Applied Thermal Engineering 43 (2012) 118e127

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Heat pump integration in a cheese factory

Helen Becker a, *, Aurlie Vuillermoz b, Franois Marchal a
a
b

Industrial Energy Systems Laboratory, Ecole Polytechnique Fdrale de Lausanne, CH-1015 Lausanne, Switzerland
EDF R&D, Eco-efciency and Industrial Process Department, Centre des Renardires, F - 77818 Moret sur Loing Cedex, France

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 15 August 2011
Accepted 23 November 2011
Available online 10 December 2011

The aim of process integration is to increase the efciency and to reduce the energy consumption, operating
costs and CO2 emissions of an industrial process. The proposed methodology is applied to a real case study
of a cheese factory with non-simultaneous process operations and uses the time average approach
combined with restricted matches. This work focuses on appropriate heat pump integration and two
different integration strategies are proposed. In the rst option, process modications and direct heat
exchange between the process and heat pump streams are not allowed, which means that the integration of
heat pumps has to be realized through intermediate heat transfer networks. On the contrary, in the second
option, direct heat exchange and process modications are possible. The results of both options are
analyzed and compared for the French and German context. Depending on the industrial constraints and
the location, it is shown that the nal choice of new heat pump installations may be different. Saving
potential in operating costs is higher for option 2 where the cost savings can be higher than 40% for both
countries. Furthermore the potential CO2 emissions and primary energy savings are compared.
2011 Elsevier Ltd. All rights reserved.

Keywords:
Pinch analysis
Utility integration
Industrial heat pumps
Total site
Restricted matches
Time average approach

1. Introduction
When studying the energy efciency of an industrial process,
the analysis of heat pump integration has to be considered as part
of a complete methodology, beginning from the data collection and
modeling of all process unit operations up to the nal utility integration. The use of pinch analysis techniques allows identication
of heat recovery potential in the process. Opportunities to integrate
heat pumps and other utility systems can be identied by analyzing
the grand composite curve. The curve shows the enthalpy
temperature prole of the heat to be supplied to the process and of
the heat excess to be evacuated by a cold utility. Heat pump integration became interesting in the late 80s and the early 90s due to
increasing fuel prices. For example, the rules for optimal placements of a heat pump in an industrial process have been introduced
by Linnhoff and Townsend [8] in 1983. Later, Wallin and Berntsson
[13] demonstrated that, characteristics of both, industrial process
and heat pumps, must be taken into account. Kapustenko et al. [5]
analyzed heat pump integration based on selected streams of
a cheese factory. Also Pavlas et al. [11] analyzed heat pump integration for a gasication process. However, both approaches are
limited to ammonia refrigeration cycles. Recently Kapil et al. [4]

* Corresponding author.
E-mail address: helen.becker@ep.ch (H. Becker).
1359-4311/$ e see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2011.11.050

proposed a methodology for low grade heat recovery, by

combining the total site approach with heat recovery models such
as heat pumps, ORCeORC or absorption refrigeration. The main
disadvantage of their method is that self-sufcient pockets are not
considered and as result not all heat pump opportunities may be
identied. More generally, the potential of industrial heat pump
integration was demonstrated by Becker et al. [3]: a mixed integer
linear programming (MILP) formulation of the heat cascade is used
to optimize simultaneously the ow rates in heat pumps and other
utility systems.
Two process integration options will be analyzed in this paper:
rst, process modications are not allowed and a newly integrated
heat pump cannot exchange directly with the process. In the
second option, process modications and direct heat pump process
integration are possible. Saving potential becomes higher, but also
the complexity of process congurations increases. When utilities
(e.g. heat pump in the rst option or co-generation unit) cannot
exchange heat directly with the process, heat exchange restrictions
have to be included and the method proposed by Becker and
Marchal [2] can be applied.
In the food industry, most of the process operations are performed in batch mode. However, consideration of different
temperature levels and assumption of heat storage through available hot water tanks, makes the heat recovery between periods
possible. The time average approach ([6,7],) can be applied to
realize the process integration analysis, where heat loads will be

H. Becker et al. / Applied Thermal Engineering 43 (2012) 118e127

Nomenclature

DTmin=2
Q_
Q_ *
A
B
COP
du
dtot
ep
f
InvC
M
mCO2
OpC
PR
T
U
eel
ef

Minimum delta T/2 [K]

Heat load [kW]
Instant Heat load [kW]
Heat exchanger area [m2]
Annual benet [V/year]
Coefcient of performance [-]
Yearly operating hours of unit u [h/year]
Yearly operating hours of the process [h/year]
Primary energy [MJ/t]
Installation factor for estimating investment costs [-]
Investment costs [V]
Yearly tons of product [t/year]
CO2 emissions [kg/t]
Operating costs [V/t]
Payback time [year]
Temperature [K]
Heat transfer coefcient [kW/m2]
Specic electricity consumption [kW h/t]
Specic fuel consumption [kWh/t]

expressed by a constant mean value during the operating time over

all periods. The concept of restricted matches between subsystems, introduced by Becker and Marchal [2] can also be used
to restrict the direct heat exchange between non-simultaneous
process operations. The nal purpose of this paper is to realize
a process integration analysis for the previously described options
in a cheese factory and to integrate heat exchange restrictions at
different levels (e.g heat exchange restrictions between process
units not working at the same time and heat exchange restrictions
between all process units and the heat pump).
2. Process integration methodology
2.1. Introduction to the case study
The industrial process that is used for this analysis is a cheese
factory. The main process operation steps are summarized on Fig. 1.
The current energy bill corresponds to the energy consumption of
2895 kWh of natural gas and 194 kWh of electricity for one ton of
produced cheese.

