8 replies to this topic

[sheiko]

Hi,

I usually advise people to use the following equation to monitor fouling factor:

fa = fd + 1/Ua - 1/Ud, with

fa: actual fouling resistance
fd: design fouling resistance (from heat exchanger data sheet)
Ua: actual overal heat-transfer coefficient
Ud: design overal heat-transfer coefficient (from heat exchanger data sheet)

However, I now realize that Ua vary significantly with process conditions (temperature, pressure, flowrate...), and that actual and design conditions are always different.

So my questions are:
How is it possible to accurately monitor fouling factor, given that the actual process conditions are continuously changing versus the design process conditions?
In other words, won't the difference (1/Ua - 1/Ud) be equal to the SUM of:
1- the difference between actual and design fouling factor (fa-fd)
AND
2- the change of 1/Ua due to the change of actual process conditions wrt design conditions?

Edited by sheiko, 15 January 2012 - 01:15 PM.

[S.AHMAD]

1. I was on the same boat as you before. That was the time where I used slide rule.
2. Two years ago, I helped a process plant (ammonia/urea) develop computer program for exchanger monitoring (water coolers) using EXCEL spreadsheet.
3. In this program, to compute the fouling factor we compute the individual heat transfer coefficient (using actual flowrate and temperatures) to arrive at the U clean and the actual U from Q = UALMTD and hence calculate the fouling factor using these figures.
4. With speed of computing, above program run very fast. The longer time require is for obtaining the plant data.

Edited by S.AHMAD, 15 January 2012 - 09:56 PM.

[DB Shah]

As said by S Ahmad, we too were using the method you mentioned couple of years back. Now we rate individual exchanger in HTRI to find out Rd.

If you are using HTRI, HTRI has a feature "R Trend" where you can link any one file wtih one R trend (xls) file. With excel as front end and HTRI at back end, it calculates, tabulates and gives various graphs for the number of readings specified (ie you specify to run "From date" to "To date"). R trend file is available frm HTRI support. I am not uploading the file as I am not sure if it is an open source or binded by copy right.

Alternate method is as suggested by S Ahmad by acutally calculating hi/ho in excel.

[kkala]

Your formula seems to assume that (clean) heat transfer coefficients of shell side and tube side remain as designed (always have their design value). This can be approximately true in some cases*, even if conditions (stable, not transient) are changing. Then you see when fouling exceeds the design value.
Of course mentioned software, to calculate heat transfer coefficients and fouling out of actual data, will give much more realistic results.
It is noted that scale formation in evaporators can be monitored by plotting (1/Ua)² versus time, giving a straight line. Extrapolation can predict when Ua is to get a specific lower value. Probably mentioned R trend does something similar in a more precise manner, not limited to evaporators.
A presentation of mentioned plot for evaporators can be seen at "Introduction to Chemical Engineering" by W Badger and J Banchero, McGraw-Hill 1955, Chapter 5 Evaporation, p. 230, Fig 5-30. A theoretical justification can be found in "Principles of Unit Operations" by A Foust, L Wenzel, etc, Wiley 1960, Chapter 19, p. 370-371.

* Especially if clean Ua is adjusted once to actual performance data.

Edited by kkala, 16 January 2012 - 10:41 AM.

[GS81Process]

I use measured temperature and flowrates for the heat exchanger to calculate the Uservice value. This requires both flowrates and at least three temperature values (the fourth can be determined by exchanger energy balance).

I calculate the fouling resistance,Rf, based on the equation:

Rf= 1/Uservice - 1/Uclean

I like to determine Uclean from operating data after the exchanger is cleaned (using the same method as for Uservice described above) rather than from the datasheet. I find that this Uclean value can sometimes differ from the datasheet value because the process in general is being run differently than as specified for the datasheet.

Edited by GS81Process, 16 January 2012 - 02:04 PM.

[sheiko]

Thanks all.

@S. AHMAD,
Your method is rigorous but I believe the results will be as precise as the fluid physical properties measurements (thermal conductivity, viscosity, density, ...).
May I know however which correlation do you use to calculate the individual convective heat transfer coefficients?

@GS81Process,
I like your method: simple and sensible.

Edited by sheiko, 16 January 2012 - 09:20 PM.

[S.AHMAD]

Sheiko

As attached for your perusal

Attached Files

•  HEAT EXCHANGER PERFORMANCE MONITORING_R1.pdf   56.39KB   274 downloads

[breizh]

Hi ,

It may help to maintain the performance of HX :

http://heatexchanger...6_Krueger_F.pdf


Breizh

[guillermolineroa]

Hi all,

(At first you'll excuse the grammar with which I write because I do not speak the language perfectly and I am supporting the google translator)

Take the matter under discussion to comment on my perception of the assessment method that I use in heat exchangers shell and tube, as recently I had to make an assessment of 345 exchanger tubes and 36 tubes clogged by having low thickness , then determine if he could continue working with the same load of the plant. (Interstage cooler)

The required area is considered to be solely dependent on the process conditions when fixing the clean transfer coefficient (Ulimpio), this is done for starting a heat exchanger design, however, when the exchanger has designed, is a dynamic calculation which not only depends on the process conditions, is affected by the fixed parameters are: dimensions of the exchanger and fouling factors.

