Retrofit of ammonia plant for improving energy efficiency

Abstract

The aim of this work is to perform a retrofit study of an ammonia plant, in purpose of improving energy efficiency. As a common practice, one can divide an ammonia plant into two parts: the hot-end and the cold-end. In the hot section, two different options are investigated that both lead to a threshold condition and achieve maximum energy saving. The first option covers only process-to-process energy integration, while the second option considers some modification in the convection section of the primary reformer through a new arrangement of the heating coils. Thus, a considerable reduction in cooling water, HP steam and fuel gas consumption is achieved. In the cold section, retrofit study is dominated by reducing the amount of shaft work or power consumption in the refrigeration system. Application of the Combined Pinch & Exergy Analysis revealed that part of the shaft work, which was originally being used, was inefficient and could have been avoided in a well-integrated design. Therefore, by proposing optimum refrigeration levels, reasonable saving (15%) in power consumption was observed without the need for new investment.

Full text

Energy 33 (2008) 46–64
Retrofit of ammonia plant for improving energy efficiency
M.H. Panjeshahi, E. Ghasemian Langeroudi

, N. Tahouni
Department of Chemical Engineering, University of Tehran, Tehran, Iran
Received 27 February 2007
Abstract
The aim of this work is to perform a retrofit study of an ammonia plant, in purpose of improving energy efficiency. As a common
practice, one can divide an ammonia plant into two parts: the hot-end and the cold-end. In the hot section, two different options are
investigated that both lead to a threshold condition and achieve maximum energy saving. The first option covers only process-to-process
energy integration, while the second option considers some modification in the convection section of the primary reformer through a new
arrangement of the heating coils. Thus, a considerable reduction in cooling water, HP steam and fuel gas consumption is achieved. In the
cold section, retrofit study is dominated by reducing the amount of shaft work or power consumption in the refrigeration system.
Application of the Combined Pinch & Exergy Analysis revealed that part of the shaft work, which was originally being used, was
inefficient and could have been avoided in a well-integrated design. Therefore, by proposing optimum refrigeration levels, reasonable
saving (15%) in power consumption was observed without the need for new investment.
r2007 Elsevier Ltd. All rights reserved.
Keywords: Exergy analysis; Pinch technology; Threshold problem; Heat exchanger network; Energy saving
1. Introduction
Ammonia production is an energy intensive process, so
the recovery of relatively small quantities of heat can
accumulate to become sizeable energy savings. The highly
energy consuming nature of the process is the key driving
force for improving the technology and reducing the overall
cost of manufacturing. Hence, any attempt for energy
conservation in the process goes a long way in many aspects.
In recent years, some potentially significant developments
and concepts have been done, that may impact the way in
which ammonia is produced. Some of these manufacturing
routes are being tested or employed at a few plants around
the world, but have still to be fully developed into
commercial processes. This includes reformer combustion
air preheat, control of steam to carbon ratio [1], hydrogen
recovery from purge gas, improved CO
2
removal system
[2,3] and co-generation [4,5]. Other attempts include
optimization [6],safety[7],advancedcontrol[8], improved
catalysts [9] and heat integration [10–12].
The design and retrofit of energy efficient process plants
require tools that enhance the engineers understanding of
the complex interactions between process plants and utility
systems and point to potential solutions. The three schools
of Process Integration methods are the use of graphical
diagrams and thermodynamic concepts, use of mathema-
tical modeling and subsequent optimization and use of
heuristics in design and economy.
Pinch Technology [13–15] as an example of first school,
is a methodology, comprising a set of structured techni-
ques, for systematic application of the first and some
aspects of second laws of thermodynamics. The application
of these techniques enables process engineers to gain
fundamental insight into the thermal interactions between
chemical processes and the utility systems that surround
them. Such knowledge facilitates the improvement of the
overall utility consumption, and setting process and utility
system configurations prior to final detailed simulation and
optimization.
Exergy analysis [16,17] as another example of first school
helps to identify the inefficiency in the processes, so the
engineers can understand the cause and magnitude of
exergy loss in each process unit.
ARTICLE IN PRESS
www.elsevier.com/locate/energy
0360-5442/$ - see front matter r2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2007.08.011

Corresponding author. Tel.: +98 2166957788; fax: +98 2144475465.
E-mail address: ghasemian.el@gmail.com (E. Ghasemian Langeroudi).

