S. H. A. NEJAD et al., Modification of Heat Exchanger Network Design by …, Chem. Biochem. Eng. Q.

26 (2) 79–87 (2012) 79

Modification of Heat Exchanger Network Design

by Considering Physical Properties Variation
S. H. A. Nejad,a F. Shahraki,a,* M. R. S. Birjandi,a A. Kovac Kralj,b and F. Fazlollahia
aDepartment of Chemical Engineering, Faculty of Engineering,
University of Sistan and Baluchestan, P.O. Box 98164-161, Zahedan, Iran Original scientific paper
b
Faculty of Chemistry and Chemical Engineering, Received: December 6, 2011
University of Maribor, Smetanova 17, Maribor, Slovenia Accepted: June 6, 2012

The temperature dependency of streams’ physical properties can have a significant

effect on heat exchanger networks (HENs) design. An average value can be utilized dur-
ing each temperature interval in order to determine the physical properties of the streams.
Another method is the use of the physical properties’ values, which vary with regard to
temperature along each stream.
In this paper, the energy targeting for an ammonia plant has already been done be-
fore using both methods, with and without stream segmentation. Pinch temperature,
DTmin, and the values of the hot and cold utilities in both cases are different and so the
HEN has to be changed. The results demonstrates that when splitting streams into two or
more segments within the existing ammonia plant, the pinch point increases by about
3 °C, and hot and cold utilities consumption increases by approximately 23 % and 11 %,
respectively. The area of the heat exchanger network also is increased by about 50 %. Fi-
nally, an aggregation diagram is introduced for the evaluation of specific heat capacity
distribution along the whole streams of the network.
Key words:
Heat exchanger networks, heat integration, pinch technology, physical properties varia-
tion, aggregation diagram

Introduction sign problem: tree searching algorithm method,4

neural networks,5 mixed integer nonlinear program-
Energy integration technology is a remarkable ming,6 genetic algorithms,3,7 R-curve and Site
approach for decreasing energy consumption and Source Sink Profile (SSSP) analysis,8 shortcut
improving profitability throughout the chemical method,9 graphical technique,10 etc. Although pinch
process industries. Pinch analysis is an efficient analysis relies on heuristic rules and does not guar-
method for developing better-integrated process de- antee a global optimal HEN solution, it is one of the
signs during process heat integration, these advances most notable approaches to designing HENs and
being achieved as pinch technology by Linnhoff maximizing heat recovery.3,10 Therefore, we used
and Vredeveld.1 The main objective of this method pinch analysis in this paper.
is to save expenses by maximizing process-to-pro-
cess heat recovery. The concept of “target before Energy management can be achieved along
design” was introduced by Linnhoff and Hindmarsh,2 two routes; paying less per unit of energy, or reduc-
for designing those of individual processes which ing energy consumption per unit of product.11
consider pinch rules. The first step in the targeting Co-generation of heat and power is an example of
phase is to calculate the minimum heating and cool- the first route, whilst energy integration of the pro-
ing requirements for the heat exchanger network. cess plant is an example of the second. In the theory
Well-designed HENs can significantly contribute to of process integration, it is generally supposed that
a decrease in energy consumption and, therefore, the values of the physical properties are constant.
energy expenses.3 This is accomplished by mini- This consideration can achieve solutions very far
mizing the use of utilities and maximizing heat re- from the industrial application point of view.12
covery between existing hot and cold streams. However, physical property variation can have a
Several techniques and analyses have been de- significant effect during the targeting phases. In the
veloped recently for the heat exchanger network de- works of Panjeh Shahi,13 Polley et al.,14 and Polley
and Panjeh Shahi,15 a relationship between the pres-
*Correspondence:Farhad Shahraki; e-mail: [email protected]; sure drop and the individual heat transfer coeffi-
Tel.: +985412447039; Fax: +985412447186. cients was proposed.
80 S. H. A. NEJAD et al., Modification of Heat Exchanger Network Design by …, Chem. Biochem. Eng. Q. 26 (2) 79–87 (2012)

