CHEN 4232: Process Integration

Lecture Note on Module I:

Introduction to Process Integration:

Increased energy cost, increased focus on environmental aspects and generally harder competition have
all contributed to increasing demands regarding efficiency within the process industry during the last
decades. Higher efficiency is often obtained through designs with increased process integration. Simple
examples of process integration are separation and recycling of unreacted feed in the outlet stream from a
reactor (material recycle), or utilizing the excess heat in a hot outlet stream from an exothermic reactor to
preheat the feed using a heat exchanger (energy recycle). An efficient overall solution can usually not be
achieved from one single recirculation, and tight process integration is often required. For instance, all hot
process streams may be integrated with all cold process streams in a heat exchanger network. In addition
to this heat integration, there may be a number of material recycles together with integration of turbines,
heat pumps etc.

Process integration is motivated from economic benefits, but it also changes the characteristics of the
plant: Instead of addressing each process unit independently, the total plant has to be investigated as a
whole (a systems approach). This issue should be particularly emphasized when considering operation
and control of the plant, mainly due to the two following points:
1. Process integration increases interactions in the plant. That is, a change in a disturbance or manipulated
input may not only give local effects, but can spread to a larger part of the plant.
2. Process integration may dramatically change the plant dynamics. As a severe example, integration of
units that are all stable may result in an unstable plant.

In addition, process integration often introduces extra degrees of freedom that may be utilized for
optimization during operation. This may be thought of as a possibility for further savings, but usually the
situation is exactly the opposite: A process design is based on the assumption that the process is operated
optimally without disturbances, and therefore the extra degrees of freedom may have to be used for
optimization during operation in order not to deteriorate the benefits that process integration can give. For
the reasons pointed out above, it is clear that operation of integrated processes can be both difficult and
important for the total economy of the plant. This gives motivation for further research within these two
areas:
1. Design of integrated processes which also are easy to operate /control.
2. Design of operation/control systems for integrated processes.

1
Both of these areas are challenging and they are somewhat related to each other. Despite that the first
point in some cases may be a prerequisite for obtaining good results in the second point;

Process Integration in Chemical Engineering:

Process integration is a term in chemical engineering which has two possible meanings. A holistic
approach to process design which emphasizes the unity of the process and considers the interactions
between different unit operations from the outset, rather than optimising them separately. This can also be
called integrated process design or process synthesis.
Pinch analysis, a technique for designing a process to minimise energy consumption and maximise heat
recovery, also known as heat integration, energy integration or pinch technology. The technique
calculates thermodynamically attainable energy targets for a given process and identifies how to achieve
them. A key insight is the pinch temperature, which is the most constrained point in the process. This
definition reflects the fact that the first major success for process integration was the thermal pinch
analysis addressing energy problems and pioneered by Linnhoff and co-workers. Later, other pinch
analyses were developed for several applications such as mass-exchange networks water minimization
and material recycle. A very successful extension was "Hydrogen Pinch", which was applied to refinery
hydrogen management .This allowed refiners to minimise the capital and operating costs of hydrogen
supply to meet ever stricter environmental regulations and also increase hydrotreater yields.
In the context of chemical engineering, process integration can be defined as a holistic approach to
process design and optimization, which exploits the interactions between different units in order to
employ resources effectively and minimize costs.
Process integration is not limited to the design of new plants, but it also covers retrofit design (e.g. new
units to be installed in an old plant) and the operation of existing systems. It is said that with process
integration, industries are making more money from their raw materials and capital assets while becoming
cleaner and more sustainable.
The main advantage of process integration is to consider a system as a whole (i.e. integrated or holistic
approach) in order to improve their design and/or operation. In contrast, an analytical approach would
attempt to improve or optimize process units separately without necessarily taking advantage of potential
interactions among them.
For instance, by using process integration techniques it might be possible to identify that a process can
use the heat rejected by another unit and reduce the overall energy consumption, even if the units are not
running at optimum conditions on their own. Such an opportunity would be missed with an analytical
approach, as it would seek to optimize each unit, and thereafter it wouldn’t be possible to re-use the heat
internally.

