See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/316820035

Power output fluctuations in large PV plants

Article · April 2012

DOI: 10.24084/repqj10.676

CITATIONS READS

6 215

4 authors, including:

Javier Marcos Eduardo Lorenzo Pigueiras

Universidad Pública de Navarra Universidad Politécnica de Madrid
39 PUBLICATIONS   922 CITATIONS    149 PUBLICATIONS   3,709 CITATIONS   

SEE PROFILE SEE PROFILE

Miguel García
Universidad Pública de Navarra
37 PUBLICATIONS   942 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Solar radiation and PV applications View project

PVCROPS View project

All content following this page was uploaded by Eduardo Lorenzo Pigueiras on 06 October 2017.

The user has requested enhancement of the downloaded file.

 European Association for the International Conference on Renewable Energies and Power Quality
Development of Renewable Energies, Environment (ICREPQ’12)
and Power Quality (EA4EPQ) Santiago de Compostela (Spain), 28th to 30th March, 2012

Power output fluctuations in large PV plants

J. Marcos1, L. Marroyo1, E. Lorenzo2, M. García1

1
Departamento de Ingeniería Eléctrica y Electrónica,
Edifico Los Pinos, Universidad Pública de Navarra,
Campus Arrosadía, 31006, Pamplona, Spain. [email protected]
2
Instituto de Energía Solar,
Grupo de Sistemas, ETSI Telecomunicación,
Ciudad Universitaria, s/n, 28040, Madrid, Spain

Abstract. The power generated by large plants located in Spain and with output powers
photovoltaic (PV) plants is of an intermittent nature, ranging from 1 MWp to 9.5 MWp (a total of 20
due to cloud passage, which can create considerable MWp). The research focussed on studying the
fluctuations in output. Therefore, if the penetration of irradiance and power fluctuations for time intervals of
PV power in the overall electricity system is high less than 10 minutes, paying particular attention to the
enough, these fluctuations could have a serious effect characterisation of the magnitude and frequency of
on the quality and safety of the power supply. This occurrence of these fluctuations. The study also
article aims to characterise these fluctuations and to analysed the influence of the PV plant size and the
look at the smoothing effect of the PV plant size and number of PV plants grouped in a multi-site system,
that of the geographical dispersion of the PV plants. on the magnitude of the fluctuations. Finally, a PV
plant was modelled, making it possible to simulate the
Key words: dynamics of the power generated by a PV plant based
on irradiance data.
large PV plants; grid-connected; power fluctuations;
PV plant model; geographical dispersion. 2. Experimental set-up

1. Introduction A review of the scientific literature prior to this study

reveals a significant lack of experimental data on the
The variations in irradiance produced by changes in magnitude of fluctuations. The number of studies on
cloud cover can cause rapid fluctuations in the power the subject is significantly small and it has only been
generated by large PV plants. For intervals of less over the last few years when the study of the
than 10 minutes, these fluctuations are directly fluctuations has gradually taken on greater
absorbed by the electricity system in the form of importance. In this way, irradiance fluctuations have
frequency variations, yet without the utility operator been monitored and analysed in countries such as
having the response capacity to correct the Germany [1], Japan [2], Belgium [3] and Australia
imbalances. If these frequency deviations exceed the [4]. On the other hand, as far as the study of power
permitted limits, then there is a risk of a power system fluctuations is concerned, the number of studies made
failure. This fact, together with the high levels of is even more limited. The literature available only
penetration achieved by PV power generation sector includes measurements with a 5 minute resolution on
over the last few years, alerted the grid operator in 100 PV systems (243 kWp in total) in Germany [5];
Spain, Red Eléctrica de España (REE), who was led data taken at 10 second and 1 minute intervals at a
to promote a research initiative to study these fixed 4.6 MWp system and measurements taken every
fluctuations. The Instituto de Energía Solar (IES), the 10 minutes at three 100 kWp plants in Arizona (USA)
INGEPER Research Group of the Universidad [6] and, finally, data recorded with a one minute
Pública de Navarra, and Acciona Solar also took part resolution over a 3 month period on 52 PV systems in
in the project. Japan, with an approximate mean power per system of
3.2 kWp [7].
This paper presents the main conclusions reached in
the project. The results are shown for the In an attempt to cover this lack of prior empirical
synchronised continuous monitoring, with a one research, within the framework of this project,
second resolution, from April 2008 onwards, of 7 PV monitoring equipment has been designed,

