Floating PV solar on dam reservoirs

L. Deroo, ISL, France, and Chair of ICOLD Technical Committee on Emerging Challenges
and Solutions for Reservoirs
The increasing number of floating solar panels on hydro reservoirs around the world is demonstrating the increasing enthusiasm for this
technology. This article, based on the on-going work of one of ICOLD’s Technical Committees, discusses experience gained so far, as well as
the benefits and challenges associated with floating solar PV on dam reservoirs.

T
he adoption of floating PV solar (FPV) is rapid- The last annual meeting of ICOLD Committee T, on
ly increasing, as shown by the World Bank & ‘Emerging challenges and solutions for Reservoirs,’
SERIS report ‘Where Sun Meets Water: was dedicated to floating PV solar on dam reservoirs.
Floating Solar Market Report’ [World Bank Group, It brought together professionals from the dam and
20191,2]. Dam reservoirs are natural companions to this FPV industries. This paper presents some of the con-
technology. They provide the surface on which the clusions of the meeting, and suggests ways in which
panels can be installed. They may also provide addi- FPV can be used on large dam reservoirs.
tional benefits: energy transmission facilities, cooling
of the panels and potential for hybridization. But some 2 .Case studies and lessons learnt
progress still needs to be made in both the knowledge 2.1 Ten years of FPV projects
base and the design options for FPV on dam reservoirs
(FPV-DR), particularly as installing FPV on large dam Floating solar emerged less than 10 years ago, in 2011.
reservoirs involves some specific challenges. In ten years, more than 2 GW of peak capacity has
been installed, with installations ranging from less
than 100 kWp to up to 100 MWp. The pace of growth
1. The context is very high, and several very large FPV plants are
The World Bank report mentioned above illustrates the planned. Most of the FPV projects have been installed
rapid growth in FPV. on water bodies other than large dam reservoirs:
In addition, researchers at the DoE National ponds, quarry lakes, industrial water reservoirs, small
Renewable Energy Laboratory (USA) published a dam reservoirs. The largest projects to date utilize
report [US DoE, 20203] last September, stating that the lakes created by subsidence of coal mines.
combination of floating solar panels and hydropower Installation on dam reservoirs is less frequent.
could produce up to 40 per cent of the world’’s elec- However, some landmark projects provide useful
tricity. lessons. The Dami FPV was installed on the Dami hydro
Large dam reservoirs are beginning to be used for reservoir, in Vietnam and commissioned in June 2019.
FPV projects. It is important that this is done with due The FPV plant has a peaking capacity of 47.5 MWp, and
consideration for the specific features of reservoirs at the hydro installed capacity is 175 MW. The site selec-
large dams. Anchoring and mooring are more com- tion for this project considered the following advantages
Fig. 1. A graph plex, because of the deeper water and larger water of the Dami reservoir: the hydro plant and FPV have the
from the World level variations. Physical impacts on the panels are same owner; the site is protected by mountains against
Bank and SERIS higher, especially with higher waves on large reser- powerful winds and typhoons; and, there are low fluctu-
report illustrating voirs. Dam safety standards require assessment, to ations in the lake level (less than 3 m).
the rapid growth check that FPV would not represent any specific new In Europe, EDP has installed a pilot project on the
in FPV. threat. Alto Rabagão reservoir, and is installing a 4.5 MWp
FPV powerplant on the Alqueva reservoir. The Alto
Rabagão project provides interesting feedbacks
regarding mooring on a water body with a 60 m depth
and 30 m of water variation, as well as waves up to 1
m high. Another pilot project was installed at Les
Toules dam in Switzerland, by Romande Energie. The
conditions, at an elevation of 1810 m are demanding:
the water level variation is 50 m, there is snow cover
of up to 50 cm, and ice cover of up to 60 cm on the

Fig. 2. A (very small) FPV island, with the FPV panels,

showing the FPV structures, the mooring lines and anchors.

62 Hydropower & Dams Issue Three, 2021

panels, temperatures vary from -25°C to 30°C, and
there can be wind of up to 120 km/h. But the electric-
ity yield at this high-altitude location is 20 per cent
higher than in the valleys.