This paper presents energy integration and ever, for greater temperature differences, the
modification of heat exchanger network design for physical properties and, therefore, the amount of
an ammonia plant, using recent advances in pinch enthalpy are related to the temperature. In this con-
technology in both cases. In the first case, targeting dition, streams should be segmented into different
has been carried out and the physical properties ranges of temperature; in order to accurately por-
over the whole of each stream are supposed to be tray the CP value of a stream as it is heated or
constant. In the second case, various physical prop- cooled.
erties have to be considered as some streams have
to be divided into several segments containing con- Segmenting streams
stant properties. On the other hand, the streams that A constant specific heat capacity can provide
have various physical properties within the range a sound approximation for the behavior of a stream
of their temperatures have to be segmented into if any changes in heat capacity, CP, are small. On
several sub-streams using an approximation of the other hand, dividing a stream becomes in-
their physical properties. Various specific heat dispensable when the heat capacity of a stream
capacities can lead to a shifting of the pinch point, fluctuates over its temperature range across the
and thereupon produce differences during utility heat exchanger network (HEN). Fig. 1 illustrates
targeting. how to determine segmentation regarding hot and
cold process streams. Hot process stream segments
Energy targeting must always predict a lower temperature than the
Targets are theoretical values that illustrate an actual temperature in order to preserve a conserva-
ideal or perfect situation. They are significant as ex- tive approach. Likewise, cold process stream seg-
amination tools as they provide a comparison of ments must always predict a temperature higher
how close the current design is to the optimal de- than the actual temperature.17,18 Process streams can
sign. Energy targets are the minimum amounts of be segmented by defining different heat capacities
utilities needed to satisfy the process streams’ re- and local heat transfer coefficients for different
quirements.16 During energy integration, extracting temperature ranges of the temperature-enthalpy
process streams is the first step when considering plot.
energy targets. Therefore, as CP is assumed to be
constant for each cold or hot stream, from a supply
temperature (Ts) to a target temperature (Tt), the to-
tal heat added or removed will be equal to the
stream enthalpy change, eq. (1):

Q = òT CPdT = CP(Tt - Ts ) = DH
Tt
(1)
s

where, CP is the product of the specific heat capac-

ity (CP) and the mass flow rate of the process
stream. On the other hand, differential heat flow
dQ, when added to a process stream, will increase
its enthalpy (H) by CP dT.
In the conventional pinch method, CP is the
mean heat capacity flow rate for each stream be- F i g . 1 – Segmenting hot and cold process streams
tween the supply and target temperatures, which
can be calculated by arithmetic or geometric means.
Heat transfer coefficient
The mean value theorem is another way of calculat-
ing mean heat capacity flow rate, as given by eq. The heat transfer coefficient (HTC) is used for
(2): calculating the area of the heat exchanger network.
Within the targeted area, the effect of individual
òT CPdT
T2
stream film heat transfer coefficients within the to-
CPm =
1
(2) tal area of the network can be represented by the
òT dT
T 2

1 following expression in eq. (3), as demonstrated by

Linnhoff and Ahmad:19
The CP of a stream rarely varies when the
temperature difference is satisfactory and the K æ I q J
q j ,k ö
1 ç ÷ (3)
enthalpy of each stream with fixed heat capacity is Atotal = å DT ç å h
i,k
+ å h ÷
constant throughout the temperature range. How- k =1 LM , k è i=1 i,k j =1 j ,k ø
S. H. A. NEJAD et al., Modification of Heat Exchanger Network Design by …, Chem. Biochem. Eng. Q. 26 (2) 79–87 (2012) 81

