Another Visit to El Hierro

Guest post by Hubert Flocard(1)

The Gorona del Viento plant on the island on El Hierro has been hailed all over the world as an example of how a remote island that has historically been 100% dependent on fossil fuel generation can produce all of its electricity from renewable energy sources. Is this a correct assessment?

Well, it seems that no one other than Energy Matters is making any effort to find out – certainly the plant operators are in no hurry to publish any results. Thus we are doubly indebted to Hubert Flocard for this exhaustively detailed and carefully reasoned analysis, which was completed from scratch in a remarkably short period of time.

Hubert’s conclusions are summarized immediately below so I won’t bother to repeat them here. As Hubert graciously notes in his Acknowledgements section I am not entirely in agreement with them, but rather than make this introduction interminably long I will put my alternative scenario forward in comments.



An analysis of the El Hierro island electric data found on the Red Eléctrica de España site for the period from June 26 to August 31, 2015 shows that the renewable contributions have covered 49.5 % of the electric demand of the island. It also shows that with the present wind plus storage system this renewable fraction can’t exceed 80.1 %(2). Neither the capacity of the smaller reservoir of the pumping system, the power of the pumps, nor their efficiencies appears to be the limiting factor. Increasing the active wind power appears as the most effective option to reach a higher renewable fraction.


In his post “El Hierro revisited” of Sept.2 2015 [RA1](3) R.Andrews presents an analysis of more than one year of the data that the Spanish grid Red Electrica de España (REE) publishes on its site for the El Hierro island electric system(4). The present study uses the same data to extend his analysis.

The figure below extracted from [RA1] shows the Gorona del Viento “wind park + hydraulic storage system” of El Hierro.

Figure 1:  Gorona del Viento plant layout.

The pumping plant has a total power of Pu = 6 MW(5) while the power of the hydraulic turbines is Pd = 11.32 MW(6) . The useful volume capacity of the lower reservoir is 150000 m3(7) which, given the 655 m head corresponds to a theoretical potential mechanical power of R0 = 267 MWh. There is a 40 % loss in the transformation of mechanical energy into electric energy due to the efficiency of the pumping (ru) and of the turbines (rd).

The upper reservoir is larger: 380 000m3(8) . It has other functions than electric production. For instance it provides irrigation services. This explains why the sea can’t be used as lower reservoir.

The sum of the nominal power of the five wind turbines is 11.5 MW(9). Moreover, for its electric energy, the island can rely on the Llanos Blancos diesel plant which consists of seven diesel-fired units for a total of 11.78 MW. This information on the characteristics of the “wind + storage” system was taken from [RA1]. In the same reference one finds the following figure.

Figure 2:  Daily average power Gorona de Viento generation, July 27, 2014 to      August 30, 2015

The upper envelope of the coloured zone (yellow topped by green) gives the electric demand. It oscillates around a 5 MW daily average value. Before June 2015 it is almost entirely covered by the production of the diesel-fired plant(10) . Only towards the end of June 2015 begins one to see a significant renewable energy contribution. Because we are interested in the performances of the El Hierro “wind+storage” system, we thus neglect all data prior to the end of the month of June 2015.

Fig.2 shows also that the electric demand is not modified by the modification of the balance within the production mix. This tells us that the islander’s activities which required the level of the electric demand observed before June 2015, remained unchanged after the transition.

(2) This value is obtained from the knowledge of the wind production over only two months and 5 days. This calculated maximum limit may change depending on the evolution of wind productivity over a full year.
(5) “Installed in a newly-constructed building, it comprises 2×1500 kW pump sets plus 6×500 kW pump sets, with a total power of 6 MW the system is equipped with 1500/500 kW variators”
(6) “4 Pelton groups of 2,830 kW of power each, for a total power of 11.32 MW. The maximum flow during generation is 2.0 m3/s, with a gross head of 655 meters”
(7) ”Located in the vicinity of T.C. “Llanos Blancos”. It has a useful capacity of 150 000 m3. A dam was built to create it. It uses PVC-sheet waterproofing; it is repairable under the water”.
(8)“Located at the “La Caldera” crater, it has a maximum capacity of 380 000 m3 two drain intakes with PVC-sheet waterproofing; it is repairable under the water.”
(9) “5 wind turbines (Enercon E-70) with 2.3 MW of power each, for a total power of 11.5 MW”.
(10) It seems that filling the lower reservoir with that part of the production capacity of the desalination plant which could be diverted from the tasks this plant was initially built for (human usage), took a very long time. This may well explain why nothing much happened over more than a year from June 2014 to June 2015.



On its site the Spanish grid operator REE provides statistics on the electric demand and productions of El Hierro island. This work uses these data from the months of June, July and August 2015. The data consist of the electric demand and the three productions (Fuel, Wind, Hydro, in MW) with a 107 time resolution. In fact, only three pieces of data are independent as it can be checked that, at any time the demand is equal to the (algebraic since Hydro is sometimes negative) sum of the three productions.

