T23 Boiler Tube - Curious Fracture 7

[Guest102023]

T23 Reheater Tube Fracture in Weld -- Anyone seen this fracture morphology before?


Several T23 reheater tubes failed in a similar manner after less than 6 months and ~250 cycles from commissioning of the HRSG. The circ. fracture originates in the weld, at the front (where you would expect bending force to be greater) then moves into the base metal as it propagates around to the back side. Meanwhile, a ~4" long tear happened on the front side. It seems like there is more than one thing wrong here.

Unfortunately I did not see the header side of the story (another case of repair first, ask questions later), so it is hard to confirm the lack of penetration that me and the repair contractor suspected.

Particulars: 2" T23, welded to P22 header with manual TIG, ER9018-B3, PWHT. Unknown tube origin, but they are significantly older than the plant. All zones appear to have normal microstructure (no ferrite at least). The correct filler was used. Weld hardness is 285HV1, HAZ up to 335HV1, base ~192HV1. I have found both tube surfaces decarburized. Scattered corrosion pits up to 3 mils deep on outer surface. A few very short secondary cracks and sometimes oxide along prior austenite GBs.

Any thoughts or suggestions? The hardness seems on the high side for a PWHT joint. I thought about reheat cracking, but did not think that can affect the weld also(?)

I am a newbie here so I hope this gets posted correctly ...

 

[Guest102023]

unclesyd - indeed, I have read these threads in the past and was hoping metengr would lead the cavalry over the hill

first question, I wrote 'flattened' when I meant to say they ground off header material to present a planar surface to the tube end. Sloppy writing on my part. But this was hearsay from the repair contractor; I have not seen it personally. Don't know if that is standard industry practise; it would be simpler and more reliable than dressing the tube end to a saddle shape I suppose.

second question, I cannot answer that one

If anyone can point me to good references that address these HRSG operational issues I would appreciate it greatly. I have downloaded all the available issues of Comb Cycle Journal and a ton of EPRI stuff.

One more thing - I would like an opinion on the weld zone hardness I reported previously - it seems a bit high for a PWHT joint ...

 

[metengr]

brimstoner
Thanks for the kind words.

This paper would be of interest to you on Grade T23 material (welding and fabrication). Also, best of luck to you and your new career.

http://www.msm.cam.ac.uk/phasetrans/2005/LINK/68.pdf

 

[rmw]

The way HRSG"s are designed and built, there should be some real good job security in this new career. They introduced a new buzz phrase to the industry; "cut your way in - weld your way uut".

You might want to rummage around on this site:

http://www.hrsgdesign.com/

Best of luck.

rmw

 

[Guest102023]

Thanks, rmw.
I now know what you mean by 'cut your way in' - I have some pics from the repair contractor on this job. I look at the MBA beancounters on this projects as my job security, because they are so fixated on upfront and material costs.

In the refinery business they use the term 'opportunity crude'. The opportunities are for corrosion engineers and failure analysts ...

cheers

 

[Ron]

The weld macro shows little solutioning...does further observation show fusion problems?

 

[unclesyd]

The fracture at the weld is a little troublesome from several points, the high hardness and the fact it is a repeat occurrence. Just looking at the literature on T/P 23 it looks very good. But after reading a little about the welding and PWHT it appears to require more attention than was probably given during the fabrication and/or repair. One thing is that requirement for PWHT is waived for material less than 10 mm, thk. until you read caveats. Even though it does show some higher hardness in as deposited it mentions temper embrittlement in almost the same sentence. Looking at the difference in impact properties 200 J to 20 J, some say 10 J get scary.

Here is a little information on welding of Cr-Mo boiler steels.

http://www.pdfone.com/download/20_keywordweldingtp23steel/home.pdf

http://www.t-put.com/EN_welding_CrMo.pdf

 

[Guest102023]

unclesyd,

The weld was PWHT due to P22 on the header side of the joint. The filler is -B3.

I noticed a few prior austenite GB penetrations along the weld fracture, the most prominent of which is in the attached pic. This one has oxide, which is very interesting. I cannot be sure if the fracture (which was ductile but not very) was intergranular or not. I would certainly think that impact properties are closer to 20J than 200J ...