119

Subscripts
c
Cold streams
cog
Co-generation engine
h
Hot streams
hp
Heat pump
ref
Reference value
u
Index for unit
Superscripts

Entering the system


Leaving the system
max
Maximum value
min
Minimum value
Acronyms and convention
MILP
Mixed integer linear programming
MVR
Mechanical vapour compression
TVR
Thermal vapour compression
bold characters Optimization variables
small letters Specic quantities

previous section, rst the relevant process operations for the

energy integration are identied. It is important to concentrate only
on signicant process operation units, in order to keep the problem
as simple as possible and not to waste time for modeling non
signicant heat exchange requirements [10]. The overall process
including auxiliary units (e.g. hot water production, space heating,
cleaning in place,..) are considered. Several pasteurization units are
used to remove bacteria from milk, cream or water. The evaporation
unit, one of the main consumers, consists of 5 effects and one
thermal vapour re-compression. Before entering the evaporation in
the rst effect, the whey is rst preheated to reach predened
operating conditions that leads to a rst evaporation. Then, the
remaining liquid is sent to the second effect at a lower pressure, and
the vapour boiled off in the rst effect is recovered to provide heat
to the second effect. The same operation is repeated in the
following effects. A part of the steam from the third effect is reused
in the thermal vapour re-compression (TVR) driven by high pressure steam. The remaining heating and cooling requirements
concern process units like forming, product rening, packaging or
cold stores.

2.2. Process operation units

The rst step is to identify the most signicant energy
consumers. Starting from the global energy bill given in the

Fig. 1. Process description.

2.2.1. Process requirements and their representations

The applied approach includes the process analysis methodology of Muller et al. [10]. Once the process operation units are
identied, they can be modeled in order to dene process heat
requirements and the corresponding hot and cold streams for the
process integration. Furthermore, all process and waste streams

Fig. 2. Triple representation of a process requirement [9].

120

H. Becker et al. / Applied Thermal Engineering 43 (2012) 118e127

leaving the system are systematically cooled down to the ambient

temperature, in order to maximize the heat recovery potential.
Fig. 2 summarizes the possible representations to dene heat
requirements of process operation units.
 The thermodynamic representation corresponds to the process
heat requirements where the hot streams have the highest
possible temperature and the cold streams the lowest possible
temperature. The enthalpy temperature prole results from the
thermodynamic analysis of the process operation unit. This
representation is chosen when the process and heat supply can
be modied.
 The technology representation denes hot and cold streams, as
they are realized in the process. The unit is modeled as a blackbox and only heat supply through heat exchangers can be
modied, but not the process unit itself.
 The utility representation corresponds to the process heat
requirements by dening their hot and cold utility streams. In
this case neither the process nor the utility supply can be
modied.
With the assumption of no heat losses, the representations are
characterized by the same amount of heat, but with different
temperature proles. The exergy losses can be visualized in Fig. 2
on the right, using the Carnot composite curves [12].
The streams of the evaporation units are rst modeled regarding
the existing technology (5 effects with thermal vapour recompression). This means that the evaporator will not be modied. For the second heat pump integration option, the process can
be modied and the evaporation unit is modeled in the thermodynamic representation without thermal vapour re-compression. It
becomes also possible to modify the layout of the evaporation unit
and to integrate a mechanical vapour re-compression (MVR). A
ow-sheet of the evaporation unit is given in Fig. 3.
All other units are dened if possible in the thermodynamic
representation.

a mean value related to the energy consumption of 1 ton of cheese

(kWh/t). To make batch processes continuous and easier to solve,
Linnhoff et al. [7] introduced the time average approach.
Heat exchange becomes more difcult for multi-period process
and it has to be indirect through heat transfer units and storage
tanks. For this, all units of different periods are dened as independent sub-systems. The heat transfer between those subsystems can only be realized by intermediate heat transfer
networks. For the cheese factory, following process units are not
always working simultaneously and are therefore dened in subsystems with no possibility of direct heat exchange (evapo, pasto
1e5 and proc6).
Furthermore, the DTmin=2 has to be adapted to include the fact
that the real instant heat load will be higher than the mean heat
load. Thus, a higher heat exchange area and investment costs for
the same heat amount will be required. The instant heat load
*
Q_ h=c of a hot or cold stream belonging to unit u can be expressed
by Eq. (1). Q_ h;c is the mean heat load over all periods, dtot is
the total operating time and du is the effective operating time of
unit u.
*
dtot
Q_ h=c Q_ h=c
du

DTmin=2u

*
Q_ h=c

UA

(1)

*
Q_ h=c
DTmin=2ref
Q_

(2)

h=c

Combining Eqs. (1) and (2), the DTmin=2u for each unit and their
streams can be deduced. The DTmin=2ref is considered to be 2  C and
is used for all permanently working units. For all other units, the
DTmin=2u value is calculated with Eq. (3).

dtot
DTmin=2ref ;
du

DTmin=2u x

(3)

As given in Eq. (4), the heat loads are expressed in kWh/t.

Q_ h=c $dtot
M

2.3. Streams denition

qh=c

The process operations are not simultaneous. In order to avoid

multi-period problems, the energy consumption is expressed by

Finally, the complete list of streams with corresponding heat

loads and DTmin=2 values is reported in Table 1.

PH

PH

PH

Effect 1

Effect 2

Effect 3

T1
P1

T2
P2

T3
P3

(4)

Milk inlet

PH

To thermal vapour compression

....

steam

To effect 4 and 5 ...

product
steam
condensates

Fig. 3. Flow-sheet of evaporation unit.

