On Module Temperature in Floating PV Systems

Ian Marius Peters André M. Nobre

Helmholtz Institute Erlangen Cleantech
Nuremberg Solar
Erlangen, Germany Singapore
[email protected] [email protected]

Abstract—Conventional wisdom has it that floating PV systems installation floating on a water reservoir and the other covering
have a performance advantage due to a, typically not further the roof of the inverter room. The layout of the site and images
specified, cooling effect of water. But how much does the presence of the installations as well as a list of measured parameters are
of water really contribute to the cooling of PV panels? In this case shown in Figure 1.
study, we explore the temperatures of two PV systems located next
to one another at a site in Cambodia. One of these systems is
mounted on a roof, and one is floating on a pond nearby. For these
two installations, we show simulated and measured temperature
data. Our analysis suggests that the presence of water does not
have a direct effect of the temperature of the floating installation.
Differences in the performance of the two sites are best explained
by different cooling due to wind. We additionally observe
differences in thermal inertia. Overall, we find that, at least
without further measures, water bodies only have an indirect
effect on PV panel temperature via other factors such as ambient
temperature or wind.

Keywords—Floating PV, Modelling and Simulation

I. INTRODUCTION
A floating installation is one of the rising trends in Fig. 1. Aerial image of the site with the two PV installations. Also included
photovoltaics (PV). Floating PV has surpassed the GW mark for are the location and functions of the different sensors that produced the data
used in this work.
cumulative installed capacity in 2018 [1] and analysts project a
further significant grow, with one source mentioning an The sites are equipped with multiple sensor stations,
expected additional 13GW until 2024 [2]. A major advantage of
measuring insolation, temperature, humidity, wind speed, wind
floating PV is the unlocking of new ground for PV installations,
direction and the output of the PV plant at several locations. For
making this concept especially attractive for areas in which
available land is scarce and water area is available. Examples the presented analysis, a subset of these measured parameters
are islands like Singapore and countries with large hydroelectric was used. Specifically: The temperature of the floating PV
dams like China, Japan and Korea [3]. The proximity to electric installation (TPV,water) measured at the rear side of a floating
systems at damns is considered a further advantage, as it module, the temperature of the modules on the rooftop (TPV,roof)
provides convenient access to the grid and unlocks synergies measured at the rear side of a floating module, the water
between the two power generators [1]. Proponents of floating temperature (Twater) measured at a depth of 0.3 m in the
PV cite additional advantages: covering water could reduce reservoir, and the ambient temperature (Tambient) measured close
evaporation losses and improve water quality by reducing to the shore of the reservoir. Insolation was measured on several
excessive algae growth. Quantitative evidence for either effect stations with differences that are insignificant for this study,
on actual PV installations is scarce; reduced evaporation of a hence only the value of one sensor close to the shore (Ishore) was
floating PV system was reported in [4], other papers frequently used. Humidity was measured directly at the two installations
cite experiments with other types of cover (for example [5]). (Hwater and Hroof). Wind speed (vwind) and direction (dwind) were
Passages on algae reduction frequently cite [6], a study that measured with a single sensor on top of the roof. Measurements
considers the effects of various illumination conditions on algae commenced in early-October 2019 and 215 days’ worth of data
growth that allow conclusions how floating PV panels may were collected for this study, with recordings ongoing.
affect it.
The “cooling effect of water” is another frequently II. OBSERVATIONS
mentioned advantage [7-9]. Yet, the exact mechanisms that Figure 2a shows the median temperature for each time-
contribute to this alleged cooling are generally not specified. In stamp for both PV installations. Contrary to expectations, we
this paper, we explore if and how the presence of water affects observe that the floating PV system during the day is
module temperature. We present our observations for two PV significantly hotter than the installation on the rooftop. The
installations located in Cambodia (~100 km south of Phnom median temperature difference in the three hours around noon is
Penh). The two sites are within meters from each other, with one 8.7 ± 2.2 K. Also shown in the figure are the median water- and

978-1-7281-6115-0/20/$31.00 ©2020 IEEE 0238

Authorized licensed use limited to: UNIVERSITAS GADJAH MADA. Downloaded on October 15,2023 at 14:59:31 UTC from IEEE Xplore. Restrictions apply.
ambient temperature. Also shown in the figure are the median system on the roof. The similar temperature of rooftop
water- and ambient temperature. An indication about which installation and ambient suggest that they are thermally coupled.
factor determines the thermal state of the PV systems can be The 1K difference between the floating installation and ambient
obtained when looking only at the nighttime temperatures, can be explained by radiative cooling [10]. We have no
recorded between midnight and 3am in Figure 2b. At night, the indication that the water temperature is relevant for determining
floating PV system really is cooler than the rooftop system by the temperature of the floating PV system.
about 1 degree with a small seasonal variation (lower part of
Figure 2b). We attribute this difference to a different thermal III. TEMPERATURE MODEL
inertia of the two systems. The cooling rate of the rooftop system We used the model by Veldhuis [11] to simulate the
is lower than that of the floating installation, as is evident from temperature of the roof-mounted PV modules. A comparison
the “tail” in the temperature curve of the rooftop system after between simulated and measured temperatures is shown in
6 pm (blue curve, Figure 2a). figure 3a. We find that the model provides a decent description
a) median temperature with an R2 of 0.99, a slope of 1 and an RMSE of 1.77K. The
model underestimates rare events with very high module
Nov 19 Jan 20 Mar 20 May 20
temperatures.
'Tnoon [K]