On the longitudinal tube tear, very rough chevron marks point to an origin at or near the weld. The fracture surface had a lot of ups and downs of up to almost 1 mm in scale; I'm wondering if this suggests anything untoward in the tube condition. Comparison with the textbook 45° shear edge at the circumferential terminus confirms it was very dynamically loaded.

Thanks for the references. I would be interested in opinions/philosophies on filler metal for this joint - matching for the tube or for the P22 header. Also I am interested to know failure mechanisms in these joints after 5-10 years (including type IV creep I assume).

 

[davefitz]

There are several severe thermal stresses that can be imposed on a RH stub to header weld during startup operations on a HRSG.
a)reheater spray water causes alernating dry-out overheat folowed by wet subcooling of teh tube. Whiel the tube itself can accept fast temperteru cahnges, the close vicinity of the thicker header surface will not respond to the fast heat-quench-heat cycles the same way. The result can be severe alternating shear stresses at teh stub to header interface- the telltale indication of shear is that it is a maximum at the wall centerline thus the crack should originate there.
b) same alternating tube temperature variations as (a) , but if also causes a high tube to tube temperature difference which can lead to a high alernating bending moment on the tube if there is inadequate flexibility in the tube bundle design. For example, 3 rows of tubes from inlet hdr to outlet hhdr with no tube bends intervening is a low flex- high bending moment configuration that is not tolerant of such temp unbalances.
c) Water layout on the bottom of header or a header which has tubes stubs only welded on the bottom side of a top supported header- fast heatup of the header can lead to a top to bottom header temperature differential that would tend to cause the header to temporarily bow or "bannanna" , and if this deflection is restrained by the tubes, then it is overstressing some tubes. see below (d) for one such event .
d) If unit has a dry reheater during startup ( ie does not have a HP steam to reheater bypass system ) the tubes will be heated to 1150 F prior to turbine synchronization, and the instant after synchronization, the initial flow of steam thru the reheater will cuase sudden downshock of teh tubes and sudden upshock to the header.
e) typical causes of water layout in the header or alternating wet/dry conditions of the tubes are (1) OEM's are notorious for using insufficient straight transfer pipe lengths downsteram of the spray water attemporator, leading to un-evaporated spray water slugs to etner the inlet header that distribute unequal enthalpy fluids to the neext ros of tubes. Some tubes get hto steam, some get wet steam , and some get water. The min lenght of straight pipe downsteram of thw spray nozzle should be 0.35 seconds min transit time to next elbow and 0.7 secs min transit time to downstream header. (2) connecting the reheater header drains to the blowdown tank is asking for trouble- the reheater can operate under vacuum prior to turbine synch on a dry reheater startup system and it sucks water from the blowdown tank up into the header( as much as a 25 ft draw) .

Put some thermocouples on the headers and tubes and monitor with a Digital data aquisition system , with fast trace reset on a 1 sec basis, and prepared to be very surpised at the massive temperature differences you will see during startup. Don't even think of pluggin these values into a finite element model of the tube/header joint- you won't believe the stresses.

 

[Guest102023]

davefitz,

Thanks! Thats a lot to think about but clearly you know your way around a HRSG (and that startup is a tricky business).

Could you define 'water layout'?
What is the specific function of reheater spray water (besides temperature/thermal expansion management in general)?

I revisited the many photos and see that the 3rd tube that was sent elsewhere did not have a vertical tear, but otherwise similar morphology in the weld fracture. It is located at the end of a tube row, while the two tubes I have (one damaged beyond usefulness) are from near the middle - the one I am examining in the middle row, the other in the front of three rows).

It is clear that operational issues are at the root of these failures, as the customer has acknowledged. I will offer the last piece of your advice as a recommendation.

 

[davefitz]

brimstoner:

The water layout at the header invert is just remnant moisture that was separated from a 2-phase flow mixture that may have discharged from the smaller tubes. IN the individual tubes, the steam+ water velocity may have been high enough to entrain the water mist as droplets, but the lower velocity in the header may allow the water to drop out of entrainment and layout on the header bottom.