2
Typically, process integration techniques are employed at the beginning of a project (e.g. a new plant or
the improvement of an existing one) to screen out promising options to optimize the design and/or
operation of a process plant. Also it is often employed, in conjunction with simulation and mathematical
optimization tools to identify opportunities in order to better integrate a system (new or existing) and
reduce capital and/or operating costs.
Most process integration techniques employ Pinch analysis or Pinch Tools to evaluate several processes
as a whole system. Therefore, strictly speaking, both concepts are not the same, even if in certain contexts
they are used interchangeably. A review explains that in the future, several trends are to be expected in
the field. In the future, it seems probable that the boundary between targets and design will be blurred and
that these will be based on more structural information regarding the process network. Second, it is likely
that we will see a much wider range of applications of process integration. There is still much work to be
carried out in the area of separation, not only in complex distillation systems, but also in mixed types of
separation systems. This includes processes involving solids, such as flotation and crystallization. The use
of process integration techniques for reactor design has seen rapid progress, but is still in its early stages.
Third, a new generation of software tools is expected. The emergence of commercial software for process
integration is fundamental to its wider application in process design.

A plant transfers raw materials into products by means of chemical reactions, separation steps etc. Often a
plant comprises several feed streams being converted into a main product together with wanted or
unwanted byproducts in a complex process structure. In addition, utility systems for thermal energy,
mechanical and electrical energy, pneumatics and so on are required. In the subsequent, the term utility
refers to thermal energy, i.e. addition and removal of heat.

The purpose of a plant is to manufacture the desired products as economically as possible. However,
government regulations (e.g. regarding environmental pollution), safety and health demands also have to
be fulfilled. Three tasks that have a major impact on the overall properties of a plant are the selection of:
1. Reaction path and reactor(s).
2. Recycle and separation network.
3. Heat exchanger network and utilities.

There are two essentially different approaches to process design:

1. Sequential development of a process structure.
In this approach, the designer starts by making decisions about the most important part of the

3
process and then moves to less important parts. In process design the “onion model”
described by Smith and Linnhoff (1988) is usually followed, see figure.1.

Figure 1. Onion diagram in process design.

The design starts with the “core” of the process which is the reactor. When reaction path and
reactor have been selected, the designer can move outwards in the onion and make decisions
on separation and recycle system, and so on. The flowsheet is developed sequentially and
there is no guarantee that a series of optimal local decisions leads to an optimal overall
process. For example, decisions made in the reactor and separation layers may give little
potential for heat recovery in the outer layers. The designer may then go back to the inner
layers and investigate other alternatives, hoping to find an overall solution at least close to the
global optimum. Despite the danger of being trapped in local optima and the work required
in investigating several alternatives, this approach also has advantages: The designer (or
design team) interacts actively in the design process and can use his/her knowledge and
experience. With the designer in command of the decisions taken a large variety of
knowledge can be considered, but perhaps more important - the designer learns from the
design process and will become a better designer in subsequent projects.

2. Simultaneous development of a process structure.

The second approach is based on a superstructure of the process. A superstructure (or hyperstructure)
includes all process structures that will be considered. That is, the designer starts by setting up all possible
reactors, all possible recycles and separators and so on. This results in a large number of candidate
process structures and the design problem is formulated as a mathematical model which is solved by

4
means of optimization. The approach usually result in an MTNLP problem where the integer part
represents discrete decisions such as whether a processing unit is present or not. (Note that the term
“integer” in MTT.P or MINLP problems often actually refers to binary variables and not to integers).
Logical propositions such as “choose exactly one reactor’ or “if reactor A is chosen, then separator D
cannot be chosen” are formulated as mathematical constraints and included in the model.

