Leaking head gaskets in a large heat exchanger with high delta-T's 8

[JEFF1]

The heat exchanger is about 14 feet long, has "high temperature water" (primary side) entering at 380 F, approximately 32" head gasket diameter, 4 pass, and secondary water entering at approximately 150 F. High water temperature drop on the primary (say 200 F) and minimal temperature rise on the secondary. The head is very rigid in construction. The portion of the head where primary high temperature water enters regularly leaks because of loss of bolt tension, and the gasket blows out (not just leaks). Gasket is "corraguard" by Garlock, considered a good gasket. Has anyone heard of a problem like this before?

 

[ScottFree]

Yes. What type of bolting are you using? ie. A193 B7?<br>
What size are the bolts and what do you torque them up to? Do you do a retorque after a period of time? What makes you think that you have a loss o ftension problem? <br>
Any other info?

 

[Carmagen]

Yes. The temperatures are not all that high though. With proper flange, gasket, and bolting design and boltup procedures, we seal flanges in refinery applications at 800F and more. I initially suspect that there is a boltup procedure or gasket design/installation problem. How long has the exchanger been in service and has it always had the leakage problem? Does the leakage only occur during particular phases of the operation?

 

[JEFF1]

The studs are 193B7. We do three star pattern torque sequences, followed by a sequential pass, and then a second sequential pass going the opposite direction. The bolts are 3/4"-NC 10, and final torque is 220 lb-ft. We do not retorque, because we were concerned that the studs could be yielding (because of loss of tension). Loss of tension was determined with an ultrasonic device which measures transit time of sound waves in the studs. For three identical heat exchangers, loss of stud tension was noted only in the head quadrant where the 380 F water enters. The exchangers have been in service for two to three years, and leakage occurs in the fall and spring when there is no thermal load. The metal substrate in the gasket has reduced thickness in the quadrant of the head where the leakage occurs (as measured with vernier calipers). Only one true "blowout" occurred, and this may have been due to poor tensioning. But even with tensioning verified with ultrasonic equipment, the gasket still leaked at the same place. The heat exchanger is oversized, and a disproportionate amount of temperature drop occurs on the primary side in the first pass; hence,the head has large thermal gradients.

 

[Carmagen]

I guess we're all making the basic assumptions that the flange has been "properly" designed per ASME Code procedures for the gasket and bolting being used and that there is no facing damage that could aggravate the problem. Here are some ideas to consider.<br>
<br>
When you torque in the "star" pattern, are all passes done at the maximum torque value or is it done in stages? "Staged" torquing helps keep gasket compression more even as the flange is bolted up. Detailed boltup torque procedures that I've been involved with have typically used ~40% of max torque on first pass, ~70% of max torque on next pass, then two passes at 100% torque.<br>
<br>
The 220 ft-lb torque is a bit high based on some "rules of thumb" I've used, but still should be nowhere near yielding the studs. Assuming the studs, nuts, and contact surfaces are well lubricated with a colloidal nickel type lubricant, ~190 ft-lb torque on a 3/4" stud should get ~50 ksi tension (which is what I would typically shoot for). The 220 ft-lb you quote should get ~58 ksi, still under the 95 ksi yield point of 3/4" B7's. "Hot bolting" a leaking flange at this torque level should not be a problem and is commonly done. It does not always solve the problem, but shouldn't yield the studs.<br>
<br>
I have seen some exchanger applications where there was a "high" delta T on the tube side that caused leakage problems at the channel cover flange. Once we identified the services where this was a problem, we designed future exchangers so that the pass partition plate did not extend to the channel cover. It was terminated short at an enclosure plate that ran between it and the channel shell. This isolated the channel inlet nozzle higher temperature flow from the channel cover, so that the channel cover was at a relatively uniform tempeature.<br>
<br>
I'm not familiar with the Garlock "corraguard" gasket you mentioned so cannot comment. If you haven't already, perhaps talking to Garlock might be helpful. I am familiar with Corrugated Metal Graphite Covered type gaskets (e.g., by M&P Graphonic) and Grooved Metal Graphite Covered type gaskets (e.g., Kammprofile by Flexitallic). Both of these types tend to be more forgiving to flange defects, movement, and thermal gradients. Using one of them might be worth a try if all else fails. I used the Kammprofile gasket as a replacement on an air cooled exchanger cover plate that was causing lots of problems during field hydrotesting (replacing a jacketed gasket) and it worked great the first time.<br>
<br>
Another "trick" I've used in situations where thermal transients have been a problem is to place a Belleville disc spring washer between the back of the nut and the flange surface. This tends to keep a more even load applied to the flange during thermal transients.<br>
<br>
Anything that could be done on the operations side to slow down the temperature change to give the flange more time to equilibrate would obviously be helpful.<br>
<br>
Good luck. Let us know how it all turns out.