The REE data tables contain columns for other types of production including “Solar fotovoltaica”. These columns contain only zeros. It thus seems that any solar photovoltaic production existing on the island is either not injected into the grid or not known by REE at the moment if prepares its data file(11).

We make here the assumption that the goal of the Gorona del Viento system is the demonstration of the technical viability of a 100% renewable coverage of the electric demand. This implies that diesel-fired turbines are never called to produce energy for the pumping of water. Only wind produced energy feeds the storage system.

The hydraulic power registered by REE is either positive or negative. We denote H+ the positive contribution which is produced by the Pelton turbines with down-going water and H- the negative contribution which is fed into the pumping part of the storage system with the aim, whenever possible, to raise water from the lower to the upper reservoir(12).

In this work, we will then distinguish different usages for the wind produced energy. Let us denote Wt (t for total) the wind energy produced by the wind park and registered as “Wind” in the REE data table. A fraction of Wt energy can be sent directly to the grid to cover electric demand. We denote it Wg (g for grid). The rest of wind energy which we denote (Wp, p for pumping) is set aside to feed the pumping station. On the other hand, for diverse reasons (lower reservoir empty or Wp power exceeding the power of the pump Pu), sometimes, there may be an excess of wind energy. Thus we will distinguish the wind energy that can be used for storage Wu (u for useful) and the wind energy that is lost Wl (l for lost). Whether we discuss power or energy we have at any time the following formal “equalities”:

Wt = Wg + Wp = Wg + Wu + Wl.

Later, we mention how it is possible to destroy energy within a hydraulic storage system.
Since the production of the diesel-fired plant is not used to raise water in the storage system which only relies on wind production one also has the additional “equalities”:

|H-| = Wt – Wg = Wp = Wu + Wl,

where |H-| is the absolute value of H- ( |H-|= – H-).

The list of assumptions made in this analysis of the RRE data is:

• The limit on the wind energy that can be stored is determined by the capacity of the lower reservoir (the smaller of the two). This reservoir corresponds to a total useful “mechanical” energy R0.

• The upper reservoir which is multipurpose does not introduce any limit on the system.

• The amount of water present in the lower reservoir is only modified by transfers to and from the upper reservoir. We do not take into account phenomena such as evaporation (El Hierro is a sunny and windy island) or rain. We also assume that there is no other man-made injection of water in the lower reservoir as may result for instance from the diversion of spring water or the production from the desalination plant. The relevance of this assumption will be clear from the results of Sect.III.

• The flux of energy going to the upper reservoir and leaving the upper reservoir is limited by the power Pu = 6 MW (up-going) of the pumps and Pd = 11.3 MW (down-going) of the turbines.

• Since the storage system has two independent water pipelines, energy storage of wind energy and production of hydraulic electricity can be simultaneous.

• Diesel-fired turbines are never called to assist in the pumping of water.

• The electric demand column includes both the usual electric losses (keeping the diesel turbine ready to function and any grid losses) AND the wind losses (wind energy that cannot be used because either the lower reservoir is empty or the power would exceed the nominal power of the pump). As we do not use the redundant electric demand data column until Sect.IV, this means that when Hydro is negative (H-), its absolute value corresponds exclusively to the sum of wind power used to pump water (Wu) plus any wind power produced but lost (Wl).

• Since there are two independent water pipelines, it is possible to destroy wind energy (Wl) within the storage system by pumping with one pipeline and releasing water with the other pipeline while the Pelton turbines are disconnected.

• The wind electric energy which must be used to lift a quantity X of mechanical energy from the lower reservoir is X(1+ru) where ru measures the efficiency of the up-going process.

• When a quantity of electric energy Y is produced by the turbines (when in the column Hydro the figure is positive H+) a quantity Y(1+rd) of mechanical energy has left the upper reservoir to refill the lower reservoir.

• REE does not discard some wind production in a hidden way by not accounting for it anywhere in its data sheet. As a result, if the wind power is larger than real demand minus the diesel production, the remaining wind power is integrally transferred to a negative Hydro. Then, it may occur one day that the absolute value of delivered wind power minus wind power directly sent to the grid exceeds the pumping power of the storage system.

(11)In continental France, the information on all electric production directly injected into the low voltage grid (managed by distribution operators), such as solar photovoltaic for which the number of producers already counts in several hundred thousands, is not transferred to the high voltage grid manager RTE (equivalent of REE) until after few days.
(12) Because the pumping phase of hydraulic storage (called H- here) and the hydraulic production of energy (H+) are of a distinct nature – consumption instead of production -, in the data tables provided by the French grid operator RTE on its eCO2mix web site, hydraulic storage (negative power) and hydraulic production (positive power) are given in separate columns. REE chooses to combine them in just one column.