The first law analysis method is widely used to evaluate
thermodynamic systems; however, this method is con-
cerned only with energy conservation, and therefore it
cannot show how or where irreversibilities occur in a
system or process. To determine the irreversibilities, the
exergy analysis method is applicable; Exergy is defined as
maximum amount of work, which can be produced by a
system when it comes to equilibrium with a reference
environment. The standard conditions of the earth atmo-
sphere are considered as the thermodynamic state of the
environment [18]. Therefore, combination of pinch and
exergy analysis could be an appropriate tool for simulta-
neous study of heat and power. Particularly in subambient
systems, where in general all cooling and heating require-
ments have a direct shaftwork implication. The most
important application of combined pinch and exergy
analysis is retrofit study of low-temperature processes
and refrigeration cycles, thus in this research we put into
practice Pinch Analysis and Exergy analysis tools to reduce
the exergy loss of refrigeration cycle of the ammonia plant.
Optimization methodologies enable the engineer to
address all multiple and complex trade-offs, which would
not be possible to be exploited manually.
Hierarchical Analysis [19] and Knowledge Based Sys-
tems are rule-based approaches with the ability to handle
qualitative (or fuzzy) knowledge.
2. Threshold problems
As mentioned previously, Pinch Analysis is a thermo-
dynamic approach to energy integration based on simple,
powerful and graphical representation. Several tools
including, Driving Force Plot (DFP), Composite curves
(CCs), Grand Composite Curve (GCC) are applied for
thermal analysis of processes [11]. If the thermal system is
presented in the pinch diagram (HT diagram) by CCs, the
minimum vertical distance between hot and cold
curves represents the minimum approach temperature for
thermal exchanges, DT
min
. This point of minimum
temperature difference for heat exchanges in the
process represents the critical point of the heat recovery
and is called pinch Point. The pinch point divides the
process in two different thermodynamic regions, above
and below the pinch point. Nevertheless, one cannot
achieve the pinch point for all thermal systems. There are
thermal systems that reach a point in which one of the
thermal utilities reduces to zero; the DT
min
value diminishes
by moving the hot and cold CCs on the horizontal axis.
The DT
min
value, at which this happens, is known as
‘‘DT
Threshold
’’.
2.1. Definition of process–process/utility–process divider
There are three zones in the HT diagram which are
made by the position of CCs: the heating zone in the
extreme right of the curves that represents the minimum
hot utility requirements for the process, the exchange zone
where the CCs overlap represents the heat recovery
potential through the heat transfer between streams of
the process, and the cooling zone in the extreme left of the
curves that represents the minimum cold utility require-
ments for the process. In the threshold problems, the hot
ARTICLE IN PRESS
Nomenclature
A
1–1
area with 1–1 exchanger (m
2
)
A
1–2
area with 1–2 exchanger (m
2
)
A11min:rremaining area with 1–1 exchanger (m
2
)
A12min:rremaining area with 1–2 exchanger (m
2
)
Aheat transfer area (m
2
)
A
t
total area of tubes (m
2
)
aMDEA activated Methyl di Ethyl Amine
BFW boiler feed water
HP Steam high-pressure steam
LP Steam low-pressure steam
HT enthalpy–temperature diagram
L height of tubes in kettle (m)
llevel of liquid in kettle (m)
mc
p
heat capacity flow rate (kW/1C)
P
0
reference pressure (Pa)
Qheat transfer rate (W)
T
0
reference temperature (K)
t
C
temperature of cold stream in grid diagram
t
H
temperature of hot stream in grid diagram
Uoverall heat transfer coefficient (W/m
2
K)
U
g
heat transfer coefficient of gas (W/m
2
K)
U
l
heat transfer coefficient of liquid (W/m
2
K)
Wcompressor shaftwork (kW)
Greek letters
DEX
process
exergy supplied to the process (kW)
DH
h
load of hot utility (kW)
DH
c
load of cold utility (kW)
DT
min,r
remaining minimum temperature difference
(1C)
DT
Threshold,r
remaining threshold temperature differ-
ence (1C)
DT
min
minimum temperature difference (1C)
DT
Threshold
threshold temperature difference (1C)
DT
LM
log mean temperature difference between fluids
(K)
(sT
o
)
HEN
exergy loss in a heat exchanger network (kW)
a
max,DP
area efficiency of a network by assumption of
fixed pressure drop
a
max,h
area efficiency of a network by assumption of
fixed heat transfer coefficient
Z
ex
exergetic efficiency of the refrigeration system
Z
c
Carnot factor
M.H. Panjeshahi et al. / Energy 33 (2008) 46–64 47

CC, for example, has such a thermal load that from
DT
Threshold
approach value it stays completely inside
the cold CCs. There are two zones in this situation:
one of heating and one of overlap zone (cold-end
threshold). The point that divides the CCs in two
zones is proposed to be called Process–Process/Utility–
Process Divider. The same thing can happen when
the cold curve is inside the hot curve (hot-end threshold)
resulting only in one zone of cooling and one of overlap
(Fig. 1).
2.2. Retrofit of threshold problems
Retrofit study for heat exchanger networks (HEN)
regarding Threshold problems, are found on the basis of
vertical heat transfer [20]. As we have given the two
conditions discussed (hot-end and cold-end threshold), the
method of retrofitting the network is different. But, due to
the identical theory governing both methods, we only
explain one of them in this paper.
2.2.1. Retrofit of a hot-end threshold problems
The design policies for retrofitting a hot-end threshold
problem are as follows:
(a) Elimination of exiting heaters in the network.
(b) Substitution of existing coolers in process-to-process
region with new heat exchangers.
Therefore, the HEN should carefully be examined to
explore all the available possibilities, based on design
policies. For example, designing a new process-to-process
exchanger for the network may lead to removal of a heater
and also reducing the duty of a cooler, which transfers heat
from above the PP/UP divider to below the divider. This
new heat exchanger results in thermodynamic improve-
ment through substitution of utility consumption with
process heat recovery. It is obvious that such retrofit
options have priority.
3. Case study
The main sections of the understudy ammonia plant
which located in south of Iran, are as follows:
Natural gas purification: For the removal of H
2
S and
organic sulfur with cobalt/molybdenum catalyst, sulfur
content of stream poisons the catalyst of the primary
reformer.
Reforming: The natural gas is mixed with steam and
heated in the convection section of the primary reformer.
The gas and steam mixture will then react in the primary
and secondary reformer. Preheated air is introduced in the
secondary reformer as a source for N
2
needed for the
ammonia reaction, whereas O
2
reacts with H
2
and CH
4
to
provide the heat needed for the steam reforming reaction.
Hot gases leaving the secondary reformer are mainly used
in high-pressure steam production.
CO-conversion and CO
2
absorption: CO is converted to
CO
2
in two stages; high and low-temperature shift
converters. CO
2
is absorbed by aMDEA.
Methanation: Remaining CO and CO
2
are converted to
CH
4
in order to avoid poisoning the ammonia converter
catalyst.
Synthesis and refrigeration:H
2
and N
2
react in the
ammonia converter to form the ammonia product. Hot
gases leaving the converter are cooled down in a number of
chillers (using ammonia as a refrigerant in three tempera-
ture levels) where ammonia is condensed and the unreacted
gases are circulated back.
The plant can be divided into two parts: the hot-end and
the cold-end (Figs. 2 and 3). The hot end consists of all
plant sections that involve in preparing the gases needed
for the ammonia reaction such as: desulphurization,
primary reforming, secondary reforming, shift conversion,
CO
2
removal and methanation. The cold-end section is
mainly the ammonia synthesis and refrigeration cycles
(Figs. 2 and 3). Detail study of hot section has been done
using Pinch Analysis and also the cold section has been
rechecked by the application of Combined Pinch & Exergy
Analysis.
4. Hot section
Stream data extracted for the current case study
consist of 19 hot streams that require cooling and
16 cold streams to be heated. The grid diagram includes
15 heat exchangers, 11 coolers using cooling water, sea-
water, air coolers and four heaters using HP and LP
steams. The existing process requires a total hot utility
load of 6076 kW, and a total cold utility of 119,786.2 kW.
CC and GCC of the existing plant are shown in Figs. 4
and 5.
In the hot section, two alternative options can be
proposed:
The first option covers process-to-process energy inte-
gration.
ARTICLE IN PRESS
Fig. 1. A hot-end threshold.
M.H. Panjeshahi et al. / Energy 33 (2008) 46–6448

