Renewable Energy 21 (2000) 471±493

www.elsevier.com/locate/renene

Modelling of a thermosyphon solar water

heating system and simple model validation
Soteris A. Kalogirou a,*, Christos Papamarcou b
a
Department of Mechanical Engineering, Higher Technical Institute, P.O. Box 20423, Nicosia 2152,
Cyprus
b
Electricity Authority of Cyprus, P.O. Box 4506, Nicosia, Cyprus
Received 20 October 1999; accepted 17 February 2000

Abstract

A thermosyphon solar water heater consisting of two ¯at plate collectors of total
aperture area of 2.7 m2 and 150 l storage tank is modelled using TRNSYS. Simple
experiments were conducted in order to validate the model. During the experiments weather
conditions were measured every 10 min and integrated over an hour. The temperature of
the water in the storage tank was also measured at the beginning and at the end of the day.
The storage tank temperature rise was used to validate the model by using the actual
weather data as input to the program. Validation tests were performed for 25 days spread
over 6 months and the mean deviation between the predicted and the actual values of water
temperature rise is 4.7% which is very satisfactory. Subsequently, long term system
performance is estimated by using TRNSYS model run with the weather values of TMY
®le for Nicosia, Cyprus. The annual solar fraction obtained was 79% and the system could
cover all the hot water needs of a house of four people during the three summer months.
The maximum auxiliary energy was needed during the months of December and January
(about 280 MJ/month). In addition, an economic analysis of the system was carried out.
The pay-back time of the system was found to be 8 years and the present worth of life
cycle savings was found equal to C£ 161. 7 2000 Elsevier Science Ltd. All rights reserved.

* Corresponding author. Tel.: +357-2-305030; fax: +357-2-494953.

E-mail address: [email protected] (S.A. Kalogirou).

0960-1481/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 9 6 0 - 1 4 8 1 ( 0 0 ) 0 0 0 8 6 - 0
472 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

1. Introduction

Cyprus has no natural oil resources and relies entirely on imported fuel for its
energy demands. The only natural energy resource available is solar energy.
Cyprus has a very sunny climate with an average annual solar radiation of 5.4
kWh/m2 day (on a horizontal surface).
Solar water heating units are extensively employed in Cyprus. In fact the total
number of units installed are such that constitute Cyprus to be among the leading
countries in the world in this area. These units are mostly of the thermosyphonic
type. This type of solar water heater consists of two ¯at-plate solar collectors
having an absorber area between 3 and 4 m2, a storage tank with capacity of 150±
180 l and a cold water storage tank, all installed on a suitable frame. An auxiliary
electric immersion heater and/or a heat exchanger, for central heating assisted hot
water production, are used in winter during periods of low solar insolation.
Owing to the manufacture of solar water heaters mainly of the thermosyphon
type that has expanded rapidly in Cyprus, there is a need to study in depth and
model this type of systems. It is also required to validate the model using simple
physical experiments. In this way the model can be used to investigate the e€ect of
design changes and therefore improve the solar water heater performance.
Many computer software programs have been developed concerning the
modelling and simulation of thermal systems. The most popular are WATSUN,
TRNSYS and TSOL. As part of this study TRNSYS 14.2 is selected as the most
appropriate because this program allows detailed analysis of all components of the
system and it is widely accepted as giving accurate prediction of solar system
performance.
The objective of this work is to model a thermosyphon solar water heater with
TRNSYS and validate the model by performing a number of simple experiments.
Subsequently, the long-term system performance is presented by using TRNSYS
model run with the weather values of TMY ®le for Nicosia, Cyprus.
There have been extensive analyses of the performance of solar water heaters,
both experimentally and analytically by numerous researchers. Some of the most
important are shown here.
Gupta and Garg [1] developed a model for thermal performance of a natural
circulation solar water heater with no load. They represented solar radiation and
ambient temperature by Fourier series, and were able to predict a day's
performance in a manner that agreed substantially with experiments.
Ong performed two studies [2,3] to evaluate the thermal performance of a solar
water heater. He instrumented a relatively small system with ®ve thermocouples
on the bottom surface of the water tubes and six thermocouples on the bottom
surface of the collector plate. A total of six thermocouples were inserted into the
storage tank and a dye tracer mass ¯ow meter was employed. Ong's studies
appear to be the ®rst detailed ones on a thermosyphonic system.
Kudish et al. [4] in their study measured the thermosyphon ¯ow rate directly by
adapting a simple and well-known laboratory technique, a constant level device,
to a solar collector in the thermosyphon mode. The thermosyphon ¯ow data
 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493 473

gathered were utilised to construct a standard eciency test curve, thus showing
that this technique can be applied for testing collectors in the thermosyphon
mode. They also determined the instantaneous collector eciency as a function of
time of day.
Morrison and Braun [5] have studied system modelling and operation
characteristics of thermosyphon solar water heater with vertical or horizontal
storage tank. They found that the system performance is maximised when the
daily collector volume ¯ow is approximately equal to the daily load ¯ow, and that
the system with a horizontal tank did not perform as well as a vertical one.
Hobson and Norton [6] in their study developed a characteristic curve for an
individual directly-heated thermosyphon solar energy water heater obtained from
data of 30 days' tests. Using such a curve, the calculated annual solar fraction
agreed well with the corresponding value computed from the numerical
simulation. Furthermore, the analysis was extended, and they produced a simple
but relatively accurate design method for direct thermosyphon solar energy water
heaters.
Shariah and Shalabi [7] have studied optimisation of design parameters for a
thermosyphon solar water heater for two regions in Jordan represented by two
cities, namely Amman and Aqaba through the use of TRNSYS simulation
program. Their results indicate that the solar fraction of the system can be
improved by 10±25% when each studied parameter is chosen properly. It was also
found that the solar fraction of a system installed in Aqaba (hot climate) is less
sensitive to some parameters than the solar fraction of a similar system installed
in Amman (mild climate).