https://doi.org/10.24084/repqj10.676 1276 RE&PQJ, Vol.1, No.10, April 2012

programmed and installed in a number of PV plants 4. Fluctuation at a PV plant
owned by Acciona Solar. This equipment, which has
been operating since April 2008, makes a Fig. 2 shows the irradiance fluctuation distributions
synchronised recording every second of the irradiance observed at the Cintruénigo site over the course of a
and the power generated by 7 PV plants located in year (from May 2008 to April 2009) for ∆t equal to
Spain. The plants have powers ranging from 1 MWp 10, 20, 60 and 600 s. All the distributions are
to 9.5 MWp (a total of 20 MWp). Likewise, data are normalized, so that the total area below the curve is
available from two sections of the same plant (48 equal to one. Visual appearance gives a different
kWp and 143 kWp), also recorded with a one second impression because the frequency axis has a
resolution. Six of the plants are located in the South of logarithmic scale. A positive correlation between the
Navarre, separated by distances ranging from 6 km to fluctuation magnitude and the sampling time is
60 km. The seventh plant is situated in the locality of observed again. Table 2 shows the values of some
Socuéllamos (Ciudad Real) at a distance of 345 kms fluctuation intervals. Note that a significant
from the other plants in Navarre (Fig 1). The most fluctuation of, let us say, more than 3% in 1s, is
representative details of these plants are shown in the relatively rare. In fact, this is why frequency is
following table: presented in logarithmic scale in Fig. 6. However, the
relative frequency is close to 41% at 600s.
Peak Power Transformer Power Area
PV plants
(kWp) (kW), P* (Ha), S 10
2

1s 10
2

Arguedas 958 775 4,1 10

1 20s
60s
600s
Sesma 990 800 4,2 10
0

Relative frequency (%)

0
10
-5 -1 +1 +5
-1
Cintruénigo 1438 1155 6,4 10

-2
Rada 1780 1400 8,7 10

-3
10
Castejón 2640 2.000 11,8
-4
10
Milagro 9500 7243 52
-5
10
Socuéllamos 2600 1975 18
-6
10
-100 -80 -60 -40 -20 -1 +1 +20 +40 +60 +80 +100
Total 19906 15348 - Irradiance Fluctuations (%)

Fig. 2: Distributions of the irradiance fluctuations, ∆G∆t(t),

Table 1: Characteristics of the photovoltaic (PV) plants. recorded at the Cintruéingo site in the course of a year, for
∆t = 10, 20, 60 and 600s.

It is noteworthy that all the irradiance fluctuation

features calculated at the rest of places (and not
presented here), essentially coincide with the features
at Cintruénigo site, in accordance with the fact that all
sites belong to the same climatic region.
Consequently, the power fluctuations will not be
affected by the effects of geographic location.
∆t (s)
∆G (%)
1s 20 s 60 s 600 s
Fig. 1: Location of the 7 photovoltaic (PV) plants studied.
0%≤ ∆G ≤ 3 % 98.94 % 92.51 % 87.50 % 59.37 %
3. Fluctuation definition 3%≤ ∆G ≤ 10% 0.92 % 4.47 % 7.61 % 28.24 %
10%≤ ∆G≤ 50% 0.16 % 2.71 % 4.66 % 11.25 %
The magnitude of a power fluctuation ∆P∆t (t) at -4
50%≤ ∆G≤ 100% 2.35·10 % 0.32 % 0.70 % 1.35 %
an instant t for a given sampling period, ∆t, is
calculated as the difference between the two power Table 2: Influence of ∆t on the frequency distributions of
outputs, normalized to the transformer power P* of the fluctuations.
the plant under consideration, Eq. (1). That is: On the other hand, Fig. 3 shows the maximum

∆P∆t (t ) =
[P (t + ∆t ) − P (t ) ] (1)
irradiance fluctuations recorded at the various sites for
an entire year in relation to the different sampling
P* times ∆t considered. The value of the maximum
fluctuation is extremely useful for the power utility
This definition can also be applied to irradiance operator, making it possible to check the robustness of
fluctuations, ∆G∆t, normalizing by G* = 1000 W/m2. a grid. The values observed for ∆t greater than 2
seconds are over 80%. Therefore, although the

https://doi.org/10.24084/repqj10.676 1277 RE&PQJ, Vol.1, No.10, April 2012

irradiance fluctuations can reach very high values, this
140
is a relatively rare phenomenon in annual terms.
140 120