The second option considers some modification in the
convection section of the primary reformer through a
new arrangement of the heating coils.
Cost data: The correct cost data are essential for a
successful retrofit project. The cost needs to be annualized
to study the economics in terms of yearly saving and
payback time. The basic economic data consists of yearly
operating hours (8000), plant life (20 years) and interest
rate (15%).
Operating costs: The heating utilities currently include
low and high-pressure steam, whilst the cold utilities
include air, cooling water, and sea-water. The average
costs of these utilities are reported in Table 1.
Capital costs: The capital cost of heat exchangers is
reported in Table 2.
4.1. Retrofit—scheme 1
By considering the range targeting result (Table 3), the
minimum temperature approach used in this analysis is
12.4 1C which leads to a threshold condition and achieves
maximum energy saving. Minimum utility of plant, total
saving and total investment at DT
Threshold
¼12.4 1C are
illustrated in Table 4. Accordingly, the CCs and GCC of
the process are obtained, as illustrated in Figs. 6 and 7.
Minimum hot and cold utility requirements (DH
h
and DH
c
)
are found to be 0 and 113,710.2 kW, respectively.
As hot utility equals zero, the PP/UP Divider divides the
CCs into two zones, one zone of cooling and one of
overlap. The PP/UP Divider position, which corresponds
to, DT
Threshold
=12.4 is found at 79.75 1C on the hot CC
and at 10 1C on the cold CC. The grid diagram, Fig. 8,
confirms that exchangers E1, E6, E8, E14 and E15 transfer
ARTICLE IN PRESS
Fig. 2. Simplified flow diagram of ammonia plant.
Fig. 3. Schematic of ammonia plant.
M.H. Panjeshahi et al. / Energy 33 (2008) 46–64 49

heat from below the PP/UP Divider to above this point,
while coolers C2, C3, C5, C7, C8, C10 and C11 are located
in the overlap zone (above the PP/UP Divider).
The reduction of hot utility is equivalent to 4460 kg/h of
HP steam, which decreased the fuel gas consumption in the
boiler, and also 5910 kg/h reduction of LP steam, that
indirectly affects the fuel gas consumption by increasing
the enthalpy of BFW in the cycle. Moreover, reduction of
446.319 T/h of cooling water and 562.222 T/h of sea-water
of cold utility has been reported.
ARTICLE IN PRESS
Fig. 4. Composite curve of the existing plant.
Fig. 5. Grand Composite Curve of the existing plant.
Table 1
Operating costs for the hot section of ammonia plant
Existing hot utility (kW) 6076 Interest rate (%) 15
Existing network area (m
2
)P
N
i¼1
Heat exchanger area ¼14;827:3Hot utility cost ($/kW yr) 23
Average area per shell (m
2
)P
N
i¼1
Heat exchanger area=no:shell ¼864:65 Cold utility cost ($/kW yr) 1.52
Plant life time (yr) 20 Power cost ($/kW yr) 200
M.H. Panjeshahi et al. / Energy 33 (2008) 46–6450

4.1.1. Utility diagram
As mentioned before, the feasibility study of installing
new process-to-process exchangers in the network and for
transferring heat between streams, over which there is a
heater or a cooler, is the priority of retrofitting the
threshold problems. Therefore we introduce a new dia-
gram named utility diagram, which facilitates decision at
this step.
A utility diagram only contains a set of process streams
passing through heaters and coolers in order to show
inappropriate uses of utility units. As shown in the utility
diagram of Fig. 9, the two vertical lines represent the
PP/UP Divider temperatures (from the hot and cold CCs
of Fig. 6). In Fig. 9 cold streams are presented in blue and
hot streams in red. In the utility diagram heat load of all
existing heaters/coolers and utilities name (according to
grid diagram of Fig. 8) are tabulated in the extreme
left of the diagram and the amount of heat that each
utility exchange above PP/UP Divider, are defined at the
right end of each stream. In this diagram, hot and cold
streams’ temperatures have been shifted by the size of
DT¼12.4/2 1C. Thereby the overlap section between
hot and cold stream is an appropriate guide for study
the possibility of designing new process-to-process heat
exchangers. Of course these decisions should be confir-
med by remaining problem analysis, DFP and process
limitations.
Nevertheless, as the case study shows in Fig. 9, there
are some opportunities for designing new matches;
cold stream C13 (Fig. 10) and a split of cold streams
C14 and C2 make overlap with some hot streams.
However, because of small duty demand of cold stream
C2 and also according to the limitations in total invest-
ment and additional area (Table 4), there is no benefit
splitting such exchanger. As there were no further
beneficial modification, analyzing the existing network
was considered.
4.1.2. Analysis of existing network
The potentials for energy saving in the retrofit pro-
ject have been identified at the targeting stage. Table 5
shows the result of remaining problem analysis of the
existing heat exchangers in both constant heat trans-
fer coefficients and fixed pressure drop, using PILOT
ARTICLE IN PRESS
Table 3
Range targeting result for retrofit—scheme 1
DT
min
(1C) Saving ($/yr) Investment ($) Payback (yr) Hot utility (kW) Cold utility (kW)
12.4 148,969 295,773 1.985 0.0 113,710.8
12.6 141,009 245,633 1.742 325.2 114,034.8
12.8 133,047 298,047 2.24 649.9 114,359.5
13 125,086 247,375 1.978 974.6 114,684.2
13.2 117,126 234,216 2 1299.3 115,008.9
13.4 109,165 221,217 2.026 1623.9 115,333.6
13.6 101,204 208,265 2.058 1948.6 115,658.2
13.8 93,243 195,307 2.095 2273.3 115982.9
14 85,281 182,340 2.138 2598 116,307.5
14.2 77,321 134,657 1.742 2922.6 116,632.2
14.4 69,360 124,002 1.788 3247.3 116,956.9
14.6 61,400 113,225 1.844 3571.9 117,281.6
14.8 53,438 102,270 1.914 3896.6 117,606.2
15 45,477 91,064 2.002 4221.3 117,930.9
15.2 37,516 79,507 2.119 4546 118,255.6
15.4 29,555 67,443 2.282 4870.6 118,580.3
15.6 21,595 54,624 2.53 5195.3 118,904.9
15.8 13,633 40,553 2.974 5520 119,229.6
16 5673 23,909 4.215 5844.7 119,554.3
16.1 1692 12,841 7.591 6007 119,716.6
Table 2
Capital cost for the hot section of ammonia plant
Area capital cost ($) ¼A+B(Area)
C
Area capital factor A4600$
Area capital factor B920
Area capital factor C0.7
Table 4
Targeting result at DT
Threshold
¼12.4 1C for retrofit—scheme 1
Retrofit point target
Min. temp. diff. (1C) 12.4 Minimum 1–2 area (m
2
) 11,762.52
Cold utility (kW) 113,710.12 Target area (m
2
) 17,053.41
Hot utility (kW) 0.0 Existing area (m
2
) 14,827.30
Exist. hot Ut. (kW) 6076.00 Additional area (m
2
) 2226.11
Hot Ut. reduce. (kW) 6076.00 Inc. (a) 0.690
Total saving ($/yr) 148,969 Total investment ($ 295,773
Payback yr 1.985
M.H. Panjeshahi et al. / Energy 33 (2008) 46–64 51