What I see, and I support the theory is that by reducing the number of tubes, the exchanger has less area and the overall transfer coefficient increases dirty to absorb the loss of area, it also increases the local coefficient transfer tube side ( htubos) as this is directly proportional to the velocity of flow within the same elevated to 0.8 (as scaled the area through which the gas, the speed will increase), the calculation method does not consider htubo estimate using empirical expressions, such rate as calculated with the other data we have and solid results (as mentioned below)

Then I show the method used to discuss what I see fit:

The area required to call you Arequerida the calculation using the following equation:

Arequerida = Q / (Ulimpio * DT)

Conditions that remain constant are: the thermal load and DT, however, I need to estimate the coefficient of heat Ulimpio (This is where the number of tubes affects the calculation of Arequerida to the same process conditions), for this use the following equation:

Ulimpio = 1 / (1 / (htubos * (di / do)) + 1 / hcoraza)

Where:

* - Say, inner diameter of the tube.
* - Do: pipe diameter.
* - Hcoraza: local transfer coefficient of shell side, this ratio correlated with what I consider applying for the conditions we use, and is unaffected by changes in the number of tubes or temperatures, depends on the dimensions of the shell The organization of the tubes (pitch) and the flow of cooling water.

* - Htubos: local transfer coefficient of tube side, this ratio as calculated using the equation of the resistance to transfer, as I know all the conditions, they represent them in the following equation:

htubos = 1 / ((di / do) * ((1 / Usucio) - (1 / hcoraza) - Rdtotal - Rk))

Where:

* - Rdtotal: Resistance to heat transfer associated with fouling (fouling factor), this parameter is the amount of fouling in the tubes but the impurities in the shell and consider the design value.

* - Rk: Resistance to heat transfer associated with the conductivity of the material, this introduced an error rate, because it made me hard to find the specific conductivity of the material we are working, as the conductivities between the different materials used does not varies considerably, then kept the value of the conductivity of the shell. (May be an improvement point in the evaluation of the heat exchangers).

* - Usucio: corresponds to the overall heat coefficient is calculated with the conditions established in the system, this parameter affects the calculation of Arequerida as to change the number of plugged tubes, amending Area available (Adisponible) and this in turn generates an increase in Usucio that the chain of calculations described earlier, a direct impact on the Arequerida, the effect is seen in the following equation to calculate the Usucio:

Usucio: Q / (Adisponible * DT)

Where:

* - Adisponible: reports directly to the number of tubes and is calculated using the expression:

Adisponible = Npt * Nt * P * do * Lt

Where:

* - Npt: Number of tube passes.
* - Nt: Number of tubes
* - Lt: length of the tubes.

ADDITIONAL COMMENTS

The way that the required area remains constant for the same process conditions by changing the number of tubes is that it iterates Rdtotal parameter (fouling factor) to obtain the required area is desired, in this way could know as increased fouling, the explanation is the following: to clog a tube increases the velocity of flow of gas (as mentioned above), this increases the mass transfer coefficient of film tube side and therefore the area I require decreases (increases area in excess), here the parameter that limits clogging of pipes should be the pressure differential therein, however, considering the pressure that is handled in the process side, the pressure drop not limit the operation of the equipment (a unless very large) and the compressor is adapted to the conditions, it is likely that the overall system efficiency decrease slightly, but you can work with these conditions to change the exchanger, we are helped by the fact that the exchanger has a overdesign so great.

Another detail to be seen in the results that are important is the relationship between the local coefficients of transfer tubes and shell side, this indicates that controls the transfer, and it is vital that the control transfer process and not the side Water cooling at all times the water side coefficient must be greater than the coefficient side tubes, this parameter must be careful. Evaluate the condition in which covered a number of tubes until they are equalized local transfer coefficients, obtaining the following table:

84310.00 process mass flow kg / hr
Thermal load 7,538,368 kcal / hr
Percentage of Load 100%
Process inlet temperature 165.00 ° C
Process outlet temperature 50.00 ° C
cooling supply temperature 35.50 ° C
Outlet cooling temperature 48.50 ° C
Mass flow - Water Cooling 1,275,350 lb / hr
578,494.92 kg / hr
Thermal load 7,538,368 kcal / hr
Global transfer coefficient
Clean 4,111 kcal / h m2 ° C
Dirty 1,265 kcal / h m2 ° C
Dirty 1,265 kcal / h m2 ° C
Transfer area required for 44 m2
Available area in exchanger 143 m2
Area 99 m2 used for dirty

Corrected dtml 41.59 ° C

Local coefficients
Tubes 9,391 kcal / h m2 ° C
Coraza 9,391 kcal / h m2 ° C
Hc / ht 1.00
Fouling factors
Tubes Rd 0.000400 kcal / h m2 ° C
Shell Rd 0.000100 kcal / h m2 ° C
Total Rd 0.000500 kcal / h m2 ° C
Rk 0.000047 kcal / h m2 ° C
Percentage of excess by fouling 69.22%
Number of tubes in service 196
Number of plugged tubes 158
Speed ​​in the tubes 619.58 ft / s
3491816.85 mass velocity lb / hr ft2


Highlight that should cover 158 tubes to reach that condition and the heat exchanger can do their job, but in this case it is convenient to calculate the DP to see that both increased and set a value to limit the clogging of tubes. I calculate the parameter of differential pressure, but I'm looking for the data I require: gas viscosity and other physicochemical properties.

I hope the information provided here will be of some

PS: If you have information in Excel to estimate the viscosity of a gas stream, would appreciate the information greatly.

Regards,

Guillermo Linero:.