Performance Analysis of PV Green Roof Systems

Abstract

The common assumption is that plants underneath PV modules will cause a cooling effect resulting in a higher energy yield due to the negative temperature coefficient of PV power. The PV module temperature is measured to verify this thesis on a flat roof PV plant in Winterthur, Switzerland. Thirteen different mounting test field types with two different mechanical PV mounting systems are used, 20° or 15° tilted angles. Each type of mechanical mounting system is combined with planting beneath and with and without irrigation in the different test fields. The difference of the module temperatures, weighted according to the energy production, of all 13 analysed test fields is between ± 1.8 °C. This temperature difference results in a calculated energy yield difference of about ± 0.7 % for the used crystalline silicon modules. This value is in the range of the measurement uncertainty of the power measurement (± 1.2 %). Therefore the green roof has only negligible influence on the temperature reduction of the PV modules on the base of the used system components. However, a combination of PV and green roof is absolutely doable and recommendable provided that the mounting system is optimized for green roofs. Other more powerful benefits of such combination of green roofs and PV are the improvement of water retention in the city.

Full text

32nd European Photovoltaic Solar Energy Conference and Exhibition (EU PVSEC 2016), Munich, June 2016 5CO.14.3
T. Baumann et al. 1
PERFORMANCE ANALYSIS OF PV GREEN ROOF SYSTEMS
Thomas Baumann1*, Daniel Schär1, Fabian Carigiet1, Andreas Dreisiebner2,
Franz Baumgartner1
1 ZHAW, Zurich University of Applied Sciences, SoE, Institute of Energy Systems and Fluid Engineering,
Technikumstrasse 9, CH-8401 Winterthur, Switzerland, www.zhaw.ch/~bauf
*phone: +41 (0) 58 934 71 87; e-mail: thomas.baumann@zhaw.ch
2 Solarspar, Bahnhofstrasse 29, CH-4450 Sissach, Switzerland, www.solarspar.ch
ABSTRACT: The common assumption is that plants underneath PV modules will cause a cooling effect resulting in a
higher energy yield due to the negative temperature coefficient of PV power. The PV module temperature is
measured to verify this thesis on a flat roof PV plant in Winterthur, Switzerland. Thirteen different mounting test
field types with two different mechanical PV mounting systems are used, 20° or 15° tilted angles. Each type of
mechanical mounting system is combined with planting beneath and with and without irrigation in the different test
fields. The difference of the module temperatures, weighted according to the energy production, of all 13 analysed
test fields is between ± 1.8 °C. This temperature difference results in a calculated energy yield difference of about
± 0.7 % for the used crystalline silicon modules. This value is in the range of the measurement uncertainty of the
power measurement (± 1.2 %). Therefore the green roof has only negligible influence on the temperature reduction of
the PV modules on the base of the used system components. However, a combination of PV and green roof is
absolutely doable and recommendable provided that the mounting system is optimized for green roofs. Other more
powerful benefits of such combination of green roofs and PV are the improvement of water retention in the city.
Keywords: PV System, System Performance, Thermal Performance, Green Roof
1 INTRODUCTION
According to the energy strategy 2050 of the Federal
Council and Parliament of Switzerland electricity
originated from nuclear power has to be completely
covered by electricity from new renewable energies and
21 % attributed to PV power until 2050 [1]. Thereof, the
share PV corresponds to twelve square meters PV
module area per capita, which is available on the present
Swiss roofs.
Cities all over the world are promoting green roofs to
reduce further inner city heat generation. Chicago forced
to build over half a million square meters green roofs
stipulated by law in recent years [2]. The Swiss cities
Zurich, Basel, Bern, Winterthur, Lucerne and St. Gallen
have some requirements about goals for green roofs. In
some cities, the laws are set into force recently and they
are not same stringent everywhere [3]. Today, there are
nearly no PV installations on the green roofs, yet.
Because of the space requirement conflict between
PV on the roof and green roofs, it’s essential to combine
these two systems. The goal of this project is to analyse
the potential and benefits of PV panels with plants
underneath. Outdoor measurements of PV module
temperature and power are needed to estimate the
potential energy surplus if there is a significant cooling
effect observable.
2 APPROACH
2.1 Project goal
A project was set up to verify the PV performance on
a flat green building roof in Switzerland. This was
achieved with project partners consisting of green roof
experts and photovoltaic specialists [4]. One sub goal was
to check if the PV panels are cooled caused by the plants
on the green roof.
2.2 Setup
A PV plant of 78 kWp was installed on a flat roof
(84 m x 16 m, -3° azimuth) located in Winterthur
Switzerland and the situation is shown in Figure 1. The
roof is divided in 13 test fields. Eight test fields are
equipped with 20° tilted commercial ZinCo mounting
systems and five are equipped with 15° tilted commercial
Hilti mounting systems. The two mounting types are
shown in Figure 2.
Figure 1: Aerial photo of the test facility with the
installed PV capacity of 78 kWp located in Winterthur,
Switzerland. All modules have an azimuth orientation
of -3°.
Figure 2: The two used mounting systems on the PV
plant. On the left side the 20° tilted ZinCo mounting
system for green roofs and on the right side the 15° tilted
Hilti mounting system mostly used for flat roofs without
greening.