2. Description of the system

The system consists of two ¯at plate solar collectors, having a total surface area
of 2.7 m2 tilted at 408 from horizontal, a thermally insulated horizontal storage
tank of 150 l capacity and interconnecting piping. It also incorporates a cold
water storage tank of 1000 l capacity.
The construction of the solar collector is important and relative to the
operation and eciency of the whole system. The external casing of the collector
is made of high corrosion resistant galvanised steel sheet, sprayed with aluzinc
paint.
The casing is covered with a 4 mm thick low iron single glass and sealed with a
rubber gasket. The absorber plate is made of copper. High radiation absorption is
achieved by the use of black ®ne matt ®nish on the copper surface, which has a
high absorption coecient. The undersides of the absorber plate and the side
casing are well insulated to reduce conduction losses with 50 mm and 30 mm
®breglass insulation, respectively.
The two ¯at plate collectors are connected in parallel through the supply
headers, each employing 12 evenly spaced parallel copper pipes 15 mm in
diameter embossed by semi-circular grooves formed in the ¯at plate absorber.
474 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

Hot water is stored in the 150 l hot water storage tank. Great care has been
taken in the design and production of the tank since it is the only part of the
system which is directly connected to all other parts. It is made of copper and is
thermally insulated with polyurethane. It is also protected by a galvanised outer
shell 0.6 mm thick. Connections to the other parts of the system are made by
copper pipes. In addition, the storage tank has an electric immersion element of 3
kW for heating water in the event of low solar insolation. The speci®cations of the
system are tabulated in Table 1.

3. Modelling of the system

The proper sizing of the components of a solar system is a complex problem

which includes both predictable (collector and other performance characteristics)
and unpredictable (weather data) components. In this section it is indicated how
all the necessary components of the solar water heater can be formulated into a
single model with the aid and use of TRNSYS simulation program.
Computer modelling of thermal systems presents many advantages the most
important of which are the following:
1. Eliminates the expense of building prototypes.
2. Complex systems are organised in an understandable format.
3. Provides thorough understanding of system operation and component
interactions.
4. It is possible to optimise the system components.
5. Estimates the amount of energy delivery from the system.
6. Provides temperature variations of the system.
7. Estimates the design variable changes on system performance by using the same
weather conditions.
Simpli®ed analysis methods have the advantages of computational speed, low
cost, rapid turnaround, which is especially important during iterative design

Table 1
System speci®cations

Description Value/Type

Total aperture area 2.7 m2

Storage tank capacity 150 l
Riser tubes material Copper
Number of riser tubes 12
Absorber surface Painted mat black
Glass type 4 mm low iron glass
Collector insulation Fibreglass 30 mm sides,
Fibreglass 50 mm back
Auxiliary heater 3 kW electric element
 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493 475

phases, and are easy to use by persons with little technical experience. The
disadvantages include limited ¯exibility for design optimisation, lack of control
over assumptions and a limited selection of systems that can be analysed. Thus, if
the system application, con®guration or load characteristics under consideration is
signi®cantly non-standard, a detailed computer simulation may be required to
achieve accurate results.
The initial step in modelling a system is the derivation of a structure to be used
to represent the system. It will become apparent that there is no unique way of
representing a given system. Since the way the system is represented often strongly
suggests speci®c modelling approaches, the possibility of using alternative system
structures should be left open while the modelling approach selection is being
made. The structure that represents the system should not be confused with the
real system. The structure will always be an imperfect copy of reality. However,
the act of developing a system structure and the structure itself will foster an
understanding of the real system. In developing a structure to represent a system,
system boundaries consistent with the problem being analysed are ®rst established.
This is accomplished by specifying which items, processes, and e€ects are internal
to the system and which are external.

3.1. Procedure followed

The thermosyphon solar water heater mounted in Nicosia, Cyprus was ®rst
modelled with TRNSYS program. In the program all the system characteristics
are required like the collector performance indicators of slope and intercept of the
standard collector test, dimensions of all components and piping, hot water
storage tank size, distance between the various components of the system and
many others.
The model was then simulated with weather data gathered at the location where
the thermosyphon solar water heater under investigation is installed. These values
were used in the TRNSYS program in order to predict the system performance.
In order to validate the model the initial and end of day mean water temperatures
in the storage tank were also measured and the water temperature rise was used as
a validation parameter. This procedure was conducted for a number of
consecutive days for a few months. The outcome of the simulation process
(predicted results) were then compared with the actual test measurements in order
to determine the accuracy of the simulation program.
The temperature of the water in the storage tank was chosen to validate the
model. This is because this temperature is a very important parameter both for
the designer and for the end user. No draw o€ was considered to simplify the
validation process and the auxiliary electric immersion heater remained o€ during
the testing period. Also, to take into account the storage tank strati®cation e€ects,
the top and bottom temperatures were measured and the mean of the two was
considered as the mean storage tank water temperature. These temperatures were
measured by drawing o€ a small quantity (about 1 l) of water.
The system can then be simulated with the TRNSYS model using Typical
476 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

Meteorological Year (TMY) data constructed for Nicosia, Cyprus. TMY is

de®ned as a year which sums up all the climatic information characterising a
period as long as the mean life of the system. The TMY is considered as a
representative year for the Cypriot environment. Using this approach the long-
term integrated system performance can be evaluated and the dynamic system's
behaviour can be obtained. In addition, an economic analysis of the system was
performed.