This second approach will not be trapped in local optima due to sequential decisions as in the
first approach. However, the second approach also have disadvantages: The optimal process
structure may not be embedded in the superstructure given initially by the designer. Provided
that the optimal structure is embedded within the superstructure, the numerical solution
algorithm may fail to find the global optimum. In general such problems are non-convex and
there is no guarantee that the global optimum will be found. Thus, this approach may also
result in a local optimum. In addition, the combinatorial nature of such problems may impose
severe limitations for the simultaneous approach when applied to medium or large size
problems. The most serious disadvantage, however, is probably that the designer does not
play an active role in the decision making. This means that topics like safety and health
considerations or e.g. costs related to adaptation to existing infrastructure hardly will be
considered in the conceptual design stage. Such considerations may be done by an
experienced designer in early decisions of the sequential approach, but will be more difficult
to implement in a optimization model required for the simultaneous approach. Also, the
learning of the designer will be less prominent in the second approach. Despite the fundamental
differences between the sequential and simultaneous approaches, ost practical designs will have elements
of both. The difference between sequential and simultaneous methods may not only be applied to
flowsheeting of a complete plant, it may be equally relevant for each subtask. In the next section, it will
be clear that the main difference in synthesis of HENs is also between sequential and simultaneous
methods. The rest of this section is devoted to decomposition of the HEN from the remaining parts of the
plant and to explain why good operation and control of the HEN is important for the overall process.
These issues are illustrated by an example. The example has been made as simple as possible while
retaining the essential properties of a typical integrated plant.
Consider the plant in figure 2 where decisions regarding reaction path, reactors and separation have been
made. The process does not have any material recycles.

5
 Fig.2 Simple process with two reactors and a separator.

The two feed streams require heating and PRODUCT 1 requires cooling. Table 2.1 gives the data
necessary for considering heat recovery between these three streams. Note that the classification of hot
and cold streams is based on whether a stream requires cooling or heating, it does not depend on
temperature.

Table 1. Thermal stream data for the simple heat recovery problem.

Using the data in Table1, focus here is on the heat recovery problem only. The heat
recovery problem has been decomposed from the rest of the process. An obvious way to
satisfy the heating and cooling demands for this problem is to install two heaters and one
cooler as shown in figure.3. The total heating and cooling duties are satisfied using
external utilities (steam and cooling water) and the requirements are QH= 17.5 MW and
Qc = 16.0 MW for heating and cooling, respectively. This solution does not involve any heat
recovery.

6
 Fig 3. Trivial solution using external utilities only.

Figure 4 shows one possible HEN for the simple process. The countercurrent arrangement
with hot streams on top from left to right and cold streams below from right to left (grid
diagram) was introduced by Linnhoff and Flower (1978).In this simple HEN, two process-to-process heat
exchangers (I and II) have been introduced and stream C2 does not need external heating. Compared to
figure 3, the utility duties have been reduced to QH = 8.0 MW and Qc = 6.5 MW while the number of heat
transfer units has increased from three to four. Note that the reduction of both hot and cold utilities is
equal to the total process-to-process heat transfer (QI + QII = 9.5 MW). This significant reduction in
external energy consumption due to process integration is of course important. Linnhoff and Turner
(1981) reports results from industrial applications where savings in both energy and capital costs have
been achieved through heat integration.

Fig. 4 HEN for the simple process.

The HEN in figure.4 reduces utility consumption, however, it also introduces interactions.
As an example, a disturbance in the supply temperature of stream HI will not only affect the
outlet temperature of the same stream but it will also affect the outlet temperatures of the two
cold streams as this disturbance also will travel from hot to cold side of process-to-process

7
heat exchangers. To see how these interactions in the HEN may influence the dynamics of
the overall plant, the complete process including the HEN is drawn in figure 5.

Fig 5 Complete plant with heat integration.

We see that the interactions in the HEN causesfeedback in the plant, e.g. the exit temperature
from REACTOR 1 will affect the inlet temperature of the same reactor through the heat
exchangers. This may significantly change the overall plant dynamics, and may lead to slower
response, increased sensitivity to disturbances and even instability. Of course, feedback may
also give desired effects (e.g. feedback control). However, feedback introduced by heat
integration makes it more difficult to deduce the plant dynamics from the dynamics of the
individual plant units.

As illustrated by figure 4 and figure 5, a HEN can be considered from two different points of view:
View point 1. A HEN is a part of the process with its own properties such as heat recovery level and
dynamic behavior, and it can be considered separately from the rest of the process.

View point 2. A HEN affects the overall plant properties as it connects the plant [Link]
of the HEN as a separate unit has little interest.
The first point of view which focuses on the HEN as a separate unit is often employed in the synthesis of
HENs. That is, the stream data is first selected from reactor, separation system and possibly recycle
considerations (sequential approach). Then, the HEN is designed as a separate task even if it is clear that
the optimal supply and target temperatures may change somewhat for the overall process. Regarding the

8
dynamics and control of an integrated plant, it could be easy to draw the conclusion that the second point
of view has to be applied since it is the overall plant dynamics that is important. This conclusion is not
quite true despite the fact that the dynamic behavior of the HEN itself may not be important. Consider an
integrated plant where a separate control system has been designed for the HEN. Assume that the control
system yields perfect control meaning that the outlet temperatures from the HEN are kept exactly at
their target values (setpoints) also when disturbances are present. Now, the controlled HEN will not
introduce any interactions or feedback in the overall plant. That is, if perfect control of the HEN is
obtained, the overall plant dynamics will be as if heat integration was not present. However, the steady
state benefit of heat integration in terms of reduced energy costs will be preserved. In practice, perfect
control can not be achieved but we can conclude that “good” control ofthe HEN is desired as this will
give “small” interactions and “small” effects of unwanted feedback in the overall plant. Good control of
the HEN has a decoupling effect on the dynamics of the overall plant and will simplify control system
design for the rest of the process. Of course, controlling the outlet temperatures from the HEN is also
important since deviations from the target temperatures may affect for example reactor conversion or
quality of separation in downstream units. In conclusion, dynamics and control of the HEN are
important for the overall plant behavior.

HEN synthesis, a quick overview The total design effort (on a systems level) required for a HEN typically
involves the following three stages:

1. Nominal design. Synthesize one or more networks with good properties (regarding heat
recovery and capital cost) for nominal stream data.

2. Flexibility and controllability. Investigate the networks with regard to flexibility and controllability,
and possibly introduce some modifications (e.g. increased area) such that at least one HEN shows
satisfactory results.

3. Operation. Design a control system to operate the HEN properly. This involves control structure
selection and possibly some method for on-line optimization.

HEN synthesis is often regarded as the most mature research field within process design, and progress
during the last decades has been made in parallel by the two main “schools”; pinch technology and
mathematical programming.

9
Pinch Analysis;

An important concept within pinch technology is the merge of all hot process streams and all
cold process streams into one temperature/heat diagram, known as the composite curves. The
stream data required to draw the composite curves for the simple problem used in the
previous section is given in table 1, and the data will be used as a simple example. The
composite curves for this example are shown in figure 6.

In this simple example, there is only one hot stream and the hot composite curve (upper line)
is trivial. Both of the cold streams are present in the temperature range from 80°C to 130°C,
and the slope of the cold composite curve is 1/(0.15+0.05) = 5.0 [°C/MW] in this temperature interval.
The horizontal axis is change of heat, thus the horizontal positions of the curves are
not unique. The vertical difference between the composite curves denotes the driving forces
for heat transfer. For the solid composite curves in figure 6, the smallest temperature
difference between the curves, ∆Tmin, is selected to 20°C. At the highest temperature, the

10
process needs addition of hot utility QH (e.g. steam) and at lowest temperature the process
needs removal of heat to cold utility QC (e.g. to cooling water). In die region along the
horizontal axis where both curves are present, heat may be recovered between hot and cold
process streams. The point where the composite curves are closest to each other is known as
the pinch. With the smallest driving forces, the pinch represents the bottleneck for heat
transfer. The pinch also decomposes the heat recovery problem into two sub problems; (1) above pinch
where only hot utility is needed, and (2) below pinch where only cold utility is
needed. Transferring heat across the pinch using process exchangers will increase both hot
and cold utility accordingly (“double penalty”) and should be avoided.
Before the pinch design method can be applied, a value of ∆Tmin has to be selected. This
specification actually trades off utility consumption (operating cost) against capital cost (e.g.
heat exchanger area). It is thermodynamically feasible to let the composite curves touch at
the pinch (dash-dotted cold curve in figure 6), yielding the lowest external utility consumption possible.
With no driving forces at the pinch, however, infinite heat exchanger area is required and the resulting
HEN is definitely not cost optimal. A reasonable value of ∆Tmin can be found from optimizing the trade-
off between utility consumption and capital cost. This is known as super targeting,

The pinch design method consists of three stages:

1. Targeting
2. Synthesis
3. Optimization.
.
Targeting:

An important feature related to HEN synthesis is that it is possible to quantify targets for
minimum utility consumption, minimum number of units and minimum area ahead of the
actual synthesis stage. Minimum utility consumption (Q H min and QCmin ) can be found using
e.g. the problem table algorithm or the heat cascade, or it may be found graphically from the
composite curves (when they are aligned according to the specified value of ∆Tmin). From
figure 6, the targets for utility consumption are identified to Q Hmin = 5.5 MW and QCmin
= 4.0 MW. The targets for utility consumption are constrained by the value of ∆Tmin.
The target for number of units normally arises from the “N-1” rule: Minimum number of
units (including utility exchangers) is equal to the number of streams (including utilities)
minus one. This rule is based on graph theory - no thermodynamic considerations are made,
and there are cases where the rule does not give a correct target value. A rigorous target value

11
for minimum number of units may be found using mathematical programming. For the
simple example, the target for number of units (for the complete HEN) is 3+2-1 = 4.
Applying the N-1 rule below and above the pinch separately, the target for minimum number
of units for the decomposed HEN is (3+1-1) + (2+1-1) = 5.

A target value for total area may be found assuming vertical heat transfer. This target,
however, is not compatible with the target for minimum number of units. Cost optimal HENs
often have close to minimum number of units, and the area will often deviate from the target.

Synthesis.

In this step, a HEN that satisfies the target for utility consumption (maximum energy
recovery) is designed. The design starts where the process is most constrained, at the pinch,
and is carried out separately above and below pinch. ∆Tmin is the agreed minimum approach
temperature for each heat exchanger, and exchangers are placed between streams such that
this requirement is fulfilled. To achieve this, stream splitting may be necessary. The duty of
each heat exchanger is made as large as possible, constrained by the stream with the smallest
heat demand. This “tick-off’ rule minimizes the number of units in the design. Utility
exchangers are placed on streams that do not meet the target temperatures when using only
process exchangers. Figure 7 shows the maximum energy recovery HEN for the example
and the pinch is shown by the dashed vertical line. The utility targets are met and the number
of 5 units corresponds to the target for the decomposed HEN.

Fig.7 Maximum energy recovery network (∆Tmin = 20°C).

Optimization;

12
In this stage, the maximum energy recovery HEN from the synthesis stage is improved in order to achieve
a more cost optimal HEN. All pinch designs from the synthesis will have at
least one unit more than the minimum number when the complete HEN is considered.
Removing units may reduce the total HEN costs. The network in figure 7 has one unit
above the target for the complete HEN, and excess units form loops. For this example, there
is only one loop:

The units are physically connected by the streams as indicated above. To form a loop, the
first and last stream must be identical. A loop always consist of an even number of units, and
every second unit may increase and decrease its duty by the same amount of heat while
maintaining the heat balance in the HEN. Manipulating a loop this way represents a degree of freedom in
the design stage, and it may be utilized to remove a unit. A heuristic rule in
pinch design is to remove the unit with the smallest duty. Applied to the example, heater h2
is removed by reducing the duty to zero. This results in increased duties of 2.5 MW for heater
hi and exchanger II and the same reduction in exchanger I. Examining the resulting HEN,
one will find that the constraint on ∆T min is violated (hot side of exchanger I). ∆T min is
restored by exploiting paths in the HEN. In the example, there is a path from hot utility
stream (ST) to the cold utility stream (CW, cooling water).

A path is manipulated similarly as a loop; every second unit is manipulated by opposite duties
of equal magnitude. A path always comprises an odd number of units, meaning that the first
and last unit (utility exchangers) are changed in the same direction. In the example, the hot
outlet temperature of exchanger I is increased by reducing the duty of that exchanger. Hence,
∆Tmin can be restored by reducing the duty of exchanger I which implies increased duties of
the heater and cooler. In general, restoring a violation of ∆Tmin implies that the duty of both
utility exchangers in a path are increased. The completed HEN after ∆Tmin is restored is
shown is shown in figure 4. The composite curves for the finished HEN are shown in figure
6 (dashed cold line). Note that the smallest temperature difference between the composite
curves are larger than the ∆Tmin = 20°C at the hot end of exchanger I. The reason for this is

13
that the completed HEN includes non-vertical heat transfer and heat is transferred across the
pinch. This increases utility consumption but saves one unit.

In general when using pinch analysis, the designer is faced with a number of choices in the
synthesis stage and particularly in the optimization stage. In practice, a number of alternative
designs should be carefully investigated with respect to utility consumption, capital cost (e.g.
area), operability aspects etc. before the final HEN design is settled.