III.a)  REE data

III.a.1) A very special day: August 9 2015

For a discussion of the “two-hour-100% renewable” event which attracted media attention on this specific day we direct the interested reader to [RA1]. Figure 3 presents the evolution of the electric demand and of the three types of production during this day. Figure 4 shows how the demand was covered by a mix of the three productions.

Figure 3:  Electric demand and the three productions on August 9, 2015. The vertical scale is in MW.

Figure 4:  Coverage of the electric demand by the three different productions for the 24 hours of August 9 2015. The vertical scale is in MW.

During a two hour period between 12:30 and 14:30 the electric demand was integrally covered by a mix of wind and hydraulic production. It appears that the diesel-fired plant which provided a baseload production turned off (or was turned off) suddenly at 12:30 and that the engineers only restarted it two hours later after which it returned to the same level of baseload mode. One also sees that around 13:00 the wind production, which exceeded the demand until noon, suddenly collapsed. Balancing the demand was then made possible by a surge of the hydraulic production resulting from a flow of water from the upper to the lower reservoir.(13)

From Figure 3 it can be seen that before noon Hydro was just H-. The electric demand was almost flat and the diesel production was flat. Thus the contribution of the wind to the electric demand (Wg) was also almost flat. On the other hand, during this period, all details of the fluctuations of the wind production curve can be observed in reverse in the H- curve. This is coherent with our assumption: |H-| = Wt – Wg = Wp = Wu +Wl.


(13)This event might be due either to the fortuitous conjunction of unwanted events or to the decision to make a short test designed to attract a much needed media attention after more than one year without any outstanding result to present to the funding agencies. Such a staged test would then involve turning off the diesel turbines and half an hour later disconnecting two wind turbines.


III.a.2)  Analysis of the summer production 26 June to August 31st.

Figure 5 shows the evolution of the electric demand and three productions. Demand oscillates between 7.6 and 3.3 MW. Wind varies between 0 and 7.6 MW. This may indicate that all the installed wind power (11.5 MW) has not yet been used as part of a strategy of careful gradual implementation of the El Hierro system by the Spanish engineers. Using nevertheless 11.5 MW as installed power the wind efficiency is 38.8 % a value intermediate between that of the Belgian and Danish offshore wind parks. El Hierro belongs to the good wind spots.

Figure 5:  Electric demand and the three productions powers for the period June 26-August 31. The vertical scale is in MW.

During these two summer months, 50.5 % of the demand was covered by diesel production; a figure which relativizes the importance of the “2-hour-100% renewables” event. Figure 5 also shows that the diesel power adjusts in steps to compensate periods of low or high wind production. The fluctuating part of the wind production and the oscillations of the demand are retrieved in the H- curve.

According to the assumptions the part of the Hydro curve which is negative (H-) corresponds to wind power set aside for storage (Wp) and not sent directly to the grid. One thus finds that only 54.9 % of the produced wind energy has been sent directly to the grid (Wg/Wt). This leads us to Figure 6 where the 9 August event appears as a vertical thin green line reaching the horizontal axis.

Figure 6:  Coverage of the electric demand by the three different productions from June 26 to August 31. The vertical scale is in MW.

The time-cumulated energy consumption and productions over the considered period are given in Figure 7. The largest fraction of the demand coverage is at any time provided by the fuel-fired plant. The total contribution of the hydraulic storage system is small as could already be guessed from the cumulated area of the blue zones in Figure 6.

Figure 7: Evolution of the cumulated electric demand and contributions of the three productions over the period June 26 to August 31. The vertical scale is in MWh.

We now use the assumptions to calculate how the wind energy set aside has been used or discarded.

III.b) Analysis, management of water and of wind production

III.b.1) parameters and outcomes

Let us first recall the parameters of the storage system used in this calculation. For the pumping power we use Pu = 6 MW and Pd =11,3 MW for the Pelton turbine power. For the efficiency of the pump and turbines we choose ru = rd = 18 % so that (1+ru)(1+rd)-1 ~ 40 %. These efficiencies may well be on the optimistic side. They can be modified. Anyhow, they only slightly affect the results.

For the useful mechanical energy (water) of the lower reservoir we take the same value R0 = 230 MWh as in [RA1]. This may be an overestimate and 210 MWh might be a more realistic value. Anyhow, we will show later that decreasing or increasing this value by 20 % (180 MWh or 280 MWh) does not affect much the results.

Our analysis covers a two-month period. It begins at the end of June when there is obviously a transition in the Gorona del Viento mode of operation (see. Figure 2). As the period under study is short, it is important to minimize the effect of the initial date and time. We have chosen June 26 00:00. The arbitrariness of this choice will decrease as months pass and more data are incorporated into the study. We also assume that the lower reservoir is at its full mechanical capacity R0 at initial time.