 

[JEFF1]

We are working with a consultant, and the calculations thus far show no design problems. The head is to be shipped out for measurement of trueness of the face shortly. However, three identical exchangers have all leaked in the same quadrant of the head (where the hot fluid enters), so trueness doesn't seem to be the likely problem.<br>
<br>
We do torque in stages, and if I remember right, its 60 ft lbs, 120 ft lbs, 180 ft lbs, all star pattern, and then two final sequential passes at 220 ft lbs, one CW and the other CCW. (The one that leaked most recently was actually a little lower than 220 ft lbs final torque, perhaps more like 190 ft lbs). The stud elongations were measured ultrasonically to confirm, and were a little over 50% of yield. R.e. hot bolting, my sources frown heavily on this practice because of the safety ramifications.<br>
<br>
We have been toying with a variation of the idea of isolating the channel cover thermally. This was to cut some weep holes in the pass partitions so that some high temperature water would trickle through, keeping the entire head warm. This, however, would result in some flow through the tube bundle when no load is required, and could lift the shell relief valve if there is no secondary flow. Another variation of this idea that I like is to put an insulation liner in the hot quadrant of the channel to keep the hot fluid out of contact with the steel. High temperature plastics exist that could work. The plastic would not have to be watertight, since no significant conduction path would in the water between the plastic and the steel. But you'd have to find a means of fastening, and allow for expansion of the plastic. Further, if the plastic ever broke away, it would block the tube sheet, and there's the question of setting up different kinds of thermal stresses in the channel. I've heard of insulating with concrete, but this seems a bit much. We've also thought of installing a sleeve in the hot quadrant that would be quarter-cylinder shaped, and result in an annular space of trapped fluid in this same shape between the sleeve and the channel. This trapped fluid would act as an insulator.<br>
<br>
I got the name of the gasket wrong. There are two similar ones, Corragraph and Graphonic. They are similar, and I think that we're using the Corragraph.<br>
<br>
R.e. Belleville washers, we've been told that it's hit and miss unless you do it just right with respect to the mating surfaces. We have been working on smoothing things out on the operational side.<br>
<br>
Thanks for you very helpful input.

 

[Carmagen]

Just a couple of more thoughts. All things considered, some form of the "thermal isolation" idea might be the surest bet. Using insulating refractory to reduce metal temperature is commonly done in certain high tempearture refinery applications, but we're normally dealing with much higher temperatures than here and are covering entire surfaces (not quarter panels). You'd have to fine tune the particular refractory to use and its thickness to get the "right" metal temperature. Resco (and other companies) have a wide range of materials available.<br>
<br>
Of course you could just use a thermal insulation material on the inside surface of the quarter panel. In this case, for sure you'd have to use a piece of plate to retain it (i.e., build a box to hold it there). If differential thermal expansion and local thermal stress of the "box" would be a concern, you could probably design its end attachments to the channel and partition plate to be conical rather than straight to provide some flexibility.<br>
<br>
Another thought. Is the channel cover externally insulated? If not, adding this would also help "smooth out" the local gradients and is an "easy" first thing to try to see if it helps.<br>
<br>
Agree that Belleville washers have to be installed right and the mating surfaces have to be "good" in order for them to be helpful. Like so many things, we're just kidding ourselves if we don't cover all the bases.<br>
<br>
I understand the reluctance to "hot bolt" if there is limited local experience with it. Besides, I don't know your particular application or the ramifications if something "goes wrong." All I can say is that it is commonly done when needed for leaking flanges in refineries for high temperature hydrocarbon services. It is even done in some cases before the flanges start leaking in services that are known to be bad actors (e.g., high temperature cyclic applications).<br>
<br>
Regards.

