas?
A non-metallurgist asking the question.
Thanks for any insight!

 

[metengr]

kossu;

Assuming all heat treatment operations have been done correctly by original component mfg as well as assembly fabricator but somehow the chemical distribution was not uniform to begin with especially re the trace elements like V for example, could this uneven distribution result in the soft areas?

From my experience with this material, the answer to your question is highly unlikely. The rationale is that given the advances in steel processing technology today and the hot working operations performed to produce seamless pipe, this would be very low probability.

Portable hardness testing on pipe must be done with care and with an understanding of principles behind hardness testing. Soft spots reported from hardness testing may be caused by improper surface preparation. I would strongly recommend you have a field hardness testing procedure developed and qualified for Grade 91 material.

 

[kossu]

Thanks, but...
The field hardness tests were done properly by grinding down about a sixteenth, polished, then lab tested using 3 different instruments, core samples taken and hardness measured thru the wall, all with same results. Material some 6 inches away tested fine.

 

[metengr]

kossu;
If what you report is correct, lets work our way down the list of possibilities. The heat treatment could still be the culprit if what you reported is correct. I have heard of cases where pipe spools and fittings were stacked in the furnace during austenitizing and subsequent quenching. After removal from the furnace as a batch, rapid cooling was not achieved. Despite the heat treatment report as being N&T, the end result was a bad microstructure from lack of quenching.

Did you verify the microstructure as 100% tempered martensite in the core samples? If you have verified the microstructure, run same EDS on the core samples to decide if the unlikely occurred (chemical segregation). Please report back.

 

[kossu]

Thanks meteng!

Had to go get more info.
What you suggested about the stacking of components and the resulting improper cooldown has been suggested as a possible culprit. The locations and the randomness of the soft areas still leave open questions.
Went back to the core sample reports I referred to earlier. A chem analysis was done on each and it was in accordance with specs. The hardness was low throughout but...the microstructure was void of tempered martensite and contained only carbides and ferrite. To a non-expert this does not mean anything except that the process was not successful.
Now that raises more questions. Can the normalizing and tempering operations be done more than once? Lets assume the pipe manufacturer stacked the pipes so that there was inadequate or uneven air flow during cooldown which resulted in non tempered martensite. The fitting mfg did the same. Then the fabricator welded the pieces together, added a induction bent bend to it (which after bending was heat treated independently) and stuck the whole assembly in the oven. If thats allowed I would think that tempered martensite would develop and the likelihood of soft spots disappear. Or would Murphy's Law apply?
Back to my original question about uneven distribution of trace elements. The chem test results on these core samples do not support that theory but I wonder if they could somehow still play a part here? If the components were heat treated independently and as an assembly I would want to look at other options.
What say you inspector Sherlock Holmes?

 

[metengr]

kossu

Can the normalizing and tempering operations be done more than once?

Yes.

Then the fabricator welded the pieces together, added a induction bent bend to it (which after bending was heat treated independently) and stuck the whole assembly in the oven. If thats allowed I would think that tempered martensite would develop and the likelihood of soft spots disappear.

Correct.

The chem test results on these core samples do not support that theory but I wonder if they could somehow still play a part here?

No.


Case closed.

 

[stanweld]

kossu,
Have you examined the heat treating records and have you inspected the furnace and temperature recording methods employed by the heat treater? I have seen many cases of what you describe (localized through wall ferrite); all of which have had their origins during the "normalizing" treatment. P91 has often been improperly heat treated by the manufacturers and by spool fabricators.

Additionally we see surface effects similar to that described and have associated them with local overheating from the use or oxy-fuel gas torches/rosebuds or proximity to burners during furnace PWHT. These surface effects have been seen to penetrate over 1/4" from the surface.

 

[kossu]

Thanks stanweld.
I have no access to the info you refer to. Between yourself and metengr I think the issue is resolved. Others have offered similar explanations.
Some of this stuff is almost 10 years old and if there were records nobody can find them. All we have is the piping itself.

Hippo

 

[unclesyd]

I just made a received a call that verified a statement in the post by stanweld concerning the proximity the proximity of the parts to the burners.

This incident involved quite a number of pipe spools that had been N&T and PWHT in a furnace where the initial charge had cocked the shields on several burners. The end result was that there were numerous areas that were improperly heat treated, determined by both hardness and microstructure.

 

[stanweld]

unclesyd,
Furnace control is so important with P91/P92, yet how often are furnaces surveyed? Most have not been surveyed in years, if ever, after initial installation. Also a delta of 100 F is often acceptable in a survey. If one were to stress relieve or PWHT at 1450 F using furnace T/C's only as controllers, the chances of exceeding 1500 F are quite real in specific locations. I have yet to see furnace inspection/maintenance records from any of the spool suppliers/heat treaters with which I have been associated and audited. I have also never seen an Engineering Specification requiring surveys to be performed. It would appear that our industry requirements are somewhat lacking in this regard.

I suppose that I'm just venting after [the company] spending over 10 million dollars to correct P91 heat treatment caused problems.

 

[kossu]

Thank you all: metengr, unclesyd and stanweld!!
Your advice has been most appropriate and it all makes sense.
From your frustrating comment stanweld it took some backbone from management to spend 10 million on a fix up. It must have been a serious screwup and large in scope. Care to elaborate some? Was it replaced or somehow fixed?
From my recent exposure to this, all of your comments offer a logical explanation as to the randomness of the soft areas.
In order to have spent 10 million bucks this problem must have been discovered some time ago. Timewise it takes months if not years to do the corrections of this magnitude.

kossu

 

[davefitz]

I think that we could learn something from the way that similar alloys are treated in europe. Since the 1970's , the EU had used a more difficult alloy, I think it was called X20 ( since replaced by P91). This X20 required the development of special techniques to ensure accurate PWHT and N+T practices, including design and op of furnaces, as well as QC practices such as monitoring each peices temp vs time , and hardness checks. Also, required training for operators of the furnaces and of engineers that specify the heat treatment for these difficult alloys is different.

A survey of EU practices vs US practices for difficult alloys would be a contribution, in my opinion.

 

[stanweld]

The problems were discovered through dumb luck during the waranty period; I can not elaborate how. Once discovered, and through audits made of the heat treating practices at the spool fabricator, extensive testing at all weld locations and in all induction bends (at multiple locations) and fittings was done. Over 10,000 individual hardness tests and hundreds of replication micros were made, resulting in partial spool, fitting and bend replacements; the finals of which will be made later this year.