We already know the amount of wind power directly sent to the grid (Wg, the green zone in Figure 6 and the orange curve in Figure 7). In addition to any curve or figure that can be produced directly from the REE data (sect III.a.2) the expected outcomes of the calculation are:

• The amount of wind energy that is effectively used by the storage system (Wu) and the amount of wind energy it is able to return to the grid ( Hw =Wu/((1+ru)(1+rd) ). By difference with values calculated directly from REE data (Wt and Wg), we determine the amount of wind energy which had to be destroyed (Wl).

• The absolute wind park efficiency and also the effective wind park efficiency (a smaller value).

• A comparison of the amount of hydraulic energy delivered to the grid H+ to the total amount of energy it would have been able to deliver thanks to a prior storage of wind energy Hw.

• The evolution of the mechanical energy Rd still present in the lower reservoir and available to pumping (at initial time, Rd=R0).

III.b.2) Results

From the curves in Figure 5 it is apparent that the green Hydro curve is negative during a large fraction of the time and is only positive for short and far apart periods. Except for a few occasions (such as August 9) H+ power sent to the grid is small while on the negative side, values of H- power between -4 and -6 MW are frequent. In other words a significant amount of wind energy is made available to the storage system which only returns a small fraction of it to the grid (H+).

One should thus expect that within our algorithm the lower reservoir – supposed to be full at initial time – is rapidly emptied and remains so most of the time thereafter. The blue curve in Figure 8 shows exactly this behaviour.

Figure 8: Time evolution of the energy capacity (in MWh) of the lower (blue curve) and upper reservoir (red curve) over the period June 24 to August 31. By convention the energy content of the upper reservoir is set to 0 at initial time.

Given the amount of wind production not sent directly to the grid, it takes only few days before all the water (potential energy content) of the lower reservoir is transferred to the upper reservoir. With our algorithm which involves pumping water whenever possible, the process is completed by July 4. From then on, the stored energy is only sparingly used. After that date, the largest excursion corresponds to August 9 when an emergency situation (or a staged test) required calling hydraulic production. Expectedly, the deviations of the blue curve from zero correspond to the blue zones in Figure 6.

If a large amount of wind energy (H- = Wp) is set aside for storage continuously over the period of two months while no more energy can be transferred (lower reservoir empty), one expects that most of this set aside wind energy will have to be destroyed (Wl). This is what is shown in Fig.9 which compares the time-cumulated total wind production (Wt), the fraction directly sent to the grid (Wg), the fraction set aside for storage which is lost (Wl), the fraction set aside for storage which is used (Wu) and finally the energy that can be returned to the grid once the efficiency of storage is taken into account (Hw).

Figure 9: Time-cumulated total wind production (purple), fraction of it directly sent to the grid (yellow), fraction used for storage purpose (red) and fraction lost (blue). The black curve gives the time-cumulated energy which the storage system is in position to return to the grid. The vertical scale is in MWh.

As expected starting from July 4, the blue curve (Wl) departs from 0 and keeps on growing over the two months.

From this analysis, we deduce that, after two months, only (Wg+Hw)/Wt = 63 % of the produced wind energy (Wt) has been used either directly (Wg, yellow curve) or transformed by means of the storage system (Hw, black curve). Once transformed in wind efficiencies (assuming that the 11.5 MW of installed wind power are producing) it leads to a decrease from 38.8 % to 24.4%. By comparing the hydraulic production sent to the grid over the two months (H+) to the amount of energy made available by storage (Hw), one also finds that only 66.4 % of the latter energy has been returned to the grid.

Personal comments on this poor usage of the present El Hierro electric system are given in the final section. I now turn to an evaluation of what could be the upper limit of the coverage of electric demand that such an electric wind + hydraulic storage system can achieve.


In this section we work with only two pieces of REE data: electric demand and wind production. The goal now is to make the best possible usage of the wind production to cover electric demand and in this way reduce as much as possible calling the fossil-fired production. We then determine the optimal result that can be reached by the present El Hierro system.

Because we are looking for an upper bound of the renewable coverage of the electric demand, we make another strong assumption: the fuel-fired and hydraulic productions can adjust instantaneously to the requests, whether “start” or “stop”, imposed by the fluctuations of the wind production. This is not fully unrealistic if one recalls that in the present French system, fuel-fired plants which cover only about 1% of the yearly electric energy production provide ~18 % of the balancing needs of the grid. Hydraulic production is also known to be highly reactive.

We proceed in two steps.

IV.a)  Step 1: Maximal use of the wind production while neglecting the hydraulic storage system

Any wind power used by the storage system Wu is returned reduced by the factor 1/((1+ru)*(1+rd)). It is thus preferable to inject first the wind power directly into the grid whenever it is possible.

This is what we do in this Step 1: use wind power as much as possible to cover demand. If there is an excess of wind production, it is simply destroyed (that Wp=Wl and Wu=0). If the electric demand exceeds the wind production Wt, call the fossil-fired plant.
Using the El Hierro electric data of the last two months leads then to the energy mix shown in Figure 10.

