Surface Condenser Design (TEMA Type BEM) 1

[MHS1988]

Hello everyone,

I’m a mechanical engineer working for a vendor that designs and fabricates heat exchangers. While we primarily focus on mechanical design, clients sometimes ask us to guarantee thermal performance as well. This usually involves minor tweaks in HTRI, like adjusting tube layouts or baffle arrangements, to align with our mechanical design.

However, a couple of days ago, a client sent us a process datasheet and asked us to provide our own thermal design. Here are the key design details:

SERVICE: Surface Condenser, TEMA TYPE: BEM, Qty.: 1
Tube Length= 3000mm, SHELL/CHANNEL ID= 1360mm, Tube OD= 25.4mm, Tube Thk.=1.65mm, Tube Pitch=35.5mm, Layout: 30°, No. Tubes: 888 (to be finalized by vendor),
Baffle: Single Seg. 30% Cut (to be finalized by vendor),

Shell Side Data:
Fluid: STEAM, Mass Flow: 16600 kg/hr, Inlet/Outlet Press.: 0.09341/- Bar(absolute), Allow. Press. Drop: By Vendor, Inlet/Outlet Temp.: 44.8/40 °C, Vapor Fraction In/Out: 1/0,
Design Press.: 3 Bar(Gauge), Design Temp.: 120 °C, No. Passes: 1, Inlet/Outlet Nozzle Size: NPS22/NPS3, Material: SS304L, C.A.: 0

TUBE Side Data:
Fluid: COOLING WATER, Mass Flow: 134309 kg/hr, Inlet/Outlet Press.: 3.1/2.6 Bar(absolute), Allow. Press. Drop: 0.5 bar, Inlet/Outlet Temp.: 32/42 °C, Vapor Fraction In/Out: 0/0,
Design Press.: 6 Bar(Gauge), Design Temp.: 120 °C, No. Passes: 4, Inlet/Outlet Nozzle Size: NPS10/NPS10, Material: SS304L, C.A.: 0

I ran the data through HTRI, but no solution was reached. As a troubleshooting step, I left out the cooling water flow rate, and the reported value in HTRI was almost 7 times higher than the specified value in the datasheet. A friend of mine (also a mechanical engineer) checked the data using HYSYS and confirmed the HTRI result. He also used EDR’s design module, which suggested 8 parallel exchangers of this size to meet the requirements.
Now I am wondering whether this kind of exchanger (a Surface Condenser) is just another story and too complex for us to handle, Or is there a possibility that the client’s provided data is incorrect?

I’ve already reached out to their process engineer, but he wasn’t much help. He mentioned that the data came from a licensor, but couldn’t provide further details.

Any insights or advice would be greatly appreciated!
Thanks in advance.

 

[EdStainless]

They have picked unreasonable criteria.
They have locked in every variable, but the thermodynamics don't work.
Those tubes are crazy thick.

Go back to his basic inputs and design a unit.
Either he will take it of he needs to go back and pay what the licensor wants.
Being cheap isn't a good option for him.

 

[MHS1988]

They have picked unreasonable criteria.
They have locked in every variable, but the thermodynamics don't work.
Those tubes are crazy thick.

Go back to his basic inputs and design a unit.
Either he will take it of he needs to go back and pay what the licensor wants.
Being cheap isn't a good option for him.

Thanks a lot.
I just needed to make sure there is something off with the provided data.
And I guess we have to present them our own design as you suggested. Whether they accept it or not is their own decision.

As to tube thickness, it's a common choice for high alloy steel among our business associates. The minimum specified in API660 is just one gauge smaller.

 

[r6155]

I don't know your activity: manufacturing? Process design and manufacturing?
Be careful with the scope of work, think about money and prestige.
Avoid conflicts and headaches.

 

[MHS1988]

I don't know your activity: manufacturing? Process design and manufacturing?
Be careful with the scope of work, think about money and prestige.
Avoid conflicts and headaches.

Well, where I’m from, they expect you to be a jack of all trades!

We’re an engineering company and manufacturer, but most of our work revolves around mechanical design rather than process engineering. Cases like this one don’t come our way very often.



BTW, Here’s an update on the exchanger:

After some trials, I’ve come up with a preliminary design of 4 parallel exchangers, each with a size of 1800 x 5000 mm (IDxL), to handle the steam mass flow rate!
We’re planning to have a meeting with the client’s team soon to discuss this matter.

 

[georgeverghese]

What Ft value ( log mean temp diff correction factor) does the simulator report for the 4 tube side pass setup?
Did you split the tubes equally amongst the passes ? What is the steam velocity at the first inlet pass?

A stable thermal design will have an Ft value > 0.8. There seems to be about 4degC subcooling of the steam condensate at the shellside exit, so this may require the last tubeside pass to be partially flooded on the shellside. Otherwise, you could install a separate HX just for the 4degC subcooling duty.

I've not heard of a TEMA BEM HX called a surface condensor, since the latter is a different beast altogether. Nevertheless a TEMA HX can be used to condense steam turbine exhaust. Single SEG baffles are not well suited for low dp on shellside, other baffle arrangements are better ( TEMA G or H shell ).

A pure countercurrent arrangement may be indicated here since there is little temp differential between shell and tubeside in normal operation. However, if the shellside were to be close to 120degC and the tubeside at 30-40degC, a BEM HX may give way at the tube to tubesheet joints. If so, go for a BEU HX to give you better mechanical integrity at 120degC shellside and 30-40degC on tubeside.

 

[EdStainless]

Make the tubes as thin as you can.
I am used to power plant condensers where we go to 0.020" commonly.
I am with George, concerned about steam velocities.
Though the SS will resist erosion high velocity can drive a lot of vibration and fatigue.
Minimizing pressure drops will likely require a different configuration.