 

[Tom584]

The ASME offers a bolted joint course based on a book called &quot;Introduction to the Design and Behavior of Bolted Joints&quot; written by a gentleman named Bickford who spent his working lifetime studying bolted joints. He has solved many problems for the nuclear industry sealing difficult heat exchanger closures. Check this book out. <br>
<br>
The only way to know that all your bolts are torqued properly is by using an ultrasonic extensometer. Otherwise, it is all a SWAG. <p>Tom Worthington<br><a href=mailtoworthi@astro.as.utexas.edu>pworthi@astro.as.utexas.edu</a><br><a href= > </a><br>

 

[abdul]

I recommend that you purchase bolts made from a mettallic alloy which has a low coefficient of thermal expansion, such as Invar. That way, bolt tension will not be lost.Just make sure that the material that you use has a minimum tensile yield of about 80,000 psi, and that the bolt material is compatible with the process(corrosion, and other issues).<br><br>Another alternative, is to fabricate from fiberglass, large thick disks which are sandwiches of stainless steel plates with fiberglass in between. These disks length should be about 10% of the bolt lengths, and fit between the nut and the heatexchanger.The large thermal growth of fiberglass, will compensate for the loss of stud tension.You will of course, need to insulate this area with thermal insulation, to maintain a high temperature on the disks. Another alternative is to use an excellent product called &quot;Superbolts&quot;.Just search under superbolts in the internet and you will find the company.Good luck.

 

[johngo]

The message below was posted by abdoul, on May/17/2000; I just joined this forum & I hope this reaches the right person):

&quot;I recommend that you purchase bolts made from a mettallic alloy which has a low coefficient of thermal expansion, such as Invar. That way, bolt tension will not be lost.Just make sure that the material that you use has a minimum tensile yield of about 80,000 psi, and that the bolt material is compatible with the process(corrosion, and other issues).&quot;

abdoul,
the low ceff. of thermal expansion with &quot;Invar&quot; bolts drew my attention; what type of bolts are they? what is the ASTM spec.?
Regards.

 

[abdul]

Hello,

There is one ASTM number for INVAR that I am aware of and that is ASTM-B-388.However, like I said earlier, make sure that the material used has a tensile yield strength of 70,000 to 80,000 psi( by heat treatment or coldworking).Search under &quot;INVAR&quot; in the internet, and contact the manufacturers or metals suppliers for detailed info.


abdul

 

[diamondjim]

I would like to know the length of the studs, if they
are short in length, putting a thicker washer at each
end would help to increase the length and improve the
design with little extra cost.

 

[scrobe]

In doing some research looking for technical information why one should not insulate flanges due to bolt relaxation resulting in leaks, I came across a paper from PVP94 that may be useful to you in your problem:
Go to

http://www.dynamicanalysis.com/pvp94.htm

It's a pretty good paper.

 

[laphroaig]

I have had success with gaskets type kammprofile and hydraulic tension of bolt. With hydraulic tension of bolt you get the same tension on all, or at least 50% of, the bolt at the same time.

 

[LTD]

Jeff1,

This should not be a new problem, and piping bolting flanges manuals might cover it somewhere...

But if that fails - you can make a &quot;bolt system&quot; with ANY CTE that you want by using differential expansions of dissimilar materials. Exactly what the fiberglas suggestion was. But I'd look at using aluminum too, or be very careful about what fiberglas you use at 350...

NUT/Aluminum washer/flange/aluminum washer/nut

The above config can be tailored by altering the thickness of the washers. If the washers get quite thick, more like &quot;stand-offs&quot; you can actually get a negative CTE for the system. You probably wouldn't leak if your bolts got tighter when they got hot... But it would look funny with long bolts (themselves a partial solution) and funny aluminum pipes for washers. (I'd use the largest diameter aluminum you could for obvious reasons.)

I've used this method to bolt together things with almost zero CTE that were 40' long and went through a 650F delta. It was quite difficult before I figured out this method. Fortunately I could counterbore my materials quite deeply and could therefore reduce the overall lengths and hide the fancy bits from curious eyes...

Larry

 

[jdsewell]

question,
could this be an issue of vibration? could vibration even be caused by fluid flow? the temperatures involved do not seem to be high enough in my mind to cause a great deal of thermal expansion (enough to cause loss of bolt tension).

 

[jhambham]

Jeff,
We have dealt with leaking heat exchanger (two pass) heads on our reactor water cleanup and fuel pool cleanup systems for years. These heat exchangers see a delta T of about 300 deg. F. We have also had some problems with vessel manways. Needless to say, at a nuclear plant, this is not a desireable condition. We had tried changing bolt materials, gasket designs, installing leak repair (KOPPL) injection fittings.

An old boilermaker came up with the best fixes-spring washers and/or hot torquing. Perhaps recalculating the your torque values in your procedure may be helpful. The most success we have had is with completing final torque pass once the heat exchangers are being brought online.

It looks like you've had a lot of advice. Hope you resolve the issue. Good luck,