Half-pipe coil overall coefficient 3

[mfowler]

I am currently designing a larger replacement vessel for an existing stainless steel reactor at our plant. To minimize the steel thicknesses we are investigating using half-pipe coils as the heating/ cooling source. In order to evaluate the new design, I need to calculate the steady state heating and cooling capability of the new design. I am having trouble finding some general recommendations for overall heat transfer coefficients to finish this calculation.

The vessel contains a light organic solvent (ether). The vessel will be heated with 80psig steam and cooled with 40F chill water.

Does anyone have a suggestion for a overall heat transfer coefficients?

 

[Montemayor]

mfowler:

I've done this twice, both times involving a kettle-type, batch liquid phase reactor. Both applications were done on a "best-effort" basis since the heat transfer calculations were predicting a very poor heat transfer rate and quantity.

The wrap-around, half-pipe, welded heat transfer coil has the following traits- as I recall:

1. It adds shell re-inforcement; you can get by with less shell thickness;
2. Unless you employ 2 (or more) parallel coils, you will wind up with a very long, single-pass, flow trajectory that yields either laminar flow or a very large pressure drop. Neither the laminar flow or the pressure drop are conducive to effective heat transfer;
3. The fabrication cost (carbon steel half-pipe on stainless shell) was very high; the welding labor was exhaustive and required care and precision due to localized overheating and warpage of the shell. This method requires so much welding bead (& resultant heat) that there is a concern for how it affects the parent stainless shell.
4. I calculated an overall "U" of 120 as I recall, but actually measured around 70 to 75 btu/hr-ft2-oF from field data; this data was taken on the cooling down done with cooling water in the pipe;
5. The heat up cycle with steam is even less efficient due to the two phase flow spiraling down the pipe(s).
6. The net effective shell heat transfer area is that shell portion directly inside the welded pipe; this is 60 -65% of the total area available - at best.

If I were to rely on jacket heating or cooling today, I would design for a conventional carbon steel jacket - a cylindrical jacket around the cylindrical shell (1/2 - 3/4" separation between both). If you want to accentuate the heat transfer film coefficient, you can use a spiral baffle under the thin jacket plate and get the same theoretical performance as you would with a welded half-pipe - and be able to carry out normal maintenance and reparirs on the jacket (which is not something easily done with the half-pipe. Either way, you certainly have to rely on an internal, efficient agitator to maximize the internal wall wetting and wiping action in order to have commendable heat transfer through the shell wall. I've found that the agitator design and type play a strong role in the effectiveness of the reactor's heat transfer. This means that you would be well advised to put a jacket on the reactor's bottom head as well.

I realize (and have suffered with) that internal coils are a mess with batch reactors, due to the cleaning and maintenance required. But one positive tradeoff that they offer is that they are efficient and they do not affect the shell and both heads with different stresses and expansion rates as they heat the reactor. Jackets will put different stresses on the shell and the heads due to their inherent location. I'm not advocating internal coils; I just mention them because they are an alternative. I certainly would never recommend dimpled jackets! They are a fad that appeared sometime in the 50's-60's and that turned out to be a maintenance nightmare. Their cost must be prohibitive by now.

My advice would be to take a detailed and serious critical look at the half-pipe coil - especially at the Reynolds number - pressure drop expected as well as at the amount of welding you will require. For practical purposes you will probably be looking at 2" or 3" pipe. Schedule 40 may be required for forming the spiral. This adds a lot of weight to the structure as well. Doing the job with a cylindrical jacket and spiral flat-bar baffles takes less steel, less welding, less cost, gives more area, and the same virtual effect. Plus, you can do repairs and maintenance much simpler.

I hope the above experience gives you some ideas.


Art Montemayor
Spring, TX

 

[mfowler]

Mr. Montemayor:

Thank you for your detailed input. I greatly appreciate the advice and the time you took to share it.

It appears that I have even more to consider than before.

mfowler

 

[aswainpkt]

Art & Mfowler-

I greatly appreciate your discussion on this topic as I am facing a very similar problem as well.

I am concerned that the heat transfer coefficient is as low as 70, but I am curious as to what were the operating parameters of that measurement (ie, flow rate, size of coil, coil and reactor temepratures of fluids etc.) .

Essentially we are building a reactor and chiller system to match. We have a proposed system using chill water and a half pipe coil but this post by Art is a cause for great concern. ANY insight or details you can proivde would be most appreciated. Thanks!

-Adam

 

[Latexman]

Bust the half-pipe jacket up into multiple parallel zones and then you don't have laminar conditions, or high pressure drop, or as bad of a two-phase flow problem.


