in connection with Energitinget at Alvsjo Fair, Stockholm, March 12-13, 2008
Editor: Erik Dahlquist, Malardalen University, Sweden
ISBN 978-91-977493-2-9
Scientific session on Energy saving and Green energy
Chairman: Erik Dahlquist, Mälardalen University
1. Investigations on CO2 absorption using amine solvents in hollow fiber membrane
contactors by experiments .Yuexia Lv1, Xinhai Yu1*, Shan-tung Tu1, J. Yan2, E.
Dahlquist2,School of Mechanical and Power Engineering, East China University of Science and
Technology, Shanghai, 200237, China; 2. Dept. IST, Mälardalen University, Västerås, Sweden
2. Intensification of transesterification for synthesis of biodiesel using microchannel
reactors. Zhenzhong Wen1, Xinhai Yu1*, Shan-tung Tu1, J. Yan2, E. Dahlquist2 1. School of
Mechanical and Power Engineering, East China University of Science and Technology,
Shanghai, 200237, China; 2. Dept. IST, Mälardalen University, Västerås
3. Continuous bioethanol production by fermentation. Bernt Lie, Juan I. Videla.Telemark
University College, Porsgrunn, Norway
4. Optimization and Economic Analysis of Marnoch Thermal Energy Conversion System,
C.C. Chukwu1, G.F Naterer2, M.A. Rosen3, E. Dahlquist4 I.A Marnoch5 ,1,2,3,4 University of
Ontario Institute of Technology, Oshawa. Ontario, Canada, 4 Malardalen University, Vasteras
Sweden, 5 Marnoch Thermal Power Inc. Toronto, Ontario, Canada,
5. Analysis of the coal based polygeneration system of hybrid type Guoqiang Zhang1,2, Lin
GAO1, Hongguang JIN1,* 1 Institute of Engineering Thermophysics, Chinese Academy of
6. Power transformer parameters identification using extended kalman filter Kourosh
Mousavi Takami, TDI researcher and phd student in Malardalen University , Jafar Mahmoudi ,
Erik Dahlquist, Malardalen University
7. An advanced energy saving method for Aluminium - iron - copper bonding using
explosive welding. Mohammad Tabatabaee. TDI institute, Teheran, Iran and Mälardalen
Scientific session on Energy savings in houses
Chairman: Jurek Pyrko, Lund University
8. Design of hot water user profiles for Swedish households based on time
diaries.Magdalena Lundh, Ewa Wäckelgård, Kajsa Ellegård. Department of Engineering
Sciences The Ångström Laboratory Box 534
751 21 Uppsala, Sweden
9. Modeling Household Electricity Load from Time-Use Data. Joakim Widén, Kajsa Ellegård
och Ewa Wäckelgård. Department of Engineering Sciences,Solid State Physics,Uppsala
10. Investigation of electricity consumption in single houses with electrical heating alone
and together with complementary heating systems. Cajsa Bartusch, Fredrik Wallin, Erik
Dahlquist, Mälardalen University.
11. Differences in electricity and hot water consumption in apartments of different sizes.
Iana Vassileva, Cajsa Bartusch, Erik Dahlquist. Mälardalen University.
12. Novel Architect Designs Energy-Saving Iranian Homes. S.M. Esmail Mousavi-Takami,
TDI Technical Development Institute, Teheran, Iran
13. Inform me please! - Energy use information to households and Demand Response.
Jurek Pyrko, Energy Sciencees, Lunds University
14. "Consumers, energy and IT. IT and other tools to support efficient energy
I.C. MariAnne Karlsson
Chalmers University of Technology, Product and Production Development
15. Influence of Residential Ventilation on Radon Mitigation with Energy Saving emphasis
Keramatollah Akbari, Mälardalen University and TDI in Teheran, Iran
16. Extracting alumina from bauxite ore in jajarm factory process simulation using aspen
Mohammad Farrokh and Jafar Mahmoudi, Mälardalen University and TDI in Teheran, Iran
17. Description of techniques for hot spot monitoring and detection in power transformers
Hassan gholinejad, Kourosh Mousavi Takami and Jafar Mahmoudi, Mälardalen University and
Tanesh Company, Iran
18. Payam
[email protected]; [email protected]; [email protected]; Cletus
[email protected]; [email protected]; [email protected];
[email protected]; [email protected];
[email protected]; [email protected]; [email protected];
[email protected]; [email protected]; [email protected];
[email protected];
Yuexia Lv1, Xinhai Yu1*, Shan-Tung Tu1, Jinyue Yan2,3, Erik Dahlquist2
1. School of Mechanical and Power Engineering, East China University of Science and
Technology, Shanghai, 200237, China
2. School of Sustainable Development of Society and Technology, Mälardalen University
SE-721 23 Västerås, Sweden
3. School of Chemial Science and Engineering, Royal Institute of Technology
SE-100 44 Stockholm, Sweden
2. Dept. IST, Mälardalen University, Västerås, SE-721 23 Sweden
Email address: [email protected]
Carbon dioxide absorption from nitrogen and carbon dioxide mixture has been investigated by
experiments using deionized water and low concentration amine solvents including
methyldiethanolamine (MDEA) and monoethanolamine (MEA) in polypropylene (PP) hollow
fiber membrane. A relatively low solvent concentration of 0.05 mol/l MEA and MDEA was used
in the experiment. The dependency of CO2 removal efficiency and mass transfer rate on
operating parameters were studied which include the mixed gas flow rate, the volume
concentration of CO2 at the feed gas inlet, liquid flow rate as well as concentration of absorbents
using lower absorbent concentration. It has been found that the CO2 removal efficiency increases
with the increase of the liquid flow rate and solvent concentration, while the CO2 mass transfer
rate increases with the increase of liquid flow rate, CO2 volume fraction in the feed gas, solvent
concentration and gas flow rate. It has been concluded that the absorbent concentration should be
compromised between absorption efficiency and the membrane wetting to ensure a stable and
efficient removal of CO2 with a long life of the hollow fiber membrane. It was observed by the
experiments that a long operation period may lead to partial membrane wetting even when using
relatively low solvent concentrations, resulting in a significant decrease of membrane absorption
performance, which might be a main drawback for industrial applications with the studied
membranes and solvents.
Keywords: Carbon dioxide; Hollow fiber membrane absorption; Amine solvents; Gas separation
The CO2 emissions are mainly contributed by power generation and transport sectors. CO2
capture and storage (CCS) has been recognized as one of the approaches for mitigating
greenhouse gases. Current CO2 capture systems includes post-combustion capture, oxy-fuel
combustion capture and pre-combustion capture, which make it possible to capture CO2 from
large, centralized sources like power plants and large industries (IPCC, 2005). CO2 capture from
flue gases produced by fossil fuels combustion is referred to as post-combustion capture, which
is corresponding to the most widely applicable option in terms of industrial sectors and is
compatible to a retrofit strategy.
At present, a wide variety of technological options have been used to capture CO2 from flue
gas of power plants, such as chemical and physical absorption, solid adsorption, cryogenic
distillation, and membrane techniques (Mandal et al., 2006; Moon et al., 2006; Andrea Corti et
al., 2004). This paper is focused on the membrane techniques for the CO2 capture. Membrane
gas absorption technology is a hybrid technology that combines membrane separation
technology and chemical absorption technology with advantages of both membrane contactor
and solvent absorption processes. The principle of a membrane contactor is schematically shown
in Fig. 1. Instead of depending on the membrane selectivity, the liquid flowing in a hollow fiber
contactor provides the selectivity and the unselective membrane only acting as the physical
barrier between the liquid and gas phases. CO2 is absorbed in the membrane contactor when the
gas stream contacts with the liquid phase flowing on the opposite side of the membrane.
Gabelman and Hwang (1999) studied the advantages and disadvantages of a membrane
contractor in details. Compared to conventional absorption devices such as packed towers or
bubble columns, the membrane contactor has advantages (Falk-Pedersen et al., 2000), such as
the gas and fluid phase can be manipulated independently avoiding the problems such as
flooding, foaming, channeling and entrainment which are commonly encountered in
conventional absorption devices;
• compact structure, high specific surface area and less voluminous is less energy-consuming;
• the modularity of membrane modules makes the design simple and easy to be scaled up
linearly with predictable performances .
Even though the mass-transfer coefficient of the membrane contactor is inferior to the
conventional devices for the flow of gas and liquid are normally laminar where the turbulent
flow is power-consumption, the membrane gas absorption due to its large interfacial area, is still
considered as one promising alternative to conventional and potential large scale application
technology for CO2 recovery and removal (Li et al., 2005).
Fig. 1. Schematic drawing of CO2 absorption in a hollow fiber membrane contactor.
Zhang and Cussler (1985) were the pioneers who proposed the idea of CO2 absorption by
sodium hydroxide in a hollow fiber membrane contactor. In recent years, hollow fiber membrane
technology has been extensively studied. Various liquid absorbents including pure water and
aqueous solutions of NaOH, KOH, monoethanolamine (MEA), diethanolamine (DEA), 2Amino-2-mechyl-1-propanol (AMP), N-methyldiethanolamine (MDEA), CORAL and the
potassium salt of glycine and taurine were used as absorption liquids in polyethylene (PE) or
polypropylene (PP) or polytetrafluoroethylene (Teflon) microporous hydrophobic hollow fiber
membrane contactors, in which the MDEA and MEA aqueous solutions in PP hollow fiber
membrane contactor are the most widely used for CO2 absorption (Kumar et al., 2002; Kim et
al., 2000; Mandal et al., 2001). Compared with other fibers, PP fibers are less expensive and
commercially available in a wider size range. Wang et al. (2004) performed a theoretical
simulation to study CO2 absorption using three typical alkanolamine solutions of AMP, DEA and
MDEA in a hollow fiber membrane contactor. The effects of operating parameters, membrane
configuration, module structures and different solvents on absorption flux and removal efficiency
were investigated. Lu et al., (2007) studied the effects of activators 2-Amino-2-mechyl-1propanol (AMP) and piperazine (PZ) on mass-transfer enhancement using MDEA alkanolamine
solutions in a hollow fiber contactor and concluded that the mass transfer fluxes of the activated
MDEA solutions are significantly higher than that of MDEA solution and effects of operation
conditions on mass-transfer enhancement are limited. Theoretical simulations and corresponding
experiments were carried out to describe the CO2 absorption by distilled water and aqueous
diethanolamine (DEA) solutions for better understanding of CO2 absorption in a hollow fiber
membrane contactor (Zhang et al., 2006). A theoretical model was developed based on CO2
absorption simulation by water and DEA under the wetted and the non-wetted operation modes
to study the influence of membrane wetting on CO2 capture in microporous hollow fiber
membrane contactors (Zhang et al., 2007). Matsumiya et al. (2005) developed a novel facilitated
transport membrane system where the feed gas and aqueous diethanolamine solutions were
supplied to the lumen side of the hollow fiber ultrafiltration membrane module with an upward
flow. In addition, they evaluated the energy consumption and compared with conventional
separation processes. Liu et al. (2005) examined the mass transfer performances using coiled
hollow fiber membrane modules. Their results showed that coiled hollow fiber modules can
remarkably enhance the mass transfer compared with conventional straight module. Most
researches were limited to atmospheric pressure applications using aqueous absorption solvents,
so CO2 absorption in propylene carbonate at elevated pressure using hollow fiber membrane
contactor was investigated by Dindore et al. (2004). The results showed that the decrease in the
binary gas phase diffusivity and hence the membrane mass transfer coefficient due to gas
pressures increase does not have a significant impact on the overall mass transfer coefficient.
As mentioned above, a variety of experiments and simulations have been carried out to study
the membrane absorption performance using different kind of solvents in the hollow fiber
membrane contactor. The solvent concentration used by most researches was varied from 1mol/l
to 3mol/l. However, it was reported that the surface tension of most absorbent solutions
significantly decreases with increasing solution concentration (Vázquez et al., 1997; Rinker et
al., 1994), which resulted in PP membrane wetting according to the Laplace-Young equation
(Franken et al., 1987), thus reducing the membrane performance. According to Laplace-Young
equation, lower solvent concentration can reduce membrane wetting and prolong the membrane
service life.
In this paper, the effects of operating parameters on membrane absorption performance using
solutions of relative low concentration were investigated in the pilot-scale hollow fiber
membrane module. The wetting phenomenon of PP membrane, which is often neglected in
reported studies, was also observed in prolonged operation. In addition, the optimal operating
parameters were obtained for the specified membrane module.
The experiments were performed to obtain the operating parameters effect on the absorption
using deionized water and MEA and MDEA aqueous solutions of low concentration as the
absorbent in a PP hollow fiber membrane contactor.
Table 1. Specifications of the hollow fiber membrane module
Module outer diameter (mm)
Module inner diameter (mm)
Module length (mm)
Fiber inner diameter (μm)
Fiber outer diameter (μm)
Fiber length (mm)
Number of fibers
Fiber porosity
Pore size (μm)
Contact area (m2)
A sketch of CO2 absorption in hollow membrane contactor is shown in Fig. 2. And the photo
of experimental setup is shown in Fig. 3. The HDMF-100-1 type microporous polypropylene
hollow fiber membrane module (Tianjin Blue Cross Membrane Technology Co., Ltd.) was used
as the contactor in this study. The characteristics of the membrane module are listed in Table 1.
A gas mixture containing CO2, N2 with various volume ratios was selected as the feed gas.
Deionized water, aqueous solutions of Monoethanolamine (MEA), methyldiethanolamine
(MDEA) were chosen as absorption liquids. The 99.5% grade MEA and MDEA (Shanghai
Bangcheng Chemical Co., Ltd.) were dissolved in deionized water to make aqueous solutions of
0.05mol/l-0.25mol/l. The gas phase and liquid phase flowed countercurrent through the module
(the gas passed through the shell side, and the absorbent flowed countercurrent through the
lumen side of the hollow fibers).
In the experiment, the feed gas was introduced into the system from compressed gas cylinders
and the flow rate was adjusted by Mass Flow Controller (Sevenstar Electronics Co., Ltd. MFC
D07) which can precisely control the gas flow rate. Then the gas was introduced into the static
mixer where the mixtures can be mixed uniformly. Pressure gauges at the inlet and outlet of the
membrane module measure the gas pressures, and outlet gas flow rate was measured by a mass
flow meter (Sevenstar Electronics Co., Ltd.). The inlet and outlet gas compositions were
analyzed on-line by a 9790 Ⅲ gas chromatograph (FuLi Analytical Instrument Co., Ltd.) using a
thermal conductivity detector (TCD). A stainless steel peristaltic pump (Tian Li Liquid Industrial
Equipment Factory) was used to pump the liquid into the lumen side of the hollow fibers from
solvent container, and the flow rate of liquid was controlled by a rotational flow meter. The
concentrations of the inlet and outlet absorption liquids were measured by a chemical titrimetric
Pressure gauge
Static mixer
Liquid out
Sample analysis
Gas chromatograph
Gas in
Flow meter
Gas out
Liquid in
Pressure gauge
Fig.2. Sketch of experimental system.
Fig.3. Experimental setup.
Before each run of an experiment, the system was operated for 15 min by deionized water to
eliminate the influence of the former experiment. All data were collected at steady state, after at
least 20 min of operating time. Steady state will be indicated by a constant CO2 concentration in
the outlet gas stream. Under the same operating conditions, five samples of the gas and absorbent
samples were taken and the average value was calculated. All experiments were carried out at
atmospheric pressure (0.1MPa) and at room temperatures.
Removal efficiency (η) and mass transfer rate of CO2 ( J CO ) were used to describe the
separation properties of hollow fiber membrane module using low concentration absorbents,
which can be calculated by Eq.(1) and Eq.(2) (Kumar et al., 2002; Yeon et al., 2005):
Qin × Cin − Qout × Cout
Qin × Cin
J CO2 =
( Qin × Cin − Qout × Cout ) × 273.15
0.0224 × Tg × S
Where η denotes the CO2 removal efficiency, %; J CO2 is the CO2 mass transfer rate, mol/ (m2·h);
Qin and Qout represent the inlet and outlet gas flow rate respectively, m3/h; Cin and Cout are the
CO2 volumetric fraction in the gas inlet and outlet respectively, %; Tg is the gas temperature, K;
S represents the gas-liquid mass transfer area and herein equals to the effective membrane area,
3.1. Effect of liquid flow rate on CO2 removal efficiency and CO2 mass transfer rate
In Fig. 4 and Fig. 5, the CO2 removal efficiency and mass transfer rate are plotted against
liquid flow rate using 0.05 mol/l MEA, 0.05 mol/l MDEA and deionized water as the absorbent
respectively. The CO2 volume fraction in feed gas is 40 % and the gas flow rate is 150 ml/min.
As shown in Fig. 4 and Fig. 5, the liquid flow rate has an important influence on both CO2
removal efficiency and CO2 mass transfer rate. As the liquid flow rate increases, CO2 removal
efficiency and CO2 mass transfer rate increase. Even though physical absorption using deionized
water, the CO2 removal efficiency could reach 60.3% as the liquid flow rate is 68 ml/min. With
the increase of liquid flow rate, the liquid disturbance is enhanced, which results in a higher
speed of CO2 diffusing into the liquid. The consumed absorbents at the membrane boundary
layer could diffuse into the liquid phase at a higher speed due to the increase of liquid flow rate.
Therefore, the gas-liquid interface could be maintained at a higher solvent concentration, which
increases the CO2 removal efficiency. In addition, as the increase of liquid flow rate, the
thicknesses of gaseous and liquid-phase boundary layers decrease, leading to enhancement of the
mass transfer rate. However, CO2 removal efficiency and mass transfer rate increase slowly at
higher liquid flow rate due to the limited boundary layer thickness, which are also shown in the
Fig. 4 and Fig. 5.
CO2 removal efficiency (%)
0.05mol/L MDEA
Liquid flow rate (ml/min)
Fig. 4. Influence of liquid flow rate on CO2 removal efficiency.
JCO2 E2 ( mol/m .h)
0.05mol/L MEA
0.05mol/L MDEA
Liquid flow rate (ml/min)
Fig. 5. Influence of liquid flow rate on CO2 mass transfer rate.
3.2. Effect of CO2 volume fraction in feed gas on CO2 removal efficiency and mass transfer
The influence of inlet CO2 volume fraction at the inlet of feed gas on the removal efficiency
and mass transfer rate is shown in Fig.6 and Fig.7, respectively. Deionized water, 0.05 mol/l
MEA and 0.05 mol/L MDEA were used as the absorbent. The gas flow rate is 150 ml/min and
the liquid flow rate is 17 ml/min. The CO2 volume fraction in the feed gas varied from 10 % to
40%. Fig. 6 demonstrates that the CO2 removal efficiency decreases with the increase of CO2
volume fraction in the feed gas. As the CO2 is absorbed in the liquid through physical absorption
or chemical absorption, more liquid is consumed with higher CO2 concentration at the liquid
membrane. The liquid will be insufficient relative to higher CO2 concentration at a constant
liquid flow rate, which results in a decrease of CO2 removal efficiency.
Fig.7 shows that the increase of CO2 volume fraction could effectively enhance the mass
transfer rate. With the increase of CO2 volume fraction, the CO2 concentration gradient at the
liquid-gas boundary layer increases. That is, the CO2 driving force of mass transfer in the gas is
enhanced, which leads to the increase of CO2 diffusion mass transfer rate. Therefore, more CO2
is absorbed in the liquid by permeating the membrane module.
CO2 removal efficiency (%)
0.05mol/L MEA
0.05mol/L MDEA
CO2 volume fraction in feed gas (vol %)
Fig. 6. Influence of CO2 volume fraction at the feed gas inlet on the CO2 removal efficiency.
JCO2 E2 ( mol/m .h)
0.05mol/L MEA
0.05mol/L MDEA
CO2 volume fraction in feed gas (vol %)
Fig. 7. Influence of CO2 volume fraction at the feed gas inlet on CO2 mass transfer rate.
3.3. Effect of solvent concentration on CO2 removal efficiency and mass transfer rate
The influences of the absorbent concentration on the CO2 removal efficiency and mass
transfer rate are plotted in Fig. 8 and Fig. 9. Deionized water, MEA and MDEA with the liquid
concentration varying from 0.05 mol/l to 0.25 mol/l were used as the absorbent. The gas flow
rate is 150 ml/min with 40 vol.% CO2 in the feed gas and the liquid flow rate is 17 ml/min. As
shown in Figs. 8 and 9, the CO2 removal efficiency and mass transfer rate obviously increase
with the increase of the concentration of MEA and MDEA. With the increase of absorbent
concentration, the effective component absorbing CO2 in the liquid boundary layer increases,
resulting in higher CO2 transfer rate into the liquid. As CO2 enters the liquid and reacts with the
corresponding solvent, the CO2 concentration decreases in liquid-gas boundary layer. It enhances
the CO2 solubility rate and increases the CO2 removal efficiency. The CO2 removal efficiency
can be as high as 95 % with the MEA concentration of 0.25 mol/l.
As shown in Fig. 8, the CO2 removal efficiency of MEA and MDEA is much higher than that
of deionized water under the same operating conditions. This might be caused by more CO2
consumed in the presence of MEA and MDEA by the chemical reaction. In addition, Fig. 8 also
indicates that the CO2 removal efficiency of MEA is higher than that of MDEA especially at
high absorbent concentration due to the higher rate of MEA reacting with CO2.
As discussed above, a higher removal efficiency and mass transfer rate can be effectively
achieved by increasing the solvent concentration. However, with the increase of concentration of
the solvent, the PP membrane is more likely to be wetted and its performance is deteriorated. It
can be explained by that the surface tension of aqueous MEA and MDEA solution decreases
with the increase of the amine concentration according to the Laplace-Young equation. The
membrane is successfully operated for about 60 h without being wetted by using MEA and
MDEA at the concentration of 0.05 mol/l. Therefore, the absorbent concentration should be
compromised between removal efficiency and the wetting to ensure a stable and efficient
absorption of CO2 over a long life of the hollow fiber membrane. Many researches on improving
non-wettability have shown promising results. Studies have shown that the performance of
polyethylene (PE) membranes could be greatly improved by a hydrophobic treatment to its
surface using fluorocarbonic materials (Nishikawa et al., 1995). The wetting problem can also be
achieved by coating the membrane with a very thin permeable layer (Kreulen et al., 1993).
However, a hollow fiber membrane with non-wettability is still required to be further
investigated and developed.
CO2 removal efficiency (%)
Solvent concentration (mol/L)
Fig. 8. Influence of solvent concentration on CO2 removal efficiency.
JCO2 E2 ( mol/m .h)
Solvent concentration (mol/L)
Fig. 9. Influence of solvent concentration on the mass transfer of CO2.
3.4. Effect of gas flow rate on CO2 removal efficiency and CO2 mass transfer rate
Effect of gas flow rate on the CO2 removal efficiency and mass transfer rate are shown in Figs.
10 and 11. Deionized water, 0.05 mol/l MEA and 0.05 mol/l MDEA were used as the absorbent.
The gas flow rate was varied from 75 ml/min to 200 ml/min with 40 vol. % CO2 in the feed gas
and the liquid flow rate is 17 ml/min. It is clearly shown in Fig.11 that there is a difference in the
effects of gas flow rate on CO2 mass transfer rate of MEA, MDEA aqueous solutions and
deionized water due to some uncontrolled factors. With the decrease of gas retention time, the
CO2 concentration at the gas-liquid interface increases, resulting in an increase of the mass
transfer rate for MEA and MDEA aqueous solutions. Although increase of gas flow rate can
reduce the thickness of gas boundary layer and enhance the gas mass transfer, which is favorable
for the CO2 removal. However, it simultaneously decreases the residence time of gas in the
membrane contactor, which is unfavorable for the CO2 removal. The combined effects result in
the tendency as shown in Fig. 10 and Fig. 11. It indicates that the residence time plays an
important role in the removal of CO2.
0.05mol/L MEA
0.05mol/L MDEA
CO2 removal efficiency (%)
Gas flow rate (ml/min)
Fig. 10. Influence of gas flow rate on CO2 removal efficiency.
JCO2 E2( mol/m .h)
0.05mol/L MEA
0.05mol/L MDEA
Gas flow rate (ml/min)
Fig. 11. Influence of gas flow rate on CO2 mass transfer rate.
Experiments have been carried out to investigate the effect of operating parameters on the
efficiency of carbon dioxide removal using deionized water and low concentration MDEA and
MEA in polypropylene hollow fiber membrane. It has been found that the CO2 removal
efficiency increases with the increase of the liquid flow rate and solvent concentration, while the
CO2 mass transfer rate increases with the increase of liquid flow rate, CO2 volume fraction in the
feed gas, solvent concentration and gas flow rate. It is concluded that the absorbent concentration
should be compromised between absorption efficiency and the membrane wetting to ensure a
stable and efficient removal of CO2 with a long life of the hollow fiber membrane. It has also
been observed by the experiments that a longer operation period render partial membrane
wetting even using relative lower solvent concentration, resulting in significant decrease of
membrane absorption performance. It is indicated that the decrease of residence time plays more
important role in the removal of CO2 than the intensification of the gas mass transfer.
The work described in this paper has been carried out with the financial support of the Swedish
Research Links Programme and the China Natural Science Foundation (contract No. 20606011)
for which due acknowledgement is given. We also would like to thank Mr. H. L. Li and Ms. Y.
H. Huang for the help in the experimental setup.
IPCC 2005. IPCC Special Report on Carbon Dioxide Capture and Storage. Chapter 3.
B.P. Mandal, S.S. Bandyopadhyay. 2006. Absorption of carbon dioxide into aqueous blends of
2-amino-2-methyl-1-propanol and monoethanolamine. Chemical Engineering Science 61: 54405447
S.H. Moon, J.W. Shim. 2006. A novel process for CO2/CH4 gas separation on activated carbon
fibers-electric swing adsorption. Journal of Colloid and Interface Science 298: 523-528.
A. Corti, D. Fiaschi, L. Lombardi. 2004. Carbon dioxide removal in power generation using
membrane technology. Energy 29: 2025-2043.
A. Gabelman, S.T. Hwang. 1999. Hollow fiber membrane contactors. Journal of Membrane
Science 159: 61-106.
O. Falk-Pedersen, M.S. Grønvold, H. Dannström, D.B. Stuksrud. Gas treatment using membrane
gas/liquid contactors, Proceedings of the Fifth International Conference on Greenhouse Gas
Control Technologies, August 13-16, 2000. Cairns Convention Centre, Collingwood, Australia.
J.L. Li, B.H. Chen. 2005. Review of CO2 absorption using chemical solvents in hollow fiber
membrane contactors. Separation and Purification Technology 41: 109-122.
Q. Zhang, E.L. Cussler. 1985. Microporous hollow fibers for gas absorption: Mass transfer in the
liquid. Journal of Membrane Science 23: 321-332.
Q. Zhang, E.L. Cussler. 1985. Mocroporous hollow fibers for gas absorption: Mass transfer
across the membrane. Journal of Membrane Science 23: 333-345.
P.S. Kumar, J.A. Hogendoorn, P.H.M. Feron, G.F.Versteeg. 2002. New absorption liquids for
the removal of CO2 from dilute gas streams using membrane contactors. Chemical Engineering
Science 57: 1639-1651.
S.-H. Yeon, K.-S. Lee, B. Sea, Y.-I. Park, K.-H. Lee. 2005. Application of pilot-scale membrane
contactor hybrid system for removal of carbon dioxide from flue gas. Journal of Membrane
Science 257: 156-160.
Y.S. Kim, S.M. Yang. 2000. Absorption of carbon dioxide through hollow fiber membranes
using various aqueous absorbents. Separation and Purification Technology 21: 101-109.
B.P. Mandal, M. Guha, A.K. Biswas, S.S. Bandyopadhyay. 2001. Removal of carbon dioxide by
absorption in mixed amines: modeling of absorption in aqueous MDEA/MEA and AMP/MEA
solutions. Chemical Engineering Science 56: 6217-6224.
R. Wang, D. F. Li, D. T. Liang. 2004. Modeling of CO2 capture by three typical amine solutions
in hollow fiber membrane contactors. Chemical Engineering and Processing 43: 849-856
J.G. Lu, Y.F. Zheng, M.D. Cheng, L.J. Wang. 2007. Effects of activators on mass-transfer
enhancement in a hollow fiber contactor using activated alkanolamine solutions. Journal of
Membrane Science 289: 138-149.
H.Y. Zhang, R. Wang, D.T. Liang, J.H. Tay. 2006. Modeling and experimental study of CO2
absorption in a hollow fiber membrane contactor. Journal of Membrane Science 279: 301-310.
H.Y. Zhang, R. Wang, D.T. Liang, J.H. Tay. 2007. Theoretical and experimental studies of
membrane wetting in the membrane gas-liquid contacting process for CO2 absorption. Journal of
Membrane Science. Doi: 10.1016/j. memsci. 2007.09.050
N. Matsumiya, M. Teramoto, S. Kitada, H. Matsuyama. 2005. Evaluation of energy consumption
for separation of CO2 in the gas by hollow fiber facilitated transport membrane module with
permeation of amine solution. Separation and Purification Technology 46: 26-32.
L.Y. Liu, L.J. Li, Z.W. Ding, R. Ma, Z. Yang. 2005. Mass transfer enhancement in coiled hollow
fiber membrane contactor. Journal of Membrane Science 279: 301-310.
V.Y. Dindore, D.W.F. Brilman, P.H.M. Feron, G.F. Versteeg. 2005. CO2 absorption at elevated
pressures using a hollow fiber membrane contactor. Journal of Membrane Science 235: 99-109.
G. Vázquez, E. Alvarez, J.M. Navaza, R. Rendo, E. Romero. 1997. Surface tension of binary
mixtures of water + monoethanolamine and water + 2-amino-2-methyl-1-propanol and tertiary
mixtures of these amines with water from 25Ԩ to 50Ԩ. Journal of Chemical Engineering 42:
E.B. Rinker, D.W. Oelschlager, A.T. Colussi, K.R. Henry, O.C. Sandal. 1994. Viscosity, density
and surface tension of binary mixtures of water and mixtures of these amines with water over the
temperature range 20-100Ԩ. Journal of Chemical Engineering 39: 392-395.
A.C.M. Franken, J.A.N. Nolten, M.H.V. Mulder, D. Barrgeman, C.A. Smolders. 1987. Wetting
criterion for the applicability of membrane distillation. Journal of Membrane Science 33: 315328.
N. Nishikawa, M. Ishibashi, H. Ohta, N. Akutsu, H. Mastsumoto, T. Kamata, H.Kitamura. 1995.
CO2 removal by hollow fiber gas liquid contactor. Energy Conversion and Management 36: 415418.
H. Kreulen, C.A. Smolders, G.F. Versteeg, W.P.M. van Swaaij. 1993. Microporous hollow fiber
membrane modules as gas-liquid contactors Part 2. Mass transfer with chemical reaction.
Journal of Membrane Science 1993: 217-238.
Zhenzhong Wen1, Xinhai Yu1*, Shan-Tung Tu1, Jinyue Yan2,3, Erik Dahlquist2
1. School of Mechanical and Power Engineering, East China University of Science and
Shanghai, 200237, China
2. School of Sustainable Development of Society and Technology, Mälardalen University
SE-721 23 Västerås, Sweden
3. School of Chemial Science and Engineering, Royal Institute of Technology
Stockholm, SE-100 44 Stockholm, Sweden
Email: [email protected]
In this paper, transesterification of soybean oil with methanol in the presence of concentrated
sodium hydroxide as a catalyst has been experimentally investigated in a zigzag micro-channel
reactor. The operating parameters have been optimized which include molar ratio of oil to
methanol, catalyst quantity, volumetric rate, and reaction temperature. Compared with the
contact time for complete conversion from several minutes to several hours in conventional
reactors, the corresponding contact time is significantly reduced to 28 seconds at the conditions
of oil-to-methanol molar ratio of 1: 9 and temperature of 56oC with catalyst of 1.2 % (weightbased). Intensification of the mixing of the immiscible reactants in transesterification reaction by
the zigzag micro-channel is the most influence factor, which is characterized by optical method.
It is shown that chaotic intensification mixing in micro-scale is an effective way to improve the
efficiency of the biodiesel synthesis process.
Keywords: Biodiesel; Micro-channel reactor; Mixing; Transesterification
Biodiesel is a renewable fuel that can be produced from vegetable oils, animal fats, grease, or
waste frying oil. Due to the increased price of the petroleum and the environmental concern
including both local pollution from the vehicle exhusts and global climate changes, biodiesel is
becoming more attractive biofuel of high interests. (Marchetti et al, 2007). In addition, biodiesel
is an alternative of non-toxic and biodegradable fuel, and essentially free of sulfur and aromatics
(Karmee et al., 2005). Biodiesel has been produced by transesterification of triglyceride to
methyl esters using homogeneous basic catalysts (Vicente et al, 2004), such as sodium or
potassium hydroxides, carbonates or alkoxides, as represented by the following equation
(Marchetti et al, 2007):
+ 3R’OH
Most of biodiesel production are batch processes, which need more reagents and are
generally of lower efficiency. Recently, new production technologies have been studied
including e.g., ultrasonic irradiation (Stavarache et al. 2007), supercritical transesterification (van
Kasteren et al., 2007) and fixed-bed reactors using heterogeneous catalysts (Xie et al.,
2006)..Micro-channel reactors have high volume/surface ratio, short diffusion distance, fast and
efficient heat dissipation and mass transfer (Kobayashi et al., 2006). Those features have been
used in organic synthesis (Mason et al., 2007). In this paper, a zigzag micro-channel reactor has
been developed and applied in the transesterification process. Process performance has been
evaluated on intensification of mass transportation by the chaotic mixing.. Based on
experimental results, a highly efficient and compact reactor is proposed for biodiesel production
in a mini- fuel processing plant.
Objectives of the experiment
The objectives of the experiments are investigating various parameters influencing biodiesel
production and mixing performance between immiscible reagents. Parameters in micro-channel
reactions such as catalyst rates, temperatures, methanol/oil ratios, residence time were discussed.
The catalyst rates varied as 0.55, 0.67, 0.88, 1, 1.2, 1.5, 2% (by wt. of oil). The temperatures
were varied as 40, 45, 50, 56, 62, 70, 75°C. The molar ratios (methanol/oil) were varied as 4:1,
5:1, 6:1, 9:1, 10.5:1, 17:1. The residence times were varied as 14, 18, 28, 32, 39seconds. Mixing
performance influenced by different residence times, batch to micro-channel under the same
condition were discussed. Before testing mixing performance, methyl violet was premixed in
methanol so as to be observed in optical microscope conveniently.
Experimental set-up
The sketch of the experimental setup is presented in the Fig. 1. Two syringe pumps (Harvard
Apparatus, USA) were used to inject the fluids of different flow rates. The soybean oil and
methanol aggregated through a T shape three-way junction (the outlet diameter of 200μm). The
outlet of the junction is linked with the inlet of the zigzag micro-channel reactor. The reactor is
immersed in a water bath, to ensure the accuracy of reaction temperature. The outlet of the
micro-channel reactor is connected with a collection flask placed in an ice-water vessel in order
to terminate the transesterification reaction quickly. Sodium hydroxide was dissolved in
methanol as the catalyst by electro-stirring before experiments. The biodiesel samples were
analyzed by using a gas chromatograph (FULI9790) with a capillary column for identification
and quantification. Methyl Undecanoate (Fluka) was used as the internal standard solvent.
Residue recycled
Premixing of
NaOH and Methanol
And Refining
Residue recycled
Fig.1. Schematic drawing of the continuous transesterification of soybean oil in a micro-channel
Micro-channel reactor fabrication
The micro-channel reactor was made of stainless steel of 316L (Fig.2). Three types of
patterned sheets were prepared to construct the reactor: cover sheet, medium sheet and bottom
sheet. The cover sheet has two holes, which act as the flow paths, The medium sheet, shown in
Fig.3, has one zigzag micro-channel. The dimension of micro-channel is 200μm (W) X 300μm
(D) X 1.07m (L). The thickness of the bottom sheet is 0.5 mm without holes.. Surfaces of all
sheets in three types were polished to a roughness of 2μm followed by cleaning in acetone prior
to diffusion bonding. The dimensions of the micro-channel reactor are about 72 mm × 72 mm ×
4 mm, respectively. The effective volume of the micro-channel reactor is 0.125 cm3, including
the micro-channels and the inlet and outlet zones.
Fig.2. Photo of the micro-channel reactor.
micro-channel sheet.
Fig.3. Schematic drawing of the
Characterization of the experimental set up is conducted by applying the method of optical
measurement. Determining droplet size distribution requires a few minutes, coalescence of
methanol droplets emerge immediately as the sample removed from collection vessel. Therefore,
a stabilizing agent or surfactant needed to be added to the dispersion to slow the coalescence
phenomenon (Wu et al, 2007).
Decaglycerol monooleate ester as a surfactant was dissolved in methanol before mixing to
avoid coalescence phenomenon, which is adverse for measurement. The experiment on mixing
was carried out in the micro-channel reactor and the three-necked batch reactor without the
catalyst of sodium hydroxide. The size of the droplets of the emulsion was identified using an
optical microscope (Sharpscope SF-1 Industrial Microscope, Nanjing, China) with supporting
software (VIS ver 2.90 professional-Digital Imaging and Measured System) that allows 950X
enlargement and has a color camera, which sends a signal to a personal computer. The software
was used to measure the diameters of methanol droplets directly on a monitor by a three-point
measurement technique. As the diameter distribution is dependent on the population (Lemenand
et al., 2003), a large number of samples (600) for the measurement were measured to ensure the
accuracy of the results.
Since the measurements of droplet diameter are based on sampling a large number of
experiments (Lemenand et al. 2003), It is importantly to determine the number of droplets of the
samples so that the diameter distribution is independent of the population. The empirical method
used here entails concerning errors between two distributions, the error is evaluated by a
normalized standard deviation, defined by
SD n =
∑ i (f i − f i' )
are the droplet frequencies of the same class diameter and Nc is the class
number. For this case, the normalized standard deviation between two successive distributions is
as a function of number of droplets, this minimal number of 600 is needed for droplet diameter
measurement so as to assure the reproducibility.
Comparison between the micro-channel reactor and impeller batch reactor
Fig. 4 shows the difference in the soybean oil conversion between the impeller batch-reactor
under the conditions: temperature of 56 °C, methanol/oil molar ratio of 9:1, catalyst amount of
1.2% (weight fractions of the oil supplied). The experimental results show the time scale for
stirring in the impeller batch reactor compared to the contact time for the micro-channel reactor.
It can be seen from the Fig. 4 that the conversion of soybean oil in micro-channel reactor is
higher even if the time for the micro-channel reactor is only one tenth the corresponding value
for the batch reactor. In the absence of the catalyst, the mixing experiments were carried out in
the same conditions.. Two photos of the emulsions formed in the mixing experiments are
presented in Fig. 5. It is indicated that the diameter of the droplets in the emulsion by the microchannel reactor is smaller than that by the batch reactor with an agitator. Several big droplets can
be seen in the emulsion by the batch rector whereas the droplets by the micro-channel reactor are
of approximately the same size. The distribution and mean droplet size were calculated by 600
droplet samples. The results are shown in Fig. 6. The mean droplet size for micro-channel is
4.21μm, smaller than that of 4.92μm for the batch reactor. As immiscible fluids mix using a
batch stirrer, convection is induced, resulting in a turbulence and chaotic mixing. In a mixture of
two immiscible liquids, there is a critical agitator speed, NC, at which a separated immiscible
layer disappears (Shinji, 1975). Nc can be calculated by the empirical equation:
1/ 9
⎛μ ⎞
N c = KD −2 / 3 ⎜ c ⎟
⎝ ρc ⎠
⎛ ρc − ρd ⎞
⎝ ρc ⎠
100 200 300 400 500 600
Residence time, s
Fig.4. Comparison between the zigzag micro-channel reactor and batch impeller.
In which K is a proportionality coefficient, ρc and ρd are the densities of the continuous and
dispersed phase, respectively, ρc is the viscosity of the continuous phase, and D is the vessel
diameter. As for the mixing of methanol and soybean oil, the following parameter values are
used to estimated this speed (Wu et al, 2007), K was taken as 750 for the agitation in center of
the vessel, ρc is 920 kg/m3, ρd is 791 kg/m3, μc is 0.05 Pa s, and D is 17cm. The minimum critical
agitator speed is 493 rpm, which means the agitator speed of 600 rpm in the experiment is
acceptable. The shear forces that cause the convection are obviously dampened away from the
stirrer and most of the bottom flask and axis of the impeller. Therefore, longer time is needed for
mixing to avoid the disadvantages of convective dead zones (Alvarez et al., 2002). In addition,
the dampened shear force also results in the broad distribution of droplet size as shown in Fig. 6,
which in accordance with a unsteady-state of the emulsion by an agitator. Taking advantage of
the recirculation phenomenon, the mixing efficiency of zigzag micro-channels could be
significantly improved compared to straight micro-channels (Virginie et al., 2002). These
recirculations induce a transversal component of velocity, which improves the mixing process.
Therefore, the smaller and size and narrow distribution of droplet can be effectively achieved in
the zigzag micro-channel as shown in Fig. 5 and Fig. 6.
Fig.5. Photograph of droplets in the emulsion generated by (a) batch impeller, (b) zigzag
microchannel reactor.
