19 replies to this topic

[NoobMi]

If let say i have two heat exchanger, E-1 and E-2 (performing the same function e.g. cooling water with brine).
E-1 has a twice the area of E-2 but having twice the flow of water and brine (for both its shell and tube side). Other properties of the 2 heat exchangers all remaining the same.
 
Will this difference in area affect the heat transfer coefficient or heat transfer coefficient is independent on the area of heat exchangers, but only a factor of the liquid properties and materials of heat exchanger?

From what i understand:
1/UA=1/(ho*Ao)+dx/(k*A)+1/(hi*Ai)
so only when the area Ao and Ai is the same as A, will the U be constant despite the increase in Area of heat exchanger. But how true is it for real-life cases?

Edited by NoobMi, 13 February 2013 - 01:24 AM.

[Ahsan67]

NoobMi

hio and ho are functions of mass velocity and physical propertiesof the fluid; if these(mass velocity and physical properties) value varies
there heat transfer coefficient will vary, if these(mass velocity and physical properties)stands same in both exchanger there heat transfer coefficient will be same ,that increase in area may be given to get the same mass velocity , Nusselt number,prandtl number etc as in E-2

Regards
Ahsan
(sharing is a procedure of learning)



 

[Art Montemayor]

No!  The Heat Transfer Coefficient is NOT Dependent On the heat transfer Area Of a Heat Exchanger.

 

It is the other way around: The heat transfer area of a heat exchanger is dependent on the heat transfer coefficient.

[markymaark]

This may be a nomenclature issue.

1)hio and ho are the film heat transfer coefficients.

 

2)Your equation is to find U, the overall heat transfer coefficient.

 

3)hio and ho are not dependent on the area (Though some of the dimensionless numbers given by post #2 have reliance on geometry)

 

4)U is a function of Area.  BUT, as post #3 says, in "real-life" you would find the U needed for the correct amount of heat transfer, THEN find A.  So technically A is a function of U.

 

Mark

[NoobMi]

Thank you all for the reply.

 

To add on to my question:

My heat exchanger is actually to cool organic vapor using brine. Organic vapor = tube side, Brine = shell side

Vapor inlet = 150 degC

Vapor outlet = -10 degC

Vapor dew point = 30 degC

Brine Inlet =-20 degC

Brine outlet = -15 degC

 

Because of the phase change occuring within the heat exchanger itself, so should my convective heat transfer (h,inner) for my organic vapor depends on the liquid or vapor phase when doing a prelim design of the heat exchanger?

Or are there any correlations which i can refer to as i will need to do a hand calculation instead of using simulator for this project.

 

Thanks again.

[breizh]

NoobMi ,

Here you are talking subcooling , the outlet of your HX will be liquid ( phase will change from gas to liquid). You may find good info in the Forum (search ) or reading book like Process Heat transfer by Kern .

Note : Google is always an option

http://www.wlv.com/p...ok/databook.pdf

Hope this helps

Breizh

Edited by breizh, 17 February 2013 - 01:24 AM.

[kkala]

Probably this is a rough preliminary design for cost estimating.

Assuming single component organic vapor, I would first try simulating three exchangers in series

1st, cooling the vapor from 150 oC to 30 oC (i.e. to its condensation temperature for the operating pressure)

2nd, condensing the vapor (at 30 oC)

3rd, cooling condensate from 30 oC to - 10 oC.

Organic fluid in tubes  and (counter current) brine in shell, for all three exchangers.

Sum of heating surfaces may not be far from  heating surface of a single exchanger doing the whole function in counter current flow (care is needed in fouling factors selection). I think heating surface is the principal parameter for capital cost estimate. Cases of single exchanger deviating from pure counter current flow could be investigated later.

Hand calculation for the virtual three exchanger system is possible, yet simulator use is much more convenient. We did the latter for three real exchangers (in series) cooling waste water of a refinery desalter.

Probably there is no simplifying correlation for the overall heat transfer coefficient of this single exchanger; in the past we would be instructed to virtually "separate" the exchanger into mentioned three pieces, which is more clear in the simulation.

[sheiko]

NoobMi,

- At design stage (before the exchanger is built), the surface area A is calculated based on an estimated overall heat transfer coefficient U.
- During operation (the design is defined, the exchanger is built and in service), U will only depend on flowrate, physical properties and fouling (tube-side and shell-side).