Here qi,k and qj,k are the individual heat loads Process description
on the hot stream i or the cold stream j in the seg-
ment k; likewise hi,k and hj,k are the individual film A grass root design for an ammonia plant was
heat transfer coefficients. Obtaining the total area of applied as a case study in order to survey the effects
a heat exchanger network using this method re- of physical properties’ variations within the heat
quires knowledge of film heat transfer coefficients. exchanger network design. Natural gas was used as
For the shell and tube sides of the heat exchanger, feedstock to produce 1000 tons of ammonia per day
the value of HTC relates to Reynolds and Prandtl with purity of about 99.99 %. Firstly, the natural gas
numbers can be evaluated from eqs. (4) and (5) re- had been desulphurized in order to remove sulphur
spectively: compositions from the feed stream. The natural gas
was heated in the convection section of a primary re-
For shell side: former after mixing with steam, so that it would then
react within the primary and secondary reformers.
1æ
m ö
0.14
h× D The nitrogen requirements for the ammonia reaction
= 0.36 Re 0. 55× Pr 3ç
ç ÷ ÷ (4)
k è mw ø were obtained from preheated air introduced into the
secondary reformer. The required heat for secondary
reformer was provided from the reaction between
For tube side: O2, H2, and CH4. The hot gases leaving the second-
1æ
ary reformer have high heat content and are primar-
m ö
0.14
h× D ily employed for the production of high-pressure
= 0.023 Re 0. 8× Pr 3ç
ç ÷ ÷ (5)
k è mw ø steam in thermosiphon heat exchangers. In two
stages consisting of high and low temperature shift
converter, carbon monoxide was converted into car-
Which:
bon dioxide after mixing with H2O, and then the
D× r× u CP × m CO2 was absorbed by MEA. The residual CO and
Re = , Pr = (6) CO2 were converted into CH4 in the methanator, in
m k order to prevent catalyst poisoning from the ammo-
nia converter. H2 and N2 reacted in the ammonia
The eqs. (4) and (5) are used to calculate the converter to form the ammonia product after com-
HTC based on the specified physical properties val- pression with steam turbines. The hot gases leaving
ues consisting of density, viscosity, heat capacity, the converter were cooled over a series of exchangers
and conductivity (according to eq. (6)); while these where the ammonia was condensed and the non-re-
properties are also related to temperature. active gases were circulated back (Fig. 2).

F i g . 2 – Schematic diagram of ammonia plant

82 S. H. A. NEJAD et al., Modification of Heat Exchanger Network Design by …, Chem. Biochem. Eng. Q. 26 (2) 79–87 (2012)

Results and discussion

The crucial part of the pinch analysis is the

stream data extraction from the process flowsheet
that is required for the pinch analysis.20 The process
streams are those streams that contain the fluid that
should change in heat load but not in composition.
The stream data extracted for the current ammonia
unit consists of 11 hot streams that required cooling
and 12 cold streams to be heated, as shown in Table
1. There is a trade-off between energy and capital
cost, and an economical amount of energy recovery
when choosing optimum DTmin.21 The composite
and grand composite curves when considering the
optimum DTmin for constant heat capacity are
shown in Fig. 3. Required are 55.10 MW of hot
utility and 49.03 MW of cold utility that are not
serviced by heat recovery and should be provided
by external utilities.
The heat exchanger network for the existing
ammonia unit is presented in Fig. 4, and consists of
21 process-to-process heat exchangers that ex-
change approximately 305 MW duties between hot
and cold process streams.
Using the basic problem table method, streams
were approximated using constant CP and inde- F i g . 3 – (a) Composite curve, and (b) grand composite
curve, without considering the segmenting streams
pendent temperatures. However, heat capacities are
apparently always related to temperature and will
tend to vary according to temperature. If it is ig- has been divided into 3 segments within its temper-
nored it may lead to serious targeting errors. Thus, ature range (Fig. 5), whereas stream H-6 has greater
in the second considered streams with sensible tem- fluctuation within its temperature range and is di-
perature dependency, the variations of physical vided into 4 segments (Fig. 6). On the other hand,
properties are segmented into certain sub-streams, stream C-12 as a cold stream is segmented into 2
as shown in Tables 2, 3. For example, stream H-3 sections that modify the heat capacity value for this