2.2 Technical challenges of FPV on dam reservoirs

Installing FPV on large dam reservoirs implies specif-
ic technical challenges, associated with: the specific
loads that occur on such reservoirs; and, specific
requirements of dam safety.
Waves induced by wind on large reservoirs are high-
er than on smaller water bodies. Wave loads require a
careful assessment and FPV islands must be designed
accordingly, which is not so easy.
• Conventional wave height computation (based on
wind velocities at nearby weather stations, fetch
assessment and empirical formulae from technical lit-
erature) is not precise enough. They do not take into The Yamakura FPV occupied 30 per cent of the 2.3 Fig. 3. Rough
account, for instance, the possible wave swell caused km² surface of the reservoir. The installed capacity was sketch to show the
13.7 MW. It was designed for a wind velocity of 41.5 configuration of the
by reflection on reservoir rims. Yamakura FPV
• There are coupling effects: wind and wave loads not m/s. The typhoon provoked a failure of the mooring
lines of the FPV island, which then drifted, broke apart island.
only depend on natural conditions at the dam site, but
also on the geometry and features of the FPV-DR pro- and caught fire.
ject. Wave loads are dampened along a series of PV A thorough review of the causes of failure was car-
floating modules, or by water barriers. ried out by a Technical Working Group initiated by
• The FPV support structure is subject to fatigue Ministry of Economy, Trade and Industry (METI) of
because of the waves. As shown at the Alto Rabagão Japan. Various possible causes for the failure of the
pilot plant, the specific dam reservoir conditions (high- mooring lines were considered: wind velocity beyond
er and more frequent waves) could lead to early ageing. the design value; an actual wind force larger than had
been estimated in the design; unexpected additional
Anchoring is challenging in dam reservoirs, because force by wind waves; deterioration of anchors and
of the water depth. This usually prevents a complete connection pins from repeated loading; insufficient
and precise appraisal of the land surface topography foundation strength of the anchors; and, unexpected
and geotechnical conditions at the bottom of the reser- ununiform force balance acting on the anchors and
voir. Mooring lines need to be adapted to local condi- connection pins because of the specific island shape.
tions, and especially the often large water level varia- It was shown that the five first hypotheses could be
tions of dam reservoirs. Specific designs might be disregarded. The failure was initiated by an unantici-
required to keep tension in the cables, to prevent the pated repartition of mooring forces, caused by the
PV island from drifting. In some cases, the FPV design island shape and the orientation of the various mooring
also needs to consider a possible grounding at the bot- lines. Fig. 3 shows the FPV island, its 420 anchors and
tom of the reservoir; rugged topography, remaining 828 mooring lines, and the direction of the wind dur-
trees or immersed objects could harm the FPV mod- ing the typhoon event.
ules in case of grounding. The potential presence of According to the METI working group, the failure
unexploded ordnance (UXO) can be an additional began with several anchors being pulled out (red rect-
issue in some reservoirs. angle). This caused force concentration on the adjacent
Large FPV islands are designed under the various mooring lines and connection pins between panels,
loading cases that they will have to endure (as regards resulting in the propagation of the failure.
intensity and direction), accounting for the uncertain- It is worth mentioning that, even if the wind design
ties associated with the geometry of the mooring lines speed and the design wind force were not exceeded,
(variability caused by water level variations and inac- the METI group findings show that they were actually
curacy of the anchors locations), integrating the vari- reached.
ous scales?? that interact (island-module-connectors)
and considering fatigue and wear is challenging.
This has an impact on dam safety: conventional FPV 3.Dam safety issues
on water ponds are typically designed for 50- or 100- 3.1 Dam-safety related issues
year wind loads and for a service life of 25 years. A possible blockage of a spillway induced by a failed
Considering these figures and the fact that, as men- FPV island, that could drift towards the spillway and
tioned above, the design procedures struggle to account impair its discharge capacity, is the main threat.
for all the loading cases, the probability of failure of an Two other consequences of the failure of FPV
FPV-DR island, designed as a classical FPV island, anchors, and consequent drift of the loose FPV islands,
could easily exceed 10-2 per year. And, in certain cases, are: additional loads on structures (on intake
this may have an impact on the dam safety. towers/gates/dams) mainly induced by wind loads
(drag) on drifting islands that would have docked
2.3 The Yamakura accident again these structures; and, fire near dam safety struc-
The Yamakura accident was the failure of an FPV tures in the event of a failed FPV island catching fire.
powerplant that was installed on the Yamakura dam Additional safety issues are raised for dams with
reservoir in Japan. It was caused by typhoon No.15, upstream impermeable elements, when this watertight-
which occurred in September 2019. ness could be damaged by a loose FPV drifting island.

