Engineering, 2013, 5, 75-80
doi:10.4236/eng.2013.51b014 Published Online January 2013 (
Copyright © 2012 SciRes. ENG
Construction of a small scale laboratory for solar collectors
and solar cells in a developing country
Gentile Niko1*, Davidsson Henrik1, Bernardo Ricardo1, Gomes Joao2, Gruffman Christian3, Chea
Luis4, Mumba Chabu5, Karlsson Björn6
1 Energy and Building Design, Lund University, Lund, Sweden
*Corresponding Author email:
2 Solarus Sunpower Sweden AB, Stockholm, Sweden
3 Finsun AB, Älvkarleby, Sweden
4 Universidade Eduardo Mondlane, Maputo, Mozambique
5 University of Zambia, Lusaka, Zambia
6 Division of Energy Engineering, Mälardalen University, Västerås, Sweden
Received 2013
In the field of renewable energy, self-provided research in developing countries is barely present, but most welcomed.
The creation of know-how and self-development of technologies should reduce the dependence on industrialized coun-
tries for both materials and knowledge. This work presents technological and social issues related to the construction of
a low budget solar laboratory in Mozambique. The goal is to demonstrate that scientific level research can be carried
out in developing countries by using affordable solutions without sacrificing quality of the results. For this investiga-
tion, a solar laboratory was built in 2011 at Universidade Eduardo Mondlane of Maputo. The laboratory enables meas-
urements to evaluate solar thermal and photovoltaic-thermal hybrid collectors. Thanks to the flexibility of the sys-
tem, students and teaching staff can add/remove equipment and develop customised local research programs. In addi-
tion, a course on the principles of solar energy and collector simulation for local students was taught. The needed data
acquisition devices usually used in Europe were compared with cheaper and easy-maintenance ones. Calibration and
estimation of the uncertainty were successfully performed. Approximately 9% of inaccuracy in the measurement was
introduced by the cheaper equipment, but the investment cost was reduced by more than 90%. Other issues, results and
future recommendations are shown.
Keywords: solar thermal; solar hybrid; small-scale laboratory; scientific research; developing country; Mozambique.
1. Introduction
The flow of foreign aid to developing countries during
the year 2006 was 103.6 billion USD, while the global
amount disbursed in the past 50 years is over 2.3 trillion
USD [1]. The resulting benefits are often negligible,
even if not detrimental, as revealed by different studies
[2, 3]. Nevertheless foreign aid is still an essential condi-
tion to stimulate investments and to reduce the gap be-
tween industrialized and developing countries [4].
Moreover, the consequences of a drastic reduction in
foreign aid can be dramatic for the population (e.g.,
about half of the Government budget in Mozambique is
underpinned by foreign aid [5]). The discussion nowa-
days is consequently oriented toward improvements in
the efficacy of aid. Williamson argues that the possibility
that both donors and recipients do not have adequate
knowledge to achieve development goals must be ad-
dressed [6]. In this view, long term collaboration be-
tween Universidade Eduardo Mondlane (UEM) in Ma-
puto (Mozambique), University of Zambia and Lund
University (LU) permitted characterization of Mozambi-
can primary needs in terms of technology, as well as the
weaknesses of previous aid projects. By means of re-
sources mostly provided by the Swedish International
Development and cooperation Agency (SIDA), a low
budget solar laboratory for scientific research was thus
built. Several reasons oriented both donor and recipients
to this new approach to foreign aid.
Firstly, according to the 2012 appraisal, about 1.6 billion
people worldwide have no access to electricity, of whom
more than a third are living in the African continent [7].
Copyright © 2013 SciRes. ENG
Particularly in Sub-Saharan Africa less than a quarter of
the population can use electricity [8]. In Mozambique the
percentage is as low as 6% [9]). Lack of access to af-
fordable electricity is a major determinant of poverty
[10], confirmed by barely $500 of GDP/pp in Mozam-
bique [11]. In view of the availability of solar radiation
in the country [12], recent projects are largely based on
the installation by donors of stand-alone photovoltaic
systems in remote areas. The effectiveness of interven-
tion is strongly related to both local expertise and the
availability of material and spare parts, which are often
disregarded [13]. Furthermore, the systems themselves
are directly produced and imported from industrialized
countries, which is the cause of the continued depend-
ence status of the developing countries for further rural
electrification projects.
