Economy & Energy
Year VIII -No 49:
April-May  2005  
ISSN 1518-2932

seta.gif (5908 bytes)No 49 Em Português






other e&e issues

e&e No 49


Main Page

The Future of the Brazilian Electric System

A “Destination Port” for the Brazilian Electric System


No 48 in adobe pdf for download

e&e links



Characteristics of the Brazilian Integrated
Electric System and its Projection up to 2035

Carlos Feu Alvim (coordinator)

José Israel Vargas

Othon Luiz Pinheiro da Silva

Omar Campos Ferreira

Frida Eidelman


The Brazilian electricity generation system is characterized by its continental dimension and by the strong predominance of hydroelectric generation. This makes it unique in the world.

Practically all electric systems have to satisfy the daily and seasonal demands variation. Furthermore, a predominantly hydroelectric system must adapt itself to the supply oscillations due to the seasonal rainfall regime and the variation of this regime along the years.

For these reasons the Brazilian system in its origin has adopted multi-annual reservoirs that can compensate all predictable variations in supply and demand.

The construction of large reservoirs, with flooding of large areas, was possible only at times when restrictions concerning the environmental and land use were less tight and when social or economic objections were less intense. Undertakings that made Sete Quedas and São Simão Channel disappear would not be acceptable in the present circumstances. Likewise, flooding large areas as carried out in the past cannot be considered anymore and energy planning must adapt itself to this reality.

As a consequence of this fact, the flooded area of the Belo Monte project was reduced by 1/3 (from 1,200 to 400 km2) without reduction of the power to be installed (about 11,000 M).[i]

The main objective of the present study is to sketch a scenario for the Brazilian electric system that is compatible with the use of the different energy sources available in the country. For this purpose priority will be given to renewable sources in order to guarantee the necessary autonomy and sustainability of the system at environmental costs and investment efforts that are acceptable to the Brazilian society. This scenario, that will be denominated “destination port for the Brazilian electric planning”, has as a “starting point” the present system, fundamentally based on hydroelectric generation

The study presents first a brief description of the seasonal character of the hydroelectric generation in Brazil and with the help of a simple computational model (described in Annex 1), describes the effect of introducing additional power with smaller accumulation regarding affluence. In the next step, the foreseen growth concerning electricity demand in Brazil and the transition from the existing generation base to the new configuration are examined.

In the first step are presented:

·          Description of seasonal character of the existing system ;

·          Description of the results obtained through simulation with a model that can describe the generation behavior and that of the  associated storage;

·          Test regarding verification of the model applied to real situations and examination of the model’s variables behavior for the more general case of the four integrated systems (SE, NE, South and North);

·          Description, in Annex 1, of Simple Model that Simulates Hydroelectric Systems and its application to typical cases.

In the second step it is described the transition under way in Brazil, from a system that is almost essentially hydroelectric and of large capacity to a system still predominantly hydroelectric but where the thermal plants will have an important role.

In order to describe this transition it is necessary to:

·          Introduce thermal generation in the simulation model (Annex 2)

·          Evaluate the economic growth in the considered horizon, for a  reference scenario, using the projetar_e program based on a semi-empiric macroeconomic model for Brazil (Annex 3)

·          Evaluate the total energy and electric energy demands corresponding to the considered scenario using the module integrated to the projetar_e model and the concept of equivalent energy (Annex 4)

·          Evaluate the future storage capacity considering the existing hydroelectric potential and the expected trends for storage,

·          Describe the electric energy production in the reference economic scenario, specifying the shares to be supplied by hydraulic, conventional thermal and nuclear energies and the respective installed capacity.

The Seasonal Character of the Brazilian Electric System

In Figure 1 it is shown the annual variation of the Affluent Natural Energy (ANE)[ii]   that represents the energy that can be generated from the water that flows to the dams. The curves for each region were made using data from ONS (National Electric System Operator) and are referred to the month of maximum affluence

It should be noticed that since most part of future generation will come from the North Region (of strong seasonal character) and considering that it does not seem politically possible the construction of large reservoirs in the region, it is expected that the problem concerning the variation of monthly available energy (along the year) will be aggravated if there is no change in the generation park profile.

Figure 1:  Affluent Natural Energy relative to monthly maximum value. It should be noted that the seasonal character of the North Region (in the integrated plants) is rather strong and the affluence from August to November is only 10% of the expected maximum value (month of March).

