Thermal Performance of Unknown Shell & Tube. 4

[tc7]

I have a shell/tube heat exchanger which has no nameplate and is of unknown origin. I know all about the physical dimensions. We need to use it as a cooler with cooling water in the tubes and process fluid (ethylene glycol) in the shell. How can I ‘reverse engineer’ this thing to estimate its cooling capacity? (I don’t have the luxury of budget or facility to set up a test to take measurements.)
Description is as follows:
107 Finned U-tubes end-to-end length varies but ~ total combined tube length is 32,852 inches.
U-Tube data: Copper 90/10 Nickel alloy, 5/8” O.D. x .035 wall thk.
Baffle Configurations: 3 each “edge flow” baffles; each baffle has 17.25 square inches of edge flow area.
2 each “center flow” baffles; each baffle has 13 square inches of center flow area.
Shell Data: Steel, 21 5/8” O.D. x 20” I.D.
Shell inlet and outlet flow areas; 6.5 square inches each).
Shell length (inside head to head distance): 13’-0”
Shell assembly is single pass arrangement.

For my present purposes, I can probably estimate pressure drops, all I really need is advice on estimating thermal performance. Thanks to anyone who can respond.
TC7

 

[pmover]

i would forward posted information (i.e. physical data) and proposed process conditions to a shell and tube mfg and have the mfg conduct an analysis/thermal rating of exchanger.
if not, post the proposed process conditions (i.e. pressure, temperature, fluid type, and flow rate for both process streams) and i am certain that someone will take further action.
an area of concern is the pressure rating, per asme code, of the exchanger. since no nameplate exists, are there any stampings on the shell itself that may provide useful information? conducting such an exercise may be futile if the insurance company will likely not insure this equipment. recommend investigating with corporate risk/assets group and/or insurance reps before proceeding with the actual use of exchanger.
the case may be that since you have physical data, the exchanger may be thermally, mechanically, and hydraulically rated and tested to conditions you proposed and then the company/insurance company may accept that as a "valid" and an insured piece of equipment.
regardless, good luck!
-pmover

 

[TBP]

To quickly calculate the required area of a heat exchanger, I use: A = Q/U X LMTD

A = surface area, in sq ft
Q = BTU/hr
U = heat transfer coef (maybe 200 for a shell & tube)
LMTD = log of the mean temp diff

See if the calculated surface area required is less than what you get when you figure the actual surface area of the tubes.

Have you pressure tested your heat exchanger? Glycol will cheerfully leak where water doesn't. It's always disappointing to have glycol getting into your cooling water, or, usually worse, having water dilute your glycol.

 

[tc7]

PMOVER & TBP
The heat exchanger will be used as part of a test set up. I will NDT the shell and heads and hydro everything before use. There are no pressure vessel stampings on the shell but the drawings I have allow the tubes as well as the shell to be hydro'ed to 800 psi. I will eliminate all structural concerns before I operate this. I expect the shell (process side) to see ~400 psi. The desirable temp range of the process fluid ( glycol) can be between 100-125 deg F but may not exceed 175 deg F. Heat input may be on the order of 1.3 million BTH/hr. I don't know what pressure or available temp the tube side (cooling water) will require; I guess that is part of my investigation, when I get a handle on the thermal capability of this thing, then I'll be able to know what the cooling water flow and pressure will require?

TBP - you mention the Q and LMTD but I don't know what assumptions I can make to permit me to estimate or guess at these values.
Hey thanks for thinking about the problem.
TC7

 

[pmover]

Q = U x A x LMTD
where,
Q = overall duty, Btu/hr
U = overall heat xsfer coefficient, Btu/hr-ft^2-°F
A = Surface area, ft^2
LMTD = log mean temp difference, °F

the duty can determined by:
Q = mass flow x Cp,fluid x dT
for both fluid streams.

b certain units are consistent in analysis.

again, i highly recommend using the gpsa data book for an initial analysis and/or contacting a heat exchanger mfg to conduct the rating.

you or the the responsible engineer will need to establish pressure, temperatures, and flow rates for both streams before any rating can be accomplished.
-pmover

 

[TBP]

I'm not sure that I'd put 400 PSI of anything in a heat exchanger with no nameplate, and unknown origin. However...

You have to work backwards - identify what it is you need to have happen at the process. Q = 1,300,000 BTU/hr. If you've got the flow, and % of glycol, you can get the inlet/outlet temps across the glycol side of the HX.

If you've got, say a 10*F rise on the cooling water through the HX, then: 1,300,000 / (10 X 500) = 260 GPM. If you can double the delta-T, the water flow gets cut in half.

The "500" converts lbs/hr to USGPM for water. 8.33 lbs/gallon X 60 minutes per hour = 499.8.

 

[Montemayor]

tc7:

Having been in the same situation in the past, the following is what I have done with successful results:

1) First and foremost, do what pmover and TBP have correctly pointed out: you must thoroughly identify what you have in an engineering manner. You must have a mechanical rating of the unit, such as it's MAWP, total heat transfer area (bare tubes - the fins are ineffective), the number of tube passes, the flow area per pass, the inlet & outlet nozzle sizes & ratings on both shell & tube side. You haven't furnished this information, but you obviously can, and should, get it.

2) Don't worry that much about the thermal rating; you're going to have to accept a conservative design. This is definitely not an "optimization" or academic attempt to calculate the answers down to the decimal place. The practical way it is done in actual industrial applications is that a conservative "U" is established. A value of 200 Btu/hr-ft2-oF is, in my experience, too optimistic. I would rely on a value of 150 as being more realistic for this liquid-liquid, U-tube application.