Figure 10: “Optimal” coverage of the electric demand by the wind and diesel productions only, from June 26 to August 31. The vertical scale is in MW.

A comparison with Figure 6 shows a significantly reduced area of the yellow zone. Now the renewable (wind only in fact) fraction is Wg/demand = 74.6 %. The fraction of the wind production that could be used directly by the grid is Wg/Wt = 90.8 %. The energy wind efficiency is slightly reduced from 38.8 % to 35.2 %.

On the other hand, even if it is much cheaper to improve renewable coverage by using the wind power directly as much as possible, by throwing away the surplus wind energy and by asking the dispatchable standard system to adjust to the fancies of the god Eole, this step 1 strategy defeats the advertised purpose of the El Hierro demonstration experiment which is to show how a storage system can be used to optimize wind production(14).


(14)I use the adjective “optimal” from a purely technical perspective which is completely distinct from the economic perspective.


IV.b)  Step 1: Maximal use of the wind production first directly then indirectly via the storage system.

We have seen that in Step 1 a fraction Wl/Wt = 9.2 % of the wind production is destroyed. We now use the same algorithm as described in Sect.III to rescue it by means of the storage system. As soon as some energy has been transferred to the upper reservoir, we use it to reduce any present or later call to diesel production. This leads to this modified picture of the energy mix.

Figure 11: “Optimal” coverage of the electric demand by the wind, hydraulic storage and diesel productions, from June 26 to August 31. The vertical scale is in MW.

The comparison with Figure 10 shows that, as expected, the hydraulic production further reduces the calling on the fuel-fired production. Before going into numbers, Figure 12 illustrates the process by which the wind energy Wp is transformed into Hw and how the process affects the water content of the lower reservoir.

Figure 12: Time evolution of the energy capacity (left scale in MWh) of the lower reservoir (red curve) over the period June 26 to August 31. The green curve (right scale in MW) shows the excess of wind production over demand power. The blue curve shows the hydraulic power (right scale in MW) of the storage system within our “optimal” scenario.

The green curve which gives the excess of wind power over demand (Wp) corresponds to periods when storing energy is possible. It corresponds to a down going portion of the red curve which measures the potential energy content of the lower reservoir. The rate of decrease might be reduced if the upper reservoir is at the same time delivering water to avoid calling the diesel production. The blue curve shows when and how much power is sent from the storage system to the grid. The peaks in the blue curve correspond to up-going sections of the red curve. The hydraulic production H+ stops when the water level of the lower reservoir has reached its upper limit.

Since the value of the red curve on the right side of the figure is equal to R0=230 MWh we know that the storage system has transformed all wind energy excess into hydraulic energy and used it later (but certainly before August 31) as H+ to feed the grid. One notes that the potential mechanical energy content (water mass in fact) of the lower reservoir never goes below 100 MWh. This indicates that the size of the lower reservoir is presently sufficient at least if one assumes that the pumping system remains steadily operational at its nominal power (6 MW). In particular it shows that the saturation situation described in Sect.III.b.2 Figure 8 results from a deliberate choice by the people in charge of Gorona del Viento.
A comparison of Figure 11 with Figure 10 shows a further reduction of the yellow area by the blue zones. Now the renewable (Wind + Hydro) fraction is 80.1 %. As said above, the fraction of the wind production that could be used directly by the grid is (Wg+Wu)/Wt =100 %. On the other hand the efficiency of the storage system makes that only 97.4 % of the initial wind energy is returned as electric energy to the grid. The reduction of the energy wind efficiency is smaller than in step 1: from 38.8 % to 37.9 %.

IV.c)  Sensitivity of the “optimal” performances to the “Wind + storage” system characteristics

Improving the efficiency (i.e. decreasing (1+ru)(1+rd) ) will obviously lead to better results. It is more interesting to investigate the influence of two parameters in our calculation: first the capacity of the lower reservoir, second the wind production.
If one redoes the calculation of Sect. III.b with R0 = 180 MWh or R0 = 280 MWh, the results barely change. The amount of electric energy that is extracted via the storage + turbine system varies from Hw = 534 MWh to 619 MWh. However, since both values are larger than what, according to the REE strategy last summer, the turbines effectively returned to the grid (H+ = 382.6 MWh), one does not observe any effect on the system performance.

If one considers the influence of R0 on the performances of the “optimal strategies” presented in Sect. IV, there is also no effect. In fact, this could have been predicted from the red curve in Figure 12 which shows that only for a value of R0 lower than 130 MWh, would one have observed a lowering of performances. Obviously within our set of assumptions, no improvement can be expected from an increase of the lower-reservoir capacity. In particular there is no technical incentive to use the sea as the lower reservoir.