3.2. TRNSYS simulation program

TRNSYS is an acronym for a ``transient simulation program'' and is a quasi-

steady simulation model. This program was developed by the members of the
Solar Energy Laboratory of the University of Wisconsin [8]. It is written in ANSII
standard Fortran-77. The program consists of many subroutines that model
subsystem components. The mathematical models for the subsystem components
are given in terms of their ordinary di€erential or algebraic equations.
With a program such as TRNSYS which has the capability of interconnecting
system components in any desired manner, solving di€erential equations and
facilitating information output, the entire problem of system simulation reduces to
a problem of identifying all the components that comprise the particular system
and formulating a general mathematical description of each.
Once all the components of the system have been identi®ed and a mathematical
description of each component is available, it is necessary to construct an
information ¯ow diagram for it. The purpose of the information ¯ow diagram is
to facilitate identi®cation of the components and the ¯ow of information between
them. Each component is represented as a box, which requires a number of
constant parameters and time dependent inputs and produces time dependent
outputs. An information ¯ow diagram shows the manner in which all system
components are interconnected. A given output may be used as an input to any
number of other components. A simpli®ed information ¯ow diagram for the
thermosyphon solar water heating system under investigation is shown in Fig. 1.
From the ¯ow diagram shown in Fig. 1 a deck ®le has to be constructed
containing information on all the system components, weather data ®le, and the
format the output is given.
Simulations generally require some components, which are not ordinarily
considered as part of the system. Such components are utility subroutines and
output producing devices. The type number of a component relates the component
to a Fortran subroutine, which models that component. Each component has a
unique type number. The unit number is used to identify each component (which
can be used more than once) in the deck ®le. Although two or more system
components can have the same type number, each must have a unique unit
number (like Type 14, Load shown in Fig. 1).
It is noteworthy here, that the deck ®le reads information provided by the
Typical Meteorological Year (TMY) weather data which was generated by
Petrakis et al. [9] for Nicosia, Cyprus. A copy of this ®le is modi®ed by employing
 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493 477

the actual weather data of the days when the experiments were performed and
used for the program validation.

3.3. Type 45: thermosyphon collector storage subsystem

This component models the thermosyphon system. This is the major component
of the present model as depicted in Fig. 1. It models a system consisting of a ¯at
plate solar collector, a strati®ed storage tank (either vertical or horizontal
cylinder), and a check valve to prevent reverse ¯ow, which employs water as the
working ¯uid.
The ¯ow in the loop is assumed to be steady state. The system is analysed by
dividing the thermosyphon loop into a number of segments normal to the ¯ow
direction and applying Bernoulli's equation for incompressible ¯ow to each
segment.
Flow rate is obtained by numerical solution of the resulting set of equations.
Strati®cation in the tank is modelled using Type 38 algebraic component, which is
embodied in Type 45. The number of segments (nodes) in this model is not ®xed,

Fig. 1. Information ¯ow diagram for the thermosyphon solar water heating system.
478 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

but depends on many factors, i.e. the simulation time step, the size of the
collector, load ¯ow rates, heat losses and auxiliary input [10]. In this model
simulation starts with a certain number of segments. As hot water leaves the top
of the collector and enters the storage tank from a certain point at the top, it
mixes with water at this level if the temperatures are within 0.58C. If its
temperature is lower than that at the top by more than 0.58C it ¯ows down and
mixes with the water of a segment where the temperature is within 0.58C of it. In
the case when the temperature of water entering the storage tank is higher than
that at the top of the storage tank by more than 0.58C, a new segment is created
at the top, increasing the number of segments by one. When hot water is drawn to
the load, the same case applies for cold water from the mains entering the tank
from the bottom, this water mixes with that at the bottom if the temperatures are
within 0.58C of each other, otherwise a new segment is created.
The system's design parameters and the input data required for the operation of
Type 45 are outlined in Tables 2 and 3, respectively, which also list the actual
values adopted for the thermosyphon solar water heater under investigation. It
should be noted that the intercept and slope of the eciency against …Tci ÿ Ta †=IT
curve (Parameters 2 and 3 in Table 2) have been obtained from the performance
test of the solar panel of the collector carried out by the Ministry of Commerce
and Industry, Nicosia, Cyprus. The output information Type 45 can provide is
shown in Table 4.