32nd European Photovoltaic Solar Energy Conference and Exhibition (EU PVSEC 2016), Munich, June 2016 5CO.14.3
T. Baumann et al. 2
Each test field consists of different soil conditions on
the ground including gravel, different substrates and
different substrate layer thickness, respectively. The
whole green roof part was seeded with the herbs seed
mixture “UFA-Kräuter Solardach CH” [5] from the Swiss
company “UFA Samen”.
2.3 Measurements
Several module temperature sensors are installed in
every test field, total 48 PT100 sensors over the whole
PV plant site. The measurement uncertainty of the
temperature sensors depends on the actual temperature
resulting in ± 0.55 °C (k=1) at 50 °C. The used
measurement electronic in the data logger contributes
another 0.3 °C to the measurement uncertainty (k=1). The
ambient temperature, the wind speed and direction and
the power of three PV modules are measured each
10 seconds as well. The measurement accuracy of these
three power measurements are ± 1.2 % (k=1). The
measurements campaign was started in autumn 2014. The
power of the PV modules is also measured and controlled
by the SolarEdge module optimizers in approximately
15 minute intervals with a measurement accuracy of
± 5 % (k=1). Additionally three modules were selected
and their power was measured with a resolution of
10 second as a reference to the SolarEdge measurement.
3 INNOVATION AND RELEVANCE
Sealed floor surfaces in cities, waste heat of vehicles
and houses which build heat islands enhance the warming
of the air during the day and decrease the cooling of the
air during night. Thus, summer nights are four to five
degrees warmer in cities compared to rural regions [3].
The economic and ecological advantage of green roofs is
undisputed. Today, the decision between PV and green
roof is often made on flat roofs. If it’s decided to install
PV, about 16 % of heating power is converted into
electricity and then bitumen or gravel is used regularly
under the PV panels. The green roof has several
advantages like water retention, reduction of peak water
runoff, protection of the roof seal, additional insulation,
cooling and air humidification of the ambient
atmosphere, habitat for animals especially insects,
filtering air pollutant etc [6]. With PV on green roofs, the
symbiosis of these two components can be combined. In
this project, the advantages and disadvantages of this
symbiosis is analysed with focus on the cooling effect on
the PV panels caused by the plants.
4 ANALYSES AND MEASUREMENT RESULTS
The energy weighted module temperatures were
analysed at the different test fields from 01.10.2014 to
31.05.2016. The days were classified into seven classes
according to the operating hours. In each of these classes,
the differences in the energy weighted module
temperatures between the test fields were analysed.
4.1 Data availability
Some days of data were missing due to a hardware
failure of the module temperature measurement system in
summer 2015. Furthermore, the days taken into account
are those that have every 10 second one measurement
between 08:30 in the morning and 16:00 in the afternoon
available. Thus, 527 days of measurement data are
analysed in the period from 01.10.2014 to 31.05.2016
resulting in an electricity production of 99 MWh for the
whole PV plant according to the SolarEdge web portal
data for these 527 days. Thus a specific AC yield of
974 kWh/kWp results for the annual period starting in
June 2015.
4.2 Calculation of the energy weighted module
temperatures
The temperature measurements are available in a
resolution of 10 seconds as well as three PV module
power measurements. The SolarEdge module power
measurements are available for all installed modules.
Unfortunately, the SolarEdge power measurements,
exported from the online monitoring portal, have no
constant time interval. The calculation of the energy
weighted module temperatures according to Formula 1
needs a synchronised data set between the SolarEdge data
and the temperature data.

