Continuous Monitoring of Distillation Exchangers with Fluctuating Rates

[AlbusTheEngineer]

Hello All!

I'm creating monitoring tools for our distillation columns and I am trying to determine the best way to monitor the Condensers (CW and Propane Refrigerant) and Reboilers (75# IPS and 25# LPS) on multiple towers for fouling. From school, my understanding is a trend of the UA curve would let you know at what rate your Overall Heating Coefficient is degrading. However, when I graph the UA curve it is constantly jumping around due to our constantly changing feed rates into these towers. I am currently trying to graph the steam valve % Open vs tower feed to determine if the Reboiler is degrading, but it seems very "Hand-wavy." What is generally used to continuously monitor these types of heat exchangers with constantly changing rates? Any other tips you have I would greatly appreciate

Note: All are S&T exchangers, counter-current. Attached is an example that comes out of my current Reboiler graph that I made, that I can't really make any sense of.

Have a wonderful day and thank you for your insight!

AlbusTheEngineer

 

[don1980]

It's easy to monitor fouling when the heat load is relatively constant. In such cases the increasing pressure on the condensing side (that is, increasing dT) is a direct indication of fouling. When fouling occurs, the U decreases, and this has to be offset by increasing the dT (increased condensing side pressure). As long as the pressure (dT) can continue to rise, then the exchanger will continue to perform at the desired Q. However, eventually the pressure (dT) can no longer rise because the control valve is fully open (pressure in the exchanger is the same as the supply pressure). At that point you either have to tolerate running at a lower rate, or shutdown the exchanger for cleaning.

Similarly, the condensing side pressure (dT) is a direct indication of load (Q) as long as the U is constant (fouling isn't occurring).

So, dT (condensing side pressure) varies with load and fouling. If both are occurring (fouling and load changes) then dT can't be used to distinguish between the two.

 

[AlbusTheEngineer]

So in this case the heat load is not constant, as we are constantly changing our feed rates in and out of these towers. I apologize if that wasn't clear. What I want to ensure is that if we are running at a lower rate (lower heat load) and we want to come back up to our max load, that we will have enough Heat Exchanger Capacity and aren't degrading past the point where we either run into a heating issue (Reboiler maxed out) or don't have enough condensing duty.

 

[don1980]

In order to distinguish between load changes and fouling, you need some performance data for a clean exchanger. Specifically, you need to know how the dT varies (condensing side pressure varies) with load (Q). U = Q/(dT*A). "A" can be disregarded since it's constant, so we can say that U is directly proportional to Q/dT. Of course, to generate a Q/dT curve you'll need the necessary instrumentation to measure Q. Assuming you can do that, gather dT and Q data over a range of loads (using a clean exchanger), and plot that curve. As fouling occurs, the observed ratio Q/dT will decrease. Tracking that decreased Q/dT ratio is a direct way to monitor the fouling as it progressively gets worse. You'll know when you reach the point where you can't maintain the load because of fouling, and the Q/dT value that corresponds to that point is your "red-line". In actual practice, this red-line is readily apparent by just monitoring the opening position of the control valve. Once the valve approaches the full open position, the exchanger is too fouled to deliver the needed Q.

 

[georgeverghese]

This is a common problem for many operations engineers trying to keep a handle on HX performance. The reason you arent getting reasonable results is due to the clean U value changing as flows decrease on both sides of the HX.

If you've got the process / thermal datasheet for each of these units, try to compute the clean U value (or the design case h values with given values of fouling coeff for each of tube and shellsides) for the HX for a range of process side flows and corresponding utility side flows, assuming you have a fixed target temp for the process side temp for all operating flows. Then compute overall U. To gain some confidence on the calculation routine, first see if you can match the design case overall U for the design case flows stated on the thermal datasheet. Dont forget to take into account tubeskin temps and how film coeff changes with viscosity, where applicable. Finally compute the clean UA value for each case given this flow adjusted clean value of U. Then compare with the field derived UA value.

What may make data reconciliation more difficult is when temperature transmitters show considerable response lag ( due to the thermal inertia in thick wall thermowells) to actual process temp for transient process / utility side flows. So you should only jot down process parameters when stream temperatures on both sides have had some time to stabilise.

 

[georgeverghese]

The tedium of deriving flow adjusted values for clean U in a HX may be considerably reduced if you have access to a process simulator which can rate an existing HX when all required configuration and operating parameters are filled in - one example is the HX rating routine in Pro - II from Simsci.