Good luck,
Latexman

 

[25362]

If heating and cooling time are of importance, an alternative approach worth of investigation is heating or cooling with an external heat exchanger using a circulating pump, circumventing the need of having a jacket and a mixer.

I do not pretend to say that it would be cheaper, however, depending on the circulating rate, it would enable shortening the time needed to achieve the required temperature whether on cooling or on heating.

The overall HTC may be much better than that obtained by the use of a jacket and a mixer, in particular when one considers that the largest resistance to heat transfer stays with the liquid in the vessel.

 

[mfowler]

Since my last post, I took the opportunity to actually run the numbers to see if I could confirm the heat transfer coefficient suggested by Montemayor. I was able to find three well written articles on this subject:

Garvin, "Understand the Thermal Design of Jacketed Vessels", Chemical Engineering Progress, June 1999

Bondy and Lippa, "Heat Transfer in Agitated Vessels", Chemical Engineering, April 4th, 1983

Dream, "Heat Transfer in Agitated Jacketed Vessels, Chemical Engineering, January 1999 (beware of the error on line 23, Table 2, the exponent for the Reynolds number should be 0.8)

By my calculations, I found the overall heat transfer coefficient using cooling water in the half-pipe coil to be ~70 Btu/(hr)(ft2)(F). I compared this with data from Pfaudler on non-baffled annular jackets with agitation nozzles and found the HTC in this case to be ~65.

In my scenario, I am using steam to heat the vessel and water to cool the vessel. I was not able to find any data on steam in half-pipe coils other than the normal outside coefficient of ~1000 Btu/(hr)(ft2)(F). This would suggest that the annular jacket with agitation nozzles would perform equally well from a heat transfer standpoint. However, I am planning to build a 316L SS vessel which has a real crappy thermal conductivity. Therefore, the resistance of the vessel wall begins to play a part in the overall HTC. To achieve the pressure rating required by my process (100psig/ FV in both vessel and jacket with 1/16in corrosion allowance), the bottom head of the vessel became quite thick (1 1/4in) upon calculation with an annular jacket. This again pushes me toward a half-pipe coil jacket. As usual, there is one other complication. Applying a halp-pipe coil jacket to a vessel requires a lot of welding on the vessel wall therefore inducing a lot of heat stress to the stainless steel. Without proper post treatment this could lead to corrosion problems in the future that could seriously detract from the vessel's useful life.

In conclusion, there are many factors affecting the decision to use a half-pipe coil jacket or an annular jacket with agitation nozzles. All factors should be weighed carefully and completely to ensure the longest life for the money spent.

MFowler

 

[aswainpkt]

My situation is prohibitive against a pump around basically because we are dealing with propellants.

Here are the details:

I need 700,000+ Btu/hr cooling in a stainless jacketed reactor that holds 500 gallons of fluid. We are looking at a half pipe coil design that will cover about 50-60 ft^2 of area. We have proposed running 44 gpm of chill water thru a 3" half-pipe coil at 88 F would do the job but it looks like I have overestimated the heat transfer coefficient (I took roughly 150).

Basically Im horribly stumped now, and I need a solution...any and all help would be greatly appreciated!.


-Adam

 

[aswainpkt]

Mfowler-

I am very interested in getting my hands on a copy of these articles....where can I get them?

-Adam

 

[mfowler]

Aswainpkt,

The only suggestion I may have is to run your process at the boiling point of the reactor contents. By using an overhead condenser at total reflux, you can take advantage of the heat of vaporization to remove some extra heat. You are going to be hard pressed to remove that much heat through a jacket alone.

MFowler

 

[mfowler]

aswainpkt,

We had the back issues of Chemical Engineering in our department. I believe you can order back issue articles from the magazine web sites. If you still can't get your hands on them, let me know and maybe I can get them to you.

MFowler

 

[aswainpkt]

MFowler-

Anyway I coudl get a fax-copy of those articles would be great..we are kind of under the gun over here. I can compensate you in some fashion if need be for your time--we are running out of time lol

-Adam

 

[25362]

Aswainpkt, if you're time-pressed consider chilling the water to a lower temperature to improve the LMTD.

 

[mfowler]

aswainpkt,

Email me your fax number at mike.fowler@clariant.com and I will get them coming. I can definitely understand your position and would be happy to help out a fellow.

MFowler

 

[Montemayor]

mfowler & aswainpkt:

Guys, first allow me to apologize for taking a sabbatical from this thread. I didn’t lower the importance I originally gave the subject at hand; my complaining about boring weekends catalyzed my wife’s eagerness to replace all our guest bathroom faucets and the related plumbing. I’m through complaining about boredom – the Honey-do projects don’t pay well.