T a b l e 1 – Process streams with constant physical properties

Hot streams Cold streams
Stream TS (°C) Tt (°C) CP (MW °C–1) Enthalpy (MW) Stream TS (°C) Tt (°C) CP (MW °C–1) Enthalpy (MW)
H-1 998 360 0.084 53.6 C-1 4.1 4.6 22.00 11
H-2 325 40 0.042 11.9 C-2 –12 –11.4 27.17 16.3
H-3 89 40 0.116 5.7 C-3 50 300 0.041 10.2
H-4 445 26 0.241 101 C-4 –6 25 0.235 7.3
H-5 106 50 1.211 67.8 C-5 150 400 0.142 35.5
H-6 214 38 0.234 41.2 C-6 33 150 0.235 27.5
H-7 414 173 0.079 19 C-7 100 500 0.013 5.2
H-8 288 26 0.041 10.7 C-8 38 500 0.015 7
H-9 103 89 0.581 8.1 C-9 60 380 0.254 81.3
H-10 20 –6 0.627 16.3 C-10 60 450 0.201 78.3
H-11 100 38 0.314 19.5 C-11 69 100 1.242 38.5
C-12 114.5 115 85.2 42.6
S. H. A. NEJAD et al., Modification of Heat Exchanger Network Design by …, Chem. Biochem. Eng. Q. 26 (2) 79–87 (2012) 83

F i g . 4 – Grid diagram of existing ammonia unit with constant heat capacity

T a b l e 2 – Hot streams segmentation

TS (°C) Tt (°C) Viscosity Conductivity Density CP Heat load HTC

Streams
(kg/m–1 s–1) (W m–1 K–1) (kg m–3) (kJ kg–1 °C–1) (MW) (kJ h–1 m–2 °C–1)
325 60 0.000016 0.151 6.67 3.486 10.80 902
H-2
60 40 0.000015 0.119 13.97 4.644 1.10 1292
89 79 0.000039 0.068 66.63 31.917 2.593 3253
H-3 79 61 0.000092 0.138 184.07 13.926 2.075 5733
61 40 0.000156 0.175 257.20 5.963 1.037 5416
445 364 0.000020 0.195 21.31 4.217 19.40 2053
364 253 0.000018 0.174 24.95 4.137 25.76 2119
H-4 253 33 0.000014 0.136 38.07 4.063 50.30 2377
33 29 0.000014 0.118 58.65 12.780 3.068 4044
29 26 0.000017 0.130 76.68 11.980 2.099 4675
214 178 0.000018 0.090 11.87 2.126 2.798 728
178 126 0.000041 0.162 121.78 11.907 22.95 5748
H-6
126 86 0.000086 0.252 264.92 6.142 9.027 8069
86 39 0.000169 0.263 313.09 3.572 6.208 6575
103 102.8 0.000016 0.030 1.29 2.293 0.0043 109
102.8 100.3 0.000025 0.047 30.66 51.101 1.466 2149
H-9 100.3 95.1 0.000053 0.107 123.97 59.435 3.675 7107
95.1 92.4 0.000082 0.167 213.63 47.012 1.470 10849
92.4 89 0.000100 0.200 264.67 36.780 1.470 12149
20 7 0.000021 0.141 91.03 10.066 8.874 4866
H-10
7 –6 0.000032 0.168 128.03 8.408 7.395 5710
100 43 0.000010 0.035 11.20 2.430 2.114 441
H-11 43 42 0.000052 0.218 268.10 1137.74 15.60 46836
42 38 0.000102 0.426 550.76 30.776 1.772 28303
84 S. H. A. NEJAD et al., Modification of Heat Exchanger Network Design by …, Chem. Biochem. Eng. Q. 26 (2) 79–87 (2012)

T a b l e 3 – Cold streams segmentation

TS (°C) Tt (°C) Viscosity Conductivity Density CP Heat load HTC

Streams
(kg/m–1 s–1) (W m–1 K–1) (kg m–3) (kJ kg–1 °C–1) (MW) (kJ h–1 m–2 °C–1)

