The 13th Asia Pacific Confederation of APCChE 2010

Chemical Engineering Congress October 5-8, 2010, Taipei

Process streams temperature flexibility for energy savings in HENs

Abdelbagi Osman*, M. I. Abdul Mutalib, M. Shuhaimi

Chemical Engineering Department, Universiti Teknologi PETRONAS,

Seri Iskandar, 31750 Tronoh, Perak Malaysia
*
Corresponding Authors E-mail: abdelbagi2001@yahoo.com

Keywords: HENs retrofit, energy savings, streams temperature flexibility.

Abstract

The increasing pressures from current economic uncertainties and the tighter environmental
regulations have substantiated more energy conservation projects to be reviewed and analyzed properly.
Retrofit of Heat Exchanger Networks (HENs) is among the common projects to reduce plant operational
cost. Based on the plus/minus principle of Pinch Technology, this paper is addressing a new concept of
varying the process streams temperature as a means of increasing the heat recovery for the process heat
exchanger network (HEN). This concept mainly depends on the process streams flexibility to changing
the inlet and outlet temperature. Essentially, the idea involves systematic steps in varying the hot and/or
cold streams temperatures in a manner that will increase the Heat Recovery Approach Temperature
(HRAT) in the HEN beyond the existing value to allow for more heat recovery between the hot and cold
streams. Using the utility paths, the adjustment of the heat loads for both hot and cold utilities resulting
from the heat load shifting through the HEN is made and assessed accordingly. Given that major changes
in process operations are rarely desired, the changes in the temperature made to the streams while
exploring each alternative has been kept within a small magnitude of flexibility, i.e. around 1 5oC. A
HEN example of several utility paths is used to demonstrate the proposed idea.

1. Introduction

Industry is currently considering energy conservation projects more favorably as a result of increasing
pressure from current economic uncertainties and tighter environmental regulations. The significant hike
in oil and gas prices as reported by the Energy Information Administration (EIA), has impacted the
energy price tremendously, (EIA, 2009). The increasing price trend is expected to continue over the long
run as the oil and gas supply depletes with time. A recent report by Campbell highlighted that most oil
and gas producing countries have attained peak production and the decline is forecasted to be at 2-3% a
year (Campbell, 2006). The new oil discoveries made were insufficient to cope with the shortfall of
supply. Additionally, the situation is made worse by the population growth and massive industrialization
effort by China and India, (Campbell and Laherrere, 1998). Therefore, projects leading to improved heat
recovery in chemical processes will be expected to continue receiving support from the industry. In
particular, the energy conservation projects featuring small capital investment and quick economic returns
are thought of as attractive for industries to adopt. Besides the energy crisis, stricter environmental
regulations relating to atmospheric discharge have become another prime driver for energy conservation
projects. Process plants contribute to the level of CO2 emissions through their central utility system which
mostly fires fuel to generate heat and power.

1
The 13th Asia Pacific Confederation of APCChE 2010
Chemical Engineering Congress October 5-8, 2010, Taipei

The previous researches conducted on the HEN design and retrofit for better heat recovery have
considered the operating conditions to be fixed. The HEN system has been treated as a subsystem in the
process plants and changes in process conditions have not been significantly highlighted. Moreover, it is
found that most of the conducted research in HEN retrofit concentrated on reconstruction and topological
changes in order to justify acceptable economic returns. The massive topological changes always require
high capital investment and massive plant changes which is risky in the light of current economic
situations.

In this paper, minor retrofit using the available utility paths in the HEN is proposed while considering
the process streams temperature flexibility (TF) based on the plus/minus principle in order to reduce the
energy consumption in the HEN. The minor HEN retrofits which involve only area addition to the
existing matches in the network are more suitable (advisable) as they are cheap and require less structural
adaptation and civil work.

2. Background for the use of utility paths in the HEN retrofit

Varbanov and Klemes presented the rule of path construction for the HEN retrofit based on the
techniques of Pinch Technology, (Linnhoff and Tjoe, 1985; Varbanov and Klemes, 2000). They also
extended the methodology of the Network Pinch which has been explored by Asante and Zhu (Asante and
Zhu, 1996; Asante and Zhu, 1997). In certain HEN systems, if there is no available utility path to increase
the heat recovery or, the potential of existing paths is exhausted to handle more heat recovery the
Network Pinch could not be established (Varbanov and Klemes, 2000). Based on this fact, the rule for
path construction was developed and used for the HEN retrofit when the Network Pinch could not be
identified. Similar to Asante and Zhu, Varbanov and Klemes have proposed an ordered topology changes
for this special case of the HEN retrofit. Five Topological changes were suggested, namely they were:
match relocation, match addition, match removal, splits addition and splits removal.