4. Model validation

Once the TRNSYS model of the thermosyphon solar water heater under
investigation had been constructed the next step would be to simulate it with
actual weather data. Validation tests of the model can vary from simple to very
detailed. The latter ones are performed on an hourly basis by comparing the
system's actual and predicted performance for all the parameters predicted by the
program. In the present work a simple experimental procedure was chosen in
order to validate the model which does not require very complicated and
expensive equipment. Only one parameter is used for validation, i.e. the increase
of temperature in the storage tank. This parameter inherently includes all other
performance parameters of the system. For this reason physical experiments for a
number of consecutive days were performed at the location where the solar
heating system is installed. The weather parameters during experimentation were
monitored and their values were used as input to TRNSYS. Furthermore, the
simulated performance obtained with TRNSYS and measured weather parameters
were compared with the measured (experimental) performance. This model
validation study was conducted with the purpose to determine the degree to which
the TRNSYS model serves as a valid and representative model for the real
thermosyphon solar water heater.
Table 2
Type 45: system design parameters

Parameter Description Value

Ac Collector area (m2) 2.71

FR …ta†n Intercept of the eciency vs. (TciÿTa)/IT curve 0.792
FRUL Negative of the slope of the eciency vs. (TciÿTa/ITcurve (kJ/h m2 8C) 23.944
Gtest Mass ¯owrate per unit collector area for test conditions (kg/h m2) 96
b0 Incidence angle modi®er constant 0.1
b Collector slope (degrees) 4.0
LU for LU < 1 collector pressure drop calculated internally ÿ1
NR Number of parallel collector risers 12
dR Riser diameter (m) 0.0135
dH Header diameter (m) 0.026
H Header length (m) 1.9
Nx Number of collector nodes for thermal head calculations 3
Hc Vertical distance between collector outlet and inlet (m) 0.78
Ho Vertical distance between outlet of tank and inlet of collector (m) 0.78
di Diameter of collector inlet pipe (m) 0.02
Li Length of collector inlet pipe (m) 1.93
NB1 Number of right angle bends (or equivalent) in inlet pipe 2
Ui Loss coecient of collector inlet pipe plus insulation (kJ/h m2 8C) 5.1
do Diameter of collector outlet piping (m) 0.02
Lo Length of collector outlet piping (m) 0.46
NB2 Number of right angle bends (or equivalent) in collector outlet piping 2
Uo Loss coecient of collector outlet pipe plus insulation (kJ/h m2 8C) 5.1
Tank modes 1. Fixed inlet positions 2. Variable inlet positions 1
Vt Tank volume (m3) 0.15
S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

Ht Tank height (if vertical) or diameter (if horizontal) (m) 0.60

HR Height of collector return to tank above bottom of tank (m) 0.29
Cp Fluid speci®c heat (kJ/kg 8C) 4.19
rs Fluid density at standard conditions (kg/m3) 1000
kw Thermal conductivity of ¯uid in the tank (0 for no conduction) or e€ective thermal conductivity of ¯uid and walls (kJ/h m 8C) 2.3
(continued on next page)
479
 480

Table 2 (continued )

Parameter Description Value

Tank con®guration 1. Vertical cylinder 2. Horizontal cylinder 2

UA Overall UA value for tank (kJ/h 8C) 17.9
ri Ratio of insulation at the top and bottom of a horizontal cylindrical tank (=1 if tank is concentric with jacket) 1
TI Initial temperature of preheat portion of tank (8C) 18
Qhe Maximum rate of energy input by auxiliary heater (kJ/h) 10800
Ha Height of auxiliary heater above bottom of tank (m) 0.245
Hth Height of thermostat above bottom of tank (m) 0.245
Tset Set temperature (8C) 50
DTdb Thermostat temperature dead band (8C) 1
(UA )f Conductance of heat loss to ¯ue (kJ/h 8C) 0
S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493
 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493 481

Table 3
Type 45: input dataa

Input Description Value/Unit

IT Incident radiation (kJ/m2 h) [16,6]

IH Horizontal radiation (kJ/m2 h) [16,4]
Id Horizontal di€use radiation (kJ/m2 h) [16,5]
y Incidence angle (degrees) [16,9]
rg Ground re¯ectance 0.3(constant)
Ta Ambient temperature (8C) [9,5]
TL Temperature of replacement ¯uid from load (8C) [14,1]
mL Mass ¯owrate from load (kg/h) [11,2]
Tenv Environmental temperature for losses from storage (8C) [9,5]

a
The unit numbers refer to the number of output of a unit used as input to unit 45.

4.1. Description of the experimental set-up

A typical thermosyphon solar water heating system installed in Nicosia was

chosen to carry out the experiments. Its characteristics are shown in Tables 1 and
2.
Each experiment was started and ®nished by measuring the initial and end day
water temperatures in the storage tank. In order to take into account the storage
tank strati®cation e€ects the top and bottom water temperatures were measured
by a thermometer and the mean of the two was considered as the mean storage
tank temperature. To accomplish the above, the storage tank was modi®ed by
installing valves at its top and bottom points, and drawing a small quantity of
water (about 1 l) and measuring its temperature. Also, each experiment was
started by emptying the water tank and charging it with fresh tap water.