H. Becker et al. / Applied Thermal Engineering 43 (2012) 118e127

Table 1
Process streams.
Unit
other

evapo tech

pasto1

pasto2
pasto3
pasto4
pasto5
proc6

proc7
proc8
proc9
proc10
heat
CIP

Name
other_c1
other_h1
other_h2
evapo_c1
evapo_h1
evapo_h2
evapo_h3
pasto1_c1
pasto1_c2
pasto1_h1
pasto2_c1
pasto2_h1
pasto3_c1
pasto3_c2
pasto4_c1
pasto4_c2
pasto5_c1
proc6_c1
proc6_c2
proc6_c3
proc7_c1
proc8_c1
proc9_c1
proc10_c1
heat_c1
clean_c1

121

Energy
Electricity Fuel

Tin
[ C]

Tout
[ C]

Heat load
[kWh/t]

DTmin=2u

dtot =du [-]

100
5
0.3
100
44
44
44
6
48
75
79
54
74
6
69
8.5
66
105
78
95
15
70
35
32
35
85

190
0.5
2.5
190
5
25
44
48
75
4
85
4
80
28
75
26
76
105
78
95
55
70
35
25
35
85

367.8
307
56.9
993.7
32.9
198.1
627.8
568.2
344
904.6
5
41.9
84.1
308.2
32.1
83.3
54.2
131
49.6
49.6
40.4
62.6
33.8
175.5
153
238

2.0
2.0
2.0
3.1
3.1
3.1
3.1
3.6
3.6
3.6
6.8
6.8
2.9
2.9
4.5
4.5
2.9
2.8
2.8
2.8
2.0
2.0
2.0
2.0
2.0
2.0

[ C]

Support
Water Air Inert Gas
Sub-system "all"

Global Heat transfer system (GHTS)

Common
units (CU)

Raw
materials

Heat transfer
units (HTU)

Energy
services
Products
Byproducts

1.7
Sub-system 1n
Heat transfer system (HTS) of
parent sub-system 1n

Sub-system
2n

1.8

3.4

Sub-system 1nn
Heat transfer system
(HTS) of parent
sub-system 1nn

Sub-system
2nn

1.5
2.3

Sub-system Sub-system
1nnn
2nnn

1.4
1.4
Heat losses

1
1
1
1
1
1

2.4. Process integration

Having well dened the process and its thermal streams, the
maximum heat recovery and the minimum energy requirement is
calculated. Suitable utilities can be integrated and particular
attention has been given to the heat pump integration. Two options
are considered. In the rst option, the heat pump cannot exchange
heat directly with process, while in the second option process
modications and direct heat exchange between the heat pump
and the process are allowed. Direct heat transfer between nonsimultaneous process operations is not possible, but indirect heat
transfer can be realized trough heat transfer units and storage
systems. Atkins et al. [1] pointed out the importance of heat
recovery loops for the feasibility, especially in large processes and
when process operations do not occur at the same time. Time
average approach is very useful, but it requires the application of
restricted matches between non-simultaneous streams. The
concept of restricted matches between sub-systems, introduced
Becker and Marchal [2] can be applied. By denition, one subsystem cannot exchange heat directly with another sub-system
and an additional heat demand can only be satised by the heat
transfer system.
The problem is formulated as a MILP problem and considers
simultaneously the integration of heat recovery loops, energy
conversion units and heat pumping systems. Since the method is
also valid for several levels of restrictions, the method can be
applied for our case study with both, normal heat exchange
restrictions and restrictions due to non continuous process operations. Fig. 4 explains the concept. The problem formulation is an
extension of the formulation introduced by Becker and Marchal
[2]. The detailed problem formulation and explanations are given in
Appendix A.

Solids
Water
Waste

Gas

No direct heat exchange possible

Direct heat exchange possible

Fig. 4. Denition of sub-systems with several levels.

systems to minimize the operating costs of a cheese factory. As the

fuel and electricity prices vary signicantly from one country to
another, the optimal utility integration may be different. For this
work, the process integration has been performed with the energy
prices for Germany and France. Once the targeting phase is nished,
complementary global results, such as CO2 emissions, primary
energy consumption and also a rst evaluation of investment costs
can be evaluated and compared. Specic heat pump results and their
coefcient of performance (COP) are also analyzed and presented.

3. Case study
First, the maximum heat recovery is computed. Exergy losses
can be visualized on Fig. 5a, which shows the hot and cold Carnot
composite curves. The process grand composite curve, presented in
Fig. 5b, is used to dene optimal temperature levels for the energy
conversion system (utilities).
Considering the self-sufcient pockets of Fig. 5b, it can be
deduced that the heat recovery between process streams concerns
process streams below the pinch point. As process streams without
heat exchange restrictions are mainly involved, the heat recovery
penalty of restricted matches due to non-simultaneously process
operations is rather small. The penalty can be evaluated by
comparing the case with heat exchange restrictions and no indirect
heat recovery to the case without any heat exchange restrictions.
The use of intermediate heat transfer networks could only recover
a small amount of heat. The exact penalty due to non-simultaneous
operations can be evaluated using the methodology presented in
Becker and Marchal [2]. In fact, the penalty of 27 kW h/t corresponding to 1% of the hot utility and 3% of cold utility, is quite small
and can be neglected in the following. This means that no supplementary heat transfer units will be integrated to reduce this penalty.
Regarding the necessary temperature levels for heat supply on
Fig. 5b, following utility types are proposed to the process:

2.5. Results analysis

The main goal of this work is to demonstrate the application of
the optimal integration of utility and especially heat pumping

 Conventional steam boiler, refrigeration cycle and cooling

water are available on-site and no supplementary investment
costs are necessary.

122

H. Becker et al. / Applied Thermal Engineering 43 (2012) 118e127

Fig. 5. Minimum energy requirements. (a) Hot and cold composite curves using the Carnot scale, (b) Grand composite curve.