Number distribution,%
Batch experiment
Microchannel experiment
Droplet size,um
Fig.6. Comparison of differential droplet distributions for batch impeller and the zigzag microchannel reactor
Banks and coworkers (Banks et al., 2003) derived rate expressions for a liquid-liquid
interfacial reaction system for copper extraction from an aqueous solution into an organic phase.
If mixing in both the dispersed and continuous phase is good, then the bulk concentrations of
each phase will be near to that of the interfaces and the change in concentration with time is
d [ B]
k 4π r 2
N [ B]
Where V is the volume of the dispersed phase, r is the droplet radius and N is the total number of
particles. The effective rate constant for the process should correspond to the surface rate
constant by
K eff ( r ) =
k 4πr 2
S droplet
As the mean droplet size for the impeller agitation is 1.17 times larger than that for the microchannel reactor, the effective rate constant for the micro-channel reactor could be as much as 1.4
times higher than that of the conventional agitation system. Therefore, the zigzag micro-channel
might improve the efficiency of the transesterification effectively by reductions in the mean
droplet size.
Performance of the micro-channel
The effect of catalyst amount on the performance of the micro-channel is shown in Fig. 7.
The catalyst rate was varied in the range of 0.55 -2%. These percentages were weight fractions of
the oil supplied for this reaction. It can be seen that the conversion rate was increased first with
the increase of catalyst rate from 0.55% to 1.2%. But, with further increase in the catalyst rate,
the conversion rate was decreased, which might be due to the formation of soap (Meka et al.,
In the presence of homogeneous catalyst, the reaction mixture constitutes of a two-phase
system. One is methanol-catalyst, and another is oil. In general, the rate of the transesterification
reaction can be accelerated with the increase of the reaction temperature. As shown in Fig. 8, the
conversion increased with reaction temperature to nearly 97.3% at 56 °C and then decreased with
the further increase of temperature above 56 °C. A large number of bubbles were observed at the
outlet of the micro-channel reactor as the temperature was increased to 65 °C. This might be
caused by the vaporization of methanol at temperatures around 56oC and higher. The bubbles
reduce the interface of oil and methanol. On the other hand, the flow rate of the mixture was
accelerated in the existence of the bubbles. The combination of both phenomena presents
unfavorable impacts on the conversion of soybean oil. Based on this, the appropriate reaction
temperature shall be considered to be around 56 °C.
Catalyst amount,(wt,%)
Fig.7. Influence of catalyst amount on the conversion.
Reaction conditions: methanol/oil molar ratio 6:1, residence time 18 seconds and temperature 63
35 40 45 50 55 60 65 70 75 80
Temperature( ℃ )
Fig.8. Influence of reaction temperature on the conversion.
Reaction conditions: methanol/oil molar ratio 6:1, catalyst amount 1.2%, at the residence time 28
Stoichiometrically, the methanolysis of soybean oil requires three moles of methanol for each
mole of oil. However, in practice, the methanol/oil molar ratio should be higher than that of
stoichiometric ratio in order to drive the reaction towards completion and produce more methyl
esters as product. As shown in Fig. 9, when the methanol loading amount was increased from 4
to 6, the conversion was increased considerably. The maximum conversion was very close to
12:1. However, with further increase in the molar ratio there was only little improvement in the
conversion. Therefore, we could conclude that to elevate the conversion an excess methanol feed
was effective to a certain extent.
The effect of the residence time on the conversion of soybean oil is illustrated in Fig. 10. The
conversion was increased with the increase of the reaction time from 14 to 28 seconds, and
thereafter remained nearly constant as a representative of a nearly equilibrium conversion. The
effect of the residence time on the mean droplet size is shown in Fig.11 (a). It can be seen that
the mean droplet size increases with the increase of the residence time from 14 s to 31 seconds
and decreases thereafter. As shown in Fig.11 (b), an increase of the residence time has little
influence on the distribution of the droplet size. On the one hand, the increase of residence time is
favorable for the mixing of the immiscible flows and the conversion of soybean oil due to the
increase of contact time. On the other hand, the secondary flow in the zigzag microchannel is
weakened by the decrease of the flow rate, which is unfavorable for the mixing and formation of
small droplet (Virginie et al., 2002). Thus, the suitable reaction time for the synthesis of biodiesel
shall be considered to be around 28 seconds.
10 12 14 16 18 20
Molar ratio of methanol/oil
Fig.9. Influence of methanol/oil molar ratio on the conversion.
Reaction conditions: catalyst amount 1.2%, residence time 28 seconds and temperature 56 °C.
Residence time,s
Fig.10. Influence of residence time on the conversion.
Reaction conditions: Methanol/oil molar ratio 9:1, catalyst amount 1.2% and temperature 56 °C.
Number distribution,%
Mean Droplet size, um
Residence time, s
Droplet size, um
Fig.11. Differential droplet distribution and mean droplet diameter of various residence times.
(a) Mean droplet size of different residence times. (b) Distribution of droplet size for different
residence time.
Experimental studies for the improvement of biodiesel by using micro-channel reactor were
conducted. A zigzag micro-channel reactor was developed by electric spark and diffusion
bonding for the transesterification of soybean oil with methanol in the presence of concentrated
sodium hydroxide as a catalyst. Compared with the contact time for complete conversion from
several minutes to several hours in conventional reactors, the corresponding contact time by the
micro-channel reactor is reduced sharply to 28 seconds at the following conditions: oil/methanol
molar ratio of 1: 9, temperature of 56 °C , catalyst amount 1.2% (wt, %). This might be resulted
from the intensification of the mixing of the immiscible reactants in transesterification reaction
by the zigzag micro-channel, which is characterized by a microscope method. It is shown that
intensification of the chaotic mixing on the micro-scale is a desirable way to promote the
efficiency of biodiesel synthesis process.
Even though the total throughput of biodiesel by the zigzag micro-channel reactor is small, the
important effect of intensification by the micro-channel is crucial to the high efficiency for the
production of biodiesel. Further work will include, for example, designing new type micromixers to enlarge the total volume of biodiesel production, and studying new solid catalysts so as
to avert strenuous after-treatment process.
This study has been financially supported by the Swedish Research Links Programme and
the China Natural Science Foundation (contract No. 20606011). Auhtors would like to
acknowledge both organizations.
J. M. Marchetti, V.U. Miguel, A. F. Errazu. 2007. Possible methods for biodiesel production.
Renewable & Sustainable Energy Review, 11: 1300-1311.
S. K. Karmee , A. Chadha,. 2005. Preparation of biodiesel from crude oil of Pongamia pinnata.
Bioresource Technology 96: 1425-1429.
G. Vicente, M. MartÍnez, J. Aracil. 2004. Integrated biodiesel production: a comparison of
different homogeneous catalysts systems. Bioresource technology 92:297-305
C. Stavarache, M. Vinatru, Y. Maeda, H. Bandow. 2007. Ultrasonically driven continuous
process for vegetable oil transesterification. Ultrasonics Sonochemistry. 14:413-417.
J. M. N. van Kasteren, A. P. Nisworo. 2007. A process model to estimate the cost of industrial
scale biodiesel production from waste cooking oil by supercritical transesterification. Resources
Conservation & Recycling.50:442-458.
W. Xie, H. Peng, L. Chen,.2006. Calcined Mg-Al hydrotalcites as solid base catalysts for
methanolysis of soybean oil .Journal of molecular catalysis A: Chemical. 246:24-32.
J. Kobayashi, Y.Mori, S. Kobayashi. 2006. Multiphase Organic Synthesis in Microchannel
Reactors. Chemistry-An Asian Journal 1-2:22-35
T. Lemenand, D. Della Valle, Y. Zellouf, H. 2003. Peerhossaini. Droplets formation in turbulent
mixing of two immiscible fluids in a new type of static mixer .International journal of
multiphase flow 29:813-840.
B. P. Mason, K.E. Price, J. L. Steinbacher, A. R. Bogdan, T. McQuade, 2007. Greener
approaches to organic synthesis using microreactor technology. Chemical reviews. 107: 23002318.
N.Shinji 1975. Mixing: principles and applications 1st edn. Wiley, New York.
P.Wu, Y.Yang, J. A. Colucci, E. A. Grulke, 2007. Effect of Ultrasonication on Droplet Size in
Biodiesel Mixtures
Journal of the American Oil Chemists' Society.84:877-884
C. E. Banks, O. V. Klymenko, R. G. Compton, 2003. Liquid-liquid processes and kinetics in
acoustically emulsified media.
Physical Chemistry Chemical Physics. 5:1652-1656.
M. M. Alvarez, J. M. Zalc, T. Shinbrot, P. E. Arratia, F. J. Muzzio. 2002. Mechanisms of mixing
and creation of structure in laminar stirred tanks. AIChE Journal. 48:2135-2148.
V. Mengeaud, J.J.Hubert, H. Girault, 2002. Mixing processes in a zigzag microchannel: finite
element simulations and optical study. Analytical Chemistry 74:4279-4286.
P. K.Meka, V. Tripathi, R. P. Singh, 2007. Synthesis of biodiesel fuel from safflower oil using
various reaction parameters. Journal of Oleo Science.56 (1):9-12.
3. Continuous bioethanol production by fermentation
Bernt Lie_, Juan I. Videla
Telemark University College, Porsgrunn, Norway
February 15, 2008
Alcoholic fermentation is an important biochemical process which has been known for some
5000 years, e.g. in beer production. The process is relevant for producing bioethanol fuel, thus
avoiding the introduction of fossil carbon dioxide into the atmosphere. The aim of this paper is to
give a simple introduction to the fermentation process and to dynamic models of such systems in
order to familiarize control engineers with the topic. Based on equations for a fermentation
reactor in the literature, a rational model description is given, with parameter values and
operating conditions. The model description includes extensions of the literature model.
In this paper, several alternative reaction rates are discussed, and a more general model of
oxygen uptake is introduced, e.g. by including the e¤ect of the substrate.
The various models are compared by simulation, and a simple analysis of controllability is suggested to illustrate possible use of the model. The presented model is complete in the sense that
all necessary information is given to implement the model in a simulation tool. Several
applications of the model are discussed, and various model developments are indicated.
Swedish National Energy Convention, March 12-13, 2008
4. Thermal Optimization and Economic Analysis of a Marnoch Heat Engine
C.C. Chukwu1, G.F. Naterer1, M.A. Rosen1, E. Dahlquist2, I.A. Marnoch3
Faculty of Engineering and Applied Science, University of Ontario Institute of Technology,
2000 Simcoe Street North, Oshawa, Ontario, L1H 7K4, Canada
Faculty of Engineering and Applied Science, Malardalen University, Vasteras, Sweden
Marnoch Thermal Power Inc., 420 Portage Rd., Port Severn, Ontario, L0K 1S0, Canada.
The Marnoch Thermal Energy Conversion (MTEC) device is a novel heat engine that
recovers heat to produce electricity, using lower temperature differences than normally utilized
in conventional heat recovery systems. A laboratory-scale proof-of-concept device has been built
and tested at the University of Ontario Institute of Technology, Canada. It produces electricity by
using heat exchangers and a piston assembly at temperatures and gauge pressures as low as 65oC
and 1.03 MPa (150 psi), respectively. A performance assessment and thermodynamic
optimization of the MTEC system are reported in this article. A thermodynamic model and
control system developed for improved operation of the system are described. Results are
provided of a numerical optimization of key design variables with appropriate constraints, and an
economic analysis for different configurations is presented which demonstrates the practical
viability of the system.
Keywords: Heat Engine, Optimization, Economic Analysis, Conservation, Efficiency
- Area (ft2)
- Specific heat capacity (Btu/lb oF)
- Inner diameter (ft)
- Thickness (ft)
- Friction factor
- Fouling factor
- Thermal conductivity (Btu/hr ft oF)
- Irreversibility (Btu)
- Number of variables
- Length (ft)
- Mass flow rate (gpm)
- Pressure (psi)
- Heat (Btu)
- Heat transfer rate (Btu/hr)
- Temperature (oF)
- Internal energy (Btu)
- Overall heat transfer coefficient (Btu/hr ft2 oF)
- Mean velocity of fluid (ft/s)
- Density (lb/ft3)
- Environment
- Tube
- Shell
- Air
- Volumetric
- Boundary
1 Introduction
Although fossil fuels will continue to remain prominent in the world energy mix for
many years, renewable energy sources are almost certain to become increasingly important to
reduce greenhouse emissions. Correspondingly, alternative technologies that facilitate the use of
renewable energy sources are expected to be increasingly developed.
The paper examines a promising alternative energy technology, the Marnoch Thermal
Energy Conversion (MTEC) system, which can operate using renewable energy as well as other
sources of thermal energy. The MTEC device is intended to produce electricity from a small
temperature difference, as little as 15˚C. Operation over such small temperature differences
allows for a highly flexible energy generation system, which can operate using a variety of heat
sources and cooling reservoirs. Potential sources include waste heat from industrial facilities and
co-generation applications, including those with steam cycle-based power generation. Solar
thermal systems can also be used to provide heat in commercial applications of this device.
Jeong et al. [1] have outlined the methods with which systems like the one considered here can
be applied in waste heat recovery.
The MTEC device is a heat engine that consists of four cylindrical tanks, each fitted with
helical copper tube coil heat exchangers. Each tank is connected to both a hot source and cold
sink, via tubes and valves that allow a controlled mass flow at any step of the process. Initially,
all tanks are kept at a constant pressure and mass. One-half of the Marnoch cycle is driven by
heat supplied to the tanks, which drives pistons external to the tanks. The other half of the cycle
connects to a colder tank, facilitating thermal expansion in the hot tank and thermal contraction
in the cold tank, via a piston assembly.
The MTEC system utilizes an innovative sequence of steps that together achieve a
desired electrical output. The operation of the MTEC device involves three steps: 1) conversion
of internal energy to a pressure difference between two pressure vessels; 2) conversion of flow
work into mechanical energy by means of a specialized piston assembly; and 3) conversion of
mechanical energy to electrical energy by means of a generator.
The amount of electricity generated from a MTEC system strongly depends on the size
and design of the heat exchangers, the pressure difference and the overall temperature difference.
The working fluids for this device are water and dry air. The cylinders are arranged such that at
any given time there is at least a pair each of a hot source and cold sink. A photograph of the
MTEC system and the arrangements of the heat exchangers are shown in Figs.1 and 2,
Figure 1: Photograph of the MTEC device
MTEC system
Figure 2: Schematic of
2 Thermodynamic Model of the MTEC System
Modeling of the MTEC device and control system has been reported by Armstrong et al.
[2]. The first step of the device operation consists of developing a pressure difference between
two pressure vessels, through a temperature difference. When device operation commences,
these vessels have the same initial condition. Heat exchangers placed within the vessels transfer
heat from an external source to their interiors.
A general configuration of the heat exchanger and pressure vessel is illustrated in Fig. 3
Depending on whether individual tanks are used for heating or cooling, a thermal energy source
or sink is connected accordingly.
Figure 3: Schematic of the heat exchanger and pressure vessel
By increasing or decreasing the temperature within the pressure vessel, the pressure
increases or decreases for each case, respectively. Once a pressure difference has been generated
between a pair of vessels, a specialized piston assembly is actuated to convert the pressure
difference to mechanical energy. The piston assembly consists of two chambers separated by a
piston. The piston moves back and forth, thereby converting the energy in the transferred mass to
boundary work. As the first chamber’s volume increases, the second chamber’s volume
decreases. A valve system is configured so that the chamber with less volume is connected to the
vessel with a higher pressure, while the lower pressure vessel is connected to the chamber with
the larger volume, as shown in Fig. 4.
The pressure difference between the two chambers results in a net force on the piston,
moving it to the right in Fig. 4. Once the piston has reached the end of the cylinder (the right side
in Fig. 4), a valve system is activated, which causes chamber 2 to be connected to the highpressure vessel and chamber 1 to the low-pressure vessel. This yields a net mass flow from the
high-pressure vessel to the low-pressure vessel, as the gas in the chamber with decreasing
volume is transferred to the low-pressure vessel. The second half of the piston’s movement,
which returns the piston to its starting position, is shown in Fig. 5. The cycle continues as long as
there is a pressure difference sufficient to move the piston.
Figure 4: Figure 4: Schematic of the piston
Figure 5: Schematic of a piston reaching
the right end of the cylinder
Through the operations of the piston, the internal energy of the working fluid piston
is directly converted to mechanical energy which, via a flywheel connected to a generator, is
converted to electrical energy.
Two thermodynamic models of the system are developed to predict the system
performance. These models are intended to simulate the conversion of internal energy to flow
work and mechanical energy of piston movement.
Performing an energy balance for the heat exchanger,
Q = U 2 − U1 = m ⋅ cv ΔT
Here U is the internal energy, m is the mass of air transferred, cv is the volumetric specific heat
capacity and ΔT is the temperature difference. Alternatively, the heat transfer rate can in a given
time interval can be expressed as
∫ Q(t )dt = mcv ΔT
By substituting the known boundary values at the initial condition, the following temperature
expression can be derived:
1 2.
Tgas(t) =
Q(t )dt + Tgas(0)
mcv t∫1
The heat transfer rate Q can be expressed as follows:
Q =U ' AΔT
where the overall heat transfer coefficient U’ is determined by means of a generalized thermal
resistance network, i.e.
U '=
+ f water + f gas +
hc hwater
Combining the above equations, an expression for T2 with respect to time is obtained:
U' A 2
Tgas (t ) =
(Twater − Tgas (t )) dt + T1 (0)
mcv ∫t1
This function is integrated numerically. The model is applicable to both the heat exchanger that
supplies heat to the tank, as well as the heat exchanger that rejects heat from the tank. The value
of Q is negative for the side that is rejecting heat. In this manner, the instantaneous gas
temperature and pressure within the pressure vessel may be determined. This model assumes a
closed system, with no mass entering or leaving the pressure vessel.
The model is further extended to predict the power output and number of strokes, based
on the initial conditions within the pressure vessels. This model is based on the principle of
conservation of mass and Bernoulli’s equation. For a fixed time interval, it is assumed that each
side of the system is at constant temperature and no frictional losses occur during pressure
transfer. By utilizing the Bernoulli equation (3), it can be shown that
(1− k )
= P1
CA ( k − 1) g c Rk 3 ⎜ T0
+ (t 2 − t1 ).
⎜ (1− k )
⎜ p 2k
⎝ 0
(1− k )
= C ⋅ Apipe
( k +1)
⎟ . ⎡ 2 ⎤ ( k −1)
⎟ ⎢⎣ k + 1 ⎥⎦
⎛ 2⎞
⎛ k +1 ⎞
⎜ ⎟
p2 ⎝ k ⎠ ⎟
2 ⋅ p1 ⋅ ρ1 ⋅ ⎜ ⎜ ⎟ − ⎜ ⎟
⎝ p1 ⎠
⎜ ⎝ p1 ⎠
Using this equation, it is possible to determine the mass flow rate from each tank into the
Based on the mass flow rates, numerical integration leads to the predicted change of mass
with time in each tank as follows:
m1 (t ) = m1 ( 0) − ∫
Rearranging (9) gives
m 2 (t ) = m 2 ( 0 ) +
dm 2
These equations are based on the pistons and pressure vessels being configured as shown in Fig.
6. Based on the mass in each of the piston chambers, the pressures at points 3 and 4 are
calculated as a function of time. With this information, the boundary work produced by the
system during a given power stroke is determined as follows:
W b = ∫ p 3 dV − ∫ p 4 dV
These calculations are repeated for several strokes, until the difference between p1 and p2
becomes negligible. The resulting boundary work terms of all strokes are added to obtain a total
value, and the result is divided by the time required for the strokes to occur to determine the
average power output of the system.
Figure 6: Sequence of steps in the piston motion
3 Results and Optimization
Using the heat transfer model presented in the previous section, results are generated for
a variety of heat source and sink temperatures. In this section, results are presented for initial
pressures of 200 psi and 2000 psi, and simulations over 60 seconds using 0.1 second time
intervals. The conditions are intended to replicate actual working conditions of the prototype at
UOIT, as well as planned future extensions to higher pressure units. In modeling the MTEC
device, heat transfer within the pressure vessels and pressure changes in the piston-cylinder
device are included.
The characteristics of the system at a given initial pressure and varying heat and
source temperatures are presented in Figs. 7-10 for a pair of heated and cooled cylinders over a
period of 60 seconds. The source temperature ranges from 113oF to 194oF, and the sink
temperature from 4oF to 50oF. The pressures at varying source temperatures are shown in Figs.
113 F
140 F
167 F
194 F
50 F
32 F
14 F
Pressure (psi)
Pressure (psi)
Figure 7: Pressure variation with time for
several source temperatures for the
heating tank (69.8˚F and 200 psi gauge
pressure initially in the tank)
Figure 8: Pressure variation with time for
several source temperatures for the cooling
tank (69.8˚F and 200 psi gauge pressure
initially in the tank)
Pressure (psi)
P res s ure (ps i)
50 F
32 F
14 F
113 F
140 F
167 F
194 F
Figure 9: Pressure variation with time for
several source temperatures for the
heating tank (69.8˚F and 2000 psi gauge
pressure initially in the tank)
Figure 10: Pressure variation with time
for several source temperatures for the
cooling tank (69.8˚F and 2000 psi
gauge pressure initially in the tank)
The pressure model presented earlier is used to predict the power output of the device.
Results are obtained on a per-stroke basis and simulations are performed for two initial
pressures, varying the working temperature conditions. Results are presented in Figs. 11 and 12
Energy Output (Btu)
Energy Output (Btu)
that depict the performance of the heat exchangers in terms of energy output after 60 seconds
with initial cylinder pressures of 200 psi and 2000 psi, respectively.
Pressure Change in Cylinders (psi)
Figure 11: Variation of energy output
with pressure difference for a tank
initial pressure of 200 psi 103
Pressure Change in Cylinders (psi)
Figure 12: Variation of energy output
with pressure difference for a tank
initial pressure of 2000 psi Irreversibilities in the system are also assessed. The irreversibilities in the system follow
closely those outlined by Cornelissen and Hirs [3]. The total irreversibility of the system is the
sum of thermal and frictional irreversibilities:
I = I ΔT + I ΔP
The thermal component of the irreversibility is given by
T ⎤
I ΔT = T0 ⎢ m1 c p1 ln 1out + m2 c p 2 ln 2 out ⎥
T2in ⎦
and the frictional component by
⎡ .
I = ⎢ ( p1in − p1out ) + 2 ( p2in − p2 out ) ⎥
⎢ ρ1
Note that
mcp ( T1out − T1in ) = hAΔT
T2out = T1in + ΔT , T2in = T1out + ΔT
Substituting eqs. (15) and (16) into (13) yields
⎜ T + 2 c p 2 (T1out − T1in ) ⎟ ⎥
⎢ .
⎜ 1in
h1 A1
= T0 ⎢ m 1 c p1 ln 1out + m 2 c p 2 ln ⎜
m 2 c p 2 (T1out − T1in ) ⎟ ⎥
⎜ T1out +
h1 A1
⎠ ⎦⎥
where A1 = π D 1 L1 and h 1 is obtained according to Kern [4].
For the frictional component,
= 1 1 Δp1 + 2 2 Δp2
T2 out
where Δp1 and Δp2 are expressible as
Δp1 = p1in − p1out = 2 f1ρυ12
Δp2 = p2in − p2out = 2 f 2 ρυ22
L ( D2 + ( D1 + 2d1 ) )
( D − ( D + 2d ) )
The mean velocity of the gas in the cylinder is assumed zero and the mean velocity of the water
in the tube can be written as
υ2 =
ρπ ⎡ D22 − ( D1 + 2 d 1 ) ⎤
I ΔP =
3 2 f 2 m 23 L 2 ( D 2 + ( D 1 + 2 d 1 ) )
π 2 ρ 22
− ( D1 + 2 d 1 )
where the friction factor along the tubes f 2 is obtained from the friction law of Blasius [5], given
⎛ μ2 ⎞
f 2 = 0.0791 ⎜
⎝ ρν 2 D2 ⎠
The bounded constraints for the size of the heat exchangers and the flow rate are set to
minimize the pressure drops. Having expressed L1 and D1 in terms of L2 and D2 respectively, the
constraints are
737.23 ≤ L1 ≤ 1105.84
0.083 ≤ D1 ≤ 0.117
m1 = 20000
Design Expert, a design of experiment (DOE) software package is used here, specifically
for its optimization capability in predicting the optimal variable combination for the MTEC
device. The Responsive Surface Method (RSM) is used, where experimental and predicted
values are combined to determine the optimal process design. Three optimization variables:
length of the tube (L), Diameter of the tube (D) and mass flow rate in the tube ( m ), labeled A, B
and C respectively are considered. Weights are also assigned to these variables since our
problem is multi-objective. Constraints are set in terms of limits and target and are given in
Table 1. These limits are based on the results from the models in appendix 1. Neglecting the gas
side of tanks, the pressure drop in the tube is minimized so as to reduce the irreversibility due to
frictional components. Different transformations are tried to find the best curve fit for this
variables. For k number of variables, the logarithmic transformation gives the best curve fit and
is related to the pressure drop according to equation (27).
Δp / = log10 (ΔP + k )
737.23 1106
Mass flow
25000 1
Upper Lower Upper
weight weight
Table 1: Constraints and weighting factors
used for the optimization problem
The objective is to find the combination of the three variables that achieves a pressure
drop of 5 psi or less within the constraints given in Table 1. The solution of the optimization
after ten runs using Design Expert is shown in Table 2. The plots of the relationships when
different variables are combined are shown in Figs. 13 and 14. The desirability is an indication of
how good our result is compared to the target, with a desirability of 1 being the best. The
prediction indicates a desirability when 2 or 3 variables are combined in a run. The color code
indicates the trend from low to high with red being the highest. The individual and combined
desirability of the varaibles are shown in Figs. 15 and 16.
L (in)
D (in)
m (gpm)
ΔP (psi)
Table 2: Results of the optimization after 10 runs
C: m 1.22381
B: D
Figure 13: Combination of flow rate and tube diameter
C: m 17500.00
A: L
Figure 14: Combination of flow rate and tube length
Figure 15: Desirability of individual and combined variables
Figure 16: Desirability of individual and combined variables with constraints
4 Economic Analysis of MTEC Device
The economic viability of the MTEC device is examined based on the amount of
electricity that can be effectively harnessed from the device and the cost of producing this
energy, given a certain configuration. The relationship between cost and anticipated revenue is
examined, and a payback analysis is carried out to help determine the project viability.
The model provided in Appendix 1 with a configuration of 8 tanks of 432 US gallons
each is used. The initial pressure of air in the tanks is 5000 psi and the heat transfer rate is 556
KW/h. The Carnot efficiency for this configuration is 23% and the system efficiency for
mechanical components is assumed to be 50%. The projected output and annual revenue for
continuous operation using steady solar heat source and waste heat source assuming that
electricity is sold at $110 per MWh are given in Table 3.
Tank configuration/hour (3456 gallon)
257.245 kWh
10 hours solar source
10 hours peak solar and 14 hours low intensity at 50%
4373.27 kWh
24 hrs steady output
6173.89 kWh
Annual revenue using direct solar heating only
Annual revenue using steady waste heat source
Annual revenue using waste heat + solar sources
Table 3: Projected output and annual income for a given configuration using 8 cylinders
Using the cost tables for the components in Appendix 2 and assuming 2% and 3%, for
solar and waste heat maintenance costs, respectively, the payback periods are determined and
shown in Table 4. The revenue discount rate charts are shown in Figs. 17 and 18.
Capital expenditure ($) 886,975 563,975 698,558
Annual revenue ($)
103,284 247,881 175,582
Payback period
Table 4: Cost, revenue and payback periods for the METC system
Interest Rate (%)
0.1 6
0.1 8
0.1 2
0.1 4
Revenue ($)
0.0 8
Capital cost
Revenue ($)
0.0 4
0.0 6
capital cost
0.0 2
C ap ital C o st/R e ven u e ($)
Capital Cos t/Rev enue ($)
Interest rate (%)
Figure 17: Revenue vs. discount rate
Figure 18: Revenue vs. discount rate
using waste heat source
using solar heat source
Table 4 shows that the MTEC device will pay back the capital investment in less than 3
years if the source of heat is derived from waste heat of a process industry and in 9 years if solar
panels are deployed for the of heat. When both heat sources are efficiently combined, the
payback period on the investment would be 4 years. Figs. 17 and 18 show the profitability of the
project with various interest rates based on yearly returns. Fig. 17 shows that the MTEC device is
profitable even beyond a 20% interest rate on borrowed capital. This is a case where the heat
source is obtained from waste heat. Fig. 18 is profitable up to a 10% interest rate. This occurs
when the heat is from a solar source. These calculations do not consider subsidies from the
government which would make the system more profitable
5 Conclusions
The novel MTEC technology is demonstrated to be a promising system, both technically
and economically. Using waste heat from nuclear plants or other sources, the system can
continuously generate electricity and pay back the initial capital investment in a few years. By
not using fossil fuels, this device is capable of reducing the greenhouse gas emissions associated
with conventional fossil fuel energy systems. Much work is ongoing to improve the transmission
system and scale up the system. Future thermal optimization processes are expected to include
the gas side of the heat exchangers. Physical measurements will also be done to validate the
results obtained.
[1] Jeong, S., Kang, B.H., Karng, S.W., Dynamic Simulation of an Absorption Heat Pump for
Recovering Low Grade Waste Heat, Applied Thermal Engineering, vol. 18, nos. 1-2, pp. 1-12,
[2] Armstrong, A., Haseen, F., Marnoch, I., Weston, J., Naterer, G.F., Lu, L., Rosen, M.A.,
Thermodynamic Design and Control of the Marnoch Thermal Energy Conversion Device,
Proceedings of 21st Canadian Congress of Applied Mechanics, Toronto, pp. 284-285, 2007.
[3] Cornelissen, R.L., Hirs, G.G., Thermodynamic Optimization of Heat Exchanger,
International Journal of Heat and Mass Transfer, vol. 42, pp. 951-959, 1999.
[4] Kern, D.Q., Process Heat Transfer, McGraw Hill, pp. 216-217, 1950.
[5] Roger, G.F.C., Mayhew, Y.R., Engineering Thermodynamics, Work and Heat Transfer, 4th
edition, Longman, London, 1992.
Appendix 1: Data used in calculations for heat exchangers
Tank and Coil Data
Tank inner diameter
Tank total length
US gal.
Operating pressure
Design pressure
Operating pressure
Number of coils
Water Flow Data
Thermal conductivity of
ft2 F
Specific heat capacity of
Coil tube outer diameter
Coil tube thickness
Tube inner diameter
Flow cross sectional area
Mass velocity
Reynolds Number (Re)
Number of turns in coil
Coil outer diameter
Coil length
Tube length
Pressure drop
Tube volume
US gal.
Percent of total coil to tank
Total surface area of coil
Tank inside area of coil
Water temperature change
Volume of gas in tank
Heat Balance Data
Mass of gas
Rate of temperature
Rate of pressure change
psi/min 1059.59
Required pressure change
Calculated time
Calculated time
Gas temperature change
Approximate water heat
transfer coefficient
ft2 oF
Water fouling factor
h ft2
Gas fouling factor
h ft2
Wall thermal conductivity
ft oF
Wall resistance
h ft2
Overall heat transfer
coefficient U’
ft2 oF
Heat Transfer Data
Gas thermal conductitity
ft oF
Molecular weight of gas
Specific heat capacity of
Assumed water
Btu/lboF 0.24
Assumed gas temperature
Gas viscosity
ft2 s-1
Gas density
Coefficient of thermal
Kern kf factor
Kern p2 factor
Gas heat transfer
coefficient from
ft2 oF
Heat transfer rate Q
Gas heat transfer
coefficient calculated
ft2 oF
Appendix 2: Material costing for the MTEC device
No. Items
cost ($)
Solar panels for 1 acre (1600 at $300 each)
Pressure vessels (8 x 432 gallons at $30,000
Generator and transformer
Grid connection/interconnection
Foundation/construction work
Piston assemblies
Total capital cost
cost ($)
2 shell and tube type heat exchangers
Pressure vessels (8 x 432 gallons at $30,000
Generator and transformer
Grid connection/interconnection
Foundation/construction work
Piston assemblies
Total capital cost
Publication in the proceedings of the International Scientific Conference at Energitinget, March 1213, 2008 at Alvsjo, Stockholom
Guoqiang Zhang1,2, Lin GAO1, Hongguang JIN1,*
Institute of Engineering Thermophysics, Chinese Academy of Sciences,
Graduate School of Chinese Academy of Sciences
P. O. Box 2706, Beijing 100080, P. R. China
E-mail: [email protected]
A Polygeneration system with production of both chemicals and power is a promising option
for fossil fuel utilization. The configuration of a polygeneration system has a significant effect on its
performance, energy and component utilization character, and flexibility of production capacity
ratio. The polygeneration system of hybrid type, which bears the configuration characters of both
parallel and sequential polygeneration systems, is proposed and its performance feature is
investigated. The relation between the main parameters and performance are described by analytical
equations. For this system, the main parameters affecting the energy utilization level are the ratio of
the exergy consumed in the chemical subsystem to that consumed in the power subsystem, and the
exergy efficiency of the chemical subsystem. The parameters effect on the system is described
qualitatively and quantitatively. Through system simulation the energy utilization level and
effective component conversion character of the system are analyzed. It is identified that the hybrid
polygeneration system has a wide range of energy saving rate, from 2% to 14%, covering the
energy saving range of parallel and sequential type. Compared with the parallel and sequential type,
the hybrid polygeneration system has more freedom on performance distribution with better
performance, that is to say it has an excellent performance at each designed production capacity
Key words: hybrid polygeneration system, configuration character, performance analysis,
flexibility of production distribution
Coal is the main energy resource and will last for a long time in China. But coal is a dirty
energy resource with low quality, resulting in many environmental problems. High efficiency and
clean utilization of coal is attracting more and more attention. - Coal gasification technology is a
promising option for clean coal technology [1-3]. The present chemical production energy systems
show their disadvantages in energy integration for pursuing high production rate and the power
energy systems suffer much exergy destruction in the combustion process. The polygeneration
system can deal with these negative impacts borne in the individual energy systems by integrating
the power generation system and chemical production system, and the environmental problem can
also be mitigated [4]. Some polygeneration systems have been studied mainly from the aspect of
chemicals production [5-7]. The main reason for the high efficiency of the polygeneration system is
system integration under the principle of cascading utilization of chemical and physical energy
[8,9]. The configuration of a polygeneration system has a significant impact on the degree of system
integration. In the parallel polygeneration system, the captive power plant is replaced by a
combined cycle with high efficiency and the heat is integrated between the chemical subsystem and
power subsystem [10]. The energy saving effect of a parallel polygeneration system is less
significant than that of a sequential one [11]. In the sequential polygeneration system, the
component adjusting unit is canceled and partial recycle of the unreacted syngas is applied, which
result in reasonable component conversion and chemical energy conversion. But there is a strong
constraint on the production capacity ratio for the small range of recycle ratio for the proper
component conversion rate. This may has strong impact on the design of a sequential
polygeneration system.
To avoid the disadvantages of the sequential and parallel polygeneration system, in this paper a
novel polygeneration system of hybrid type is proposed. The configuration characters and
performance is analyzed and the internal relation between its configuration character and its
performance is discussed.
Description of the reference system
As shown in Fig. 1 (a), the individual coal gas based power system (IGCC) is based on a
TEXACO O2-blown and slurry-feed gasifier with O2 provided by an air separation unit (ASU). The
hot coal gas from the gasifier firstly is cooled in a waste heat boiler (WHB), and then flows to a
fines removal and low temperature purification unit. The cleaned coal gas fuels the following power
generation subsystem including gas turbine (GT), heat recovery steam generator (HRSG) and steam
turbine (ST).
Exhaust Gas
Fuel Gas Preparation Subsystem
Power Generation Subsystem
Purge Gas
Mixed Gas
Methanol Production Subsystem
Fresh Gas Preparation Subsystem
Fig. 1 Flow diagrams of individual processes
The methanol production (MP) system is show in Fig. 1 (b). The coal is gasified in the gasifer
similar to that used in IGCC, and to fit the methanol synthesis reaction the pressure of the gasifier is
increased to 6.8MPa. The hot gas from the gasifier is quenched from about 1619K to 513K. After
the quenching process, about 56% of the gas from the quenching unit is divided to the shift unit to
adjust the CO/H2 ratio via the water-gas shift reaction. The rest of the stream bypasses the shift unit
and mixes with the shifted gas. After cooled down in a heat recovery unit where the low pressure
steam and medium pressure hot water are produced, the mixed gas flows into the purification unit
where impurities like sulfur compounds, carbon dioxide and etc. are absorbed by MDEA (methyl
dimethyl aether). From the purification process, the clean syngas is sent to the methanol synthesis
(MS) unit equipped with Lurgi technology. The temperature of the synthesis process is kept at about
523K by controlling the evaporation temperature of the cooling water. After cooled down, the
unreacted gas is compressed and recycled to the synthesis unit and the raw methanol is purified in
the flash drum and then the distillation unit. In the distillation unit, there are two columns (L-C and
H-C) to remove the light components containing H2, CO, DME, and aldehyde, and the heavy
components including ethanol, water and higher alcohols.
Description of the hybrid polygeneration system
As shown in figure 2, the flow diagrams of the parallel polygeneration system (a) and
sequential polygeneration system (b) are presented. In the parallel polygeneration system, the
characters of the individual production systems (IGCC system and methanol production system) are
still obvious. The chemical subsystem and power subsystem are rather similar to the individual
methanol production system and IGCC power system respectively, except some material and heat
interactions between the two subsystems are present. For the sequential polygeneration system, the
power subsystem is arranged downstream the chemical subsystem fueled by the unreacted syngas
from the chemical subsystem. Thus the chemical process upstream has a constraint and significant
effect on the power subsystem. It is necessary to stress that the chemical subsystem is reformed by
canceling the shift unit and partially recycling the unreacted syngas. The expander in both 2(a) and
Mixed Gas
Purge Gas
Exhaust Gas
Fig. 2 (a) The flow diagrams of the parallel polygeneration system
Acid Gas
Cycled Gas
Synthesis Unit
Crude Methanol Distillation Output
Uncycled Gas
GT Unit
Material Stream
Heat of Steam
Work Stream
Fig 2 (b) The flow diagrams of the sequential polygeneration system
2 (b) is used to reduce the pressure of the stream for the power subsystem and produce some power.
Syngas Waste Heat Syngas
ΔE desC
ΔE desS
Chemical subsystem
Syngas Preparation Subsystem
Gas Turbine
Steam Turbien
Power to the ASU
Power subsystem
Power to the Rcycle
ΔE desP
Exhausted Gas
Power Output
Material Stream
Power Output
Steam Heat Stream
Power Stream
Fig. 3 The flow sheet of hybrid polygeneration system In figure 3, the flow sheet of the hybrid polygeneration system is presented. Coal is gasified in
the gasifier similar to that of the individual chemical methanol production system. The raw syngas
is first cooled down in the waste heat boiler, then divided into two streams in the R 1 (to stand for it,
r1 is defined as the fraction of the stream flow to the chemical subsystem.), one to the chemical
subsystem the other to the power subsystem after purified by MDEA. In the purification unit, CO2
is not removed from the stream that will go to the power subsystem, while for the stream will to the
chemical subsystem the CO2 is removed. The clean syngas to the chemical subsystem is firstly used
to produce methanol, after cooled down in the heat regenerator the raw methanol and the unreacted
syngas are separated. Part of the unreacted syngas is divided in the R 2 (to stand for it, r2 is defined
as the fraction of the unreacted syngas stream flow recycled to the synthesis unit) to the recycle
compressor (RC) through which the unreacted syngas is recycled to the synthesis unit. Then the raw
methanol is distillated in the distillation unit. The other part of the unreacted syngas is mixed with
the off gas from the distillation unit and then goes to the power subsystem. The syngas from the
purification unit in the syngas preparation subsystem and the unreacted syngas from the chemical
subsystem go the power subsystem to fuel the combined cycle. The power needed in the syngas
preparation subsystem and chemical subsystem is provided by the power subsystem, and the heat
integration among the three subsystems is carried out by water/steam with proper pressure.
It can be seen that both configuration characters of the parallel and the sequential
polygeneration systems can be identified in this novel polygeneration system. Firstly, the chemical
subsystem and the power subsystem are arranged in parallel style, which gives the system more
freedom on the production distribution. Secondly, in the chemical subsystem the component
adjusting unit is canceled and partially recycling unreacted syngas.
Evaluation criterion and basic simulation assumption
The performance of the hybrid polygeneration system is evaluated by the energy saving rate
(ESR), which can be define by
ESR=[(Epi+Eci)-Epg] / (Epi+Eci) ] (1)
Where Epi and Eci represent the total energy input to the electricity generation and the
methanol production system, respectively, and Epg represents the energy input to the
polygeneration system. It must be sure that the polygeneration system and the individual systems
should have the same amount of production when they are compared. Then the ESR stands for the
fuel saving rate of the polygeneration system compared with the individual systems with the same
The hybrid polygeneration system is simulated by means of Aspen Plus 1.11, in which the
mass and energy balances are assured. The pressure drops of all streams flowing throw a heat
recover, a combustor, a gas cleaner and other resistance units are set to 5%. For the chemical
subsystem, the synthetic pressure and temperature are set to 78bar and 255Ԩ respectively. The
simulation is carried out according to the state of art for the present technology, and the main
assumptions during the simulation are listed in table 1.