Table 4
Type 45: output information

Output Description

1. Th Temperature of hot ¯uid entering storage (8C)

2. Qu Useful energy from collector (kJ/h) (not including pipe losses)
3. TR Temperature return to collector (8C)
4. mw Flowrate to collector (kg/h)
5. TD Temperature of ¯uid delivered to load (8C)
6. mL Flowrate to load (kg/h)
7. Qenv Storage losses (kJ/h)
8. Qsup Energy supplied to load (kJ/h)
9. DE Change in internal energy of storage since start of simulation (kJ)
10. QAUX Auxiliary energy input to tank (kJ/h)
11. QU Useful energy from collector (kJ/h)
12. Tav Average storage temperature (8C)
482 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

Furthermore, the auxiliary heater was not activated during the experimentation
and no water was drawn o€ from the system during daytime.
Solar insolation was measured by means of an Eppley pyranometer installed on
a horizontal surface near the thermosyphon solar water heating system. This
pyranometer had been compared with the Eppley group of reference standards.
The accuracy of its measurements is estimated to be 21 percent of the entire
temperature range employed.
The ambient conditions were measured by the means of temperature sensors
and wind velocity sensor (anemometer). During the experiments solar insolation,
ambient temperature and wind velocity were measured and recorded every ten
minutes and integrated over an hour. This time interval was identical to the time
interval employed in the study of Shitzer et al. [11] as well as Braun and Fanney
[12].

4.2. Model validation

The experiments were performed during December 1998 through May 1999 for
a total of 25 days. The number of consecutive days for each month when the
experiments were conducted are shown in Table 5.
This time span is more than what others have followed to reach to the same
conclusions concerning the accuracy of the simulation programs.
The weather data gathered, namely the horizontal global radiation, ambient air
temperature, and wind velocity were used to modify a copy of the TMY ®le,
which was subsequently used with TRNSYS model in order to predict the mean
water storage temperature. For simpli®cation of the validation process, no water
was drawn o€ from the storage tank, and the electric element remained inactive
during experimentation.
Table 6 tabulates a sample of the recorded data. The data refer to the average
hourly values of the weather conditions for the 23rd of February. The actual
(experimental) and predicted (modelled) initial and end day storage water
temperatures are also shown in Table 6 for comparison. The temperature rise …DT†
of the water in the storage tank is the parameter which is used for validation of

Table 5
Number of consecutive days when the experiments were conducted

Year Month Number of consecutive days

1998 December 3

1999 January 4
February 3
March 5
April 5
May 5
Total 25
Table 6
System performance and comparison of actual and model results (sample)

Time Ambient temperature Horizontal Global radiation Direct radiation Wind velocity Storage tank temperature Storage tank temperature
(8C) (W/m2) (W/m2) (m/s) (actual) (modelled)
(8C) (8C)

8.00 13.7 278.4 674.1 1 23.6 21.1

9.00 14.7 443.7 816.3 1.5
10.00 16.1 603.3 860.7 1
11.00 16.7 692.1 896.2 1.5
12.00 17.3 673.3 905.1 1
13.00 17.7 656.6 913.9 1
14.00 17.6 612.2 940.6 1.5
15.00 17.2 479.1 798.6 1
16.00 16.9 292.8 674.4 1.5 49.7 48.0
Daily temperature rise …DT): 26.1 26.9
S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493
483
484 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

the accuracy of the TRNSYS model. An attempt was made to have the actual
initial storage tank temperature as close as possible to the modelled one to avoid
possible errors.
A comparison of the water temperature rise between the actual and predicted
results together with their percentage di€erences for all the days considered in the
experiments are shown in Table 7. As can be observed the predicted results agree
well with the actual values and the mean percentage di€erence between the two is
within 4.68%, which is quite acceptable and reasonable.
From the results tabulated in Table 7 it can be concluded that the model
constructed for the thermosyphon solar water heating system is representative of
the real system and can be used con®dently to estimate its dynamic behaviour and
long term integrated performance. Also, it can be concluded that this modelling
approach can be used for an in-depth analysis of real systems and can assist in
e€orts for improving a real system by optimising the parameters a€ecting its
performance.

4.3. Possible causes of the di€erences between simulation and actual results

As can be seen from Table 7 in a few days the percentage di€erence between
the actual and the modelled temperature di€erence is near or above 10%. The
reason for this deviation may be due to weather conditions, which were very
unstable during those days. This means that the mean hourly values used as input
to the program may not be representative of the actual input to the system. Also
during those days, by re®lling the tank with fresh tap water at the beginning it
was not possible to achieve initial temperatures very close to the modelled ones.
The relatively high values of some of the percentage di€erences are also due to the
magnitude of the values. For example the biggest di€erence of 21.3% corresponds
to a temperature di€erence of only 3.38C.
Other possible causes that have contributed to the deviation between the actual
and model results are the following:
1. The measuring equipment may not be as accurate as it should be or as
indicated by their manufacturers.
2. Determining the mean storage tank temperature by measuring the temperature
of the water at the top and bottom of the storage tank and taking the mean
value may not represent the actual mean storage tank temperature due to the
fact that the ``thickness of each temperature layer'' in the tank cannot be
determined. In order to estimate more representatively the mean storage water
temperature (taking into account the strati®cation e€ects) the cylinder should
be modi®ed by inserting a number of thermometers along its side at various
heights. This is even more dicult in the case of horizontal storage tanks.
However, the horizontal storage tanks exhibit very poor strati®cation which is
their main disadvantage. This type of tank is used mainly in an attempt to
reduce the height of the whole system.
3. The performance characteristic curve may not be as accurate as it should be.
Table 7
Comparison of actual and modelled results and their percentage di€erencesa

Test day Actual data Modelled data Absolute % di€erence between actual and modelled
water temperature rise values
Initial tank Final tank Initial tank Final tank
temperature (8C) temperature (8C) temperature (8C) temperature (8C)