 A co-generation engine providing heat and electricity can be

interesting because of rather low temperature levels for
a signicant part of the hot utility (<120  C). Corresponding
investment costs are considered for the results analysis.
 Heat pump units are integrated for two options; The rst
option is described in 3.1 when process modications are not
allowed and the second option is shown in 3.2 where the
process can be modied. Investment costs for new installations
are included in the results analysis.
All utilities will be integrated simultaneously and heat sources
for the refrigeration cycle and heat pumps are considered in the
optimization. For example, the cooling water will satisfy a part of
the condensation heat of the refrigeration cycle and heat pumps
ideally satisfy hot and cold heat demands at the same time. The
nal overall electricity balance contains the electricity consumption (e.g. from the process or heat pumps) and the electricity
production (e.g. co-generation units).
The closed cycle heat pump and the co-generation engine are
integrated through an intermediate heat transfer network which
means that they cannot exchange heat directly with the process.
The results are compared for the French and German context. The
cost, CO2 emissions and primary energy for the French and German
electricity mix are given in Table 2.
Results of analyzed scenarios will be summarized in Table 3.
Several cases are compared to the current case (Case 0). First (for
cases 1a-d) the current evaporation unit including its thermal
vapour re-compression is modeled in the technology representation. Three scenarios with different hot utilities have been

evaluated: boiler (Case 1a), boiler and heat pump (Case 1b), boiler
and engine (Case 1c) and boiler, heat pump and engine (Case 1d).
For heat pump integration option 2 (Cases 2,3 and 4) the evaporation units are modeled in the thermodynamic representation:
MVR and boiler (Case 2a) and MVR, boiler and engine(Case 2b). For
Cases 3a/b and Cases 4a/b the evaporation unit is rst modied.
For all cases, the on-site available utilities (boiler, refrigeration
and cooling water) are integrated. The rst lines of Table 3 indicates
the selected option and additional proposed utilities.
3.1. Option 1 - No process modications allowed
In the rst approach, the process can not be modied. Thus, the
direct heat exchange between a potential heat pump and the
process is not allowed. The approach presented in appendix is
applied. In the higher level two sub-systems (in this case the
process and the heat pump) cannot exchange heat directly. Moreover, heat cannot be exchanged directly between different periods
and hence the process operation units of different periods are
dened in independent sub-systems (e.g. evaporation unit,
pasteurization units) of the sub-system process. The evaporation
unit can therefore not directly exchange heat with the pasteurization units or the heat pump, but it is possible to exchange heat with
the other process units. Satisfying the required temperature levels,
a closed cycle heat pump using the refrigerant R245fa has been
chosen. By analyzing the shape of the grand composite curve,
appropriate operating conditions for the heat pump and the
intermediate heat transfer units can be estimated. Simultaneously,

Saving potentials higher than 200% are not displayed on this graph
Saving potentials higher than 200% are not displayed on this graph

Savings [%]

150

200

Cost
Fuel
Electricity
CO2 emissions
Primary energy

150

Savings [%]

200

100

50

Cost
Fuel
Electricity
CO2 emissions
Primary energy

100

50

50

50
1a

1b

1c

1d

2a
2b
Case

3a

Fig. 6. Saving potentials in France.

3b

4a

4b

1a

1b

1c

1d

2a

2b

3a

Case

Fig. 7. Saving potentials in Germany.

3b

4a

4b

H. Becker et al. / Applied Thermal Engineering 43 (2012) 118e127

Table 2
Cost, CO2 emission and primary energy for given electricity mix in France (FR) and
Germany (DE).

FR
DE

Euro/
c
fuel
kWh

c
Euro/
el
kWhel

c
Euro/
el
kWhel

CO2el kg/
kWhel

Eprel MJ/
kWhel

0.0392
0.0500

0.0620
0.1080

0.0496
0.0864

0.092
0.631

11.788
10.945

utility, heat pump and heat transfer units are considered in the
MILP model to dene the corresponding optimal ow rates that
minimize the operating costs. The potential of a closed cycle heat
pump without direct heat exchange is illustrated in Fig. 8a. The
saving potential can be increased by minimizing the minimum
temperature difference in the heat exchangers between the heat
pump and the intermediate heat transfer uids. The advantage of
this approach is that only the investment costs related to the heat
pump, heat exchangers and pipes have to be accounted. The
process itself remains unchanged. Thus, the safety and product
quality aspects are also maintained.
3.2. Option 2 - process modications allowed
In the second option, process modications are allowed. This
leads to more heat pump integration opportunities. For example,
operating pressures of the process can be modied to improve heat
recovery, and mechanical vapour re-compression (MVR) units can
replace the thermal vapour re-compression unit. Case 2 replaces
the thermal vapour re-compression with a mechanical vapour recompression. The layout and the pressure levels of the evaporation effects are kept. The steam leaving the third effect at about
60  C is compressed mechanically to 74  C. As the DT is less than
18  C, a dynamic compressor is suitable and therefore selected
(Fig. 8b). In Case 3, all effects are realized in parallel and mechanical
vapour re-compression is integrated. The temperature difference
for a mechanical vapour re-compression in Fig. 8c is small (<10  C).
The pressure levels of the ve effects are adapted, so that all effects
evaporate at about 70  C. In Case 4, the pressure of effects are
modied and mechanical vapour re-compressions are included:
rst the new temperature levels have to be dened. Assuming that

123

the heat exchange area in the effects will not be changed, the
temperature levels of effects 4 and 5 are reduced (see Eq. (5)). The
new temperature levels are 48  C (effect 4) and 32  C (effect5).
Theoretically, it could be possible to raise waste heat at 32  C with
a heat pump to satisfy a part of the heat demand in effect 1.
However, high temperature lifts make such heat pump integration
not optimal. The use of successive mechanical vapour recompressions between the different effects have therefore been
preferred (Fig. 8d). The temperature of effect 5 is quite low. If these
operating conditions can not be reached in the unit, it could also be
possible to raise the temperature levels of all effects.