–12 –11.7 0.000190 0.564 647.26 4.937 0.0205 19421

C-2
–11.7 –11.4 0.000098 0.293 324.59 4858.89 16.25 97712

38 277 0.000016 0.059 16.89 2.725 3.204 617

C-8
277 500 0.000021 0.100 9.96 3.498 3.844 539

60 187 0.000303 0.663 926.09 4.202 14.77 21930

C-9 187 247 0.000126 0.646 834.44 4.857 8.182 31363

247 380 0.000066 0.338 404.61 15.808 58.30 22836

60 237 0.000288 0.642 897.09 4.368 21.52 21724

C-10 237 374 0.000068 0.344 413.63 15.503 58.80 23018

374 450 0.000026 0.064 13.32 2.286 4.833 395

114.50 114.54 0.000300 0.564 973.88 1634 25.34 150462

C-12
114.54 115.0 0.000293 0.546 940.92 106 17.31 58082

F i g . 5 – Segmentation of stream H-3 into three sub-streams, F i g . 6 – Segmentation of stream H-6 into four sub-streams,
(a): without segmentation and (b): with segmenta- (a): without segmentation and (b): with segmenta-
tion tion
S. H. A. NEJAD et al., Modification of Heat Exchanger Network Design by …, Chem. Biochem. Eng. Q. 26 (2) 79–87 (2012) 85

F i g . 7 – Segmentation of stream C-12 into two sub-streams, F i g . 8 – (a) Composite curve and (b) grand composite
(a): without segmentation and (b): with segmentation curve, when considering the segmented streams

stream (Fig. 7). Targeting by considering various H-6, when reaching pinch temperature above the
physical properties and segmentation of the process pinch, should exchange heat with a cold stream,
streams has been repeated. The composite and which has a higher heat capacity flow rate, based
grand composite curves then plotted again (Fig. 8). on pinch technology rules. As shown in Fig. 9,
An aggregation diagram has been introduced for none of the cold streams has this condition. The
better understanding of the quality of heat capacity, heat capacity accumulation for cold streams 9 and
and enthalpy distribution. For example, hot stream 10 are at higher temperatures than the pinch tem-

F i g . 9 – Aggregation diagram for demonstrated CP distribution

86 S. H. A. NEJAD et al., Modification of Heat Exchanger Network Design by …, Chem. Biochem. Eng. Q. 26 (2) 79–87 (2012)

T a b l e 4 – Energy and area targeting with and without segmenting network

DTmin (°C) Hot pinch temperature (°C) Cold pinch temperature (°C) QH,min (MW) QC,min (MW) Area (m2)
Without segment 6 103 97 55.10 49.03 64.712
With segment 5.5 106 100.5 68.07 54.64 96.155

perature. There is no cold stream using the favored DTmin- minimum approach temperature, °C
CP that is near enough to the pinch temperature to hi,k - individual film heat transfer coefficients on the
absorb the additional heat of hot stream H-6. If hot hot stream ‘i’ in segment ‘k’, W m–2 °C–1
stream H-6 is connected to one of these cold hj,k - individual film heat transfer coefficients on the
streams, the heat exchanger will violate the DTmin cold stream ‘j’ in segment ‘k’, W m–2 °C–1
condition; because the CP of the hot stream is K - thermal conductivity, W m–1 °C–1
higher than that of the cold streams. So, this stream
Pr - Prandtl number, –
is used for generation of the LP steam. Overall, the
more CP accumulation of the hot streams above the qi,k - individual heat loads on the hot stream ‘i’ in seg-
pinch or the cold streams below the pinch, the less ment ‘k’, kJ h–1
heat load of the utility streams would be obtained. qj,k - individual heat loads on the cold stream ‘j’ in
segment ‘k’, kJ h–1
Summary of targeting results for both the cases
of constant and varying physical properties, are Re - Reynolds number, –
shown in Table 4 as a demonstration of changing Ts - supply temperature, °C
targeted values. These variations cause targeting Tt - target temperature, °C
correction and improvement in heat exchanger net-
work design.
Greek symbols

