20

th
European Symposium on Computer Aided Process Engineering ESCAPE20
S. Pierucci and G. Buzzi Ferraris (Editors)
2010 Elsevier B.V. All rights reserved.
Effects of fouling on performance of retrofitted
heat exchanger networks; a thermo- hydraulic
based analysis.
Francesco Coletti
a
Sandro Macchietto
a
and Graham T. Polley
b

a
Department of Chemical Engineering, Imperial College London, South Kensington
Campus, SW7 2AZ U.K. s.macchietto@imperial.ac.uk
b
Department of Chemical Engineering, Universidad de Guanajuato, Mxico.
Abstract
In refineries, fouling in crude pre-heat trains (PHTs) causes several thermal-hydraulic
inefficiencies which lead to increased operating costs (from reduction in throughput and
extra fuel burnt at the furnace), carbon emissions, and maintenance issues. The energy
recovery performance of PHT can be severely affected over time. Such time varying
effects are normally not considered in the design or retrofit of heat exchangers
networks. In this paper, an existing PHT network is simulated including its fouling
behaviour of over ca. two years. For this purpose, a dynamic, distributed mathematical
model for shell-and-tube heat exchangers undergoing crude oil fouling (developed and
validated against refinery data in previous work) is used. Three retrofit options aimed at
maximizing overall heat recovery are proposed. Simulation results show that networks
designs that maximize energy recovery in clean conditions (following traditional pinch
rules) may not be best when fouling occurs and that a proper retrofit design must
include consideration of time varying fouling effects.

Keywords: Fouling, crude oil, refinery, efficiency, network.
1. Introduction
Two thirds of the energy necessary for primary fractionation in oil refineries are
recovered in the pre-heat train (PHT), an extensive network of heat exchangers (Yeap et
al. 2004). However, PHTs efficiency is hindered by the progressive accumulation of
unwanted material (fouling) on the heat exchange surfaces. Mitigating fouling is
paramount for refinery operators to avoid production loss, health and safety hazards and
increased fuel consumption (with related GHG emissions) at the furnace downstream
the PHT. Typically, PHTs are optimised to maximise energy recovery through pinch
technology or MINLP techniques. However, these techniques have several limitations:
1. They consider steady state performance of exchangers (i.e. no fouling dynamics)
2. They seek to match the hottest process streams with the crude at its highest
temperature. This leads to wall temperatures being maximised. In turn fouling, which
is a function of temperature and velocity, is exacerbated.
3. They rely on simple, lumped models that are limited by a number of assumptions on
physical properties and temperature profiles inside the units
It was previously noted (Wilson et al. 2002) that failure to consider fouling behaviour at
the design stage could lead to poor network performance. This is also confirmed by
practical refinery experience that highly optimised networks will not work due to
fouling dynamics (personal communication). The objective of this paper is to use high
fidelity thermo-hydraulic simulations that capture complex dynamic interactions in
F.Coletti et al.
Residue
Lower
P/A
Upper
P/A Kerosine GO
Crude
E-01 E-02 E-04
HGO

E-07 E-06 E-05 E-03
Fired heater
Desalter

Figure 1 Pre-heat train structure in the base-case configuration C1.
the network to analyse alternative retrofit options of the network. For this purpose, a
mathematical model by Coletti and Macchietto (2009), is used here to assess the fouling
behaviour of an existing network. The model is dynamic, distributed and takes into
account the exchanger geometry and configuration, the variation of temperature,
velocity, physical properties and fouling rate along the length of each unit. It also
accounts for localized fouling growth as a function of process conditions through the
Ebert-Panchal model (Panchal et al. 1999):