Q_ U*A* Tvap  Tprod

(5)

4. Results & discussion

As mentioned above, the applied MILP formulation minimizes
the operating costs (including fuel and electricity costs). Furthermore, the utility integration is optimized to satisfy the process
demand and the product quality. Keeping the process operations as
they are, this paper focuses on the rational conversion of energy. To
study the impact of different fuel and electricity prices and the
corresponding electricity mix, the results of the case study have
been compared for France and Germany.
4.1. Global saving results
All cases are compared to the current case before heat recovery
(Case 0). The results are presented in Table 3. Beside the operating
costs, the global electricity and fuel consumption, CO2 emissions
and primary energy consumption are presented in this table. Bold
values show the best option for each criterion.
It is interesting to remark that the proposed utilities in most of
the cases (Cases 2a/b,3a/b and 4a/b) are integrated in the same way
for France and Germany. A difference can be recognized in Cases 1c/
1d, where co-generation engines are favored in Germany.
Compared to the French case, the optimal size of a co-generation
engine in Germany would be more than twice as high in Case 1c.
In Case 1d, the co-generation is not integrated at all in France. A

Table 3
Process integration results for France (FR) and Germany (DE).
Case

1a

1b

1c

1d

2a

2b

3a

3b

4a

4b

Option

hp

cog

hp cog

mvr

mvr cog

mvr

mvr cog

3mvr

3mvr cog

126.0
166.0
2895.0
2895.0
194.0
194.0
e
e
602.6
707.2
15314.4
15150.8
e
e
e
e
e
e
e
e
e
e

97.8
127.3
2358.5
2358.5
86.6
86.6
955.3
955.3
484.4
531.0
11.6
11.6
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0

87.6
117.4
1921.7
1921.7
197.1
197.1
672.7
672.7
406.3
512.6
11.0
10.8
246.8
246.8
794.6
794.6
0.0
0.0
0.0
0.0
4.0
4.1

95.9
117.7
2527.1
2758.6
63.2
234.2
945.2
968.9
504.6
409.5
10.6
9.9
0.0
0.0
0.0
0.0
334.4
716.0
413.4
552.7
11.2
3.0

87.6
114.6
1921.7
2293.0
197.1
0.0
672.7
798.0
406.3
463.2
11.0
10.3
246.8
133.4
794.6
456.9
0.0
326.7
0.0
410.5
4.0
3.5

88.9
117.9
2023.7
2023.7
154.8
154.8
1003.6
1003.6
423.0
506.5
10.9
10.8
239.1
239.1
516.7
516.7
0.0
0.0
0.0
0.0
3.0
2.8

85.5
93.3
3041.6
3041.6
L680.3
L680.3
1017.1
1017.1
551.8
185.1
5.7
6.2
264.5
264.5
565.9
565.9
1880.4
1880.4
961.7
961.7
6.4
2.3

86.2
117.7
1770.5
1770.5
270.2
270.2
464.1
464.1
382.5
528.2
11.2
10.9
641.9
641.9
1256.9
1256.9
0.0
0.0
0.0
0.0
5.5
6.7

81.3
94.0
2605.3
2605.3
419.9
419.9
469.8
469.8
487.6
261.3
6.8
7.1
651.7
651.7
1274.2
1274.2
1546.9
1546.9
846.9
846.9
6.6
3.3

69.4
95.3
1394.9
1394.9
237.0
237.0
512.9
512.9
303.6
431.3
9.1
8.9
525.9
525.9
1160.3
1160.3
0.0
0.0
0.0
0.0
2.1
1.9

65.5
79.6
1893.0
1893.0
174.6
174.6
516.3
516.3
366.3
272.2
6.5
6.6
539.1
539.1
1185.4
1185.4
927.2
927.2
628.6
628.6
2.9
1.9

Utilities
OpC [e/t]
ef [kWh/t]
eel [kWh/t]
Cooling [kWh/t]
mCO2 [kg/t]
ep [MJ/t]
E_ hp kW
InvChp [ke]
E_ cog kW
InvCcog [ke]
PR [years]

FR
DE
FR
DE
FR
DE
FR
DE
FR
DE
FR
DE
FR
DE
FR
DE
FR
DE
FR
DE
FR
DE

124

H. Becker et al. / Applied Thermal Engineering 43 (2012) 118e127

Fig. 8. Comparison of Carnot composite curves. (a) Integrated composite curves Case 1d, (b) Integrated composite curves Case 2b, (c) Integrated composite curves Case 3b, (d)
Integrated composite curves Case 4b.

sensitivity analysis of the fuel to electricity ratio could identify

different integration zones for the heat pump and the cogeneration engine. A major difference between the two countries
concerns the CO2 emissions. Taking the current case (Case 0), the
CO2 emissions in Germany are higher, however the possibility of
integrating heat pumps and in particularly co-generation engines
leads to a higher saving potential in CO2 emissions. As the electricity in France is mainly produced by nuclear and hydro power
plants, the interest of co-generation engines is smaller. Whereas in
Germany, the CO2 content in electricity (mainly produced by coal) is
relatively high, and thus efcient co-generation engines reduce not
only the energy consumption of the process but also the CO2
emissions. Another advantage of co-generation engines is the
possibility to sell the surplus of produced electricity to the grid (e.g.
Cases 2b, 3b and 4b in France and Germany). However, this depends
strongly on the given selling prices. Figs. 6 and 7 show the saving
potential in France and Germany respectively. The saving potential
of a well integrated process without any supplementary heat pump
(closed cycle heat pump or MVR) or co-generation unit is about 22%
in France and 23% in Germany. This means that the heat recovery is
fully exploited and for this, new heat exchangers are necessary, but
no further utility units are considered. It is important to remark,
that the optimal integration of heat pump units has to be compared
to the case with optimal heat recovery of the process. The integration of heat pumps and co-generation units will reduce the
operating costs. Either option 1 (up to 30%) or option 2 where the
maximum saving potential reaches up to 50% of operating costs for
the German case and up to 45% for the French case, can be chosen.
The electricity consumption in some cases is higher than in the
current case, this is due to the integration of heat pumps.

