Pressure Vessel PWHT Distortion 1

[WestCoast]

We have just finished repairing a pressure vessel that was distorted due to PWHT. The direct cause of the distorition is not clear (support, overheat, or flame impingement problem) and I was wondering if there is a someone who has dealt with this (I'm sure there are a few) and what kind of 'expertise' they could provide? It was an internally fired vessel that was only externally supported. the diameter is 14ft with a tan length of ~100ft (1-3/4" wall).
Any thoughts??

 

[carljpme]

This sounds like the case in Saudia Arabia where the company performed a 5 day PWHT.

 

[WestCoast]

The time to reach temperature was a concern, it took 2 days...

 

[WestCoast]

This case was very similar to the one in Saudi Arabia (I only wish I would have seen the post before it happened to us!). We repaired it using four 200 ton jacks spaced apart, moving them along the vessel and making many jacking cycles with small displacements. This required an intricate setup inside as you can imagine the hardware required was very hefty. We were very careful to displace the loads as much as possible and we managed to get it within code (within 1% and within external pressure requirements). We now need to determine the exact cause of this and help to avoid it in the future (for everybody).

 

[unclesyd]

Come back with a little more description on the mechanics of how you were attempting to accomplish the PWHT. Direct fire or indirect fire and anything else like was the distortion near a thermocouple?

Where was the distortion and how much was it?
What is the material of construction?

 

[bvi]

The whole vessel can ovalize if it is not supported. So if it is general ovalization that you have observed, it is likely to be creep deformation of the vessel during heat treatment, as a result of insufficient support for the heat treatment conditinos. You can get a good indication as to expected creep deformation by running a finite element elastic plastic analysis using an isochronous creep curve.

 

[WestCoast]

The vessel was only partially stress relieved. It was a 60' section that was internally fired through two radial manyways and fully insulated (one near the center of the 60' section, the other near the end). The distortion was in a 40' section between a head and a transition cone. The manyways were fired directly without any deflection or baffles and no inner supports were used (entire vessel was supported on three 24" wide saddles). We questioned this configuration and were assured that it was adequate. There were 19 thermocouples total (only three on each of the firing locations and not directly on the flame impinged locations).

When the firing first began, the vessel 'bananad' (for lack of a better term) from the top portion heating first and it eventually evened out. They then had difficulty getting up to temperature so the upper limit was increased to 1250F to get the coolest point to 1150F min, the whole process took about 52 hrs. The charts did not show a greater than 250F temperature differential between TC, but it also did not show the initial firing (started at around 300-400F).

We of course now realize that they were not experienced enough on a vessel of this size, and we paid for it. We need to find the exact cause and where the scope of responsibility is.

Was the distortion due to the flame impingement (no deflection) or just the prolonged temperature (creep as cb4 pointed out)?

Was the temperture not even and was it higher in areas away from the TC due to thermal pockets?

Was there lack of support under the vessel or would it have still sagged with additional saddles?

To answer your questions unclesyd, The material was SA516 GR70N. There was a distortion of max-min diameter of 4" at the worst spot which was directly where the one of the two firings occured and it was very 'egg shaped' as cb4 asked. The other firing was on the transition cone which was not affected.

As I said, we did manage to fix this successfully, but it was costly (not nearly as much as replacing cans). Any help on this problem would be very much appreciated. Thanks

 

[bvi]

The head and cone will normally act as stiffeners, preventing ovalization, so the ovalization due to creep sagging would be greatest at the point furthest from these, or the area of highest temeprature. Considering it is ovalized, and considering the high temperature and duration, creep distortion has to be considered to be a likely candidate. The way to confirm it is a creep analysis to predict the creep distortion of the vessel under the heat treatment conditions.

 

[unclesyd]

It looks as though you overheated a section due to the direct firing. I think the temperature of the area in question got considerably above the 1250°F recorded. Burner design and placement is most critical on a large vessel. I have seen two schools of thought large burners or smaller ones only more of them. The secret of either system is heater/burner output control. You have to have a good turndown on each burner/heater to control the approach and not overshoot. Thermocouple placement is also very critical not only to monitor metal temperature but control the heaters. One way to control the vessel temperature is to ramp up the temperature. This gives you time to adjust the burners/heaters and level out any temperature differences on the vessel. The thermocouple differentials should be no more than 250°F during the heat up. The ramping will give you a chance to narrow the range on the approach. A good heating rate is 300°F-400°F divided by the thickness in inches. I prefer the 300°F for the any material above P-1.