I consider this subject to be an important, hard-core engineering problem that is cost-driven but is pivotal to a successful kettle operation. Recent basic data furnished make it imperative, in my opinion, to get all our information out front as quickly and accurately as possible. We now know mfowler is at the design stage while aswainpkt is on the verge of field installation. I’m going to give priority to aswainpkt’s situation and discuss that in as much detail as I can within the framework of this forum. I’m going to assume that the operation is a batch kettle operation that requires fast and effective clean-out of the kettle once the reaction is complete and products are pumped out. I’m also going to assume the kettle has a pump-around capability with a reasonable capacity to recirculate and evacuate liquid. The kettle operation is carried out in the liquid phase only.

Allow me to summarize my comments and suggestions; I will explain further at the end:

1) Time is of the essence and the reaction heat has to have a sink that will provide a reasonable batch time and not be a production bottle neck; the best engineering step is that recommended by 25362: employ an external, conventional heat exchanger as a cooler. However, this is not applicable to aswainpkt’s kettle.

2) If aswainpkt has to work in the liquid phase, a total reflux condenser is also not applicable. This leaves one outstanding option: an internal coil or panels. If this is also not acceptable because of contamination or workability, then the reaction will be controlled by the ability to extract heat through the kettle’s shell and into the jacket. A 44 gpm flow in half a 3” pipe is going to give a horrendous Reynolds Number and a similar “U”, in my opinion, but I can’t come up with other options except what 25362 suggests: use a colder cooling fluid in the jacket – maybe brine. In this case, batch times will be longer than anticipated and unit operating costs are going to start taking on a heavier toll in the profit picture – something everyone dreads, because it then is out of engineering hands and in the MBA’s lap.

3) I estimate that aswainpkt’s kettle is about 4 ft diameter by 5 ft straight length with a net working volume of 500 gallons. For this physical size of kettle, I would use 1” SS condenser tube to fabricate a welded coil bundle that would have two internal headers to emulate multiple passes. This will be tough if the kettle has no liberal-size of manway to access the internal volume. An internal coil allows you to engineer the proposed results and gives the operators a direct control on the reaction’s heat release – something that is paramount to ensure a safe and predictable result. I don’t like internals – much less coils – but when you’re out of other options, it has to be considered even though you might have cause to reject it. There is no credibility attached to the assertion that an operator can control an exothermic reaction with a cooling jacket. I’ve never come across an acceptance of this type of process scope. An external cooler is the preferred, credible answer and an internal coil is the unavoidable alternate.

Internal cooling coils bring a lot of bad baggage into a kettle operation: potential contaminating leaks, clean-out obstacles, maintenance nightmares, potential vessel entry requirements, additional costs, etc. However, in this case, they offer the operating heat removal control that is sadly lacking in an inherently limited kettle jacket. Cooling jackets, in my opinion, cannot be regarded as serious, engineered controls for the effective and predictable removal of process heat. They are merely “best effort” type of coolers where the engineer exploits the given limited area simply because it is there. External coolers with variable speed, re circulating pumps offer the opportunity to seriously design and predict the batch times and reduce the operating risks while lending flexibility with the possibility of introducing brines, for example, if additional capacity is required. With the ability to vary flow rate, cooling medium, and heat transfer area you have the complete operating toolbox with which your personnel can take control. External coolers not only give you the complete, desired process control, they also inherently introduce a recirculating pump that yields additional returns in evacuating the kettle in short time – thus adding more effective kettle time to production efforts.

Needless to say, I’m elated that my prediction of a 70 Btu/hr-ft2-oF turned out to be not only a credible, but also a confirmed estimate on a kettle’s overall “U”. I have not read any of the articles cited. Since I’ve been involved directly in the past with this type of problem, I tend to depend on my personal experience rather than on published material. All of this serves to reinforce the basic design criteria confronting kettle reactor operations: Do not rely on the kettle’s cooling jacket to furnish reaction heat removal control. Take what is given to you, but design on a firm and dependable method to achieve complete and predictable reaction results. Batch reaction times in US and Canadian processing plants are too critical and expensive to take them lightly, in my opinion. Everything must be done that can add to the assurance that the heat removal will be effective and in the complete, safe control of the operator.

I wish I could add a quick, cheap, and effective answer to aswainpkt’s problem. However, I find the response of all the contributors to this thread to be positive and very important in assisting to bring a successful solution to this problem. The respondents to this thread have taken serious looks at what I consider an important subject and have come up with important contributions. I hope my token comments add to the useful help that has already been furnished. Needless to say, I wish you both all the luck in going forth.