Energy utilization and component utilization characters
3.1.1 Qualitative evaluation of chemical
Table 1 Basic Assumptions for the simulation
components conversion
Assumption for Gas Turbine
From the configuration character
analysis, it can be seen that the hybrid Parameter
polygeneration system has both of the
characters of the parallel and sequential Turbine inlet temperature (C)
polygeneration systems. Thus from the point
Pressure ratio
of view of energy, the hybrid polygeneration
system bears the energy utilization and Mechanical efficiency, Turbine and
component conversion characters of the other Compressor (%)
two polygeneration systems. Compared with
the parallel polygeneration system, the hybrid Isentropic efficiency, Turbine (%)
polygeneraton system cancels the component Isentropic efficiency, compressor (%)
adjusting. Consequently, there is no need for
Assumption for Steam Cycle
quenching the raw coal gas from the coal
gasifier, and the heat of the raw coal gas can
Turbine inlet temperature (C)
be used to generate high-pressure steam. The
low-heat-value of the syngas will not decrease Turbine inlet pressure (bar)
without the exothermic shift unit, which does
a great mater to the power subsystem. Mechanical efficiency (%)
Partially recycling the unreacted syngas
enables the effective components (CO/H2) to be converted sufficiently and properly to methanol,
and the rest of the components that are not suitable for methanol synthesis will be burned as fuel in
the power subsystem. Thus the chemical energy of effective components realizes their cascading
Compared with sequential polygeneration system, the hybrid type has a relative excellent
performance of the chemical subsystem, but not all of the components from the syngas preparation
subsystem go to the chemical subsystem. Thus the first stream division r1 has a negative effect on
the performance. However, when the production capacity requires more power production, the
hybrid polygeneration system, because of its flexible configuration, can perform better than the
sequential one. For the power production requirement, the sequential polygeneration has to break
the principle of partial recycling of the unreacted syngas to choose a not sufficient recycling. The
chemical subsystem of the hybrid polygeneration system manages the proper utilization of the
effective components chemical energy at each of the first split ratio. That is to say for the same
production capacity ratio the hybrid polygeneration has a relatively better performance than the
sequential type.
3.1.2 Analytical analysis
The hybrid polygeneration system is characterized by the two divisions, and to investigate how
the two stream divisions effect on the system performance, the role of the main parameters are
extracted and analyzed. The system is composed of three subsystem, syngas preparation subsystem,
chemical subsystem and power subsystem. Figure 4 is the summarizing exergy balance sketch,
which is detailed in figure 3. In the Figure 4, the E f , EM , and W represent the total exergy input of
fuel, exergy output of methanol production, and power output, respectively; the EdesS , EdesC , and
EdesP represent the exergy destructions in syngas preparation subsystem, in chemical production
subsystem, and power subsystem, respectively.
Due to the exergy balance we can get:
Chemical Subsystem
E f = EdesS + EdesC + EdesP + EM + W (2)
E f (Caol)
Material and energy
In the hybrid polygeneration system, the exergy
ratio of the chemical and power production to the
Power subsystem
exergy input can be expressed:
EM E f − EdesS EdesC + EM
E f − EdesS EdesC + EM
W E f − EdesS EdesP + W
E f − EdesS EdesP + W
E f − EdesS
Fig. 4 the summarizing exergy balance sketch
; αp =
EdesP + W
EdesC + EM
E + EM
E +W
; β p = desP
And β c = desC
E f − EdesS
E f − EdesS
Then equation (3) and (4) can be expressed respectively by:
= α f ⋅ β p ⋅ α p (5)
Make α f =
= α f ⋅ βc ⋅α c
; αc =
In the equation (5) and (6), α f represents exergy efficiency of the syngas preparation subsystem;
α c represents exergy efficiency of the chemical production subsystem; α p represents exergy
efficiency of the power subsystem; β c represents the proportion of the exergy consumption in
chemical subsystem accounting for the exergy output from the syngas preparation subsystem; β p
represents the proportion of the exergy consumption in power subsystem accounting for the exergy
output from the syngas preparation subsystem. From the equation (2), it’s easy to prove that
βc + β p = 1 .
Based on the equation (5) and (6), the evaluation criterion, energy saving rate expressed by equation
(1), then can be expressed by:
ESR = [( EPi + ECi ) − E pg ] /( EPi + ECi )
(W / η Ep + EM / η Ec ) − E f
= 1−
= 1−
W / η Ep + EM / η Ec
(W / E f ) / η Ep + ( EM / E f ) / η Ec
α p / η Ep + β c • (α c / η Ec − α p / η Ep )
It is clear that the main parameter that effect the
performance of the system are α f exergy
efficiency of the syngas preparation subsystem;
α c exergy efficiency of the chemical production
subsystem; α p exergy efficiency of the power
subsystem; β c the proportion of the exergy
consumption in chemical subsystem accounting
for the exergy output from the syngas
preparation subsystem.
= f (α f , α c , α p , β c )
r 2
Sensitivity analysis of the hybrid Fig. 5 exergy consumption proportion in the chemical subsystem duo
polygeneration system
to the two stream splits
From the Analytical analysis above, the
main parameters that affect the performance are
exergy efficiency of the syngas preparation
subsystem α f , the proportion of the exergy
consumption in the chemical subsystem
accounting for the exergy output from the
syngas preparation subsystem β c , and the exergy
efficiencies of chemical and power subsystem
α c and α p . For this polygeneration system, the
exergy efficiency of the syngas preparation
subsystem α f is almost a constant, since the
two stream divisions have no effect on it. The
exergy efficiency of power subsystem α p varies
Fig. 6 exergy efficiency of the chemical subsystem duo to the two
in a very small range from about 0.55 to 0.56.
stream splits
Thus the parameters that significantly affect the
system’s performance are the proportion of the
exergy consumption in the chemical subsystem accounting for the exergy output from the syngas
preparation subsystem β c and the exergy efficiency of the chemical subsystem α c . The two
divisions both decide the value of β c , and the second division decides the exergy efficiencies of the
chemical subsystem α c .
Figure 5 illustrates the simulation results of the proportion variation of the exergy consumption
in the chemical subsystem accounting for the exergy output from the syngas preparation subsystem
β c due to the two stream divisions. From figure 5 it can be seen that, the β c increases with both the
increase of the two stream divisions r1 and r2, and the tendency of the increase due to the second
stream division r2 seems strengthened. But from the analytical formulae expression, it can be seen
that the increase of the β c benefits the system performance.
Figure 6 illustrates the simulation results of the variation of exergy efficiency of the chemical
subsystem α c due to the two stream divisions. And it can be seen that, the α c decreases with the
increase of the second division r2, and the tendency of the decrease is more significant than the
increase tendency of the β c . The main reason for the decrease of the α c is the sharp increase of the
work consumed by the recycle compressor. The decrease of the α c results in a bad energy utilization
level. On the other hand, the value of the first stream division r1 has almost no effect on the exergy
efficiency of the chemical subsystem α c . It just decides the β c .
Compromising both effects of the
proportion of the exergy consumption in
chemical subsystem accounting for the exergy
output from the syngas preparation subsystem
β c and the exergy efficiency of the chemical
subsystem α c , the performance simulation
result of the system is shown in the figure 7,
the performance of the hybrid polygeneration
system due to different values of the two
stream divisions. The energy saving rate
increases as the first stream division r1
increases, which result in the increase of the
Fig. 7 performance figure of hybrid polygeneration system duo to
different values of the two stream splits
β c . For the second stream division r2, there is
an optimal value corresponding to the highest
energy saving rate, and the optimal value of the second stream division r2 is between 0.8 and 0.9.
The line of the highest energy saving rate along the first stream division r1 represents the
polygeneration systems that apply a proper recycling ratio of the unreacted syngas in their chemical
subsystem. The line of the lowest energy saving rate along the first stream division r1 represents the
polygeneration systems without unreacted syngas recycle. The right line corresponding to the first
stream division r1 =1 represents the sequential polygeneration system, in which all the syngas
passes through the chemical subsystem without bypass. The highest point thus is the sequential
polygeneration system with proper recycle of the unreacted syngas from the synthesis unit.
Character of different design working condition
From the configuration character analysis
The hybrid polygeneration system
of the hybrid polygeneration system, it is
concluded that compared with the other two
polygeneration system, this system deserves a
better performance when the power production
capacity increases. To quantitatively analyze
this character, the performances of the two
The sequential polygeneration system
polygeneration system is compared at different
production capacity ratio. In figure 8, we can
identify that both the hybrid and the sequential
The ratio of chemical production to the power production
performance than the parallel type at each
production capacity ratio. Compared with the
Fig. 8 the performances at different production capacity ratio
sequential polygeneration system, the hybrid
polygeneration system is better, and the
advantage increases when the power production capacity increases. When the capacity ratio of
chemical production to power production increase the critical value about 1.4, the two systems have
the same performance for in that case the hybrid polygeneration system have transferred to the
sequential type in configuration.
This character of more flexible production capacity production with higher performance
provides a flexible option for the designing of a polygeneration system. It may be the first choice
for its flexible production capacity, easier management, and relative higher performance.
A new type of polygeneration system, a hybrid polygeneration system is proposed. The idea of
constructing this system is based on the principle of energy cascading utilization and the
combination of the parallel and sequential polygeneration systems. The configuration character of
the hybrid polygeneration system is investigated, and its effects on system performance are
analyzed. Firstly, the system is evaluated qualitatively from the general principle of cascading
utilization of energy. Then the system is analyzed analytically and the main parameters with relation
to the system performance are extracted. How the main parameters effect the energy utilization and
effective component conversion is detailed. The result is that the hybrid polygeneration system has
an energy saving rate of 2% to 14%, and it has a relatively flexible configuration, which may bring
forward a convenience on designing a polygeneration system. At each capacity ratio the hybrid
polygeneration system performs an excellent efficiency. In all, the hybrid polygeneration system
provides an option of high efficiency utilization of fossil fuel, and gives a convenience for selecting
a reasonable capacity ratio with better performance.
[1] DOE /USA. 2000. Clean Coal Technology Demonstration Program, U.S. Department of
Energy, DOE/EF-0419
[2] Zhen Fan, Archie Robertson, 2006. Co-production of Hydrogen and Electricity Using
Pressurized Circulation Fluidized Bed Gasification Technology, U.S. Department of Energy
Final report, DE-FC26-00NT40972
[3] Dr. Benjamin,C. B. Hsieh, 1996. Overview of Clean Technology in the United States. The
United States of America and the People’s Republic of China Experts Report on Integrated
Gasification Combined-Cycle Technology (IGCC), 1996
[4] Jin H., 1999. Future Research on Exergy and its Application: Simultaneous Resolution of
Energy and Environment. Proceedings of the ECOS99, June 8-10, 1999. Tokyo, Japan.
[5] Jackson, R. G., 1989. Polygeneration System for Power and Methanol Based on Coal
Gasification, Coal Conversion 3: 60-64.
[6] Brown W. R., Moore R. B., 1994. Flexible Electric Power Generation--The integrated
Gasification/Liquid Phase Methanol (LPMEOH Demonstration Project, Third Annual Clean
Coal Technology Conference, September 6-8, 1994. Chicago, Illinois,
[7] Brown W.R., Fenduto F.S., Fuel and Power Coproduction-The Integrated Gasification/LiquidPhase Methanol (LPMEOH Demonstration Project, First Annual Clean Coal Technology
Conference, September 1992. Cleveland, Ohio.
[8] Sun, S., Gao, L., Jin, H., 2007. System Study on Polygeneration System for Methanol and
Power Production Based on Coke Oven Gas and Coal Gas, Proceeding of ECOS2007, 2007.
Padova, Italy.
[9] H. Jin, W. Han, L. Gao, 2007. Multifunctional Energy System (MES) With Multifossil Fuels
and Multi Products, AMSE Trans., Journal of Engineering for Gas Turbines and Power, 129:
[10] Gao, L., Jin, H., and Liu, Z., Exergy Analysis of Coal-Based Poly-Generation System for
Power and Chemical Production, Energy -The International Journal, 2004, 29, pp. 2359-2371
[11] L. Gao. 2004, Investigation of Coal-Based Polygeneartion Systems for Production of Power and
Liquid Fuel. Dr. thesis, Institute of Engineering Thermophysics, Chinese Academy of Sciences.
Publication in the proceedings of the International Scientific Conference at Energitinget, March 1213, 2008 at Alvsjo, Stockholom
6. Power Transformer Parameter Estimation with online data ACQUISITION and USING
the Kalman Filter method
Kourosh Mosavi-Takami
Professor Erik Dahlquist
IST dep.,MDH University, Box 883, SE 721 23, Västerås, Sweden
[email protected]
[email protected]
Power transformer parameters change after many years due to aging, hot spot temperature, short and
long term over load, short circuit, isolations degradation an so off line parameter determination by
manufacturer. For this reason, on line identification is the best option; it can be followed using load
changing that can make the disturbances on voltage, exciting voltage and currents. These disturbances
can help to perform the parameter estimation, convergence in value and computations method. Recent
advances in computational power has allowed the use of online parameter estimation techniques in
varied applications such as reconfigurable or adaptive control, system health monitoring, and fault
tolerant control. The combined problem of state and parameter identification leads to a nonlinear
filtering problem; furthermore, many systems are characterized by nonlinear models as well as noisy
and biased sensor measurements. The Kalman filter (Kalman 1960), which assumes Gaussian
distribution for the uncertainties in system dynamics and utilizes the first two moments of the state
vector (mean and covariance) in its update rule is an optimal sequential linear estimator ideally suited
for recursive implementations Kalman Filter is a commonly used algorithm for recursive parameter
identification due to its excellent filtering properties and is based on a first order approximation of the
system dynamics. Using Kalman filter estimator, estimated the 230/63 KV, 250 MVA transformer
non-dynamic parameters.
A new parameter identification method is proposed in this paper to estimate the parameters of large
power transformers. A sudden change of load is imposed to the transformer when it is under load
operation respectively in no load condition, while its transients of voltages, currents and temperatures
are recorded. For parameter identification and so verification of models authors used MATLAB
software for the coding and computation process. Parameters were identified and then measured and
so predicted outputs were compared.
Keywords: Estimation. Kalman, parameter, power transformer
Two-winding transformers (TWT) are widely used in power system applications. Determination of its
equivalent circuit parameters is useful in performance computations, power system load flow studies,
monitoring of rise in winding temperature and health of transformer, design of relays in protection
circuit of transformer, etc.
Parameter estimation and System Identification techniques allow the engineer to form a mathematical
model of a system using measured data. The power industry has placed great emphasis on system
identification of various equipments since the resulting mathematical models are useful in the system
design and management process, especially for developing elaborate simulation environments and
control systems design.
System identification techniques can roughly be classified in two groups: offline techniques and
online techniques.
Offline system identification techniques tend to depend on iterative methods that exploit the
advantage of having a complete set of data available for processing, whereas online or recursive
system identification techniques must use the data as it becomes available. Recursive system
identification is a valuable tool in the design of adaptive control Laws health monitoring algorithms,
and the design of fault tolerant systems. Increasing availability of onboard computational power
indicates further emergence of applications employing recursive parameter identification algorithms.
Recursive system identification techniques handle loading data as it is measured through onboard
sensors and estimate the required state variables derivates in real-time. Measured data can contain
considerable amount of noise. Furthermore there might be biases and unobserved states in the system
model which must be estimated; hence filtering techniques are generally employed. Fundamental to
all stochastic filtering methods is a two step Bayesian procedure consisting of prediction, or time
update; and correction, or measurement update. The Kalman filter (Kalman 1960), which assumes
Gaussian distribution for the uncertainties in system dynamics and utilizes the first two moments of
the state vector (mean and covariance) in its update rule is an optimal sequential linear estimator
ideally suited for recursive implementations. However, most power systems involve a nonlinearity of
some form, furthermore the method of parameter estimation through state.
The Kalman filter is a suitable algorithm to use for fault detection in large-scale power transformer.
It will be possible to correlate the parameters of the equivalent circuit to a fault type as shown in Table
1. This approach is currently under investigation. In table 1 some examples of faults that can be
detected on-line through parameter estimation are shown.
Table 1 Correlation of model parameter changes to transformer fault type
Capacitance between
Type of fault
- Disc deformation
- Local break down
- Winding short
- Disc movement
- Buckling due to large mechanical forces and moisture ingress
- Shortened or broken disc
- Partial discharge
- Aging of insulation
Development of the Model:
In this section, we will describe an arrangement by which the state equations of a two-winding
transformer can be implemented in a computer simulation. In this work, currents are chosen as state
variables. The equivalent circuit representation of Figure 1 has a cut set of three inductors. Since
their currents obey Kirchhoff’s current law at the common node, all three inductor currents can not
be independent.
V1 is chosen as an input variable, while V2 is and output variable. Core losses are included by
approximating them as losses proportional to the square of the flux density in the core, or the square
of the internal voltage em shown in Figure 1; then an appropriate core loss resistance is connected
across em, in parallel with the magnetizing inductance, Lm. The resultant equivalent circuit would
then be the same as that derived from steady-state considerations. From Figure 1, current of Ll1 is
represented as x1, x2 for L22 and x3 for Lm.
Given Data:
The following data have been extracted from the nameplate and transformer catalogue:
R1 =0.002 pu
Rbase=(230KV2/3*250MVA)=70.5 Ω , R1 =0.002*70.5=0.14 Ω
L1=0.08pu, Lbase=70.5/(2*3.14*50)=0.2245 H, L1=0.08*0.2245=0.018H
L1pu= L2pu, R1pu= R2pu
Magnetization resistance and reactance
Using the resistance and inductance can simulate the core active and reactive losses. When
selected, the p.u. values are based on the nominal power Pn and on V1. For example, to specify 0.2%
of active and reactive core losses, at nominal voltage, use Rm = 500 p.u. and Lm = 500 p.u.
Rm must have a finite value when the inductance of winding 1 is greater that zero.
voltage, V1
voltage, V2
Figure 1. Full equivalent circuit of a two-winding transformer (no loaded), 230/63 KV, 250MVA
The general form of the state equations is defined as:
x& = Ax + Bu
y = Cx + Du
step1: no load and open circuit in secondary side
For no load condition, x2 =I2 =0, so KVL in primary winding of transformer is equal to:
= v1 − x1 R1 − ( x1 − x3 ) R fe
R fe
R fe
x&1 =
v1 −
R1 x1 −
x1 +
x&1 = −(
R fe + R1
R fe
) x1 +
x3 +
i1 = i0 = i fe + im
R fe
R fe
= i fe R fe ⇒ Lm x& 3 = ( x1 − x3 ) R fe ⇒ x& 3 =
x1 −
The state space equation is:
⎡ R fe + R1
⎢( −
⎡ x&1 ⎤ ⎢
⎢& ⎥ = ⎢
⎣ x3 ⎦ ⎢ R fe
⎣ m
R fe ⎤
L1 ⎥ ⎡ x1 ⎤ ⎡⎢ ⎤⎥
⎥ ⎢ ⎥ + ⎢ L1 ⎥ v1
R fe ⎥ ⎣ x3 ⎦
⎢⎣0 ⎥⎦
Lm ⎥⎦
⎡1 ⎤
[B] = ⎢⎢ L1 ⎥⎥ [C ] = ⎛⎜ R
⎜ fe
⎢⎣0 ⎥⎦
v 2 = ( x1 − x3 ) R fe
− R fe ⎟
⎠ and [D ] = [0] .
⎞ ⎡ x1 ⎤
− R fe ⎟ ⎢ ⎥
⎟ ⎣ x3 ⎦
y = ⎜ R fe
Of course:
i1 = i fe + im
i0 = ( x1 − x 2 )
⎡ R fe + R1
⎢( −
[A] = ⎢
⎢ R fe
⎣ m
R fe ⎤
L1 ⎥
R fe ⎥
Lm ⎥⎦
step2: loaded condition
In this step an R load has been embedded in secondary terminals. The load energized by the
transformer, gives I2 a known value. From the load characteristics, the load resistance is a given
voltage, V1
voltage, V2
Figure 2. Full equivalent circuit of a two-winding transformer (loaded), 230/63 KV, 250MVA
( x1 − x 2 ) = i fe + x3
i fe = x1 − x 2 − x3
From the KVL in primary loop and situation I with state variable x, the following equations can be
L1 1 = v1 − x1R1 − i0 R fe
R fe + R1
R fe R fe
x&1 = −(
x3 +
) x1 + x 2
= E1 − x 2 R2 − x 2 Rload ⇒ L2 x& 2 = ( x1 − x 2 − x3 ) R fe − x 2 R2 − x 2 Rload
R2l = R2 + Rload
x& 2 =
R fe
x1 −
R fe + R2l
x2 −
R fe
= R fe ( x1 − x 2 − x3 ) ⇒ Lm x& 3 = R fe x1 − R fe x 2 − R fe x3
R fe
R fe
R fe
x& 3 =
x1 −
x2 −
v 2 = i2 Rload = x 2 Rload
⎡x ⎤
⎤⎢ 1 ⎥
y = ⎢0
0⎥ ⎢ x 2 ⎥
⎥⎦ ⎢ ⎥
⎣ x3 ⎦
⎡ R fe + R1
⎡ x&1 ⎤ ⎢
⎢ & ⎥ ⎢ R fe
⎢ x2 ⎥ = ⎢ L
⎢ x& 3 ⎥ ⎢ 2
⎣ ⎦ ⎢
⎢ fe
⎢ Lm
⎡ R fe + R1
⎢( −
⎢ R fe
⎢ L2
⎢ fe
⎢ Lm
R fe
R fe + R2l
R fe
R fe
R fe + R2l
R fe
R fe ⎤
⎡1 ⎤
L1 ⎥
⎢L ⎥
R fe ⎥ ⎡ x1 ⎤ ⎢ 1 ⎥
⎢ ⎥ + ⎢0 ⎥ v1
L2 ⎥⎥ ⎣ x3 ⎦ ⎢ ⎥
⎢0 ⎥
R fe ⎥
⎢ ⎥
⎣ ⎦
Lm ⎥⎦
R fe ⎤
⎡1 ⎤
L1 ⎥
⎢L ⎥
⎢ 1⎥
R fe ⎥
, B = ⎢0 ⎥
L2 ⎥
⎢ ⎥ [C ] = ⎜ 0
⎢0 ⎥
R fe ⎥
⎣ ⎦
Lm ⎥⎦
[D] = [0]
Figure 3. a) Kalman filter dynamic concept and b) block diagram of kalman filter for state and finally
parameter estimation
Figure 3 represents one cycle of consecutive prediction and filtering updates in KF method.
The Kalman Filter is a linear, discrete time, finite dimensional time-varying system, whose inputs are
the system inputs, uo, u1……. uk−1, and the system measurements, y0, y1, . . . , yk. The output is the
process ˆx(k|k −1) or ˆx(k|k).
Where; ˆx(k|k −1) is the estimate of x(k) given past measurements up to yv(k-1). ˆx(k|k) is the
updated estimate based on the last measurement yv(k).
The time-varying Kalman filter is a generalization of the steady-state filter for time-varying systems
or LTI systems with no stationary noise covariance. Given the plant state and measurement equations:
x[n + 1] = Ax[n] + Bu[n] + Gω [n]
yv [n] = Cx[n] + υ [n]
The time-varying Kalman filter is given by the recursions.
Measurement update:
[ ] [ ]
[ ]
M [n] = P[n n − 1]C T ( R[n n]+ CP[n n − 1].C T ) −1
P[n n] = ( I − M [n]C ) P[n n − 1]
xˆ n n = xˆ n n − 1 + M [n]( y v [n] − Cxˆ n n − 1 )
Time update:
[ ] [ ]
P[n + 1 n] = AR[n n]AT + CQ[n].G T
xˆ n + 1 n = Axˆ n n + Bu[n]
In this simulation, for simplicity we get the white noise in coding then, G=Q=R=1
The presented equations have been coded in Mfile of MATLAB software, using the Kalman estimator
method. The Kalman estimator is the optimal solution to the following continuous or discrete
estimation problems. The Kalman filter is the filter that obtains the minimum mean-square state error
The first plot in figure 4 shows the filtered output (solid line). The second plot compares the
measurement error (dash-dot) with the estimation error (solid). This plot shows that the noise level has
been significantly reduced. This is confirmed by the following error covariance computations.
MeasErr = y-yv;
MeasErrCov = sum(MeasErr.*MeasErr)/length(MeasErr);
EstErr = y-ye;
EstErrCov = sum(EstErr.*EstErr)/length(EstErr);
The error covariance before filtering (measurement error) is MeasErrCov.
MeasErrCov = 1.1138
While the error covariance after filtering (estimation error) is only EstErrCov.
EstErrCov = 0.2722
This value is smaller than the theoretical value errcov and close to the value obtained for the steadystate design.
Time-varying Kalman filter response
x 10
No. of samples
Diffrence of output with mesured and estimated values
No. of samples
Figure 4. Measured and estimated output
Table2: Comparison between given and estimated data using KF method
Parameter name
Given parameter
Fault diagnosis using parameter detection
Parameter detection can be used to diagnose the health of a transformer and foresee any developing in
faults and failures. In online parameter estimation, the parameter is estimated in real time and all of
measured input and output will be considered. While the states are monitored, fault detection can be
performed by use of identified parameters. The parameters in healthy work of the transformer have a
constant value, but in faulty conditions, these parameters change due to internal fault and make a
difference between estimated and given parameters. For example, in internal turn to turn fault, due to
electrically connection of two or more turn to them (for short circuit, deformation or other faults
connect), the resistance and reactance change and deviate from the given value. In on line parameter
estimation, estimator can estimate the parameters in real time and can check with given value, and
then it can analyze the deviation and send an alarm (for small deviation) or trip for large deviations.
Therefore, with comparing the estimated parameters over given parameters, the faults can be detected
and identified. It is obvious that the reactance and resistance are proportional with turns; every change
in turns can make a change in these parameters. Turns contributed in flux making in transformers and
with an increase or decrease in turns, the flux and all of its related parameters will change. The short
circuit or turn to turn faults can make a residual currents or zero sequence values. Some of small turn
to turn faults cannot be diagnosed by residual current relay or differential relays so this method is
effective and applicable.
The schematically fault detection and data acquisition processes are shown in figure5.
Field Bus- Fiber
Control and parameter estimation
Figure 5. Online monitoring, fault diagnosis, control and parameter estimation graphical systems
Using the Simulink models in MATLAB software for input and output generation, we have provided
virtual data for performing the parameter estimation. The white and Gaussian noise was applied to the
model and the IEEE guide line data used for initial conditions. Virtual measured data (from model
running in Simulink) and estimated data had a satisfactory best fit, see table 2. Estimation has been
performed using Kalman filter and estimated parameters were close to the given parameters. The
variation is for noise and so error rate of iterations.
We have shown that the prediction method can be applied as an on-line method for parameter
estimation. With parameter estimation we can do also parameter estimation for use in power flow
calculation in power transmission network with high accuracy. The electrical utilities can monitor and
detect several different types of internal faults in power transformers, for example turn to turn in
winding faults, wax bulk between discs, local break down, insulation aging and others, see table1.
The innovation gain M is chosen to minimize the steady-state covariance of the
estimation error given the noise co variances
process white noise
measurement white noise
known inputs
measurements ouputs
noise covariance data
noise covariance data
MeasErrCov The error covariance before filtering (measurement error)
the error covariance after filtering (estimation error)
state variable
steady-state error covariance
per unit
winding resistances in secondary and primary, respectively
winding inductances in secondary and primary, respectively
core resistance
core inductance
load resistance
mutual inductance in primary side of transformer
mutual inductance in secondary side of transformer
estimated value
Erik Dahlquist, ‘’Process technology and process simulation’’, November 2006, Malardalen
University Press, Västerås, Sweden, ( Book)
Michael Athans, ”Dynamic Stochastic Estimation, Prediction and Smoothing,” Series of Lectures,
Spring 1999.
T. Kailath, Lectures Notes on ‘’Wiener and Kalman Filtering,” Springer-Verlag, 1981.
Thomas Kailath, Ali H. Sayed, Babak Hassibi, ” Linear Estimation,” Prentice Hall, 2000, (for Book)
Peter S. Maybeck, ”The Kalman Filter: An Introduction to Concepts,” in Autonomous Robot
Vehciles, I.J. Cox, G. T.Wilfong (eds), Springer-Verlag, 1990.
J. Mendel, “Lessons in Digital Estimation Theory”, Prentice-Hall, 1987, (for Book)
K. S. Miller, “Multidimensional Gaussian Distributions,” John Wiley & Sons, 1963, (for Book)
J. M. F. Moura, “Linear and Nonlinear Stochastic Filtering,” NATO Advanced Study Institute, Les
Houches, September 1985.
A. Papoulis, “Probability, Random Variables and Stochastic Processes,” McGraw-Hill, 1965.
M. Isabel Ribeiro, “Gaussian Probability Density Functions: Properties and Error Characterization,”
Institute for Systems and Robotics, Technical Report, February 2004.
H. L. Van Trees, “Detection, Estimation and Modulation Theory,” John Wiley & Sons, 1968. [1] D.
Dolinar, J. Pihler, and B. Grcar, “Dynamic model of three phase power transformer,” IEEE Trans.
Power Delivery, vol. 8, Oct. 1993.
P. Bastard, P. Bertrand, and M. Mevnier, “A transformer model for winding faults studies,” IEEE
Trans. Power Delivery, vol. 9, Apr. 1994.
T. Leibfried and K. Feser, “Monitoring of power transformers using the transfer function method,”
IEEE Trans. Power Delivery, vol. 14, Oct. 1999.
D.J. Wilcox, “Time-domain modeling of power transformers using modal analysis,” IEE Proc.
Electric Power Applicat., vol. 144, Mar. 1997.
Maria Isabel Ribeiro, ‘’ Kalman and Extended Kalman Filters:Concept, Derivation and Properties’’,
Lectures Notes, February 2004, Institute for Systems and Robotics Instituto Superior Tcnico,
Lisboan Portugal
Kourosh Mousavi Takami, ‘’on line identification of large scale synchronous generator parameters
using extended kalman filter estimator’’, electric power engineering conference, November 2000
A part of coding is:
n = length(t)
randn('seed',0);Q = 1; R = 1;
w = sqrt(Q)*randn(n,1);
v = sqrt(R)*randn(n,1);
sys = ss(A,B,C,0,-1);
y = lsim(sys,u+w);
% w = process noise
yv = y + v;
% v = measurement noise
P = B*Q*B';
% Initial error covariance
x = zeros(3,1);
% Initial condition on the state
ye = zeros(length(t),1);
ycov = zeros(length(t),1);
for i=1:length(t)
% Measurement update
Mn = P*C'/(C*P*C'+R);
x = x + Mn*(yv(i)-C*x);
P = (eye(3)-Mn*C)*P;
% x[n|n]
% P[n|n]
ye(i) = C*x;
errcov(i) = C*P*C';
% Time update
x = A*x + B*u(i);
% x[n+1|n]
P = A*R*A' + C*Q*G';
% P[n+1|n]
Mn =
P = [ 0.6124 0.1318 -0.4144
0.1318 0.7301 0.3890
-0.4144 0.3890 0.9888 ]
7. An advanced method for Aluminium - iron - copper bonding using explosive welding
Mohammad Tabatabaee
Jafar Mahmoudi
Ph.D. Student in Mälardalen University
Professor in Mälardalen University
TDI institute and Mälaradalen University, Box 885, SE 721 23, Västerås, Sweden
[email protected]
[email protected]
Explosive welding occurs when adjacent surfaces of appropriately positioned metals are properly
thrust together by energy released from an explosive source. Experimental tests have been performed
to explosively welded aluminium to copper, Iron to Iron in flat and non flat surface in three steps. The
tests have been carried out using various stand-off distances and explosive ratios. Various interface
geometries have been obtained from these experiments. In this study, all the experiments carried out
were simulated using the finite element method. The conventional equations were used to describe the
behaviour of explosive. These equations were coded into the COMSOL software. Most aspects of the
explosive welding parameters were properly simulated. The flyer plate and collision velocities
obtained from the analysis were validated by the pin-measurement experiments. The numerical results
showed that very high localized plastic deformation produced at the bond interface.
Keywords: Explosive welding, plate, impact, detonation, velocity, test, FEM
Explosion Welding (EXW) is a solid state process where welding is accomplished by accelerating one
of the components at extremely high velocity through the use of chemical explosives. This process is
most commonly utilized to clad carbon steel plate with a thin layer of corrosion resistant material
(e.g., stainless steel, nickel alloy, titanium, or zirconium).
Unlike other forms of welding such as arc welding, the process for which was developed in the early
1800s, explosion welding was developed rather recently, in the decades after World War II, but its
origins go back to World War I. It was observed that pieces of shrapnel that were sticking to armor
plating were not only embedding themselves, but they were actually being welded to the metal. Since
the extreme heat involved in other forms of welding did not play a role, it was concluded that the
explosive forces that the pieces of shrapnel were being subjected to was the cause. These results were
later duplicated in laboratory tests, and not long afterwards, the process was patented and put to use.
The fearsome destructive power of explosives can be harnessed to provide a unique joining method,
known as Explosive Welding. To form an explosive weld the following conditions need to occur:
• Two surfaces that need to be joined are initially spaced at a small distance (standoff distance).
• An explosive force brings these two surfaces together progressively at a collision front. The collision
front's velocity must be lower than the speed of sound in the materials, so that the shock wave
precedes the bond being formed. If not, the shockwave would interfere with the contacted surfaces
preventing a bond occurring.
• The interfacial pressure at the collision front must exceed the yield strength of the materials, so that
plastic deformation will occur.
A jet of metal is formed just ahead of the collision front, comprising of the two component surfaces,
which is finally ejected from the interface. The surfaces and any surface contaminants are removed in
the jet. Behind the collision front, the now clean surfaces bond, under extreme pressure, in the solid
state. This dynamic welding situation is shown in Fig.1. In cross section, the materials usually bond
together in an undulating wave form and the process can weld a parent plate of thickness 0.025mm to
over 1m (the maximum flyer plate thickness is one third that of the parent plate). Up to 30m 2 can be
welded in one explosion.
Fig.1. Dynamic situation at the collision front showing the jetting mechanism
In figure 1 Prime component is placed either parallel or at an angle to the base, Explosive is
distributed over top surface of prime component, upon detonation, prime component collides with
base component to complete welding.
(a, b)
Figure2: Effect of Velocity on Explosion
Weld Geometry To analyze the process, the hydrodynamic analogy were used by various authors
[1,2,3] due to the creation of the high localized pressure and the material fluid like behaviour at the
collision zone. The process parameters are the impact velocity, the collision point velocity, the
dynamic angle, the stand-off distance, the type of the explosive used and the detonation velocity,
density and size and distributions of the explosive mix. Welding windows were proposed to show the
weld ability ranges of process parameters i.e. impact velocity (or collision point velocity) versus the
dynamic angle for various materials [3,4,5,6]. Nevertheless, the data were obtained by means of large
number of experiments performed. However, the process could be simulated using the finite element
method and most aspects of the welding process could be obtained. Few attempts have been reported
in the literature to simulate the process. Al-Hassani [7] treated the problem as a normal transient
loading of plane stress elements of rectangular shape. In this analysis, kinematically equivalent
concentrated loads at the nodes represented the uniformly distributed explosive load. Explosive
welding process was simulated by Oberg [8] by means of Lagrangian finite difference computer code,
but only produced jetting. The explosive welding process was also modelled by Akihisa9. He only
produced waves but no jetting. In addition, the author assumed that symmetric or asymmetric shear
flow distribution was generated in the flyer and parent plates and the modelling was performed based
on this supposition. No attempts have been made in the literature to analyze and test the explosive
welding of liquid sources and curvy form surfaces. This matter would be presented in this article.
This is a solid state joining process. When an explosive is detonated on the surface of a metal, a high
pressure pulse is generated. This pulse propels the metal at a very high rate of speed. If this piece of
metal collides at an angle with another piece of metal, welding may occur. For welding to occur, a
jetting action is required at the collision interface.
This jet is the product of the surfaces of the two pieces of metals colliding. This cleans the metals and
allows to pure metallic surfaces to join under extremely high pressure. The metals do not commingle,
they are atomically bonded. Due to this fact, any metal may be welded to any metal (i.e.- copper to
steel; titanium to stainless). Typical impact pressures are millions of psi. Fig. 1 shows the explosive
welding process.
The commonly used high explosives are shown in table 1.
Table1: detonation velocity for many kinds of material
Detonation velocity ,
RDX (Cyclotrimethylene trinitramine, C3H6N6O6
PETN (Pentaerythritol tetranitrate, C5H8N12O4)
TNT (Trinitrotoluene, C7H5N3O6)
Tetryl (Trinitrophenyl methylinitramine,
Lead azide (N6Pb)
Ammonium nitrate (NH4NO3)
Applications: Joining of pipes and tubes, Major areas of the use of this method are heat exchanger
tube sheets and pressure vessels, Tube Plugging, Remote joining in hazardous environments, Joining
of dissimilar metals - Aluminium to steel, Titanium alloys to Cr – Ni steel, Cu to stainless steel,
Tungsten to Steel, etc., Attaching cooling fins, Other applications are in chemical process vessels, ship
building industry, cryogenic industry, etc. and …
Advantages of explosive welding are: Can bond many dissimilar, normally unweldable metals,
Minimum fixturing/jigs, Simplicity of the process, extremely large surfaces can be bonded, Wide
range of thicknesses can be explosively clad together, No effect on parent properties and Small
quantity of explosive used.
Limitations of this method are: The metals must have high enough impact resistance, and ductility,
Noise and blast can require operator protection, vacuum chambers, buried in sand/water, The use of
explosives in industrial areas will be restricted by the noise and ground vibrations caused by the
explosion, The geometries welded must be simple – flat, cylindrical, conical.
Experimental tests have been performed to explosively welded aluminium, copper and Iron in three
step. In step one; The flyer plate had a thickness of 0,5 mm. The parent plate made of Iron with
thickness of 3mm. The tests have been carried out using various stand-off distances and explosive
ratios. Various interface geometries have been obtained from these experiments. The explosive
material was positioned above the buffer and flyer plate.
In this study, all the experiments were simulated using the finite element method. The conventional
equations were used to describe the behaviour of explosive. The explosive properties used were
tabulated in Table 2. These equations were coded into the COMSOL software.
Table 2. explosive property
K g/mm2
HK g/mm2
Iron 3 mm
Aluminium 5mm
A geometrical analysis [11] shows that the plate velocity bisects the angle between the initial plate
and the deformed plate orientations. On a per unit of time basis, A goes to B with Vp when point O
goes to B with Vw, see figure 1. Triangle OAB is isosceles; for triangle OBC one has
V p = 2Vd sin( )
When β<10, one can use the approximation Vp≈Vd.sin β and Vp are the most important parameters of
explosive welding. The formation of a jet is a necessary prerequisite for explosive welding.
Wittman [12] and Deribas [13] developed an explosive welding window (Fig. 3), in which the
collision angle β is plotted in the ordinates and the welding velocity, Vw; is plotted in the abscissa.
They studied jet formation, the critical impact pressure, the maximum impact velocity and wavy–
smooth transition velocity. For example, if Vw reaches a supersonic value, there is no jetting. The
same happens if the impact pressure is too low. The wavy–smooth transition parameter does not
appear in this plot and needs to be studied more closely.
A model capable of predicting the wave geometry would be able to determine the smooth–wavy
transition. Deribas [20] developed the first model, based on experimental results. It is a hydrodynamic
model, which describes wave formation as being analogous to fluid flow behind an obstacle. Thus, the
smooth–wavy boundary corresponds to the laminar-turbulent transition. In fluid mechanics the
Reynolds number characterizes this transition. Cowan et al. [15, 16] introduced the following
Reynolds number Rt for explosive welding:
Rt = ( ρ a + ρ b )Vt 2( H + H )
Where ρa and ρb are the densities of the materials and Ha and Hb are their Vickers hardnesses. This
new parameter was an important advance in the domain, but the hydrodynamic analogy contained
limitations reported by Jaramillo and Szecket [17]. In the hydrodynamic model, the waves generated
behind an obstacle reach a stable configuration after a certain distance while the waves at the weld
interface must be created at the collision point. Furthermore, even if the fluid analogy applies to the
collision point, what kind of fluid is assumed and what is the response of the regions removed from
this area? Nevertheless, Cowan et al. [16] obtained correct results with regard to the smooth–wavy
transition. In all of their experiments, Eq. (2) in which Rt is equal to 10.6 describes the transition very
well. The limitation is that they studied only one collision angle, β = 12°.
This relationship is not applicable for a generic collision angle. In order to generalize, it is necessary
to obtain a relationship between the Reynolds number and the collision angle. A model was developed
by Szecket [17–21] for the three following material couples: Fe–Fe, Cu–Cu, Al–Al and Al-Cu (Al
alloy) (Flyer plate thickness: 0,5, 3 and 5 mm).
Fig. 3. Theoretical and practical boundaries for wave formation and jetting [12, 13].
When all the data for these four different systems are combined, the following general relationship
between the Reynolds number and the collision angle is obtained with a correlation factor=0.948:
Rt = K EP = 92.87 ( ±8.92) − 12.87 ( ±2.87 ) β + 0.89( ±0.34) β 2 − 0,032 ( ±0.12) β 3
Eq. (3) provides a very good approximation of the smooth–wavy transition for the systems studied.