Returning the parameter R0 to its original value (230 MWh), one now investigates how an increase of the installed wind power affects the upper limits of renewable fraction according to the optimal strategies described in Sect. IV. We assume that the addition of new wind turbines corresponds to a scaling up of the instantaneous wind power by a factor F within the interval from 1 to 1.5.

In Figure 13, the red curves show the evolution of the renewable fraction as a function of F. The dashed curve corresponds to step 1 of our research of an optimal scheme. As F increases from 1 to 1.5 the renewable fraction grows from 74.6 % to 83.1 %. When one introduces the hydraulic storage (Step 2) , the renewable fraction (solid curve) increases from 80.1 % to 98.4 %.

However, in addition to the cost of the new turbines there is a price to pay which is already evident from the shape of the red curves. While the total wind energy grows linearly with F, the renewable fraction curves saturate. As a matter of fact, for F = 1.3, the produced wind energy Wt is almost equal to the entire electric demand. Only wind production does not occur always when needed. This means that as F grows an ever larger fraction of the energy produced by the wind turbines finds no use whether by a direct injection to the grid (Wg in step 1) or via hydraulic storage (Wu in step 2). In Figure 13 this is illustrated by the blue curves which show the effective wind fraction. Again the dashed blue curve corresponds to step 1. The solid blue curve which corresponds to step 2 shows that for F=1.5 almost 20 % of the total wind energy is lost. There is obviously an optimal value for F which ultimately will be decided by the “internal” economy of this “wind + storage” system(15).

Figure 13: Evolution of the upper limit of the renewable fractions (red curves) and of the effective wind fraction (blue curves) as a function of the wind power scaling factor F. The dashed curves correspond to step 1 of the “optimal” scenario while the solid curves are obtained when the hydraulic storage system is being used additionally.

Figure 14 illustrates how the wind energy is used as a function of F. The useful part is the sum of the yellow and red areas. The blue area corresponds to the energy Wl which has to be destroyed as no use can be found for it, increases steadily with F. One should keep in mind that only a fraction 1/((1+ru)(1+rd)) of the red area is transformed in electric energy.

Figure 14: Evolution of the fractions Wg (yellow area), Wu (red area) and Wl (blue area) of the total wind energy as a function of the wind power scaling factor F.


(15) I use the adjective “internal” as opposed to “external” in which the economy of this system would be compared with that other more conventional ways of covering the electric demand of El Hierro. An “external” comparison would most probably show that the Gorona del Viento system is uneconomical.



In Section IV, in order to estimate the maximal performance that Gorona del Viento can reach, we have played with numbers making in particular extreme and probably unrealistic assumptions on the reactivity of diesel and hydraulic plants. On the other hand engineers do not play. They work with real machines for real people. They are assigned goals and they generally are under strong pressure from their bosses to reach success.

In [RA1] it was shown that for a year after its started working, the El Hierro wind + storage system was almost not used. The electric production relied almost entirely on the production of the fuel-fired plant. The situation changed in the last week of June 2015 so that the average dependence on fossil fuel is now lowered to ~50.5 %. However from the present analysis, it appears that, this summer, a large fraction (~37 %) of the energy produced by the wind turbines was wasted. Moreover, over two months, ~32 % of the energy made available via storage has not been sent to the grid to reduce the contribution of the fuel-fired plant.

It may well be that what we are presently witnessing, since two months, is only one more step towards a fuller implementation of the Gorona del Viento Wind+ storage system. One suspects that other steps will follow such that the call to fuel-fired production becomes each time less important.(16)

The Spanish engineers know that the eyes of the world, but mostly of the Spanish (national and regional level) politicians and of the European civil servants who decided to use public money to subsidize the entire operation, are focused on them. The engineers want to effect a successful technical demonstration and in particular avoid any blackout that may give them bad publicity (publicity is central for renewable energies). For some time, the engineers will satisfy themselves with occasional feats such as the “2-hour-100%-renewables” of last August 9 and count on the unabated enthusiasm of the ecological people to ensure the usual drumbeat towards the gullible media. In the background, the competent people will strive towards the desired goal by means of a steady and careful progression.

With the kind of wind production that occurred over the last two months, it seems that the present system can at most reach  80 % renewable coverage of the electric demand. To achieve higher renewable fractions more wind production is needed. On the other hand, the wind production figures which have never exceeded 7.6 MW over the summer seem to indicate that the 11.5 MW installed power which is announced is not yet fully operational (or operated). In Sect IV we have shown that, within an optimal scenario, more wind power leads to higher renewable fractions. The price to pay is that an ever growing fraction of the produced wind energy is destroyed. Presently boosting the capacity or the performances of the hydraulic storage system does not seem to be a priority.

