T23 Boiler Tubes - Fracture in HAZ 3

[Guest102023]

I suspect reheat cracking as the failure mechanism in a 2"Ø T23 reheater tube that failed in the HAZ. Following are the findings so far:

(1) The tube is joined to a P22 header with ER70S-B3 filler (undermatching for hardness and creep strength w.r.t. the tube).
(2) The fracture surfaces are well oxidized (leak-before-break it appears).
(3) There are almost no indications of creep microcracks sub-fracture surface.
(4) HAZ hardness approaches 300HV, decreasing to 205HV in unaffected base metal.

I am not looking for anyone to solve my problem, but I would like to know what conditions could predispose T23 to reheat cracking, and what is the incidence of this type of failure in HRSGs?

 

[metengr]

I would think given your findings above, this is more fatigue crack propagation than reheat cracking given the location of failure, hardness mismatch and assumed cycling duty.

Normally, reheat cracking is intergranular for this alloy is in thicker wall sections.

 

[Guest102023]

I should have added that I see what looks like multiple fatigue origins on the sides where stresses should be greatest (6 and 12; in the vertical plane normal to the header direction). SEM shows mixed IG and ductile fracture modes in some areas, but these are more recent than the heavily oxidized surfaces where features have been destroyed.

Puzzling to me that the tube could have been leaking for what looks like a long time before complete fracture. But then it may have been broken a long time; it was a single fracture, and one leaker might have gone unnoticed.

 

[Guest102023]

p.s., any thoughts on the hardness levels? They seem a bit extreme.

 

[metengr]

The 300 HV looks reasonable to me for the T23 base metal HAZ with no PWHT. I believe you had a leak in this tube for some time before final failure. This happens more often then you think in tube to header attachment welds, basically from fatigue crack initiation and propagation.

 

[Guest102023]

Sorry again, should have mentioned the weld was in fact PWHT @ 1325/1425°F, which should give HV numbers closer to 200. Assuming it was done correctly of course. I don't know how much error in hold temp. would cause this hardness. Too high a temp. (although less likely) as well as too low could do it I think.

 

[davefitz]

some more info would be useful.

For example,

was it a full penetration weld with nozzle reinforcement or was it a partial pen weld with add'l header wall thickiness for reinforcement?

What was its operating life to date- is this an infant mortality or has it been thru a series of severe startup/shutdown/spray water cycles?

T23 is bainitic and requires a fast post weld cool down to ensure bainite- if the header was very thick, then the cool down rate may have been inhibited.

 

[Guest102023]

Thanks davefitz
(1) Full pen + fillet to ~1" thick header. The site pics just received show fracture coinciding with the weld toe.
(2) Operating cycles are in the low hundreds; under 3 years in service. They start the unit too hard (more recently backed off somewhat) and have had severe draining/water hammer problems.
(3) Bearing your comments in mind, the high hardness is a bit puzzling, but it has previously been found at the opposite end weld, where there were several weld metal creep-fatigue failures ~1 year ago. That was less surprising considering the undermatched B3 filler.

I am now wondering how much tube vibration could affect random tubes. Tube length is >40 ft.

What is the 'optimum' heat treat temperature for T23? I have googled for lots of information, but there's not a lot of fundamental data out there; maybe its time to shell out for the V&M book or the HRSG Users Group text ...

 

[metengr]

Based on the original and subsequent code case revisions I have reviewed for Grade T23, the PWHT temperature range is the same as that for P-No 5A base material. Despite this being a creep strength enhanced ferritic steel, it is not like Grade 91.

The most important feature associated with this alloy is the chemical composition to ensure a fully hardened bainitic or bainitic/martensitic microstructure, prior to tempering for original heat treatment. Subcritical PWHT is not as critical to exceeding the lower critical transformation temperature, as compared with Grade 91.

 

[davefitz]

Although not stated above, I guess this tube was installed on a HRSG superheater header.

The very fast startups of these units plus the disparity in the tube wall thickness vs hdr thickness plus common occurrence of overspraying of the attemporator often leads to fatigue failure at the tube stub to header weld.

The rate at which the metal response to temperature fluctuations varies by the square of the wall thickness. The tube is likely 6 times thinner than the header, so it responds 36 times faster to transients, such as caused by subcooled water spray hitting a hot tube or header during startup . The tube responds quickly and cools down to 650 F, but the header remains at 1050 F ( guestimate), leading to a high shear stress at the weld interface between the tube and header.

You can prove or disprove this theory by placing a few thermocouples on the header and also monitor 10+ tube stubs close to the header- the plot ot T vs time during startup ( with 1 sec scans) will show how severe is the magnitude of the temperature difference during cold startups as the superheater spray is activated.

So, it is just as wild to assume the failure is fatigue as to assume it is weld reheat, only the metalllurgiacl analysis and field test data can yield a correct answer.

 

[stanweld]

Agree with davefitz. This type of failure has occured before on HRSG T23 tubes at the weld toe of connecting tube to header welds. Additionally you have indicated water hammer. Combination of thermal cycling and vibration = low cycle fatigue failure.

 

[Guest102023]

Thanks everyone, you are confirming what I most suspected, that it was fatigue. I have had a further look at the micros, and even with the oxidation I think the fracture is too far away from the CG-HAZ to consider reheat cracking. I am curious about the fairly heavy decoration of the prior austenite grain boundaries - what is it and whether it is normal (I have also seen it in published pics). The high hardness levels still puzzle me.

They have run the duct burners very aggressively (for the 1st year at least), but of course they are not monitoring tube and header temperatures. I am not sure where the duct burners are situated - the upper end? [I don't want to spell out the name of the constructor, but you may have guessed it reading between the lines.]

 

[stanweld]

Poor PWHT practice can lead to high HAZ hardness. I have seen such in the more distant past by some Korean manufacturers. While I know it is seldom done, I would recommend that the Owner and EPC firm make a shop visit to the manufacturer to ascertain their PWHT practices, especially if not previously known. We've had 3 costly incidents in the past when PWHT was inadequate, providing very hard HAZ's in the tube or weld metal.

 

[Guest102023]

I suspected poor PWHT already from a previous failure. I don't know where these components were sourced; I do know the tubes have surface decarb, indicating that they are not V&M or equivalent.

davefitz, temperature monitoring would have been invaluable in solving this case (or even preventing it).

 

[Guest102023]

How do I source the V&M book on T23/T24 (also the one on T91/T92)? The V&M website and googling have not yielded results.

 

[metengr]

You need an actual contact within V&M to obtain a copy OR you can purchase several mill runs of tube

 

[Guest102023]

Which is easier?

 

[Guest102023]

Received a package with two small books in the post today ... I just had to ask them. Good thing, because my garage isn't long enough to warehouse 40 foot long tubes

 

[metengr]

I did even better, had my pocket-sized T23 books from VM hand delivered to me at last weeks ASME Code meeting by one of their engineering folks.

 

[Guest102023]

metengr,
I can't top that; you win. Did they give you tubes too?