, 0.66 0.33
,
Re Pr exp
f n
f n
dR
E
dt RT
o t


=
| |
|
\ .
(1)
The model also captures the interactions between the fouling layer and the fluid-
dynamics by solving a moving boundary problem. Simulation results for industrial
single units indicated that output temperatures are calculated within 2% error even
when the model is tested for its predictive capabilities over 1-2 years.
To illustrate the main point of the paper, first we build the network model based on an
industrial case study of a small refinery. The existing network is used as base case.
Parameters that characterize the fouling behaviour in each unit are estimated so as to fit
this base case. Once the fouling behaviour is captured, the same values are used to
assess alternative retrofit options, with the goal of increasing overall energy recovery.
Three network structures are proposed. The first follows pinch rules to maximize energy
recovery based on clean exchanger performance. The remaining two seek to improve
the energy recovery while also minimizing the fouling behaviour. The results for the
proposed retrofit designs are analyzed with respect to the coil inlet temperature (CIT)
achieved and the heat load at the furnace over the entire period.
2. Case study
In order to protect proprietary information, the heat and mass balance for the existing
network (Figure 1) has been adjusted and some features of the plant (reported in Table
1) have been changed. Nonetheless, these changes do not affect the validity of the
conclusions. This network structure will be referred to as configuration C1 and
represents the base-case for the study. The performance of the pre-heat train was
monitored at start-up (after cleaning) and after 8000 hours operation (Table 2). The
performance of units E1, E2 and E4 did not change significantly over the operating
period whereas the other units exhibited severe fouling. The observations of the refinery
operators were that fouling in units E5 and E6 mainly occurred on the tube-side (the
crude oil flowed through the tubes) and this was the only side of these units cleaned.
Unit E3 fouled on the shell-side (which handled column residues) but not on the tube-
side. Unit E7 fouled heavily on both sides. For the following analysis it is assumed that
the desalter temperature is optimally controlled and that its performance does not affect
the fouling behaviour of the network. Crude is on the tube-side and hot fluids flow on
the shell-side in all units. Stream properties are estimated though API relationships.
Effects of of fouling on performance of retrofitted heat exchanger networks. A thermo-
hydraulic based analysis.
Table 1 Exchangers geometry. All units have 20 ft (6.1 m) tubes of inch outside diameter.
Unit No. Tubes No. Tube pass Baffle cut Baffle spacing No. Baffles Shell diameter
[-] [-] [%] [mm] [-] [mm]
E01 152 6 22 178 32 444
E02 308 8 17 146 40 584
E03 336 2 17 130 43 584
E04 620 4 17 153 37 798
E05 1130 6 17 136 41 1100
E06 240 2 17 138 41 520
E07 968 4 17 137 41 740
3. Proposed network revamps
By analyzing the stream temperatures reported in Table 2, it is clear that more energy
can be recovered from the residue stream. If pinch rules are applied, the area of E-07
would be increased. This can be done by adding an extra unit E-07x, identical to E-07 in
geometry. The resulting network structure, referred to as configuration C2, is shown in
Figure 2(a). In this design, the hot residue stream is matched with the crude at its
highest temperature. Whilst this ensures maximum heat recovery in clean conditions,
over time the fouling behaviour is expected to penalize the overall heat recovery of the
network as the wall temperatures, on which fouling depends, are also maximized. To
increase the performance over time, we analyse an alternative network retrofit,
configuration C3 in Figure 2(b). This goes against traditional pinch rules by matching
the hot residue stream with the crude at an intermediate temperature (at the exit of E-
05). In his configuration, the residue stream enters the extra unit E-07x, placed between
units E-05 and E06 and subsequently enters E-07 and E-03. As a result, heat recovery in
clean conditions is expected to be less than that achievable with configuration C3,
however, the crude at the highest temperature exchanges heat with a lower temperature
residue stream. A third alternative configuration, C4 in Figure 2(c), was proposed. In
this configuration the residue enters first unit E-07 as in the base-case structure and then
unit E-07x which as in structure C3 is placed between E-05 and E06. The performance
given by the different designs will be assessed by considering the value of the CIT
achieved over time and the total energy required by the furnace to maintain a constant
value of the coil outlet temperature (COT) of 360 C. The total (cumulative) energy
required for each configuration Cj, is defined as:
( )
0
( ) ; for 1, 2, 3, 4
t
Cj p
E t mc COT CIT dt j = =
}
(2)
Where mis the massflowrate of the crude, c
p
its specific heat capacity and COT is the
coil outlet temperature. The difference in performance (extra energy recovered at the
Table 2. PHT performance in clean conditions (
0 0
,
in out
T T ) and after 8000 hours (
* *
,
in out
T T ).
Shell-side (Hot streams) Tube-side (crude oil, cold stream)
Unit Flow
0
in
T
0
out
T
*
in
T
*
out
T Flow
0
in
T
0
out
T
*
in
T
*
out
T
[kg/h] [C] [C] [C] [C] [kg/h] [C] [C] [C] [C]
E01 10000 205 44 205 44.1 100000 32 51.3 32 51.3
E02 20000 290 83 290 83 100000 51.3 96.5 51.3 96.5
E03 50000 254.6 206.5 295.4 246.1 100000 96.5 120.5 96.5 121.2
E04 40000 220 148.2 220 147.5 100000 116 147.1 116.7 147.5
E05 60000 280 193 280 203.5 100000 147.1 199.3 147.5 193.4
E06 10000 280 208.5 280 207.2 100000 199.3 206.2 193.4 200.5
E07 50000 360 254.6 360 295.4 100000 206.2 256.4 200.5 231.3
F.Coletti et al.