r - stream density, kg m–3

Conclusions
m - stream viscosity, kg m–1 s–1
Against the fundamental problem table method mw - viscosity of water, kg m–1 s–1
for designing a heat exchanger network, in practical u - stream velocity, m s–1
problems dependency of physical properties such as
heat capacity according to temperature has been
considered, and ignoring this dependence will often Abbreviations
lead to serious targeting errors. Energy targeting for
an ammonia plant has been investigated before us- CC - Composite Curve
ing both methods, with and without stream segmen- GCC - Grand Composite Curve
tation. Pinch temperature, DTmin, and the values of HEN - Heat Exchanger Network
the hot and cold utilities in both cases are different,
and so the HEN has to be changed. When splitting
streams into two or more segments within the exist- Indices
ing ammonia plant, the pinch point increased by
i - hot stream
about 3 °C, and hot and cold utilities consumption
increased by approximately 23 % and 11 %, respec- j - cold stream
tively. The area of the heat exchanger network also k - segment
is increased by about 50 %.
References
List of symbols 1. Linnhoff, B., Vredeveld, D. R., Chem. Eng. Prog. 80 (7)
Atotal - total area of the network, m 2 (1984) 33.
2. Linnhoff, B., Hindmarsh, E., Chem. Eng. Sci. 38 (5) (1983)
CP - effective specific heat capacity, kJ kg–1 °C–1 745.
CP - heat capacity flowrate, kW °C–1 3. Allen, B., Savard-Goguen, M., Gosselin, L., Appl. Therm.
CPm - mean heat capacity flow rate, kJ kg–1 °C–1 Eng. 29 (2009) 3437.
D - flow area diameter, m 4. Pho, T. K., Lapidus, L., AlChE Journal 19 (6) (1973) 1182.
5. Bittanti, S., Piroddi, L., Journal of Franklin Institute 334B
DH - enthalpy difference, MW (1) (1997) 135.
DTLM,k - log mean temperature difference in the seg- 6. Yee, T. F., Grossmann, I. E., Kravanja, Z., Comp. Chem.
ment ‘k’, °C Eng. 14 (10) (1990) 1151.
S. H. A. NEJAD et al., Modification of Heat Exchanger Network Design by …, Chem. Biochem. Eng. Q. 26 (2) 79–87 (2012) 87

7. Dipama, J., Teyssedou, A., Sorin, M., Appl. Therm. Eng. 15. Polley, G. T., Panjeh Shahi, M. H. M., Trans. on Inst of
28 (2008) 1763. Chem. Eng. 69 (1992) 445.
8. Matsuda, K., Hirochi, Y., Tatsumi, H., Shire, T., Energy 34 16. Papoulias, S. A., Grossmann, I. E., Comp. Chem. Eng. 7
(2009) 1687. (6) (1983) 707.
9. El-Halwagi, M., Harell, D., Spriggs, H., Appl. Eng. 86 (6) 17. Saboo, A. K., Morari, M., Comp. Chem. Eng. 11 (4) (1987)
(2009) 880. 399.
10. Kovac, A. K., Energy 34 (2009) 1372. 18. Saboo, A. K., Morari, M., Comp. Chem. Eng. 11 (5) (1987)
11. Lababidi, M. S., Alatiqi, I. M., Nayfeh, L. J., Appl. Therm. 457.
Eng. 20 (2000)1495. 19. Linnhoff, B., Ahmad, S., Comp. Chem. Eng. 14 (7) (1990)
12. Ravagnani, M. A. S. S., Caballero, J. A., Comp. Chem. 729.
Eng. 31 (2007) 1432. 20. Kemp, I., Pinch Analysis and Process Integration, 2nd ed.,
13. Panjeh Shahi, M. H. M., Ph.D. Thesis, UMIST (1992). Elsevier Ltd. 2007.
14. Polley, G. T., Panjeh Shahi, M. H. M., Jegede, F. O., 21. Krajnc, M., Kovac, A. K., Glavic, P., Appl. Therm. Eng.
Trans. on Inst of Chem. Eng. 68 (1990) 211. 26 (2006) 881.