Power fluctuation (%)

120 100
Irradiance Fluctuations (%)

80
100
Max-48 kWp
60 Max-143 kWp
80 Max-958 kWp
40 Max-990 kWp
Max-Arguedas Max-1438 kWp
60
Max-1780 kWp
Max-Castejon 20 Max-2640 kWp
Max-Cintruenigo Max-9500 kWp
40
Max-Milagro 0
Max-Rada 1 2 5 10 20 40 60 120 300 600
20 Max-Sesma Sample period (s)
Max-Socuellamos Fig. 5: Maximum power fluctuations recorded at each PV
0
1 2 5 10 20 40 60 120 300 600 plant vs. ∆t.
Sample Time (s)
Fig. 3: Maximum irradiance fluctuation recorded at each PV The relationship between thee smoothing effect
plant site vs. ∆t.
and PV plant size is embodied in the following
With regard to the power fluctuations produced by proposed empirical Eq. (2), which determines the
a PV plant, the magnitude of these fluctuations is peak power fluctuation value of a PV plant of any size
influenced not only by ∆t, but also by the PV plant S [Ha] and for any sampling time ∆t:
size. It is therefore to be expected that, the larger the
Max(∆P∆t) =90%·(1-e-0.24·∆t)·S –c (2)
plant size, the lower the power fluctuations in relation
to the irradiance fluctuations. This reasoning is based Smoothing coefficient c for ∆t= 1s is 0.5, thus the
on the fact that a cloud will always take longer to smoothing effect on the fluctuations can be described
cover an entire PV plant than an irradiance quite accurately by a 1/√S law. However, for ∆t=600s,
measurement module. Thus, the shorter the sampling coefficient c is equal to 0.02 which supports the idea
time, the greater the smoothing effect. This fact can that, for these sampling times, the plant size is not a
clearly be observed in Fig. 4, showing the annual smoothing factor for the fluctuations. A more detailed
power fluctuation distributions for the powers and supplementary account of the results shown in
recorded at each PV plant for ∆t = 20s. As was to be this and the previous section can be found in [8].
expected, the filtering effect on the fluctuations
increases in line with the increased plant size whilst, 5. Fluctuations in a multi-site PV plant
on the other hand, it decreases as ∆t increases.
system
Likewise, this smoothing effect can be clearly seen in
Fig. 5, showing the value of the maximum power
We shall now go on to quantify the smoothing of PV
fluctuation recorded over a year at each plant in
power fluctuations by geographical dispersion.
relation to ∆t. However, for ∆t= 600s, the smoothing
Thanks to the synchronised monitoring made, it is
effect is non-existent, and fluctuations of 90% can be
possible to determine, second by second, the
reached, regardless of the plant size.
combined power generated by a group of PV plants,
10
2 by monitoring plant behaviour, in the same way as the
48 kWp 10
2
system operator.
1 143 kWp
10 958 kWp Fig. 6 shows the normalised distributions for the
0
990 kWp
1438 kWp
fluctuations recorded in 2009 for ∆t = 60s
Relative frequency (%)

10 0

1780 kWp 10
-10 -1 +1 +10 corresponding to just one plant (Arguedas,
2640 kWp
10
-1
9500 kWp P1* = 0.775 MW), to a group of 3 plants (Castejón,
48 kWp
Milagro and Sesma, P3* = 10.043 MW, COMB3) and
-2
10 a further group of 6 plants (all located in southern
-3
Navarre, P6* = 13,373 MW, COMB6). It can clearly
10
9500 kWp be observed how the magnitude and frequency of the
10
-4 fluctuations decreases in line with an increase in the
number of plants grouped together, N. It is important
-100 -80 -60 -40 -20 -1 +1 +20 +40 +60 +80 +100
to emphasise the fact that the smoothing effect
Power Fluctuations (%) extends as far as the long sampling times (∆t = 600 s)
in contrast to the effect observed for plant size.
Fig. 4: Distributions of the power fluctuations, ∆P∆t(t),
recorded at each PV plant over an entire year, for ∆t = 20 s.