Edited by sheiko, 17 February 2013 - 11:34 AM.

[katmar]

In your first example, where you are using water and brine with no phase change, if you have geometric similarity and hi and ho are the same for the two exchangers then the overall heat trasnfer coefficient will be the same for both exchangers.

From your equation 1/UA=1/(ho*Ao)+dx/(k*A)+1/(hi*Ai) you can see that

1/U = A/(ho*Ao) + dx/k + A/(hi*Ai)

If you double the area A then you will also double the areas Ao and Ai and you have

1/U = A/(ho*Ao) + dx/k + A/(hi*Ai) = 2A/(ho*2Ao) + dx/k + 2A/(hi*2Ai)

But this is not necessarily true for your second example where you have a phase change.  As kkala has pointed out, you need to calculate this as though it were 3 separate exchangers, even if physically it is all in one unit. In this case you cannot speak of a constant hi for the whole exchanger.  You can work out an average hi for the exchanger, but it will have to be weighted according to the area of each zone (gas cooling, condensing and liquid cooling).

 

If the ratios of these 3 sections are the same for your 2 exchangers then the average hi will be the same and then you can say that the overall heat trasfer coefficient is the same. But you will have to check your geometry to confirm that the ratios are the same and that the hi's for the corresponding zones are the same.

[NoobMi]

NoobMi,

- At design stage (before the exchanger is built), the surface area A is calculated based on an estimated overall heat transfer coefficient U.
- During operation (the design is defined, the exchanger is built and in service), U will only depend on flowrate, physical properties and fouling (tube-side and shell-side).

If let say i am interested in the design stage, how do i actually get the overall heat transfer coefficient for my system? Got quite confused as several parts requires Area.
And also will a delta T of 5 degC possible for cooling of organic vapor with brine? i believe normal rule is delta T of 10 degC, not sure if there are exceptions.

what i did:
* Q=U.A.LMTD - i have Q and LMTD but not U and A.
 
1) Split into the 3 parts as mentioned by kkala and katma. [thank you for the suggestions ]
2) Get all the physical properties required for the calculations (Cp, viscosity, density, thermal conductivity, etc)
2) Find the Reynold's number for brine and organic vapor (for this i need the velocity so do i assume a diameter of the tube and shell for the flow of my fluids and the number of tubes??)
3) Find the Prandtl number
4) Find the Nusselt number - using Sieder-Tate (for laminar) and Dittus-Boelter (for turbulent)
For condensation - using Coluburn's-Kirkbride film condensation theory (turbulent) or Nusselt's analytical method for laminar film condensation
5) With Nu found i can get my h for the inner tube and outer tube.
6) With that i will need to find my overall U (but again area is needed - so how do i proceed? Do i based on the calculated UA and LMTD to compare against my current Q and re-iterate by changing the area?)
 
Need some advice on my directions. Please point out my mistake if i am not proceeding in the correct way.
 
*didnt managed to get a copy of Process Heat transfer by Kern, but is currently using Fundamentals of Momentum, heat and mass transfer by Welty/Wicks/Wilson/Rorrer. Hope it is sufficient?

Edited by NoobMi, 20 February 2013 - 08:55 AM.

[thorium90]

WWWR is a good book too, its got all those equations you need, including those correlations and graphs. I've added some on wikipedia if you do not know where to look.

[Shivshankar]

NoobiMi,

http://nptel.iitm.ac...es/103103027/16

Check above link.

Regards
Shivshankar

Attached Files

•  Process and Mechanical design of Heat Exchanger.pdf   2.95MB   69 downloads

Edited by Shivshankar, 24 February 2013 - 05:01 AM.

[kkala]

I understand that this is preliminary design and you can use a simulator. Indeed the task is not simple, but simulation can make it easier, loosing quite little insight in this case. I would try following (partially applied, comments welcomed).

1a. Adopt  overall heat transfer coefficients Us from literature (including Perry) for the three virtual exchangers (true counter current flow, post no 7). These would be rough approximations.

Fouling factors have to be included in mentioned Us. This can be uniform in the external tube side (brine), advice in the internal tube side would be welcomed.

1b. Run the simulation using above Us, resulting in surface area A for each exchanger, as well as inlet - outlet temperatures.