Other procedure based on the path analysis for the HEN retrofit was introduced, (Van Reisen et al.,
1995). The method is a prescreening approach to analyze an existing HEN system for energy saving
purposes. From the existing HEN, the approach initially identified all the possible sub-networks that
contain at least one heat shifting path. Then, energy conservation using the identified path was established
for each sub-network and analyzed economically. Within the path analysis approach, two important rules
must be met while decomposing the HEN:

All the sub-networks should be heat balanced.

At least, a heater and a cooler connection within one path should be included in each sub-network.

Increasing the heat recovery using the identified path(s) would save energy at the expense of adding
new area to the existing exchanger on the existing path, or adding new matches which create new paths.
Adding area is preferred due to the minor impact on the existing exchangers.

The path analysis method was later extended where they incorporated the structural targeting
methodology for the HEN retrofit that best fit the large network aiming at reducing the retrofit design
effort, (Van Reisen et al., 1998). Besides the energy savings and area investment, this method gives a
target for structural modifications where the location of the topology changes is essentially considered.
Also, the HEN layout, functionality and operability have been taken into account but ignored the effect of
the matches and streams pressure drops which are considered as crucial parameters that could affects the

2
The 13th Asia Pacific Confederation of APCChE 2010
Chemical Engineering Congress October 5-8, 2010, Taipei

design cost. Since the method is based on the path analysis approach, it identifies part of the network with
high energy savings to investment potentials.

Recently, new approach based on the path analysis method for the HEN retrofit was introduced,
(Osman et al., 2009). Their approach termed as the Paths Combination Approach for HENs Retrofit
which uses a combinatorial method to combine the available utility paths systematically to generate
several heat load shifting options. The pressure drop constraint of the existing exchangers in the HEN was
considered for calculating the heat transfer coefficients and the heat transfer area. The obtained retrofit
options using the path combination approach were considered to be minor retrofit solutions where only
area addition was involved without topology changes.

3. Theory of using the utility paths in enhancing the heat recovery in the HEN

The utility path in the HEN is defined as a connection between heater(s) and cooler(s) through definite
match(es) in the HEN, (Shenoy, 1995). A certain amount of heat load could be shifted along this path
from the heat source and the heat sink within the HEN system while increasing the heat recovery. Using
this path, if certain amount of heat load (x) is to be subtracted from the cooler (C) and the heater (H) in
the HEN; it must be added and subtracted alternatively to and from the exchangers (E) lying on the path.
Figure 1 explained this concept more obviously.

Minimum practical temperature difference must be maintained for individual matches in the HEN
while undergoing the heat shifting. This understanding could be followed for the HEN retrofit to enhance
the process-to-process heat recovery either with or without topological changes.

Figure 1: Utility path to shift heat load between heater and cooler in the HEN

It has been shown that Tmin will no longer be a constraint to limit the heat recovery while optimizing
the HEN using loops and paths, (Smith, 2005). However, impractical temperature driving force for
individual exchangers must be avoided. Therefore, in this work the existing heat recovery approach
temperature (HRAT) is maintained instead of Tmin to ensure heat transfer between the exchange streams
is feasible. Meanwhile, the HRAT also becomes the constraint parameter which limits the extent of heat
load shifting along the utility path in the HEN.

3.1 The plus/minus principle and process changes for heat recovery in the HEN

The energy requirements in the HEN could be reduced by changing and maneuvering the mass and
energy balance of the process. It is possible to identify changes in an appropriate process parameter that
would have a favorable impact on energy consumption together with applying the thermodynamic rules of
the Pinch Technology. The concept of playing with process parameters called the plus-minus principle
which was introduced and described in the Pinch Technology, (Linnhoff and Parker, 1984; Linnhoff-

3
The 13th Asia Pacific Confederation of APCChE 2010
Chemical Engineering Congress October 5-8, 2010, Taipei

March, 1998). Changing the heat and mass balance would imply changes in the composite curves of the
HEN as shown in Figure 2 below:

Figure 2: Demonstrating the plus-minus principle

From the figure, the plus-minus principles could be summarized in the following points:

Increasing the hot stream(s) duty above the pinch and/or decreasing the cold stream(s) duty
above the pinch would result in reducing the hot utility target.