An example of integrated composite curves is given for France for

the Cases 1d, 2b, 3b and 4b in Fig. 8. In Case 1d, the evaporation unit is
modeled in the technology representation including the thermal
vapour re-compression. A heat pump, a co-generation unit and
intermediate heat transfer networks are integrated to ensure the
indirect heat transfer between the process and the new utility units.
The temperature difference for the heat sink and heat source of the
heat pump is relatively high. Nevertheless the overall operating costs
and especially the fuel consumption can be reduced. In the following
cases, the evaporator is modeled in the thermodynamic representation and more saving opportunities can be pointed out. The necessary
temperature difference is smaller in case 2b, where the thermal
vapour re-compression is replaced by the mechanical vapour recompression. By modifying the layout of the evaporation effects,
temperatures differences of a mechanical vapour re-compression can
be reduced and the saving potential becomes higher. Fig. 8 shows the
Carnot composite curves. The surface between the utility and and
process composite curves represents the exergy losses.
4.2. Evaluation of heat pump performance
Table 4 elucidates the characteristics of each heat pump. The

COP of a heat pump is dened as shown in Eq. (6). Q_ H is the optimal

heat delivered by the heat pump and E_ hp is the necessary electricity

consumed by the compressor. For this, the isentropic efciency of
the compressor is assumed to be 0.76.

COP


Q_ H

E_

hp

(6)

H. Becker et al. / Applied Thermal Engineering 43 (2012) 118e127

Table 4
Specic heat pump results.
Case

Type

Fluid [kW]

Teva [ C]

Tcond [ C]

E_ hp=mvr [kV]

COP [-]

1b FR/DE
1d FR
1d DE
2a FR/DE
2b FR/DE
3a FR/DE
3b FR/DE
4a FR/DE
4a FR/DE
4a FR/DE
4b FR/DE
4a FR/DE
4a FR/DE

HP
HP
HP
MVR
MVR
MVR
MVR
MVR1
MVR2
MVR3
MVR1
MVR2
MVR3

R245fa
R245fa
R245fa
water
water
water
water
water
water
water
water
water
water

30
30
30
57
57
66
66
28
46
57
28
46
57

90
90
90
74
74
74
74
36
52
74
36
52
74

386
386
209
239
264
642
652
141
92
294
141
92
307

3.6
3.6
3.6
14.8
14.8
32.3
32.3
28.5
40.4
14.8
28.5
40.4
14.8

4.3. Estimation of investment costs and payback time

All cases are compared regarding their estimated investment
costs and related payback time (Table 3). To estimate the instantaneous power of heat pump or MVRs, the mean value (heat load
per tons of product) is multiplied by number of tons of products per
week divided by the weekly total operating time of the evaporation
unit. From selected quotations, the investment costs [V] can be
estimated with Eq. (7) as a function the compressor power (E_ hp in
[kW]). The installation factor f is supposed to be 1.5. Investment
costs for co-generation units [V] can be estimated from Eq. (8). In
Cases 1a-1d, the process streams connected to the heat pump are
not simultaneous. Thus, the investment can not be calculated
without considering heat storage. It could be possible to evaluate
two extreme cases: the maximum value corresponds to the peak
power required by the heat source, while the minimum value is
estimated by assuming that the heat source is stored and
progressively upgraded leading to a more constant operation of the
heat pump. However, storage is necessary in both cases and further
calculations will be necessary. Finally without considering the
discount rate, the payback rate is estimated by Eq. (9).

PR InvCtot =B

a co-generation unit are proposed to the process; In Germany a heat

pump and a co-generation engine are integrated and satisfy a part of
the hot utility. The engine also satises the electricity consumption of
the process. As the temperature levels of the co-generation engine is
similar to the condensation level of the heat pump, both utility are in
direct competition. In France, only a heat pump is integrated.
Compared to the German case, this heat pump is bigger and satises
more of the hot utility than the heat pump in France.
5. Conclusion

Although these denitions are widely used in industry, theoretically they are only correct for a heat pump having its heat source
at ambient temperature. It is important to note that a heat pump
with a high COP can even be counterproductive when it is not
appropriately integrated. In order to be compatible with the well
known conventions, the COP will be related to a performance of the
heat pump machine by using this denition. But to be rigorous
when comparing heat pump cycles, global results and the exergy
efciency is preferably used.