2. For more precise results, estimate fluid properties at average (inlet-out) temperatures for each exchanger, assume a tube bundle configuration (same for all three, only individual length will be different according to heat exchanged and adopted As per 1b) and calculate individual Us. Then go to 1a to repeat procedure.

Note: velocity in shell side (brine) had better approach 1 m/s to avoid excess scaling, though it may not be easy <http://www.cheresour...eat-exchangers/>.

Note': ΔT in vapor condensation section will be about 47 oC. Critical temperature drop for boiling is often 22-28 oC  (McCabe - Smith - Harriott, "Unit Operations of Chemical Engineering", Maximum flux and critical temperature drop), not sure whether applicable to condensation. Nevertheless see <http://www.nzifst.or...httrtheory8.htm>.

Note": Approach of 10 oC (brine=-20 oC, liquid=-10 oC) seems adequate, <http://www.cheresour...ature-approach/>.

3. Stop procedure when differences in individual As get insignificant  (say less than 5%). No need for precise estimate, fouling factors are uncertain, bundle may eventually change, etc.

4. Have you looked into passing brine from tubes due to scales? Tubes are more easy to clean. Above procedure would be similar.

Note: In detail engineering a "precise" form of exchanger will result from some software (e.g, bjac), or vendor.
 

Edited by kkala, 21 February 2013 - 07:04 AM.

[NoobMi]

Thanks all for the suggestion and advices.   I have currently some difficulty in finding the reynold's number for the shell side to determine the heat transfer coefficient for my brine.   I have attached the necessary information in the excel file.   My calculated Re number is only 700+ which i think is a bit too low and most probably wrong. Appreciate if anyone can help to point out the mistakes.   My reference for my calculations was based on: http://web2.clarkson...s/shelltube.pdf  

edit: sry forgot to include attachment

Attached Files

•  Reynold Number for Shell Side.xlsx   9.82KB   28 downloads

Edited by NoobMi, 24 February 2013 - 08:50 AM.

[breizh]

Post your calculation , someone will help!

 

Note : I've added a power point presentation to support .

 

Breizh

Edited by breizh, 24 February 2013 - 05:56 AM.

[NoobMi]

Post your calculation , someone will help!

 

Note : I've added a power point presentation to support .

 

Breizh

 

Attaching my calculations for advices.

Attached Files

•  Initial Calculations.xlsx   29.56KB   50 downloads

[kkala]

I was writing a response to post no 14, when post No 16 came with revised spreadsheet. Following concerns brine in tube site.
Spreadsheet in post no 14 did not consider "surface area" in the calculation of Re, so resulting value was low. This was corrected in "Initial Calculations.xlsx", where clearence=5.95 mm was also noted. According to it, surface area Sm= 0.009639 m2, which gives velocity for Re (combined with brine mass flow = 22000 kg/h).
Perry, Heat transfer equipment (Section 11 in the 7th edition), thermal design of heat-transfer equipment, would calculate Sm as below.
pitch (triangular)  p' = 25E-3 m
baffle spacing ls = 90E-3 m
tube outside diameter Do = 19.05E-3 m
shell inside diameter Ds = 45E-2 m
shell outer tube limit Dotl = Ds - clearance (as understood) =0.45 - 0.00595 = 0.44405 m
Sm = Is(Ds-Dotl+(Dotl-Do)/p'(p'-Do)) = 0.09*(0.45-0.44405+(0.44405-0.01905)/0.025*(0.025-0.01905))=0.009639 m2
Thus OK, calculated surface area complies with Perry's. I have calculated Sm for checking (reference in post no 14 and spreadsheet calculation were unclear to me).

Edited by kkala, 25 February 2013 - 09:34 AM.

[kkala]

"Initial Calculations.xlsx" indicates a diligent work, to the extent that I can understand.

1. Consider fouling factors in shell and tube side. Investigate brine passing from tubes, now or later (post no 13).

2. Critical heat flux does not apply in condensation, only in boiling; this is understood through reading  relevant books. Confirmation is welcomed.

[thorium90]

Critical heat flux is explained in lotsa detail here

http://en.wikipedia....ucleate_boiling

http://en.wikipedia....tical_heat_flux

[kkala]

Thanks;  and critical flux seems to concern boiling only, not condensation.