Decreasing the hot stream(s) duty below the pinch and/or increasing the cold stream(s) duty
below the pinch would result in reducing the cold utility target.

Referring to the Pinch Technology approach (Linnhoff-March, 1998), it is even possible to change
temperatures rather than heat duties so as to further reduce the heating and cooling duties. Therefore, the
plus-minus principle ought to be beneficial to increase the temperature of hot streams and/or reducing the
temperature of cold streams which make it easier to extract heat from them. Changing the temperature of
streams in this manner would improve the driving forces in the HEN while at the same time decrease the
energy targets of the process.

Based on the plus/minus principle, Zhang and Zhu have introduced the idea of adopting process
changes with the HEN retrofit, (Zhang and Zhu, 2000). They have developed a systematic method to
investigate the impact of process temperatures and flow-rates changes while undertaking the HEN retrofit.
Likewise the plus-minus principle, Zhang and Zhu have represented their insight in a T-H diagram to
show how energy consumption reduces by managing the process conditions changes. The interaction
between process changes and HEN retrofit was explored based on the retrofit model of topology
modification, (Asante and Zhu, 1997). Accordingly, their finding has revealed a significant cost savings.
However, the streams and exchangers pressure drop have not been considered despite their great impact
on the network units operation as well as operational cost especially when the streams flow rate varies.

4. Streams temperature flexibility for energy savings in the HEN

From a practical point of view, process conditions are known to change to a certain extent for most
processes. In accordance with this understanding, the plus/minus principle and process changes for the
HEN retrofit have been presented as discussed above, (Linnhoff and Parker, 1984; Zhang and Zhu,
2000). Based on these two ideas, the process streams temperature variation is adopted and integrated with

4
The 13th Asia Pacific Confederation of APCChE 2010
Chemical Engineering Congress October 5-8, 2010, Taipei

the developed path combination approach to further enhance the heat recovery within the HEN system in
this work.

Given the situation where process stream temperature could undergo slight changes, the pinch point
(corresponding to the HRAT) could actually be relaxed within a certain temperature limit beyond its
original value using the streams temperature flexibility (TF). Proper maneuvering of such an effect
within the HEN could result in better heat recovery as demonstrated previously by the plus-minus
principle. Therefore, sensible process changes could be taken positively as a means to reduce utilities
consumption and hence increase the plant energy efficiency, (Zhu and Vaideeswaran, 2000).

The energy savings opportunities provided by shifting the heat load from the HEN utilities through the
utility paths could be increased by relaxing the HRAT beyond the current value. The HRAT could be
relaxed by making the hot streams of the HEN a bit hotter and the cold ones a bit colder according to the
available temperature flexibility while keeping the utility requirements unchanged as shown in Figure 3
(a). Consequently, more heat load could be shifted from the HEN utilities using the paths combination

approach until the HRAT is again restored to its original value, Figure 3 (b). Simply adding ( t hst ) oC to

hot streams and/or subtracting ( tcst ) oC from cold ones would result in increasing the HRAT beyond the
current value. The range of the temperature added to the HRAT value is termed as the temperature
flexibility range which is represented by equation (1) below:
t t TF (1)
hst cst range

For a given maximum value of ( t hst ) and ( t cst ), all the possible TF ranges could be identified and
arranged in a simple matrix R of i rows and j columns. The matrix will display all the possible TF
ranges that resulted from equation (2). The entries of this matrix rij, which correspond to the TF ranges,
could be determined as follows:

i,j { 0,1,2,3,...n} ; rij i j (2)

Therefore, if the maximum allowable ( t hst ) and ( t cst ) is 5oC, then the TF ranges could be represented
by the following square matrix:

0 1 2 3 4 5
1 2 3 4 5 6

2 3 4 5 6 7
R
3 4 5 6 7 8
4 5 6 7 8 9

5 6 7 8 9 10

The repeated values of the TF ranges shown in the matrix (indicated by the diagonal arrows) would
have similar energy savings in the HEN since the same value is to be added to the HRAT. However, each

of the values has been obtained from different combinations of ( t hst ) and ( t cst ). Accordingly, this
provides a degree of freedom for expanding the HRAT value in different manners; i.e. based on the extent
of flexibility for the hot and cold streams to increase and/or decrease their temperature.