This equation is the first step toward a dynamic plasticity model and implies to redefine Rt by the
elastic–plastic constant KEP.
The advances achieved by Deribas [13], Wittman [12], Cowan [15] and Szecket [18] enable the
construction of a plot that Szecket named the ‘‘weld ability window’’, which includes both a straight
and wavy interface domain. This plot is applied to the aluminium alloy used in this investigation. The
appropriate equations for the inclined plate configuration used in this investigation are presented next.
The establishment of a weld ability window requires the relationship between the initial conditions
(the initial angle α and the characteristics of the explosive) and the collision angle β. The geometrical
considerations applied to this configuration give the following equations [4]:
V p = 2 sin(
β −α
)V d
Vw =
2Vd sin( β )
4Vd2 − V p2
Vf =
β +α
sin( β )
Knowing a; Vp and Vd; β can be obtained from Eq. (1). Vp is calculated from the Gurney equation [26,
27] for the specific explosive welding geometry
V p = 2E (
(1 + 2 ) 3 + 1
6(1 + 1 )
+ 1) 3
R =C/M is the mass of explosive for unit mass of flyer plate. 2 E is the Gurney energy, which is
experimentally known for common explosives. The Gurney equation only predicts a terminal
velocity; the problem of the flyer plate acceleration is intentionally left out. Vw and Vf can be
calculated by Eqs. (5) and (6). Known values of α; β; Vd; Vp; Vf ; Vw and the properties of the material
enable the design of the weldability window. This diagram can be drawn in both the Vw vs β and Vp
vs. β plane. Eq. (8) gives the lower limit for welding (due to Deribas et al. [14]). In Eq. (8), β is in
radians, k1 is a constant, H is the Vickers hardness in N/m2, and ρ is the density in kg/m3. The value of
k1 is 0.6 for high-quality
β = K1
ρ .Vw2
Pre-cleaning of surfaces, and 1.2 for imperfectly cleaned surfaces. Eq. (9), due to Deribas [13] and
Wittman [12], gives the upper limit for welding. k3 should be evaluated experimentally at a value of
Vw; which is equal to half of the compressive wave velocity Cf (Vw=2645 m/s for pure aluminium); t
is the thickness of the flyer plate:
sin( ) =
t 0.25V w1.25
K3 =
Cf = K
,K = E
3(1 − 2v)
Eq. (10), due to Szecket [11], gives the smooth– wavy transition zone for the Al alloy. This zone has
been built with experimental results. For Al _ Cu :
Rt = K EP = 13.87 ( ±11.92) − 22.87 ( ±5.07 ) β + 2.89( ±0.14) β 2 − 0,012 ( ±0.02) β 3
Correlation factor = 0.926.
Szecket [12] developed a weld ability zone, which contained left and right boundaries. His results for
Cu will be merged with data for the aluminium alloy used in this investigation. Specific parameters
are: ρ=2700 kg/m3; Hv=62kg/mm2; Cf=6400 m/s.
Fig. 3. Weldability window of the aluminum alloy. Upper and lower limits due to Deribas [13], wavy–
smooth transition due to Szecket [12].
The transition zone is given for the aluminium alloy; nevertheless, the parameters corresponding to
the lower limit for less perfectly cleaned surfaces (Vw=3000–5000 m/s) should enable a welding
without waves. These values correspond to the collision angles β between 4°and 15°. If PETN and a
value of R =1/3 are chosen, the range of the initial angle a will be from 4° to 10°. Fig. 3 shows the
application of Eqs. (1)– (10) to the Vw vs. β space. Two different flyer plate thicknesses t are used:
0.5, 3 and 5 mm. The smooth–wavy interface transition is shown in this plot, and two regimes are
clearly seen.
Flyer plate is copper with 0.5 mm thickness and parent plate with 3 mm thickness is iron.
aa´line: β = 10(Vd − 5.5)
6.5 7
7.5 8
8.5 9
bb´ line : (1.2-1.3)*sound velocity=1.3*4900= 6370 m/s
cc´ line: β=3°
dd´ line: β=31°
Rt=10.6 then Vd= 1310 m/s
ff´line=Vp(min)~180m/s and with
V p = 2Vd sin( )
9.5 10
3.52 1.72 0.86 0.57 0.44
gg´line: with
sin( ) =
t 0.25V w1.25
12,25 7
5,1 4,06 3,4 2,95 2,61
Flyer plate is iron with 3 mm thickness and parent plate with 3 mm thickness is iron.
aa´line: β = 10(Vd − 5.5)
bb´ line : (1.2-1.3)*sound velocity=1.3*6000= 6370 m/s
cc´ line: β=3°
dd´ line: β=31°
Rt=10.6 then Vd= 1460 m/s
ff´line=Vp(min)~271 m/s and with
V p = 2Vd sin( )
3,18 1,55 0,77 0,51 0,4
gg´line: with
sin( ) =
0.25 1.25
Flyer plate is aluminium with 5 mm thickness and parent plate with 10 mm thickness is copper.
aa´line: β = 10(Vd − 5.5)
bb´ line : (1.2-1.3)*sound velocity=1.3*4900= 6370m/s
cc´ line: β=3°
dd´ line: β=31°
Rt=10.6 then Vd= 1466 m/s
ff´line=Vp(min)~150 m/s and with
V p = 2Vd sin( )
2,34 1,14 0,57 0,38 0,29
gg´line: with
sin( ) =
t 0.25V w1.25
The finite difference and finite element communities have used Eulerian methods for over 30 years to
analyze problem with explosive loading, but until comparatively recently, they were too
computationally demanding and inaccurate to be attractive for solving problems in solid mechanics.
Fig. 4 shows a representative 2D configuration used in Raven. The dimensions are in millimeters, the
time is expressed in microseconds, and colors differentiate the materials in this and subsequent
figures: the explosive is yellow; the flyer plate is green; the parent plate is blue.
The computational results are presented as time sequences. Fig. 5 shows the simulation with an initial
angle of approximately 10° at 7, 10, 13, 15 and 20 μs. Calculations were also run for the other
experimental a angles. The sequence of Fig. 5a shows that the angle β is not constant; rather, it
increases with time. Fig. 5b represents a plot of collisions angles, measured from the computed
welding sequences as a function of time. These measurements were made for different values of α: 4°,
6°, 7.8°, 10 ° and 14°. This is an important result, and it is consistent with the metallographic
characterization of the weld morphology reported in figure 6. It should also be noted that the thickness
of the flyer plate is not constant either, contrary to the assumptions made until now.
As the collision angle β increases (and this angle increases with time, as shown in Fig. 5) the wavy–
smooth boundary is traversed for the three cases.
From that point on the welding interface is smooth. Again, this is in full agreement with the
observations made in figure 6. The computations also reveal the effect of a: The smaller a; the more
rapid is the increase of β and Vp: For α =4°, the interface has a very short smooth–wavy transition
part, and the wavelength is not constant. For a=8°, the interface has a large smooth–wavy transition
part, and its wavelength should not be constant. Furthermore, the transition zone is reached earlier
than for α =4°. For α =10°, the wavy part might be divided in two zones: the first one with decreasing
wavelength as in α =4° and 8°, the second one with constant wavelength.
The transition region is larger than for 4° but shorter than for 8°. These predictions from the numerical
results are confirmed by the experimental observations. Note also, in regard to the wave shapes, that
the wavelength depends on the impact velocity Vp whereas the amplitude depends on the initial angle
a: Melting appears when Vp is constant during a few microseconds (see α=10°).
Fig. 4. 2D section used for the computations (dimensions in mm); yellow:explosive; green flyer plate;
light blue: parent plate; red and blue:air. Explosive welding 9.46 degree.
In Fig. 5 c (configurations 8°, 10°, 12° with 25 μs), jetting first appears in the 10° configuration, and it
is also the configuration for which the jet lasts the longest. One concludes from this that the quality of
the welding is better at 10°. The weldability window recommends an initial angle range of 5–10° with
PETN and R =1/3 All the theories suggest that when jetting occurs, it occurs during the entire welding
process. Once again, this is based on a constant collision angle assumption. For Deribas [10], jetting is
responsible for the formation of the waves.
Unfortunately, these diagrams do not give information about the microscopic mechanisms.
As one cannot see the jet along the entire welding process, the conclusion could be that it is only
made with materials located on the last welded millimeter. However, Fig. 5d shows the 10°
configuration with the sacrificial layer. One can observe that this third material is part of the jet. In
fact, with the current simulations, one cannot reach a reliable conclusion about the mechanism behind
the formation of the waves. However, a relationship between the quantity of ejected matter and the
formation of the waves seems reasonable; calculations involving a constant impact angle would help
to establish this relationship. Another possibility is that when the melting temperature is reached along
the interface, the wavy solid phase welding process ends, and a smooth, liquid phase welding process
for 7.68 degree
for 9.46 Degree
for 12 Degree
for 9.46 degee
Fig. 5. (a) Numerical simulation of the explosive welding process: material behaviour in function of
time. Initial angle β~10°. It should be noted that the collision angle is not a constant, but increase with
weld propagation; (b) impact angle β in function of time for different values of initial angle α. (c)
Jetting at 25 μs for three different initial configurations (α~8°, 10°, and 12°). Note minute points
ahead of collapse region. (d) α~10° configuration with an additional layer of Al (red material) on the
flyer plate. One can see that the interlayer material forms the jet.
Figure6: Metallographic tests, this figures describes on acceptable continuity and analogous welding
in joint point
The objective of this study was to establish the conditions for straight, smooth way weld formation
in the explosive welding of Aluminium vs. copper and iron and ... The smooth and straight domains
defined by Szecket [17] were used to successfully predict the two regimes. Szecket’s [17] results for
Al were supplemented by the constitutive response for Cu and yielded a plot applicable to the
experimental results containing both wavy and smooth domains. It was possible to calculate the
relationship between the terminal velocity Vp and the flyer thickness with the Gurney equation. The
agreement between the calculation and the computational simulation proves that the assumption
made on the flyer plate acceleration is reasonable. Experimental observations (by optical
microscopy) on explosively welded specimens suggested that the Vp– β relation was not constant
during the welding process since, in all cases, a region of wavy weld was followed by smooth weld.
It should be noted that the thickness of the materials influences the welding process and
consequently the collision angle β: For the configuration chosen for the capsules in the Mars Return
Mission, the flyer plate thickness is not a constant, and thus, the impact velocity Vp should vary.
Finite element calculations using Raven were conducted in a 2D geometry. The Johnson–Cook
constitutive equation was used with experimentally obtained constitutive parameters for Al obtained
from quasistatic and dynamic experiments carried out over a broad range of temperatures. From a
numerical point of view, the results are particularly convincing. However, it is manageable to reach
a higher level of accuracy utilizing a model that would include higher rate sensitivity. Although the
individual wave formation could not be monitored because of mesh size limitations, the results
demonstrate that the collision angle increases with propagation distance for all initial configurations
analyzed. This change in collision angle is directly responsible for the change in interface
morphology from wavy to smooth at the welding front.
Furthermore, the correlation between the experiments and the simulations demonstrates that the
model is good enough to simulate the process.
The numerical simulation shows the formation of a jet under some initial conditions but does not
reproduce the micromechanics of the process. The wave formation occurs on a micrometer scale
that cannot be captured by the continuum mechanics computation. Nevertheless, jet formation was
observed for 8°, 10°, and 12°.
[1] T.Z. Blazynski, Explosive Welding, Forming and Compaction, Applied Science, London (1983).
[2] G.R. Cowan and A.H. Holtzman, “Flow Configuration in Colliding Plates: Explosive Bonding”,
J. Appl. Phys. 34, 928-939 (1963).
[3] H.K. Wylie, P.E.G. Williams and B. Crossland, “Further Experimental Investigation of
Explosive Welding Parameters”, Proc. 1st Int. Sym. on Use of New Properties and Possibilities of
Application thereof in the Chemical Industry, pp. 45-70, 1970.
[4] A. Szecket, “An Experimental Study of the Explosive Welding Window”, Ph.D. Thesis, (1979).
[5] R.H. Wittman, “The Influence of Collision Parameters on the Strength and Microstructure of an
Explosion Welded Aluminum Alloy”, Proc. 2nd Sym. on Use of Explosive Energy in Manufacturing
Metallic Materials of New Properties and Possibilities of Application thereof in the Chemical
Industry, pp. 153-168, (1973).
[6] A.A. Deribas, V.A. Simonov and I.D. Zakharenko, “Investigation of Explosive Welding
Parameters for Arbitrary Combinations of Metals and Alloys”, Proc. 5th Int. Conf. on High Energy
Rate Fabrication, pp. 4.1.1- 4.1.24, (1975).
[7] Dolgin B, Sanok J, Sevilla D, Bement L. Category V compliant container for Mars Sample
Return Missions. 00ICES-131 2000.
[8] Birkhoff G, MacDougall DP, Pugh EM, Taylor G. Explosives with lined cavities. J Appl Phys
[9] Deribas AA. Explosive welding. Siberian branch of academy of sciences 1967 in explosive
welding. Fiz Goreniya Vzryva 1967;3(4):561–8.
[10] Crossland B. Explosive welding of metals and its applications. Oxford:Oxford Science
Publication; 1982.
[11] Meyers MA. Dynamic behavior of materials. New York:Wiley Interscience; 1994.
[12] Wittman RH. The influence of collision parameters on the strength and microstructure of an
explosion welded aluminum alloy. Proceedings of the Second International Symposium on the Use
of Explosive Energy in Manufacturing, Marianskie Lazni (Czechia), 1973. p. 153–68.
[13] Deribas AA, Simonov VA, Zakcharenko ID. Investigation of the explosive parameters for
arbitrary combinations of metals and alloys. Proceedings of the Fifth International Conference on
High Energy Rate Fabrication. 1975. p. 4.1.1– 24.
[14] Deribas AA, Kudinov VM, Matveenkov FI, Simonov VA. Explosive welding. Fiz Goreniya
Vzryva 1967;3(1): 111–8.
[14] Cowan GR, Holtzman AH. Flow configuration in colliding plates:explosive welding. J Appl
Phys 1963;34:928–39.
[16] Cowan GR, Bergman OR, Holtzman AH. Mechanism of bond zone wave formation in
explosion-clad metals. Met Trans 1971;2:3145– 55.
[17] Jaramillo D, Szecket A, Inal OT. On the transition from a waveless to a wavy interface in
explosive welding. Mater Sci Eng 1987;91:217– 22.
[18] Szecket A. An experimental study of the explosive welding window. PhD, thesis, Queen’s
University of Belfast, 1979.
[19] Szecket A, Mayseless M. The triggering and controlling of stable interfacial conditions in
explosive welding. Mater Sci Eng 1983;57:149–54.
[20] Szecket A, Inal OT, Vigueras DJ, Rocco J. A wavy versus straight interface in the explosive
welding of aluminum to steel. J Vac Sci Technol 1985;3(6):2588–93.
[21] Szecket A, Vigueras DJ, Inal OT. The cyclic pressure distribution of explosively welded
interfaces. In:Murr LE, Staudhammer KP, Meyers MA, editors. Metallurgical applications of shockwave and high-strain-rate phenomena. New York:Dekker; 1996. p. 887–903.
Magdalena Lundh1, Ewa Wäckelgård2, Kajsa Ellegård3
Dep. of Engineering Sciences, Uppsala University
P.O. Box 534, SE-751 21 Uppsala, Sweden
[email protected]
[email protected]
Technology and Social Change, Linköping University
SE-581 83 Linköping, Sweden
[email protected]
Energy demand from hot tap water usage on individual level has been modelled from data in time
use diaries of 179 Swedish households living in detached houses and apartments. Data of activities
that involve hot tap water was extracted and converted into energy – time diagrams (user profiles).
The report describes the procedure to first convert each type of activity to an amount of hot water
and secondly into energy. It was found that the energy use profiles for hot water consumption vary
considerably between individual households. The average user profile of all households was
compared to models of hot water use in two of the more frequently used simulation tools for solar
thermal systems. It is shown that the profiles in the simulation tools are significantly different from
the profiles modelled in this study. In a following study it will be investigated how these differences
influence an optimal use of solar energy in a solar thermal system.
A solar thermal system should, when functioning optimally, load the hot water storage with as much
solar energy from the solar collectors as possible and minimize the use of auxiliary heating. In order
to achieve a maximal solar energy usage the heat storage must be designed to keep temperature
stratification in the tank. It is also important to charge the tank when the solar collectors are
running. This implies the importance of the time for discharging the tank, i.e. when hot water is
used in the household. It has to our knowledge not been studied in real cases how different unload
strategies influence the solar fraction of the system. Jordan et al however performed system
simulations with hot water profiles built on probabilities for different activities and water usage
based on studies from Germany and Switzerland. It was shown that the fractional energy savings
are reduced by more than three percentage units by introducing a probability based user profile
instead of the more simple hot water profiles usually applied in simulation of solar thermal systems.
The reason for reduced fractional savings is foremost the much larger number of draw-offs
compared to the simplified profiles with only three draw-offs during the day. Furthermore, the main
difference to the simplified model is during summer, when the solar energy gain is less with the
realistic hot water profile. The importance of taking holidays with no hot water consumption into
account is stressed. The probability for different activities to take place on different weekdays, on
the other hand, only causes negligible reductions in energy gain. (Jordan et al., 2000) Creating
realistic profiles without separating different weekdays, but only weekdays from weekend days,
therefore seem reasonable.
Models for predicting domestic hot water use have previously been developed for Swedish
conditions as well. For example, Wollerstrand (1997) describes a statistical model based on the
empirical work and probabilistic model developed in Holmberg (1987). The models are developed
to enable simulation of hot water loads in different types of residential buildings, where the
stochastic variables were derived based on measurements. Those models however focus on peak hot
water load to enable sizing of components for district heating and do not take the actual hot water
consuming activities in a household into account.
In this study we have used data from time use diaries, presenting household activities, to construct
profiles for hot water usage on single household level. Such profiles are “true stories” of how the
activities follow each other in sequences that makes sense in how a household organises its
everyday life. An approach based on probabilities for activities do not contain this dimension of
organisation. This dimension is further highly valuable when working out strategies for hot water
usage with optimal use of solar energy.
This paper presents the results from construction of hot water profiles for 179 Swedish households
living in detached houses and apartments. The time use study on which the hot water profiles are
based will first be introduced. The method to convert the data on activities to energy use will then
be described and important assumptions presented. The results, in form of the generated hot water
profiles for detached houses and apartments as well as profiles used in two simulation tools, are
then shown and compared. Finally, the future use of and possibilities with the results are discussed.
The aim of this study is to model hot water user profiles based on time use data from Sweden. By
translating activities recorded in time diaries to hot water use, average as well as individual hot
water profiles can be generated. This means that the profiles will be based on people’s statements
about their activities over the day.
To create hot water profiles the outcome of a Swedish investigation on residents’ use of time has
been studied. The pilot study was performed in the autumn 1996 by Statistics Sweden, while the
data was coded at Linköping University (Ellegård et al., 2004). The material has previously been
used in time geographic studies as well as in a study on residential electricity use (Widén et al,
2008) (Kall et al., 2007).
The time use study includes a total of 464 persons from 179 different households. The households
both live in apartments and detached houses and the participants are of varying age. Household
members older than 10 years have written so called time diaries during the same days; one weekday
and one weekend day, predefined by Statistics Sweden. The persons then fill in all activities they
perform during the day. The activities were defined by the persons themselves and are therefore
given with different degrees of detail in the material. The number of activities defined in the
material is consequently extensive. The time resolution of the data is one minute, but it is seldom
used. Five minutes intervals are more frequently used, and therefore the detail of the hot water user
pattern is set to 5 minutes resolution.
In the background data family composition, such as number of children older and younger than 10
years, type of living (apartment or detached houses) and occupation among others can be found.
The data also tells if the person is at home or not and if the person is alone or in company with one
or more family members, or other persons. In this study the activities in which the persons use hot
water at home have been extracted. This yields a profile of demanded hot water from the solar
heating system, and a load profile representing the household energy use, while inclusion of hot
water use during daytime at work would yield a more extensive picture of a person’s total hot water
use. That is however beyond the scope of this study.
Detached houses
Out of the total, 271 persons live in detached houses, divided into 92 households. This study mainly
focuses on detached houses, since solar heating is most often installed for single family houses. All
households were then considered, from one person households to those with up to six persons.
All households with complete time diaries living in apartments were also considered. A possible
difference in hot water use between the two types of living can then be investigated. The time use
study comprised a total of 128 persons living in apartments, divided into 64 households. The
number of household members varied between one and six persons with a majority of small
To enable creation of hot water profiles the time use data had to be converted to energy. This was
done in several steps. First, all activities associated with hot water consumption were identified
from the list of activities specified in the time diaries. The hot water can either be used before or
during the activity. For example, tapping or taking a bath often means that the bathtub is first filled,
and then the person takes the actual bath. Hot water is in fact not used during the activity, but
before. In the same way hot water is only assumed to be drawn in the first time interval to do the
dishes in the sink, and the same dishwater is used in the other time intervals during the activity.
A certain volume of hot water was then connected to every activity, either as a fixed volume of
water per activity or a volume per time unit depending on the type of activity. For those activities
requiring large volumes of water, often also extended in time, such as showering or taking a bath,
the hot water was defined as a certain volume per five minutes of activity. This means that a certain
volume or power was equally added to those five minutes intervals where the activities took place.
However, a bath requires a limited volume of water. When this maximum volume is reached the
energy use for hot water in the next time step will be zero.
To find the required hot water volumes rules of thumb on average consumption connected to
different activities were used in the first place, but when necessary reasonable assumptions were
made for example by trying different activities. By programming in the computer based calculation
program Matlab profiles could then be generated for every person in the study by adding the hot
water used in the different activities performed during the days. Profiles could also be generated for
households, groups of people or all participants. To get average consumption patterns the hot water
use of all persons living in detached houses and apartments were combined respectively. Finally, the
generated profiles were compared to the simplified hot water profiles from two standard solar
heating simulation tools, Winsun (Perers et al., 2002) and Polysun (SPF, 2004). The comparison
aims at showing the resemblance between the shape of the stylized profiles and those generated by
the time use data, since the latter will show when hot water consuming activities are actually taking
place. The magnitude of the hot water use will however be analyzed and evaluated in detail in
future work focusing on validating the model.
To find exceptions from the average, profiles were also generated for every single household. They
were compared to the average profile to find “extreme” hot water behaviour.
The model to design hot water profiles could also be used to separate different activities or tap
places. Thereby the hot water demand connected to a certain activity, or the relative distribution of
hot water on different activities within a household or a group of people, can be studied in detail.
The hot water temperature is assumed to be 55 °C, representing the tank storage temperature. The
water is then mixed with cold water of 10 °C at the tap place to reach the desired, usable,
temperature which differs slightly between different activities.
Statistics on hot water use during different activities are hardly found in literature. The Swedish
Energy Agency performs measurements in Swedish households to increase the knowledge about hot
water use at different tap places in dwellings (Stengård et al, 2007). That study is however not yet
finished and will not separate different activities taking place at the same tap. Therefore values on
hot water demand for shower, bath and dishes, used as rules of thumb by Swedish energy
companies are applied when generating hot water profiles in this study. The amounts of hot water as
well as energy connected to some of the most important hot water consuming activities are
presented in Table 1. The volumes in the middle column are the amounts of 55 °C hot water
required to reach usable temperatures when mixed with cold water. The amount of energy in the
right side column is the increase of heat compared to cold water, here 10 °C.
Most households are assumed to have traditional taps, both in the shower and at the tap places. The
flow in a traditional shower tap could be as high as 35 litres per minute, while in a modern tap it is
about 12 litres per minute according to armature manufacturers. Water flow in modern taps at
washstands is about 5 litres per minute and for kitchen taps about 7 litres per minute.
Table 1. Common hot water activities assuming 55 °C water temperature.
Litres of water
40 l/5 min
2.1 kWh/5
100 l
5.2 kWh
Wash hands/go to
0.7 l
0.04 kWh
Dishes, by hand, in
16 l
0.8 kWh
Hand wash
0.7 l
0.04 kWh
0.5 l/5 min
0.02 kWh/5
3.3 l/5 min
0.2 kWh/5
Washing clothes by 13 l
0.7 kWh
For cooking and baking one hand wash or rinsing household goods was assumed to occur twice
every 15 minutes, see Table 1. For washing clothes by hand a temperature of 40 °C was assumed,
but also washing up once and rinsing three times.
Washing machines and dish washers have not been considered in this study. At present most of this
equipment are fed with cold water and heated with electricity in Swedish households.
To try the hypothesis of the importance of hot water use, simulations were first performed in the
solar heating simulation tool Polysun 3.3. The simulations were done for an average new-built
house in Stockholm, Sweden, with a combisystem supplying both space heating and hot water. The
solar collector system comprises 12 m2 ordinary flat plate solar collectors and 1 m3 storage tank.
The four different hot water profiles available in Polysun 3.3, which can be found in Fig. 7 below,
were introduced in the model. It was found that the profile did affect the solar utilization. The main
difference between profiles with morning and evening peaks respectively was in winter, where hot
water peaks in the morning increased the solar contribution by up to 14 percent per month due to
the decreased need of auxiliary heating to keep the tank temperature during nights. On the contrary,
in summer hot water peaks in the evenings were favourable, increasing the solar contribution by up
to 3.5 percent or 17 kWh per month.
Hot water use in detached houses
To enable simulations with load profiles based on time use data and compare different user patterns,
hot water profiles were modelled. The hot water consuming activities of all 271 persons living in
detached houses were combined to one profile; see Fig. 1 for the weekday pattern and Fig. 2 for the
weekend pattern. The total hot water use in the weekday is 82 litres or 4.3 kWh per person and day
and 86 litres or 4.5 kWh per person and day for the weekend day. Varying figures on average hot
water demand in Sweden are given mainly by the energy companies, from 66 to 100 litres per
person and day. This, however, indicates that the assumptions made on hot water use for different
activities are reasonable.
Fig. 1: The average hot water use over the day (weekday) for all persons in the study living in
detached houses.
The main difference between weekdays and weekends is the later morning peak and the broader
peaks in the weekend. Furthermore, the morning and evening peaks in the weekend are rather
similar, while for weekdays the morning peak is much higher and narrower than the evening peak.
Peaks both in the morning and in the evening can however be clearly distinguished for both
weekdays and weekends.
Fig. 2: The average hot water use over the day (weekend) for all persons in the study living in
detached houses.
Peaks in hot water use for most households fall within the peaks of the average curve in Fig. 1, i.e.
in mornings and evenings. Nevertheless, there are households that deviate significantly. In Fig. 3 the
hot water use of three families with different “extreme” hot water behaviour are presented. The
evenly distributed pattern is for a two person family, the morning peaks for a four person family and
the evening peaks for a six person family. The curves represent an average family member. For
comparison, the average profile is also shown.
Fig. 3: The weekday hot water use of three families with different “extreme” behaviour, one only
using hot water in the morning, one only in the evening and one evenly distributed during daytime.
The average use is also shown.
Hot water use in apartments
The hot water use of all 128 persons living in apartments was also combined to create an average
profile. The hot water use for an average person during a weekday is found in Fig. 4 and for a
weekend day in Fig. 5. These profiles resulted in a much higher daily hot water demand compared
to the average person in detached houses; see Fig. 1 and Fig. 2. In the apartments a total of 123
litres of hot water per person and day was used during a weekday and 140 litres per person and day
for a weekend day. This corresponds to 6.4 kWh per person and day for weekdays and 7.3 kWh per
person and day for weekend days. It is considerably higher than the standard values given by the
energy companies (66 to 100 litres per person and day), which may indicate that the assumptions on
hot water use for the different activities should be corrected. This will however be investigated in
detail in future papers.
Fig. 4: The average hot water use over the day (weekday) for all persons in the study living in
The weekday and weekend profile for apartments differs mainly in the more constant use of hot
water during weekdays, with the morning peak being the exception. During weekends there are two
well distinguished peaks, while in weekdays only the morning peak is clearly seen.
Fig. 5: The average hot water use over the day (weekend) for all persons in the study living in
Comparison between detached houses and apartments
The main difference between the hot water profiles generated for detached houses and apartments is
the much higher hot water consumption in apartments. Furthermore, the similarities between
weekdays and weekends are less for apartments than for detached houses. The morning and evening
peaks are not similar in apartments and in weekdays there is no clear evening peak at all. On the
other hand, the morning peaks for weekdays are similar for both types of living, although the total
hot water use is more or less doubled in apartments compared to detached houses. The higher hot
water use in apartments compared to detached houses is also shown in the preliminary results from
the measurements performed by the Swedish Energy Agency (Stengård et al, 2007). The common
opinion about the reason for higher hot water consumption in apartments is that hot water is usually
included in the rent and not individually billed. A qualitative study of effects from introducing
volumetric billing in an apartment building in Sweden however shows small, but no substantial,
changes in hot water routines (Krantz, 2005). Possible reasons for the minor reduction are
discussed, such as the household selection where several participants were already interested in
those issues, as well as all households having enough monetary resources to pay for their water use.
The participants though consider themselves less wasteful after the introduction of volumetric
Comparison to standard hot water profiles
Rather simple hot water profiles are usually used in solar heating simulations. In the simulation tool
Winsun the relative demand profile in Fig. 6 is used. The load is scaled depending on the total hot
water demand defined by the program user. All days have the same profile, making no difference
between weekdays and weekends.
Comparison between the simplified profile in Winsun and the ones based on time diaries, i.e. Fig. 1
and Fig. 2 for detached houses and Fig. 4 and Fig. 5 for apartments, shows that the peaks appear
more or less at the same times for weekdays in detached houses, around 6-7 a.m., 11-12 a.m., 5-6
p.m. and 8-9 p.m., but not for weekdays in apartment, since those have almost no clear peaks. The
hot water profile based on time diaries is also much more complex with several smaller peaks and
does not coincide very well for the weekend day, neither for detached houses nor for apartments,
since the peaks are differently distributed during weekends.
Fig. 6: The hot water demand profile used in the solar collector simulation tool Winsun.
The solar heating simulation tool Polysun enables choice between the four profiles presented in Fig.
7; constant use, morning peaks, evening peaks or daily peaks. The total hot water load is again
scaled depending on the chosen daily consumption. Different amounts of hot water can be defined
for different months, but not for separate days. Holiday periods with no hot water demand can also
be selected.
Fig. 7: The four different hot water demand profiles (constant use, morning peaks, daytime peaks and
evening peaks) that can be chosen in the solar heating simulation program Polysun.
The four profiles in Polysun 3.3 do not coincide with the profiles generated from time use data to as
high extent as those from Winsun. The peaks appear too late in the mornings and too early in the
evening. The extreme cases are however well represented in Polysun, even though the examples in
Fig. 3 are even more extreme with almost no peaks at all at other times and the peaks appearing at
different times of the day. The profiles in Polysun also generate a base load of hot water throughout
the day, only excluding a few hours during the night.
Comparison to previous work
Similar studies focusing on creating realistic hot water user profiles have been performed before,
for example by Jordan et al (2000), which was mentioned earlier. The main difference between her
work and what is done in this study is the type of background data used. In Jordan et al probabilities
for different activities were used to create activity patterns over the day, which means that probable
distribution of activities performed by people, distributed over the day and over the week, is shown.
This study, on the other hand, starts from peoples own statements about their activities, what people
actually consider themselves doing. That means a background data set built on real activities, and
activity sequences that are logical to the individual performing them. This therefore yields more
realistic profiles of hot water use in households.
Detailed measurements at different tap places in dwellings have seldom been performed. In
technical measurements information on individual energy use by different households members are
however missed, since only total use at different tap places can be recorded. By generating hot
water load profiles based on time diaries the hot water use can be identified to a certain person and
feedback can be given to increase the household members’ awareness of energy use at home.
Extensive and detailed measurements of hot water load in dwellings would still be a highly
advantageous complement to time use studies. Hot water use in eight households is measured in
detail by the Swedish Energy Agency (Stengård et al, 2007). This will be extended to include
incoming water to another 60 households. To generate reliable and representative data on hot water
use at different tap places detailed measurements on a larger number of households would however
be desirable.
Domestic hot water user profiles have been modelled for an average person living in a Swedish
detached house based on time diaries from 271 persons in 92 households. Profiles were also
generated for an average person living in an apartment based on 128 persons from 64 households.
The difference between detached houses and apartments was significant, both regarding total hot
water use and the shape of the profiles. During weekdays only a morning peak was clearly
distinguished in the apartments, while both a morning and an evening peak is seen for households in
detached houses and for weekends in the apartments. It can be concluded that the hot water usage
follow considerably different patterns when comparing individual households. How this will finally
influence the solar fraction for the same solar heating systems placed in the different households
will be the task for our next study. It has been shown here that the hot water profiles used in solar
heating simulation tools do not agree with the average hot water profile obtained from the time
diaries in this study.
The next step in this study is to validate the model created to convert data from time diaries to
energy use. This will be done by introducing time use data from a household where energy supply
to the hot water tank has been measured simultaneously. This will yield an idea of how well the
model works, but also if the assumptions are within reason. Moreover, by comparing the generated
average profiles to average measured data from several Swedish households the model can be
further validated. There may however be a certain difference due to the selection of people included
in each study. This work will be done and reported in cooperation with a study on electricity use
based on the same time use data (Widén et al, 2008).
After validating the model, the hot water profiles will be introduced in solar heating simulations to
study how the solar fraction is affected by different profiles. This will yield more realistic dynamics
between the solar collector system and the users. The difference between the simplified and the time
diary based hot water profiles can then be studied, as well as how different family compositions
influence the ability to utilize solar heating.
Solar heating systems are most often installed for single family houses. By investigating profiles
based on time diaries from people living in apartments, larger systems for blocks of flats can also be
studied. Additionally, the profiles generated from time diaries can be used to study benefits and
drawbacks with joint and individual solar heating systems for residential areas. When applying a
larger cooperative solar heating system to a residential area the effect of coincident load behaviour
is utilized. This means that the hot water load will be more evenly distributed over the day, since
different households perform certain hot water consuming activities at different times of the day
according to the example presented in Figure 3. This may be beneficial to the solar fraction.
The work has been carried out under the auspices of The Energy Systems Programme, which is
primarily financed by the Swedish Energy Agency. Thanks to Joakim Widén, Uppsala University,
for converting the original data in text format to matrixes to be used in this paper.
Ellegård K. and M. Cooper (2004). Complexity in daily life – a 3D visualisation showing activity
patterns in their contexts. eIJTOUR (Electronic International Journal of Time Use Research), Vol 1.
Holmberg S (1987). Flow rates and power requirements in the design of water services. Doctoral
thesis from Department of Heating and Ventilation Technology, Royal Institute of Technology,
Jordan U. and K. Vajen (2000). Influence of the DHW load profile on the fractional energy savings:
A case study of a solar combi-system with TRNSYS simulations. Solar Energy, Vol 69, p. 197-208.
Kall A-S. and J. Widén (2007). Det stora i det lilla och det lilla I det stora. Arbetsnotat 37, ISSN
1403-8307, The Energy Systems Programme. In Swedish.
Krantz, H. (2005). Matter that matters: A study of household routines in a process of changing
water and sanitation arrangements. Doctoral thesis from Department of Water and Environmental
Studies, Linköping University, Sweden.
Perers B. and C. Bales (2002). A Solar Collector for TRNSYS Simulation and System Testing. A
Report of IEA SHC – Task 26, SERC Dalarna University, Borlänge.
SPF Solartechnik Prüfung Forschung (2004)., “Polysun 3.3“.
Stengård L. and A. Johansson (2007). Presentation at the Swedish Energy Agency’s hearing on
improved energy statistics for built environment 2007-11-08 in Stockholm. Available online
Widén J., E. Wäckelgård and K. Ellegård (2008). Modeling household electricity load from timeuse data. Proceedings of Green Energy with Energy Management and IT, Stockholm 12-13 March
Wollerstrand J. (1997). District heating substations: performance, operation and design. Doctoral
thesis from Department of Heat and Power Engineering, Lund Institute of Technology, Sweden.
Joakim Widén*, Ewa Wäckelgård**, Kajsa Ellegård***
*Department of Engineering Sciences, Uppsala University
Box 534, SE-751 21 Uppsala, Sweden
[email protected]
**Department of Engineering Sciences, Uppsala University
Box 534, SE-751 21 Uppsala, Sweden
[email protected]
***Tema Technology and Social Change, Linköping University
SE-581 83 Linköping, Sweden
[email protected]
Increased knowledge of electricity load patterns in households is needed for many different purposes,
a few of which are simulation of distributed electricity production and studies of electricity use in the
context of everyday life. This paper describes a modeling approach aimed at obtaining reliable load
profiles for household electricity from time-use data. The objective has been to minimize the number
of assumptions while still keeping the resulting model output sufficiently accurate. The main material
used as input data to the model is a time use survey collected by Statistics Sweden in 1996. The model
outputs hourly data on electricity use in eleven appliance/activity categories for one weekday and one
weekend day per household. The modeled data covers 169 households and distinguishes between
detached houses and apartments. Validation was done against preliminary measurements from the
Swedish Energy Agency’s on-going measurement study on household electricity and showed close
correspondence between modeled and measured data. Possible uses of the model are identified, as
well as benefits of using time-use data instead of, or as a complement to, electricity measurements.
Previous studies of energy use in the Swedish residential sector have shown discrepancies between
future scenarios and actual energy use (Nässén & Holmberg, 2005). Evidently, development of
technical efficiency of household appliances is relatively easy to predict, while activities in
households, which govern the use of the technology, are much more complex and unpredictable.
There is thus a need for better understanding of the dynamics behind energy use in the residential
sector. Moreover, there is an increasing interest in high-resolution statistics on electricity use in
households. One example is the current survey by the Swedish Energy Agency (Bennich, 2007).
From a technology point of view, increased knowledge of daily load patterns can contribute to
better dimensioning of renewable energy technology in distributed forms (solar power, wind power,
micro-CHP, etc.). Since the main idea is to produce electricity close to the load, it is crucial that load
and production match each other well. For studies of this, daily load patterns are needed. From the
electricity user’s point of view it is important to understand how electricity use is constrained by the
complex patterns of habits and activities in the everyday lives of household members (Green &
Ellegård, 2007). For example, which are the opportunities for change in electricity-demanding
This paper reports on work aimed at filling the above needs by a modeling approach. The objective
of the study has been to model household electricity use from a detailed data set on the time use for
every-day activities in Swedish households. The work is by definition cross-disciplinary, combining
the field of modeling and prediction of electricity load with time-geographic studies of household
activities. The modeling has been done in a bottom-up manner. Since this modeling approach often is
complex and in need of numerous assumptions, thus making the exact modeling procedure
nontransparent and hard to reproduce, an important goal has been to use as simple assumptions and
modeling techniques as possible, while still keeping the results sufficiently accurate.
There are a number of benefits of using time-use data for modeling electricity load, apart from it
yielding detailed data on electricity demand. First, it is possible to express electricity use in terms of
the everyday activities in households, which brings electricity use closer to the actual doings of the
household members. Second, it adds different dimensions to electricity use. For example it is possible
to split energy use between household members, genders and age groups, to study who is performing
which electricity-demanding activities. Third, it could be a cost-effective way of estimating electricity
use in households. Instead of installing expensive measurement devices, the electricity could be
modeled from self-reported activities.
When considering electricity use in the Swedish residential and service sector, one usually
distinguishes between household electricity, direct electric heating and electricity for common
purposes. In the official energy statistics of Sweden, electricity for common purposes is electricity
used in premises, direct electric heating covers heating in both premises and households, and
household electricity is the electricity used for electrical devices (lighting, appliances, etc.) in
apartments and detached houses. (Swedish Energy Agency, 2007)
While the use of electric heating peaked in the 1990’s and since then decreased somewhat, reaching
29 TWh in 2006, the use of household electricity, 22,1 TWh in 2006, has increased steadily since the
1970’s. The increase in household electricity is partly due to an increased number of households,
partly due to more appliances in each household. (Swedish Energy Agency, 2007)
There have been measurements done on household electricity at appliance level in Sweden (e.g.
NUTEK, 1994) but none that reveal detailed daily patterns of use on different appliances. To improve
the statistics on household electricity, the Swedish Energy Agency is currently performing a
measurement survey of Swedish households, running between 2005 and 2008. The study, which
covers 400 households, will show how the electricity use varies during the day and between different
appliances. (Bennich, 2007)
The objective of the model presented here is to model electricity with both a high time resolution
(hours) and a high ‘spatial’ resolution (single appliances in individual households). Most models do
not achieve this; high resolution in time usually implies a low spatial resolution, and vice versa.
Examples of models with a high time resolution are load forecast models that predict the power
demand for a cluster of households, typically within a utility’s area of supply (for a review of such
models, see Alfares & Nazeeruddin, 2002). What these models have in common is that the model
structure can be simplified by application of a low spatial resolution. For example, variations in power
demand between large numbers of households can be modeled as noise or implicit variations in timeseries. Examples of models with low time resolution and high spatial resolution include econometric
models where total annual electricity use is broken down into use per appliance or household activity,
e.g. by use of statistics on appliance ownership or sale figures for appliances (Parti & Parti, 1980;
Sanchez, et al., 1998; Larsen & Nesbakken, 2004).