, ∑
∙
,
∑,
° (1)
Three methods are analysed by calculating the energy
weighted module temperature using the SolarEdge power
data. The results is compared to the the energy weighted
module temperature of one module that has power and
temperature measurements with 10 seconds resolution
available. The methods are applied on to the SolarEdge
data set using different time periods.
- Method A “Linear interpolation”: The interpolated
value at a query point is based on linear interpolation
of the values at neighbouring grid points in each
respective dimension. The time resolution of the
received values is 10 seconds.
- Method B “Nearest neighbour interpolation”: The
interpolated value at a query point is the value at the
nearest sample grid point. The time resolution of the
received values is 10 seconds.
- Method C “closest own measurement point”: For
each SolarEdge point the closest own measurement
point is searched in a range of ± 10 seconds around
the SolarEdge point. The time resolution of the
received values is the same at the SolarEdge values.
Table 1 Energy weighted module temperature for one
module with 10 second power measurement (Ref)
compared with the three methods linear interpolation (A),
nearest interpolation (B) and closest own measurement
(C) calculated over different time periods.
Calculation
Method
Time period
Energy weighted module
temperature for 1 Module
Ref [°C] A [°C] B [°C] C [°C]
2 days
(08-09.11.2014) 21.08 21.20 21.20 21.22
1 month
(September 2015) 32.32 32.39 32.41 32.48
Whole analyse period
01.10.2014 to
31.05.2016
31.65 31.67 31.68 31.72
Table 1 shows the results for these three methods
compared to the values calculated with the 10 second
data (Ref). The linear interpolation method A provides
the most accurate result with a deviation of + 0.07 %
compared to the Ref value.
Figure 3 illustrates the result for the 4th August 2015.

32nd European Photovoltaic Solar Energy Conference and Exhibition (EU PVSEC 2016), Munich, June 2016 5CO.14.3
T. Baumann et al. 3
The green graph corresponds to the 10 second power
measurement, the SolarEdge power measurements are
indicated by black x-markers and the blue graph shows
the linear interpolated SolarEdge power measurements.
Figure 3: The power measurements for the 4th August
2015 are shown for the 10 second power measurement
(green), the SolarEdge power measurement (x) and the
linear interpolated SolarEdge measurement (blue).
4.3 Results
In Table 2 and Figure 4, the resulting energy
weighted module temperatures calculated according to
Formula 1 are shown for the whole analysed period and
for each test field. The average energy weighted module
temperature over all test fields is 31.7 °C. Table 2 shows
also the configuration of all test fields.
Table 2 Overview over the test fields data with name,
characteristics and energy weighted module temperature
analysed from the 1st October 2014 to the 31th Mai 2016.
Two different mounting systems are used, 8 test fields
from ZinCo and 5 test fields from Hilti, respectively. At
each type of mounting system, there is one field without
planting and there are fields with and without irrigation
(* = raincover).
Test field
Energy weighted module
temperature [°C]
Deviation from the
average [°C]
Mounting system
Irrigation
Planting
Vegetation layers [mm]
Vegetation material
Z50K 31.6 -0.1 ZinCo No No 50 Gravel
Z70R 30.4 -1.3 ZinCo No Yes 70 (60+10) RieFa/ZinCo
Z70 30.3 -1.4 ZinCo No Yes 70 ZinCo
Z100VA 31.8 0.2 Zi nCo No Yes* 70/100/130 ZinCo
Z100V 30.5 -1.2 ZinCo No Yes 70/100/130 ZinCo
Z70B 33.3 1.6 ZinCo Yes Yes 70 ZinCo
Z70BR 31.3 -0.4 ZinCo Yes Yes 70 (60+10) RieFa/ZinCo
Z100VB 30.6 -1.1 ZinCo Yes Yes 70/100/130 ZinCo
H70 33.5 1.8 Hilti No Yes 70 ZinCo
H100V 32.1 0.5 Hilti No Yes 70/100/130 ZinCo
H70R 32.7 1.0 Hilti Yes Yes 70 (60+10) RieFa/ZinCo
H70Ro 30.8 -0.9 Hilti No Yes 70 (60+10) RieFa/ZinCo
H50K 32.9 1.2 Hilti No No 50 Gravel
average 31.7
Figure 4: Average energy weighted module temperatures
in °C of each test field during the measurement period
from the 1st October 2014 to the 31th May 2016. The
overall averaged energy weighted module temperature is
shown in the last bar and corresponds to 31.7 °C.
The temperature differences result in a lower or
higher energy yield because of the negative temperature
coefficient (-0.39 %/K according to the module
datasheet). Figure 5 shows these energy yield differences
relative to the overall average.
Figure 5: Differences of the energy yield relative to the
average of all test fields in % calculated with the power
temperature coefficient (-0.39 %/K) and the energy
weighted module temperatures during the measurement
period from the 1st October 2014 to the 31th May 2016.
4.4 Distribution of the resulting energy weighted module
temperatures according to the operating hours of all test
fields
For each day, the nominal operating hours are
calculated resulting in 1356 hours over the 527 days. The
distribution of the number of days according to the
nominal operating hours is build and illustrated in
Figure 6.
Figure 6: Distribution of the number of days according
to the nominal operating hours.