https://doi.org/10.24084/repqj10.676 1278 RE&PQJ, Vol.1, No.10, April 2012

 2
10 2
10
2 10
Arguedas-Castejon
Arguedas-Socuéllamos
1
10

0 0
Relative frequency (%) 10 10

Relative frequency (%)

0
10
-10 -1 +1 +10

-2
-2 Arguedas 10
10

COMB3 COMB6
-4
10
-4
10 1s
60s 20s
600s
-6
-100 -80 -60 -40 -20 -1 +1 +20 +40 +60 +80 +100 10
-100 -80 -60 -40 -20 -1 +1 +20 +40 +60 +80 +100
Power Fluctuations (%) Power Fluctuations (%)

Fig. 6: Distributions of the power fluctuations ∆P∆t,N, Fig. 8: Distribution of the power fluctuations ∆P∆t,2,
recorded in the course of an entire year (2009) at the recorded over an entire year (2009) at the combination of
Arguedas PV plant and the combination of N = 3 (COMB3) Arguedas–Castejón PV plants (distance = 6 km) and
and N = 6 (COMB6) for multi-site PV plants, all in Navarre, Arguedas–Socuéllamos (distance = 345 km) for ∆t = 1, 20,
for ∆t = 60s. 60, and 600 s.

Fig. 7 below shows the value of the peak power The data shown in Fig. 7 was used to obtain an
fluctuation recorded versus the sampling time ∆t empirical Eq. (3) which provides the value of the peak
based on the number of plants grouped together, N, power fluctuation for a group of N PV plants of a
for all plants located in Navarre. It is possible to certain size S, for any ∆t of less than 10 minutes.
observe a considerable reduction in the magnitude of
the fluctuations, irrespective of the ∆t. Furthermore, it 99th(∆P∆t,N)= 99th(∆P600,1)·(1-e-0.24·∆t)·S -c·N–a; (3)
was found that for a given N, no plant combination a,c> 0
had a greater smoothing effect than another. In other
words, the location of each PV plant in relation to the This equation describes the smoothing of the
others within the group, had no effect on the fluctuations resulting from the plant size and also the
magnitude of the smoothing effect. Therefore, the geographical dispersion. Whenever coefficients a are
fluctuations of a group of PV plants will be smoothed greater than c, as in Table 4, the smoothing due to
provided that there is sufficient distance from one dispersion is greater than that due to plant size.
plant to another. This leads on to the question of what
is the minimum distance between plants in order to ∆t c a
ensure that the plant fluctuations remain independent 1s 0.49 25.55
of each other and to guarantee smoothing. For this, 5s 0.29 64.09
20s 0.11 88.2
Fig. 8 shows the distribution of fluctuations for the
60s 0.05 91.6
entire year of 2009 for the Arguedas-Castejón group, 600s 0.02 94.07
separated by a distance of 6 kms, and the distribution
for the Arguedas-Socuéllamos pair, 345 kms apart. As Table 4: Estimated parameters a and c for empirical
can be seen, there are hardly any differences between Eq. (3).
the fluctuations of any two-plant combinations.
Therefore, a distance of 6 kms, being the shortest Based on Eq. (3), Fig. 9 shows the fluctuations of
distance between any plant and our observatories a total PV power of 100 MW in relation to the number
(located at Arguedas and Castejón) is sufficient to of plants amongst how this power is distributed and
guarantee the independence of plant power sampling time ∆t. As can be seen, after approximately
fluctuations for time intervals of less than 10 minutes 100 plants (1 MW each) the fluctuations are smoothed
or, to put it another way, to guarantee smoothing by considerably, and it is debatable whether the power
geographical dispersion. needs to be distributed still further, given the fact that
this would have little effect on the smoothing
achieved. All these conclusions, together with some
additional analyses can be found in [9].

Fig. 7: Peak power fluctuations ∆P∆t,N, recorded over an

entire year for all possible combinations for N=1…6 plants
(all in Navarre). Fig. 9: Estimated Max(∆P∆t,N) values for different ∆t and N,
P* PV plants combinations;

https://doi.org/10.24084/repqj10.676 1279 RE&PQJ, Vol.1, No.10, April 2012

6. Fluctuations in the frequency domain empirical Eq. (4) to provide the value of the cut-off
frequency in relation to the area S of a PV plant:
The periodicity character of irradiance and the power
generated by a PV plant due to solar resource, invites fc = 0.021·S -0.5 (4)
an analysis of these two variables in the frequency
As a result, it is now possible to propose a PV
domain. The study of the PV power fluctuations in
plant model relating the irradiance received in the
this domain makes it possible to consider the plants as
tracker plane and the power generated, Fig. 11:
low-pass filters. Indeed, what has up to now been
called smoothing due to plant size, could be
interpreted as a filtering of the signal representing
irradiance. The Discrete Fourier Transform was used
to convert to the frequency domain.