Residue
Lower
P/A
Upper
P/A Kerosine GO
Crude
E-01 E-02 E-04
HGO

E-07 E-06 E-05 E-03
Fired heater
Desalter
E-07x

(a)
Residue
E-07x
Lower
P/A
Upper
P/A Kerosine GO
Crude
E-01 E-02 E-04
HGO

E-07 E-06 E-05 E-03
Fired heater
Desalter

(b)
Lower
P/A
E-05
Residue
E-07x
Upper
P/A Kerosine GO
Crude
E-01 E-02 E-04
HGO

E-07 E-06 E-03
Fired heater
Desalter

(c)
Figure 2 Network retrofit configurations C2 (a), C3 (b) and C4 (c).
furnace) of each retrofit is:

1 1
; for 2,3, 4
C Cj C Cj
E E E j

A = = (3)
Equation (3) is used to assess the performance of the three retrofit configurations
proposed over the base case. As the same additional heat exchanger E-07x is used in all
three retrofit options, the capital costs will be the same.
4. Results and discussion
Figure 3 shows the CIT over this period for the base-case and the three retrofit options.
It can be seen that in clean conditions (i.e. t=0), for configurations C2 and C4 the CIT is
over 13 C larger than that achieved in C1, indicating a good extra energy recovery.
Despite the extra heat transfer area available in configuration C3 compared with C1, the
initial CIT in configuration C3 is 3.5 C lower than in the base case. This is because the
performance of existing units E6 and E7 falls due to reduced temperature driving force.
Over time, however, things change significantly because of fouling. Fouling rates in E-
07 are highest for configurations C1 and C4, and lowest for C2 and C3 (Figure 4(a)).
Unit E-07x fouls more in C2 then in C3 and C4 (Figure 4(b)). As a result, after less than
a month of operations, the CIT in C3 is maintained at a higher value compared to that of
the base-case C1. After 150 days C3 starts recovering more energy than C2. The
structure generated according to pinch rules, C2, results in the worst performance over
time. After 300 days the CIT in the retrofit structure C4 also falls below that of C3.
Effects of of fouling on performance of retrofitted heat exchanger networks. A thermo-
hydraulic based analysis.
0 100 200 300 400 500 600
200
210
220
230
240
250
260
270
280
C3
C2


C
I
T

[

C
]
Time [days]
Configuration C1
Configuration C2
Configuration C3
Configuration C4
C1
C4

