Modeling 3 pass shell and tube steam surface condenser.

[chemEcaleb]

I'm working on replacing/debottlenecking a steam surface condenser that's used in our "vacuum boiler" (it's a draft tube baffle crystallizer built by Swenson). Maarky has sent me a proposal, but I just want to do some design verification to make sure this will work for us in the event that we increase production capacity by 15%. I'm 90% sure it won't be a bottleneck because we currently have 10% of the tubes plugged and it works just fine, but I want to mathematically verify. I think worst case scenario is we will have to get a vacuum pump with an increased capacity. I have no heat exchanger modeling software. I have no Aspen/ProII/Hysis... Just Mathcad. Aside from diving in to McCabe Smith & Harriot and doing some Q=UAdT calcs, what should be my approach?

Right now my idea is this: with an 15% increase in production capacity, we will have a 15% increase in steam flow rate. I can calculate the change in outlet temperature of the gas that is going to the vacuum pump and see how much higher of a volumetric flow rate that will be from the base case and see if I have excess capacity in my vacuum pump...

 

[Compositepro]

The vacuum pump is for removal of non-condensibles. This depends on your leak rates, not your steam rates.

 

[georgeverghese]

The actual U of the surface condensor in it current operating condition would be somewhat removed from that in the manufacturer's thermal datasheet. A rate check of the unit to get the actual value of U would be required at the current effective exchange surface area in order to extrapolate to the increase flow condition.

Also note that 15% extra flow will mean 1.15 ^ 2 = 1.35 x times the current pressure drop, so, for a fixed inlet pressure to the condensor, the exit pressure will be less by a factor of 1.35x, so that needs to be accounted for in the selection of the new vac pump, if this is required.

Is this condensor operate with steam and cooling water flowing in pure countercurrent flow, or countercurrent - cocurrent, or cross flow ? A sketch will help. More later.

 

[apetri]

you can model an exchanger with Mathcad (or Matlab or similar tools),
there are many resources discussing how to apply these tools for solving chemical engineering problems,
for Mathcad see for example

"

http://www.harveyhensley.com/blog-che-old-school-and-new-school"

you may consider different alternatives, for example a rigorous model (based on correlations given in Kern or similar textbooks) or some simplified model (a correlation which correlates, for example, flow vs. dP vs. heat exchanged etc.)

 

[MJCronin]

Steam surface condensers work across a range of operating pressures, massflow rates and heat transfer conditions.

You state: but I just want to do some design verification to make sure this will work for us in the event that we increase production capacity by 15%----What exactly do you mean ? An increase of 15% can be tolerated by most HXs of this type, but with an increase in back pressure. What cannot be tolerated by you ?

Additionally, you probably reject heat to cooling towers. These influence the performance of the condensers. How will these limit your maxima conditions as the seasons change ????

Lastly, there are several excellent books on condensers and performance..

http://www.eng-tips.com/viewthread.cfm?qid=169345

MJCronin
Sr. Process Engineer

 

[chemEcaleb]

Hi guys, thanks for all of your responses. Sorry it took me so long to get back to this... so busy. THIRD AUDIT OF THE SUMMER TOMORROW... but I'm going to try and dig in to this tomorrow regardless. OK here are some responses.


Compositepro - yes the vacuum pump is for removal of air, however, if you increase the flow rate of steam your outlet temperature is going to be higher (i.e. more condensables in outlet gas). Maybe that change in volume would be insignificant; I'm not sure.

georgeverghese - it's 3 pass tube side, 1 pass shell side with two longitudinal baffles (I'm pretty sure it's split flow shell side). I actually don't know what the internals look like exactly because it's "intellectual property" -.- I would attached a drawing, but I'm pretty sure Graham would sue me.

MJ Cronin - I can't tolerate an increase in back pressure because that will result in a temperature increase in the crystallization unit which will cause certain unwanted precipitation products. What that point is exactly depends on operating concentrations.

I'm going to think on this some more tonight. Tomorrow I'll dig in to my books and see what kind of model I can come up with.

 

[georgeverghese]

Okay, with steam on the shellside with 2 long baffles in crossflow, and cooling water on the tube side with 3 passes, ( which is a TEMA G shell, yes ? ), Perry Chem Engg Handbook 7th edn says Ft ( LMTD corr factor) will always = 1.0. So you can work out the corrected LMTD for a current operating case ( preferably some flow which is close to the max acceptable performance you can get from this unit).

For this current case, work out the Q based on Q = M Cp dT, for which you can use flows and dT from the cooling water side or the steam side. Then work out the UA value for this unit from

Q = UA. LMTD. Ft, where LMTD = t2-t1 / {ln[(Ts-t1)/(Ts-t2)]}, where t = cooling water temps ( in / out), Ts = condensing temp for steam, and Ft = 1.0

For the new case, refer to DQ Kern - Process Heat Transfer, pg 306, fig 12.28 for the new value of U at the higher steam rate, then find a way to derive the new value for UA given that A is invariant. (Suggest we get a pseudo value for the current case from Fig 12.28, and a value for the new case, then UA new = (Unew / Ucurrent) x UAcurrent)

Check how much cooling water you can push through this unit for the new case, which will be determined by the current dp across the tubeside and the CW pump delivery pressure.

With the new value for UA, work out a few trials for the 2 following equalities so that all the values are converged for the new cooling water rate, new value for Q (will be obviously be 15% more than the old case)

Q = M Cp dt and Q = UA . LMTD. Ft. - This will then tell you what the new steam rate will be.

The procedure in DQ Kern for a similar surface condensor re rating exercise (on page 308) assumes that steam subcooling duty is minimal in comparison the condensing duty and is hence ignored.

So what will you do with the 35% higher dp you will now incur with higher steam flow ?

 

[georgeverghese]

Just realised from the drawing in Kern that this unit isnt a TEMA G shell, but nevertheless, Kern says also that for this crossflow arrangement, Ft = 1.0 as in pure countercurrent flow.

 

[chemEcaleb]

I'm not worried about higher dP at this point. Mainly focused on the condenser as a bottleneck.

I think it is a TEMA G except with double split flow since it has two longitudinal baffles. I wish I could give you guys drawings, but like I said I'm worried about intellectual property violation. Attached is a photo of the tube sheet. That will give you an idea of what's going on with this heat exchanger.

So what I've done so far is calculate the Q from the latent heat of vaporization from the condensate flow rate. Then I back calculated my overall heat transfer coefficient from U*A*LMTD and I got the exact same heat transfer coefficient as the manufacturer... They got 246 I got 245 BTU/hr*ft^2*F. That's a good sign. For a 15% increase in steam flow rate I will need to increase this to 283.5 BTU/hr*ft^2*F.

I currently have limited control over the chilled water flow rate (I have room in the volute to change the impeller diameter), but I do have control over the chilled water temperature.

I need to think about this more. I'm going to make a sketch for you guys and define the problem statement a little bit more. I'll get back to this tomorrow.

 

[georgeverghese]

Does look like a regular S/T HX, so agree this would be a TEMA G shell.

Yes, if you could drop the chilled water temp and increase the LMTD, that will help . You could then split out the required increase between the increase in U due to increase in steam rate and the increase in LMTD to get an overall 15% increase in Q.