Also, water can be inducted up the header drain connection into the reheater header if the header drain was erroneously intertied to a superheater drain line or if the reheater is under vacuum during startup.

 

[Guest102023]

davefitz,
So, water is going where it should not, or else its presence (if unavoidable) is not being handled correctly?

 

[davefitz]

The proof of whether the cause in this instance is liquid water would be found from review of test thermocouples placed on the tubes and headers and also a review of other plant data , such as RH spray water flow rate, position of the header drain valves, etc.

If the data shows 200-400 F temperature differences between the top and bottom of the header , or between an individual tube stub and the header or to the average of the other tubes, then it is pretty certain that liquid water found its way to where only superheated steam is supposed to be.

It goes without saying that the reheater tube bundle was not designed for such temperature diffences. But it gets better- the temperature differences may be alternating at dozens of cycles per second , adding a prompt fatigue failure component to the problem.

So, add the thermocouples, and have fun figuring out where the water is coming from.

 

[davefitz]

error- dozens of cycles per hour.

 

[Guest102023]

Thank you, davefitz.

BTW, can someone tell me the status of grade T23 in the Code? I don't have access to the most recent editions. I saw SA-213, T23 on a data sheet - can I infer T23 is no longer a Code Case?

 

[Guest102023]

Another question re: Vickers hardness (see my original post).

Could I consider the weld and HAZ somewhat high for PWHT material? The literature indicates HAZ hardness for T23 of under 350 HV in the as-welded condition.

 

[metengr]

Grade T23 is still in Code Case for Section I applications.

 

[deco0404]

The hardness values are what you might expect after PWHT for P91 material! I'm afraid I'm not too familiar with T23, but use P91 & P22 a lot. You should find out how the hardness was conducted, and in particular the surface preparation prior to doing the test. In order to do a site evaluation of HAZ's on small diameter tubing, the weld needs to be polished and etched ( as the HAZ is so narrow ) and a UCI hardness tester, rather than a rebound type ( indentation too large)should be used

 

[deco0404]

Just had a look at specs on T23. Hardness is very high after PWHT in HAZ! I'm not sure how much of an effect the W & V make, but wouldn't expect that. 350Hv is on the very upper edge of what is acceptable on P91.

We recently did a couple of new builds on HRSG boilers, and where we did find difficulties was in the PWHT of stubs to headers. The difficulty is that in order to do a proper PWHT, the header needs to be heated up as well as the stub, and it is also quite difficult to get good contact between heaters and stubs, and even thermocouple placement is difficult and ultimately very critical on these. With these materials it is easy to get up to & through soak temp, the trick is the controlled cooling, and 95% of the difficulties we had came from problems related to cooling............

 

[Guest102023]

Thanks deco0404,

I am aware of the difficulties of effective PWHT on some of these components - in the past it has kept me gainfully employed doing failure analysis and field hardness testing. (Don't even get me started on the kids swinging their 'calibrated' Brinell hammers for some heat treaters.) As-welded P-91 hardness is somewhat higher than that of T/P23.

For this job I had hardness testing done on a prepared metallographic section using Vickers (1kg). I am aware 1kg may overstate hardness relative to the standard 10kg, but from my experience I think it would be very minor. I have compared the mean weld and HAZ hardnesses to 4 available PQRs, which are all significantly lower. Except for the 1kg test load and variations in PWHT times/temps it is an apples to apples comparison.
Also, my results are similar to published data (simulated HAZ) for non-PWHT P23. I am trying to locate a CCT curve for T23.

Starting to look as if PWHT was not very effective. Not surprising in the field, but a bit shocking for an OEM shop welded component. But its all very consistent with the brittle-looking weld fracture.

 

[metengr]

brimstoner;

Also, my results are similar to published data (simulated HAZ) for non-PWHT P23. I am trying to locate a CCT curve for T23.

.

Did you obtain a copy of the paper I referenced above? There is a CCT for Grade T23 on page 3!