software
1
[Appendix A]. It enables to make a more precise
analysis in order to integrate thermal utilities in the
process.
4.1.3. Diagnosis stage
1. By thermal shifting the exchanger E14 to the lower
temperature in such a way E14 moves to the left of PP/UP
division, a heat pocket on the hot stream H15 is released
before getting imported into the heat exchanger. It causes
considerable heat recovery by reasonable matches between
the hot stream (H15) and cold streams (C2, C16 and split
of C14), thus 4141 kW reduction in duty of cooler C9 is
obtained.
To eliminate the heater H4 from cold stream
C16, exchanger D has been proposed which re-
duces the temperature of hot stream H15 down to
238.4 1C.
Hot stream H15 is cooled from 238.4 to 233.4 1C
through heat exchanger C2.
Cold stream C2 is completely heated by a thin split of
hot stream H15 via exchanger A.
By analyzing and rating the shifted heat exchanger E14,
it is noticed that its heat transfer area should be increased
to accommodate new thermal position. This results adding
new shell to it, called E14A.
Here, we should mention that one stream passes
through E14 and Cooler C9, but because the stream phase
changes in cooler C9 and large difference in mc
p
has been
reported; we have separated H15 and H16 streams in grid
diagram.
2. According to the utility diagram (Fig. 9), decreasing
1934 kW from the load of cooler C11 makes it possible to
omit remained heaters from two cold streams C13, a split
ARTICLE IN PRESS
Fig. 7. Grand Composite Curve of the first retrofitted proposal.
Fig. 6. Composite curve of the first retrofitted proposal.
1
PILOT 2.02, Software for process integration, Polley and Panje shahi,
UK/Iran, 2002.
M.H. Panjeshahi et al. / Energy 33 (2008) 46–6452

of C14 and replace them by new process-to-process heat
exchangers, B and C1 (Fig. 11).
Each new heat exchanger and also, step-by-step cumu-
lative effects of removing them from HEN have been
checked by remaining problem analysis [Appendix A], and
the results are shown in Table 6.Fig. 11 clarifies the
representation of retrofitted scheme in grid diagram.
4.2. Retrofit—scheme 2
In the production process of ammonia, natural gas is
mixed with steam and charged to the primary reformer.
The function of reforming is to produce hydrogen by
reacting the natural feed gas with steam over a nickel
catalyst. The reaction products are H
2
, CO and CO
2
.
ARTICLE IN PRESS
Fig. 8. Grid diagram of existing HEN showing the PP/UP division.
M.H. Panjeshahi et al. / Energy 33 (2008) 46–64 53

The reformer configuration shall be on the basis of plant
experience to maximize the overall efficiency of the
equipment. The primary reformer shell consists of a
radiant box and a convection section for heating coils. In
the existing process the flue gas leaves the convection
section of the primary reformer at 185 1C(Fig. 12) whereas
the lowest temperature the flue gas can be taken to be the
acid dew point (140 1C in this case). In the current study as
illustrated in Fig. 13, the stack flue gas temperature has
been lowered to 153 1C. As a result, heater H3 on the cold
stream C14 might be canceled in case; coil C can satisfy the
target temperature.
Accordingly, rest of the network is targeted and retro-
fitted using the network Pinch method. Nevertheless,
ARTICLE IN PRESS
Fig. 9. Utility diagram of first retrofitted scheme.
Fig. 10. Possible opportunities for designing new process-to-process matches between cold stream C13 and other hot streams.
M.H. Panjeshahi et al. / Energy 33 (2008) 46–6454

