train systems that may have up to 60 exchangers excluding product and pump-around
coolers.
With the right tools it may be possible to manually adjust exchanger outlet temperatures to come up
with acceptable heat balances across the preheat train system; however, this is not a trivial undertaking
due to the complexity of interactions between exchangers. The final solution is also subjective as it
involves a set of sequential decisions as to which exchanger to start with and whether to adjust the
tube- or shell-side temperature in order to achieve a heat balance. There are many possible heat
balance solutions. The question is: Which is the most representative solution?
Mathematical data reconciliation can effectively answer this question. It is a vital
component of any successful fouling monitoring application. Reconciliation is a least squares
optimization process where the objective function minimizes the deviation between reconciled, heat
balanced data and the raw plant data. Temperature, fluid flow, and flow split data can be included in the
data reconciliation model. The relative weight the reconciliation model places on different data types is
governed by assigning trust factors or typical errors to each plant measurement. Occasionally, though
not often, it may be necessary to include local exchanger bypasses in the reconciliation problem.
Given that local bypass flows are never measured, bypasses must be estimated and thus very low trust
factors (or high typical errors) should be assigned to these variables.
A measure of “acceptability” should be established for any optimized solution, to determine whether to
accept or reject the current data set. If reconciled plant data deviations on a particular plant
measurement are consistently higher than expected instrumentation errors, the flow meter or
temperature indicator should be checked.
Heat Exchanger fouling monitoring [10], read more on [8] [9]
To track the amount of fouling in an exchanger one can calculate Uo frequently (weekly) and plot Uo versus
time. The pattern of the fouling curve will usually repeat itself after each cleaning. 1/Uo = R= RCLEAN +RDIRT
If conditions have changed significantly from design condition, h i and ho can be adjusted to current
conditions using the equations from the Natural Gasoline Processing Handbook summarized in
Key Formulas. RCLEAN can then be recalculated. The current RDIRT can be compared to the design RDIRT.
Many heat exchange services are regulated by instrumentation to provide a stream to a downstream
process at a consistent temperature. This temperature is accomplished by bypassing part of the
exchanger feed around the exchanger and mixing this bypassed material with the exchanger effluent.
The degree of bypassing is regulated by a temperature controller which senses the downstream mixture
temperature. When the exchanger is clean, the maximum amount of feed is bypassed; and when the
exchanger is fouled to the maximum, no feed is bypassing the exchanger. By recording the clean
bypassing condition (percent of feed bypassed), you can easily estimate the amount of exchanger
fouling by observing the amount of feed bypassed with the partially fouled exchanger.
The following equations which can be readily derived from the
above and are often quoted in the literature:
Fraction dirty = rn/rd = (Ud / UN)[UC – Un ] / (UC – Ud )
where:
rd = RDIRT design
rn = RDIRT current (now)
UC = Clean U with no fouling factors
UD = Design U which includes fouling factors
Un = Current U (now)