 

[MHS1988]

What Ft value ( log mean temp diff correction factor) does the simulator report for the 4 tube side pass setup?
Did you split the tubes equally amongst the passes ? What is the steam velocity at the first inlet pass?

A stable thermal design will have an Ft value > 0.8. There seems to be about 4degC subcooling of the steam condensate at the shellside exit, so this may require the last tubeside pass to be partially flooded on the shellside. Otherwise, you could install a separate HX just for the 4degC subcooling duty.

I've not heard of a TEMA BEM HX called a surface condensor, since the latter is a different beast altogether. Nevertheless a TEMA HX can be used to condense steam turbine exhaust. Single SEG baffles are not well suited for low dp on shellside, other baffle arrangements are better ( TEMA G or H shell ).

A pure countercurrent arrangement may be indicated here since there is little temp differential between shell and tubeside in normal operation. However, if the shellside were to be close to 120degC and the tubeside at 30-40degC, a BEM HX may give way at the tube to tubesheet joints. If so, go for a BEU HX to give you better mechanical integrity at 120degC shellside and 30-40degC on tubeside.

I'm sorry, I can't find this parameter in the software report. However, I uploaded the latest design (4 parallel shells) report. I would appreciate if you could find any time to check it out.
The tubes are split somewhat equally in a mixed H layout.
The steam line works at vacuum pressure. The operating temperatures aren't even close to that 120°C. That's the specified design temperature. A vacuum pump is also connected to the shell side.

 

[georgeverghese]

a) There must be some reason why the UDT=120degC was selected. The User may insist that the HX show mechanical integrity at 120degC shellside / 30degC tubeside.
b) Okay, so you've got 4shells in parallel on shellside, 4 shells in parallel on tubeside i.e. 25% of flow on steam and feed CW to each shell
c) Presume the fouling htc you've selected for CW on tubeside is in accordance with User requirements
d) MTD correction factor =1.0 throughout as shown in the report, and that is to be expected in this case for present selection of pure countercurrent flow
e)There is something weird in this simulation result : there is no section of the HX assigned for condensate subcooling - see shellside report. The simulator does not report a shellside liquid htc at the last few points where shellside should be all liquid. It seems to have assigned this sensible heat cooling duty to condensing htc, which is wrong. There has to be one shell assigned merely to subcooling duty. Be very careful how you pipe up this subcooler HX on the shellside.
f)This is only a cursory review - a full review would take hours. But given (e), a thorough review of this whole report is not warranted at the moment for this simulation result.

 

[georgeverghese]

While you are checking out this HX simulation report, tell me if you find the tubeside htc of approx 2500w/m2/degK (440btu/hr/ft2/degF) to be correct. Its been many years since I did similar work, but I clearly dont recall htc of liquid water, whether on tubeside or shellside, running up to values as high as this.

 

[MHS1988]

While you are checking out this HX simulation report, tell me if you find the tubeside htc of approx 2500w/m2/degK (440btu/hr/ft2/degF) to be correct. Its been many years since I did similar work, but I clearly dont recall htc of liquid water, whether on tubeside or shellside, running up to values as high as this.

From the beginning I've had this dout that the mass flow rate of steam was reported incorrectly. I'm still not sure if the client is entirely confident about these numbers. However, we decided to give them our technical offer based on our own design.

 

[pierreick]

For you and others,
A memo from Art Montemayor about HX.
Good stuff
Pierre

 

[pierreick]

Hi,
For reference a list of heat transfer coefficients:

https://www.accessengineeringlibrary.com/content/book/9780071624084/back-matter/appendix7

The overall heat transfer coefficient for your application should be between 1500 -4000 W/M2 K , roughly 2 times higher than the one calculated.
Pierre

 

[MHS1988]

Hi,
For reference a list of heat transfer coefficients:

https://www.accessengineeringlibrary.com/content/book/9780071624084/back-matter/appendix7

The overall heat transfer coefficient for your application should be between 1500 -4000 W/M2 K , roughly 2 times higher than the one calculated.
Pierre

First, thank you very much for sharing these contents. That was so nice of you.
Second, from what you said about overall heat transfer coefficient, I guess this is not an efficient design, is it? Maybe to opt for another type of exchanger?

 

[EdStainless]

One thing that concerns me about using multiple shells is how you are going to remove non-condensable gases.
There will also be a lot more flow losses with multiple shells.
High tube-side flow velocities will improve cleanliness and overall heat transfer, but at the cost of higher pressure drops.

 

[pierreick]

First, thank you very much for sharing these contents. That was so nice of you.
Second, from what you said about overall heat transfer coefficient, I guess this is not an efficient design, is it? Maybe to opt for another type of exchanger?

I believe so.
The only way will be to perform simulations, this is out of my area of expertise.
Pierre

 

[pierreick]

Hi,
The point made by EdStainless about Non condensable is very valid.
A refresher about HX:

https://www.pdhonline.com/courses/m371/m371content.pdf

Pierre

 

[MHS1988]

Hi,
The point made by EdStainless about Non condensable is very valid.
A refresher about HX:

https://www.pdhonline.com/courses/m371/m371content.pdf

Pierre

A vacuum pump is considered for each shell to extract the non-condensibles. I guess that does the job.

 

[pierreick]

A vacuum pump is considered for each shell to extract the non-condensibles. I guess that does the job.

Yeap!

 

[georgeverghese]

My recollection of the overall U was more for hydrocarbon - hydrocarbon applications, so not valid in this case. Here, with water - steam, U is much higher - Perry Chem Engg Handbook 7th ed states in Table 11-3 values of U ranging from 2400w/m2/degK to 5500w/m2/degK with minimal fouling allowance. Low tubeside velocity of 0.4m/sec may be increased to get better overall U - can you increase 4pass / shell to 6pass /shell ?