20 'Tnoon,med = 8.7 ± 2.2 K

10
a) rooftop system
Density
0
'Tnoon PV on water high
RMSE = 1.77 K
50 PV on roof 50
T [C]

40

Tsim [C]
ambient T 40
waterT
30

30
0 6 12 18 24
hour
b) night temperature 20 low
20 40 60
Tmeas [C]
36 PV on water
water
b) floating system
Tnight [C]

32 PV on roof
ambient
28
Reduced wind speed
60
24 slope = 0.99
RMSE = 2.3K
20 50
Tsim[C]
'Tnight [C]

TPV,roof - TPv,water 2
40

0
Nov19 May 20 30 Full wind speed
Jan 20 Mar 20
slope = 0.90
RMSE = 5.4K
20
20 40 60
Fig. 2. a) Median temperature for each timestamp of the floating PV installation
(red line), the rooftop installation (blue line), the ambient (cyan) and the water Tmeas[C]
(blue). b) Average temperatures between midnight and 3am for the two PV
installations (water – red, roof – blue), ambient (light blue, dashed) and water Fig. 3. a) Comparison between simulated and measured module temperatures.
(dark blue, dashed). The lower part shows the temperature difference between The plot shows the density of points with red colors indicating rare events and
floating and rooftop installation between October 2019 and May 2020. blue colors indicating frequent events. Comparison of simulated and measured
temperatures of the floating PV system (b). The simulation assuming full wind
speed is shown in brown, the one assuming a reduced wind speed in orange.
Ambient temperature continues to decrease during the night, The best agreement was achieved assuming wind speed was reduced to 20% of
causing a continuous cooling of the panels in both installations. its original value.
The higher thermal inertia of the rooftop system causes it to lag
behind in cooling compared to the floating installation, resulting We showed that temperatures for the floating PV system
in a higher temperature. Water, acting as a heat reservoir, is 3 K during the day are higher, and at night are similar to the rooftop
above ambient and ambient is 1K above module temperatures system. To explain this behavior, we consider altitude and wind
for the floating PV installation, while being close to the PV speed. The panels on the roof are about 3m above ground,