December 20 26.1 [25.8±26.4] 43.7 [41.4±46.3] 27.2 44.4 2.27

December 21 30.2 [29.9±30.5] 48.5 [46.0±51.0] 27.2 46.5 5.46
December 22 29.3 [29.1±29.5] 44.8 [42.4±47.2] 23.7 42.5 21.29
January15 27.4 [27.3±27.5] 49.0 [46.5±51.5] 26.2 47.5 1.39
January16 28.5 [28.2±28.8] 48.7 [46.2±51.2] 27.7 48.5 2.78
January17 24.5 [24.4±24.6] 42.1 [39.9±44.3] 23.9 41.8 1.70
January18 25.2 [25.0±25.4] 42.4 [40.2±44.6] 24.5 42.0 1.74
February21 27.0 [26.8±27.2] 49.5 [47.0±52.0] 27.3 50.2 1.78
February22 26.4 [26.1±26.7] 54.2 [51.2±57.2] 19.3 49.2 7.55
February23 23.6 [23.5±23.7] 49.7 [47.2±52.2] 21.1 48.0 3.07
March 12 25.6 [25.3±25.9] 55.2 [53.2±57.2] 26.2 56.4 2.03
March 13 26.7 [26.6±26.8] 57.4 [54.2±60.6] 26.1 56.9 0.33
March 14 24.4 [24.2±24.6] 56.8 [53.7±59.9] 23.1 56.0 1.54
March 15 25.0 [24.9±25.1] 54.4 [51.9±56.9] 24.6 53.2 2.72
March 16 26.7 [26.4±26.9] 56.5 [53.9±59.1] 24.1 57.2 11.07
April 10 28.1 [27.9±28.3] 67.2 [63.2±71.2] 26.3 62.4 7.67
April 11 27.3 [27.0±27.6] 67.8 [63.7±71.9] 25.9 64.2 5.43
April 12 31.4 [31.2±31.6] 66.7 [62.8±70.6] 28.4 66.1 6.37
April 13 30.1 [29.9±30.3] 64.4 [60.7±68.1] 27.5 65.6 9.97
April 14 29.6 [29.3±29.9] 65.4 [61.4±69.4] 28.1 63.5 1.11
May 16 25.7 [25.5±25.9] 64.2 [60.6±67.8] 25.0 62.1 3.64
S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

May 17 29.1 [29.0±29.2] 61.7 [58.5±64.9] 30.4 61.9 3.37

May 18 31.4 [31.2±31.6] 63.6 [60.2±67.0] 30.1 60.4 5.90
May 19 26.4 [26.1±26.7] 60.7 [57.6±63.8] 26.1 59.6 2.33
May 20 30.3 [30.0±30.6] 67.4 [64.4±70.4] 30.0 65.4 4.58
Mean percentage di€erence: 4.68

a
485

The numbers in square brackets represent the storage tank lower and upper temperatures. From these the degree of strati®cation in the storage tank
can be seen.
486 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

This is very important as the performance of the solar collector determines how
much energy is intercepted from the quantity falling on the collector aperture.
4. Heat losses to the surroundings may be di€erent from those calculated by
TRNSYS program due to variation of the real insulation properties.

5. Simulation with TMY weather data

Following the validation studies presented in the previous section it can be

concluded that the TRNSYS model serves as a valid simulation tool for the
thermosyphon solar water heating system being considered. In the present section,
TRNSYS model is simulated with Typical Meteorological Year (TMY) data and
with a speci®c hot water consumption pro®le. With this approach the dynamic
behaviour of the system and the long-term integrated performance can be
evaluated. In addition, an economic analysis of the system is performed.
The selection of typical weather conditions for a given location is very crucial in
computer simulations for performance predictions and has led various
investigators to either run long periods of observational data or to select a
particular year, which appears to be typical from several years of data. Klein et al.
[13] have constructed the ``average year'' by selecting the monthly data from an 8-
year period which corresponded most closely to the average monthly insolation
and ambient temperature. The present investigation was conducted through the
use of a Typical Meteorological Year (TMY) for Nicosia, Cyprus. Petrakis et al.
[9] have generated the TMY from hourly measurements, of solar irradiance
(global and di€use on horizontal surface), for a seven-year period, from 1986 to
1992 using the Filkenstein±Schafer statistical method. The measurements were
recorded by the Cyprus Meteorological Service at the Athalassa region, an area at
the suburbs of the town of Nicosia.
The adequacy of using an average or typical year of meteorological data with a
simulation model to provide an estimate of the long-term system performance
depends on the sensitivity of system performance to the hourly and daily weather
sequences. Regardless of how it is selected, an ``average'' year cannot be expected
to have the same weather sequences as those occurring in the long-term. However,
the simulated performance of a system for an ``average year'' may provide a good
estimate of the long term system performance if the weather sequences occurring
in the average year are representative of those occurring in the long term or if the
system performance is independent of the weather sequences [13].
With regard to the thermal load, although the hot water demand is subject to a
high degree of variation from day to day and from consumer to consumer it is
impractical to use anything but a repetitive load pro®le. This is not quite correct
during the summer period, when the consumption pattern is somewhat higher.
However, during this period, the temperature requirement for hot water is not as
high as during winter. Consequently, the total thermal energy requirement is
 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493 487

reasonably constant throughout the year. For the present simulation, the hot
water consumption pro®le illustrated in Fig. 2 is employed, which assumes a daily
hot water consumption of 120 l at 508C for a family of four (30 l/person).