before reducing DT
min
, by moving the hot and cold CCs
closer to each other to reduce energy consumption, we
should remove the cold stream C14 from grid diagram.
There are two points of concern for omitting cold stream
C14 from grid diagram: first, flue gas that flowing into the
primary reformer is not considered to be a part of process
streams so, it is not possible to let it enter the grid diagram
and make a match with the cold stream C14. Second point,
the flue gas resulted from combustion is not considered to
be hot utility as well. Furthermore, no extra expense is paid
for obtaining this energy and by not making use of it there
will be no cost reduction. In other words it is like disposal.
The minimum temperature approach used for the HEN
in this case is 12.4 1C, which leads to a threshold condition,
similar to the first retrofitted option. PP/UP Divider
position, which corresponds to, DT
Threshold
¼12.4 is found
at 81.23 1C on the hot CC and at 10 1C on the cold CC.
Accordingly, minimum hot and cold utility requirements
ARTICLE IN PRESS
Table 5
Remaining problem analysis of existing heat exchangers
DT
Threshold
¼12.4 DH
h
¼0 Constant heat transfer coefficient A11min ¼11;539:67 Fixed pressure drop A12min ¼11;762:5
Exchanger name DT
Threshold,r
A
1–1
A11min:ra
max,h
A
1–2
A12min;ra
max,DP
E1 12.19 668.3289 11,102.48 0.980 1227.5 11,310.38 0.938
E2 12.40 431.6074 11,195.72 0.993 714.30 11,423.74 0.969
E3 12.40 124.5314 11,471.32 0.995 150.30 11,736.80 0.990
E4 11.83 96.45125 11,557.12 0.990 151.00 11,811.00 0.983
E5 15.19 2279.877 9511.030 0.979 2750.0 9709.76 0.944
E6 12.40 1757.118 10,128.79 0.971 1454.0 10,305.76 1.000
E7 11.99 415.3133 11,217.25 0.992 944.10 11,491.78 0.946
E8 12.40 125.3511 11,460.23 0.996 223.80 11,713.73 0.985
E9 12.32 672.0271 10,963.98 0.992 1543.4 11,170.71 0.925
E10 10.77 182.2555 11,812.80 0.962 244.20 12,194.92 0.946
E11 12.4 4472.728 8317.120 0.902 965.70 8366.300 1.00
E12 12.40 238.3825 11,442.71 0.988 468.00 11,768.42 0.961
E13 12.40 290.2002 11,406.57 0.987 483.00 11,699.74 0.966
E14 9.858 1822.481 10,179.52 0.962 2938.0 10,448.36 0.879
E15 12.40 424.9766 11,373.47 0.978 570.00 11,589.15 0.967
Fig. 11. Grid diagram of first retrofitted scheme.
M.H. Panjeshahi et al. / Energy 33 (2008) 46–64 55

are found to be 0 and 116,910.3 kW, respectively. Also total
saving and investment at DT
Threshold
¼12.4 are illustrated
in Table 7. Here, we follow the same steps as first option.
The hot utility reduction is the same as the first retrofit
but the reduction in the cold utility is equivalent to
309.976 T/h of cooling water.
4.2.1. Diagnosis stage
1. By thermal shifting the exchanger E14 to the lower
temperature in such a way E14 moves to the left of
PP/UP division, the heat interval on the hot stream
H15 is released before getting imported into the heat
exchanger. This rearrangement offer two new matches
(A, D) between cold streams C2 and C15 and hot stream
H15 by removing heaters H1 and H4 (Fig. 14).
2. Reducing the load of cooler C8 makes a possibility of
designing new exchanger called B, which transfers heat
between cold stream C13 and hot stream H13. Hence,
heater H2 can be omitted (Fig. 14).
Fig. 14 is the representation of the retrofitted scheme in
the grid diagram. Each new heat exchanger and also, step-
by-step cumulative effects of removing them from HEN
have been checked by remaining problem analysis shown in
Table 8.
5. Cold section
Refrigeration systems are cyclic processes that employ
refrigerants to absorb heat from one place and move it
to another. Mainly, a refrigeration system consists of a
condenser, an evaporator, a compressor, and an expansion
valve. Fig. 15 shows the schematic of an ammonia
refrigeration cycle which uses three ammonia temperature
levels—18.3, 2.1, and 9.3 1C.
ARTICLE IN PRESS
Table 6
Remaining problem analysis for the new heat exchangers
DT
Threshold
¼12.4 DH
h
¼0 Constant coefficient A11min ¼11;539:67 Fixed pressure drop A12min ¼11;762:5
Exchanger name DT
Threshold,r
A
1–1
A11min:ra
max,h
A
1–2
A12min;ra
max,DP
A 12.33 56.325 11,495.19 0.999 96.5500 11,615.2 1.000
B 12.40 6.6419 11,513.11 1.000 73.1400 11,735.61 0.996
C1 11.57 89.733 11,456.89 0.999 149.990 11,689.56 0.993
C2 11.42 15.899 11,588.82 0.994 49.7700 11,742.67 0.997
D 12.40 66.987 11,504.95 0.997 158.810 11,713.53 0.991
E14A 12.40 1659.6 10,420.67 0.956 1701.98 10,406.57 0.971
A+B 12.33 62.966 11,536.47 0.995 169.69 11,674.51 0.993
A+B+D 12.33 129.95 11,502.09 0.992 328.50 11,643.94 0.983
A+B+D+C1 11.49 219.68 11,420.33 0.991 478.49 11,572.25 0.976
A+B+D+C1+C2 10.45 236.21 11,473.08 0.986 528.26 11,651.71 0.966
A+B+D+C1+C2+E14A 10.45 1895.8 10,350.64 0.942 2230.2 10,453.24 0.927
Fig. 12. The coil arrangement of the existing plant.
M.H. Panjeshahi et al. / Energy 33 (2008) 46–6456