5
The 13th Asia Pacific Confederation of APCChE 2010
Chemical Engineering Congress October 5-8, 2010, Taipei

T (oC)
(a)

(b)

H (kW)
Process to process heat Hot utility requirement
recovery (QHU) Cold utility requirement (QCU)

Composite curves after TF Composite curves before TF Smallest driving force in HEN
has applied has applied
(Tmin or HRAT)

Figure 3: Representation for the composite curves when applying the TF concept
(a) Before shifting the heat load, (b) After shifting the heat load

For further explanation, lets suppose the TF range value of 2oC is selected from the matrix above. It
shows that the value is repeated three times, and the three possible combinations for the hot and/or cold
temperature changes that could lead to the TF range value being attained are;

1. Adding 2oC to the hot streams while maintaining the cold ones,
2. Adding and subtracting 1oC to and from hot and cold streams, respectively,
3. Subtracting 2oC from cold streams while maintaining the hot ones.

5. The proposed procedure

The procedure followed to apply the TF concept for further energy reduction in the HEN using the
available utility paths starts with data collection for an existing HEN. It is then followed by the paths
identification and heat load shifting from one utility source to the other while maintaining the heat
balance for each device on the HEN. The heat transfer area for each exchanger is roughly predicted to
calculate the film heat transfer coefficients from the pressure drop correlations. Accordingly, the actual
heat transfer area for each exchanger and for the entire HEN is determined. Energy-Area tradeoff is then
made to judge the reduction in the energy consumption against the additional area requirement. The
temperature flexibility concept is then applied for more heat load shifting. The concept is applied by

increasing the temperature step wisely from 1oC to 5oC for both ( t hst ) and ( t cst ) with the aim to increase
the HRAT value and then resume the heat shifting process. The respective heat loads from the HEN
utilities are shifted until the HRAT is restored back to its original value. This part of the procedure is
indicated by the loop featured in the methodology flow diagram shown in Figure 4 below:

6
The 13th Asia Pacific Confederation of APCChE 2010
Chemical Engineering Congress October 5-8, 2010, Taipei

Existing HEN data

HRAT
check Path(s) identification

Heat load Shifting and adjustment: Q = CP.(T1 T2); EnergyConsumption QH QC

Abefore A after
Exchangers area prediction:
Q before Q after

ht and hs from P equations: PT K PT1 .A.hT 3.5 K PT 2 .hT 2.5; PS KS1.hS2.86 KS 2.A.hS4.42 KS 3.A.hS4.69

1 1 Q
T 1 T 2
Actual area calculation: A ; LMTD ; AHEN A
hT hS LMTD FT T 1
ln
T 2

Energy-Area trade-off

Acceptable No Applying TF in
savings ascending order

Yes

End

Figure 4: Methodology flow chart

Where:

A: exchanger heat transfer area (m2).
CP: Stream heat capacity flow rate
(kW/oC).
KPT1, KPT2, KS1, KS2, KS3: dimensional LMTD: logarithmic mean
constants depend on the stream
physical properties and the exchangers
geometrical configuration.
Qbefore, Qafter: Exchanger heat load
before and after the heat load shifting
(kW).
T1,T2: Exchanger input and output
temperature for either hot or cold
stream (oC).
Abefore and Aafter : Exchanger AHEN: overall HEN heat transfer
heat transfer area before area (m2).
and after the heat load
shifting (m2).
FT: LMTD correction factor hT and hS: exchanger tube and
2o
(assumed to be 1.0 as for shell film coefficients (kW/m . C).
the base case HEN).
Q: Exchanger heat load (kW).
temperature difference.
QH and QC: the heat load of EnergyConsumption: the overall
heater and cooler (kW). heat load of the heaters and
coolers required by the HEN
(kW).
PT and PS: Exchanger T1 and T2: Exchangers hot
tube and shell pressure drop side and cold side temperature
(kPa). difference (oC).