Models with both high time resolution and high spatial resolution need large amounts of data. A
typical technique for achieving an arbitrary resolution is bottom-up modeling. Starting from the
smallest possible units of a system, bottom-up models successively aggregate these to reach higher
system levels. The so-called ‘Capasso model’ constructs load curves by adding the power demand of
single appliances together. Detailed data on demography, socio-economical status and lifestyle from a
variety of sources, including time diaries, are used (Capasso, et al., 1994). It is obvious that this
dependency on detailed and specific data narrow the possibilities of applying the model in other
contexts. Another bottom-up model, developed at Helsinki University of Technology, uses statistical
mean values and general statistical distributions, which lowers the model precision, but also decreases
the amount of data needed (Paatero & Lund, 2006).
As depicted in figure 1, the model presented in the following chapter outputs household electricity
demand, as defined above, for all households covered by the time-use data, used as input to the
model. The model is dependent on a number of parameters that determine the electricity use of
appliances and a daylighting scheme for modeling lighting. Input data, model structure, parameters,
implementation and output are all described in detail below.
Appliance powers
Time-use data
Fig.1. Overview of data, parameters and output of the model.
4.1 Time-use data
The basic material for this study is a pilot time-use survey by Statistics Sweden from 1996
(henceforth called ‘SCB 1996’). Data on the time use for different activities in households were
collected by letting each person in every participating household write a diary reporting the timing of
activities, defined by the diary-writers themselves, together with information about geographic
location or means of transport, and with whom they were (Ellegård & Wihlborg, 2001). Most
households wrote diaries for one weekday and one weekend day and all were collected in the autumn
of 1996, with dates ranging from August to December. Individuals aged between 10 to 97 years took
part in the study. The resulting data set is mainly based on 5-minute intervals (although a few
households have reported on 1-minute basis) and the activities are organized in a detailed hierarchic
activity scheme, defining activities at different levels of abstraction (Ellegård, 1999). An example of
the logic behind this scheme is shown in figure 1. The figure shows the different activities involved in
cleaning, as a subcategory of “Room care”, which in turn is a subcategory of “Household care”.
Cleaning can be subdivided into more detailed activities as shown on the fourth level of detail
(“Routine cleaning”, “Airing” and “Thorough cleaning”) and subsequently in even more detailed
activities (such as “Sweeping” or “Beat carpets”). Considering energy-demanding activities, “Vacuum
cleaning” (marked with bold text in figure 1) is the only activity in need of electricity. The scheme
gives an impression of the way in which energy use is embedded in larger activity schemes.
Routine cleaning
Scrub floor
Vacuum cleaning
Thorough cleaning
Beat carpets
Scrub floor
Clean windows
Vacuum cleaning
Wash up (not food)
Fig.2. Activity code scheme for cleaning, a section of the complete scheme used for categorization of
activity data. Cleaning is a subcategory to ‘Room care’, which in turn is a subcategory to ‘Household
care’. Both of these include various subcategories, apart from those shown here. ‘Household care’ is
also one out of seven main categories. Vacuum cleaning, which is the only electricity-demanding
activity considered in the modeling, is shown in bold text.
Previous work with the time-use data set has involved the creation of a computer program, developed
to visualize the data (Ellegård & Cooper, 2004; Vrotsou, et al., 2007). Figure 3 shows how the
cleaning activities defined by the scheme above are distributed among the persons in the data set for
weekdays and weekend days. The horizontal axis orders individuals by gender and age, and the
occurrence of each person’s activities in time is depicted by vertical lines, with time shown on the
vertical axis. From the figure it is easy to see the difference between men and women as well as the
difference between weekdays and weekend days.
Fig. 3. Screenshots from the Visual-TimePAcTS program described in Ellegård & Cooper, 2004,
showing the occurrence of the cleaning activities defined in figure 2, for weekday (a) and weekend
day (b). The horizontal axis orders the individuals by age and gender; men ordered by increasing age
to the left, and women ordered by increasing age to the right. The vertical axis shows the time of day.
A number of observations are easily made. Women do more cleaning than men, there is more
cleaning on weekend days, and cleaning is more concentrated to mid-day on weekend days.
Subsequent time-use surveys were conducted after the 1996 study. One set of data from 2000 and
2001 is currently being imported into the software and it is hoped that the model described here can
be applied to this data set to assess differences between 1996 and 2000/2001. Collection of time-use
data from diaries has also been carried out in a project in the Swedish Energy Agency’s electricity
survey, and these were used for validation of the model in section 5.
4.2 Model structure
Prior to modeling, all activities that were assumed to involve use of electricity were examined. Some
activities were then excluded, although electricity use could be assumed, because the occurrence in
the material was low (typically in one household) in combination with ambiguities in modeling. None
of these activities were assumed to contribute substantially to the total energy use of the households
where they appeared. Four different modeling schemes were used, depending on the type of activity
and appliance involved:
a) Power demand not defined by activities. This applies to the use of cold appliances. This was
modeled as a base load in the form of a constant power.
b) Power demand constant during activity. This applies to cooking, ironing, cleaning, and use of
TV, audio appliances and computer. The power demand is considered constant during the time given
for the activity, as shown in figure 3. This is more of a simplification for certain activities than for
others, typically for cooking, which covers a variety of different appliances and patterns of usage not
captured by the time-use data. It is a simplification for ironing as well, which is thermostat-controlled,
but it is the best that can be done without complicating the modeling unnecessarily. Different
modeling routines were tested to improve the modeling of cooking, for example making the power
demand dependent on the type of meal reported in the “eating” activities subsequent to cooking.
However, the simplified approach gave more realistic results both for single households and for the
mean load curve. To take into account the possible use of multiple appliances, the company data for
each person was used. All persons reporting that they were accompanied by another family member
while watching TV or using audio appliances and computer are assumed to use one appliance
together, while persons reporting that they were on their own are assumed to use one appliance each.
This means that explicit assumptions about the number of appliances in the households are avoided.
Fig. 3. Modeling scheme for electricity use for activities where electricity is used during the entire
course of the activity. Throughout the activity, power of level Pmax is demanded.
c) Power demand constant after activity. This applies to dish-washing, washing and drying. The
activities used in the model are of the type “Fill washing machine”, so that the actual appliance use
can be assumed to begin after the activity has finished. Power is then demanded until a maximum
time has elapsed or until the activity starts anew.
Fig. 4. Modeling scheme electricity use for activities where electricity is used after the entire course of
the activity and with a maximum time limit. When the activity is finished, power of level Pmax is
demanded, until maximum time tmax has elapsed or the activity starts anew.
d) Activities with time-dependent power demand. This applies to lighting and is similar to scheme
(b) above, with the difference that the power varies with time. In order to make the modeled load
curves exhibit realistic daily and seasonal variations in lighting demand, they were made dependent
on the daylighting level. All illuminances above a certain limit value (Lmax) imply minimum demand
of lighting power (Pmin). For illuminances L between the maximum value and zero, the power demand
P is set to vary according to the following formula:
P = Pmin + (Pmax – Pmin)(1 – L/Lmax)
This varying power is demanded by each person at home and awake.
4.3 Parameters
The model is dependent on parameters describing the power demand for activities or appliances used
during activities. The objective has been to find a “standard appliance set” that gives a realistic mean
energy use when applied to all households. Estimates of standard powers and runtime for appliances
were obtained through product tests, mainly from the Swedish Consumer Agency, reflecting the range
of appliances available at certain times. To the extent possible, the mean lifetime of appliances was
taken into account. This method of course gives rough estimates of appliance characteristics, since it
is hard to find reliable data on the actual scope of appliances. In the results reported here, the same
appliance set has been applied to all households. However, the model is flexible and allows different
appliance sets to be defined for each household. The resulting parameters are shown in table 1.
Another important set of parameters, not shown here, is the set of illuminance values used to
determine the lighting power scheme. Such data can be obtained as monthly means of hourly values
from the online database SatelLight (SatelLight, 2007). In the following simulations SatelLight data
for Stockholm were used. The illuminance limit for minimum power demand was set to 500 lux.
Table 1. Parameters used in the model: power demand for activities/appliances and maximum
runtime where applicable.
Appliance/activi Power
Cold appliances
Lighting (min /
80 / 200
4.4 Implementation and data preparation
The actual program code for extracting data from the activity database and implementing the model
was written partly in Java, partly in Matlab. Before applying the model to the SCB 1996 data, some
households were excluded for different reasons. In a few households, there were incomplete data for
some persons and in another household two weekend days were reported instead of one weekday and
one weekend day. Apartments and detached houses were also separated. The subset finally contained
431 persons in 169 households, with 103 families living in detached houses and 64 in apartments. 2
households’ type of residence was undefined and therefore left out when separating detached houses
and apartments.
4.5 Output
Although time-use data are recorded on 5-minute intervals, or even 1-minute intervals, one-hour
means were chosen as output. This is because the simple assumptions used in the modeling approach
are sufficient on an hourly basis, but would be too rough on shorter time intervals. There is obviously
a trade-off between model complexity and output resolution, but since hourly intervals are generally
needed for applying the model in different simulation programs, this seemed to be a reasonable aim.
The general output of the model, when applied to the SCB 1996 data, consists of hourly values of
demanded power for the appliances/activities in the eleven categories, shown in table 1, for weekday
and weekend day in each household. The lighting demand also varies with each month, which affects
the total load curves. Figure 5 shows the December and June load for detached houses and weekday,
indicating the effect of the varying lighting demand.
Lighting power demand
Total demand except lighting
Power (W)
Power (W)
Time (hours)
Lighting power demand
Total demand except lighting
Time (hours)
Fig. 5. An example of output from the model when applied to SCB 1996, showing how the power
demand for lighting varies between months. The remaining power demand is the same in the two
figures. The figures show total load curves, averaged over all households, for detached houses and
weekday, for the month of December (a), and for the month of June (b). The morning peak for
lighting, distinctive in the winter, almost disappears in the summer.
Ideally, the model should be validated against measured electricity load in the same households as
where the time-use data were collected. This is possible for the small number of households that took
part in the time-use subproject in the Swedish Energy Agency’s measurement survey. When applying
the model to the SCB 1996 data, no such measurements are available, neither on the surveyed
households, nor on any households with the required level of detail. The aim has therefore been to
validate the mean output from the SCB 1996 data and parameters reflecting today’s ‘standard’
appliance ownership against the preliminary mean results of the Swedish Energy Agency’s
measurement survey. While the first type of validation shows the model errors, the second reflects
both model errors and errors from using SCB 1996 data for predicting today’s electricity use. These
errors are described in more detail below.
5.1 Possible errors
The whole process from the collection of the SCB 1996 time-use data to the validation of the final
model output contains a number of possible sources of errors. These are important to identify in order
to interpret the validation results correctly. The most important are the following:
(1) Bias in the household selection process.
(2) Diary-writers not reporting with care.
(3) Problems with interpreting exactly the diary-writer’s notice and applying activity scheme to it.
(4) Modeling assumptions.
(5) Estimation of ‘standard’ appliances.
(6) Non-representativity of the surveyed households relative to the targeted households when
5.2 Comparison with measurements
Two comparisons with measurements were made: (a) between total average load curves from the
model and the Swedish Energy Agency’s preliminary results, and (b) between model output and
measurements for a household where both time use and electricity use were followed. The first
comparison shows how well the model estimates the electricity demand of today when applied to
SCB 1996. Differences between model output and measurements can then be caused by any of the
above error sources. The second comparison shows how well the model works when the households
with time-use data and the targeted measured households are the same. In this case, bias errors (1) and
non-representativity problems (6) are irrelevant. Differences are only due to error sources (2) – (5).
Figure 6 shows the correspondence between model output and measurements for detached houses
and apartments on weekdays and weekend days.
Detached houses, weekend day
Power (W)
Power (W)
Detached houses, weekday
Time (hours)
Power (W)
Power (W)
Time (hours)
Apartments, weekend day
Apartments, weekday
Time (hours)
Time (hours)
Fig. 7. Comparison between the output average load from the model when applied to SCB 1996 (–)
and the measured average load from the Swedish Energy Agency’s measurement survey (- - -). The
figures show total load curves, averaged over all households, for detached houses and apartments on
weekdays and weekend days. The generally close agreement between measurements and model
output is evident. Source of measurement data: Peter Bennich, Swedish Energy Agency, personal
A close overall correspondence can be observed for all four cases in figure 7. The main difference is
that the model output is more variable while the measured data curve is smoother. This is because the
number of households is larger in the measurement survey (200 detached houses and 200 apartments
in the measurement survey, compared to 103 detached houses and 64 apartments in the time-use
survey) and because the number of days per household is much greater in the measurement survey
(months or years in the measurement survey and one day per household in the time-use survey).
Considering this, and the above overview of possible sources of differences between model output
and actual energy use, the model gives very good resemblance with measured data. What is of
importance is that the general pattern is captured by the model, for example that the power demand is
generally lower in apartments than in detached houses, and that the morning load is shifted towards
mid-day on weekend days.
A comparison was also made between each appliance/activity category. For ease of comparison, the
model categories were rearranged to match those of the Energy Agency. An estimate of total annual
use was calculated by scaling the daily values output by the model to annual values and weighting the
weekend and weekday figures appropriately for each month. The result of this comparison, shown in
figure 7, indicates once again that the model yields reasonable results.
Detached houses
Fig. 7. Comparison between predicted annual electricity demand per household from the model and
measured annual means per household from the Energy Agency’s survey. Categories are A: cold
appliances, B: lighting, C: cooking, D: dishwashing, E: washing and drying, F: TV, VCR, DVD, etc.,
G: computer, H: audio appliances, I: additional. The ‘additional’ category includes cleaning and
ironing in the modeled data. Source of measurement data: Bennich, 2007.
Although the exact magnitudes differ somewhat, the general pattern is the same for the measured and
modeled figures. Note for example that the relative magnitudes of cold appliances, lighting and
cooking are roughly the same in both detached houses and apartments. The most obvious difference
between measurements and model output is the computer category. This is probably a nonrepresentativity problem, reflecting the increased use of computers since 1996.
The result of comparison (b) between measured electricity demand and model output for a single
household, where both time use and actual electricity demand were followed, as described above, is
shown in figure 8.
Power (W)
Time (hours)
Fig. 8. Comparison between measured electricity demand and the model output for a single
household. The model gives considerably good correspondence with measurements although the
standard parameters are used.
The figure shows the total load curves for the four successive days covered in the time-use data for
the household. The correspondence between measured and modeled load is not perfect, but follows
the same overall pattern. The number of peaks and the height of the peaks are approximately the
same. The biggest difference occurs on the third day and is due to misplacement of the power demand
for cooking (error source 4) and non-reported dishwashing (error source 2). The cooking power
misplacement is not surprising, given that it is the roughest model assumption. Quite contrary, it is
surprising that the modeled load curve shows such similarity to the measurements on single household
level, although the modeling assumptions are kept simple and the standard appliance set is used. It
should also be noted that there are four other households to validate the model against. This will be
covered in more detail in a coming Elforsk report.
To summarize the validation results, the overall resemblance between measurements and model
output is good, especially considering the simplicity of the model structure and parameters. The
preliminary measurements should not be considered as true either, being subject of a selection process
with possible bias. Combining the results of the two comparisons, these indicate that error sources (2)
– (5), reflecting the modeling assumptions and time-use data quality, do not have any substantial
effect on the model results, while error sources (1) and (6) seem to have some impact on single
appliance categories, computer use being the most obvious one.
The aim to use simple and transparent assumptions to model household electricity loads while
keeping the results sufficiently accurate has evidently been achieved for the 1-hour resolution output.
Comparing with previous models, like the Capasso model, the assumptions are few. An obvious
difference, which is a main reason behind this, is of course that this model is not stochastic. Although
it would be possible to create such an extension of the model, for example by creating synthetic
activity data, there is a point in not doing so and instead keep the original activity data. With the
whole pattern of activities kept intact, it is possible to analyze activity use in context, one of the
benefits of using activity data, as mentioned in the introduction. Methods for such systematic analyses
of activity patterns are currently being developed for the Visual-TimePAcTS program (Vrotsou et al.,
It is possible to identify a number of uses for the model. Mentioned briefly in the introduction was
the possibility to use the model as an alternative to actual measurements of electricity use. With data
on the specific household appliances and detailed time-use data describing the use of them, it should
be possible to model the electricity use with reasonable accuracy, as indicated by the validation of the
model. Both the model and the time-use survey method could then be improved, the latter by making
it more specific and restricted to the use of household appliances.
In the near future, the usefulness of the model will be shown in a number of coming papers. Hot
water demand is modeled from the same data in a parallel study (Lundh, 2007). Since there are no
available statistics covering hot water and electricity demand in the same households, a model
providing such data would be beneficial. Other studies that utilize the modeled load curves in
simulations are under way. Options for load matching in households with photovoltaic generation are
currently being investigated, as well as the limits to the penetration level of distributed photovoltaics
in low-voltage distribution grids in Sweden.
The main conclusion is that the proposed model gives sufficiently accurate results with a minimum of
assumptions. When validating the model against preliminary results from the Swedish Energy
Agency’s on-going survey of household electricity, the model applied to time-use data from 1996
yielded data in good agreement with the mean measurements for all households. When applying the
model to a household with both time use and electricity use measured, it also showed good overall
agreement, with a few explainable discrepancies.
The work has been carried out under the auspices of The Energy Systems Programme, which is
primarily financed by the Swedish Energy Agency. The ELAN Programme financed the work on
model development. The authors want to thank Peter Bennich at the Swedish Energy Agency for
providing measurement data and preliminary measurement results for validation of the model, and
Kristina Karlsson, Linköping University, for help with extracting time-use data for the additional
households used for validation.
Alfarez, H.K. and Nazeeruddin, M., 2002. Electric load forecasting: literature survey and
classification of methods. International Journal of Systems Science 33: 23–34.
Bennich, P., 2007. Mätning av hushållsel (Measurements of household electricity). Presentation the
Swedish Energy Agency’s hearing on improved energy statistics for the built environment 8/11
2007. Available online at (2007-12-16).
Capasso, A., Grattieri, W., Lamedica, R., Prudenzi, A., 1994. A Bottom-Up Approach to
Residential Load Modeling. IEEE Transactions on Power Systems 9: 957–964.
Ellegård, K., 1999. A time-geographic approach to the study of everyday life of individuals – a
challenge of complexity. GeoJournal 48: 167–175.
Ellegård, K. and Cooper, M., 2004. Complexity in daily life – a 3D-visualization showing activity
patterns in their contexts. electronic International Journal of Time Use Research 1: 37–59.
Ellegård, K. and Wihlborg, E., 2001. Fånga vardagen – ett tvärvetenskapligt perspektiv (Capture
the daily life – an interdisciplinary perspective). Lund: Studentlitteratur.
Green, A. and Ellegård, K., 2007. Consumer behaviour in Swedish households: routines and habits
in everyday life, ECEEE Summer Study, June 4–9, 2007. La Colle sur Loup, Côte d’Azur, France.
Larsen, B. M. and Nesbakken, R., 2004. Household electricity end-use consumption: results from
econometric and engineering models. Energy Economics 26: 179–200.
Lundh, M., 2007. Manuscript submitted to the conference.
NUTEK, 1994. Hushållsel i småhus (Household electricity in detached houses), B 1994:11.
Stockholm: NUTEK.
Paatero, J. V. and Lund, P. D., 2006. A model for generating household load profiles. International
Journal of Energy Research 30: 273–290.
Parti, M. and Parti, C., 1980. The total and appliance-specific conditional demand for electricity in
the household sector. The Bell Journal of Economics 11: 309–321.
Sanchez, M. C., Koomey, J. G., Moezzi, M. M., Meier, A., Huber, W., 1998. Miscellaneous
electricity in US homes: Historical decomposition and future trends. Energy Policy 26: 585–593.
SatelLight, 2007. The European Database of Daylight and Solar Radiation, (2007-12-16).
Swedish Energy Agency, 2007. Energy in Sweden – Facts and Figures 2007. Available online at (2007-12-16).
Vrotsou, K., Ellegård, K., Cooper, M., 2007. Exploring time diaries using semi-automated activity
pattern extraction. IATUR XXVIIII Conference, October 17–19, 2007. Washington, DC, USA.
Iana Vassileva, Cajsa Bartusch and Erik Dahlquist
Department of Public Technology, Mälardalens Högskola, Box 883, 721 23 VÄSTERÅS,
SWEDEN; communications: [email protected]
Measurements of electricity and hot water consumption were performed in 24 apartments during the
years 2005 and 2006. Buildings, located in Västerås, Sweden, are characterized by the fact that
energy costs are included in the rent. Data obtained from the company renting out apartments
MIMER were analyzed in order to find out if there were any general trends interrelated with the
level of surface area, number of rooms, or the main driving force is the individual behaviour of
tenants. It was found that the analysis by various parameters is as important as analysis of each
individual apartment. Some tendencies were well defined such as increase of total electricity
consumption during the winter months and lowest both electricity and hot water consumption in
June and July. The number of rooms at equal surface area could not be accepted as a criterion of
electricity consumption. Bearing in mind the interactive contact between the energy company and
the respective tenants, a positive trend of lowering both studied parameters was well manifested if a
comparison of data obtained during 2005 and 2006 was carried out. Obviously, further
improvements in company-tenants relations in terms of more detailed mutual information would
probably change tenants’ behaviour thus improving energy saving tendency in both buildings.
For the EU-25 as a whole, growth in electricity consumption was particularly strong in the service
sector, followed by households (European Environmental Agency, 2007). However, over the period
from 1990 to 2004, this increase is less than 1 % for Sweden. Energy consumption particularly in the
building sector accounts for 39 per cent of Sweden’s total final energy consumption and for about 50
per cent of total electricity consumption in Sweden (Swedish Energy Agency, 2007). The energy is
used to heat areas and water, as well as to operate various types of equipment. Although energy
consumption in the buildings has remained relatively stable since the early 1970s, heating and hot
water account for more than 60 per cent of energy consumption in this sector. During the last 10-15
years, a significant improvement in the energy use in the household has been achieved through
introduction of modern, energy-saving and available technologies. Nevertheless, it is widely accepted
that interventions that target consumer behavior should be reinforced as it presents large number of
possibilities within the sustainable development (Wood and Newborough, 2003). In fact, ssustainable
energy consumption is a key element within sustainable development approach. The hot spot in
sustainable energy consumption is mainly oriented to reducing consumption through tenants’ behavior
analysis and further measures to efficiently change it.
Two points should be considered when assessing energy consumption in multifamiliar buildings in
Sweden. The first one concerns the fact that the cost of utilities (heating, electricity, cold and hot
water) forms part of the total rent per apartment. Therefore, tenants are not encouraged to save energy
like in individual households (individual houses or apartments where tenants pay for utilities). For
example, results showed that apartments for which heat is included in the rent in are kept warmer than
those where tenants pay for heat. The temperature differences is largest when no one is home,
indicating that tenants who pay for heat are more likely to take simple conservation measures such as
turning down the thermostat when leaving home (Levinson and Niemann, 2004)
The second important point is that in Sweden, information related to individual energy consumption is
not available without tenants’ permission which makes difficult deeper data analysis (5). From a
logical point of view the hypothesis that the energy usage in apartments correlate well to the size of a
household seems reasonable. On the other hand, based on the literature analysis we may consider
individual habits as more important factor. A reason for this is that the public company renting out
apartments in Västerås, MIMER, has seen that the overall electricity as well as heating energy
consumed in one area with “energy for free” is almost 50% higher than in a similar area where the
inhabitants pay individual consumption directly. Similarly, in the USA it was recently demonstrated
that tenants who rent apartments with utilities included behave differently than they would if they paid
heating costs separately from rent: they use more heating and turn back thermostats less when away
from home (Sjogren et al., 2007).
Therefore, to investigate the above hypothesis we have studied the usage of electricity and hot water
in apartments in Västerås. There is a plan to install individual measurements of both electricity and
hot water in about 1300 apartments, and thus this information can later be compared to those who
have already had an individual metering for several years. This study was performed in the frame of a
larger project for the development of a method for interactive (between tenants and local energy
companies) information on energy consumption in individual apartments. It is also interesting to see if
information about the individual energy usage can lead to a decrease of the usage.
Data included in this study were collected during two years (2005 and 2006). Measurements were
carried out for both electricity and hot water consumption on hourly basis. Further, data was sent to
central data base using a server existing at MIMER and at Metrima (the company which provided the
meters and collects the hourly values). Presented to the inhabitants by displays placed next to fusebox.
The buildings included in the study were constructed in 2001 and, therefore were provided with
energy efficient facilities to fulfil modern comfort standards as well as with the state of the art with
respect to energy conservation. Data on energy and hot water consumption were available from
apartments situated in the Östra Mälarstrand area at the shore of the lake Mälaren, Sweden. The total
amount of apartments with individual measurements was 24 with a total surface of 1894 m2. Of these,
8 were with 2 room and 62 m2, 1 with 2 rooms and 79 m2, 1 with 3 rooms and 79 m2, 6 with 3 rooms
and 80 m2, 8 with 3 and 95 m2, and 2 with 4 rooms and 95 m2. In each building there were 12
apartments, occupied by 10 and 21 tenants, respectively. Consequently, the average number of persons
is two per apartment.
In Table 1 we can see the annual consumption of electricity and hot water for different groups of
apartments independently of the number of rooms during 2005 and 2006 years.
Table 1. Total electricity and hot water consumption during 2005 and 2006 years
Hot water
group (m2)
62 (average/apt)
2 Builds (24 apts)
(total, all apts)
79 (average/apt)
(total, all apts)
80 (average/apt)
(total, all apts)
95 (average/apt)
(total, all apts)
1894m2 all apts (per m2)
Analyzing the data achieved a general trend for increasing the average electricity and hot water
consumption per apartment should be noted with increasing the apartment surface. However, data
from 80 m2-apartments, particularly obtained for hot water, were no responding to this observation.
Another interesting trend was the decrease of both parameters during 2006 year for all apartments
excluding hot water consumption in 79 m2-apartments. This difference was more pronounced in 80
m2-apartments where 5 208 and 4 775 kWh less electricity and hot water total consumption were
reached. Overall, average values of 47 and 41 kWh of electricity and 59 and 53 kWh of hot water per
m2 were consumed in 2005 and 2006 year, respectively. These values are far lower than 150-230
kWh/m2 of annual energy consumption in the EU and even than 120-150 kWh/m2 in low-energy
buildings in some Scandinavian countries (Balaras et al., 2000).
Fig. 1 a, b. Comparative data on electricity monthly consumption by different groups of
apartments (averaged per apartment)
Comparison between different groups - EL
consumption 2006
62m2, 2r
79m2, 2,3r
80m2, 3r
95m2, 4r
95m2, 3r
80m2, 3r
79m2, 2,3r
62m2, 2r
Comparison between different groups - EL
consumption 2005
95m2, 3r
95m2, 4r
Data, presented on Fig. 1 (a, b) show a well difined tendency of high total electricity consumption in
winter months (December, January, and February). Apartments of higher surface, mainly of 95 and 80
m2 consumed more electricity compared with apartments of lower surface area. The lowest level of
electricity was registered in July independently of the year. Some exceptions should be mentioned that
were not related to the surface area. For example, within the group of 95 m2, it seems that apartments
with 4 rooms consumed significantly higher amount of electricity during the 2005 and particularly in
the “cold” months in 2006 while maximum of monthly electricity usage during 2006 was noted in
apartments of 79 m2.
Fig 2 a, b. Comparative data on hot water monthly consumption by different groups of
(averaged per apartment)
Comparison between different groups - HW
consumption 2006
95m2, 4r
80m2, 3r
95m2, 3r
79m2, 2,3r
80m2, 3r
62m2, 2r
79m2, 2,3r
62m2, 2r
Comparison between different groups - HW
consumption 2005
95m2, 3r
Data obtained for hot water monthly consumption are presented on Fig 2 a, b. In general, the usage of
hot water along the respective 2005 and 2006 year was more stable in comparison with that of
electricity. Lowest consumption was observed in both July and August. Again, as in the case of
electicity total consumption, there were two groups of apartments with higher level of hot water usage
(95 and 79 m2).
Table 2. Maximum electricity (EC) and hot water (HWC) consumption during 2005 and 2006
on monthly and hourly basis (taken for individual apartment basis)
Apartment Maximum EC
Maximum HWC
m2number rooms 2005/2006
Maximum EC
Maximum HWC
Taken on individual basis per each group of apartments, data for maximum values of electricity and
hot water monthly and hourly consumption offers another possibility for analysis. For example, it
appears that the number of rooms affects the consumption but in different manner. For apartments of
95 m2 increase in number of rooms caused a logic enhancement of consumption of both electricity
and hot water although single maximum measurement data did not correlate with this trend. On the
other hand, apartments of 79 m2 demonstrated totally different trend as a higher consumption was
registered in apartments with lower number of rooms while single maximum measurement data
correlate with the maximum monthly consumption.
If we take as indicator the total monthly consumption of electricity and hot water for an average
apartment (Table 3), independently of its surface, data confirmed the observations already discussed
(Fig 1a, b and Fig 2a, b) – higher consumption during the winter months (a trend better pronounced
for electricity and 2005) and lower consumption in June, July and August for both electricity and hot
Table 3. Total monthly electricity and hot water consumption for average* apartment during
2005 and 2006
Hot water
* Total consumption /24 apartments=average monthly consumption per apartment
More detailed view on electricity and hot water consumption for each individual apartment during
2005 and 2006 is presented on Fig. 3a, b and Fig. 4a, b. It is important to note that individual
apartments in general follow the same trend in consumption if we compare 2005 with 2006 years.
There were several exceptions as for example apartment Nº 16 which showed significantly lower
consumption of both electricity and hot water in 2006 compared to those in 2005. On contrary, data of
hot water consumption of apartment Nº 23 was significantly higher in 2006 that that consumed in
Fig. 3a, b. Electricity consumption for 2005 and 2006 for each individial apartment
Total EL consumption for each apartment 2005
kWh 4000
1 2
3 4 5 6
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Total EL consumption for each apartment 2006
1 2 3 4 5 6 7
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Fig. 4a, b. Hot water consumption for 2005 and 2006 for each individial apartment
Total HW consumption for each apartment 2005
kWh 4000
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Total HW consumption for each apartment 2006
kWh 4000
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Another important parameter that should be taken into account and further studied is the ratio
between average and median values, both calculated per year for each individual apartment (data
not shown).
Some of the apartments showed significant difference in their consumption compared to the rest of
the apartments studied. For instance, there were 3 apartments, one of 62m2 and two of 95m2, with
respective ratios of 0.50, 1.36 and 1.62 (for electricity consumption, 2005) compared to 0.98 to 1.02
for the rest of the apartments. For the electricity consumption 2006, the situation was similar to the
previous year - two apartments, one 62m2 and one 95m2, showed ratios of 0.84 and 1.38,
Considering hot water consumption, 2005, there were eight apartments with ratios varying between
0.89 and 1.2. For hot water, 2006, the apartments with higher difference in their ratios were six,
being the lower ratio value 0.89. In all these cases (hot water, 2006) the median was higher than the
average and therefore all the ratios are below 1.0.
Analyzing the results presented in this work, it should be concluded that each group of data is
important, alone but also within the frame of overall interrelations with all other groups. First of all,
increasing the surface area causes enhancement of both hot water and electricity. A well defined
tendency of increased total electricity consumption during the winter months and lowest both
electricity and hot water consumtion was observed in June and July for all groups of apartments. The
number of rooms can not be accepted as a criterium of consumption as different groups of apartments
demonstrated different behaviour using this parameter. One very important observation concerns the
generalized tendency of each apartment to follow in 2006 the same consumation as that registered in
2005 although at lower level. This conclusion is probably the most important as it confirms the
statement that the individual habits are the dominanting reason for the consumption profile of each
apartment although all other parameters should be taken into account when assessing the detailed
electricity and hot water usage. In any case, the system applied in these two buildings from the real
estate, MIMER obviously resulted in decrease in energy consumption if we compare it in 2005 with
2006. Bearing in mind the presented data analysis and conclusions, further work should be oriented
towards changing consumption behaviour of tenants on individual basis and considering data obtained
from each individual apartment but with its peculiar characteristics such as surface area and number
of rooms amongst others. Interactive relation between individual tenants and energy companies and/or
the renting apartments company presenting energy savings and respective money savings will
probably provoke further changes in tenants’ habits that may decrease electricity and hot water
consumption. In this sense, our group is preparing a computerizing system that will enable each tenant
to be well informed about his particular instantaneous consumption.
Further work will focus on continuation of the observation of electricity and hot water consumption of
these two buildings. It would be also interesting the creation and distribution of some questionnaires
that will be given to the tenants. The households will also be informed by diagrams and figures (some
of them presented in this article) about their consumption during 2005/2006.
Afterwards consumption during 2008 will be compared with data studied in this paper and both
possible change in behavior patterns and a lowered consumption will be expected.
We want to thank the Swedish Energy Agency for financial support, as well as Mimer, Mälarenergy,
Eskilstuna Kommunfastigheter and Eskilstuna Energi och Miljö for support.
Balaras, C.A., K. Droutsa, A.A. Argirou, D.N. Asimakopulos. 2000. Potencial for energy
conservation in apartment buildings. Energy and Buildings 31:143-154.
European Environmental Agency, Indicator: EN18 Electricity Consumption (2007.04).
Levinson, A. and S. Niemann. 2004. Energy use by apartment tenants when landlords pay for
utilities. Resources and Energy Economics 26:51-75.
Sjogren, J.-U., S. Andersson and T. Olofsson. 2007. An approach to evaluate the energy
performance of buildings based on incomplete monthly data. Energy and Buildings 39:945-953.
Swedish Energy Agency. 2007.
Wood, G. and M. Newborough. 2003. Dynamic energy-consumption for domestic appliances:
Environment, behavior and design. Energy and Buildings 5: 821-841.
S.M. Esmail Mousavi Takami, M.Sc in architect, [email protected]
The results from implementing building intelligence technologies and processes are buildings that
cost less to operate and are worth more to their occupants. For projects that are owner occupied, such
as corporate, government, and institutions, the benefits of an intelligent building provide an
immediate ROI in terms of higher employee productivity and reduced operating expenses. For
commercial developments, these projects are expected to result in above market rents, improved
retention, higher occupancy rates, and lower operating expenses. Intelligent buildings designing are an
important title in last decades. The definition of an ‘intelligent building’ depends on point of view.
Two last decades building intelligence was constrained by both technology limitations and economic
viability. Recently a whole series of Internet and network communications developments truly
elevated average structures to Intelligent Buildings.
From this vantage point, it is clear that Intelligent Buildings are a reality. Often people confuse
Intelligent Buildings with High-Tech Buildings. This is totally wrong. An intelligent building is not
gadget-oriented. It must be designed to suit the present needs of the occupants. A major problem in the
study of intelligence is: Are we talking about intelligence as an abstract set of capacities, or as a set of
biological mechanisms and phenomena? These two questions set up two dimensions of discussion
about intelligence.
The first dimension: Is intelligence separable from other mental phenomena?
And the second dimension in discussions about intelligence involves the extent to which we need to
be tied to biology for understanding intelligence. Can intelligence be characterized abstractly as a
functional capability which just happens to be realized more or less well by some biological
This piece of work discusses the above aspects as a goal of creating buildings with adaptive
environments of high quality, energy efficiency, security and safety, permitting optimized internal and
external communications to qualify the actual meaning of intelligence.
This issue is followed by analyzing different buildings which already located in order to place the
multiplicity of proposals into perspective.
Keywords: building, design, intelligent, façade, envelope
The word "intelligent" was first used to describe buildings at the beginning of the 1980s, and its use
has been accompanied by the American term "smart", used to imply the same kind of abilities in
materials, structures and buildings. Other synonymous expressions are used, such as “Home System”,
“Smart House” and “Home Automation”. Another expression is “Integrated House”. Many of the
early examples of buildings called "Intelligent Buildings” IB simply represented an attempt to portray
and exploit the prevailing trend for incorporating increasing quantities of information technology into
The idea of the intelligent building has, for many people means the use of information technology and
control system to make the functioning of the building more useful to its occupants, in relation to its
management, or in respect of the building's operational purposes.
The Intelligent Building Institute has proposed: "an intelligent building is one that provides a
productive and cost-effective environment through optimization of its four basic elements - structure,
systems, services and management - and the interrelationships between them. Intelligent buildings
help business owners, property managers and occupants to realize their goals in the areas of cost,
comfort, convenience, safety, long-term flexibility and marketability" [1].
To be able to asses the degree of intelligence we have to look at the specific needs of a specific
building. Today, architects and designers are beginning to look toward developments in new "Smart"
or "Intelligent" materials and technologies for solutions to long-standing problem in building design
Intelligence can be used to improve the performance of the building fabric by making it more capable,
so as to reduce the need for imported energy for heating, cooling, lighting and ventilation. Thus, the
goal is to reduce the operating cost of the building while maintaining the desired environment for the
building occupants [3].
The façade of a building can account for between 15% and 40% of the total building budget, and may
be a significant contributor to the cost of up to 40% more through its impact on the cost of building
So, the builders are giving the customers a chance to maintain a comfortable and healthy home
environment without taking a huge bite out of their monthly budget [4].
Envelope of the building:
The main relevant design feature which affects some interactions of the building with the environment
is the envelope (skin) of the building.
In dealing with the subject of the building’s envelope there are many perspectives of effects such as:
1. Natural ventilation
2. External color of the building envelope.
3. Shading conditions.
4. Properties of building materials.
5. Windows design.
6. Effect of orientation.
Natural ventilation:
In all climatic regions of the world there are times when the outdoor temperature is pleasant and
natural ventilation can be simplest and most effective way to provide indoor comfort [5].
Natural ventilation takes place mostly through windows, so that building design for natural ventilation
means decisions by the designer concerning the location, number, size, orientation and design details
of the windows [5].
The main design features which affect the indoor ventilation conditions are [5]:
1. Type of the building which determines the options of providing effective cross-ventilation.
2. Orientation of buildings and openings to provide good natural ventilation at least one of the
walls of the unit should face the wind.
3. Vertical location of windows: the height of the inlet opening may determine the level of the
main indoor air flow. When openings are near the ceiling, poor ventilation conditions may
exist in the occupied zone unless the flow is directed down ward by the openings details of the
4. Windows types and ways of openings: different types of windows, when serving as inlets,
produce different patterns of indoor air flow and provide different options for controlling the
direction and level of the flow.
5. Subdivision of the interior space, because of the reduction of the flow rate air has to pass
through more than one space. Therefore, if internal subdivision of the building enables
independent cross ventilation of individual spaces, better over all ventilation is produced.
Building skin color:
The color of the external walls and the roof has a tremendous effect on the solar impact on the
building and its indoor climate. Such a solution can be provided by careful design of some building
elements and a selective choice of colors for different parts of the building. With very light colors of
the walls and the roof, the indoor average temperature is close to that of the outdoor average so that of
the small indoor diurnal swing can result in a maximum significantly than the outdoors’ maximum.
However, with dark colors of walls and the roof, the indoor average may be about 4°C-5°C (7.5°F9.5°F) above the outdoors’ average. Also the wind speed mainly affects the external surface
temperature of dark walls or roofs, while the wind speed effect in case of light colored walls is much
As a result from the effect of the envelope color on the walls and the roof affect greatly cooling load
of the building and the need for thermal insulation in hot regions [5].
Shading devices:
It is divided as follows:
• Operable shading devices:
The common feature to all external operable devices is that they can be adjusted to either exclude or
admit solar radiation; they can reduce solar heat gain to about 10% to 15% of the radiation impinging
on the wall while enabling day lighting to enter. Most of the operable shading devices can be applied
with about the same effectiveness to all orientations and to any shape of window. Their color may
have little effect on their thermal performance unless they are completely closed. They also usually
need maintenance to keep them in good condition.
Fixed shading devices:
The different efficiencies of the fixed shading devices, influences how the shape of a window shaded
by such devices affects its thermal performance; There two types of fixed shading devices: horizontal
(over hang) and vertical (fins), they can also be combined in different combinations.
Fixed shading is usually an integral part of the building’s structure. Once built their patterns depend
only on the incident angle of the sun’s rays, but with proper design the over all performance can be
reasonably good. The advantage of fixed shading is that it needs no handling by the occupants and is
also free, it can be considered as an "Intelligent Element".
• Building materials:
The new materials should be used in buildings design to save as much energy as possible. The
materials of which a building, and especially its envelope, is constructed determine the relation ship
between the outdoor temperature and solar radiation conditions, and the indoor temperature (non-air
conditioned buildings) when saving energy consumption is required.
To day, the availability of modern insulating materials, together with application of passive cooling
systems, makes it possible to maintain indoor temperatures significantly below the average of the
outdoor air temperature with such design details it is possible to provide natural comfort through
longer periods of the year and to use less energy.
The thick and heavy structure of the walls and the roof suppress the swing of the external temperature
at a level of the external surfaces of the building’s envelope.
In hot areas traditional buildings are built of high mass, thick walls made of heavy materials, such as
stone, bricks, adobe and mud. Vaulted roofs, with flat impervious external finish covered with earth,
also provide high mass to the building. Windows are usually small and protected from the sun by the
thickness of the wall in which they are placed and may be provided with wooden shutters.
Windows fulfill many functions in building, such as providing visual and auditory contact with the
outdoors, natural ventilation and daylighting. In addition they can serve as elements in passive solar
heating and cooling system[5].
A correctly selected window system can provide the following benefits:
¾ Reduce and eliminate air conditioning cooling load.
¾ A lower lighting load: glazing systems with a high value for this
reduce energy use for lighting.