The absence of domestic production of photovoltaic and
solar thermal systems can be related, not only to lack of
economic resources but also to the lack of local scientific
research. Issues for African Universities concerns the
availability of funds and difficulties in receiving instru-
ments from industrialized countries because of bureau-
cracy, shipping costs and risks, and a very long waiting
time. The proposed solution is to replace expensive in-
struments with cheap, locally available and
easy-maintenance ones; precision and accuracy should
be only sacrificed inside a range of acceptability. By
offering knowledge and a basic set of tools, this work is
a contribution to the start-up for local scientific research.
The installed equipment allows general training for en-
gineers even outside the field of solar energy, e.g. data
logger programming for developers or improvement of
the solar tracking system for mechanical engineers. Na-
tional testing agencies for instruments can arise and ex-
ploit the laboratory as reference. As a direct conse-
quence, an increase in local patents could occur, with
general improvement of domestic production. In a virtu-
ous circle, economic growth should allow higher invest-
ment in research, which is an essential condition for the
transition from basic to high level scientific research.
Data-loggers, solar radiation, temperature and angle
sensors are part of the laboratory equipment. The system
was set up in the furtherance of easy customizations for
increased local researches in the future. At the present
time a prototype of concentrating Photovoltaic/Thermal
(PV/T) hybrid collector is being tested [14]. UEM can
perform reliable testing and get acquainted with the
tools, while LU can afford tests in very different climatic
In the first section of this paper the methodology fol-
lowed in choosing instruments and solutions is explained
in detail. The second section shows discussions and the
results of a comparison of the sensors, including a calcu-
lation example and related accuracy. The conclusions
highlight the results achieved and underline improvable
aspects for further similar projects.
2. Methods
2.1. Project schedule
The preliminary schedule of work was drawn up in LU
during the beginning of year 2010. The early design
phase of the project concerned the definition of several
“Involving”. Recipients must be completely involved
in the job achievement. Courses to local students
were given, an MSc student participated in the meas-
urement phase and the presence of personnel from
UEM was always requested during the building and
testing process. A local member was also assigned as
responsible for the facilities after the end of the pro-
“Customizable”. Recipients must be allowed to run
their own researches, also in other fields. Easy
adaptability of the lab to different acquisition devices
is fundamental. Adaptable data logger and sensor
were chosen.
“Cheap”, “Simple” and “Efficient”. The choice of
devices stressed the importance of affordable prices,
easy understanding and installation, without exces-
sively sacrificing the quality of measurements.
“Low maintenance” and “Locally available”. Sensors
and devices are not always available in Africa. Long
life and easy obtainable replacements are the distin-
guishing characteristics of a project with a long term
“Low water usage”. Water is a precious, rare and
expensive resource in Africa. Grid water supply is not
always guaranteed. A closed water circuit with a 250
l boiler was studied for the solar thermal part. Rather
than using fresh water from the grid, water can flow
through the system during the night to cool down the
During the second part of year 2010, the project passed
from the early design stage to a more detailed level. Dif-
ferent sensors and devices were tested in Lund and the
definitive adopted equipment was eventually chosen.
The team from LU reached Maputo in January 2011 and
worked for three months. The combination of both de-
lays in material delivery and the low availability of spare
parts in Mozambique, caused changes to the initial
schedule. Though the solar laboratory was successfully
built by the end of the visit, some planned works could
not be carried out (e.g. tracking system). Sensors were
installed to check ageing during the subsequent visit.
During the 12 weeks basic courses on solar energy were
given to UEM and University of Zambia students. They
covered topics such as: solar radiation basics and solar
conditions in Mozambique, basics on different collector
Copyright © 2013 SciRes. ENG
types and systems, optimization of solar collector sys-
tems and cost analysis for local conditions.