Fig. 10 shows the spectrum for the irradiance

Fig. 11: Proposed transfer function for a PV plant with a
normalised by G*, recorded for the entire year of surface area equal to S.
2009 at the Milagro plant, together with the
normalised power spectra for this same plant where K is the energy gain of the PV plant [m2]. This
(P* = 9.5 MWp) and that for the Sesma plant model has been validated in both the time and
(P* = 0.9 MWp), also recorded over this same period. frequency domains, thereby permitting the simulation
0
10 of fluctuations in any real electricity system, insofar
24 h
12 h
as it makes it possible to convert relatively easy-to-
-1
10 8h generate irradiance time data into power data for any
f
- 0·7 potential PV plant size. By way of example, Fig. 12
-2
10 - 1·7 shows the distributions of the real and simulated
f
Amplitude

-3
maximum daily fluctuations, for ∆t equal to 600s at
10
Milagro PV plant. As can be seen, the frequency
-4 distributions are very similar. A more detailed
10 GN
description of this model can be found in [10].
PN 0.99 MWp
-5
10
PN 9.5 MWp 40
Real
-6
10 -6 -4 -2 35 Simulated
10 10 10
f (Hz)
30
Relative Frequency (%)

Fig. 10: Spectrum of the irradiance GN recorded at Milagro,

output power PN at Sesma (P* = 0.99MW) and Milagro 25
(P* = 9.5MW) over 1 year. The linear region for the higher
20
frequencies of the power spectra can be well fitted by a
function of the form f -1.7. 15

It is possible to observe the corresponding peaks at 10

frequencies of 24, 12h etc, due to the cyclic nature of 5

the solar resource, being independent of the spectrum
0
considered. In the low frequency zone (up to <10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-9090-100
Max power fluctuation, ∆P (%)
approximately 10-3 Hz), all the spectra had a similar 600s max

performance, with a downward trend f -0.7. However, Fig. 12: Histograms of the real maximum daily power
after a certain frequency fc, which differed from one fluctuation observed in the course of a year at Milagro PV
plant to another, a change in trend was observed, plant and the simulated fluctuations for ∆t =600 s.
implying the increased smoothing of the higher
frequencies. As was to be expected, the greatest 7. Conclusions
smoothing was obtained for the plant with the largest
size. The mathematical expression of the new trend is The quality and safety of the electricity supply can be
form f -1.7 and is independent of the PV plant seriously affected by large PV plant power
considered. The cut-off point for functions f -0.7 fluctuations caused by the variability in irradiance due
and f -1.7 can be identified as cut-off frequency fc for to cloud passage. This article has presented the study
the systems in question. Furthermore, the difference of these fluctuations, obtaining the following results:
between the slope for the two functions is the unit,
which means that the corresponding low pass filter is - A unique monitoring system, in place since April
first order, with a pole at cut-off frequency fc. This 2008, and which makes synchronous recordings with
exercise was extended to the other PV plants, a one second resolution, of the power delivered to the
obtaining the corresponding value for fc. These grid by seven PV plants, with powers ranging from 1
experimental values were generalised to obtain MWp to 9.5 MWp and with a total combined power
of almost 20 MWp.