Of course, a successful technical demonstration does not tell us anything about the economic validity of this very interesting experiment nor on the EROI it achieves. The figures found in the literature on the cost of the several production and storage components of the present system does not indicate that the model of Gorona del Viento is economically viable without public subsidies. Sometimes it is also said that it pioneers a solution that one can transfer to other islands, and even, for the most extreme aficionados, to the rest of the planet. One may have doubts about that, essentially because the investment cost of some major components of the system, in particular of the upper reservoir using the specific topography of the island (Figure 1), had been made before the Gorona del Viento project was envisaged.


(16)Obstacles to that evolution could be the economic policy of the ENDESA firm which has invested in the construction of the diesel plant and has negotiated long term contracts for the furniture of fuel.



I benefited from information, many suggestions and comments provided by Jean-Jacques Hérou(17) . I also had a fruitful exchange of mail with Roger Andrews which helped me clarify my ideas on how I wanted to treat the subject. At the end of this exchange, we sort of agreed to disagree.

Roger does not dispute the results; he does not agree with the assumptions: first with my last assumption (REE wind data do indeed correspond to the real total wind production not just to that part sent to the grid) and second with the fact that my assumptions amount to saying that the storage system is hermetically separated from the rest of the economy of the island. As a miser hoarding his pile of gold, the algorithm makes sure that every molecule of water assigned to storage is not diverted away. Implicitly, the pumping building is fed by an electric line whose meter tells REE what is H-. From the Pelton turbine building, an electric line comes out whose meter tells REE what is H+. Finally the meter on the electric line crossing the fence of the wind park tells REE what it should write in the Wt column of its data table. Once one makes these assumptions, it becomes impossible not to introduce the notion of destroyed wind energy as is done in this work. This is something that Roger would like to avoid.

His point is that the isolation of the “Wind + Storage” system which is implicit within my set of assumptions does not correspond to the reality of the island. Indeed the upper reservoir has been there before the Gorona del Viento experiment started and is used for the benefit of other human activities on the island. There is no reason to forbid (as I do explicitly in one of my assumption) that water can only enter the lower reservoir via the water pipeline coming from the upper reservoir. The lower reservoir is connected to the desalination plant. This plant was indeed used to feed it water in order to create the mechanical energy potential over more than one year (2014-15). I agree with all that. On the other hand, I am afraid that in the absence of additional information on the water coupling of the “Wind + storage” system with the rest of the water economy of the island it will not be possible to make a numerical analysis beyond the plotting the bare REE data.

Well, if Roger is right, the present document would just be another occasion to remind oneself of the statement by W. Pauli a day he was shown a correct but not very informative paper: “It is not even wrong”.


(17)J.J. Hérou is an engineer who spent the greater part of his career within the hydraulic department of the EDF production system. In recent years he has worked as a consultant on many projects dealing with hydraulic renewables (tide, waves, sea currents and wind+ hydraulic storage) all over the world. In particular, he has worked on the analysis of a “wind+storage” system for the island of Crete.

This entry was posted in Energy and tagged , , , , . Bookmark the permalink.

13 Responses to Another Visit to El Hierro

  1. Hubert: Re my “alternative scenario”.

    First a brief recap. The problem we have identified with REE’s generation data is that while diesel + wind + hydro = demand, the negative hydro values show far more water being pumped uphill than the reservoirs can handle if we assume, as you do, that they are a sealed system. You account for this as follows (please correct me if I’ve misinterpreted anything):

    1. REE’s negative hydro values include wind power that could have been generated and used to pump water uphill but which had to be wasted because the reservoirs were already “charged”.

    2. REE’s wind generation values include power actually generated plus this “wasted” wind power.

    Here’s my alternative explanation:

    1. The reservoirs aren’t a sealed system. Water is being pumped up to the upper reservoir (UR) and distributed from there to the island’s towns and farms, as called for by the project plan. (I have identified what appears to be an extraction pipeline in June 2013 Google Earth images. There may be more by now).

    2. REE’s negative hydro numbers show power taken from the grid and used to pump this water to the UR.

    3. REE’s wind numbers show wind power delivered to the grid. There was no significant wastage of wind power.

    4. The water comes from a desalination plant that supplies 18,000 cu m/day of water to the lower reservoir (estimated assuming a 1.5MW constant supply from the diesel plant and 2kWh consumed per cu m of water produced.)

    5. REE’s hydro numbers show that 2,800MWh (effectively 2,300MWh allowing for 18% losses) was used to pump water to the UR from the LR in the 66 days between June 26 and August 30. The 380,000 cu m UR has an energy storage capacity of 680 MWh, which translates to a total flow of 2,300/680*380,000 = 1,180,000 cu m of water into the UR (all numbers approximate). Averaged over 66 days this also gives 18,000 cu m/day. But It also presupposes that 1,180,000 – 380,000 = 800,000 cu m, or 12,000 cu m/day, was extracted from the upper reservoir, more than the average 9,000 cu m/day demand on El Hierro. However, Gorona del Viento says that the UR actually has a capacity of 550,000 cu m, and if we assume that both the UR and the 150,000 cu m LR started empty then the number drops to a more plausible 1,180,000 – 500,000 – 150,000 = 480,000 cu m, or 7,000 cu m/day.