7
The 13th Asia Pacific Confederation of APCChE 2010
Chemical Engineering Congress October 5-8, 2010, Taipei

6. Demonstrative example

The idea of adopting the streams temperature flexibility for reducing the energy consumption in the
HEN using the available utility paths is applied in a case of the HEN taken from the literature (Marcone
and Zemp, 2000; Osman et al., 2009) with all the required data. The grid representation of the case is
shown in Figure 5. The geometrical configuration for each exchanger in the network is in accordance to
the guideline provided by Philippe, (Philippe and Louis, 2007).
o o o o
165 c 147.7 c 118.4 c 95 c
Hot 1 E2 E4 C1

o
3460kw o
o
240 c o
215 c 132.2 c 65 c
Hot 2 E1 E3 C2
5807kw

o
220 c o
158.96 c
o
143.42 c 125 c
o
Cold 1 H1
o 8485kw 2160kw 2560kw
192 c 61 c
o
Cold 2
7153kw
o o
185 c 140 c
o
70 c
Cold 3 H2
2790kw 4340kw

Figure 5: Existing heat exchanger network

From the HEN above, two short utility paths are curried forward to shift the heat load from the heaters
and coolers in order to increase the heat recovery and hence reduce the energy consumption. These paths
are the shortest in the HEN which are defined as follows:

1. Path (A): to shift heat load from both H1 and C1 through E2.
2. Path (B): to shift heat load from both H1 and C2 through E1.

The existing HRAT taken from the HEN is 7.7oC (i.e. the difference between the hot inlet and cold
outlet of exchanger 4); but it was assumed to be 7oC since lower practical HRAT allows more heat
shifting through the utility path. Beside the temperature data which is located on the grid diagram shown
above, the required physical properties data, the remaining stream data, and the exchangers area, heat
load and pressure drop for the existing network are available in work of Osman, (Osman et al., 2009).

7. Results and analysis

The method shown in Figure 4 was applied on the demonstrative HEN example which is shown in
Figure 5 above. The obtained results are mainly reduction in the energy consumption and its
corresponding heat transfer area for the overall HEN before and after applying the TF concept. The
analysis of such results is conducted in this section of the paper.

7.1 Energy-area before applying the TF concept

The existing energy consumption and the overall HEN area are 20542kW and 2187m2, respectively.
Slight changes could be observed in the both energy consumption and the overall HEN area when using
path A for energy reduction without applying the TF concept as shown in Table 1. The small reduction in
the energy consumption is justified by the HRAT limitation which has been reached earlier while
undertaking the heat load shifting from the HEN utilities as described in the method. In contrast to path A,
the energy consumption has remarkably reduced using path B at the expense of 759m2 additional area.

8
The 13th Asia Pacific Confederation of APCChE 2010
Chemical Engineering Congress October 5-8, 2010, Taipei

Therefore, the retrofit solution adopted path B considered to be very promising energy savings options
even before the TF is considered.

Table 1: Energy and area results for the HEN before applying the TF concept
E1 E2 E3 E4 Energy
AHEN (m2)
Q (kW) A (m2) Q (kW) A (m2) Q (kW) A (m2) Q (kW) A (m2) Consumption (kW)
Path A 2160 133.85 2664 638.38 7153 723.23 4340 774.66 20334 2270
Path B 3542 296 2560 587.45 7153 1321 4340 741.94 17778 2946

7.2 Energy-area when applying the TF concept

Figure 6 shows the profile of the energy consumption against the overall heat transfer area while
applying the TF concept in the HEN for both path A and path B. Generally, further reduction in energy
consumption is possible when applying the concept of temperature flexibility (TF) at the expense of
further changes in the additional area requirements.

When subjecting the heat shifting to the TF, Path A showed a trend where the energy consumption has
decreased further with slight increase in the overall HEN area along the TF range. In contrast to path A,
path B showed a trend where the AHEN decreases when applying the TF to reduce the energy consumption
in the HEN. Decreasing the AHEN together with energy consumption is considered to be attractive which
can be justified by rearranging the inlet and outlet temperatures of the HEN devices. The changes in the
temperatures between the exchangers in the HEN while applying the TF leads to affecting the heat
transfer area requirement for each exchanger on the HEN.