Hydropower & Dams Issue Three, 2021 63

 This is the case, for instance, for dams with exposed failure probabilities might be considered. On the other
geomembranes. hand, assessing the structural safety of FPV-DR
There is already experience of dam reservoirs for islands under extreme wind conditions is complex, and
very similar issues. In many reservoirs, log booms are ensuring a failure probability as low as 10-4 per year
installed to collect floating debris. The accumulation during the whole service-life seems out of reach at the
of this debris can sometime result in large floating moment. Therefore, it might be more appropriate to
‘islands’, very similar to the FPV islands. In cold cli- design for the conventional 1:100 probability that is
mates, reservoir surfaces (and sometimes also the river usually considered in FPV design, and to provide addi-
surfaces) freeze during winter, which also results in tional safety barriers between FPV-DR failure and dam
large floating islands. Both these phenomena can failure, so as to disconnect FPV safety requirements
cause similar threats. Safety barriers might be devel- and dam safety requirements.
oped based on experience with ice and debris han- For a given probability, there is a relationship
dling. between the wind speed and the duration, during
A key issue is that safety must be guaranteed which this speed is exceeded; this is a classical inten-
throughout the dam’s service life, and not only the sity-duration-frequency relationship. The selection of
FPV service life. There are specific risks during FPV the wind intensity duration to be considered will
installation before the anchors, mooring lines and con- depend on the failure mechanism that is to be consid-
nection have been completed. There are also specific ered. When computing the wind forces on the FPV
risks during FPV operation, considering that the con- modules, one should consider gust speed, that means,
ventional FPV design does not cover the resistance to peak velocities. For computing the wind forces on the
loads (winds, floods) with as low a probability as dam mooring lines, a very short duration gust speed might
safety may require. Furthermore, there are risks after not be relevant, and a 10 min average speed could be
the end of the typical FPV 25-year service life, when accounted for. When computing wave heights, it is
the design strength mooring lines and connections has necessary to consider an established wind during a
been reduced by corrosion or by fatigue. time frame long enough to generate substantial waves:
the longer the fetch, the longer the duration necessary
3.2 Design loads to reach maximum wave heights.
It should be noted that DNVGL’s ‘Recommended Once wind intensity-duration-frequency rose dia-
Practice for Design, Development and Operation of grams have been established, free-field (without FPV
Foating Solar Photovoltaic Systems’, issued in March island) wave parameters such as height and wave-
2021 [DNV, 20214] provides an interesting set of length can be computed. For large and medium-sized
design criteria for FPV, even though the document projects, it is recommended to carry out a numerical
does not address the specificities of FPV on dam reser- simulation, accounting for the reservoir shape and
voirs. bathymetry. Indeed, wave height distribution is far
Though water flow loads during floods need to be from uniform in a reservoir. For large projects, it is
assessed, especially if the FPV is located near a spill- also recommended to take local measurement at the
way, the main load factor is usually wind, with two proposed project location, to calibrate the simulation
main actions: direct wind loads (drag and uplift) on the models.
FPV islands, and wave formation.
Because of the unique location of every reservoir, 3.3 Design for wind and wave actions
wind velocities on the reservoir might substantially Computing wind and wave loads on structures and
differ (in amplitude and direction) from velocities computing stresses induced by these loads is a com-
measured at nearby weather stations. For large pro- plex task. A step-by-step design procedure is recom-
jects, wind simulations taking into account the topog- mended.
raphy around reservoirs and/or wind record with one
or several anemometers are necessary to derive robust • A global (free-field) wind model would establish a
local wind rose diagrams. For smaller projects, wind frequency-intensity-duration rose diagram.
speed velocities derived from codes that include cor- • A global (free-field) wave model would determine
rections for direction, altitude, exposition, and so on, wave parameters for various wind conditions of a
might be used. given frequency.
There is no consensus as to: • An island-scale FPV model is used to compute the
overall wind forces; this will be panel dependent (drag
• which wind speed should be considered for the
coefficient). Then anchors and mooring lines will be
design;
calculated. The island should be stable under estab-
• which frequency (1:100-year, 1:1000-year or even
lished wind speeds (typically 10 min wind speed). The
1:10,000-year); or,
computations must account for the uncertainties in the
• which wind duration (instantaneous gusts, 10 minute
geometry of the mooring lines, and so the load distri-
average wind, 1 hour average wind).
bution among the various lines will differ from the the-
Preliminary answers to these questions are as fol- oretical model.
lows. • An island- scale FPV model can be used to compute
In the case of a large dam, where the failure of an wave attenuation by the islands; this model is also
FPV island would represent a direct threat to the dam’s panel dependent.
safety, then the FPV failure probability needs to be • The structural design of the various panel connectors
very low, for example, 1:10 000 per year or even less. can eventually be carried out according to the wind
But if there are additional safety barriers, or if the fail- forces and wave parameters. The structure must resist
ure of the FPV island is not a direct threat, that is, if a gust wind speed and waves in various directions. This
risk assessment shows that the FPV island failure includes fatigue verification and provision for wear
would not directly result in dam failure, then higher and corrosion.