Figure 3 CIT for the base-case and the three alternative network configurations considered.
Whilst the analysis of the CIT highlights the importance of considering fouling
dynamics in the retrofit of PHT networks, it is not sufficient to asses which structure,
amongst those proposed, provides the largest amount of energy recovered over time.
Figure 5 shows the (cumulative) amount of extra energy required in the furnace with
respect to the base-case that can be recovered trough one of the three proposed retrofits
as calculated by Equation (3). Although barely appreciable in Figure 5, given the scale
of the graph, it should be noted that E
C1-C3
is negative for ca. the first 50 days of
operations, confirming that at clean conditions, configuration C3 has a negative impact
on the overall energy recovery. However, after roughly a year of operations
configuration C3 starts performing better than C2, the configuration proposed following
pinch rules. By analyzing Figure 5 another important aspect can be unveiled. Whilst the
CIT in configuration C3 becomes larger than that in C4 after ca. 300 days, the
cumulative extra energy recovered by the latter is constantly larger than that recovered
by the former. Whilst C4 increases the fouling resistance in E-07 with respect to the
base-case (the tube side temperatures are higher), it results in the lowest fouling
resistance in E-07x. Clearly, the placing of the extra unit and its fouling behaviour is
paramount for the overall performance of the network. In fuel costs alone, savings after
600 days of over US$450,000 (assuming a cost of fuel oil of US$27/MWh) can be
achieved by using configuration C2, ca. US$600,000 using C3 and ca.US$680,000
using configuration C4 and are expected to be much higher for larger refineries.
0 100 200 300 400 500 600
0.000
0.005
0.010
0.015
0.020
0.025
C2
C3
C2
C4
Configuration C1
Configuration C2
Configuration C3
Configuration C4


F
o
u
l
i
n
g

r
e
s
i
s
t
a
n
c
e

[
m
2
K

W
-
1
]
Time [days]
(a)
C1
0 100 200 300 400 500 600
0.000
0.005
0.010
0.015
0.020
0.025
C4
C3
Configuration C2
Configuration C3
Configuration C4


F
o
u
l
i
n
g

r
e
s
i
s
t
a
n
c
e

[
m
2
K

W
-
1
]
Time [days]
(b)

Figure 4 Average fouling resistance calculated over time for exchangers E-07 (a) and E07x (b).
F.Coletti et al.
0 100 200 300 400 500 600
0
10
20
30
40
50
60
70
80
AE
C1-C4
AE
C1-C3


A
E

[
G
B
T
U
]
Time [days]
Configuration C2
Configuration C3
Configuration C4
AE
C1-C2

Figure 5 Extra energy recovered at the furnace with respect to the base case in the three cases
considered.
5. Conclusions
A mathematical model for shell-and-tube heat exchangers undergoing crude oil fouling
has been used to show rules aiming at maximize energy recovery alone may not be the
best strategy to pursue in retrofitting PHTs networks. In particular it has been shown,
though the case study presented, that energy recovery in three different retrofit solutions
can be large. Whilst after 300 days of operation, the energy recovered in retrofit
C4>C2>C3, after 600 days C4>C3>c2. In fuel costs alone, these differences translate in
ca. US$ 230,000 over 600 days between network structures that follow pinch rules and
those that take into account fouling.
Some key conclusion can be drawn from the study presented: (i) fouling plays a major
role in the energy recovery which is not captured if steady state conditions alone are
considered at the design stage; (ii) a trade-off exists between maximum energy recovery
and fouling behaviour; (iii) a detailed mathematical model is required to accurately
assess, energy losses and, ultimately, refinery costs related to fouling.
Of course, other factors such as reduction in throughput, which were not included in this
paper, could play a crucial role in the choice of the arrangement and will be considered
in future work.
References
F. Coletti and S. Macchietto (2009). "A dynamic, distributed model of shell and tube
heat exchangers undergoing crude oil fouling." Submitted for publication.
C. B. Panchal, W. C. Kuru, C. F. Liao, W. A. Ebert and J. W. Palen (1999). Threshold
conditions for crude oil fouling. Understanding Heat Exchanger Fouling and Its
Mitigation, Castelvecchio Pascoli, Italy, Begell House Inc.
D. I. Wilson, G. T. Polley and S. J. Pugh (2002). "Mitigation of crude oil preheat train
fouling by design." Heat transfer engineering 23(1): 24-37.
B. L. Yeap, D. I. Wilson, G. T. Polley and S. J. Pugh (2004). "Mitigation of crude oil
refinery heat exchanger fouling through retrofits based on thermo-hydraulic
fouling models." Chemical Engineering Research & Design 82(1): 53-71.