3000 0
TF range from (0 - 10) oC
2900
2
Path A
2800 4
Path B
2700 6
8
A_HEN (m2)

2600 10
10
2500 8
6
2400 4
2
2300 0

2200
Existing
2100

2000
15000 16000 17000 18000 19000 20000

Energy Consumption (kW)

Figure 6: Energy-Area performance with TF range

8. Conclusion

Getting the most out of process-to-process heat recovery could be addressed by shifting the heat load
from the HEN utilities using the existing utility paths. However, this might be constrained by the existing
HRAT. Accordingly, the process streams temperature flexibility could offer further chances for extra

9
The 13th Asia Pacific Confederation of APCChE 2010
Chemical Engineering Congress October 5-8, 2010, Taipei

heat load shifting in order to enhance the potentials for the identified paths which could be further
improved. The proposed method is emphasizing the consideration of the existing exchangers pressure
drop where the geometrical configuration is of main concern. Moreover, the topological changes for the
HEN are avoided where only area addition to the existing exchangers is involved to meet the reduction of
energy consumption.

References

1. Asante, N. D. K. and Zhu, X. X. "An Automated Approach for Heat Exchanger Network Retrofit
Featuring Minimal Topology Modifications." Computers Chemical Engineering
20((Supplement)): S7-S12. (1996).
2. Asante, N. D. K. and Zhu, X. X. "An Automated and Interactive Approach for Heat Exchanger
Network Retrofit." Trans. IChemE 75(A). (1997).
3. Campbell, C. J. and Laherrere, J. H. The End of Cheap Oil. Scientific American Magazine. USA.
(1998).
4. Campbell, C. J. "Oil Depletion Protocol: The Rimini Protocol an Oil Depletion Protocol "Heading Off
Economic Chaos and Political Conflict During the Second Half of the Age of Oil"." Energy
Policy 34(12): 1319-1325. (2006).
5. EIA. "Integrated Energy Statistics, Office of Energy Markets and End Use." from
http://www.eia.doe.gov/oil_gas/petroleum/data_publications/weekly_petroleum_status_report/wp
sr.html. (2009).
6. Linnhoff, B. and Tjoe, T. N. Pinch Technology Retrofit: Setting Targets for an Existing Plant. AIChE
National Meeting, Houston. (1985).
7. Linnhoff, B. and Parker, S. "Heat Exchanger Networks with Process Modifications." IChE Annual
research meeting. (1984).
8. Linnhoff-March Introduction to Pinch Technology. Gadbrook Park, Northwich, Cheshire, U.K.,
Targeting House. (1998).
9. Marcone, L. S. and Zemp, J. R. "Retrofit of Pressure Drop Constrained Heat Exchanger Network."
Applied Thermal Engineering 20: 1469-1480. (2000).
10. Osman, A., Abdul Mutalib, M. I., Shuhaimi, M. and Amminudin, K. A. "Paths Combination for Hens
Retrot." Appllied Thermal Engineering 29(15-14): 3103-3109. (2009).
11. Philippe, W. T. and Louis, G. "Minimizing Shell and Tube Heat Exchanger Cost with Genetic
Algorithms and Considering Maintenance." International Journal of Energy Res. 31: 867-885.
(2007).
12. Shenoy, U. V. Heat Exchanger Network Synthesis: Process Optimization by Energy and Recourse
Analysis. Houston, Texas, Gulf Publishing Company. (1995).
13. Smith, R. Chemical Process Design and Integration, John Wiley & Sons Ltd. (2005).
14. Van Reisen, L. B. J., Polley, G. T. and Peter, J. T. V. "Structural Targeting for Heat Integration
Retrofit." Applied Thermal Engineering 18: 283-294. (1998).
15. Van Reisen, L. B. J., Polley, G. T. and Peter, J. T. V. "The Placement of Two Stream and Multi-
Stream Heat Exchangers in an Existing Network through Path Analysis." Computer and Chemical
Engineering 19: S143-S148. (1995).
16. Varbanov, P. S. and Klemes, J. "Rules of Path Construction for Hens Debottlenecking." Applied
Thermal Engineering 20: 1409-1420. (2000).
17. Zhang, J. and Zhu, X. X. "Simultaneous Optimization Approach for Heat Exchanger Network Retrofit
with Process Changes." Ind. Eng. Chem. Res 39: 4963 - 49731. (2000).
18. Zhu, X. X. and Vaideeswaran, L. "Recent Research Development of Process Integration in Analysis
and Optimization of Energy System." Applied Thermal Engineering 20: 1381 - 1392. (2000).

10