Experience with mulitpass condensers 1

[tmengineer]

Hello,

I'm just about finished the thermal design of a multipass condenser and subcooler system and would like to know if anyone has some real world experience with these units to compare with the theory I've been using:

I'm proposing to use a vertical shell and tube condenser then vertical shell and tube sub-cooler to avoid temperature cross. The hot stream temperature profile is 90 deg C (water + alcohol) vapour -> 80 deg C liquid -> 20 deg C liquid. The cold stream is water with a temperature profile of 15 deg C -> 21 deg C -> 65 deg C.

The flow rate of water (5.5m^3/hr) is lower than that used in similar units on comparable sites so that warm water can be used for heat integration. Due to the low water flow rate 5 passes are being used in the condenser to get the heat transfer coefficient up. I've calculated it as approx. 500 W/(m^2.K) which I think is conservative. The water velocity with 5 passes is 0.21 m/s - I've seen in textbooks (Chemical Engineering Design. Principles, practice, and economics of plant and process design by Towler and Sinnott) that the velocity should be 1 - 2 m/s to avoid scaling. I've used a fouling factor of 3000 W/(m^2.K) for the inside of the tubes, as that was quoted for water, but will this be higher due to the lower velocity?

The model that I've made is for steady state but the units are to be operated in a batch system and I'm not sure how to model the start-up. The main output that I've calculated is the cooling area for both the condenser and the sub-cooler (15 m^2 and 5m^2 respectively) for a rating of 300 kW and 50 kW. What sort of safety factor is appropriate to apply for the required area?

Any feedback you have would be greatly appreciated!

Best,

Tom

 

[EdStainless]

What will the material for the tubes be?
The flow velocity is way too low, your heat transfer coef will be very low and fouling is almost guaranteed. How good is your cooling water?
In erosion resistant materials (SS and Ti) velocities of at least 2.5m/s are used, with the upper limit only set by pressure drop.
In steel or Cu alloys you need to be concerned with erosion and the flow velocities will be need to be limited depending on the material.
I know that you want the discharge water as hot as possible, but you will sacrifice efficiency and reliability if you don't follow the design guidelines.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[tmengineer]

Hi EdStainless,

The cooling tubes are made out of copper (C106 half hard 10 gauge).

The water quality for the region is defined as "soft" on the UK water agency website and has Calcium mgCa/l = 2.90, Calcium mgMg/l = 1.1, hardness as mg/l CaCO3 = 11.76, Clark degrees 0.82, French degrees 1.18 and German degrees 0.66. The numbers seem a lot lower than regions that are defined as hard water but I'm still concerned about scaling due to low velocities. I'm not sure how to put these numbers into a model? Or should I just stick with the value of 3000 W/(m^2.K) from the textbook?

I've played with the number of passes and 11 only gives a water velocity of 0.45 m/s and puts the heat transfer coefficient at 655 W/(m^2.K). The figure I've calculated for the heat transfer coefficient (500 W/(m^2.K) is lower than the values given as guidelines in Perry's (568 - 1135) and Towler and Sinnott (1000-1500) but is in the range for Incropera and Dewitt (250 - 700)

I haven't calculated the pressure drop yet, but the limit in the design specification is 0.7 bar. I'll do that tomorrow and see how that limits the number of passes. I hadn't thought about calculating erosion - could you point me in the direction of how to do this? At these velocities I think it'll be pressure drop that is limiting rather than erosion.

Thanks again for your comments!

 

[georgeverghese]

3000w/m2/degK = 530btu/hr/ft2/degF - okay for fouling service cooling water - you'll have to manage tubeside fouling anyway with cooling water - is this open loop cooling with cooling towers - if so, you should make provisions for considerable fouling in these tubes - what dia tubes have you got?

Hypo dosing in place? C106 copper tubes - is this 90:10 CUNi? What is the tubesheet made of and how are you managing corrosion issues due to dissimilar materials here?

5pass on tubeside sounds odd for the condensor, and mechanically not sound. Why not 4pass or 6pass - that way you have a regular U tube bundle.

You havent said much about the subcooler..

Cooling water exit temp from the subcooler is 21degC, while process side exit is at 20degC - presume the LMTD Ft value here is still 1.0.