0:9
InvC f $1500$1600:1 $E_ hp


2
InvC 0:68*  0:0391$E_ cog 850:89$E_ cog 306016

0:8
125$E_ cog $1477:7=1068:3

125

A method for the optimal integration of energy conversion and in

particular heat pump units has been presented. In all cases, the
installation of a closed cycle heat pump or a mechanical vapour recompression reduces the operating costs by more than 50% (in
Germany). The potential CO2 emissions savings are more than 50%
(in France). Generally speaking, the analysis can be valid for any type
of process. Specic process characteristics can easily be integrated,
since the method allows consideration of heat exchange restrictions
due to industrial constraints and/or non-simultaneous process
operations. The principal results presented above can also be valid
for other processes with evaporation units (e.g. dairy, pulp&paper).
The interest of heat pumps is slightly higher in France, but also in
countries with higher fuel and electricity prices, heat pumps can
have a high saving potential. On the other side co-generation
engines are particularly interesting in the German context.
The advantages and disadvantages are briey summed up in the
following. Indirect heat pump integration (option 1) needs no
process modications and no supplementary investment costs. The
safety and product quality is ensured, since the heat pump has no
direct contact with the process. But the main disadvantage is that
the COP is relatively low. Moreover, when process operations are
not simultaneous, the storage problem has to be considered, which
may lead to supplementary costs. The advantage of direct heat
pump integration (option 2) is mainly the better energy efciency.
Furthermore no storage has to be considered, since the mechanical
vapour re-compression will be integrated directly with the evaporation unit. On the other hand, process modications could
become necessary and this may give higher investment costs. The
MVR is in direct contact with the product.
Finally, the investment costs are estimated and payback time
can be roughly evaluated. However, consideration should be given
to the fact that the real investment costs depend strongly on a given
case. Therefore, the nal decision has to be taken with concrete
heat pump offers corresponding to the existing facilities and
installations of the company.
Acknowledgements

(7)
The authors wish to thank ECLEER for supporting this research
and collaborating in its realization.

(8)
(9)

Also the economic analysis shows clearly that the integration of

a co-generation engine is not protable in France. Considering case 1c
where only a co-generation unit is proposed in the process, it can be
seen that the operating costs in Germany would signicantly be
reduced. Thus, the payback time of 3 years is obtained. On the
contrary in France, the co-generation engine reduces the operating
costs only by a small amount. Regarding the high payback time of 11
years, it is improbable that a company would invest in a co-generation
unit. Considering in a second step case 1d where a heat pump and

Appendix A. Process integration with the concept of multilevel sub-systems

Extended nomenclature:
Q_ htss;s;k Heat provided by the heat transfer system () to subsystem s or heat removed by the heat transfer system ()
from sub-system s in interval k [kW]
Q_ hts1;k Heat provided by the higher heat transfer system () or
heat removed by the higher heat transfer system () [kW]
Total electricity demand () or excess () [kW]
E_ el
Cascaded heat to lower interval k [kW]
R_ k

126

H. Becker et al. / Applied Thermal Engineering 43 (2012) 118e127

fu
q
rk
yu
E_ el;u
E_ f ;u
~cu
cel
cf
cu
nf
nk
nps
ns
nsub
nu

Multiplication factor of unit u [-]

Specic heat load [kWh/t]
Specic cascaded heat to lower interval k [kWh/t]
Integer variable representing the existence (1) or not (0)
of unit u [-]
Consumed ()/produced () nominal electricity by unit u
[kW]
Consumed () nominal fuel by unit u [kW]
Nominal utility operating cost (excluding fuel and
electricity costs) per tons of products [V/t]
Electricity price buying price () selling price () [V/kWhel ]
Fuel price [V/kWh]
Nominal utility operating cost (excluding fuel and
electricity costs) per hours [V/h]
Number of different fuels [-]
Number of temperature intervals [-]
Number of parent sub-systems [-]
Number of streams [-]
Number of sub-systems [-]
Number of units [-]

A unit can be a process f u 1 or a utility (f u variable) unit.

The ow rates of streams belonging to utility units are proportional
to the multiplication factor, which is limited by a minimum and
a maximum value. The associated integer variable yu denes if the
utility unit u is added to process yu 1 or not yu 0.

yu $fumin  f u  yu $fumax

The concept of restricted matches between sub-systems can be

extended to multi-level sub-systems (Fig. 4). The denition of subsystems inside sub-systems become possible. For each level a heat
transfer system is necessary. It consists in units which can exchange
heat with all sub-systems of the corresponding level. The global
problem is represented by the last sub-system which contains all
sub-systems and the global heat transfer system with no heat
exchange restrictions. For each sub-system s the heat cascade is
given by Eq. (A.6)e(A.8).
ns
h;s;k
X

f u Q_ h;s;k;u 

ns
c;s;k
X

f u Q_ c;s;k;u Q_ htss;s;k  Q_ htss;s;k R_ s;k1

cs;k 1

hs;k 1

 R_ s;k 0

Subscripts
hts
Index for heat transfer system
k
Temperature interval
s
Index for sub-system

(A.5)

ck 1:::; nk

R_ s;1 0 R_ s;nk1 0

R_ s;k  0

cs 1:::; nsub

ck 2:::;nk

A:6

cs 1:::;nsub
(A.7)

Appendix A.1 MILP formulation [2]

The corresponding equations are given below. Bold letters are
used to highlight the decision variables. Like for the conventional
heat cascade the objective is to minimize the operating costs
(Eq. (A.1)).

Fobj min@dtot $@

nf
X
f 1

_
c
el E el

nu
X

c
f

nu
X
u1

f u E_ f ;u

E_
c
el el

11

(A.1)

f u cu AA

u1

Eq. (A.2) states that the total electricity import and the produced
electricity by the process have to be greater or equal to the process
electricity consumption. The overall electricity balance is given by
Eq. (A.3). Both equations are necessary to distinguish the price for
electricity import and export.
nu
X

f u E_ el;u E_ el 

u1

nu
X


f u E_ el;u  0

(A.2)

Q_ htss;s;k  0


Q_ htss;s;k  0

Q_ h;s;k;u is the nominal heat load of hot stream h in sub-system s

and interval k and belonging to unit u. The real heat load is calculated with the multiplication factor f u. When a sub-system has
a decit or a surplus of heat in the temperature interval k, the heat
is supplied from the heat transfer system of its sub-system