¾ Superior comfort to occupants.
¾ Less fading damage caused by UV light transmission.
luminous efficacy can
Often, selecting a window is based on only some of the many possible considerations. A decision to
purchase a window can be based on the basic style, operating type, and initial cost. The problem with
this approach is that there is no certainly of performance. Often, the energy-related aspects of the
window are not factored into the decision. While this approach may save money on the initial
purchase, it may result in considerably more expense over the life of the window. Efficient windows
also protect against possible future energy price increases.
Designers must learn new technologies that improve window performance and reduce wasteful
energy use and make "Intelligent Choices".
• Effect of orientation:
When a window is unshaded its quantitative effect on the indoor temperature depends on its
orientation. A study has demonstrated that when window are effectively shaded, by shading devices
exterior to the glazing, their orientation has very little effect on the heat gain of the building and its
indoor temperatures[5]. Therefore, unshaded windows should be avoided as much as possible.
Case studies:
The purpose of the case studies is to describe the range and variations of "Natural Intelligence" (NI)
moderate energy flows through the building’s skin.
employed to
The Environmental Building [14]:
BRE’s Environmental Building provides a model for offices for the 21st century. Innovative and
environmentally advanced, it demonstrates the way for the future based on a platform of new lowenergy targets.
Garston, UK
Operable solar shading and stack ventilation
Fielden Clegg
Project Background:
The building aims not only to provide a working office with low energy consumption in use (fig.1),
but also to serve as a large-scale experimental facility for evaluating various innovative technologies.
The design strategies (fig.2) used in response to the specification
are summarized on the following points:
Fig.1. Office energy consumption data
Fig. 2. Exterior view ¾ Energy consumption.
¾ Quality and comfort of the internal environment.
¾ Day lighting levels.
¾ Performance of glass louvers.
¾ Airflow in stacks and floors.
¾ Air change rates.
¾ Open plan space utilization.
• Stresses in superstructure from construction and
occupancy loads.
• Project Description:
The new Environmental Building at Garston has
been built as a demonstration building for the
conditioning is not used in the new building - the
major energy consumer in many existing office
buildings. Other savings will be made by making
better use of day lighting and by using the
building's 'thermal mass' to moderate temperatures.
The floor plan is divided into open-plan and cellular
offices allowing cross ventilation in the open plan
arrangement while the 4.5-meter-deep cellular offices are
located on the north side with single-sided natural
ventilation (fig.3). A shallow open-office plan is coupled to
a highly glazed façade. A wave-form ceiling structure is
used. At the high point of the wave, a clerestory window
allows daylight to effectively penetrate the space. A duct
providing space conditioning and
ventilation was
Fig. 3. Floor plan with cellular offices on the
north side and open plan on the south side
Fig. 5. Cross section
through the ventilation
stack (left) and the
glazed façade (right).
placed within a hollow core at the low point of the wave-form structure.
• Ventilation and cooling:
A key feature of this building is the integration between natural ventilation and daylighting strategies.
The most striking feature of the building when seen from the south side is the five distinctive
ventilation shafts running up the façade. These form a key part of the energy-saving natural
ventilation and cooling system (fig.4).
Working rather like a greenhouse, the summer sun shines into the glass-fronted shafts,
warming the air inside. This warmed air naturally rises out of the stainless steel
'chimneys' and causes air from inside the building to be drawn through to replace it. On
a breezy day the movement of air across the tops of these chimneys increases this 'stack'
effect. On very warm, still day’s low-energy fans in the tops of the stacks can be turned
on to give greater airflow.
Fig. 4.Ventilation shafts
A stack ventilation system was designed as an alternative ventilation strategy for the
open plan offices during extreme cooling conditions. Vertical chimneys were designed
to draw hot air through the duct in the wave-form structure as well as through bottomhung, hopper, etched windows. The exterior of the stacks are glazed with etched glass
blocks, allowing daylight admission. Low-resistance propeller fans were mounted at the
top-floor level, to provide minimum ventilation and to flush internal heat gains during
the night (fig.5).
This air moving out from
the building draws cooler,
fresh air in from outside
windless days the air is
taken from the shady
north side of the building,
coming in through highlevel
warmer or windy days
(when it’s windy the air
Cross ventilation, windy summer’s day
Stack ventilation, hot still summer’s day
on the north side is not as
Fig. 6. Natural Ventilation
drawn in through passages in the curved hollow concrete floor slabs. Because of its bulk - or thermal mass the concrete cools the incoming air by absorbing heat from it (fig.6).
Additional cooling can be achieved by circulating cold water through the slab. Cold water is drawn
from a 70 meters deep bore hole where the temperature is constantly around 10 Celsius. This is passed
through heat exchangers to chill water that is circulated through under floor pipe work. The borehole
water is returned to the ground via a second,
'wasted'. Overnight, the control systems can open
ventilation paths right through the concrete slab to
cool it further, storing this 'coolness' for the
following day (fig.7). The exposed curved ceiling
gives more surface area than a flat ceiling would,
acting as a cool 'radiator', again providing summer
cooling without energy-consuming air conditioning.
During the winter months the water circulating
through the concrete slab is heated to give gentle
under floor heating. This is supplemented when
necessary by conventional radiators around the
perimeter of the office area. The water is heated by
condensing gas boilers which are 30% more
Fig. 7. Additional cooling can be achieved by circulating cold
water through the slab.
efficient than ordinary boilers, mainly by recovering much of the heat that is usually wasted in the flue
Fig. 8. Occupants can override automatic
settings for lights, louvers, windows and
Solar control and day lighting:
To make maximum use of available daylight the building has a large glass area,
carefully optimized to provide high light levels but low heat losses and solar
gain. To prevent excessive heating and glare from the sun shining in - the 'blinds
down, lights on' situation common in many offices on sunny winter days - the
building has a system of Colt motorized glass louvers on the south façade to
control the daylighting levels (fig.8). Each louver has a translucent ceramic
coating on the underside which obscures the direct sunshine whilst still letting
diffuse light through. During the day the angle of the louvers changes according
to the position of the sun. At times when direct sunshine is not a problem the
louvers are angled to act as 'light shelves' - reflecting light off their smooth
upper surface onto the ceilings of the offices. This reduces the amount of
artificial lighting needed in the parts of the offices furthest from the windows.
Commerzbank Headquarters [15]:
This building is the “world’s first ecological high-rise tower”, which
reflects a modern translation of an ancient concept. This translation provides the
benefits of energy efficiency and high-quality indoor environments, through the
thoughtful use of climate adaptive building skins.
Frankfurt, Germany
double-curtain walls with internal
Norman Foster and Partners
Project Background:
addressed Completion:
1994 - 1997
sustainability issues such as: ambient
energy used as much as possible to reduce the dependency on
fossil fuels by an innovative façade and building system that
uses natural ventilation; maintaining health and happiness for
the users by daylighting office space and allowing them to
enjoy the outdoors by operable windows and planted sky
gardens (fig.9).
• Project Description:
The Tallest building in Europe – 299 meters, 53 floors, steel
framing and green Design.
The Commerzbank building uses a combination of opaque
and transparent double-curtain walls with internal shading to
provide daylighting, natural ventilation, shading in summer,
and passive solar heating in winter. This system is so energy-
Fig. 9.Exterior view Fig. 10.Cross section 137
effective that the building relies solely on natural ventilation and daylighting for about 70% of the
Four-story sky gardens, set at different levels on each of the three sides of the tower, are the visual and
social focus for village-like clusters of offices. The gardens are places to relax and socialize, bringing
richness and humanity to the workplace, and from the outside they give the building a sense of
transparency and lightness (fig.10).
• Daylighting:
Every office in the tower is daylight and has openable windows, resulting in energy consumption
levels equivalent to half those of conventional office towers. Daylight is provided directly to the
outside offices through the windows, and indirectly to offices facing the gardens that are lit both from
the side and above (fig.11).
The climate façade is a custom made double skin envelope system that mediates the weather between
the interior and exterior of the building. It is also this innovative design that allows for individuals to
control their surroundings by operable windows and a sunshade system.
• Virendeel
This invention allows natural ventilation to be viable in a skyscraper. It
is composed of a solid pane of laminated glass on the outer layer, which
deflects strong winds and rain.
of climatic
Fig. 11. Cross section
climatic façade
. façade
• Natural ventilation:
Natural ventilation is achieved across the 16.5m floor plates by
controlled perimeter windows. The 14m high garden facades can also be
opened in good weather to ventilate the atrium space (connected to two
other gardens) and indirectly provide fresh air to the offices facing the
atrium. Outdoor-facing offices are ventilated directly from outside
Fig. 12. Natural Ventilation
The environmental designs are not new but were recently featured in Intelligent Architecture (IA).
Intelligent Skins (IS) is a valuable tool for Architects, Engineers, Surveyors, and Designers, so by
proper design of the Building Skin the size of the energy consumption system can be reduced bringing
down the initial investment, as well as reducing the operating cost the building over its lifetime [5].
This paper is intended to help architects make more informed decisions that will be important to
Intelligent Building’s Future Comfort and Finances. Also, the paper outlines a new methodology
which takes account of the Intelligent Architecture (IA) and illustrates the application of this
methodology to the analysis of a particular project, showing how this might be achieved.
Donald A. Coggan, Smart Buildings,, (last accessed
May 2005).
Michell Addington and Daniel L. Shodeck, Smart Materials and Technologies in Architecture,
Architectural Press, Elsevier, November, 2004
free, (last accessed May 2005).
Built Green, (last accessed May 2005).
Climate Considerations in Building and Urban Design, Baruch Givoni, Van Nostrand Reinholdp,
1998, pp 53, 69, 76,87,88,92, 339.
B. Chandrasekaran and Susan G. Josephson ,Architecture of Intelligence:
The Problems and Current Approaches to Solutions, Laboratory for AI Research, the Ohio State
University Columbus, OH 43210 USA., (last accessed May 2005).
Better Bricks, Climate Adaptive Building Skin, Building Envelope-Daylighting basics,
ing&indextype (last accessed May 2005).
Management, (last accessed May 2005).
Intelligent Buildings, Design and Building Management Systems,,(last accessed April 2005).
Environment,,(last accessed April 2005).
(last accessed May
Michael Wigginton and Judge Harris, Intelligent Skins, 2002.
Passive Solar Design,A source book for Green and Sustainable, (last accessed May 2005).
The Environmental Building,, (last accessed July 2005).
Building, A
21st, (last accessed July 2005).
High-Performance, Commercial Building Facades, Building, (last accessed July 2005).
Commerzbank Headquarters, Frankfurt, EIB for Integral Room Automation, Foster & Partners,
Commerzbank Headquarters, Foster and Partners, Frankfurt, Germany, 1991-1997, (last accessed
August 2005).
Commerzbank Headquarters, Frankfurt, Germany, Michelle Brinkley, February 24, 2005, (last
accessed August 2005). Brinkley
13. Inform me please! - Energy use information to households and Demand Response. Jurek
Pyrko, Energy Sciencees, Lunds University
14. "Consumers, energy and IT. IT and other tools to support efficient energy consumption"
I.C. MariAnne Karlsson
Chalmers University of Technology, Product and Production Development
Publication in the proceedings of the International Scientific Conference at Energitinget, March 1213, 2008 at Alvsjo, Stockholom
Mohammad farrokh Professor jafar mahmoudi
[email protected] [email protected]
TDI, Tehran,Iran; IST Dep., MDH University, Box 883, SE 72123, Västerås, Sweden
Installed DCS system in Jajarm alumina production factory in Iran, help to us for transferred the
signals from equipments to simulators. The simulators can help to operator for control, monitoring
and training. Authors have tried to find a new simulator using ASPEN plusTM software and write a
new protocol for discussed tasks.
Aspen PlusTM by Aspen Technology is one of the major process simulators that arewidely used in
alumina process industries today. It specialises on steady-state analysis.
For implementation, simulation and interpretive of the alumina process flowsheet in jajarm factory
was used from the Aspen Plus simulator. We have tried to be able to provide a list key operating and
design parameters and equipment selection criteria for a limited set of unit operations and processes.
And so for determinanation the economic potential of a process and to identify the operational
variables that affect the economics we have studied the process and flowsheets.
Using the general information of factory, process informations same as temperature, measured
pressure and so PFD, cinitic informations and finallay clin, bauxite and sodium sulphate
inforamations we have design a new simulation method for digestion and control process in Jajarm
factory. The process has implemented on LCR1,PU08, PU10 and PU11 units.
Key words: simulation, boxite, alumina, process
Aluminum is considered to be the most abundant metallic element in the earth’s crust and it’s the third
most common element around. It is not found in its elemental form but always tenaciously held in
many compounds, most of which contain oxygen or silica, or both. The most important aluminous ore
for the manufacture of aluminium is bauxite. It consists of several hydrous aluminium oxide phases
[gibbsite (Al(OH)3), boehmite (AlOOH) and diaspore (Al2O3.H2O)] as well as normal impurities such
as Fe2O3, SiO2 and TiO2. The difference in composition between the minerals lies in different
geographical locations.
Jajarm line production description
Factory process is shown in flowsheet form in the appendices. Extracted bauxite stone from Jajarm
mine transport to stone breacker units by trucks. The output is the less than of 20 mm, around of 0.15
cm. Clin and Na2so3 mixed in rod and ball mill in PU08 unit. Mill outputs were controlled by spiral
classifiers. Less than 90 micron materials exit from classifier and send to PU09 unit, for removing of
silice. Greater than of 90 micron were milled again. In PU10 using high pressure pumps, slurry
pressure reach to 100bar and then send to PU11, digestion unit. Slurry temperature reaches to 275
centigrade degree and then sends to piped reactors and digesters. After silica removing in
PU12, liquid with aluminium, for phase separation of liquid from solid, transmit to thickning of
separator unit (PU13). Overflow thickener that is a soluble with alumina, after filtrization transmit to
soluble cooling unit, PU17A. Underflow output of tinker is called red flower. After many steps
whashing, in PU14 unit after sulphate sodium refinding, concentrated and finally send to the waste
dam. Transmitted sluable to cooling unit, vent to the PU16 unit after cooling. This unit done the
precipitation in 2 steps, and finally create the aluminum hydrate crystals. After filteration on liqur with
aluminium in PU17, created the aluminium hydrate cacke Al2O3.3H2O and send to the hydrate store,
PU19. Produced hydrate, transmit to PU21 unit (calcinations unit) for producing of AL2O3, alumina
powder. Final product transmits to PU15 unit for storage.
In table 1 main parameters of process design compared in 3 cases:
reported parameters by first designer (TEX)
reported parameters before changing by NFC
reported parameters after changing by NFC
Table 1: main parameters of process
Al2O3 content in bauxite
Silica ratio of bauxite
Caustic concentration Na2Ok
175 g/l
171 g/l
190 g/l
αk of digested liquor
N/S of digested red mud
0.32 w/w
0.36 w/w
0.34 w/w
A/S of digested red mud
1.2 w/w
1.21 w/w
1.24 w/w
Digestion temperature
310 C
275 C
275 C
Addition of Lime
LOI of digested red mud
Diluted concentration Na2Ok
130 g/l
133 g/l
140 g/l
pregnant 1.58
αk in spent liquor
Soluble soda loss in mud 3.1 kg/t-dry mud kg/t-dry mud kg/t-dry mud
Moisture of red mud cake
Main flows rates of factory are shown for production of 1 ton alumind in 3 cases.
Table2:main flow rates for production of 1 ton alumina
Name of flow
Bauxite (dry)
2.844 2.919 2.915
to m
digestion (PU11)
0.673 0.37
12.23 13.00 10.88
water t
from PU11
3.640 2.883 2.342
after m3
diluted (PU12)
19.30 20.67 18.45
Red mud
wash t
to t
Pregnant Liquor
wash t
5.705 5.270
2.235 2.227 2.200
15.07 15.44 13.55
Evaporated water t
in PU18
2.313 2.035 2.188
Caustic make up t
50% NaOH
0.281 0.328 0.307
Alumina product
1.000 1.000 1.000
Process Description
The process is based on the Gladstone Alumina refinery. Shown in figure1 is the block diagram of the
Figure 1: Block Diagram of Alumina Manufacturing Process
Bauxite Preparation
The bauxite entering the refinery must be uniform and sufficiently fine so that the extraction of the
Al2O3 and other operations are successful. In most modern plants, the bauxite is mixed with a portion
of the process solution and is ground as slurry to prevent a dusty working environment. Rod mills and
ball mills are used most frequently, for mill of stone to less than 20 mm. The ground slurry is then
passed over screens or through cyclones, with the finer particles (less than 0.15 cm) progressing and
the course ones being returned to mills.
An innovation recently adopted is to hold ground slurry for several hours in agitated tanks. These
tanks are operated so that plant feed is uniformly blended. Sometimes, bauxites are dried to improve
handling or washed to remove clay.
In digestion, all of the Al2O3 in the bauxite must be extracted. It works by the principle that gibbsite
and boehmite are soluble in sodium hydroxide (NaOH), the solubility being temperature,
concentration of NaOH and bauxite characteristic dependent.
Most of the impurities in bauxite are not soluble in NaOH, except reactive silica (SiO2), which forms
a quite insoluble compound during processing. The following describes the two important reactions
occurring during this stage.
Reaction 1 – Digestion
AlOH NaOH  A
 lOH Na
AlOOH NaOH H O
 l OH Na
It was found that the solubility of gibbsite is higher than boehmite; hence the conditions for digesting
boehmitic bauxite must be more severe. The temperature and/or NaOH concentration must be
increased. For diaspore bauxites, CaO is charged to allow digestions at lower temperature.
It was found that the solubility of gibbsite is higher than boehmite; hence the conditions for digesting
boehmitic bauxite must be more severe. The temperature and/or NaOH concentration must be
increased. For diaspore bauxites, CaO is charged to allow digestions at lower temperature.
For this project, a ‘sweetening process’, whereby the Al2O3 concentration is kept low enough to
dissolve is introduced as the monohydrate grade bauxite contains a mixture of gibbsite and boehmite.
The monohydrate grade is digested under relatively mild conditions, producing an intermediate Al2O3
concentration. In a separate vessel, the trihydrate grade is added to the flow from the first digest to
raise the Al2O3 concentration to the desired level. This allows moderate digest conditions.
Reaction 2 – Causticization
CaO H2 O
CaOH Na2 CO3  CaCO3 +2NaOH
The reason for this reaction is primarily to make up the NaOH losses. Lime (CaO) is added to the
digest. Na2CO3 is formed by decomposition in the digest of organic material from the bauxite and
CO2 adsorbed into process solutions from the atmosphere.
The next major step in the process is to separate the solid residue from the sodium aluminate solution.
This separation must be as complete as possible as any remaining solids will cause contamination to
the product.
The slurry is fed into the center of thickening tanks (also called settlers), each 40 m in diameter. The
velocity of the slurry becomes very slow as the solution flows radially across the thickener. The
solids, having a higher specific gravity than the solution, sink to the bottom of the thickener. The fine
solids behave as a relatively stable colloidal suspension, hence they settle slowly if not treated further.
Flocculants are added to improve the clarity of the thickener overflow, the solids content of the
underflow, as well as the settling rate of the solids.
The particle size distribution of the residue solids is usually bimodal. The term sand is used to
describe coarse fraction over 100 μm in diameter whereas the rest of the solids are finer than 10 μm.
The settler underflow first has sand removed in sand traps, which is recycled to bauxite grinding. With
the residue concentrated in the thickener underflow, the next task is to wash it with fresh water so that
it can be discarded within environmental standards. This is done in countercurrent washing thickeners
similar in design to those used for settling. Fresh water is use to recover the soda and alumina content
in the residue before being pumped to large disposal dams. This result in minimising the total costs of
the soluble salts lost due to incomplete washing and the cost of evaporating the dilution introduced.
The final stage of clarification is polish filtration of the overflow settler solution. This stage is to
remove the few particles of solids remaining to protect product purity. Heavy cotton cloth provided
the early filter medium, but it was easily damaged and slowly attacked by the caustic solution.
Polypropylene is the current fabric choice and is unaffected by process conditions. The plant uses
Kelly-type constant pressure filters.
On occasions, small quantities of Al(OH)3 or DSP will partially blind the cloth. Hence, lime is added
to produce a filter aid – tricalcium aluminate (3CaO.Al2O3.6H2O) to promote the formation of a
porous, rigid filter cake.
With all the solids removed, the liquor leaving the filter area contains alumina in clear supersaturated
solution. This filtered solution must be cooled before precipitation. This is done by flash evaporation
whereby a steam jet pump is used to remove noncondensable gases from the system to create vacuum.
Steam is given off and is used to heat spent liquor (from classifiers) returning to digestion.
Precipitation of crystals is used to recover dissolved alumina from the liquor. The reaction is the
reverse of the first digestion reaction:
Al OH Na A
 l (OH) 3 +OHThe above reaction is done in rows of precipitation tanks that are seeded with Al(OH)3 to promote
crystal growth.
The first objective in precipitation is to produce Al(OH)3 that, when calcined, meets the product
specification. The second objective is to obtain high yield from each volume of the solution. Care
must be taken to achieve this as the number of new particles generated in precipitation must equal the
number of particles leaving as product for the system to remain balanced. This requires balancing
nucleation, agglomeration, growth, and particle breakage, so a combination of science and art has
developed. In the plant, the precipitation tank is agitated, with a holding time each of about 3 hours.
During the 25 – 30 hours pass through precipitation, alumina of various crystal sizes is produced. The
entry temperature and the temperature across the row, seed rate and caustic concentration are control
variables used to achieve the required particle size distribution in the product. Consequently the
process moves slowly from being too fine to too coarse and back again. The aim is to minimise the
frequency and amplitude of the changes.
Slurry leaving the precipitators is led to three stage ‘gravity’ classification tanks to be separated into
three size ranges. The primary classifiers collect the coarse fraction, which becomes the product. The
intermediate and fine crystals from the secondary and tertiary classifiers are washed and returned to
the precipitation tanks as seed.
The spent caustic liquor, which is essentially free from solids, from the overflow of the tertiary
classifier is returned through an evaporation stage. This liquor will be heated and reconcentrated, to be
recycled back to dissolve more alumina in the digesters. Fresh caustic soda is added to the stream to
make up for process losses.
The final operation in production of alumina is calcination. The temperature of Al(OH)3 is raised
above 1380 K resulting in the reaction:
2AlOH3  A
 l2 O3  H2 O
The product from the primary classifier is washed on horizontal-table vacuum filters to remove
process liquor. The quality of the wash water is of some concern because such impurities as calcium
and magnesium can be adsorbed on the surface of the product.
The resulting filter cake is then fed to a series of calcining units to remove both free moisture and
chemically-combined water. The calcining units are made up of rotary kilns each 100 m long and 4 m
in diameter. They are mounted on bearings and rotate about an axis inclined at a small angle to the
horizontal. Damp Al(OH)3 from the filters enters the upper end and slowly tumbles toward the lower
end, travelling against a stream of hot gas formed by the combustion of natural gas at the discharge
Cooling the alumina is first carried out in rotary or satellite coolers, which preheat the secondary
combustion air for kilns; and then in fluidised-bed coolers for further cooling.
The final product – pure alumina is discharged on to conveyor belts, which carry it to storage
buildings where it is stockpiled for shipment.
For this project, the composition of the bauxite used is as follows (table 3):
Table 3: Composition of Bauxite [7]
Trihydrate Grade (wt %)
Monohydrate Grade (wt %)
Available Al2O3
Other (mainly H2O)
For the monohydrate grade, of the available Al2O3, 40% is from gibbsite and 10% is from boehmite.
Whereas for the trihydrate grade, all available Al2O3 consists of gibbsite.
Aspen model diagram
The simulation is based on Gladstone Alumina refinery with the following feed rate:
Table 3: Amount of feed rate
Monohydrate grade 725
grade 90
The product will be a smelting-grade alumina with the following specification:
Table 4: Smelting grade alumina
Normal range
Particle size
10 – 200 μm
0.01 – 0.04 wt%
0.01 – 0.03 wt%
0.02 – 0.08 wt%
0.002 – 0.005 wt%
Many of the simulated units are simplified. For example, only one flash unit is being simulated while
realistically, there should be a series of flash vessels. Assumptions and basis also has to be included
for the simulation. Below gives the description, assumptions and basis for each block.
Figure 2: Aspen model block diagram for manufacture of Alumina in the bottom and PFD model of
PU11 unit in down
Digestion (B1 to B4)
B1 (Mixer)
This is used to simulate part of the bauxite preparation whereby the raw feed is mixed together with
the caustic and slaked lime to form slurry.
 Assumptions – the bauxite has been pre-grinded in compartment rod/ball mills to allow better solid
and liquid contact during digestion.
B2 (Rstoic)
This block simulates the digestion units. The units are operated at a temperature of 573K and a
pressure of 3500kPa.
Assumptions – only the following reactions takes place:
1. 2NaOH + Al2 O3
2. CO2  NaOH
NaAlO2 H2 O
Na2 CO3 H2 O
 igestion – this reaction has been described earlier. For this simulation, it is assumed that 97% of
Al2O3 has been converted to NaAlO2.
 arbonation – this reaction occurs naturally with the carbon dioxide from the atmosphere. This
reduces the effectiveness of the liquor (caustic liquor) to dissolve alumina.
B3 (Mixer)
This allows the addition of Ca(OH)2 which will be used for the later reactions in the digesters.
B4 (Rstoic)
This block again, simulates the digestion units whereby only causticization reaction occurs.
1. Na2 CO3+Ca(OH) 2
CaCO3  2NaOH  Causticization
Caustization – as mentioned previously, the above reaction is to regenerate NaOH and CaCO3 will
be removed later. It is assumed that all Na2CO3 has been reacted.
B5 (Flash2)
Flashing units are simulated in this block. Flash tanks are used for heat recovery purposes.
The simulation is done with an outlet temperature of 383K and outlet pressure of 250kPa.
Clarification (B6 to B8)
B6 (Sep)
This block is used to simulate settlers.
All unreacted raw feed (Al2O3, Fe2O3, SiO2 and TiO2 and CaCO3) is considered as solids.
This simulation is run with an underflow containing 25% solids
B7 (Sep)
 his block is used to simulate Kelly-type constant pressure filters whereby all remaining solids are
separated and the liquid leaving the filter area contains alumina in a clear supersaturated liquid.
B8 (Mixer)
 his simulates the washer units. The underflow from the previous settler is mixed together with
water from the heat recovery sections. The product is called mudwaste.
Precipitation (B9 to B11)
B9 (Flash2)
 he resultant supersaturated liquid needs to be cooled before precipitation. This is done in vacuum
flash vessels, which is simulated by this block.
The target outlet temperature is 345K.
B10 (Rstoic)
This block is used to simulate precipitators. The following ‘crystallization’ reaction takes place:
1. 2NaAlO2 H2 O  Al2 O3 +2NaOH
 Precipitation
It is assumed that all NaAlO2 has been crystallized to Al2O3.
B11 (Sep)
 lurry leaving the precipitators is being separated in 3 classifiers, namely, primary, secondary and
tertiary. But for simplicity, 1 single unit is being simulated.
It is simulated on an overflow concentration of about 3 g/L of alumina.
The resultant overflow liquid is termed as spent liquor and is heated before being recycled back to
the digesters.
Calcination (B12 and B13)
B12 (Heater)
Calcination equipment (rotary kilns) is simulated for this block. The underflow from the primary
classifier (B10) is heated to temperatures above 1373K to remove water.
B13 (Heater)
The product is then cooled in a series of equipment (initially in rotary coolers and then, in
fluidised-bed coolers) to room temperature.
Heat recovery system (B14 and B15)
B14 (Mixer)
This is used to simulate a mixer whereby fresh NaOH solution is added to the caustic liquor
recycled from the classifier (B10).
B15 (Heatx)
The flash heat generated from flasher units (B5) is used to preheat incoming caustic liquor in
tubular heat exchangers. Condensate from these heat exchangers is used to wash the mudwaste
Mass and Energy Balance
A Process Flow Diagram has been done for the above process. The following abbreviations are used
for the equipment:
CR – Crusher; V – Vessels; R – Digesters (Reactors); S – Settlers and Clarifiers; F – Filter; HX –
Heat exchangers; PR – Precipitator; CL – Calcine; M – Mixer; C - Cooler
The critical operating unit for the process is the temperature and pressure of the digester units. As
different operating values will have different conversions for Al2O3, this will affect the final yield of
the product.
From the simulation, another important variable is the flash temperatures and pressures.
At the moment, the flash temperature and pressure used are 383K and 250kPa respectively. A
change in either of the variables will affect the composition of the vapour and the liquid streams,
namely, the water and NaAlO2 concentrations. It would be ideal to have as low concentration of
NaAlO2 at the vapor stream. When increasing the temperature, it was found that water
concentration has been increased in the vapour stream, but NaAlO2 concentration did not changed
much. An increase in pressure will however result in both concentration decreases in the liquid
stream. Classifier’s overflow composition can also be changed to control the amount of NaOH
return to the digesters. This will in turn affect the amount of NaOH recycled back to the digesters
for reaction.
It has been shown that it is possible to simulate processes using Aspen software. To be as realistic
as possible, operating values for each unit is required. Otherwise, heuristic assumptions have to be
made. Process has been simplified and scenarios were introduced into the implementation protocol
and simulator.
Many pahases, robust informations bank and general equations help to desiner for cover the all of
systems. This software can help to us for simulation of liquors with solid included, electrolyte and
more, these capability are the major competences on HYSYS software an so others. Using software
we have designed the process flow diagram, and so with choose the inputs in software get the
outputs and in the first view it have been shown the good accuracy, between measured and
simulated profiles.
Recommendations for further work on the above are as follows:
To simulate the above process as a solids handling process. This will then include the pretreatment process of crushing and particle size distribution (PSD) will have to be specified. This
will make the simulation more realistic.
Impurities such as Na2O should be considered.
[1]. Caroline Crosthwaite, 1E202 Process systems analysis study guide, Department of Chemical
Engineering UQ, 2000, pages 4 and 5.
[2]. Lyle F.Albright, Processes for Major Addition-Type Plastics and their Monomers, McGraw
Hill, 1974, pages 177, 191, 213 table 6-1.
[3]. Warren et al., Process Design Principles – Synthesis, Analysis, and Evaluation, Wiley & Sons,
1999, pages 45 to 55.
[4]. John J. McKetta, Encyclopedia of Chemical Processing and Design, Vol 3, Marcel Dekker Inc.
1976, pages 57, 58
[5]. Wolfgang et al., Ulmann's Encyclopedia of Industrial Chemistry, Vol A1, 5th edition, VCH
Publishing, 1985, pages 568, 570, 572, 574, 579.
[6]. A.R. Burkin, Production of Aluminium and Alumina, Wiley & Sons, 1987, pages 15, 16, 19,
21, 25 to 27, 33, 37.
[7]. Queensland Alumina Limited Company brochure – The Gladstone Alumina Refinery,
Queensland Alumina Ltd
[8]. Stanley M.Walas, Chemical Process Equipment – Selection and Design, Butterworth, 1990,
page 320 table 11.9.
[9]. Nist Chemistry Webbook,
[10]. Engineering and Applied Science Knowledge,
[11]. Aspen PlusTMVersion 10.1 help file.
Degress of Freedom Analysis
Number of Components, n = 12 (Al2O3, SiO2, FeO2, TiO2, NaOH, CO2, CaO, Ca(OH)2, NaAlO2,
CaCO3, Na2CO3, H2O)
Number of reactions in B2, r = 2 while B4 and B10, r = 1
B1: Mixer = n(c + 2) = 5c + 10
B2: Rstoic = c + 4 + r = c + 6
B3: Mixer = n(c + 2) = 2c + 4
B4: Rstoic = c + 4 + r = c + 5
B5: Flash = c + 4
B6: Sep = 2c + 2
B7: Sep = 2c + 2
B8: Mixer = n(c + 2) = 3c + 6
B9: Flash = c + 4
B10: Rstoic = c + 4 + r = c + 6
B11: Sep = 2c + 2
B12: Heater = c + 4
B13: Heater = c + 4
B14: Mixer = n(c + 2) = 2c + 4
B15: Heatx = 2c + 7
Therefore, total = 27c + 69
Interconnecting streams = 21(c + 2) = 21c + 42, Hence, remaining = (27c + 69) – (21c + 42) = 99
Feed 1 (Monohydrate) = 9 (Pressure, Temperature, all 0% except Al2O3, SiO2, FeO2, TiO2 and
H2O (Composition))
Feed 2 (Trihydrate) = 9 (Pressure, Temperature, all 0% except Al2O3, SiO2, FeO2, TiO2 and H2O
Feed 3 (CaO) = 13 (Pressure, Temperature, all 0% except CaO)
Feed 4 (CO2) = 13 (Pressure, Temperature, all 0% except CO2)
Feed 5 (Ca(OH)2) = 13 (Pressure, Temperature, all 0% except Ca(OH)2)
Feed 6 (NaOH) = 13 (Pressure, Temperature, all 0% except NaOH)
B2 (Rstoic) = 4 (Pressure, Temperature, % conversion of both reactions)
B4 (Rstoic) = 3 (Pressure, Temperature, % conversion)
B5 (Flash) = 2 (Pressure, Temperature)
B6 (Sep) = 5 (Bottom composition - 25 % of Al2O3, SiO2, FeO2, TiO2 and CaCO3)
B7 (Sep) = 4 (Bottom composition – all Al2O3, SiO2, FeO2, TiO2)
B9 (Flash) = 2 (Pressure, Temperature)
B10 (Rstoic) = 3 (Pressure, Temperature, % conversion)
B11 (Sep) = 1 (Top composition – 3 g/L of Al2O3)
B12 (Heater) = 2 (Pressure, Temperature)
B13 (Heater) = 2 (Pressure, Temperature)
B15 (Heatx) = 1 (No vapor in the cold stream)
Total = 99
Degrees of Freedom = Remaining – Specifications: = 99 – 99= 0
Therefore, the model is solvable.
Publication in the proceedings of the International Scientific Conference at Energitinget, March 1213, 2008 at Alvsjo, Stockholom
Hassan gholinejad1 Kourosh Mousavi Takami2 Jafar Mahmoudi3
TDI Tehran, Iran and IST dep., Mälardalen University, Box 883, SE721 23, Västerås,Sweden
1. Ph.D. student in MDH 2. TDI researcher and Ph.D. student in MDH 3. Professor in MDH
[email protected]
[email protected]
[email protected]
Power transformer is a complex and critical component of the power transmission and
distribution system. System abnormalities, loading, switching and ambient condition normally
contributes towards accelerated aging and sudden failure. In the absence of critical components
monitoring, the failure risk is always high. For early fault detection and real time condition
assessment, online monitoring system in accordance with age and conditions of the asset would
be an important tool.
This paper presented an algorithm on the estimation of the hot spot temperature within a
transformer, addressing some issues concerning the accuracy of their results. Included are
summaries of articles on modelling techniques and the reasons offered by their authors regarding
their effectiveness, focussing on their advantages and disadvantages. The paper also engages in
some discussion on the current use of Artificial Intelligence (AI) systems on the dissolved gas
analysis (DGA) results of transformer oils, and how similar AI systems are being adopted to
estimate hotspot temperatures. Finally, some mention is made of the application of fibre
temperature probes that have been recently developed. Such technology has been valuable to
corroborate results and secure the validity of works done in this area. But there are limitations on
their application, and these are discussed, highlighting the economic and safety implications.
Insulation is the major component, which plays an important role in the life expectancy of the
transformer. Transformer life known to us is based on the designed parameter with respect to
normal operation and climate conditions. To determine the performance and aging of the asset,
insulation behavior is a main indicator. Most of the transformers in a system, around the world are
exceeding their designed life. In the absence of insulation assessment, good number of transformer
failed due to insulation problems, before reaching to their designed technical life. It is important to
investigate the cause(s) of the insulation degradation with respect to age.
Average age of the transformers that failed due to insulation deterioration during the last ten years
was 17.8 years. A good number of aged transformers are still performing well, it is vital to monitor
the insulation behavior rather then replacing with new one. Transformer insulation behavior is
different with respect to operation mode, climate (ambient condition) and frequency of subjected
faults. Load growth has influence on the insulation degradation. The insulation degradation trend
needs regular assessment. An accurate analysis of the insulation can suggest operating condition,
de-rating of the transformer will increase the life expectancy. The unit can be proposed for
relocation, subjected to less stress. Cost effective maintenance strategies can be developed.
Insulation aging in transformer is a complex and irreversible phenomena. To ensure higher
reliability and safety, insulation condition monitoring and trend analysis are of major concern.
Insulation trend analysis ill conclude type of failure as well as severity of the fault. This will make
easy to understand type of maintenance required, loading constraints and future management
The analysis will predict the life expectancy of the asset. It is significant to recommend insulation
assessment for the aged and suspicious behaviour transformers. With perfect condition monitoring
the rate of aging can be reduced. Online insulation condition monitoring, proper diagnostics and
accurate interpretation/ analysis will provide realistic decision for economical operation and cost
effective maintenance strategies.
The overall integrity of the asset can be assessed, with minimum risk of sudden failure. The
environmental risk can be reduced. Effect of aging rate on the life expectancy can be established.
Condition monitoring provides information on the developing insulation problems and incipient
faults. Thus early warning of any abnormality can avert the catastrophic failure.
Put into perspective, power transformers are probably the single most expensive asset within an
electrical transmission and distribution network [1], and that alone justifies the requirement of
developing appropriate reliable systems ensuring their availability and reliability.
They are designed with a nominal life expectancy in years, evaluated for a rated load under ideal
conditions as indicated by their nameplate. Excluding unusual circumstances unrelated to the
normal functioning of a power transformer, failure is expected to occur as the windings’ paper
insulation erodes past its limit to sustain appropriate structural and electrical stresses. The
degradation is cumulative and adversely affected by the presence of both heat and oil contaminants,
including gases and water [2].
Limited excessive loading beyond the transformer’s nameplate rating is permissible and often
experienced during emergency conditions, but the trade-off is an accelerated demise of the
transformer towards an end-of-life condition.
In Australia, the privatisation of the electricity utilities has met with increasing pressure by
investors and shareholders to maximise Return-On-Investment (ROI). Utilities in the electricity
industry must now re-examine operating strategies on long term, back up and failure procedures
balancing the demands for both reliable and economic performance. Obviously, under utilised
equipment, especially at the size of expense of power transformers, is undesirable. Ultimately the
target will be to set transformers to run at “Full Warp Drive” as the normal mode of operation.
However, the techniques used to predict a transformer’s loading limits are still too primitive to
provide any guarantees. Any inadvertent abuse, where the transformer’s temperature thresholds are
exceeded will compromise the performance of the transformer, a result caused by the application of
stresses beyond the anticipated design limits, and possibly generating undetected incipient faults.
Albeit the transformer may survive, avoiding premature failure, there will remain a high level of
doubt concerning its state of reliability.
An obvious resolution to this matter lies in the ability to accurately predict the safe limitations of
power transformers and the associated loss-of-life trade-off. Such calculations will enable
operational strategies that address economic and reliability issues without exceeding those
thresholds that result in transformer damage or failure.
Since 1885 transformers (0.15 MVA) are serving the power industry and are being produced with
higher rating (> 2000MVA). Majority of transformer population is serving in many of the
transmission and distribution utilities are 20 to 40 years old. As an example the installed power
transformer (United States) capacity has reduced from 185 GVA (Giga Volt Amperes) to 50 GVA
per year over the past twenty-five years.
The average load growth rate observed is approximately 2%. Transformer utilization has increased
by 22% on average, causing oil hot spot temperature to increase by approximately a 48%, at normal
peak load.
Due to gradual increase in the temperature, peak load insulation life will be reduced by a factor of
approximately 8. Economic pressures and factors such as an increasing proportion of aged power
transformers are combining to dictate more efficient plant maintenance management. Life
assessment is becoming increasingly important as the average age of the asset increases, due to
economic pressures and a relatively low load growth, with fewer major re-development projects.
A scientific remnant life assessment would be an important tool towards higher reliability of the
system and asset management. After determining the critical indicator responsible for aging as well
as asset technical assessment, the rate of ageing can be reduced by implementing the correct
operational and maintenance strategies. The early and failures due to aging can be effectively
minimized. Better asset management system can be implemented (timely relocation/ replacement
can be planned). The transformer's condition assessment can be broken down into the following
areas of concern:
Operating performance to design criteria
Aging of insulating materials due to stress imposed both thermal and electrical.
Chemical deterioration from moisture, oxygen and acidity and other contaminations.
Mechanical strength of the solid insulating and bracing materials.
Transformer assessment is mainly based on dissolved gas analysis (DGA), in particular Rogers
Ratio method.
C2H2/C2H4, CH4/H2 & C2H4/C2H6
The important issue is trending of combustible gases.
(S T
− S O ) × V × 10 −6
7.5 × T
R = Rate (ft3/day)
SO = First sample (ppm)
ST = Second sample (ppm)
V = Tank oil volume
T = Time (days)
The other important parameter is rise in winding and top oil temperature. The analysis outcome
mainly depends on the test accuracy and correct interpretation.
Core, windings, insulation oil, bushing and on-load tap changer are the main active parts of the
transformer insulation chain. The degradation of insulation systems is accompanied by phenomenon
of changing physical parameters or the behavior of insulation systems. The degradation of
insulation systems is a complex physical process. Many parameters act at the same time thus
making the interpretation extremely difficul.