https://doi.org/10.24084/repqj10.676 1280 RE&PQJ, Vol.1, No.10, April 2012

 - The power fluctuations occurring in a PV plant of Education and Science (grant number: DPI2009‐
within time intervals of less than ten minutes, are very 14713‐C03‐01).
high and relatively infrequent, although they are
smoothed in relation to the irradiance variations. The References
smoothing factor is the PV plant size, and is relevant
for times of just a few seconds, disappearing as the [1] Beyer HG, Decker B, Luther J, Steinberg-Willms R.
sampling time increases. Likewise, an empirical ‘‘Spatial and temporal characteristics of short term
fluctuations in solar radiation for PV-plant
equation is proposed, and which provides the value of
applications’’, 10th European Photovoltaic Solar Energy
the peak power fluctuation in relation to the PV plant Conference, 8–12 April, 1991 Lisbon, Portugal, 453–
size. 456.
[2] Otani K, Minowa J, Kurokawa K. Study on a real solar
- The geographical dispersion of the PV plants is a irradiance for analyzing a really-totalized PV systems.
highly effective way of smoothing the power Solar Energy Materials and Solar Cells 1997; 47: 281–
288.
fluctuations, even for ten minute sampling intervals. It
[3] Woyte A, Thong VV, Belmans R, Nijs J. Voltage
is sufficient to locate two PV plants at a distance of 6 fluctuations on distribution level introduced by
kms, one from the other, to ensure that the photovoltaic systems. IEEE Transactions on Energy
fluctuations over 10 minute intervals are independent Conversion 2006; 21: 202–209.
of each other and are smoothed out when combined. [4] Woyte A, Belmans R, Nijs J. Fluctuations in
Furthermore, an empirical equation is proposed, that instantaneous clearness index: analysis and Statistics.
will give the value of the peak power fluctuation, Solar Energy 2007; 81: 195–206.
[5] Wiemken E, Beyer HG, Heydenreich W, Kiefer K.
based on the number of plants grouped together and Power characteristics of PV ensembles: experiences
the PV plant size. from the combined power production of 100 Grid
connected PV systems distributed over the area of
- A spectral analysis of the power and irradiance Germany. Solar Energy 2001; 70(6): 513–518.
data measured over the course of a year has made it [6] Curtright E, Apt J. The character of power output from
possible to characterise a PV plant as a first order utility-scale photovoltaic systems. Progress in
Photovoltaic: Research and Applications 2008; 16: 241–
system and, specifically, as a low pass filter whose
247.
cut-off frequency is directly related to the plant size. [7] Murata A, Yamaguchi H, Otani K. A method of
This model can be a highly useful tool for utility estimating the output fluctuation of many photovoltaic
operators, in the sense that the operator simply needs power generation systems dispersed in a wide area.
to obtain irradiance data and the PV plant size in order ElectricalEngineering in Japan 2009; 166(4): 645–652.
to simulate any photovoltaic scenario. [8] Marcos, J., Marroyo, L., Lorenzo, E., Alvira, D. and
Izco, E. (2011), Power output fluctuations in large scale
PV plants: One year observations with one second
With regard to future lines of investigation, the resolution and a derived analytic model. Progress in
conclusions of this work suggest the need to continue Photovoltaics: Research and Applications, Volume 19,
studying the correlation of fluctuations for distances issue 2, pages 218-227, March 2011. doi:
of less than 6 km. Currently ongoing and still 10.1002/pip.1016.
provisional studies suggest that risky days result from [9] Marcos, J., Marroyo, L., Lorenzo, E., Alvira, D. and
the combination of high wind speed (about 60 km/h, Izco, E. (2011), PV power fluctuations smoothing by
geographical dispersion. Progress in Photovoltaics:
or more) with intermediate daily clearness index Research and Applications, Published online in Wiley
(0.35 < Kt < 0.65). Because wind speed and clearness Online Library (wileyonlinelibrary.com). doi:
index can be predicted from 1 day to another (in fact, 10.1002/pip.1127.
it is already being done in standard weather forecast), [10]Marcos, J., Marroyo, L., Lorenzo, E., Alvira, D. and
this likely opens the door for future prediction of Izco, E. (2011), From irradiance to output power
fluctuations in daily terms. Finally, there is a need to fluctuations: the PV plant as a low pass filter. Progress
in Photovoltaics: Research and Applications, Volume
study the combination of PV generators - storage 19, issue 5, pages 505-510, August 201. doi:
batteries. As well as limiting power fluctuations, this 10.1002/pip.1063.
alternative would allow the PV systems to play an
active part in frequency regulation and would also
make it possible for the plants to increase their daily
production curve, to a certain extent, in an endeavour
to adapt it to consumption.

Acknowledgements

The authors would like to thank Red Eléctrica de

España (REE) for its financial support and Acciona
Solar for authorizing measurements at its PV plants
and also for the helpful collaboration of its staff. This
work has been also supported by the Spanish Ministry

https://doi.org/10.24084/repqj10.676 1281 RE&PQJ, Vol.1, No.10, April 2012

View publication stats