An additional visit by a Lund University team to Maputo
was made in November 2011. The main activities con-
cerned finalization of the laboratory construction, a
check on the ageing effect on the products and perfor-
mance of further tests. No additional courses were of-
fered, but an MSc student from University of Zambia,
Mr. Chabu Mumba, was fully engaged in the activities as
an exchange researcher. Works concerned the imple-
mentation of the HSAT and other minor improvements
to the piping system.
Further tests had the aim to assess ageing effects on both
the panels and sensors. Cheap devices were compared
with the scientific ones.
2.2. Choice of data acquisition devices
The tests performed during 2010 made it possible to find
devices with a satisfactory quality/price ratio. Manufac-
turers of the chosen products were subsequently involved
in the project. In the following list, “a)” indicate the ref-
erence device, “b)” the tested option.
1) Data Logger.
a) Campbell CR1000 DataLogger. Analog, digital and
pulse inputs are suitable for the adopted scientific data
logger. For the mean voltage input range ±2.5V, maxi-
mum resolution is 0.67 mV and measurable through up
to 16 single-ended ports. High accuracy, versatility and
reliability allowed this product to be spread worldwide
for scientific application. The price is approximately
1500 USD.
b) MELACS®. It enables stand alone data logging and
remote collection through the built in web server. Con-
nection of multiple loggers (e.g. to increase the number
of ports) is possible through the Ethernet port. Voltage
input range is fixed to ±3.3V, corresponding to 0.8 mV
of resolution and measurable through 8 channels. Pulse
and digital channels are also available. It works with
open source GPL software. Current price is about 260
2) Water temperature.
a) PT100 Class A. High precision temperature meas-
urements were carried out through a PT100 sensor with
immersed insert. The Class A definition guarantees the
accuracy of T=±(0.15+0.002·|T|), where |T| is the ab-
solute temperature in °C. Pentronic AB was the chosen
manufacturer, which supplied and tested 30 sensors ac-
cording to EN10204. In order to have similar offset in
measurements, the two PT100 with closer response dur-
ing the test were chosen for the ∆T measurements. De-
spite the benefit of fluid immersed measurement, appro-
priate plumbing adaptation is required (Fig. 1). Adaptors
are rather expensive, approximately 60 USD, and hardly
available in Mozambique. The price of the RTD itself is
approximately 56 USD.
b) LM35CZ. LM35CZ are precision integrated-circuit
temperature sensors produced by National Semiconduc-
tor Corporation. Voltage output is linearly proportional
to Celsius temperature with 0 mV as set point for 0°C
and +10.0 mV/°C scale factor; nonlinearity typically
below ±1.4°C is guaranteed over the full range of
55-150°C. Accuracy is ±0.4°C, hence ±4 mV, at 25°C,
up to a typical value of ±0.8°C in extreme conditions.
The price is of approximately 5 USD each. Copper paste
on the surface and good insulation around the pipe must
be carefully provided to have good thermal contact and
low heat losses. Indeed, the sensor could record the air
temperature in the proximity of the pipe instead of the
pipe surface temperature (Fig. 1). Since the device is not
specifically designed for water temperature measure-
ments, it can be successfully used for other applications.
Figure 1. PT100 (left) and LM35 (right) positioning
3) Solar radiation
a) Kipp&Zonen CMP 11. Scientific pyranometer cali-
brated after purchase according to the technical regula-
tions of World Meteorological Institute. Estimated com-
bined expanded uncertainty for the used device is ±1.4%,
corresponding to 8.67 µV/W/m2 of sensitivity at normal
incidence on horizontal pyranometer. Commercial price
is about 4,000 USD.
b) Finsun SRS1000. Basic pyranometer with sensing
element made of single crystal Si-cell. The output volt-
age is 100 mV when exposed to 1000 W/m2 solar radia-
tion. Sensitivity, offset and ageing tests were performed.
Commercial price is approximately 115 USD.
4) Flow meter. Kamstrup 10EVL-MP110 energy and
flow meter was adopted. No cheaper flow meter was
5) Concentrating thermal and PV/T-hybrid collectors.