32nd European Photovoltaic Solar Energy Conference and Exhibition (EU PVSEC 2016), Munich, June 2016 5CO.14.3
T. Baumann et al. 4
Next, the daily average energy weighted temperature
is calculated for each group of daily measurements that
have the same nominal operating hour. This is shown in
Figure 7.
Figure 7: Distribution of average energy weighted
module temperatures according to the nominal operating
hours of all test fields during the measurement period
from the 1st October 2014 to the 31th Mai 2016.
Figure 8 illustrates the distribution of the energy yield
difference compared to the average of each range with
the same nominal operating hour. This is calculated by
building the differences between the energy weighted
module temperatures and its corresponding average
values in the same nominal operating hour range and by
multiplying with the power temperature coefficient of
- 0.39 %/K.
Figure 8: The distribution of the energy yield difference
in % is calculated using the power temperature
coefficient and the energy weighted module temperatures
and it is compared to the average yield of each range with
the same nominal operating hours from the 1st October
2014 to the 31th May 2016.
5 CONCLUSION
The main result of the analyses during the first one
and a half year is that the difference of the energy
weighted module temperatures between all the thirteen
test fields is below ± 1.8 °C. Therefore, the energy yield
difference is below ± 0.7 % calculated with the standard
temperature coefficient for crystalline silicon modules.
This difference is in the range of the measurement
uncertainty of a power measurement of ± 1.2 %. It’s
relevant that the mounting system is optimised for green
roofs in order that the plants do not cause shades on the
PV modules and thus leading to expensive maintenance
[7].
Because of the space requirement conflict between
PV on the roof and green roofs, it’s essential to combine
these two systems. Due to the fact that the green roof has
a negligible influence on the temperature reduction of the
PV modules on the base of the used system components a
new follow-up project with Solarspar was startet. The
vertical East-West facing bifacial modules shown in
Figure 9 combine PV and green roof and allow a cost-
effective maintenance of the green roof.
Figure 9: The PV power plant “Seniorenzentrum
Wiesengrund” a) with vertical East-West facing bifacial
modules before module mounting b) schematic of
mounted bifacial PV modules and picture of the PV
powered lawnmower working autonomously beneath the
PV modules, located in Winterthur, Switzerland is a
follow-up project of ZHAW with Solarspar and will be
start operation on July 11th, 2016.
ACKNOWLEDGMENT
Thanks to the founding agency climate funds Winterthur
and project partners Solarspar, ZinCo, intelli solar
GmbH, RieFa – BAWES GmbH, Fenaco, A777 garden
design, Fritz Wassmann, public utility Winterthur,
PlantCare, ZHAW Life Sciences and Facility
Management and Markus Klenk, IEFE contributing to the
design of the new bifacial vertical module for project
“Wiesengrund”.
a)
b)

32nd European Photovoltaic Solar Energy Conference and Exhibition (EU PVSEC 2016), Munich, June 2016 5CO.14.3
T. Baumann et al. 5
REFERENCES
[1] Swiss Federal Office of Energy SFOE; “Botschaft
zum ersten Massnahmenpaket der Energiestrategie
2050”; http://www.admin.ch/opc/de/federal-
gazette/2013/7561.pdf
[2] City of Chicago Planning and Development;
“Chicago Green Roofs”;
http://www.cityofchicago.org/city/en/depts/dcd/supp_
info/chicago_green_roofs.html
[3] Michael Soukup und Stefan Häne; “Mit grünen
Dächern gegen die Hitze”; Tagesanzeiger vom
07.09.2015;
http://www.tagesanzeiger.ch/leben/gesellschaft/mit-
gruenen-daechern-gegen-die-hitze/story/24108679
[4] Markus Chretien, Ralf Walker, Marcel Okle,
Matthias Delker, Tobias Probst, Walter Schmidt,
Fritz Wassmann, Stephan Brenneisen PV green roof
project funded by the Klimafonds Stadtwerk
Winterthur, Switzerland, 2015
[5] UFA Samen; “UFA-Kräuter Solardach CH”;
http://www.ufasamen.ch/de/dachkraeuter-
mischungen/product/dachkraeutermischungen/ufa-
kraeuter-solardach-ch
[6] Berlin Senate for Urban Development
Communications; “Rainwater Management Concepts,
Greening buildings, cooling buildings”;
http://www.gebaeudekuehlung.de/SenStadt_Rainwate
r_en.pdf
[7] VESE Tagung April 2015; Presentation Tecton
“Pflegeintensive PV-Dächer”;
http://www.vese.ch/wp-
content/uploads/Tecton20150418.pdf