5.1. Simulation results

The TRNSYS model simulated with the TMY data and the speci®c hot water
consumption pro®le of Fig. 2 produced the results that follow. TRNSYS can give
output on a daily or monthly basis. The results for a particular day, chosen
randomly, are presented here as well as total monthly values. The former can give
a better picture of the hourly behaviour of the system whereas the latter o€ers a
more representative view of its long term integrated performance. The results for
April 21 were used to plot graphs correlating the various parameters of the
system, i.e., the thermosyphonic water ¯ow rate, the eciency and the outlet
temperature.
Fig. 3 shows the variations of thermosyphonic ¯ow and incident solar radiation
with time. As expected, the water ¯ow rate in the system follows the pattern of
variation of the solar insolation. The ¯ow rate increases during the morning hours
to reach a maximum of about 70 kg/h at solar noon, corresponding to
approximately 25.8 kg/h m2 of collector area and then starts to decline in the
afternoon, in a pattern similar to the solar radiation.
Fig. 4 illustrates the variation of the system's eciency and thermosyphonic
water ¯ow rate with time. The system eciency follows a pattern, which is slightly
di€erent from that for the ¯ow rate. It increases rapidly during the morning hours
to reach a maximum of about 50% at around 11 am solar time. During the
afternoon after 4.00 pm the eciency of the system drops to values below 15%.
It is worth noting that the peak of eciency does not coincide with that of
thermosyphonic ¯ow. Flow rate is not something that the user or the system
designer speci®es as no pump is used. It is something that happens naturally due
to the thermosyphonic e€ect.
The variation of the collector's outlet temperature and eciency with time are
indicated in Fig. 5. It is clear from the ®gure that both the collector's outlet

Fig. 2. Hot water daily consumption pro®le.

488 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

Fig. 3. Variations of incident solar radiation and water ¯ow with time (April 21).

temperature and eciency start rising in the morning following the variation
pattern of the solar radiation, but they reach their peak values at around 2 pm
and 11 am, respectively. The highest temperature reached at the collector outlet is
758C at 2 pm. The variation of the mean storage tank temperature is also shown
in this ®gure. The temperature reaches a peak at about noon and declines steadily
due to the hot water use (see consumption pro®le in Fig. 2) and its replacement
with tap water, tank heat losses and reduction of the solar energy input during the
afternoon hours.
The variation of the annual solar contribution is depicted in Fig. 6. In this
®gure, f, the solar fraction, is de®ned as the ratio of the useful solar energy
supplied to the system to the energy needed to heat the water if no solar energy is
used. In other words, f is a measure of the fractional energy savings relative to

Fig. 4. Variation of thermosyphon water ¯ow and system eciency with time (April 21).
 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493 489

Fig. 5. Variation of collector water temperature, mean storage tank temperature and eciency with
time (April 21).

that used for a conventional system. It can be calculated from the following
relationship:
QLOAD ÿ QAUX
fˆ …1†
QLOAD

Fig. 6 implies that the solar fraction is lower during the winter months and
higher, reaching 100%, during the summer months. The annual solar fraction is
determined to be 79%.
Fig. 7 shows the auxiliary energy needed per month. The maximum auxiliary
energy consumption is in January and December. During the summer months the
heat requirements are fully met by solar, hence no auxiliary is needed.
Fig. 8. illustrates the monthly energy ¯ows, which include the total radiation

Fig. 6. Predicted monthly and yearly solar contribution of the thermosyphon solar water heater.
490 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

Fig. 7. Predicted monthly auxiliary energy needed by the system.

incident on the collector (Qins), the useful energy supplied to the tank (Qu-tank),
the hot water energy requirements (HWload), the auxiliary energy demand (QAUX)
and the heat losses from the storage tank (QENV). As it can be seen from the
graph of total radiation incident on collector (Qins), the maximum value occurs in
the month of August (1900 MJ). The useful energy supplied to tank (Qu-tank) is
maximised in the month of April (600 MJ). It can also be seen from Fig. 8 that
there is a great di€erence between the solar radiation and the energy supplied to
storage. This di€erence indicates a high rate of heat loss from the collector and
the interconnecting piping. Another important point is the reduced incident solar
radiation and consequently the useful energy collected during the month of May.
This is characteristic of the climatic conditions of Nicosia and is due to the

Fig. 8. Energy ¯ow of the thermosyphon solar water heater.

 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493 491

development of clouds as a result of excessive heating of the ground and thus

excessive convection, especially during the afternoon hours.
From the curve of the energy lost from storage tank (QENV) it can be seen that
during summer months the energy lost from storage to surroundings is maximised.
This is due to the fact that at these months the temperature in the storage tank is
higher and consequently, the energy lost increases.
Referring to the curve of hot water load (HWLoad), there is a decrease of hot
water load demand during the summer months. This is attributed to the fact that
during the summer months the total incidence solar radiation is higher which
results in higher temperatures in the cold water storage tank. Consequently, the
hot water demand from the hot water storage tank during these months is
reduced. Also there is a reduced quantity of hot water demand from the consumer
due to the higher environmental temperatures.