Two prototypes of asymmetrical CPC-collectors, one
PV/T and the one thermal, were installed.
6) Solar tracker. A horizontal single axis tracker
(HSAT) was installed. The tracker is based on an engine
driven by Melacs® logger. The software controls the po-
sition of solar panels using time and location as input.
This solution allows a correct positioning without the use
of additional photodiodes-based trackers, thus cutting
costs and maintenance.
3. Results and discussion
Copyright © 2013 SciRes. ENG
3.1. Data logger
Acquisition devices were connected to both the data
loggers. The resolution of CR1000 is higher than that of
Melacs® and mismatching was expected especially for
the solar irradiance measurement. On the contrary, no
significant difference in recorded voltage was observed
during the tests. At this level of approximation the two
loggers can be considered equivalent. Campbell CR1000
values were used for the following calibrations for pre-
cautionary reasons.
3.2. Pyranometer
Solar irradiation must be properly measured. Three dif-
ferent pyranometers were installed:
Kipp&Zonen CMP11 LU, the reference pyranometer,
is the property of Lund University, provided with a
recent calibration certificate;
Finsun SRS1000 new, is the proposed economic
pyranometer, installed in November 2011 (test peri-
Finsun SRS1000 old, is the proposed economic
pyranometer, installed in January 2011 to test the
ageing effects after about one year;
The pyranometers were mounted on the same flat sur-
face, built-in to the solar collectors (Fig. 2). Hence, sen-
sors follow the panels in the solar tracking.
Figure 2. Two of the installed pyranometers: a) CMP11 LU
and b) new SRS1000
The ageing effect on the SRS1000 is negligible in terms
of differentials, but not in terms of absolute values. In-
deed, a fairly constant overestimation of approximately
5% of solar irradiation was observed in the aged
pyranometer. The same parameter for CMP11 is guaran-
teed by the producer to be lower than 0,5% per year
from. Further investigations have to be carried out to
fully understand the behavior of SRS1000.
The calibration was performed on the aged SRS1000
(Gsrs) through data collected during a sunny day (9th,
November 2011). The photosensitive surface of SR1000
lies slightly below the external edge. This causes reading
errors due to shadows when the solar incidence angle is
close to 90°, i.e. typically sunrise and sunset; random
deviations up to 50% from the reference values of
CMP11 (Gref) are registered in this situation. Calibration
attempts did not take in consideration or weighted less
low those solar irradiance values.
The least squares method was applied to the ratio be-
tween global solar irradiance from the reference sensor
and the tested SRS1000 calculated over the whole day.
The aim was to find a correction curve for which Gref/Gsrs
was as close as possible to 1. Records where Gref<400
W/m2 were excluded for best curve fitting at higher solar
irradiance levels. Linear and logarithmic least squares
were applied, but the results were not satisfactory. In-
stead a slightly more complex polynomial interpolation
was chosen and equation (1) found:
 
. (1)
Fig. 3 shows calibrated values of SRS1000 (Gcal) during
the observed day. Change in the tilt of panels caused the
step around half past ten. Equation (1) was applied to the
series of data recorded on site during November 2011.
The standard deviation of Gcal from Gref was σ=±6,1%.
The provided value is intended along the whole radiation
range, though it is strongly recommended not to rely on
Gcal<200 W/m2.
Figure 3. Calibration of pyranometer SRS1000
3.3. Water temperature sensors
Immersed resistance thermometers based on PT100 are
used as reference. They should guarantee more accurate
measurements than LM35 for three reasons: 1) according
to the product data sheets, the thermal resolution differs
by an order of magnitude, 2) the sensor is immersed in
the fluid, and 3) the devices used are tested and certified
by an accredited laboratory.
Calibration was performed to investigate contingent
measurement error and time shift due to the disadvanta-
geous position of LM35. Sensors were previously tested
and calibrated at LU solar laboratory in controlled condi-
tions. Various LM35 were tested against the same PT100
to check random and/or systematic errors. Negligible
variations between them were registered and a general
behaviour was spotted.