Art Montemayor
Spring, TX

 

[aswainpkt]

Art-

Thanks a bunch for your help. It has been invaluble.

When you experienced a field htc of 70 Btu / hr ft^2 F, what were your operating parameters? I am getting close to 120 theoretically for my 44 gpm water jacket (which is similar to your operation as I see).

According to my calculations, the 44 gpm in a half 3" pipe is turbulent. The pressure drop across my jacket is in the range of 10-15 psi, does this sound reasonable to you?

Pump around recirculation coolers are out of the question due to the nature of the contents..so are internal coils.., even though they would be the best solution I feel.

Thanks!

 

[Montemayor]

aswainpkt:

I'm glad all of our comments have been of help to you. You have formidable constraints placed on your kettle design - one of which is the physical size of the vessel, as kettles go.

I don't have the specifics of the basic design data I used on the kettle I worked on. I just remember the summary information that I noted. I do recall that the following were heat transfer design criteria:

1) The heat-up phase of the initiator fluids after they're charged;
2) The cool-down required of the on-going reaction for control purposes; this involves a complex, pseudo-mixture of the reactants and some of the products as they are being formed. An average of viscosity, density, and thermal conductivity values were assumed - depending on the degree of completion we were at during the reaction phase.
3) The cool down of the reaction products at the end of the reaction; this usually was the worst-case scenario in the case of very high MW products with high viscosity and density values.

The "U" I calculated was for a relatively "light" MW product run using 50% reaction completion. I could do this because, as regarding reaction control, I was really depending on my outside, vertical heat exchanger to do the real, serious cooling. My reactor was a 4,000 gal, 316ss kettle with 8 ft diameter by 116" straight height. The external spiraling half-pipe had a considerable overall length and I couldn't modifiy it for multiple parallel flow in the half-pipe; I recall something in the order of 20-25 psi as the pressure drop. Once the half-pipe is welded on the shell, you've "bought the ranch" and have to proceed with what you're given. This may be similar to the situation that you find yourself in at present - except you have far more constraints. I seriously doubt if you can rely on obtaining a better "U" from a practical standpoint. I hope that I turn out being wrong on this point and that you're able to resolve your cooling with what you have in front of you. Good Luck!

Art Montemayor
Spring, TX

 

[MJCronin]

mfowler,

For what it is worth, there are software packages available to evaluate various combinations of half-pipe and other jackets versus agitator performance.

This type of heat tranfer problem ends up being calculated many times for a variety of conditions

The "Madison Technical Software" people have a heat transfer package that may be of interest to you:

http://www.madisonsoftware.com/page.php4?page=deltat

My opinion only

MJC

 

[negrita]

This message is for M Fowler:

Dear Mr Fowler:

On October 10, 2003, you said that since your last post, you took the opportunity to actually run the numbers to see if you could confirm the heat transfer coefficient suggested by Montemayor. Then you said that you were able to find three well written articles on the subject, and suggested that if a member was unable to get them you coud fax the articles. My fax number is: (415)768-5511

I would like to have the three well written articles on the subject of heat transfer:

Garvin, "Understand the Thermal Design of Jacketed Vessels",
Chemical Engineering Progress, June 1999

Bandy and Lippa, "Heat Transfer in Agitated Vessels", Chemical Engineering, April 4th, 1983

Dream, "Heat Transfer in Agitated Vessels, Chemical Engineering, January 1999.

In exchange for these articles, I will be glad to make a contribution to Eng-Tips Forums

My fax number is: (415)768-5511

 

[negrita]

Dear Mr. Mfowler:

I received the fax. The cover sheet indicated 26 pages, but the actual number of pages received including the cover sheet was 21.

Garvin, "Understand the Thermal Design of Jacketed Vessels",Chemical Engineering Progress, June 1999,contained pages 61 thru 66. From the way page 66 ends, it appears that the article has more pages. Some pages are missing.

Dream, "Heat Transfer in Agitated Vessels, Chemical Engineering, January 1999, contained pages 90 thru 93. From the way page 93 ends, it appears that the article has more pages. Some pages are missing.

Bandy and Lippa, "Heat Transfer in Agitated Vessels", Chemical Engineering, April 4th, 1983, contained pages 62 thru 71. It is complete

If you can, please fax me the missing pages in the first two articles. Also I want to know to whom I should send my check contributing to the Eng-Tips Forums.

Thank you again for your help.