The aging process in the oil/cellulose insulation system under thermal stress and their measurable
effects are due to chemical reactions in the dielectric. The temperature of the oil/paper dielectric is
the critical aging parameter to cause enough change in the mechanical and electrical properties of
the material. Apart from high temperatures, other important parameters affecting the aging of the
solid and liquid insulation include the presence of water and oxygen in the system [6, 7].
The monitoring and assessment of such components is vital to achieve better reliability of the
system. By implementing correct operational and maintenance strategies the insulation aging/
degradation process can be controlled and the asset life can be extended effectively. Asset’s critical
component monitoring (strict) is required for the technical assessment (normal to end of life) to
ensure economical and safe operation. Also better asset management polices can be implemented.
The types of failures which may occur within the transformer tank are many, one with serious
concerns to identify the effectiveness of diagnostics tests and various condition monitoring
techniques are listed below, The types of failures which may occur within the transformer tank are
many, one with serious concerns to identify the effectiveness of diagnostics tests and various
condition monitoring techniques are listed:
• Overheating breakdown
Overheating breakdown in core plate insulation leads to circulating currents and usually sparking at
the fault.
• Core bolt and Core clamping structure
Breakdown of insulation between parts of clamping structure results in circulating current, possibly
sparking. Breakdown of insulation between core and core clamps and the tank leads to spark (fault).
• Windings and inter-winding insulation
Overheating due to poor joints is a common fault in any part of the electrical circuit. Breakdown of
inter-strand insulation results in circulating current causing overheating of insulation and hot spots
at point of fault. This can be a result of winding movement. A turn to turn fault produces a similar
effect but with much more energy and can usually be detected and identified, whereas there is
currently no diagnostic test to identify a strand to strand fault.
Partial discharge faults can develop between various parts of the insulation structure as a result of
contamination (including moisture) or due to poor impregnation or overstressing.
Overheating of stress shields results in breakdown and circulating current. A fault between
windings - usually results in serious damage and a fault from line to ground also usually results in
serious damage.
• Tank, flux shields, and fittings
The breakdown of insulation between portions of the tank shields or between the shields and tank
can lead to circulating current, this will be a function of load current. Partial discharges may
emanate from the ground potential surfaces of the tank and parts mounted on the tank. Circulating
current in the tank due to proximity of heavy current conductors can produce hot spots in the tank
and across gasket joints.
• Deterioration and Failure Factors
The factors responsible for failures and accelerated deterioration are categorized as:
• Operating environment (electrical)
Operating voltage (50/60 Hz), transient over-voltages,
load current, short circuits (fault currents), lightening and switching surges.
• Operating environment (physical)
Temperature (operating full load with high ambient temperature-humidity index), wind, rain,
seismic and pollution.
• Operating time
Time in service and time under abnormal conditions or extreme condition (Load variation, change
in thermal stresses).
• Number of operations of tap changer
Number of on-load tap changer operation.
• Vibration effect
Sound and material fatigue.
Moisture (water content in oil), presence of oxygen and particles in oil.
A transformer’s loading capacity is related to the exposure of its insulation to heat, the highest
temperature of which is referred to as the Hot Spot Temperature (HST). The HST’s effect on the
windings’ paper insulation is used to quantify the limit of its temperature range over a calculated
period of time. The problem, however, is that these limits are dynamic due to the changing
transformer characteristics, and varying ambient climate conditions.
Development of an accurate method of estimating the dynamic loading limitations, and
consumption of service life of a transformer, has economical and operational benefits.
Economically, deferral of expenditure on expensive plant by “driving them harder”, has always
been recognised as a wise business strategy. Conversely, but equally wise, being able to correlate
the cost of depreciation on a transformer with usage in kilowatt hours (kWhs) allows for a more
financially equitable method for assessing asset performance, thereby enabling the calculation of
the most appropriate time for retiring older equipment. It also assists in planning for network
growth and reducing expensive risks.
Similarly, strategic maintenance and operational procedures are best formulated where the
performance of existing plant has been accurately assessed. Unexpected failures are rarely
welcome, especially with increasing pressure to minimise backup resources such as redundant
transformer capacity.
Over recent years, condition monitoring of power transformers has become a far more affordable
exercise; in actual fact, the true cost of not having adequate online monitoring is arguably much
greater, especially if an avoidable incident should occur resulting in damage and injury. Several
companies have already developed monitoring systems with simple algorithms that activate alarms
and protection devices when monitored inputs have exceeded static thresholds. These thresholds are
unquestionably conservative, perhaps through lack of confidence, which intuitively reduces any risk
of damage or failure, but also under utilises the transformers’ true performance.
Several Standards have been developed to estimate the temperature of hotspots within a transformer
as a function of current load and ambient climate conditions. Many electricity companies have used
these as the basis of their operating procedures, developing tables and charts that highlight their
exposure to risks within their network. As previously mentioned, there is often a policy among
decision makers in this field to bias all errors on the side of caution. In this paper are presented two
such Standards: IEC 354:1991 and IEEE Std. C57.91-1995.
5.1 IEC standard
IEC 354:1991
“IEC: Loading guide for oil-immersed power transformers”
The IEC Standard [3] provides a series of simplified equations that describe a mathematical model
for the calculation of operating temperatures in a transformer. The assumptions listed in these
standards include: a linear temperature rise in the oil from the bottom of the tank to the top, a
parallel temperature rise in the windings, and an allowance for stray losses that is used to assess the
As these Standard points out, for large power transformers, the results for hotspot temperatures
(based on temperature rise tests) may not be valid due to the significance and complexity of the
contribution of flux leakage to the heating of the windings. Therefore, this method has a limited use,
restricted at or below the transformers rated capacity.
A further note in this Standard adds that corrections to account for load losses and oil viscosity, can
be dismissed as either insignificant, or that the effects cancel each other.
5.2 IEEE standard
IEEE Std C57.91-1995
“IEEE Guide for Loading Mineral-Oil-Immersed Transformers”
In the IEEE [2] Standard there are two methods of calculating the HST. The first method, as with
the IEC Standard, is a mathematical model. In the calculation there is the assumption that the oil
temperature across the windings is equal to that of the top oil temperature. However, at the
commencement of the Standard there is an admission on the validity of this assumption. After
measurements were taken with recently available direct thermocouple and fibre optic devices, it was
revealed to be incorrect. A paper written by Lesieutre et al. [4] also points out that the method does
not adequately account for variations in the ambient temperature for which they [4] have suggested
a modification and claimed to have verified.
The second method in the IEEE Standard (Annex G) takes into account these observations and
attempts to correct them, including thermal effects and liquid viscosity during overload conditions.
While Mathematical Models, as those represented in IEEE [2] and IEC [3] Standards are based on
the simplified observation of the exponential cooling of lumped heated bodies, others such as Swift
et al. [5] have suggested that a thermodynamic and heat transfer approach would be more
They explain that by incorporating the significance of the heat transfers between mediums such as
the transformer oil and transformer tank-wall, a more practical and accurate estimation is probable.
But even this model suggests the need for accurate parameter estimation and the assumption that the
transformer is operating in an environment that supports nominal operation.
Owing to the complexity and variations in transformer design and operation, the expectation that a
reliable and accurate “one size fits all” mathematical model is possible, is questionable. In any case,
isolated experiments conducted using this method have met with the requirement for precision that
is impractical, as discovered by Tylavski et al. [6]. Ultimately, such models are not capable of
identifying and adjusting for the effects of aging, inaccurate inputs, and incipient faults. The
identification of these difficulties has ignited particular interest in two categories of AI: Fuzzy
Logic and Neural Networks.
As a research tool, the Fuzzy Inference System (FIS) has been applied to a variety of similar
projects where there has been a need to address the issue of inaccurate information and also apply
heuristic reasoning [7].
Similarly, Artificial Neural Networks (ANN) has found implication where dynamic parameters are
difficult to define mathematically [7] where training of the model can recognise trends and the
dynamic nature of some of the parameters that might once had been believed to be constant.
Fibre Optic Temperature Probes are the obvious new technology, hailed as the easy, cost effective
remedy for determining transformer HSTs. But seen objectively, the ultimate task of transformers is
to provide reliable, efficient operation at the lowest possible cost. At the bare minimum, a
transformer is little more than an iron core, wrapped with windings, bathed in a tank of oil.
Anything else adds complexity and cost through design, maintenance and supervision. It is from
this premise that the addition of measuring equipment may not always be the most advisable
Also, whilst for new transformers the task of embedding fibre optic cable into its windings is a
seemingly easy and inexpensive task, it provides no reprieve for the vast majority of older
transformers, many of which have operated undisturbed for the past few decades, and for economic
reasons, may be wisely left that way.
A multi-channel fiber optic temperature monitoring system for power transformer hot spot
measurements is shown in figure1. The new system has been developed with long-term
performance and stability in mind. This fiber optic temperature monitoring system for power
transformers offers accuracy, toughness and long-term resistance to failure.
figure1. Hot spot monitoring system using fiber optic equipments for hot spot detection
Coupled fiber-optic temperature probe provides accurate and direct temperature monitoring of
transformer windings. This solution provides a realistic, real-time view of winding conditions that is
quicker and more accurate than top oil thermocouple measurements, and greatly complements
indirect measurements based on thermal models.
System gives the exact temperature of probes in 165 millisecond per channel. Peak load or
emergency overloads are thus detected almost instantaneously. With newest technology, utility can
have a new tool to optimize high-voltage transformer performance and life expectancy.
System can specifically designed to meet power transformer industry requirements: extended
intervals between servicing, low maintenance, rugged components and the ability to withstand the
harshest conditions. All components could be specifically selected for long term performance,
including the light source that has an MTBF superior to the life of the transformer. Moreover,
compared to other technologies available on the market, like fluorescent decay, a sensor, based on
solid state semi-conductor, do not fade or drift over time, allowing a constant and absolute
temperature measurement of utility transformer windings over the lifespan of the equipment.
The fiber-optic probes are made only with dielectric materials and are designed to withstand initial
manufacturing conditions, including kerosene desorption and heat runs, as well as long term oil
immersion and vibration. Moreover, the new temperature probes are interchangeable and no
calibration or inconvenient gage factors are required when changing sensors.
The newest system is available with 1 to 16 channels. It can be mounted in an optional enclosure
with a see-through window door. The system is based on the proven solid state GaAs technology.
An original algorithm is used to analyze the signal and provide repeatable and reproducible
The system is available with a data logging option that relies on removable flash cards. This option
allows recording of temperature data points and alarms status information directly into the system
without the need for permanent connection to a remote acquisition system or to a PC. Data collected
by the system monitoring is saved in a standard SD multimedia flash card. The system can accept
memory cards of up to 128 megabytes, which is sufficient to log over 6 millions data points. This
represents more than twenty years of data logging for a transformer instrumented with 8
temperature probes! The SD card can be read by any PC through a standard SD card reader.
Moreover, data points are saved with a time stamp that comes from the internal real-time clock of
the monitoring system.
The newest system has 16 Form-C (SPDT) industrial relays with galvanic isolation which can also
be set up as Form-A or Form-B relays by user. The system can be easily connected to any
acquisition system, like a SCADA, through its built-in OPC server, its RS-232C communication
port or by the standard 4-20 mA analog outputs (0-10 Volts on request). The monitoring system
may also be ordered with DNP 3.0 and IEC 61850 connectivity.
System’s configuration is made through the indutrial grade front panel keypad or the
communication port. The monitoring front panel features 4 lines by 20 characters backlight LCD
A common method for identifying developing faults in power transformers is the Dissolved Gas
Analysis (DGA) [8]. Analysis of ratios of specific dissolved gas concentrations, their generation
rates, and the measure of total combustible gases are used as the attributes for classification.
Thresholds are designed to partition the attributes into intervals. Specific combinations of these
intervals are then used to identify the fault. However, in much the same way as determining
hotspots, the results can differ, dependent on the thresholds used, coupled with the almost infinite
variations and combinations of factors that influence the results. These are the classic characteristics
that advocate the use of Fuzzy Sets (FS) and Artificial Neural Network (ANN) systems.
For a case study, in service transformer (132/33 kV, 60 MVA,) was investigated for its abnormal
routine dissolved gas analysis behavior. Unit was commissioned in 1981 and first oil was sampled
for DGA in 1996. The key combustible gases (C2H2 & H2) were found in higher concentration.
According to the limits described in the IEEE guide for interpretation of gases, the transformer total
dissolved combustible gas (TDCG) exhibits normal conditions. C2H2 & H2 identify IEEE
condition two [Figure 2]. This shows that an active fault (early) is present. No further investigations
carried out except the transformer oil was reprocessed and the unit was energized.
Figure 2 Transformer DGA Characteristics
Loading Curves
Figure 3 Transformer
A variety of models and techniques have been assessed over the years, with claims of improvement
and higher dependability. Works done in this area by D.J. Tylavsky et al. [8] convincingly show
that there is a balance between complexity and accuracy of results, and that in fact the major
sources of error occurred from imprecision at the input.
AI systems are customarily designed for these types of problems. Coupled with physical and
mathematical models would assist in the development of a system that was both accurate and
simple to implement.
With strict monitoring, accurate diagnostics interpretations and realistic operational/ maintenance
strategies Implementation the following would be achieved effectively:
Asset economic loading conditions identification and assessment for maximum practicable
operating efficiency.
Premature failures risk minimisation.
Remnant life estimation and timely asset replacement/ retiring planning.
Asset life extension by implementing correct operational and cost effective maintenance
Improvement in the system performance ensuring good reliability as well as plant
Minimization of the long-term operational cost.
Cost saving by eliminating the unplanned maintenance.
Minimizing the outage period.
Relocation/ retirement planning.
In time procurement of spare parts to get competitive rates.
To enhance the over all reliability of the system
Accurate risk assessment.
Low insurance premiums.
It will also benefit the end user to have power supply with out interruption, in particular the
industrial sector. It will reduce the risks to human kind and the environmental damages.
[1] O. Roizman, V. Davydov, “Neuro-Fuzzy Computing for Large Power Transformers Monitoring
and Diagnostics”, Fuzzy Information Processing Society, 1999. NAFIPS. 18th International
Conference of the North American, 1999 Page(s): 248 -252
[2] IEEE Std. C57.91-1995 “IEEE Guide for Loading Mineral-Oil-Immersed Transformers”
[3] IEC 354 1991-09 “Loading Guide for Oil-Immersed Power Transformers”
[4] B.C. Lesieutre, W.H. Hagman, J.L. Kirtley Jr., “An Improved Transformer Top-Oil
Temperature Model for Use in An On-Line Monitoring And Diagnostic System”, IEEE
Transactions on Power Delivery, Vol. 12, No. 1, January 1997
[5] G.Swift, T.S. Molinski, W. Lehn, “A Fundamental Approach to Transformer Thermal Modeling
– Part I: Theory and Equivalent Circuit”, IEEE Transaction on Power Delivery, Vol. 16, No. 2,
April 2001
[6] D.J. Tylavsky, Q. He, J. Si, G.A. McCulla, J.R. Hunt, “Transformer Top-Oil Temperature
Modeling and Simulation”, IEEE Transactions on Industry Applications, Vol. 36, No. 5, Sept/Oct
[7] M. Kezunovic, “Intelligent Systems in Protection Engineering”, Power System Technology,
2000. Proceedings. PowerCon 2000. International Conference on, Volume: 2, 2000 Page(s): 801 806 vol.2
[8] Y.C. Huang, H.T. Yang, C.L. Huang, “Developing a New Transformer Fault Diagnosis System
through Evolutionary Fuzzy Logic” IEEE Trans on Power Delivery, Vol. 12, No. 2, April 1997
A review on Iranian energy resources, consumption and its role in world energy
Payam Fazilat1
Kourosh Mousavi Takami2
Jafar mahmoudi3
1. Ph.D. student in MDH 2. TDI researcher and Ph.D. student in MDH 3. Professor in MDH
TDI, Tehran, Iran; IST dep., Mälardalen University, B0x883, SE72123, Västerås, Sweden
[email protected]
[email protected] [email protected]
Iran is an important country in oil and gas production in the world. Iran’s economy relies heavily on
oil, gas export revenues. It is obvious that Iran is a considerable country in energy consumption and
CO2 production. These are our main reason for energy assessment in Iranian side. Author has
evaluated the Iranian energy production and consumptions and its role in the world.
Key words: Iran, energy, oil, production, consumption
Although Iran is a main oil producing country, due to following reasons, renewable energy (RE) is
To be in line with the world efforts to curb the world environmental problems
To improve energy accessibility for remote and isolated places in the country
To pave the way, among other things, for controlling urban air pollution
Main Policies in this paper:
Supporting private sector for dissemination of RE applications that are approaching economical
viability, such as wind, geothermal and biomass energy.
Supporting manufacturers for transferring and localization of RE technologies which are
expected to become competitive in medium terms, e.g. PV systems and solar thermal power plants.
Supporting there search centres to expand their research programs for RE technologies that are
becoming competitiveinlongerthan10 years period.
Providing sustainable and accessible energy to the poor and isolated areas. It is worth mentioning
that at present more than 96% of people in rural areas of the country has access to national electrical
Building capacity; training, supporting research centres, encouraging manufacturers to produce
required RE equipments, organizing suitable institutional framework, …
Legislation for RE development; the government is obliged to purchase electricity from private
RE power plants with the price of three times higher than the amount paid by end users.
Raising public awareness;
Pilot projects by the government; to clear the ground for the private sector to come forward.
Potential assessment studies; of RE resources with continuous upgrading of the results.
Iran's economy relies heavily on oil export revenues, with such revenues representing around 80-90
percent of total export earnings and 40-50 percent of the government budget. Strong oil prices the
past few years have boosted Iran’s oil export revenues and helped Iran's economic situation. For
2005, Iran's real GDP increased by around 6.1 percent. Inflation is running at around 16 percent per
year, though unofficial estimates place the figure at 40-50 percent. Iran’s oil export revenues have
increased steadily, from $32 billion in 2004, to $45.6 billion in 2005, with 2006 became at $46.9
billion. In 2007 it became more than $60 billion. Despite higher oil revenues, Iranian budget deficits
remain a chronic problem, in part due to large-scale state subsidies on foodstuffs and gasoline.
Thus, the country's parliament (the Majlis) decided in January 2005 to freeze domestic prices for
gasoline and other fuels at 2003 levels. In March 2006, parliament reduced the government's
gasoline subsidy allocation for FY 2006 /07 to $2.5 billion, compared with a request of $4 billion
and costs of over $4 billion for imports last year. As of July 2006, the Iranian government is still
debating how to handle gasoline subsidies. NIOC has said it has used nearly all of its $2.5 billion
budget for gasoline imports, but legislators have stated their opposition to providing the additional
$3.5 billion necessary to pay for imports through the end of the fiscal year, in March 2007.
Another problem for Iran is the lack of job opportunities for the country’s young and rapidly
growing population. Unemployment in Iran is around 11 percent, but is significantly higher among
young people. Iran is attempting to diversify its economy by investing some of its oil revenues in
other areas, including petrochemicals production. In 2004, non-oil exports rose by a reported 9
percent. Iran also is hoping to attract billions of dollars worth of foreign investment to the country
through creating a more favorable investment climate by reducing restrictions and duties on imports
and creating free-trade zones. However, there has not been a great deal of progress in this
area. Foreign investors appear to be cautious about Iran, due to uncertainties regarding its future
direction under new leadership, as well as the ongoing international controversy over the country’s
nuclear program.
In June 2005, Iran held Presidential elections in which the conservative mayor of Tehran, Mahmoud
Ahmadinejad, won a surprise victory. Ahmadinejad succeeded Mohammad Khatami, a moderate
reformist, who had been President since August 1997. Ahmadinejad ran on a populist platform and
pledged to share Iran’s oil wealth more broadly and to reduce the nation’s income gap between rich
and poor. However, policies implemented in the economic realm since President Ahmadinejad took
office are cited by analysts as responsible for high (unofficial) inflation rates, depressed housing
and stock markets, and rising unemployment. According to FACTS, reduced confidence in the
Iranian economy and the ongoing dispute with the international community has combined to cause
an estimated $50 billion of capital to flow out of Iran in the last year.
As of 2004, Iran had installed power generation capacity of about around 34.3 gigawatts
(GW). Of this total, over three-quarters was natural gas-fired, with the remainder either
hydroelectric (7 percent) or oil-fired. For 2005, Iranian power generation capacity is expected to
reach 36 GW).
Sector Organization
Although the government has considered privatization, at present Iran's power sector is run by the
state-controlled Tavanir organization. Power plant construction is handled by the Iran Power
Development Company (IPDC), a wholly owned subsidiary of Tavanir. Eventually, Tavanir may be
broken up into smaller companies as part of a privatization package. In addition to power
generation, Tavanir also is responsible for electrical transmission. Iran has three main power
distribution networks: 1) the Interconnected Network, which serves all of Iran except for remote
eastern and southern areas, using 440-kV and 230-kV transmission lines; 2) the Khorassan
Network, which serves the eastern Khorossan province; and 3) the Sistan and Baluchistan Network,
which serves the remote southeastern provinces of Sistan and Baluchistan. The government goal is
to join these three networks into one national grid. Currently, around 94 percent of Iran's rural
population has access to electricity.
Recent Capacity Additions
With power demand growing rapidly (7-9 percent annually; to 145 billion kilowatt hours in 2005),
Iran is building significant new generation capacity, with the goal of adding 18 GW over the next
five years. As a result of significant state investment in this sector, a number of new power plants
(mainly hydroelectric and combined cycle) have come online in recent years.
figure1. Iran’s power generation capacity by type
Currently, the largest hydropower projects are the 2,000-megawatt (MW) Karun 3 plant, the 2,000MW Godar-e Landar facility, and a 1,000-MW station in Upper Gorvand. Iran plans to add 6.4 GW
of hydroelectric power generating capacity over the next five years. 550 MW of new capacity was
expected to be brought online in the current Iranian year via additions of new units at the Karoun3,
Lourak, Yasouj and Menj hydropower stations. The additions will raise Iran’s hydropower capacity
from 4.8 GW in 2005 to 11.2 GW by 2011.
Natural Gas-Fired
New thermal projects include two 1,040-MW combined cycle plants in the South, a 1,300-MW
combined cycle plant at Arak, a 1,000-MW facility in Bandar Abbas, and a 1,000-MW combined
cycle plant being built by the Tehran Regional Electricity Company (TREC) in Qom. This latter
project is significant, as it is being privately financed and built by a regional - as opposed to
national - company. The plant is expected to be completed in 2007. In May 2004, a 494-MW gasfired power plant was inaugurated in Abadan. In January 2006, it was reported that Iran is to build a
1,000-MW gas power plant in the western province of Khorramabad. Construction work is planned
to start by the end of the current Iranian fiscal year and will be completed within four years.
Build-Operate-Transfer (BOT) Projects
Iran has received offers for investment in the form of loans and build-operate-transfer (BOT)
contracts, but progress has been slow. BOT contracts allow the investing company to build and
operate the generating facility for a period of 15-20 years, after which time the plant is turned over
to the Energy Ministry. Negotiations have taken place with international energy firms on expansion
plans for power plants at Bandar Abbas, Shaid Rajai, Alborz, Ramin, and Kerman.
In June 2006, Iran’s first BOT power plant became fully operational, when the last of six 159MW
open cycle gas turbine generating sets comprising the Chehelsotun power plant in South Isfahan
were brought online. The 950MW gas-fired plant, the first to be completed in Iran under a BOT
agreement, was developed by a 50:50 joint venture between the Iranian investment house IHAG
and local power contractor Mapna. The first unit at the Chehelsotun plant was brought online in
In addition to BOT plants, Iran has attempted to promote a build-own-operate (BOO) model for the
2,000-MW, Zanjan 1-4 independent power project (IPP). In September 2004, the BOO plan was
dealt a setback due to a lack of bidders, and may be re-bid. Overall, Iran is planning 5,800 MW of
BOT projects and 7,000 MW of BOO projects.
International Trade
Overall, in 2004, Iran imported around 2.2 billion kilowatthours (Bkwh) per year, up from 1.5
Bkwh in 2003,, and exported just under 1.9 Bkwh. Iran trades electricity with Afghanistan (exports
to the western part of the country), Armenia (exports and imports), Azerbaijan (exports and
imports), Pakistan, Turkey and Turkmenistan (exports and imports). In April 2003, Iran said that it
would be willing to supply Iraqi cities with electricity as well. In December 2004, a protocol was
reached on synchronizing the power grids of Iran, Azerbaijan, and Russia, with 500 MW being
exchanged beginning in 2006. In August 2004, Turkmenistan began power exports to Iran via a new
transmission line (Sarahs). Annual exports of 375 million kilowatt-hours, worth $7.5 million, are
expected. This line adds to previous electric export capacity from Turkmenistan to Iran via the
Balkanat-Gonbad line started up in June 2003. Another line is to be constructed in the short-term,
bringing total power exports from Turkmenistan to Iran to 2.4 billion killowatt-hours per year.
According to the Oil and Gas Journal, as of January 1, 2006, Iran held 132.5 billion barrels of
proven oil reserves. This figure, which includes recent discoveries in the Kushk and Hosseineih
fields of Khuzestan province, means Iran holds roughly 10 percent of the world's total proven
reserves. The vast majority of Iran's crude oil reserves are located in giant onshore fields in the
southwestern Khuzestan region near the Iraqi border. Overall, Iran has 40 producing fields – 27
onshore and 13 offshore (see table below for major fields). Iran's crude oil is generally medium in
sulfur and in the 28°-35° API range.
figure2. Iranian toatal liquids production and consumption
During 2005, Iran produced about 4.24 million bbl/d of total liquids. Of this, 3.94 million bbl/d is
crude oil, roughly 5 percent of world crude production. Iran's current sustainable crude oil
production capacity is estimated at 3.8 million bbl/d, which is around 310,000 bbl/d below Iran's
latest OPEC production quota of 4.110 million bbl/d. Through the first half of 2006, Iran’s crude oil
production was at 3.75 million bbl/d. Iran's domestic oil consumption, 1.5 million bbl/d in 2005, is
increasing rapidly as the economy and population grow. Iran subsidizes the price of oil products
heavily, which contributes to rising domestic consumption.
Iran's existing oilfields have a natural decline rate estimated at 8 percent onshore and 10 percent per
year offshore. The fields are in need of upgrading, modernization, and enhanced oil recovery (EOR)
efforts such as gas reinjection. Current recovery rates are just 24-27 percent, compared to a world
average of 35 percent. Iran also needs to increase its search for new oil, with only a few exploration
wells being drilled in 2005.
With sufficient investment, Iran could increase its crude oil production capacity significantly. The
country produced 6 million bbl/d of crude oil in 1974 but has not come close to recovering to that
level since the 1978/79 Iranian revolution. Still, Iran has ambitious plans to increase oil production
to more than 5 million bbl/d by 2010, and 8 million bbl/d by 2015. The country will require billions
of dollars in foreign investment to accomplish this.
Currently, Iran exports around 2.5 million bbl/d of oil, of which OECD countries import 60 percent
(or 1.6 million bbl/d). Iran's main export blends include Iranian Light (34.6° API, 1.4 percent
sulphur); Iranian Heavy (31° API, 1.7 percent sulphur); Lavan Blend (34°-35° API, 1.8-2 percent
sulphur); and Foroozan Blend/Sirri (29-31° API).
Figure3. a) Iranian crude oil exports by Blend(2004), b) OPEC total crude oil production
Crude Swaps
Iran's desire to become a player on the Caspian oil front has led it to push forward in the area of oil
"swaps." This arrangement involves the delivery of Caspian oil to refineries, via the Caspian port
town of Sari in northern Iran, for local consumption (See map below). An equivalent amount of
Iranian oil is then exported through Persian Gulf terminals such as Kharg Island. Shippers normally
pay a "swap fee" of $1.50-$2.00 per barrel, with swaps handled by Naftiran Intertrade Co. (Nico),
the Swiss-based trading arm of NIOC. Crude swaps have increased rapidly in recent months, with a
July 2006 report by the Oil Export Terminals Company placing the level at 130,000 bbl/d. As
recently as June 2006, Abdolrahman Kheylai, Manager of NIOC’s Sari Oil Products Export
Terminal, stated that average daily Caspian exports to Iran were about 117,000 bbl/d, consisting
mainly of Turkmen and Kazakh oil. The surge in crude swaps is attributed to higher oil prices and
the increased willingness of the Caspian Sea littoral states to offer oil in the form of direct sales.
Currently, from sari, just over 40,000 bbl/d of import oil is sent to Tehran by the existing 180,000bbl/d capacity Neka-Tehran pipeline, and roughly 80,000 bbl/d is sent to Tebriz. Eventually, Iran
hopes to upgrade its facilities in order to greatly expand oil swaps capacity first to 370,000 bbl/d
and then to 500,000 bbl/d. Iran further intends to increase storage capacity at the terminal from 1.5
million barrels to 2.5 million barrels. As a further part of Iran’s efforts to compete on costs with the
1-million-bbl/d Baku-Tbilisi- Ceyhan (BTC) pipeline, Mr. Kheylai reported that Iran plans to build
a submerged turret offshore oil loading terminal in the Caspian Sea. The technology offers a
flexible and cost-effective solution for mooring of vessels applied as shuttle tankers or storage
vessels. Mr. Kheylai stated that while currently 7,000 deadweight (dwt) tankers can bring Caspian
oil to Neka, the new turret will be able to service up to 63,000 dwt tankers.
In July 2005, Iran and Iraq signed an MOU on a swap agreement involving construction of a 24mile, 350,000-bbl/d oil pipeline from Basra to the Abadan refinery in southwestern Iran. In
exchange, Iran would ship refined products back to Iraq. In addition, Iran could allow Iraq to export
crude through the Kharg Island terminal and to import refined products through the Iranian port of
Bandar Mahshahr. One potential problem with this deal revolves around the ability of the Abadan
refinery to process Basrah Light in significant volumes. Another problem is Iran already faces a
severe shortfall in its own domestic gasoline supplies, making additional exports of gasoline
Export Terminals
Iran exports crude oil via four main terminals: Kharg Island (by far the largest), Lavan Island, Sirri
Island (reopened on April 13, 2003 for the first time since 1988, when it was damaged by an Iraqi
air raid), and Ras Bahregan. Refined products are exported via the Abadan and Bandar Mahshahr
terminals. Many Iranian oil export terminals were damaged during the Iran-Iraq War, but all have
been rebuilt. Iran operates OPEC’s largest oil tanker fleet, comprising roughly 29 ships, including
Very Large Crude Carriers (VLCCs). The fleet is run by the National Iranian Tanker Company,
which is a subsidiary of NIOC. In 2003, Iran commissioned six vessels for Caspian routes, to
promote Iran’s position as a transit option for regional crude, along with 12 LNG tankers.
Top Ten Iranian Crude Oil Export Destinations
(Thousand bbl/d)
Reporting Country
South Korea
South Africa
Reporting Total
OECD Pacific
OECD Europe
Source: Global Trade Information Services, Inc., Global Trade Atlas, July, 2006.
Upstream Projects
State-owned National Iranian Oil Company (NIOC)'s onshore field development work is
concentrated mainly on sustaining output levels from large, aging fields. Roughly 60 percent of
Iran’s oil production comes from fields more than half a century old , with some dating back to the
first oil discoveries in the Middle East . Consequently, EOR programs are underway at a number of
fields, including Marun and Karanj.
figure4. Map of Major Iranian Energy Projects and Pipelines
In February 2004, a Japanese consortium led by Inpex signed a final agreement on the $2 billion
Azadegan oilfield development project. Azadegan was discovered in 1999, representing Iran's
largest oil discovery in 30 years. It is located onshore in the southwestern province of Khuzestan, a
few miles east of the border with Iraq. Reportedly, Azadegan contains proven crude oil reserves of
26 billion barrels, but the field is also considered to be geologically complex, making the oil more
challenging and more expensive to extract. In January 2001, the Majlis approved development of
Azadegan by foreign investors using the so-called "buyback" model (see below). Inpex, which has
no upstream experience of its own, has brought in Total on the project for its technical expertise.
Initial production of medium-sour crude oil from Azadegan could come in 2009, ramping up to
160,000 bbl/d by 2012 and 250,000 bbl/d by 2014/15. At its peak, Azadegan production could
account for as much as 6 percent of Japan's oil imports. However, little forward progress has been
made on Azadegan, including the lack of an operating agreement with NIOC. In September 2005,
Iran sharply criticized Japan for the slow progress. Japan maintains that further clearing of mines
leftover from the Iran-Iraq war is needed for work to go forward and that because of an increase in
the price of field development tools, notably steel, the contract should be revised. Iran claims the
field is 96 percent clear of mines and has threatened to develop the field themselves if a final
agreement is not reached by September 2006 .
In February 2001, NIOC announced the discovery of a very large offshore oil field, named Dasht-e
Abadan, in shallow waters near the port city of Abadan. According to a top NIOC official, Dasht-e
Abadan could contain reserves "comparable" in size to Azadegan.
Since 1995, NIOC has made several other sizable oil discoveries, including Darkhovin onshore
oilfield. Located near Abadan, which contains low sulfur, 39° API crude oil. In late June 2001, Eni
signed a $1 billion, 5 1/2-year buyback deal to develop Darkhovin, with the added incentive of a
limited risk/reward element that linked payment to production capacity. Darkhovin came online at
50,000 bbl/d in July 2005, with production expected to reach 160,000 bbl/d in 2007.
Another oil discovery in western Iran was made recently by Norsk Hydro, the Anaran field, which
contains reserves of 2 billion barrels. According to Norsk Hydro, Anaran could produce more than
100,000 bbl/d of oil, possibly starting in 2010. Lukoil is a minority partner in the field. Reportedly,
development of Anaran is complicated by the need to clear landmines in the area.
Major Iranian Oil Fields (est. production, bbl/d)
Agha Jari (200,000 bbl/d)
Abuzar (140,000 bbl/d)
Ahwaz-Asmari (700,000 bbl/d)
Dorood (130,000 bbl/d, to increase
to 180,000 bbl/d)
Bangestan (around 245,000 bbl/d current production, with
Salman (100,000 bbl/d, to increase
plans to increase to 550,000 bbl/d or more)
to 150,000 bbl/d)
Bibi Hakimeh (130,000 bbl/d, to increase to 175,000 bbl/d)
Sirri A&E (95,000 bbl/d)
Gachsaran (480,000 bbl/d, to increase to 600,000 bbl/d)
Soroush/Nowruz (87,000 bbl/d)
Karanj-Parsi (250,000 bbl/d)
Marun (520,000 bbl/d, to increase to 600,000 bbl/d)
Pazanan (35,000 bbl/d)
Rag-e-Safid (180,000 bbl/d)
NIOC also would like to develop five oil and natural gas fields in the Hormuz region: Henjam
A; the A field near Lavan Island; the Esfandir field near Kharg Island; and two structures near the
South Pars natural gas and condensate field. Iran and Oman have discussed the possibility of joint
development of the Henjam A field. According to NIOC, the five Henjam fields hold an estimated
400 million barrels of oil and have a production potential of 80,000 bbl/d. Other Iranian oil fields
slated for increased development include Doroud, Nosrat, Farzam, and Salman.
As a further part of Iran’s expansion efforts, in May 2006, the head of Petroleum Engineering and
Development Company (PEDEC), Mehdi Bazargan, said that Iran will tender the development of
10 additional oilfields in its oil-rich southern provinces, including the Parsi, Shadgan, Pazenan,
Gachsaran, Karanj, Northern Azadegan, Jofair, Marun, and Mansuri fields. According to Mr.
Bazargan, the development of the fields will require roughly $7 billion dollars of investment.
Storage Capacity
In June 2006, Crude oil storage capacity at Kharg Island, the country’s largest export terminal, had
been increased over the past six years from 7 million barrels to 12 million barrels. Iran has begun a
$218 million expansion program to raise the terminal’s storage capacity to 22 million barrels. Iran’s
immediate aim should be to increase storage capacity to 10 - 15 days worth of exports, and in the
longer term to 30 - 35 million barrels.
In April 2006, NIOC planned to begin construction during 2006 of strategic crude oil storage tanks
in a number of producing regions, with a total capacity of 10 million barrels. Construction costs are
estimated at $1.165 million. NIOC had previously announced plans to construct 15 crude oil storage
tanks at the following locations: four tanks at Ahwaz with total capacity of 2 million barrels of
crude oil; three tanks at Omidiyeh with total capacity of 3 million barrels; six tanks at Goureh with
total capacity of 4 million barrels; one 500,000 barrel tank on Sirri Island; and one 500,000 barrel
tank at Bahregansar.
Foreign Investment/Buybacks
The Iranian constitution prohibits the granting of petroleum rights on a concessionary basis or direct
equity stake. However, the 1987 Petroleum Law permits the establishment of contracts between the
Ministry of Petroleum, state companies and "local and foreign national persons and legal entities."
Buyback contracts, for instance, are arrangements in which the contractor funds all investments,
receives remuneration from NIOC in the form of an allocated production share, then transfers
operation of the field to NIOC after a set number of years, at which time the contract is completed.
The first major project under the buyback investment approach became operational in October
1998, when the offshore Sirri a oil field, then-operated by Total and Malaysia's Petronas, began
production at 7,000 bbl/d. The neighboring Sirri E field began production in February 1999, with
production at the two fields expected to reach 120,000 bbl/d. Under the 1995 agreement, as the two
fields came on stream, the National Iranian Oil Company (NIOC) gradually took over as operator
after five years; but TOTAL will still be involved in the operations until its investment has been
fully recovered.
In March 1999, France's Elf Aquitaine (now Total) and Italy's Eni were awarded a $1 billion
contract for a secondary recovery program at the offshore, 1.5-billion-barrel Doroud oil and natural
gas field located near Kharg Island. The program is intended to boost production from around
136,000 bbl/d to as high as 205,000 bbl/d. Total is operator of the project, with a 55 percent
share, while Eni holds the other 45 percent. In April 1999, Iran awarded Elf a buyback contract to
develop the offshore Balal field, which was turned back over to NIOC in January 2003 when the
field reached its contract designated level of 40,000 bbl/d. Eni is also involved, with a 38.25 percent
stake. The field, which contains some 80 million barrels of reserves, started producing in early 2003
with currently has output of 40,000 bbl/d.
On March 18, 2005, a contract to develop the giant Bangestan field was awarded to Petro Iran
Development Co., after having been delayed several times since 2001. Bangestan contains an
estimated 6 billion barrels of oil reserves and produces about 158,000 bbl/d of oil, but the field is
one of the oldest in the country, requiring investment and technological applications to compensate
for natural decline. While development of Bangestan could cost $3 billion over 10 years, and aims
to raise output to 600,000 bbl/d, currently production at the field has dropped by roughly 50 percent
from peak levels and may drop a further 50 percent due to lack of gas re-injection.
In May 2002, Iran's Oil Ministry signed a $585 million buyback contract with NIOC subsidiary
PetroIran to develop the Foroozan and Esfandiar offshore oilfields. Currently producing 40,000
bbl/d, PetroIran is expected to increase production at the fields to 65,000 bbl/d. The two oilfields
straddle the border with Saudi Arabia's offshore Lulu and Marjan fields.
The Cheshmeh-Khosh field, which previously had been awarded to Spain's Cepsa for $300 million,
was re-awarded in January 2004 to state-owned Central Iranian Oil Fields Company (CIOFC). In
December 2003, Cepsa and OMV withdrew from lengthy negotiations after a reported failure to
agree on development costs and buyback terms. It remains possible, however, that Cepsa and OMV
could still be involved at Cheshmeh-Khosh in some way. The objective is to raise crude production
at the field from 40,000 bbl/d currently to 80,000 bbl/d within four years.
In late May 2002, Canada's Sheer Energy became the first foreign company since Eni's Darkhovin
deal to reach agreement to develop the Masjed-I-Suleyman, or MIS, field. Sheer's goal was to boost
MIS production from 4,500 bbl/d to 20,000 bbl/d at the historic field, which, discovered in 1908,
peaked at 130,000 bbl/d in the 1930s. However, the company was replaced by China's CNPC,
which bought the subsidiary of Sheer working on MIS. CNPC began work on the field in June
The buyback system has drawbacks for both sides: by offering a fixed rate of return (usually around
15-18 percent), NIOC bears all the risk of low oil prices. If prices drop, NIOC has to sell more oil
or natural gas to meet the compensation figure. At the same time, companies have no guarantee that
they will be permitted to develop their discoveries, let alone operate them. Finally, companies find
the short terms of buyback contracts unfavorable.
In response, Iran has considered revisions to buyback terms, including extending the length of
contracts and allowing for continued involvement of oil companies after the field is handed over to
NIOC. However, these proposals have been controversial and generally have not moved forward.
While modest changes have been made, such as increases in the number of years of buybacks and
additional incentives for E&P projects, political opposition continues to prevent any sizeable
reforms. The last major buyback left is the Eni-led development of the onshore Darkhovein oil
field, which is due to reach its targeted plateau of 160,000 bbl/d by the end of 2006.
Offshore Developments
The Doroud 1&2, Salman, Abuzar, Foroozan, and Sirri fields comprise the bulk of Iran's offshore
oil output. Iran plans extensive development of existing offshore fields and hopes to raise its
offshore production capacity significantly.