In operation of a given HEN, we want the utility consumption to be as small as possible. The
following relations between synthesis and operation can be given at this point:
Loops. Loops may be removed in the design stage, but the completed HEN design may
still include one or a few loops. As it was clear from above, manipulating loops do not
change the total utility consumption, hence we may expect that loops do not play any
important role in operation. Loops (particularly those involving only process exchangers)
introduces feedback internally in the HEN and this may change the dynamic properties of
the HEN.
Paths. Manipulating a path changes the utility consumption, and manipulating paths for
varying disturbances etc. to minimize utility consumption is important during operation.

Composite curves. While composite curves are fixed during design (when the heat
recovery level is specified), they may change during operation due to variations in supply
temperatures, heat capacity flowrates etc. In addition, the specification on ATmm (for tradeoff between
capital and operating costs) has no relevance during operation. The purpose
during operation is to vary the manipulations in order to keep the composite curves as
close to each other as possible when disturbances are present and change the shape of the
composite curves.

In the remaining part of this section, the most important methods for HEN synthesis will be mentioned. A
major difference of existing methods for HEN synthesis is between sequential and simultaneous
approaches, as it was for the design of the overall plant. Methods for synthesis of HENs can be divided
into four main categories.

1. Pinch design. Thermodynamic based sequential method.

2. Sequential approach based on mathematical programming.
3. Simultaneous approach based on mathematical programming.
4. Merge of approaches 1,2 and 3.

14
In the pinch design method, the designer participates very interactively in the HEN design. Not only are
the three stages (targeting, synthesis and optimization) carried out sequentially, but the synthesis stage
itself is indeed sequential. The designer places the heat exchangers, one by one, starting at the pinch and
moving away until the maximum energy recovery HEN is completed.

Pinch Analysis:

Pinch analysis is a methodology used to minimize energy consumption of chemical processes by

optimizing energy recovery methods. A pinch occurs at the closest point of approach between hot and
cold streams in the network. By designing around this point energy from utilities can be minimized.

Pinch Technology:

For maximum energy recovery (MER), energy should not be transferred across the pinch. The network is
separated into hot and cold regions, above and below the pinch respectively. In the hot region, streams are
matched so that heat is transferred from the coolest possible source. Below the pinch, the opposite occurs.
This yields the minimum utility requirement solution to the heat exchanger network. Minimum utility
requirement is not necessarily the optimum solution, however, the solution with the lowest overall cost is.
Minimizing the size and quantity of heat exchangers plays a large role in decreasing capital costs. There
is a trade-off between operational and capital costs. In some cases transfer across the pinch is necessary to
produce the optimal network.

Minimum approach temperature

The minimum approach temperature represents the smallest gap in temperature across which heat transfer
will occur in the system. This often only occurs at the pinch. While technically any value greater than
zero can allow for heat transfer, very small values are not often feasible. Low temperature differences
decrease the need for additional utilities, but require increasingly large heat transfer areas, meaning larger
heat exchangers. The trade-off between operating costs and capital costs should be used to choose the
minimum approach temperature for the network. Typical choices for minimum approach temperatures are
between 5 °C and 30 °C

Pinch technology is a highly used and advanced technology throughout industry. Because of this, there
have been devised multiple approaches with which to solve a pinch system.
Composite curves method

15
This method uses a graphical approach and makes obvious the presence and location of the pinch in the
system. First one combines all hot streams into single composite curve on a temperature-enthalpy graph.
The same is then done with the cold streams. Plotting both composite curves on a temperature-enthalpy
plot reveals a pinch at the point where the cold and hot curves are closest in temperature. Because change
in enthalpy is the feature of concern, the composite curves can be freely shifted along the enthalpy axis.
In this manner the minimum approach temperature can be satisfied at the pinch.
Problem table method

The problem table method, also known as the cascade method, forgoes graphs for a numerical approach.
This method is useful when solving heat-exchanger network problems manually, but has fallen into disuse
due to the advent of modern computing. The first step in the process is converting stream temperatures
into interval temperatures. To do this, one subtracts half the minimum approach temperature from the hot
stream temperatures, and adds half to the cold stream temperatures. The intervals are then ranked in order
of magnitude, ignoring any repetitions. The next step involves balancing the heat carried by streams
between each interval. Any excess heat cascades over into the next lower down interval. Add heat from
the top of the system eliminate any negative values. The amount added at the top corresponds to hot
utility duty, while the remaining value at the bottom corresponds to heat that must be removed by cold
utility.