64 Hydropower & Dams Issue Three, 2021

 This step-by-step procedure does not fully account
for the coupling between the FPV islands, but captures
the main aspects.
3.4 Additional comments
Floating debris might create additional loads, if it
accumulates along the FPV islands.
Seismic loads might sometimes also have to be con-
sidered. So far, three consequences of earthquake
actions have been identified as potential threats to recommended practices [DNV, 20214] considering Fig. 4. Design
FPV: induced landslides on reservoir rims that might Consequence Category (2), including redundancy of options as safety
the mooring lines. barriers to various
impact FPV anchors; rocks falling from nearby slopes undesirable
and potentially impacting FPV panels; and, earth- An anti-drifting device (barrier 1) could be a log
boom or any such structures that would prevent a com- situations.
quake-induced waves in the reservoir.
Sediment accumulation in a reservoir will sometimes plete free drift.
have to be accounted for. This might change the local The location of the FPV islands (barrier 2) considers
bathymetry at the location where FPV panels have the fact that it is sometimes possible to locate the FPV
been installed, with potential impact on the FPV and island in a place where a failure would not have it
mooring lines at low reservoir elevation. drifting towards the spillway (for example because the
wind direction will push the FPV panels in another
direction).
4. FPV-DR technical solutions In some cases, the dam spillway (barrier 3) could be
FPV for large dam reservoirs requires a specific appro- large enough to let the FPV modules pass, or be large
priate design. The FPV modules could be modified to enough to tolerate a partial blockage: drifting modules
mitigate the loads or to withstand higher loads. The would still float and let the discharge flow underneath.
reliability of the mooring and anchor lines needs to be This is the classical approach for dams in cold climates
improved, and the impact on dam safety must be that have to handle ice breakup events.
included in the design procedure. Barrier 4 relates to the dam sensitivity: some dam
Several technical solutions for FPV structures on types (for example, most of arch dams) can withstand
reservoirs are developing, for example: some increase in water level above the MWL without
• module geometry mitigating wind loads (lower endangering the dam.
uplift and drag coefficients); Two important topics help to control the failure prob-
• wave attenuators; abilities:
FPV modules design accommodating higher waves; • Monitoring of the FPV structures is essential, to
and, detect abnormal behaviour, such as mooring lines fail-
• monitoring systems aiming at detecting structural ures or overall ageing, to keep the FPV failure proba-
ageing of the structure or unexpected structural condi- bility as low as projected; and,
tions. • FPV projects comprising several small islands of
The safety of mooring and anchor lines can be simple and regular geometry, rather than one single
enhanced by: large island of complex geometry, are easier to design
(with positive impacts on FPV failure probability) and
• providing redundancy of the mooring lines, as the one FPV island failure is easier to handle.
failure of one or more of the mooring lines must not
lead to the failure of the complete island; 5. Opportunities for hybridization
• designing several independent FPV islands with a Hybridization means combining hydro and solar pro-
simple geometry, rather than a single complex island duction at the same grid connection point, whereas vir-
where the load distribution on the anchors would be tual powerplants refer to a combination of several gen-
more difficult to assess; eration assets delivering energy at different grid con-
• devising a monitoring system that can detect the nection points.
effects of structural ageing (fatigue, corrosion) or of
Hybridization provides value at three time scales. At
unusual structural conditions, and help programming
a minute-scale: hydro could smooth cloud effects on
predictive maintenance before an accident occurs; and,
solar power (providing a ‘smoothing service’). At the
• increasing the safety factors used for the design of
daily scale: hydro could store a share of solar daily
anchoring system to account for fatigue and unexpect-
production (‘shifting service’). At a season-scale: solar
ed load distributions (although devising an appropriate
could save water during the dry season (‘water saving
safety factor is not straightforward).
service’).
The structural safety of an FPV island will usually To provide such services, hybridization requires: a
not match the safety standards for dams. In many hydro powerplant and a reservoir, with some opera-
cases, a possibility could be to tolerate an FPV failure tional flexibility; a solar powerplant, with a peaking
(at a probability of around 1:100 per year or less), and capacity compatible with the hydro plant; and, an ener-
make sure it does not result in a dam failure (so have a gy management system to achieve the hybridization.
dam failure probability far less than the FPV failure Hybridization could eventually produce: 100 per cent
probability). A series of situations or design options renewable electricity, which is fully dispatchable; and,
might provide the desired ‘safety barriers’. an increased availability of water for dry seasons. On
The failure probability of an FPV island will depend the other hand, hybridization is likely to require more
a lot on the quality of its design and construction. It frequent adjustments of hydro generation, with
will probably be in the range of 1:10 to less than potential increases in wear and fatigue of the hydro
1:1000 per year, the latter relating to a design based on equipment.