Q_ htss;s;k or respectively removed by the same heat transfer

system Q_ htss;s;k . R_ s;k is the cascaded heat to the lower temperature interval k in sub-system s.
The heat cascade for the heat transfer system (hts) for each

parent sub-system is given by Eqs. (A.9) and (A.10). Q_ hts1;k is the

heat supplied from the heat transfer system a level above and

Q_ hts1;k is the heat transferred to the heat transfer system of the
higher level.
nsX
h;hts;k

f u Q_ h;hts;k;u 

u1


f u E_ el;u E_ el  E_ el 

u1

nu
X


f u E_ el;u 0

(A.3)

u1

E_ el  0


E_ el  0

For the electricity cost,

selling price. c
f

(A.4)
c
el

c
el

is the purchase cost and

is the

is the fuel price. E_ f ;u is the nominal energy deliv-

ered to unit u by the fuel (e.g. natural gas) and E_ el;u is the nominal
electricity demand () or excess () of unit u. cu is the nominal
operating cost per hour of unit u (excluding the fuel and electricity
costs of unit u).

nsX
c;hts;k

f u Q_ c;hts;k;u Q_ hts1;k  Q_ hts1;k

chts;k 1

nsubhts
X k


Q_ htss;s;k

s1

The corresponding thermodynamical feasibility is guarantied by

Eq. (A.4).

cs 1:::;nsub
(A.8)

hhts;k 1
nu
X

ck 1:::;nk

 R_ hts;k 0

nsubhts
X k

Q_ htss;s;k R_ hts;k1

(A.9)

s1

ck 1:::;nk chts 1:::;nps

R_ hts;1 0 R_ hts;nk1 0 R_ hts;k  0 ck 2;:::;nk chts 1:::;nps

(A.10)
To ensure that heat is cascaded correctly, a second set of Equations for the global heat transfer system is necessary. Eq. (A.11)
expresses the heat balance of the hot streams and Eq. (A.12)
expresses the heat balance of the cold streams in the heat

H. Becker et al. / Applied Thermal Engineering 43 (2012) 118e127

transfer system. The ow rates of the heat transfer units have to be

optimized in order to satisfy the remaining heat demand of all subsystems.
nsX
h;hts;k

127

_ R
_ are replaced
All heat loads and energy consumptions Q_ ; E;
with their corresponding specic values q;eel ;rk . And Eqs.
(A.2)e(A.12) can be solved using the specic values.


f u Q_ h;hts;k;u Q_ hts1;k R_ hts;k1  R_ hts;k

hhts;k 1

nsubhts
X k


Q_ htss;s;k  0

References

ck 1:::;nk chts 1:::;nps

s1

nsX
c;hts;k

(A.11)

f u Q_ c;hts;k;u  Q_ hts1;k R_ hts;k1  R_ hts;k

chts;k 1

nsubhts
X k

Q_ htss;s;k  0

ck 1:::;nk chts 1:::;nps

s1

(A.12)
Appendix A.2. Adaptation when using time average approach
With heat exchange restrictions in different levels, the combination of restricted matches due to industrial constraints and
restricted matches due to batch operations become possible. Using
mean values, heat loads and energy quantities are expressed in
kWh per tons of produced cheese [kWh/t]. By convention, small
letters are used for specic quantities. The nominal unit operating
cost cu has therefore to be expressed in Euro/t (new: ~cu ) and the
previous formulation has to be adapted. The objective function has
been re-written in Eq. (A.13).

0
Fobj OpC min@

nf
X
f 1

c
f

nu
X
u1

!
f u e
f ;u


c
e c
el eel
el el

nu
X

1
f u ~cu A

u1

(A.13)

[1] M. Atkins, M. Walmsley, J. Neale, Application of heat recovery loops for

improved process integration between individual plants at a large dairy
factory, Chemical Engineering Transactions 25 (2011) 183e188.
[2] H. Becker, F. Marchal, Energy integration of industrial sites with heat
exchange restrictions, Computers and Chemical Engineering (2011).
doi:10.1016/j.compchemeng.2011.09.014.
[3] H. Becker, F. Marchal, A. Vuillermoz, Process integration and opportunity for
heat pumps in industrial processes, International Journal of Thermodynamics
14 (2) (2011) 59e70.
[4] A. Kapil, I. Bulatov, R. Smith, J. Kim, Site-wide process integration for low grade
heat recovery, Computer Aided Chemical Engineering 29 (2011) 1859e1863.
[5] P.O. Kapustenko, L.M. Ulyev, S.A. Boldyryev, A.O. Garev, Integration of a heat
pump into the heat supply system of a cheese production plant, Energy 33
(2008) 882e889.
[6] I.C. Kemp, Pinch Analysis and Process Integration a User Guide on Process
Integration for the Efcient Use of Energy, second ed. Elsevier, ButterworthHeinemann, UK, Oxford, 2007.
[7] B. Linnhoff, G. Ashton, E. Obeng IChemE Symposium Series, Process Integration of Batch Processes, vol. 109 (1988) 221e237.
[8] B. Linnhoff, D. Townsend, Heat and power networks in process design. part 1:
criteria for placement of heat engines and heat pumps in process networks,
AIChE Journal 29 (5) (1983) 742e748.
[9] Muller, D. 2007. Web-based tools for energy management in large companies
applied to food industry. Switzerland: Ph. D. thesis, LENI - Ecole Polytechnique
Fdrale de Lausanne.
[10] D. Muller, F. Marchal, T. Wolewinski, P. Roux, An energy management method
for the food industry, Applied Thermal Engineering 27 (2007) 2677e2686.
[11] M. Pavlas, P. Stehlk, J. Oral, J. Klemes, J. Kim, B. Firth, Heat integrated heat
pumping for biomass gasication processing, Applied Thermal Engineering 30
(2010) 30e35.
[12] F. Staine, D. Favrat, Energy integration of industrial processes based on the
pinch analysis method extended to include exergy factors, Applied Thermal
Engineering 16 (1996) 497e507.
[13] E. Wallin, T. Berntsson, Integration of heat pumps in industrial processes, Heat
Recovery Systems & CHP 14 (3) (1994) 287e296.