As shown in Fig. 16, stream data extracted for the
current ammonia refrigeration cycle consist of 5 hot
streams and 1 cold streams. The grid diagram includes 1
heat exchanger, 4 chillers using ammonia as a refrigerant,
and 2 water coolers. While analyzing the existing HEM, it
is noticed that there is no hot utility and the refrigeration
cycle is in threshold condition. This derives the first
conclusion that the ammonia refrigeration cycle, which
was selected, for the case study is well integrated. Hence,
saving in total utility consumption through process-to-
process heat integration is not a promising option.
Alternatively, savings are expected to be in improving the
refrigeration levels. In the cold section, without changing
the configuration of compressors and chillers, the refrig-
eration cycle is retrofitted by application of the Combined
Pinch & Exergy Analysis [21]. Thus 15% of the compres-
sor’s shaft work could have been avoided in a well-
integrated design. The remaining questions are whether this
claimed benefit is practically applicable or not and also
which refrigeration levels should be modified.
5.1. Analysis of refrigeration levels in the existing ammonia
cycle
Exergy losses are inevitable because all natural processes
are irreversible. Technically and economically speaking,
exergy is valuable, and as a consequence, whenever it is
intended to solve a problem through the scope of exergy
analysis, a specific exergy loss should be found, which
minimizes operational costs. Several tools, including CCs,
Exergy Composite Curve (ECC), Exergy Grand Composite
Curve (EGCC), were used in the diagnosis of process
exergy utilizing. The CCs (Fig. 17) can be redrawn by
replacing the temperature axis with the Carnot factor
Z
c
¼(1T
0
/T), resulting in the ECC (Fig. 18). The
area between the hot and cold CCs is proportional to the
exergy loss, which is denoted (sT
o
)
HEN
. Thus (sT
o
)
HEN
is
proportional to the amount of ideal work equivalent
lost in heat transfer? The analogous concept applies for
the GCC.
Fig. 19 shows the utility and refrigeration level place-
ments in the EGCC. The ammonia cycle uses three
ammonia temperature levels—18.3, 2.1, 9.3 1C. The utility
system provides exergy, a part of which is lost in heat
exchange (area between EGCC and utility levels, (sT
o
)
HEN
)
whilst the remainder is supplied to the process (area inside
EGCC, DEX
process
). For subambient process the refrigera-
tion system is the utility system. Linnhoff and Dhole [16]
have shown that changes in the area representing (sT
o
)
HEN
are proportional to changes in refrigeration shaftwork
requirement.
5.1.1. Simulation of the refrigeration cycle
HYSYS v3.1 (build 4815) was used to simulate the cycle
and the Soave & Redlich—Kwong (SRK) equation of state
was selected to calculate the thermodynamic properties of
ARTICLE IN PRESS
Fig. 13. The proposed coil arrangement.
Table 7
Targeting result at DT
Threshold
¼12.4 1C for retrofit—scheme 2
Retrofit point target
Min. temp. diff.
(1C)
12.4 Minimum 1–2
area (m
2
)
10,914.38
Cold utility (kW) 116910.36 Target area (m
2
) 15,984.00
Hot utility (kW) 0.0 Existing area (m
2
) 14,827.30
Exist. hot Ut.
(kW)
2876.00 Additional area
(m
2
)
1156.70
Hot Ut. reduce.
(kW)
2876.00 Inc. (a) 0.683
Total saving ($/yr) 70502 Total investment
($)
132,845
Payback (yr) 1.884
M.H. Panjeshahi et al. / Energy 33 (2008) 46–64 57

ammonia–water mixture because it shows a good agree-
ment with the plant data.
5.2. Retrofitted scheme for the refrigeration levels
In this study, the temperature differences between
process and the refrigeration levels have been improved
to introduce the new refrigeration levels. The existing
(9.3, 2.1, 18.3 1C) and proposal (14, 2.1, 13 1C)
refrigeration level placements are shown in Fig. 20. This
figure clarifies that the area between EGCC and refrigera-
tion levels, (sT
o
)
HEN
has decreased, but there are no
changes in chiller’s duty.
Here, it should be noted that in the understudied
ammonia cycle the chillers operating pressures would
determine the refrigeration temperature levels. Therefore,
by changing the chillers’ operating pressures, the targeted
refrigeration levels can be achieved. Thus, to access the
required pressure and also by the aim of confirming the
retrofitted result, the refrigeration cycle is simulated by
HYSYS software before and after retrofit. Refrigeration
temperature levels and pressure variations of the chillers,
before and after retrofit are recorded in Table 9.
To change the chillers’ operating pressure in order to
reach the target refrigeration levels, the outlet pressure of
the valves that locates before and after each chiller were
just set.
In the ammonia refrigeration cycle, applying the
proposal temperature levels will decrease the DT
LM
of
chillers. Hence, by increasing the normal levels in chillers
[Appendix B], the heat transfer coefficient will be increased
which leads to the chillers’ heat load remaining constant.
The retrofitting, taken regarding the temperature levels,
cause just slight changes in chillers E-3408 & E-3409 flow
rates. Since these changes are not of any significance and
the fact that the chillers’ heat load are fixed after the
ARTICLE IN PRESS
Fig. 14. Grid diagram of second retrofitted scheme.
Table 8
Remaining problem analysis for the new heat exchangers
DT
Threshold
¼12.4 DH
h
¼0 Constant coefficient A11min ¼10;474:46 Fixed pressure drop A12min ¼10;914:38
Exchanger name DT
Threshold,r
A
1–1
A11min:ra
max,h
A
1–2
A12min;ra
max,DP
A 12.31 49.67677 10,434.00 0.999 257.220 10,805.86 0.987
B 12.40 14.50015 10,461.91 1.000 119.600 10,910.97 0.990
D 12.40 66.41170 10,440.30 0.996 225.510 10,815.82 0.989
E14A 12.40 491.080 10,136.76 0.985 561.46 10,623.19 0.976
A+B 12.31 64.176 10,421.44 0.999 376.82 10,802.43 0.976
A+B+D 12.31 130.58 10,387.52 0.996 602.33 10,771.88 0.960
A+B+D+E14A 12.31 621.66 10,050.25 0.982 1163.7 10,521.08 0.934
M.H. Panjeshahi et al. / Energy 33 (2008) 46–6458

ARTICLE IN PRESS
Fig. 16. Grid diagram of the existing HEN of cold section.
Fig. 15. Schematic of ammonia refrigeration cycle.
Fig. 17. Composite curve of cold section.
M.H. Panjeshahi et al. / Energy 33 (2008) 46–64 59