3) Worry more about being able to handle the required flow rates through the unit with a decent pressure drop on both sides. TBP is leading you in the right direction: you must establish what your flow rates are going to be, based on the heat load. From your very limited description of the equipment, I have doubts that you may be able to push 250-260 gpm through the unit - but, you have to furnish more detail.

4) Your shell side pressure drop can be calculated with the Delaware method. However, your description of the shell side baffles is "baffling". I have never heard of “edge flow” baffles and can't imagine what you mean. Try to use standard, conventional nomenclature as spelled out in TEMA, Kern's "Process Heat Transfer", the GPSA Engineering Data Book, or the HTRI. Otherwise, we are wasting a lot time and writing trying to guess what your nomenclature is about.

5) Be aware that the correct way to describe the U-Tube configuration is to state the number of hairpins. That way, one can safely know that there are a total of 214 total tube hole in your tubesheet. Additionally, we have to assume that there is one tube pass in the bundle - the liquid goes in one end of the hairpin and exits the other end. There could be more passes, but since you don't mention this, there probably isn't.

6) Make sure you correct the LMTD by the appropriate flow arrangement correction factor. In order to have this unit operate in the expected manner, the baffles must have a minimum of clearances and by-passing in the shell side. This is the main culprit in not getting efficient heat transfer. The baffles must be in good condition to allow for effective shell side flow.

7) Pay no heed to the fin effect on your tubes. The fins are more of a hinderance in this application. Fins are ineffectual in liquid service. They are meant for gas service where the film coefficients are much lower.

8) pmover recommends a shell and tube mfg conduct an analysis/thermal rating of the exchanger. I would not recommend this mainly because it doesn't add any value to your application. No exchanger manufacturer will, in my opinion, accept any responsibility or liability for a thermal analysis. The information, if you can get it, will only be worth the paper it's written on. You can do the thermal analysis with the same degree of confidence that they would do it under. The mechanical rating is a totally different animal. You can do a credible and professionally stamped mechanical analysis to the extent that you can furnish alloy, wall thickness, flange ratings, and other physical and measurable data. And you can do it with a conservative approach.

Hope this experience helps.

Art Montemayor
Spring, TX

 

[pmover]

tc7 and art,
i'll concur w/ art's comment #8, thinking that the exchanger is physically capable to handle required duty.

art, good job of further summarizing/detailing!
-pmover

 

[arsef]

Prior to installation the unit should have all flanges trued up. During removal and storage the flanges, especially the piping flanges may have been slightly bent. This may not be evident in visual inspection. This would cause leakage of the coolant and process fluid. Also the flange rating would need to be confirmed, after any machining.

 

[tc7]

Thanks for all of your thoughtful replies. I don’t like to let too much time pass before acknowledging but was away on travel. Everyone’s concern about safe pressure rating is well taken. Let me say that we have much experience with high and very high pressure systems. I will not leave anything to chance here. We take mag inspections, pressure vessel calcs and hydros very seriously. Art – you have given me much to think about and I am trying to do all my homework. Below are info you are suggesting is needed in each of your 19 Jun points:

1) Total heat transfer area (bare tubes - the fins are ineffective) - Total surface area of all 107 tubes is 64390 square inches.
The number of tube passes - see #5 below.
The inlet & outlet nozzle sizes - the entering and exiting tube manifold each have a flow area of 2.97 square inches.
Ratings on both shell & tube side – still working on this one, but as I said before, this is the easy part.
2) Agree, although I was going to try to calculate U by some textbook equations that I have found.
3) You said “I have doubts that you may be able to push 250-260 gpm through the unit” – You may be correct here but also I may be facility limited. Even if the unit can pass 260 gpm, I may not be able to supply this in my setup easily or cheaply. I’m hoping for flows in the realm of 100 gpm or less! (also see #6 below).
4) By “edge flow” I mean that these baffles are circular with the O.D. being 7/8” less then the I.D of the shell. The baffle perimeter has four each 1” wide x 7/16” high tabs equally spaced about the circumference to keep it centered in the shell. Therefore flow is around the outer diameter “edge” (i.e. the 7/16” gap) of these baffles. Total area for flow in each baffle is 17.25 square inches. The “center flow” baffles have an oval hole at the center. The total flow area in each “center flow” baffle is 13 square inches. (I don't know why the flow areas between these two baffles are different).
5) Yes you have interpreted correctly, 107 hairpins = 214 holes; So fluid goes in one side and out the other side one time. Now I would have thought this describes a 2-pass configuration since the fluid traverses the length of the shell two times before it exits.
6) You said “ Make sure you correct the LMTD by the appropriate flow arrangement correction factor.” I now begin to think my baffle arrangement is far from standard, but I have not yet found the references you have cited. I may have to take a very conservative guess on this factor. As for the LMTD, that is one of my biggest hang-ups right now. I just cannot get my hands around the flows I need or can achieve practically. And therefore the allowable ranges of inlet and outlet temps have got me stumped. In the 19 Jun posting by TBP, he suggests a calculation based on dT of 10 degrees and further shows another possible tradeoffs by doubling flow. So I guess determining the flow-dT relationship may be an iterative procedure, but when do you know you have it right or have it optimal?
7) and 8) Agree.

Well thanks again to all and all following advice gladly accepted.
tc7