978-1-7281-6115-0/20/$31.00 ©2020 IEEE 0239

Authorized licensed use limited to: UNIVERSITAS GADJAH MADA. Downloaded on October 15,2023 at 14:59:31 UTC from IEEE Xplore. Restrictions apply.
whereas the floating modules are close to ground level – the shown). The floating system mostly appears to be in thermal
water surface in the reservoir is a little below ground level, and contact with the ambient air.
panels are close to the water surface and are mounted with a
Just because a PV system is floating does not mean it is
small inclination of 5 degree. Wind speed changes strongly with
cooler. The main determining factors for module temperatures
height close to the ground, due to friction but also due to
are ambient temperature, insolation and wind speed. Water
obstacles nearby like bushes and grass. It is therefore plausible
bodies may affect these parameters, depending on their size and
that wind speed is reduced at the floating system compared to
their location, and, hence, may influence PV installations
measurements in the roof.
indirectly. We find no indication, however, that there is a direct
To test this hypothesis, we performed a series of simulations heat exchange between panels and water. Direct thermal
for the floating PV system with varying wind speed. Assuming contact with water is possible, but requires additional measures
full wind speed, we find that the RMSE increases significantly (see, for example [12]). As shown in this study for the example
(by a factor of 2.35), and that the slope of the linear of a small water body, constructing a floating PV installation
approximation drops to 0.90. Assuming reduced wind-cooling may not offer any thermal advantages at all – in the discussed
results in a much-improved agreement between simulation and case most likely the lower altitude of the panels reduces cooling
measurement. The result is shown in Figure 3b. The best result from wind and results in a higher temperature compared to a
was obtained for a wind speed equal to 15% of its original nearby system on a roof.
value. Apart from the wind speed, also the smoothing constant,
Our analysis was done for an artificial, shallow water body
accounting for thermal inertia, was adjusted. The best fit was
that was fully covered by solar panels. We expect our findings
obtained for a value of 7 min compared to 35 minutes for the
to hold for larger water bodies and PV installations on the
rooftop installation. The reduction is plausible and reflects the
ocean, also. There are some important differences between a
smaller thermal inertia of the floating PV system, compared to
small land based water basin and the open seas, though: wind
the rooftop system with thermal coupling to the roof.
on open waters tends to be faster than on nearby land because
IV. SUMMARY & CONCLUSIONS of a long fetch – a lack of obstacles like vegetation, hills or
buildings. Floating PV systems, under these conditions will
Floating PV is one of the emerging trends in photovoltaics
benefit from cooler temperatures and local wind conditions
(PV). Especially countries in Southeast Asia with available
should be considered for optimal solar cell performance.
water areas have started to embrace this concept. As with many
new concepts, there are claims about the benefits of this ACKNOWLEDGMENT
approach that are currently not well substantiated by supporting
The authors thank Visal Yang and Lucas Ferrand from
data. Among the advantages of floating PV, frequently the Cleantech Solar for support on the maintenance of the data
“cooling effect of water” is mentioned, but is typically not acquisition monitoring setup and its associated sensors. This
specified further. In this study, we show that the mere presence work was supported by the Bavarian State Government (project
of water is not a determining factor in the temperature of a ‘‘PV-Tera – Reliable and cost efficient photovoltaic power
floating PV installation. In our analysis, we investigate a generation on the Terawatt scale’’, No. 44-6521a/20/5).
floating PV system and a rooftop PV system next to each other
in Cambodia. For this site, we find that daytime temperatures REFERENCES
of the floating installation are significantly above those of the
rooftop installation, with temperatures differences at noon of [1] World Bank Group, ESMAP and SERIS, Where Sun Meets Water:
8.7 ± 2.2 K. At night, we find that both systems exhibit similar Floating Solar Market Report – Executive Summary. Washington, DC:
temperatures. The temperatures of the water basin on which the World Bank (2018).
floating installation sits are above ambient during the night and [2] B. Beetz, 14 trends for 2019, PV magazine, December 31st 2018.
below during the day. When modelling the temperatures of [3] IRENA, Future of Solar Photovoltaic: Deployment, investment,
these systems we find that the water temperature has no direct technology, grid integration and socio-economic aspect (A Global
Energy Transformation: paper), International Renewable Energy
impact on the temperature of the floating PV installation at all, Agency, Abu Dhabi (2019).
and that the differences in temperature are best explained by a [4] M. E. Taboada, L. Cáceres, T. A. Graber, H. R. Galleguillos, L. F.
reduced wind speed at the site of the floating PV installation. Cabeza, R. Rojas, Solar water heating system and photovoltaic floating
cover to reduce evaporation: Experimental results and modeling,
There are further notable differences between the two Renewable Energy, 105 (2017), 601-615.
configurations. Apart from a higher daily operating [5] I. Craig, A. Green, M. Scobie, E. Schmidt, Controlling Evaporation Loss
temperature, the floating PV installation also exhibits a smaller from Water Storages, NCEA Publication No 1000580/1 (2005).
thermal inertia. This is visible in Figure 1b – the floating [6] M. Alam, S. Ohgaki, Evaluation of UV-radiation and its residual effect
installation reaches ambient temperatures at night faster than its for algal growth control, in M. Tomonori, H. Keisuke, T. Satoshi,
counterpart on the roof does. When fitting with our model, we Hiroyasu SatohA2-Tomonori Matsuo KHST, S. Hiroyasu (Eds.),
Advances in water and wastewater treatment technology, Elsevier
find a smoothing constant of 35 minutes for the rooftop system Science B.V., Amsterdam (2001), pp. 109-117.
and a value of 7 minutes for the floating installation. The [7] L. Liu, Q. Wang, H. Lin, H. Li, Q. Sun, R. Wennersten, Ppower
rooftop system appears to be in good thermal contact with the Generation Efficiency and Prospects of Floating Photovoltaic Systems,
roof itself; the two also show very similar temperatures (not Energy Procedia 105 (2017), 1136 -1142.

978-1-7281-6115-0/20/$31.00 ©2020 IEEE 0240

Authorized licensed use limited to: UNIVERSITAS GADJAH MADA. Downloaded on October 15,2023 at 14:59:31 UTC from IEEE Xplore. Restrictions apply.
[8] H. Liu, V. Krishna, J. L. Leung, T. Reindl. L. Zhao, Field experiences [11] A. J. Veldhuis, A. M. Nobre, I. M. Peters, T. Reindl, R. Ruther, and A.
and performance analysis of floaing PV technologies in the tropics, H. M. E. Reinder, An Empirical Model for Rack-Mounted PV Module
Progress in Photovoltaics 26 (2018), 957 – 967. Temperatures for Southeast Asian Locations Evaluated for Minute Time
[9] A. Sahu, N. Yadev, K. Sudhakar, Floating photovoltaic power plant: a Scales, IEEE JPV, 5 (2015), 774-782.
review, Renewable and Sustainable Energy Reviews, 66 (2016), 815 – [12] M. Willuhn, Floating PV’s watershed moment, PV Magazine
824. Photovoltaic markets and technology, June 2019.
[10] L. Zhu, A. Raman, K.X. Wang, M.A. Anoma, S. Fan, Radiative cooling
of solar cells, Optica, 1 (2014), 32-38.

978-1-7281-6115-0/20/$31.00 ©2020 IEEE 0241

Authorized licensed use limited to: UNIVERSITAS GADJAH MADA. Downloaded on October 15,2023 at 14:59:31 UTC from IEEE Xplore. Restrictions apply.