ARTICLE IN PRESS
Fig. 18. Exergy Composite Curve of cold section.
Fig. 19. The refrigeration level placements on the Exergy Grand Composite Curve for existing refrigeration cycle.
Fig. 20. Exergy Grand Composite Curve with existing and proposal refrigeration levels.
M.H. Panjeshahi et al. / Energy 33 (2008) 46–6460

retrofit, there is no need for making any changes or
designing new chillers.
5.3. Exergy losses calculation for existing and retrofitted
cycle
The environment parameters are assumed as follows:
P
0
¼101 325 Pa, T
0
¼298.15 K [22]. The exergy analysis of
the existing chillers and the chillers with the proposal
refrigeration levels are carried out using DEx ¼
DH1ðT0=TÞ

. The results are summarized in Table 10.
Regarding the fact that the existing compressor’s shaft-
work is W¼12,780 kW the exergy efficiency is found to be:
Zex ¼P3
i¼1jDExj
W¼3793
12;780 ¼0:3, (1)
DsToHEN

¼XDEX 1XDEX 2
)DðsToHEN Þ¼558 kW;ð2Þ
Hence, DW
act
is found to be:
DWact ¼1
Zex
DsToHEN

)DWact ¼1860 kW:(3)
Note that the above equation assumes that Z
ex
remains
constant.
6. Discussion and conclusion
Detailed energy conservation study of the hot-end of an
existing ammonia plant has been performed and two
promising retrofit options have been investigated to realize
a energy reduction of 6076 kW, which is equivalent to
4460 kg/h reduction of HP steam decreasing the fuel gas
consumption in the boiler and also 5910 kg/h reduction of
LP steam has been reached, that indirectly effects the fuel
gas consumption by increasing the enthalpy of BFW in the
cycle. Consequently, a satisfactory conformity between the
result achieved from the design and the one obtained from
the target is reached.
The retrofit study of the low-temperature processes is
dominated by the shaft work or power consumption of the
refrigeration system and it is important to determine
optimum temperatures and pressures in a refrigeration
system. In the cold section of the ammonia plant,
application of the Combined Pinch & Exergy Analysis
revealed that part of the shaft work was inefficient.
Therefore, by optimizing the refrigeration levels, reason-
able shaft work have been saved (DW
act
¼1860 kW)
without the need of new investment. Also, to check the
integrity of other parts of process with the new refrigera-
tion levels in the cycle, simulation was performed with a
good accuracy.
Acknowledgment
This work was sponsored by Razi Petrochemical Co. in
mahshahr petrochemical zone, south of Iran.
Appendix A. Remaining problem analysis (R.P.A.)
A.1. Assumed-h R.P.A
Driving Force Plot is strictly a qualitative measure of
evaluating heat exchangers. The effect of poor alignment is
evaluated quantitatively using Remaining Problem Analy-
sis (Tjoe, 1986) [23]. This analysis can briefly be descried as
follows.
ARTICLE IN PRESS
Table 9
Pressure variation of chillers after retrofit
Exchanger name Ref. level before
retrofit (1C)
Pressure before retrofit
(bar)
Ref. level after retrofit
(1C)
Pressure after retrofit
(bar)
E-3203 +2.1 4.6 +2.1 4.6
E-3408 +9.3 6 +14 7.13
E-3409 18.3 2 13 2.5
E-3603 18.3 2 13 2.5
Table 10
Exergy losses of existing and proposed refrigeration levels
Exchanger DH(kW) Existing refrigeration level (1C) DEx (kW) Proposed refrigeration level
(1C)
DEx (kW)
E-3408 9210 +9.3 512.21 +14 353
E-3203 3390 +2.1 282.2 +2.1 282.2
E-3409, E-3603 17780 18.3 2998.2 13 2598.6
Result P
3
i¼1
jDExj1¼3793 P
3
i¼1
jDExj2¼3234
M.H. Panjeshahi et al. / Energy 33 (2008) 46–64 61

The elements of the stream data associated with the
exchanger being examined (t
H1
,t
H2
,t
C1
,t
C2
,A
i
,Q
i
of exchanger i) are removed from the grid diagram
(Fig. A1).
The DT
min
associated with the targeted energy consump-
tion given the ‘‘residual problem’’, (DT
min,r
) is then
calculated. The minimum area requirement for the
remaining data (A
min,r
) is then determined. This is added
to the area of the exchanger (A
i
). Then, by dividing the
minimum area requirement for the original full data set
(A
min
) by this sum, we determine the maximum a(a
max,i
)
that can be obtained if the proposed exchanger is accepted
as part of the final network design:
amax;i¼Amin
AiþAmin;r
. (A.4)
Exchangers having an a
max
close to unity are efficient
units and can be left alone. Those having an a
max
below the
efficiency (a) of the existing network cannot be accepted
and their contribution to the network performance need to
be improved.
This R.P.A. is conducted using the assumed-h area
algorithm and its result is referred to below as a
max,h
.
ARTICLE IN PRESS
tC1 tC2
tH2 tH1
Ai , Qi
Assumed heat transfer coefficient R.P.A.
Remaining Problem
Amin,r ,
Δ
Tmin,r
α
max,i = Amin / (Ai+Amin,r)
Proposed
Match
Fig. A1. Remaining Problem Analysis based on assumed heat transfer coefficients.
tC1 tC2
tH2 tH1
Ai , Qi
Fixed Pressure Drop R.P.A.
Remaining Problem
Amin,r ,
Δ
Tmin,r
α
max,i = Amin / (Ai+Amin,r)
Proposed
Match
ΔPtΔPs
Fig. B1. Remaining Problem Analysis based on fixed pressure drop.
M.H. Panjeshahi et al. / Energy 33 (2008) 46–6462