    6. According to local authorities 50% of the island’s electricity consumption is allocated to the production and distribution of water. According to my scenario 48% of the electricity generated on El Hierro between June 26 and August 30 went to produce and distribute water (5,500 MWh, about evenly split between desalination and pumping, out of 11,400MWh total generation).

    This scenario balances, but it’s based on assumptions and still unsupported by hard data. (I emailed Gorona del Viento about ten days ago requesting some but haven’t had a reply; I suspect I will never get one.) Therefore it does not falsify your explanation. I present it simply as an alternative hypothesis.

    Now let me add some more data I just finished putting together. Here are a couple of plots of REE data updated through September 20 (yesterday). Note that on September 7 and September 11 generation exceeds demand, I think because the demand data were duplicated in the diesel column. I’ve corrected this in the plots but not in the data I sent you.

    Two things stand out:

    1. There was an abrupt change in operating procedures after September 11.

    2. Renewables generation has been decreasing since August and is now back close to pre-startup levels.

    Hubert, if you have time I’d be interested to see whether your algorithm can make any sense of this.

    • Euan Mearns says:

      Roger, I think the only way to understand this is to contact REE and find out what the problem / strategy is. It doesn’t make much sense having the wind park and not using the wind power.

      • Roger Andrews says:

        I’ve sent an email to REE asking them to explain what their generation numbers signify and another to Gorona del Viento asking them what the capacity of the desalination plant is (an important point because that’s where all the reservoir water comes from). We’ll see what happens.

  2. Leo Smith says:

    It doesn’t make much sense having the wind park and not using the wind power.


  3. robertok06 says:

    Problem solved… but not just for El Hierro… the whole planet… and not just for electricity, but for the total energy demand.

    You don’t believe it possible? Just ask the “experts”! 🙂

    The part I like the most is the “threats” part on nuclear, page 23… scaremongering turned art, courtesy of GreenPiss International.

    Just to understand the futility/stupidity of this excercise of GP, it is sufficient to take the caption of the photo on page 25

    “… at 9.5 square Miles (25.6 square kilometers), the facility is about one-third the size of Manhattan island, or the equivalent of 4,600 football fields. when operating at full capacity, the 550-Megawatt Plant Produces enough electricity to Power about 180,000 homes. ”

    … or otherwise said the 550 MWp californian monster could be replaced by a mini-reactor occupying at most ONE football field, which would generate night and day, on demand, and not only during the day with seasonal fluctuations.

    Pathological liars.

  4. Nador says:

    I presume 1+ru = 1/efficiency. Why do you use this unnecessarily complicated notation. Or is it a standard that engineers use?
    I have to agree with Rogers: some (most) of the “H-” is actually desalination, otherwise it is hard to make sense of the data.

    Another issue: is it possible to maintain correct grid synchronisation with only wind turbines without at least some diesel or water turbine generation?

  5. jacobress says:

    Seems that the main purpose of the system is to supply water (from desalination) to the population which live mostly uphill, on the mountain.
    The wind turbines were a good pretext to get the project funded by Euro_green enthusiasts.

    • The upper reservoir was originally intended to be used as storage for irrigation water, and that’s still part of the plan.

      Then someone got the idea of using it as part of a wind-powered pumped hydro storage system as well, with the output from the wind farm used to keep the upper reservoir full while the turbines on the downstream side delivered load-following power to the island’s grid. This innovative concept got around the problem of intermittency and was basically what got the project financed.

      Then it was decided that it would be better to deliver the wind power that the grid could handle directly to the grid rather than suffer the efficiency losses involved in using it to pump water uphill, so only the wind power that the grid can’t accept is used to pump water uphill. The rest goes directly to the grid.

      But the upper reservoir is still intended to deliver (and almost certainly is delivering) water to the island’s towns and farms.

      So the plant operators are having to balance wind and hydro output with diesel generation and electricity demand while maintaining the upper and lower reservoirs in balance and at the same time delivering water to the island. The desalination plant that supplies all the water has to be figured into the equation too. It’s not surprising they’re having problems making things work.

  6. jacobress says:

    It makes sense to use wind turbines for desalination and uphill pumping, as these tasks can be performed intermittently, when the wind blows.

  7. robertok06 says:

    Off-topic, sorry, but I thought it could be interesting… after all Hawaii is an island:

    OTEC: has any of you ever heard about it?

    How much does it cost, energetically, to move/pump water from 1000 m down?
    And what volumes of water need to be moved to generate, say, 1 kWh?


  8. Roger Andrews says:

    Here is how I think the El Hierro metering system works:

Comments are closed.