Hydropower & Dams Issue Three, 2021 65

 Eventually, hybridization will probably require mod-
ification of the hydro reservoir’s operation. In many
cases, water management can be improved thanks to
hybridization (saving water in the dry season), but in
some cases, it could be a serious constraint. ◊

Acknowledgements
This article was prepared making full use of the papers
presented at the meeting of ICOLD Technical Committee T, in
December 2020. Reports on regional contexts were shared by
Musa Mukulu (UEGCL, Uganda), Harry Blohm (USA), Craig
Scott (Stantec, UK), Adama Nombre (IFEC, Burkina Faso).
Lessons learnt from completed projects and projects under
development were presented by Felipe Guerra (EDP, Portugal),
Fig. 5. The three There are several possible combinations of solar and Guillaume Fuchs (Romande Energie, Switzerland), Arnaud
time scales for hydro, with both positive and negative characteristics: Rousselin (EDF-CIH, France), Ofer Ilian (Carpi, Israel), Aries
which hybridization Firman (ICOLD, Indonesia), Michel Ho Ta Khanh (ICOLD,
can be valuable. 1. FPV-DR: FPV on a dam reservoir Vietnam), Junichi Fukuwatari (ICOLD, Japan), and Feng
No hybridization of electricity production. Weijiang (PowerChina, China). Contributions relating to the
floating panel technique were made by Adrian Constantinescu
2. ShSH : Slightly hybridized solar hydro (DN-T, Belgium), Felix Gorintin (Innosea, France), Olivier
Philippart (Ciel-et-Terre, France). Challenges and opportunities
Solar peak power up to 15/50 per cent of hydropower. associated with FPV on DR were discussed by Marine Bernicot
+: Low impact on hydro => simple operation. (ISL, France), and Zuzana Dobrotkova and Bente Brunes
(World Bank, USA). All of these people contributed greatly to
-: Only a fraction of the potential is developed. the richness of the exchanges. The Committee Chairman is
therefore very much indebted to them.
3. HhSH: Highly hybridized solar hydro
Solar peak power up to 30/100 per cent of hydropower.
References
+: High solar penetration 1. World Bank Group, “Where Sun Meets Water: Floating
-: more complex operation (including weather fore- Solar Handbook for Practitioners”, World Bank Group,
casting and adaptation) ; downstream impact on the ESMAP and SERIS, Washington, DC, USA; 2019.
river might require mitigation measures 2. World Bank Group; “Where Sun Meets Water (Vol. 2):
Floating Solar Market Report” World Bank Group, ESMAP
4. FSH : Full solar hydro and SERIS, Washington, DC, USA; 2019.
3. Lee, N. et al., “Hybrid floating solar photovoltaics-hydro-
Including a dedicated pumped-storage plant. power systems: Benefits and global assessment of technical
+: Very high solar penetration. potential”, Renewable Energy, Vol. 162, 2020.
-: High CAPEX and higher cost per kWh. 4. DNV, “Design, development and operation of floating solar
photovoltaic systems, Recommended practice”, Report
Several hybridization projects have been proposed, DNVGL-RP-0584; 2021.
but most of them still await implementation.
The main obstacle to implementation is the fact that
the LCOE is slightly higher than for pure PV projects,
and that the additional services provided by
hybridization (dispatchability, grid services, and so
on) are usually not given a value in power purchase
agreements (PPAs). Another obstacle is governance:
Fig. 6. The key two operators for a ‘single’ solar-hydro powerplant is
components for not easy to implement. There will be a contractual
hybridization. overlap between the two PPAs.
L. Deroo

L. Deroo is the Chief Executive Officer of ISL Ingénierie.

His expertise covers dams and hydropower engineering, with
experience mainly in Europe and Africa, for greenfield
projects, safety assessments and the upgrading of structures.
He is an active member of the French Committee on Dams
and Reservoirs, with which he has contributed to the drafting
of several professional recommendations regarding dam
safety, and he has served on scientific committees for
various conferences. He is Chairman of the ICOLD
Committee on Prospective and New Challenges for Dams
and Reservoirs in the 21st Century..

ISL Ingénierie, 75 Boulevard Macdonald, 75019 Paris,

France.

66 Hydropower & Dams Issue Three, 2021