A.2. Fixed-DP R.P.A.
Assumed-h R.P.A. provides a clear image about the
thermal efficiency of the exchangers. But, it tells us nothing
about how well we are using pressure drop. It will be seen
later that poor use of pressure drop results in poor use of
area (and hence capital). However, given the area
algorithm based on stream pressure drops, a R.P.A. which
accounts the combined effects of individual exchanger
pressure drop and exchanger placement on network
performance is possible (Polley and Panjeshahi, 1990) [24].
Again, for each exchanger we simply remove the stream
elements (now including the individual pressure drops)
associated with the unit from the data set and re-solve
(Fig. B1). The subsequent result of this analysis is referred
to below as a
max,Dp
.
A.3. Combined use of R.P.A.
Remaining Problem Analysis based on assumed film
heat transfer coefficients tells us how efficiently the area is
being used. Remaining Problem Analysis based on fixed
pressure drop tells us how well the exchanger is performing
overall for it considers the effects of both temperature
driving force and pressure drop.
By using the two tools in combination, we can clearly
differentiate between inefficiency due to thermal effects and
those due to pressure drop effects.
With one exchanger for example, if the a
max,Dp
is 0.66. It
shows that the exchanger is a poor one. However, it is not
clear that if the problem is one of use of pressure drop or
use of temperature driving force. However, the value of
a
max,h
which is 0.99 tells that the exchanger makes good use
of temperature driving force. So, through the combined use
of the two analyses we can see that the exchanger is poor
because of its use of pressure drop. So by reducing (in this
instance) the pressure drop through the exchanger, the
performance of the heat exchanger and consequently the
performance of network will be improved.
Appendix B. Calculation of chillers’ level after retrofit in
ammonia refrigeration cycle
Heat transfer area in kettles (chillers) is calculated from
Eq. (B.5):
A¼At
l
L;if l4Lthen l
L¼1. (B.5)
So, by comparing the two conditions (after and before
retrofit) in ammonia refrigeration cycle and considering the
fact that Q
1
¼Q
2
, we have
U1A1DTLM1¼U2A2DTLM 2. (B.6)
Since Ug5Ul, We assume
Ug¼0)U1¼U2¼Ul, (B.7)
A2¼A1
DTLM1
DTLM2
. (B.8)
Application of Eq. (B.5) to Eq. (B.8) yields
l2¼l1
DTLM1
DTLM2
, (B.9)
Dl¼l2l1. (B.10)
It should be noted that Dlreal oDl, because in real
condition U
g
6¼0.
References
[1] Alatiqi IM. Online quality control methods for steam-gas reformers.
Int J Hydrogen Energy 1990;15(3):179–85.
[2] Maclean DL, Prince CE, Chae YC. Energy saving modifications in
ammonia plant. Ammonia Plant Saf 1980;22:1–8.
[3] Seturam NK. Energy conservation on ammonia plant. J Fact Tech
Soc 1981;14(1/2):14–7.
[4] Camps JA, Gentry JF. Cogeneration capacity and energy modifica-
tions to existing ammonia plant. Ammonia Plant Saf, AIChE
1987;27:36–42.
[5] Liu M, Zhang N. Proposal and analysis of a novel ammonia–water
cycle for power and refrigeration cogeneration. Energy 2007;32:
961–70.
[6] Sreenivas IS, Rao AP. Energy conservation by process optimization
in ammonia plant. Fert News 1991;36(5):29–35.
[7] Gupta V, Borserio B. Retrofit experiences with a 32 year
old ammonia plant. Ammonia Plant Saf Relat Facilities 2002;42:
95–121.
[8] Mehta BPS. Energy integration in ammonia plant. Fert News
1993;38(3):55–61.
[9] Bhattacharyya NB, Sen B, Mukherjee DK., Energy conservation in
an ammonia plant-role of catalysts. Proc no. R&D (Fert. Assoc.
India) 1984;5:295–301.
[10] Lababidi HMS, Alatiqi IM, Nayfeh LJ. Energy retrofit study of an
ammonia plant. Appl Therm Eng 2000;20:1495–503.
[11] Wang Y, Du J, Wu J, He G, Kuang G, Fan X, et al. Application of
total process energy-integration in retrofitting an ammonia plant.
Appl Energy 2003;76:467–80.
[12] Panje Shahi MH, Ghasemian Langerodi E, Tahouni N. Heat
and power optimization in ammonia plant. In: O’Young L,
Worstell JH, editor. 2006 AIChE annual meeting, San Francisco,
California, Computing and Systems Technology Division
(10A05), Systems and process designs vol. 674 B(10a), 2006.
337pp.
[13] Polley GT, Panjeh Shahi MH, Nunez MP. Pressure drop considera-
tions in the retrofit of heat exchanger networks. Trans I Chem E
1990;68A:211–20.
[14] Polley GT, Panjeh Shahi MH. Interfacing heat exchanger network
synthesis and detailed heat exchanger design. Trans I Chem 1991;E.
69 A:445–57.
[15] De Ruyck J, Lavric V, Baetens D, Plesu V. Broadening the
capabilities of pinch analysis through virtual heat exchanger net-
works. Energy Convers Manage 2003;44(14):2321–9.
[16] Linnhoff B, Dhole VR. Shaftwork targets for low temperature
process design. Chem Eng Sci 1992;47:2081–91.
[17] Dhole VR, Linnhoff B. Overall design of low temperature processes.
Comput Chem Eng 1994;18:S105–11.
[18] Dincer I. Thermodynamics, exergy and environmental impact.
Energy Sour 2000;22:723–32.
[19] Gundersen T. A process integration PRIMER. Norway: SINTEF
Energy Research; 2002.
ARTICLE IN PRESS
M.H. Panjeshahi et al. / Energy 33 (2008) 46–64 63

[20] Townsend DW, Linnhoff B. Heat and power networks in process
design. Part II: design procedure for equipment selection and process
matching. AIChE J 1983;29(5):748–71.
[21] Sorin M, Paris J. Combined exergy and pinch approach to process
analysis. Comput Chem Eng 1997;21:23–8.
[22] Kirova-Yordanova Z. Exergy analysis of industrial ammonia
synthesis. Energy 2004;29:2373–84.
[23] Polley GT, Panjeh Shahi MH. Process integration retrofit subject to
pressure drop constraint. In: Dussemaker HTH, Ledema TD, editors.
European symposium on computer applications in chemical engi-
neering. The Hague, The Netherlands, proceeding 9: Elsevier, 1990;
p. 31–6.
[24] Tjoe TN, Linnhoff B. Using Pinch technology for process retrofit.
Chem Eng 1986:47–60.
ARTICLE IN PRESS
M.H. Panjeshahi et al. / Energy 33 (2008) 46–6464