16
Example of the problem table method. The right most column contains both the minimum heating
requirement for the system at the top, and the minimum cooling requirement at the bottom
Computer tools for HEN design

Presently, the two previously described methods of pinch analysis are not in wide use in industry; rather
analysis is performed by specialized software. These software include: Aspen HX-Net™ from Aspen
Technology Inc., SUPERTARGET™ from Linnhoff March Ltd. and UniSim™ ExchangerNet™ from
Honeywell International Inc. (Towler and Sinnott 2013). Stream data is often pulled directly from the
process simulations by the software package. Automatically generated networks must be properly
constrained lest an unfeasible amount of stream splits and loops be generated. There is great value,
however, in the ability of software packages to perform many tweaks and iterations to fine tune a
network.
Example problem
In this example problem we will use pinch analysis via the problem table method to design a heat
exchanger network given the following stream table, taken from Towler and Sinnott, and a desired
minimum approach temperature of 20°C.

17
 Number Heat Capacity Flow Rate, Source Temp. Target Temp. Heat Load,
Stream
Type kW/°C °C °C kW
1 hot 40.0 180 40 5600
2 hot 30.0 150 60 1800
3 cold 60.0 30 180 9000
4 cold 20.0 80 160 1600

The first step is to calculate the interval temperatures. Because the minimum approach temperature is
20°C, the interval temperatures for the hot stream are 10 °C less than the actual temperature and those for
the cold stream are 10°C more than the actual temperature. The next step is to rank the interval
temperatures, ignoring any duplicated values. We then determine what streams belong in which interval.
The sum of CP values for the cold streams minus that for the hot streams can then be multiplied by the
interval ΔT to give the interval ΔH, and the interval ΔH values can be cascaded to give the overall heat
flow.

The amount of heat added at the top to remove any negative values, 2900 kW, is the hot utility. The 600
kW left over at the bottom is the cold utility required. The pinch occurs where the heat transferred is zero,
that is at interval number 4, interval temperature 90 °C. This is equivalent to 100°C in the hot stream or
80°C in the cold stream.

Now we need to design the actual system:

Above Pinch

1. CPh ≤ CPc
2. It is possible to match stream 1 or 2 with stream 3 but neither stream can match with stream 4.
This is problematic, since if we match stream 1 with 3 then stream 2 will not be able to make a

18
 match at the pinch. Likewise, if we match stream 2 with 3 then stream 1 will not have a match at
the pinch.
3. Check the heat available in bringing the hot streams to the pinch temperature.

stream 1 ΔH = 40.0(180 − 100 )= 3200 kW

stream 2 ΔH = 30.0 (150 − 100 ) = 1500 kW
4. Check the heat required to bring the cold streams from the pinch temperature to their target
temperatures.

stream 3 ΔH = 60.0(180 – 80)= 6000 kW

stream 4 ΔH = 20.0(160 – 80) = 1600 kW

5. If we split stream 3 into two branches with CP of 40.0 and 20.0, then we can match the larger
branch with stream 1 and transfer 3200 kW, which satisfies (ticks off) stream 1.
6. We now have two cold streams, both with CP of 20.0, and one hot stream (2) with CP of 30.0.
We need to split stream 2 into two branches. As an initial guess these can both have CP of 15.0.
We can then match one branch of stream 2 with the smaller branch of 4 and transfer 750 kW, and
the other branch with stream 3, also for 750 kW, which then ticks off stream 2.
7. Include a heater on the larger branch of stream 3 to bring it to its target temperature:

ΔHhot = 40 (100 ) − 3200 = 800 kW

8. Include a heater on the smaller branch of stream 3 to provide the balance of the heat required:

ΔHhot = 20 (100 ) – 750 = 1250 kW

9. Include a heater on stream 4 to provide the balance of the heat required:

Δ Hhot = 1600 − 750 = 850 kW

10. Check sum of heater duties = 800 + 1250 + 850 = 2900 kW = hot utility target.

Below Pinch

1. CPh ≥ CPc
2. Note that stream 4 starts at the pinch temperature and so cannot provide any cooling below the
pinch.
3. We cannot match stream 1 or 2 with stream 3 at the pinch.
4. Split stream 3 to reduce CP. An even split will allow both streams 1 and 2 to be matched with the
split streams adjacent to the pinch, so try this initially.