Copyright © 2013 SciRes. ENG
The collected data were treated in a single elaboration in
order to find a general law of LM35 behaviour against
PT100 reference. Different mean trends were isolated. In
particular, the comparison between LM35s and PT100
Figure 4. Calibration of LM35
1) Low temperature (up to 30°C). In both steady and
dynamic condition no significant variations of the
registered temperature were found (Fig. 4.a). LM35
temperature can be assumed correct and does not
need any additional calibration.
2) Fast temperature reversal. Slight time shift for LM35
is observed (Fig. 4.b), most probably due to the sen-
sor position (Fig. 1). Subsequently to deepened inves-
tigations on a significant number of fast temperature
reversals, the delay was settled to 1 minute.
3) Warming up to 90°C. Non-linear increasing of tem-
perature underestimation is shown by LM35. The gap
is constant when the rate of warming/cooling is in-
creased/decreased (Fig. 4.c). Thus, an unequivocal
empirical logarithmic relation between the tempera-
tures of LM35 (Tlm35) and the new calibrated temper-
ature (Tcal) was proposed.
  (2)
 
 .
By defining Tpt the temperature recorded by the ref-
erence PT100, equation (2) represents a least squares
interpolation for which the Tpt/Tcal ratio should be as
close as possible to 1. Conditional setting δ=1 at
Tlm3530°C avoid an increase in the mean deviation
when the system is running at low temperature for a
long time.
4) High temperature (>30°C) and steady conditions.
Temperature is underestimated, but the LM35 profile
is quite stable (Fig. 4.d). Equation (2) is still valid.
The quality of calibration equation (2) was evaluated
through on site measurement in Mozambique. Measure-
ments were performed continuously for 30 hours. During
this time the system was heated and cooled several times.
Though the boundary conditions, as well as the used
sensors, were slightly different, the standard deviation
was σ=±1.7%. Longer time series are needed in order to
be able to estimate the annual effects for using this type
of correction.
3.4. Calculation example
The energy gain rate q of a solar collector [15] can be
defined by the simplified equation (3):
, (3)
in which the optical efficiency (τα) and the heat loss co-
efficient U are features of the considered collector. In
this example, the related uncertainty of these constants is
neglected. G is the solar irradiance and T=Tmean-Tamb is
the difference between the mean temperature of the col-
lector and the ambient temperature, with
Tmean=(Tin+Tout)/2. Tin and Tout represent respectively the
inlet and outlet water temperature in the collector. By
introduction of the standard deviations, the equation (3)
, (4)
Derived standard deviations are calculated according to
the propagation of error law for independent and sto-
chastic variables [16]:
 . (5)
In (5) the standard deviation of Tmean is the result of Tin
and Tout standard deviations combination. For a generic
flat plate solar collector with (τα)=0,87, U=3,83 W/m2K
and when G=1000W/m2, Tin=50°C and Tout=100°C, the
rate of energy gain calculated with the installed cheap
sensors will be q=679±61,45 W/m2. Deviation is ap-
proximately 9% of the measurement. The effect of G in
(4) is predominant. If the equations used were mostly
influenced by T values, the results would be more accu-
rate. Hence, the example is precautionary.
Table 1. Cost and accuracy reduction for the example.
Q [W/M2]
Copyright © 2013 SciRes. ENG
Presuming that the reference sensors give correct values
without uncertainty, the proposed calculation can be
combined with the equipment cost (tab. 1).
4. Conclusion
Scientific research requires investments in materials and
resources. Universities of developing countries interested
in starting research cannot usually rely on adequate
funding. In the majority of the cases they are obliged to
give up and the knowledge gap constantly increases. This
study concludes that:
Scientific research in solar energy can be initialized
with fairly low costs. The equipment cost can be cut
by up to 95% through using alternative sensors with a
drop in accuracy of measurement of about 9%. Ap-
proximations in the results obtained with the pro-
posed instruments should be always estimated and
underlined in report and publications;
A large part of the approximation is due to uncertain-
ty in solar irradiance measurement. Even if the man-
ufacturer could improve the design of only the sensor,
its performance could be easily improved. For future
solar laboratory, the choice should consider the aims
and the needs of the research. Thus, the decision
could consist in finding the best fitting invest-
ment/accuracy ratio for each case. For the example
discussed in this article a pyranometer more accurate
than the SRS1000 is recommended.