The use and placement of support is another area that is based on experience. The more you support a vessel such as yours the harder it is to heat in place because the supports get in the way. It is not as easy as some say as you just can’t throw in few supports.

If fuel was still cheap one might do it the old fashioned way, heat treat the cans, then do the girth welds after assembly..

As you have learned that this type PWHT is best left to one well versed in the art. If you a considering doing more vessels of similar size I would try to get some assistance from an experienced heat treater. Another possibility would be a burner manufacturer who could be of considerable help if approached right.

 

[WestCoast]

Thanks for your helpful input guys. I agree that there were several factor contributing to the problem (heat, support, Burner location/control, TC placement). I'm just wondering - if there is sufficient creep at this temperature and since this is commonly done, how do others do it? Like you pointed out, supports get in the way. What material do you use to internally support? Does it prevent the vessel from properly relieving stresses (equal thermal expansion)? I suspect that if deflectors are used and temperature is monitored more accurately, the support would probably not be an issue. I intend to get an analysis done to answer some of this...Thanks again

 

[bvi]

The creep rate at 1250F for carbon steel can be very high. You can estimate hoop bending stress using Roark and an applicable ring formula. You will be off most charts for carbon steel creep data, but can ball park creep rate using Larson Miller parameter. Just to rough out the problem.

Regarding proper set up, there is some art to this, and there is no substitute for using an experienced heat treating company.

 

[mullodon1]

You need to do a PWHT procedure. The following criteria must be known in order to correctly perform heat treatment to satisfy Section VIII, Div 1

1. P Number& Group Number of the material

2. Heat treatment method

3. Maximum charge temperature for furnace heat treatment

4. Nominal thickness

5. Heat up rate (MAX) 400 Deg F divided by metal thickness

6. Minimum & maximum hold temperature and time

7. Cool down rate (MAX) 500 Deg F divided by metal thickness

8. Cool down method

 

[WestCoast]

Yes there was a procedure - including all of this - and yes it was followed. It was a case of this being the largest vessel we have built and putting too much trust in a an inexperienced Heat Treater.

 

[carljpme]

I have seen vessels get flat spots on temporary saddle supports due to improper contact in normal PWHT oven so it is not just a problem for local PWHT or internal firing. We have started putting internal supports for large nozzles or we rotate the vessel with the nozzles at 90 degree to allow support during PWHT.

This is a question for cb4 or anyone else: Can you give some additional information about Larson Miller parameter? I was thinking that you could extrapolate the stress value for the shell material and run Zick calculations for the temporary saddles to detect potential local distortion.

 

[bvi]

You would use the Larson Miller parameter to extrapolate creep data over time and temperature. You can then use the extrapolated creep data to come up with allowable stresses. My recollection of Zick is that it does not give an assessment of the hoop bending stresses that cause ovalization, in which case it will not give you the complete answer relative to this condition. It does consider the local stresses at the saddle.

 

[GenB]

I see only one other issue.
Your vessel was overheated which violates the PWHT max temp.
now is is annealed and possible some areas were re-normalized.
you fixed the vessel after PWHT.
Now you will have to stress relieve the complete ass'y again... 'm sorry,
ER

 

[mac10]

Rarely in a failure there is only one cause, usually there is combination of factors that lead to the failure. So this case is not an exception: we have overheating, burner control and location, etc. as contributing factors to the vessel deformation.

But there are two points I would like to stress:

i) I agree with generalblr: you solved the mechanical problem, however if you have ASME Code Stamped Vessels you are probably required to heat treat the vessels again since you over heated it. Talk to your A.I.

ii) I do not concur with carljpme about running a Zick Calculation. It is not appropriate in this case, we are dealing with a vessel supported in three saddles. L.P. Zick method was specifically developed for vessels supported in TWO saddles. I would advise to use FEM.

Mac10.

 

[carljpme]

First of all Mac10 I did not say Westcoast should run a Zick calculation. I was asking if it would be a good approximation to extrapolate the stress values at PWHT temperature and use the value to evaluate the stress at two saddles using Zick.

I think CB4 has hit on the real problem, which is the creep value at an elevated temperature for such an extended cycle duration. The head and the transition will provide exceptional stiffness and the shell in the middle can oval.

I have never seen a vessel that would take 2 days to PWHT. We have performed PWHT on 7.25 inch thick steam drums using internal firing and it never took more than 30 hours.

Westcoast: Did you externally insulate the outside of the vessel before internally firing? You could take Hardness readings at many different locations along the vessel. This would give you a good indication if you overheated localized areas.