19
5. Check the heat available from bringing the hot streams from the pinch temperature to their target
temperatures:

stream 1 ΔH = 40.0 (100 − 40 ) = 2400 kW

stream 2 ΔH = 30.0 (100 − 60 ) = 1200 kW
6. Check the heat required to bring the cold streams from their source temperatures to the pinch
temperature:

stream 3 ΔH = 60.0 (80 − 30 ) = 3000 kW

Stream 4 is at the pinch temperature

7. Note that stream 1 cannot be brought to its target temperature of 40 °C by full interchange with
stream 3 as the source temperature of stream 3 is 30 °C, so ΔTmin would be violated. So transfer
1800 kW to one leg of the split stream 3.
8. Check temperature at exit of this exchanger:

Temp out= 100 – 1800/40= 55 °C, satisfactory

9. Provide cooler on stream 1 to bring it to its target temperature; the cooling needed is

ΔHcold = 2400 − 1800 = 600 kW

10. Transfer the full heat load from stream 2 to second leg of stream 3; this satisfies both streams.

20
The role of thermodynamics in process design:

How can we apply thermodynamics practically?

Most of us involved in engineering design have somewhat unhappy memories thinking back to
thermodynamics in college days. Either we did not understand, gave up hope that we ever would, and
remember with dread the horror that struck on examination day. Alternatively, we were amongst the
chosen few whose photographic memory would allow us to reiterate the definitions of entropy, Gibbs free
energy and all those differential equations faultlessly, but without real understanding. Afterwards, we
could never help asking ourselves: what is it all for? What do I do with it? In the best of cases,
thermodynamics seemed to be a fascinating science without a real application.

Pinch analysis is based on straightforward thermodynamics, and uses it in a practical way. However, the
approach is largely non-mathematical. Although (classical) thermodynamics itself may be a thoroughly
developed subject, we need to apply it the context of practical design and operation. This is the aim of the
following chapters. We distinguish between “inevitable” and “avoidable” thermodynamic losses, and
“practical” or “ideal” performance targets, to achieve both energy savings and other process benefits.

Capital and energy costs

sometimes, it is believed that energy recovery is only important if energy costs are high and capital costs
are low. Consider, for example, Figure 1.6, which shows a heat exchanger network that would seem
appropriate to most when energy is cheap and capital expensive. There is no process heat recovery – only
utility usage. Conversely, Figure 1.7 shows a network which might seem appropriate when energy is
expensive. There is as much process heat recovery as is possible in preference to utility usage. The
implicit assumption is that heat recovery (instead of utility use) saves energy but costs capital.
Consider now Figure 1.8. This shows a simpler network which still achieves maximum energy recovery.
Based on a uniform heat transfer coefficient and sensible steam and cooling water temperatures, the total
surface area for both designs has been evaluated. To our surprise, the “network for minimum capital cost”
turns out to have the higher total surface area, and is more expensive in capital cost as well as operating
(energy) cost!

From this example we realise that in networks there are two basic thermodynamic effects influencing
capital costs. One is the effect of driving forces and the other is the effect of heat loads. Evidently, as we

21
go to tighter designs (i.e. to reduce driving forces) we need less utility and the overall heat load decreases.
Capital cost then increases with reduced driving forces (we all know that) but decreases with reduced heat
load (we rarely consider this point). The design without process heat recovery in Figure 1.6
handles twice as much heat as is necessary. As a result, capital costs are increased even
though the driving forces are large!

Fig 8 (1.6) Outline heat exchanger network for “minimum capital cost

22
 Fig: 9 (1.7) Outline network for “minimum energy cost”

Although this is obviously a contrived example, it helps to shows that there is not necessarily a trade-off
between energy and capital cost, and helps to explain the frequent capital savings (as well as energy)
observed in practical case studies. Thermodynamics-based techniques can help in many other ways. For
example, the analysis of driving forces may be used not to reduce them but to distribute them. differently.

23
 Fig 10 ( 1.8.) Optimal network for minimum energy cost

This can help to clarify options in design, say, for better operability and/or lower capital costs at a
constant level of energy recovery.

24