You seem to working on some constraint on cooling water supply, which is why you've got these low velocities / more passes and re using this CW on the condensor also? By hooking up CW in series for these 2 units, pressure drop becomes critical if you cannot get it back in to the main CW return header. And pressure drop calcs are always for clean service conditions only; so tubeside CW flow will drop a lot as the tubes foul up. Typical provisions for fouling only include a degradation in film htc, but should also account for lower water flow due to higher corresponding dp.

Process controls to manage these 2 units needs to thought out very carefully - you'll a dPIC to bring some warm process gas across to the subcooler to prevent the vapor space pressure here from collapsing - how are you transferring this subcooled liquid out of the subcooler?

 

[MJCronin]

I believe that, unless there is an extremely tight space, the choice of copper tubes is a bad one.

You are putting the HX at risk considering the maintenance of purity of the cooling water loop. Cooling tower water can change in pH alykilinity etc. Just trace amounts of ammonia will cause pitting and cracking.

Although it will make the HX slightly larger, 304/316 tubes are a much better choice for a condenser with Cooling Tower water.

IMHO....

MJCronin
Sr. Process Engineer

 

[tmengineer]

Georgevergese,

The internal diameter of the tubes is 17mm. C106 copper is basically pure (99.85%) and the tubesheet will be made of brass (Cu + Zn) alloy. The use of copper is due to a process constraint; the condenser + subcooler is being used in a whisky distillery where the contact between the copper and spirit is said to be important (as it catalyses the production of esters and precipitates out sulphur). The odd number of passes is for aesthetics as there will be tourists visiting and so that it looks nice when it's piped in. I don't have much control over these parameters, I've just got to do the thermal design and make sure that the design is within pressure drop requirements. We're not going for a U tube but an AEL type - the industry standard here.

Focusing on the condenser for now as the subcooler design isn't that far down the line, and there are still discussions as to whether it should be a shell and tube or plate heat X. At the minute the Ft for the Ft for the condenser and subcooler is 0.95 and 0.78 respectively. Once I've got the thermal design and pressure drop sorted for both units then I was going to optimise then design.

The P&ID I'm working to is from another engineering company involved in the design and the cooling water is taken away on a return line without a pump so I think that's where the 0.7bar pressure drop restriction comes from.

MJCronin,

As I said above, the use of Cu is tied due to the process. Thankfully we have a good supply of water from a spring and will not need to use a cooling tower, but I'll bear your comments about ammonia and Cu in mind for future.

Thanks for the comments! I think I'm basically at the stage where I need to crack on with the pressure drop calculations.

 

[georgeverghese]

Process controls would be much easier if you had condensing and subcooling all in one shell, but it looks like you've got CW supply constraints that prevent this.

Dont know about aesthetics, but I'd check with your maintenance folks how these tube bundles are holding up to differential expansion between shell and tube bundle on the TEMA L type head.

Didnt know we need a catalyst for this esterification reaction - from memory, this happens quite readily without any catalyst if the solution is warm enough ( EtOH + HAc = Et Ac ?)

Good luck.

 

[MJCronin]

Well, now at last we have the full story.

I am sorry for my misunderstanding.... I have never been involved in a HX project where appearance trumps thermal function.....

You should tell the project manager that, since the choice of tube materials has been taken away from the engineer by the architect ( or, designer, interior decorator or whatever) he should hold that same person responsible when the HX fails, leaks or becomes extremely galvanically corroded in a couple of years.

You state that the use of copper tubes is required because it "precipitates out sulphur" ..... isn't that also called HX fouling ? ??

Your HX thermal and pressure drop calculations will be short lived,.... won't they ????

Of course, all of these problems won't exist with FDA SS materials

MJCronin
Sr. Process Engineer

 

[georgeverghese]

Something tells me this HX many be thermally more efficient if we switched sides - put CW on the shellside and alcohol vapor on the tubeside. If there is a concern with cleaning the shellside, you could use square rotated pitch (say 3/4inch tubes on 1.0inch sq rot pitch or 1inch tubes on 1.25inch sq. rot pitch). You would get (a)more water flow for the same current dp and (b) shellside htc is much more than if you placed it on tubeside.