It is strongly recommended to involve local members
from the early stages. It is very important to instill a
sense of belonging and to state clearly the hoped-for
long term aim. By investing in local research, the
weak points of the previous form of foreign aid, e.g.
pre-packed stand alone photovoltaic systems in rural
areas, could be overcome;
The executed program in Mozambique is intended to
be a suggestion for future similar projects. Different
sensors and building solutions have to be tested in
order to find the best cost-efficient solution. At the
present time, the results and experiences gained are
working as inputs for the design of modular solar la-
boratories to be built in universities and research in-
stitutes of several developing countries;
In order to make the information available and reduce
the knowledge gap, donors and recipients are en-
couraged to publish their work and results in open
access journals.
5. Acknowledgements
The authors are grateful to the Swedish International
Development Cooperation Agency for sponsoring the
[1] W. Easterly, “The white man’s burden: Why the west’s
effort to aid the rest have done so much ill and so little
good,” Penguin Books, New York, 2006.
[2] P.T. Bauer, “From subsistence to exchange and other
essays,” Princeton University Press, Princeton, 2000.
[3] W. Easterly, “The elusive quest for growth: Economists’
adventures and misadventures in the tropics,” The MIT
press, Cambridge, 2002.
[4] J.D. Sachs, “The end of poverty: economic possibilities
for our time”, New York, Penguin Books, 2005.
[5] P. Mulder, J. Tembe, “Rural electrification in an imper-
fect world: A case study from Mozambique,” Energy
Policy, Vol. 36, Issue 8, 2008, pp. 2788-2794, doi:
[6] C. R. Williamson, “Exploring the failure of foreign aid:
The role of incentives and information,” The review of
Austrian economics, Vol. 23, No. 1, 2010, pp. 17-33, doi:
[7] Alliance for Rural Electrification (ARE), “Energy access
in the world: facts and scenarios,”
[8] A. Eberhard et al., “Unpowered: The State of the Power
Sector in Sub-Saharan Africa,” The World Bank, Wash-
ington, 2009.
[9] The World Bank Group, “World Development Report
2010,” The World Bank, Washington, 2010.
[10] U. Deichmann, C. Meisner, S. Murray, D. Wheeler, “The
economics of renewable energy expansion in rural
Sub-Saharan Africa,” Energy Policy, Vol. 39, Issue 1,
2011, pp. 215-227, doi: 10.1016/j.enpol.2010.09.034
[11] The World Bank Group, “Data. GDP per capita,” 2011.
[12] M. Suri M., T. Cebecauer, T. Huld, E.D. Dunlop, L.
Wald, M. Albuisson, “Photovoltaic Solar Electricity Po-
tential in the Mediterranean Basin, Africa, and Southwest
Asia,” 2008.
[13] O. S. Ismail, O. O. Ajide, F. Akingbesote, “Performance
Assessment of Installed Solar PV System: A Case Study
of Oke-Agunla in Nigeri,” Engineering, Vol. 4, No. 8,
2012, pp. 453-458, doi: 10.4236/eng.2012.48059
[14] Bernardo R. et al., “Measurements of the Electrical Inci-
dence Angle Modifiers of an Asymmetrical Photovolta-
ic/Thermal Compound Parabolic Concentrat-
ing-Collector,” Power and Energy Engineering Confer-
ence and Engineering, Sanya, China, Dec 31, 2012 - Jan
2, 2013, unpublished.
[15] A. Ucar, M. Inalli, “A thermo-economical optimization of
a domestic solar heating plant with seasonal storage,”
Applied Thermal Engineering, Vol. 27, Issue 2-3, 2007,
pp. 450-456, doi: 10.1016/j.applthermaleng.2006.06.010
[16] J.R. Taylor, “An introduction to Error Analysis: The
Study of Uncertainities in Physical Measurements”, 2nd
Edition, University Science Books, 1997