In late 2001 and early 2002, Shell brought part of the $800 million Soroush-Nowruz development
online. The two fields are located offshore, about 50 miles west of Kharg Island, and contain
estimated recoverable reserves of around 1 billion barrels of heavy oil (20° API). The heaviness and
high sulfur content (3 percent) of the oil has made marketing Soroush-Nowruz oil difficult; in 2005
and again in the second quarter of 2006, Iran reportedly diverted Soroush-Nowruz production into
storage rather than try to sell the oil at a steep discount. In addition, production has been limited to
27,000 bbl/d at Nowruz and 60,000 bbl/d at Soroush, despite goals of 57,000 and 65,000 bbl/d,
respectively. In an attempt to sell more of Nowruz/ Soroush crude, NIOC has recently begun
blending the crude with South Pars condensate. The combined batch is then being fed into the
Bandar Abbas refinery for domestic consumption. In July 2006, Iran reported that it had processed
2 million barrels of Soroush and Nowruz crudes at Bandar Abbas. In May 2006, NIOC secured a
60,000 bbl/d term contract for Nowruz/ Soroush crude with India’s Reliance.
Another area believed to have significant potential is offshore near Bushehr, where Iran claimed in
July 2003 to have discovered three fields with potential reserves of 38 billion barrels of oil. In
March 2004, the Iranian Offshore Oil Company (IOOC) awarded a $1.26 billion contract for
recovery of NGLs and natural gas from Soroush, Nowruz, Foroozan, and Abuzar to Japan's JGC
Corporation. Ethane from the gas will feed an ethylene complex at the Kharg petrochemical
complex. Elsewhere, in May 2004, Brazil's Petrobras signed a 3-year, $32-$34 million deal to
develop the Tousan fields of the Persian Gulf.
Caspian Sea Region
Aside from acting as a transit center for other countries' oil and natural gas exports from the
Caspian Sea, Iran has potentially significant Caspian reserves of its own. Currently, Iran has no oil
or natural gas production in the Caspian region. In early 2004, a 3-D seismic survey of the southern
Caspian was conducted by Iran's Oil Survey Co.; the work yielded a number of potentially
prospective blocks, including Blocks 6, 7, 8 and 29. In June 2006, Masoud Jahdi, Director of the
state-run Caspian Sea Exploration Company, stated that Iran has completed 3D seismic work of
1,550 sq mile of the sea, and a further 11,580 sq miles has been swept in 2D surveys. According to
a 1998 study by Eni, Shell and KEPCO, the Iranian sections of the Caspian Sea have significant
potential. A tender for exploration is expected to follow after the seismic work is complete. Among
the foreign firms that have indicated their interest in the area are Statoil, Norsk Hydro, Petrobras,
Petronas and CNOOC.
At the present time, Iran continues to maintain that regional treaties signed in 1921 and 1940
between Iran and the former Soviet Union, which call for joint sharing of the Caspian's resources
between the two countries, remain valid. Iran has rejected all unilateral and bilateral agreements on
the utilization of the Sea. As such, Iran is insisting that either the Sea should be used in common, or
its floor and water basin should be divided into equal (20 percent) shares. Under the so-called
"condominium" approach, the development of the Caspian Sea would be undertaken jointly by all
of the littoral states. However, using the equidistant method of dividing the seabed on which
Kazakhstan, Azerbaijan, and Russia have agreed, Iran would only receive about 12-13 percent of
the Sea. As yet, no agreement has been reached among all Caspian Sea region states on the legal
status of the area.
According to Oil and Gas Journal, Iran has a combined capacity of 1.64 million bbl/d. Major
refineries include: Abadan (400,000-bbl/d capacity); Isfahan (265,000 bbl/d); Bandar Abbas
(232,000 bbl/d); Tehran (225,000 bbl/d); Arak (150,000 bbl/d); and Tabriz (112,000 bbl/d).
Gasoline demand is forcasted to be growing at around 11.4 percent per year. Iran plans to increase
its refining capacity to 2.54 million bbl/d by 2010. One goal of this expansion is to allow Iran's
refineries to process a heavier crude slate, while decreasing the fuel oil cut. Currently, Iran's
refineries produce around 30 percent heavy fuel oil and just 16 percent gasoline. In addition, diesel
sulfur levels are slated for a major reduction from 500 parts per million to 50 ppm by 2010,
requiring significant additional hydrotreating capacity.
The National Iranian Oil Refining and Distribution Company (NIORDC) plans to begin
construction work as early as September 2006 on three units aimed at increasing gasoline
production from the Isfahan refinery. Currently, technical proposals are being reviewed for the
construction of three units: a 32,000 bbl/d continuous catalytic reformer (CCR) unit; a 27,000 bbl/d
isomerization unit; and a 62,000 bbl/d hydrotreater. Construction is expected to cost $300 million.
NIORDC is negotiating with bidders to reduce construction time from 36 months to 30 months.
In June 2004, Japan's JGC reached an agreement with Iran to expand Arak to 250,000 bbl/d by
2009. In addition, in 2005 it was announced that a new 180,000-bbl/d-capacity refinery is being
planned for Abadan. Bandar Abbas is being expanded in several phases and is on schedule to meet
goals of adding around 250,000 bbl/d of capacity by 2010. Two planned grassroots refineries
include a 225,000-bbl/d plant at Shah Bahar and a 120,000-bbl/d unit on Qeshm Island. Under
Iranian law, foreign companies are permitted to own no more than 49 percent of Iranian oil refining
Iran plans to boost capacity at its northern refineries at Arak, Tebriz, and Tehran in order to process
additional Caspian oil. In August 2003, a $500 million tender was issued to upgrade the Tehran and
Tabriz refineries in order to handle 370,000 bbl/d of high sulfur Caspian crude. This follows a $330
million project, completed by a Sinopec-led consortium in late 2003, to expand storage at Neka and
to upgrade the Tehran and Tabriz refineries.
Gasoline Imports
Iran has imported refined products since 1982, and these imports have been increasing rapidly.
Currently, gasoline costs less than 40 cents per gallon in Iran, far below international levels,
contributing to a rapid (8-10 percent per year) growth rate in gasoline consumption. In addition, the
country imports around one-third of its gasoline. In volume terms, Iran is the second largest
importer of gasoline in the world after the United States. In 2005, Iran imported 150,000 bbl/d out
of total consumption of 400,000 bbl/d. For 2006, FACTS estimates Iran will consume 462,000 bbl/d
of gasoline and import 188,000 bbl/d, or roughly 41 percent of total consumption. According to
Petroleum Argus, around 60 percent of this comes from European oil trader, Vitol, with another 15
percent coming from India’s 600,000-bbl/d Reliance refinery However, although Iran imports large
amounts of gasoline, it is an overall net petroleum products exporter due to its large gross exports of
residual fuel oil.
Through the summer of 2006, there was heated debate among Iranian politicians as to how to
balance expanding gasoline needs and the corresponding increase needed for gasoline subsidies.
NIOC has said it has used nearly its entire $2.5 billion budget for gasoline imports but legislators
have stated their opposition to providing the additional $3.5 billion necessary to pay for imports
through the end of the fiscal year, in March 2007. Options put forward for addressing the issue
included ceasing gasoline imports in September 2006 when many contracts expire and rationing
gasoline, or implementing a two-tier price system whereby each car is allotted 2.5-3 liters a day at
the subsidized price of 8.7 cents per liter, with consumption beyond that at 60-65 cents per liter.
Going forward, FACTS reports that Iran will complete construction of three 120,000 bbl/d
condensate splitters by 2009. The facilities will produce an estimated 200,000 bbl/d of gasoline.
Along with other projects, it is possible that Iran will cease being a gasoline importer by 2010.
Given its heavily energy-centered economy, environmental issues in Iran generally have ranked as a
relatively low priority. However, ongoing and severe air pollution in urban areas have highlighted
the need to improve Iran's environmental record. The rush to develop oil and natural gas resources
in the Caspian Sea makes oil pollution in the Caspian a real environmental threat.
Huge increases in energy consumption over the past 20 years have contributed greatly to pollution
levels as Iran's carbon emissions have nearly tripled over the same time span. Large numbers of old,
inefficient cars on the road lacking catalytic converters account for much of the country's air
pollution. Energy prices are kept artificially low in Iran through heavy state subsidies, resulting in
highly inefficient and polluting consumption patterns. In addition, Iran's abundance of fossil fuel
resources has tended to discourage the country's incentive to shift to cleaner alternative energy
sources for its energy needs.
Ongoing in alternative energy area
Iranian energy minister have decided to establish of renewable and so green energy production
systems in Iran, same as solar power plant, Wind Turbine, PV systems, geothermal power plant, sea
waves energy, biomass system and micro hydro power plant.
As of 2004, Iran had installed power generation capacity of about around 34.3 gigawatts (GW).
Currently it is above of 45 GW. Of this total, over three-quarters was natural gas-fired, with the
remainder either hydroelectric (14.7 percent) or oil-fired. For 2005, Iranian power generation
capacity is expected to reach 37 GW).
It has reported [1] that the electricity production in Iran per person is 2621 KWH, and power per
person is 604 W.
Compositions of power plants in Iran are: steam station 15554 MW (37.9%), Gas station12,050
MW(29.4%), combined cycle 6832MW(16.7%), hydropower 6037MW(14.7%) and Wind and
Diesels are 530MW(1.3%). Tavanir reported that in 2005 Iranian nominal capacity of power station
conclude production by ministry of energy (38,038MW) and another parts was above the 41,003
MW, and average produced power was 37,054 MW, but maximum prepared load was 31,306 MW,
and maximum mitigated load was 32,302MW.
Based on Tavanir report In 2006 average electricity production in Iran was 43,610MW, in 2007 will
be 49,112, in 2008 so 62,722, in 2009 so 72,799, in 2010 so 78,275 and in 2011 will be 80.026
Iranian try in this area are discussed in following:
Wind Mapping Iran, Wind turbine production facilities in Iran
The goal of the project is to achieve a detailed understanding of the wind regime over Iran and to
define the basis for the purpose of wind park projects development in Iran.
For compilation of the wind map the KLIMM model is applied. The standard software tools for
wind maps are based on the WAsP model. The limitations of this program especially for complex
and semi-complex terrain are well known and have been published in numerous publications. With
the application of the three dimensional numeric mesoscale atmosperic model of KLIMM the long
term wind data at the upper boundary (geostrophic wind) and the temperature stratification of the
atmosphere can be used as input for the meteorological data. Site orography and roughness data is
used to model the site conditions. A nation-wide measurement campaign. With the installation of 50
meteorological stations will provide the near ground wind data for model calibration and fine
adjustment. This procedure guarantees high accuracy and a representative character of the
calculated wind conditions. The applied calculation model is among the most powerful calculation
procedures available for wind energy practice.
Main Data:
Land-covering wind map with a grid resolution of 200 to 1000 m for an area of 1400 km x 1200
Complex terrain
3D mesoscale atmospheric modelling with KLIMM
Wind map design
Transferability of measurements of ground wind
Identification of suitable wind park sites
Data collection and processing
Wind potential analysis
Computation of wind map using the 3D atmospheric model KLIMM
Print of wind map, incl. geographical information
Training program for local staff
Preparation of a digital wind data base
Support of wind monitoring program
Identification of suitable wind park sites
Iran Governmental energy policy and renewable energy
Iran policy about fossil fuels consumption for energy supply will be continued. However in
renewable energy resources Iran Policy is to supply 1 percent of Electricity with renewable
resources until fiscal year 2010. This amount of energy is about 550 MW electricity. This capacity
will be gained from different renewable resources as follows:
1-Renewable Power Plants which are constructed by Ministry of Energy (300MW):
- 55 MW Geothermal Power Plant in Sabalan (North West of Iran)
- 60 MW Wind Power Plant in Jarandagh( Center of Iran)
-100 MW Wind Power Plant in Manjil (North of Iran)
- 10 MW Biomass Power Plant (Urban solid waste)
- 25MW Wind Power Plant in Binalood (East of Iran)
- Photovoltaic systems in rural areas (50 MW Total Capacity)
2-Renewable Power Plants which are constructed by private sector (250MW):
Electricity produced by these power plants will be bought by government higher than 3 times of
fossil electricity price. Government considers this law (Article number 62) for promoting private
sector to erect renewable power plants.
Availability of Biomass in Iran
Considering, the availability of all types of Biomass resources in Iran, Power Ministry did a
research to evaluate the potential to use biomass for renewable energy. This research was done in
1976 and conclusions are as follows:
1- Quantity of accessible Animal Waste in Iran is 74946000 ton; Biogas could be produced by this
amount is 8668 MillionM3 and energy equivalent this biogas is 193299TJ. (TJ=1012 J)
2- Quantity of accessible Urban Waste in Iran is 10605000 ton: Biogas could be produced by this
amount is 1645.7 MillionM3.
3- Quantity of accessible Biogas from Urban Waste Water in Iran is 98.855 M3/day. (Waste Water
specifications are: 160 Litter per capita per day and BOD per capita is 33-40 gr/d.c)
4- Quantity of accessible Methane from Industrial Waste Water in Iran is 167664199 M3/year.
Energy equivalent this quantity is 6153TJ per year. ( This Waste Water is from Big Food Industry
in Iran.)
5-Total gross energy potential from Agricultural and Wood Waste is 411.896 PJ from 3285.50
Million M3 Methane and 33943344 M3 Ethanol which could be produced.
Case Study on Use of biomass for renewable energy
We analyze two case studies here: Mashhad Landfill Power Plant and Saveh Biogas Plant.
Design and Implementation of the First Landfill-Gas Power Genration Unit in Mashhad (Iran)
The first Landfill-Gas (LFG) power generation project was initiated in Mashhad, the second largest
city in Iran, through a memorandum of understanding between “Renewable energy Organization of
Iran (SUNA)” and “Waste Recycling and Processing Organization of Mashhad Municipality
(WRPO)”. Based on the mentioned memorandum, SUNA took the responsibility of preparing
technical investigations and tender documents and follow-up the electricity purchase contract
between Mashhad LFG power plant and Ministry of Energy. Investment for power plant and
commissioning was accepted by WRPO. Feasibility studies, conceptual design and tender
documents preparation were consequently allocated to Renewable Energy Department of NRI.
The first phase of this project, included design criteria for engineering landfills and gas extraction
systems and transmission lines, introduction to energy conversion and power generation
technologies for LFG, field investigations and measurement methods and monitoring of LFG. The
next phase was feasibility study for installation of power generation unit in Mashhad Landfill, in
which, estimation of landfill gas theoretical yield for Mashhad municipal solid wastes, LFG
generation modeling and gas flow prediction for long term periods, evaluation of confident power
capacity for Mashhad Landfill, technical assessment of power generation equipments for LFG and
introduction of more appropriate commercial generating sets for utilization of LFG in Mashhad
Landfill, were performed. In the third phase, economical studies, three alternatives were selected on
the base of feasibility study results. These alternatives were assessed as economical aspects and
finally, installation of generating sets near the guard room of Mashhad Landfill, which is located
near to irrigation station and a 20kV line, was selected as the most useful option for utilization of
the whole gas which can be collected through the existing gas extraction systems. The generated
electricity in this unit will be sold to power network at an average price of 616 Rls. per kWh.
The phase of conceptual design included: pipeline design for LFG transmission to central gas
station, gas blower station, gas treatment process, foundation and structure of generating house,
power generation unit arrangement and grid interconnection system plus protection and confidential
instruments. Technical characteristics and tender documents were consequently prepared on
conclusions. The contract for construction and start-up of Mashhad Landfill power generation unit
will be as an Engineering Procurement Construction (EPC) contract.
Project Results:
- Pipeline and collection systems were designed on the basis of 13 years useful operation period of a
biogas fueled generating set, and LFG generation modeling results, for average gas flow rate of 550
Nm3/h and maximum gas flow rate equal to 660 Nm3/h. A minimum power capacity of 435 kWe is
expected to be achievable until the end of 2018 by existing gas extraction systems.
- The power capacity can be increased to at least 715 kWe if Mashhad
Municipality develop the gas extraction systems to recent buried layers of solid wastes which are
dumped in period of 2004-2006.
- Generating sets with LFG fueled internal combustion engines are the best option for Mashhad
Landfill , furthermore, a list of qualified manufacturers of these equipments, has been presented in
the project documents.
- Economical assessment results showed the best alternative is the installation of power generating
unit in near of Landfill irrigation station.
A case study with capacity of 400 kWe (due to capacity of commercial models) will have the
generation cost of electricity relevant to discount rates of 8%, 12% , 16% and 20% equal to: 293,
320, 350 and 382 Rials per kWh, respectively.
- LFG power plant interconnection with regional power grid is feasible through the existing 20kV
line in Mashhad Landfill. Electrical circuit drawing and single-line diagram for connection to 20kV
network were obtained and presented in project documents.
Saveh Power Plant
The city of Saveh is in the central province, 150 km from Tehran with a population of 120,000. It
has a Semi-arid climate and the temperature during different seasons of the year fluctuates within
the range of –10°C to 40°C. In order to establish the Bio-gas power plant in this city, the usable
pollutants in Saveh were studied and they were divided into four groups:
Household garbage, sludge of sewage treatment, slaughter house waste water and sludge of
leaching pit. Through statistical studies, separation operation, sampling and executing physical, and
chemical tests, quantity and quality specifications of each of the four types of pollutants were
Some of the ongoing projects:
1- Two parabolic trough solar thermal power plants with the total capacity of17.25 Mwplants
This project consists of a 250 kW parabolic trough solar thermal power plant which has been
designed and is under construction and an integrated solar combined cycle with the solar share of 17
MW. Recently the first loop of the solar collectors of the 250 kW power plant has been tested
2- Geothermal power plant with final capacity of 55 MW
This power plant is going to be installed near Mount Sabalanin the north-west of Iran
3-398 MW Wind farm398 farmin five different locationsb)
1stlocation: 55.68 MW Manjil Wind Farm,
2ndlocation: Binalood Wind Farm in the north-east of Iran: total capacity 17 MW, These projects
consist of 660 kW turbines wheresome parts including turbine blades are manufactured in Iran.
3rdlocation: Jarandaq Wind Farm 100 MW
4thlocation: Binalood Wind Farm 100 MW
5thlocation: Siapush Wind Farm 125 MW
For these three sites private investors have completed their feasibility studies and are in the final
stage of contract agreement. The power purchase agreements of these projects are available and had
been presented to the investors together with the available wind data.
4- 90 MWSmall hydro power plants in different regions
a) With public investment:10 MW (under study)
b) Private power plants :
1-City of Mashad , landfill technology0.4 MW(PPA preparing)
2-City of Sari , plasma technology15 MW(under way)
3-City of Shiraz , landfill technology1.1 MW(under way)
6- PV systems for 40 isolated village households
In this project 40 off-grid PV systems each with the capacity of 1.1 kW would be installed and
tested in different village households in orderto pave the way for installation of about 12000 of
those systems in the isolatedand remote areas of the country with total installed capacity of 13 MW.
Main Achievements
Enabling Iranian enterprises for RE systems design and also manufacturing of wind turbines,
hydro power plant equipments, solar water heaters, parabolic trough concentrators, PV cells, panels
and systems Experiencing geothermal drilling in harsh climate areas.
Building confidence in private sector investors for producing electricity, especially through wind
Considerable up bringing of the knowledge of universities in RE fields.
Iran’s energy production, consumption and export are evaluated. Iran swap the crude oil and gas in
Caspian sea side and so Persian golf. Iran's economy relies heavily on oil export revenues, with
such revenues representing around 80-90 percent of total export earnings and 40-50 percent of the
government budget. Strong oil prices the past few years have boosted Iran’s oil export revenues
and helped Iran's economic situation. For 2005, Iran's real GDP increased by around 6.1 percent.
Inflation is running at around 16 percent per year, though unofficial estimates place the figure at 4050 percent. Iran’s oil export revenues have increased steadily, from $32 billion in 2004, to $45.6
billion in 2005, with 2006 became at $46.9 billion. In 2007 it became more than $60 billion.
Due to capital intensive nature of renewable systems its large scale application is a burden to the
national economy and therefore a supportive international financing mechanism is required.
RE technologies are mostly immature and need improvement as well as breakthrough which is
only possible through regional and international research cooperation and assistance.
We must work harder to prepare ground for private sector effective investment in the field of
renewables which is a key factor in making RE cost-effective and competitive as well as bring
about innovation, creativity and dynamism.
Iranian Ministry of energy
Agence France Presse
AP Worldstream
APS Review Gas Market Trends
BBC Summary of World Broadcasts
CIA World Factbook
Dongsheng Zhu1, Xinfang Li1, Xianju Wang1,2, Nan Wang1, Zhengdong Wang3, Shandong Tu3
Key Lab of Enhanced Heat Transfer and Energy Conservation,Ministry of Education,School of
Chemical and Energy Engineering, South China University of Technology, Guangzhou, Guangdong
510641, PR China
Navy Arms of Services Command Academy, Guangzhou, Guangdong 510431, PR China
School of Mechanical and Power Engineering,East China University of Science and
Shanghai 200237,PR China
Nanofluids have been attractive for the last few years with the enormous potential to improve the
efficiency of heat transfer fluids. In this study, the thermal conductivity of Al2O3–water nanofluids
was measured by a Hot Disk Thermal Constants Anlyser, and the effects of pH value, sodium
dodecylbenzenesulfonate (SDBS) concentration and nanoparticle concentration on the thermal
conductivity of Al2O3 nanoparticles in deionized water were discussed. The results showed that the
thermal conductivity enhancements of Al2O3–water nanofluids are highly dependent on pH values,
the weight fraction of nanoparticle and SDBS dispersant concentration of nano-suspensions. The
Al2O3–water nanofluids with an ounce of Al2O3 have noticeably higher thermal conductivity than
the base fluid without nanoparticles, For Al2O3 nanoparticles at a weight fraction of 0.0015 (0.15 wt
%), thermal conductivity was enhanced by up to 10.1%, with an optimal pH value and SDBS
concentration for the highest thermal conductivity. Therefore, the combined treatment with both the
pH and chemical dispersant is recommended to improve the thermal conductivity for practical
applications of nanofluid.
Key words: Nanofluid, Alumina nanoparticle, Zeta potential, Thermal conductivity
Heating or cooling fluids are important to many industrial sectors, including transportation,
energy supply and production and electronics. The thermal conductivity of these fluids plays a vital
role in the development of energy-efficient heat transfer equipment. However, conventional heat
transfer fluids have poor heat transfer properties compared to most solids. In fact, an amount of
studies of the effective thermal conductivity of suspensions consisting of solid particles have been
conducted. However, these studies have been confined to those produced with millimeter or
micrometer-sized particles. The lack of stability of suspensions that involve coarse-grained particles
is undoubtedly a primary reason why the fluids with dispersed milli- or micro-sized particles have
not been applied to any industrial sectors.
It was proposed that fluids containing nanometer-sized particles can be a new class of
engineered fluids with high thermal conductivity. The nanoparticle suspended fluids, named
nanofluids, have been produced as the nanotechnology producing nanoparticles developed rapidly.
Extensive research conducted on the preparation and processing technology of nanofluids in recent
years has provided an opportunity to apply this emerging nanotechnology to thermal engineering. It
cannot only solve the problems such as sedimentation, cohesion and corrosion which happen
conventionally in heterogeneous solid/liquid mixture with millimeter or micrometer particles, but
also increase the thermal performance of base fluids remarkably. Suspensions containing a small
amount of metal (e.g., Cu) or nonmetal (e.g., SiC, Al2O3, and CuO) nanoparticles have been
reported to have substantially higher thermal conductivities than those of the base fluids[Hong et
al., 2006; Li et al., 2006; Li et al., 2007; Hwang et al., 2007]. Anomalously, Patel et al. [Patel et al.,
2003] concluded that 5%–21% enhancement of the thermal conductivity of nanofluids for water
with citrate in the temperature range 30–60 Ԩ at a very low loading of 0.00026 vol% of Ag
particles. For a loading of 0.011% of Au particles, the improvement of thermal conductivity was
around 7%–14%. Kumar et al. [Kumar et al., 2004] reported an enhanced thermal conductivity of
about 20% for a nanofluid of only 0.00013% Au nanoparticles in water. Since such an anomalous
enhancement is expected to have wide applications in thermal engineering, nanofluids have
received considerable attention in thermal science and engineering. However, it is very difficult to
understand why nanofluids would have such a high thermal conductivity. Meanwhile, there are
large differences among the thermal conductivities reported by different researchers. Keblinski et al.
[Keblinski et al., 2005] further pointed out that the most exciting experimental results have not been
reproducible. Therefore, it is necessary to reconsider the reliability of the measurements reported so
In the present study we measured the thermal conductivities of the Al2O3-H2O nanofluids. The
effects of the weight fraction of the dispersed Al2O3 particles, the pH value of the aqueous
suspension, SDBS concentration, and the thermal conductivity of the base fluid on the enhanced
thermal conductivity ratio have been investigated. It is expected to provide guidance to design
nanofluids with excellent performance.
The Al2O3 powder (Alfa Aesar, Ward Hill, MA, USA) with alumina content > 99.9 % was
used in the study. It has a BET surface area of 100 m2 g-1, and a median particle size of 15-50 nm,
and a density of 3965 kg m-3. The surface area was determined using the Micromeretic Tristar 3000.
The particle size was measured using the X-ray Disc Centrifuge Particle Size Analyser Ver. 3.49.
The transmission electron micrograph (TEM) of alumina powder is shown in Fig.1. In the Fig.1,
there are a few larger particles, which are likely aggregates of the smaller ones, but the whole
distribution of the particles is relatively well-dispersed. The particles are basically spherical or near
An anionic surfactant, sodium dodecylbenzenesulfonate (referred to as SDBS) in chemical
grade, from Guangzhou Chemical Reagent Factory (China), was used. The water was purified by a
Milli-Q Academic Millipore system. The pH was controlled using hydrochloric acid (HCl) and
sodium hydroxide (NaOH) in analytical grade. All chemicals were used as received without any
further purification.
Fig1 TEM micrograph of nano-alumina
Fig.2 Schematic diagram of TPS
Measurement of zeta potential and particle size
The experiments were conducted using 0.05 wt% alumina nano-suspensions. Different
concentrations of the dispersant were added to the suspensions, which were stirred thoroughly and
ultrasonicated (KQ2200DE Ultrasonic Cleanser, 100W, Kunshan of Jiangsu Equipment Company,
China) for at least 15 minutes, 2–4 ml of suspensions was transferred into a measuring cell. Then
zeta potential and particle size were measured by a Malvern ZS Nano S analyzer (Malvern
Instrument Inc., London,UK). The measurement was run at V=10 V, T=25°C with switch time at
t=50 s. Each experiment was repeated at least ten times to calculate the mean value of the
experimental data. The pH value of system was adjusted with HCl and NaOH solution by precise
pH Meter (PHS-25, China).
Transient plane source (TPS) theory
Thermal conductivity of nanofluids is measured by means of the TPS method[Rodríguez-Pérez
et al., 2003]. In this method, the TPS element behaves both as temperature sensor and heat source.
This novel method offers some advantages such as fast and easy experiments, wide range of thermal
conductivities (from 0.02 to 200 W/m K), no sample preparation and flexible sample size.
The TPS element consists of an electrical conducting pattern of thin nickel foil (10 μm) in the
form of double spiral, which resembles a hot disk, embedded in an insulating layer made of
kapton (70 μm). (see Fig.2)
A Kapton insulated probe is dipped into the suspensions. A constant electric power is supplied
to the sensor and the increase in temperature ΔT(τ) is calculated from the variation in the sensor
resistance with time R(t) by using the equation:
1 ⎛ R (t ) ⎞
− 1⎟⎟
α ⎝ R0
where R0 is the hot-disk resistance in the beginning of the recording (initial resistance), α is the
temperature coefficient of resistance of the nickel foil, and ΔT(τ) is the temperature increase of the
sensor expressed in terms of an only variable τ, defined as:
ΔT (τ) =
, θ=
where t is the measurement time, θ is the characteristic time, which depends both of parameters of
the sensor (a is the sensor radius) and the sample (k is the thermal diffusivity of the sample).
In the case of a disk geometry, consisting of m concentric ring sources, an exact solution of
equation (1) is possible. The increase of temperature is:
ΔT(τ) = P0 ⎛⎜ π 2 aλ ⎞⎟ D(τ)
where P0 is the total output power and D(τ) is a geometric function. By fitting the experimental
data to the straight line given by equation (3), the thermal conductivity can be obtained by
calculating the value of slope for the fitting line( P0 / ⎛⎜ π 2 aλ ⎞⎟ ).
Measurement process of the thermal conductivity of nanofluids
Different concentration nano-suspensions were prepared, which were stirred thoroughly and
ultrasonicated (KQ2200DE Ultrasonic Cleanser, 100W, Kunshan of Jiangsu Equipment Company,
China) for half of an hour. A Kapton insulated probe (design 7577) was successively dipped into
the different nano-suspensions. The diameter of the sensing spiral in the probe was about 2.001 mm
and the Kapton insulation on both sides of the spiral had a thickness of 13 μm. Then the thermal
conductivity was measured by a Hot Disk Thermal Constants Anlyser (Hot Disk Inc., Uppsala,
Sweden). The measurement was run at V=0.02 V, T=25°C with switch time at t=5 s. Each
experiment was repeated at least ten times to calculate the mean value of the experimental data. The
pH value of system was adjusted with HCl and NaOH solution by precise pH Meter (PHS-25,
Influence of pH on thermal conductivity of Al2O3-H2O nanofluids
Fig.3 shows the change of zeta potential and particle size for Al2O3-H2O suspensions with SDBS
dispersant as a function of pH. According to the zeta potential values of alumina powders, pH 8.0-9.0
can be selected as an operating pH for the suspensions with SDBS dispersants. Because, in the pH,
the absolute values of zeta potential for Al2O3-H2O suspensions with SDBS dispersants is higher, so
there are more surface charges around the particles. On the other hand, the particle size of alumina
particles in water is smaller at pH 8.0-9.0, which can lead to better dispersion and stability of the
0.02 wt%
0.05 wt%
0.10 wt%
0.15 wt%
Zeta potential
Particle size
1.01 2
Particle size/nm
Thermal Zeta
Thermal conductivity ratio(k/kf)
0.02 wt%
0.05 wt%
0.10 wt%
0.15 wt%
Fig.4 Effect of pH on thermal conductivity of Al2O3‐H2O suspensions with SDBS dispersants. Fig.3 Effect of pH on zeta potential and particle size of Al2O3‐
H2O suspensions with dispersants. Fig.4 shows the change of thermal conductivity ratio for Al2O3-H2O suspensions with SDBS
dispersant as a function of pH. It can be seen that the thermal conductivity ratio increases as pH
increases from 3 to 8.0-9.0. As addressed in the literutures[Xie et al., 2003], such abnormal
enhancements are not explained by any pre-existing model. When the nanoparticles are dispersed
into water, the overall behavior of the particle–water interaction depends on the properties of the
particle surface. At the point of zero change (PZC), the repulsive forces among alumina particles is
zero and nanoparticles will coagulate together under this pH value. Therefore, when the pH value is
equal to or close to the PZC, alumina particle suspension is unstable according to the DLVO theory
[Russel et al.,1989; Hunter et al.,1987; Elimelech et al., 1995]. The hydration forces among
particles increase with the increasing difference of the pH value of a suspension from the PZC,
which results in the enhanced mobility of nanoparticles in the suspension. The microscopic motions
of the particles cause microconvection that enhances the heat transport process. So we attempt to
link the concept of surface charges to the change in thermal conductivity ratio of nanofluids. The
point to mention is that the charged surface sites seemingly provide much more effective passages
through which heats or phonons are going more efficiently. Lee et al. [Lee et al., 2006] showed that
surface charge states are mainly responsible for the increase of thermal conductivity in the present
condition by a surface complexatio model for the measurement data of hydrodynamic size, zeta
potential, and thermal conductivity. Xue et al. [Xue et al., 2003] showed from molecular dynamic
simulation that phonons felt much less resistance during travel from particle to liquid at stronger
interfacial bonding. Therefore, it looks reasonable to infer that optimizing pH or higher surface
charging condition facilitates phonon transport through increases of effective sites and transport
As depicted in Fig.3, as the pH goes away from the PZC, the surface charge increases because
of more frequent attacks to the surface hydroxyl groups and phenyl sulfonic group by potentialdetermining ions (H+, OH- and phenyl sulfonic group ), leading to an increase in the zeta potential
on the alumina powder surface. In this way, we can infer that there are more surface charges at pH
8.0-9.0, at which the dispersion behavior is better and the thermal conductivity is higher. As
expected, the surface charge is minimum at the PZC.
Influence of the SDBS concentration on thermal conductivity of Al2O3-H2O nanofluids
Thermal conductivity ratio( k/kf)
Thermal conductivity ratio(k/kf)
0.1 wt% Al2O3-H2O
Al2O3-H2O (H-C Model)
-0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16
Weight fraction(SDBS) /%
Weight Fraction /%
Fig. 5 Effect of SDBS concentration on thermal conductivity of Al2O3‐H2O suspensions . Fig. 6 Thermal conductivity ratio of Al2O3‐H2O suspensions as a function of solid weight fraction Fig.5 presents the thermal conductivity ratio of the Al2O3-H2O suspensions and the base fluid with
respect to the concentration of SDBS dispersant at pH 8.0-9.0. The weight fraction of the Al2O3H2O suspensions is 0.1 wt%. For the base fluid, the thermal conductivity ratio starts to decrease after a
certain value of the dispersant concentration. In the present experiments, the highest thermal
conductivity appears at 0.02 wt% water solution, which means when an ounce of SDBS dispersant
is added, the thermal conductivity at 0.02 wt% water solution is higher than that of pure water. In
the 0.1 wt% Al2O3-H2O suspensions, the trend of the variation of the thermal conductivity is very
similar to those in the base fluid with dispersant only. However, the thermal conductivity ratio
decreases slowly as SDBS concentration increases from 0.02 wt% to 0.10 wt%, and then decreases
very quickly with an increase in the SDBS concentration. Obviously, For the Al2O3-H2O suspensions
and the base fluid, when more SDBS is added into the systems, the thermal conductivity ratios
decrease very quickly. Owing to this trend, the addition of more dispersant seems noneffective in
the Al2O3-H2O suspensions.This is because the heat transfer area becomes narrower due to the
amount of the surfactants on the particle surface. Taking into account the combined effect of
dispersion behavior and thermal conductivity, the 0.10 wt% SDBS can be selected as an optimizing
concentration for the 0.1% alumina nano-suspensions.Therefore, it can be concluded that the
application of nanofluid with optimizing chemical dispersant is a better way among the considered
enhancement techniques in the viewpoint of the effectiveness of dispersion behavior and thermal
Influence of alumina concentration on thermal conductivity of Al2O3-H2O nanofluids
Fig.6 shows the enhanced thermal conductivity ratio of Al2O3-H2O suspensions with the
optimizing SDBS concentration as a function of the weight fraction of nanoparticle. pH 8.0-9.0 can
be selected as an operating pH for different weight fraction suspensions. The results show that an
ounce of nanoparticle suspensions have noticeably higher thermal conductivities than the base fluid
without nanoparticles. The thermal conductivity of Al2O3-H2O nanofluid is enhanced approximately
nonlinearly with the weight fraction of the alumina nanoparticle, it implies that Al2O3-H2O
suspensions can enhance the heat transfer performance. The maximum thermal conductivity
enhancements of up to 10.1% is observed at the 0.15 wt% suspension. Fig.6 also shows that the
comparison between the experimental data for Al2O3-H2O nanofluids and the values calculated
from Hamilton–Crosser model. The results show (see Fig.6) that the measured thermal
conductivities are higher than those calculated from Hamilton–Crosser model. Hamilton–Crosser
model is known as following equation [Yu et al., 2004]:
k eff
k p + (n − 1)k f − (n − 1)ϕ (k f − k p )
k p + (n − 1)k f + ϕ (k f − k p )
The notations of eff, p, and f represent nanofluid, particle and base fluids, respectively. φ and k
represent the volume fraction and thermal conductivity, respectively. Where n is the empirical
shape factor given by n = 3/ψ andψ is the sphericity. For the spherical and cylindrical shape particle,
the sphericity (ψ) is 1 and 0.5, respectively.
Bsides H-C model, other traditional models also underpredict the effective thermal
conductivity of nanofluids because these models do not completely account for particle size,
particle Brownian motion, nanolayering, and effect of nanoparticles clustering and pH condition,
which are important to nanoparticles in nanofluids. Thus, new assessments and mathematical
models of the effective thermal conductivity for nanofluids will need to be proposed.
This paper is concerned with the thermal conductivity of Al2O3-H2O nanofluid under different
nanoparticle concentration, pH values and sodium dodecylbenzenesulfonate (SDBS) concentration.
Key conclusions can be summarized as follows:
(1) We have shown that the pH of the nanofluid strongly affects the thermal conductivity of the
suspension. As the pH of the nanofluid goes far from the PZC, the surface charge increases
because of more frequent attacks to the surface hydroxyl groups and phenyl sulfonic group by
potential-determining ions (H+, OH- and phenyl sulfonic group ), and the colloidal particles get
more stable and eventually alter the thermal conductivity of the fluid. In this way, we can infer
that there are more surface charges at pH 8.0-9.0, at which the thermal conductivity is higher.
(2) The use of Al2O3 nanoparticles as the dispersed phase in water can significantly enhance the
thermal conductivity, and the enhancement increases with particle concentration under the
conditions of this work. The maximum thermal conductivity enhancements of up to 10.1% is
observed at the 0.15 wt% suspension. The experimental results are compared with H-C model
and indicate that the experimental results are remarkably higher than predicted by H-C model
for nanofluids. The comparisons also indicate that it is needed to take into account several
possible factors in enhancing the heat transfer performance of nanofluids, for example, higher
charging condition facilitates phonon transport through increases of effective sites and transport
(3) The thermal conductivity can be improved by adding optimizing SDBS dispersant. However,
the combined treatment with both the pH and chemical dispersant is recommended to improve
the thermal conductivity for practical applications of nanofluid.
The authors like to acknowledge the financial supports from the National Natural Science
Foundation of China (Grant No. 20346001), Program for New Century Excellent Talents in
University (Grant No. NCET-04-0826), Specialized Research Fund for the Doctoral Program of
Higher Education (Grant No. 20050561017), Post-doctor Foundation of China (Grant No.
20060400219) for the research work.
Hong, K., T.-K.Hong and H.-S.Yang. 2006. Thermal conductivity of Fe nanofluids depending on
the cluster size of nanoparticles. Applied Physics Letters 88: 31901.
Li, C.H. and G.P. Peterson.2006. Experimental investigation of temperature and volume fraction
variations on the effective thermal conductivity of nanoparticle suspensions (nanofluids).
Journal of Applied Physics 99: 084314.
Li,X.F., D.S. Zhu, X.J. Wang, et al.. 2006. Influence of CTAB on Stability of Copper Nanosuspensions, Proceedings of the International Symposium on Biophotonics, Nanophotonics
and Metamaterials, October 16-18, Hangzhou, China.
Li, X.F. and D.S.Zhu. 2006. Research progress and problems of heat-transfer properties of
nanofluids. Chemical Industry and Engineering Progress 25: 875-879.
Li, X.F., D.S.Zhu and X.J.Wang. 2007. Evaluation on dispersion behavior of the aqueous copper
nano-suspensions. J. Colloid Interface Sci. 310:456-463.
Hwang,Y., J.K.Lee and C.H.Lee. 2007. Stability and thermal conductivity characteristics of
nanofluids. Thermochimica Acta 455:70-74.
Patel,H.E., S.K. Das, T. Sundararagan, et al.. 2003. Thermal conductivities of naked and monolayer
protected metal nanoparticle based nanofluids: manifestation of anomalous enhancement and
chemical effects. Applied Physics Letters 83: 2931-2933.
Kumar, D.H., H.E. Patel, V.R.R. Kumar, et al.. 2004. Model for heat conduction in nanofluids.
Phys. Rev. Lett. 93:144301.
Keblinski, P., J.A.Eastman and D.G.Cahill. 2005. Nanofluids for thermal transport. Mater. Today 8,
Rodríguez-Pérez,M.A., J.A. Reglero, D. Lehmhus, et al.. 2003. International Conference
“Advanced Metallic Materials”, Smolenice Slovakia.
Xie, H., H.Lee, W.Youn and M.Choi. 2003. Nanofluids containing multiwalled carbon nanotubes
and their enhanced thermal conductivities. J. Appl. Phys. 94: 4967-4971.
Russel, W. B., D. A. Saville and W. R. Schowwalter. 1989. Colloidal Suspensions. Cambridge:
Cambridge University Press.
Hunter, R.J. 1987. Foundations of Colloid Science. Oxford: Clarendon Press.
Elimelech, M., J. Gregory, X. Jia and R.A. Williams. 1995. Particle Deposition and Aggregation:
Measurement, Modeling and Simulation. Oxford: Butterworths.
Lee, D., J.-W. Kim and B. G. Kim. 2006. A new parameter to control heat transport in nanofluids:
surface charge state of the particle in suspension. J. Phys. Chem. B., 110: 4323-4328.
Xue,L., P. Keblinski, S. R. Phillpot, et al.. 2003. Two regimes of thermal resistance at a liquid–solid
interface. J. Chem. Phys. 118: 337-339.
Yu, W. and S.U.S.Choi. 2004. The role of interfacial layer in the enhanced thermal conductivity of
nanofluids: A renovated Hamilton-Crosser model. J.Nanopart.Res. 6: 355-361.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF