Combining photovoltaic modules and food crops: first agrovoltaic prototype in Belgium

Abstract

Agrovoltaic systems (combination of biomass production and electricity production by photovoltaics (PV)) are typically installed in locations with high insolation and/or arid climates in order to protect the crops against drought and sunburn. However, even in Belgium with a temperate maritime climate, summers are getting warmer and dryer, with reduced crop yields as result. This paper describes the first agrovoltaic prototype in Belgium. By use of a coupled simulation program developed in Python, a checkerboard panel arrangement was selected as an initial validation, in order to have a homogeneous ground radiation and crop growth. Potatoes were grown below the PV modules and the microclimate was measured. Results show lower temperatures below the PV modules and less transpiration and evaporation from crop and soil. The leaf area of the potatoes was larger below the PV modules which indicates an adapted light harvesting capability. Night-time temperatures were not seen to be improved under the agrovoltaic checkerboard structure, which indicates that this arrangement may not provide much protection against frost.

Full text

18th International Conference on Renewable Energies and Power Quality (ICREPQ’20)
Granada (Spain), 1st to 2nd April 2020
Renewable Energy and Power Quality Journal (RE&PQJ)
ISSN 2172-038 X, Volume No.18, June 2020
Combining photovoltaic modules and food crops: first agrovoltaic
prototype in Belgium
Brecht Willockx1, Bert Herteleer1 and Jan Cappelle1
1 Research Group Energy and Automation
Faculty of Engineering Technology, KU Leuven
Technologiecampus Ghent – Gebroeders Desmetstraat 1, 9000 Ghent (Belgium)
Phone/Fax number:+0032 472 321410, e-mail: brecht.willockx@kuleuven.be
Abstract. Agrovoltaic systems (combination of biomass
production and electricity production by photovoltaics
(PV)) are typically installed in locations with high
insolation and/or arid climates in order to protect the crops
against drought and sunburn. However, even in Belgium
with a temperate maritime climate, summers are getting
warmer and dryer, with reduced crop yields as result. This
paper describes the first agrovoltaic prototype in Belgium.
By use of a coupled simulation program developed in
Python, a checkerboard panel arrangement was selected as
an initial validation, in order to have a homogeneous ground
radiation and crop growth. Potatoes were grown below the
PV modules and the microclimate was measured. Results
show lower temperatures below the PV modules and less
transpiration and evaporation from crop and soil. The leaf
area of the potatoes was larger below the PV modules which
indicates an adapted light harvesting capability. Night-time
temperatures were not seen to be improved under the
agrovoltaic checkerboard structure, which indicates that this
arrangement may not provide much protection against frost.
Key words. Agrivoltaic, agrovoltaic, proof-of-concept,
dual land use, ground radiation simulation
1. Introduction
The European electricity system will have to be almost
carbon-free by 2050 in order to achieve the European Union
targets [1]; for this, increasing the share of renewable
production is a requirement. Especially in Belgium, where
there is a nuclear power phase-out planned in 2025 [2], the
need to increase the share of renewable energy is high.
The share of solar photovoltaic energy in Belgium in 2017
was 3.7 % of the total electricity production [3]. Currently,
44% of the arable area of Belgium is utilised for agriculture
and horticulture, yet the most likely increasing population
(and food demand) will make sure that food production
always has priority over paving that agricultural area for
solar parks [4]. Utility-scale solar farms require large
amounts of land, which is scarce in Belgium, given its high
population density. Moreover, on average, six hectares of
open space disappears every day [5]. The combination of
energy and crop production on the same land could offer a
solution. Agrovoltaics is an innovative concept,
implemented worldwide, with expertise in Asia and
several pilot projects in Europe [6].
Most of these agrovoltaic installations are built in arid
areas and places with a high amount of annual solar
insolation (>1300 kWh/m²) [6]. The shade created by the
PV array does not necessarily lead to loss of biomass yield.
The expectation is that the shade from the PV structures
may protect crops against drought stress and sunburn and
thus be beneficial to the crop yield [6]. However, even in
Belgium (annual mean insolation 1000 kWh/m²),
summers are getting hotter and dryer [7] with reduced crop
yields as a result [8], which explains the need to test
agrovoltaic installations. This paper describes the first
agrovoltaic prototype in Belgium and is structured as
follows: section 2 describes the designing process, section
3 the building phase and section 4 the results. Finally in
section 5, a general conclusion is made.
2. Preliminary design
According to Marrou et al. [9], the main changing
parameter in agrovoltaic conditions is the ground solar
radiation. This radiation has an influence on the
photosynthetic process and transpiration of the crop, two
elements with major influence on the crop yield [10]. The
wavelengths important for crop growth are between
400 nm and 700 nm of the solar spectrum, and is called the
photosynthetically active radiation (PAR).
In order to design an agrovoltaic system with a solid
theoretical foundation, a simulation program to calculate
solar radiation below PV modules was developed in
Python [11]. This 3D simulation tool, based on an
anisotropic view-factor model, is able to calculate the
amount of direct and diffuse PAR-light on each point
below the PV modules. Additionally, the PV energy yield
is calculated by use of functions from pvlib [12]. The
weather data for this simulation is obtained from the
hourly TMY generator created by the Joint Research
Centre of the European Commission [13].

A. First design: straight-line arrangement
A first simulation was made for a typical PV lay-out facing
South, placed in Beernem, Belgium (Latitude: 51.127°N;
Longitude: 3.301°E, Cfb Köppen-Geiger classification).
The array consists of standard modules of 1 m wide and
1.66 m long, with a distance of 1.66 m between two rows,
and tilted at 5°. Figure 1 shows the design of the first
simulation.
The results from the simulation in Figure 2 shows a division
of two areas: an area with a strong radiation reduction and
an area of almost no radiation reduction. This has as result
that the crops do not grow at the same rate, with a
heterogeneous crop yield as result. This makes it difficult to
harvest the field in one go [14]. To ensure homogeneous
crop growth, homogeneous radiation exposure is preferable.
B. Second design: checkerboard arrangement
A possible solution for this heterogeneous ground radiation
can be found by the design of photovoltaic greenhouses.
Previous work [15-17] suggests that a checkerboard
arrangement (as shown in Figure 3) will have a more
uniform radiation distribution.
This homogeneous radiation distribution below PV
modules is observed in Figure 4. Another observation is
that the lowest PAR value in checkerboard arrangement is
15% higher than in the straight-line arrangement, which
offers an advantage to limit crop yield losses, especially in
Belgium where the absolute solar insolation is not that
high.
C. Energy production
The generated PV power is simulated in Figure 5 for 10
PV modules of 280 Wp with a constant module efficiency
of 19% and system losses factor of 14% (i.e. temperature
effects are neglected at this stage). The annual generated
electricity is 2447 kWh (specific yield 874 kWh/kWp).
The ground coverage ratio of the is equal for both
arrangements (straight line versus checkerboard), what
results in an equal annual energy production, expressed
per hectare, while the crop production for both
arrangements is expected to vary.
3. Building the proof of concept
After the design optimization of the module layout in our
simulation, a proof of concept has been built at the
Beernem site used for simulation [18].
A. PV structure
Because this proof of concept is installed in an agricultural
area there are some practical requirements:
Figure 2: Annual percentage PAR at ground level in
comparison with shade free environment, straight line
Figure 1: Straight line PV arrangement
Figure 3: Checkerboard PV arrangement
Figure 4: Checkerboard annual percentage PAR at ground level
in comparison with shade free environment
Figure 5: Modelled PV power production

• No concrete anchoring, but temporary and
reversible anchoring
• Enough space between pillars to allow (manual)
farming practices
Therefore, the PV structure is 10 m wide and 2 m high to
ensure that the land beneath it is cultivable and the posts are
drilled 1.5 m deep without any concrete fixation, as shown
in Figure 6.
Figure 7 shows how two areas were defined: a reference
area, without the influence of PV modules and an evaluation
area below the PV modules in checkerboard arrangement.
B. Sensors
In order to measure the change in radiation between the
reference and evaluation area, PAR sensors (Apogee SQ-
214, accuracy ± 2 % and ± 5 % at solar zenith angles of 45°
and 75°) were installed in the reference and evaluation
areas. These measure the photon flux density, given in
µmol/m²/s. An example of such a PAR sensor is shown in
Figure 8. Temperature (accuracy ± 0.5°C) and humidity
sensors (accuracy ± 3%) are added at height of 1.8 m to
measure the micro-climatic conditions.
C. Crop selection
There is very little information about the shade tolerance
of crops below PV modules (with exception of Marrou’s
study [19]). However, a first guess can be made by looking
at the light response curve of the crop. The light response
curve is expressing the photosynthesis rate in function of
the received PAR light, as can be seen in Figure 9. At
higher photon fluxes, the photosynthesis rate reaches
saturation after which further increases in photon flux no
longer affect photosynthetic rates. Crops can be divided in
three groups according to the process of photosynthesis:
C3, C4 and CAM [20]. The light saturation point is
generally at a higher PAR levels for C4 crops (maize,
sugarcane,…) than C3 crops (wheat, rice, potato,.. ), which
makes C3 crops more suitable for agrovoltaic applications.
Furthermore, it is interesting to look at the susceptibility
of drought. The shade below the PV modules will probably
result in less transpiration, which leads to a higher soil
moisture, advantageous for the biomass yield of drought-
sensitive crops [21].
An example of a typical (shade resistant) C3 crop that are
sensitive to drought are potatoes [22]. For this proof-of-
concept, the Berber (Pedigree: Alcmaria x Ropta P 365)
variety was used and planted on 15 April 2019. Figure 10
shows an example of the growing potatoes below the PV
modules.
Figure 6: Drilling process
Figure 9: Light response curve [26]
Figure 7: Division of reference and evaluation areas
Figure 10: Potatoes growing below PV modules
Figure 8: PAR sensors to measure the change in solar
radiation below the PV modules

4. Results
The measured data was collected using a Siemens S7-1200
PLC system, equipped with a SQL database. The data was
collected at a time resolution of 5 minutes. The measuring
period ran between 24 April 2019 and 13 August 2019.
A. Radiation below the PV modules
Figure 11 shows that the potato plants were shaded between
8 am and 2 pm. An erroneous measurement is observed
between 2 pm and 3 pm, where the value of the PAR sensor
below the PV modules is higher than the value of the
reference PAR sensor. This can probably be explained by
the fact that the reference sensor has been shaded by the PV
structure post.
The measured reference PAR data is used to validate the
radiation model (from section 2). The Erbs model [23] is
used to decompose the photon density flux into its direct and
diffuse components. These direct and diffuse components
are used in the radiation model and compared with real
measured data from the PAR sensor below the PV modules.
It is clear in Figure 12 that the theoretical model follows
general trend of the measured values from the sensor. Only
when the measured reference sensor is shaded (which is
used as input for the validation), the deviation error is large.
B. Temperature and humidity
When the temperature drops, the relative humidity
increases, which is logical when the water vapour content
stays the same. Colder air does not require as much
moisture to become saturated as warmer air.
With this information in mind, is it interesting to look at
the difference between the reference area and evaluation
area in Figure 13 and Figure 14. The reduction in the
amount of radiation under the PV modules results in cooler
daytime air temperatures, averaging a cooling effect of
1.65°C below the PV modules. Even during the night, the
temperature below the modules remained lower than in the
reference area, which is contradictory to results from other
studies [9], [24], which claim that the temperature at night
is higher due to the shelter effect of the PV modules. The
difference can be explained by the fact that a checkerboard
arrangement, with 50% gaps between the modules, has a
less sheltered effect. This lower temperature could be
beneficial to the crop yield, where global crop yields are
expecting to be reduced due to the rising temperature as
result of global warming [15], although this would not
provide increased protection against frost.
At night, there is no transpiration of the crops in both areas
and the water vapour content is equal in both areas. This
has as result that the relative humidity is following the
trend from the measured temperature, where the relative
Figure 11: Measured PAR below PV modules and reference
area on 12 August 2019
Figure 12: Comparison between measured and modelled PAR
Figure 13: Measured dry bulb temperature in reference area and
below PV modules on 8 August 2019
Figure 14 Measured relative humidity in reference area and
below PV modules on 8 August 2019

humidity is lower below the PV modules. During the day,
even at a significant temperature difference, there is almost
no difference in relative humidity noted between reference
and evaluation area. This can be explained by the fact that
there is less transpiration (due to less solar radiation) below
the PV modules resulting in less water vapour content in the
air. The reduced transpiration results in a reduced water
demand, saving water for irrigation.
C. Crop results
After harvesting the potatoes, both the potatoes and the
foliage were weighed and compared between the reference
area and evaluation area. Remarkable was that the total leaf
area for potatoes below the PV modules was larger than the
reference area. It shows that potatoes have the ability to
adapt to shaded conditions and can compensate the
reduction of PAR radiation by a higher light harvesting
capability, in this case with a higher leaf area. This has also
been observed for lettuce [19].
5. Conclusions
A coupled simulation program that calculates the ground
radiation influenced by PV modules and the produced
electricity was developed in Python. By use of the
simulation program, a checkerboard arrangement was
modelled and subsequently tested with a prototype field.
The first agrovoltaic prototype was built, 2 m high and
without concrete anchoring. This prototype was equipped
with PAR sensors, to measure the change in solar radiation
below the PV modules. Additionally, temperature and
humidity sensors were added.
This checkerboard arrangement ensures a homogeneous
irradiation distribution, resulting in homogeneous crop
growth, which has been validated by the measured results.
Remarkable is that potatoes are showing an improved light
harvesting capability, with higher leaf areas.
Results of the measurements of the agrovoltaic prototype
show that the temperature below the PV modules is
consistently lower than in the reference area, which likely
will be beneficial for biomass production in moderate and
hot climates. The change in relative humidity indicates that
there is less evaporation and transpiration below the PV
modules, which protects crops against drought stress and
saves water for irrigation.
In light of climate change with higher temperatures,
agrovoltaic systems may protect the crops against drought
and high temperatures. Different agrovoltaic designs would
have to be modelled and considered to provide year-round
protection in moderate climates, where frost is still an issue
in winter.
References
[1]
European Commission, “Energy roadmap 2050,
page 5,” European Union, Luxembourg:
Publications Office of the European Union, 2012.
[2]
FOD Economie, “Wettelijke grondslag van de
kernuitstap in België,” FOD economie, [Online].
Available:
https://economie.fgov.be/nl/themas/energie/energie
bronnen/kernenergie/wettelijke-grondslag-van-de .
[Accessed 25 October 2019].
[3]
Statbel (Directorate-general Statistics - Statistics
Belgium), “Gross electricity production 2017,”
2017. [Online]. Available:
https://bestat.statbel.fgov.be/bestat/api/views/3a02
2b3b-b0fa-4d1e-bc91-62a0f632238c/result/HTML.
[Accessed 28 October 2019].
[4]
Vlaams infocentrum land-en tuinbouw, “Kleine
kans op grote zonneparken in agrarisch gebied,”
VILT, 04 January 2019. [Online]. Available:
https://www.vilt.be/kleine-kans-op-grote-
zonneparken-in-agrarisch-gebied. [Accessed 10
October 2019].
[5]
Flemish Government, Department of Agriculture
and Fisheries, “CHALLENGES FOR FLEMISH
AGRICULTURE AND HORTICULTURE,”
Flemish Government, 2018.
[6]
A. Weselek, A. Ehmann, S. Zikeli, I.
Lewandowski, S. Schindele and P. Högy,
“Agrophotovoltaic systems: applications,
challenges, and opportunities. A review,”
Agronomy for Sustainable Development, Vols.
https://doi.org/10.1007/s13593-019-0581-3, p. 20,
2019.
[7]
R. Vautard, O. Boucher, G. J. van Oldenborgh, F.
Otto, K. Haustein, M. M. Vogel and S. I.
Seneviratne, “Human contribution to the record-
breaking July 2019 heat wave in Western Europe,”
World Weather Attribution, 2019.
[8]
De Standaard , “Droogte decimeert aardappel- en
maïsoogst,” De Standaard , 13 April 2019.
[Online]. Available:
https://www.standaard.be/cnt/dmf20190413_04320
855. [Accessed 27 October 2019].
[9]
H. Marrou, L. Guilioni, L. Dufour, C. Dupraz and
J. Wery, “Microclimate under agrivoltaic systems:
Is crop growth rate affected,” Agricultural and
Forest Meteorology, vol. 177, no.
http://dx.doi.org/10.1016/j.agrformet.2013.04.012,
p. 18, 2013.
[10]
C. Campillo Mario, R. Fortes and H. P. Maria,
“Solar Radiation Effect on Crop Production,” in
Solar Radiation, InTechOpen, 2012, p. 29.
[11]
B. Willockx, S. Bram and J. Cappelle,
“Masterthesis: Agrivoltaics,” BRUFACE Brussels,
Brussels, 2018-2019.
[12]
W. F. Holmgren, C. W. Hansen and M. A.
Mikofski, “pvlib python: a python package for
modeling solar energy systems,” Journal of Open

Source Software, no.
https://doi.org/10.21105/joss.00884, p. 3, 2018.
[13]
European Commission Joint Research Centre, , “
PHOTOVOLTAIC GEOGRAPHICAL
INFORMATION SYSTEM : TMY Generator,”
European Commission Joint Research Centre, 21
September 2017. [Online]. Available:
https://re.jrc.ec.europa.eu/pvg_tools/en/tools.html#
TMY. [Accessed 18 October 2019].
[14]
M. Beck, G. Bopp, A. Goetzberger, T. Obergfell,
C. Reise and S. Schindele, “Combining PV and
food crops to agrophotovoltaic - optimazation of
orientation and harvest,” in 27th European
Photovoltaic Solar Energy Conference and
Exhibition, Frankfurt, Germany, 2013.
[15]
Z. Chuang, L. Bing, P. Shilong, W. Xuhui, L.
David, H. Yao, H. Mengtian, Y. Yitong, S. Bassu,
P. Ciais, J.-L. Durand, J. Elliott, F. Ewert, I.
Janssens, L. Tao, E. Lint, Q. Liua, P. Martreu and
C. Mül, “Temperature increase reduces global
yields of major crops in four independent
estimates,” Proceedings of the National Academy
of Sciences, vol. 14, no. 10.1073/pnas.1701762114,
p. 9326–9331, 2017.
[16]
M. Cossu, A. Cossu, P. Deligios, L. Ledda, Z. Li,
H. Fatnassi, C. Poncet and A. Yano, “Assessment
and comparison of the solar radiation distribution
inside the main commercial photovoltaic
greenhouse types in Europe,” Renewable and
Sustainable Energy Reviews, vol. 94, no.
10.1016/j.rser.2018.06.001, 2018.
[17]
A. Yano, M. Kadowaki, A. Furue, N. Tamaki, T.
Tanaka, E. Hiraki, Y. Kato, F. Ishizu and S. Noda,
“Shading and electrical features of a photovoltaic
array mounted inside the roof of an east-west
oriented greenhouse,” Biosystems Engineering,
vol. 106, no.
https://doi.org/10.1016/j.biosystemseng.2010.04.00
7, pp. 367-377, 2010.
[18]
T. Van Schuylenbergh, B. Willockx and J.
Cappelle, “Het potentieel van Agrivoltaics in
Vlaanderen,” KU Leuven , Technologiecampus
Gent, 2018-2019.
[19]
H. Marrou, J. Wery, L. Dufour and C. Dupraz,
“Productivity and radiation use efficiency of
lettuces grown in the partial shade of photovoltaic
panels,” European Journal of Agronomy, vol. 44,
no. 10.1016/j.eja.2012.08.003, pp. 54-66, 2012.
[20]
D. Rintoul, R. Bear, B. Snyder, M. Smith-Caldas,
C. Herren and E. Home, Principles of Biology,
pages 535-540,
https://legacy.cnx.org/content/col11569/1.27:
CNX, 2018.
[21]
H. Marrou, L. Dufour and J. Wery, “How does a
shelter of solar panels influence water flows in a
soil–crop system?,” European Journal of
Agronomy, vol. 50, no. 10.1016/j.eja.2013.05.004,
pp. 38-51, 2013.
[22]
P. Monneveux, D. Ramírez, A. Khan, R.
Raymundo, H. Loayza and R. Quiro, “Drought and
heat tolerance evaluation in potato,” in Triennial
Conference EAPR2014, Brussels, 2014.
[23]
D. G. Erbs, S. Klein and J. A. Duffie, “Estimation
of the diffuse radiation fraction for hourly, daily
and monthly-average global radiation,” Solar
Energy, vol. 28, pp. 293-302, 1982.
[24]
G. A. Barron-Gafford , M. A. Pavao-Zuckerman, R.
L. Minor, L. Sutter, I. Barnett-Moreno, B. Daniel
T, M. Thompson, K. Dimond, A. K. Gerlak, G. P.
Nabhan and J. E. Macknick, “Agrivoltaics provide
mutual benefits across the food–energy–water
nexus in drylands,” Nature sustainability, vol. 2,
no. https://doi.org/10.1038/s41893-019-0364-5, p.
848–855, 2019.
[25]
H. Fatnassi, C. Poncet, M.-M. Bazzano, R. Brun
and N. Bertin, “A numerical simulation of the
photovoltaic greenhouse microclimate,” Solar
Energy, no. 10.1016/j.solener.2015.07.019, 2016.
[26]
J. C. Lopez, “Influence of Light on Crop Growth,”
pthorticulture, 5 October 2018. [Online].
Available:
https://www.pthorticulture.com/en/training-
center/influence-of-light-on-crop-growth/.
[Accessed 10 October 2019].