5.2. Results of economic analysis

The economic scenario used in this project is that all the costs of the solar
system are paid from the beginning (i.e., no credit payments are assumed). The
thermal performance degradation of the system is assumed to be 1% per year, the
period of economic analysis is taken as 20 years (average life of locally produced
systems), whereas all the other percentage ®gures (in¯ation rates and market
discount rate) are mean values of the last 10 years. Electricity is assumed to be
used for the auxiliary heating element. Table 8 gives a summary of economic
®gures as calculated by TRNSYS.

6. Conclusions

The objective of this study was to model a thermosyphon solar water heater

Table 8
Results of economic analysisa

Output parameter Value

Initial cost of the system C£293

Resale or salvage value C£29
Rate of return of solar investment 13.6%
Years until undiscounted fuel savings = investment 8
Undiscounted cumulative net cash ¯ow C£747
Present worth of total costs with solar C£787
Present worth of total costs without solar C£948
Annualised total cost with solar C£15/GJ
Annualised total cost without solar C£17/GJ
Present worth of cumulative cash ¯ow C£161

a
Note: 1C£ = 1.82 US$ (October 1999).
492 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493

and validate the model through simple experiments. The modelling was performed
with TRNSYS program. The validation process revealed a good agreement
between the model and experimental (actual) results with a mean percentage error
of 4.68%. This means that the model constructed represents the real system and
can be used con®dently to estimate its dynamic behaviour and integrated
performance. It can also be extracted that TRNSYS program in general is
accurate enough and can be used with con®dence to model solar systems.
From the simulation of the thermosyphon model with TMY data and speci®c
daily hot water demand pro®le the following were deduced:
1. The thermosyphonic ¯ow of water for the particular day of April generally
follows the pattern of the variation of the incident solar radiation. The ¯ow
rate increases during the morning, reaches a maximum of about 70 l/h and then
declines during the afternoon.
2. The variation of the system's eciency does not follow the pattern of the ¯ow
rate. It reaches its peak value in the morning and then decreases slowly
throughout the afternoon.
3. The yearly solar contribution of the simulated system is as high as 79%.
4. During the summer months no auxiliary heating is required, i.e., the solar
fraction reaches 100%.
5. During the summer months the demand for hot water from storage tank is
reduced.
6. There is a decrease in solar incidence radiation during the month of May due
to special conditions encountered in Nicosia, Cyprus.
7. From the economic analysis it is found that the pay-back time of the system is
8 years and the present worth of life cycle savings is equal to C£161.
Apart from the purpose for which the TRNSYS simulation program was treated
in this work it could be used directly as a design tool for optimisation of the
design parameters a€ecting the performance of the system, in order to improve it.
In general the simulations of solar systems and especially simulations with the
TRNSYS program cultivates a change in culture in a company manufacturing
solar systems from being reactive to become proactive. It is highly encouraged
that this approach should be adopted by manufacturers of solar systems who
would like to increase their potential, as this has been proved to be the major
contribution in the pursuit of excellence. In the drawbacks of this approach is the
e€ort required to learn and use the program e€ectively and its cost.

References

[1] Gupta GL, Garg HP. System design in solar water heaters with natural circulation. Solar Energy
1968;12:163±82.
[2] Ong KS. A ®nite di€erence method to evaluate the thermal performance of a solar water heater.
Solar Energy 1974;16:137±47.
[3] Ong KS. An improved computer program for the thermal performance of a solar water heater.
Solar Energy 1976;18:183±91.
 S.A. Kalogirou, C. Papamarcou / Renewable Energy 21 (2000) 471±493 493

[4] Kudish AI, Santamaura P, Beaufort P. Direct measurement and analysis of thermosyphon ¯ow.
Solar Energy 1985;35:167±73.
[5] Morrison GL, Braun JE. System modelling and operation characteristics of thermosyphon solar
water heaters. Solar Energy 1985;34:389±405.
[6] Hobson PA, Norton B. A design nomogram for direct thermosyphon solar energy water heaters.
Solar Energy 1989;43:89±95.
[7] Shariah AM, Shialabi B. Optimal design for a thermosyphon solar water heater. Renewable
Energy 1997;11:351±61.
[8] Klein, S.A. et al. TRNSYS 14.2, A Transient Simulation and Program. Solar Energy Laboratory,
University of Wisconsin, Madison, WI, 1994.
[9] Petrakis M, Kambezides HD, Lykoudis S, Adamopoulos AD, Kassomenos P, Michaelides IM,
Kalogirou SA, Roditis G, Chrysis I, Hadjigianni A. Generation of a Typical Meteorological Year
for Nicosia, Cyprus. Renewable Energy 1998;13:381±8.
[10] Fanney AH, Klein SA. Performance of solar domestic hot water systems at the National Bureau
of Standards Ð Measurements and Predictions. ASME Solar Energy Engineering 1983;105:311±
21.
[11] Shitzer A, Kalmanoviz D, Zvivin Y, Grossman G. Experiments with a ¯at plate solar water
heating system in thermosyphonic ¯ow. Solar Energy 1979;22:27±35.
[12] Braun JE, Fanney AH. Design and evaluation of thermosyphon solar hot water heating system.
In: Proceedings of the American Solar Energy Society Annual Conference, Program in Solar
Energy Minneapolis, MN. 1983. p. 283±8.
[13] Klein SA, Beckman WA, Due JA. A design procedure for solar heating systems. Solar Energy
1976;18:113±27.