In Figure 2, the rainfall regime of the Southeast Region is compared with that of the South Region using the curves that represent the projection of the affluent natural energy of ONS (based on the historical behavior[iii]). In the case of the South Region, the historical behavior of the affluent energy does not show the seasonal regularity shown in the other regions. In this region, the “expected” curve (used in ONS’ projections) does not describe well the system since the months of larger rainfall are not repeated. However, the function used by ONS still seems the available option for simulating the behavior of that region. For other regions (see Annex 1) the simulation using a regular function [iv] permits a good description of the affluence.

Figure 2: Comparison between the seasonal character of the South and Southeast Regions. In the Southeast there is a certain regularity; in the  South the affluence, along the year, oscillates around 70% of the average value, presenting rainfall peaks much different from the expected value.

Concerning generation (that in general reflects demand[v]), the seasonal oscillations are less important than that observed in other countries.

Figure 3: Seasonal variation in generation is much larger in the USA (2002/2003) than in Brazil (1996/2000) . For Brazil no data were used after 2001 because of the changes that  the “blackout” has introduced in the last years. Sources: EIA/DOE/USA and ONS (Brazil).

The average annual demand oscillation in Brazil ((Figure 3) is compared with that of the USA[vi]. While the amplitude of the generation seasonal oscillation in Brazil (difference between maximum and  minimum) is about 2,0%[vii], that of the USA is 28%.On the other hand, in what regards offer, the seasonal variation of the affluent energy (hydroelectric energy offer) in the integrated Brazilian system is 120% relative to the average. Since in the USA the hydraulic generation share is only 8,5% (1991 to 2001), this problem is not important in that country and its system can use the available potential, even separated from the demand peak[viii]. Anyway, the existing experience regarding demand regulation in other countries can be useful for creating a new configuration for the Brazilian system.

The daily demand oscillation is important in Brazil, as illustrated by the demand behavior in the state of São Paulo (Figure 4)[ix].  The question is relevant since reducing the daily oscillation could permit a better use of the installed capacity along the year.

However, this daily oscillation problem in the use of electric energy has only an indirect connection with the problem under examination. As a first approximation it could be treated separately, as long as one considers the availability of a generation capacity higher than the average demand value (generally about 20%).

Figure 4: Load variation relative to the daily average value. The amplitude of  variation along the day reaches 40%.

System Simulation

In Annex 1 (Simple Model for Hydroelectric Systems Simulation) it was tried to simulate the operation of the Brazilian interconnected systems with a simple and transparent model that permits to understand the problem.

The option adopted was to represent affluent annual energy by a simple cosine function (representing annual periodic oscillation) to which constant value was added (equal to the average value of the minimum monthly natural affluence). The values used in the simulation of the Southeast Region are compared with the historical monthly average value of the region (Figure 5). For the other systems (see Annex 1), except the South one, the representation of the systems follows rather well the function mentioned above.

Figure 5: Simulation of the affluent natural energy using a cosine-type plus a constant value function . The representation is particularly good for the Southeast Region and it is adequate for the North Region but not for the South one.

Since the objective is to supply a semi-quantitative description of the problem, the results shown in Figure 5 are rather satisfactory.

In the simulation as well as in the representation of the observed data all parameters (stock, production, spilled and affluent energy) are given in GW.month.

In this first approach, it was assumed a static situation (demand, offer and storage capacity are constant) that permits, therefore,  to conceptually separate the energy storage problem from the dynamic growth problem. The same methodology can be applied to a situation where these variables follow a growing demand, since the values used are relative and so it is sufficient to change the reference basic value for each year.

Four types of systems relative to the storage capacity are described in Annex 1:

·          Systems with multi-annual regulation.

·          Systems with regulation for a normal or typical year (monthly affluences that follow the historical  average values)

·          Systems without storage (run-of-river plants)

·          Systems with partial regulation (less than one year).

The situation for 2003 is shown in Table 1 for the four systems existing in Brazil. The systems with multi-annual storage should be able to absorb oscillations of one year or more: it will suffice that the system stores the maximum expected variation for the period in which it is desired to guarantee generation. Therefore, the storage/production ratio can be lower than one year and nevertheless guarantee the probable variations of the affluence regime of several years. Anyway, even with the electric system criterion, the present North and South systems don’t have this regulation capacity and already depend on the exchange between regions and/or thermal generation.

Table 1 Storage Capacity of the Integrated Systems


Storage Capacity (GW month)

Monthly Production (GW month) / month

Storage/ production (months)

Storage/ Production (years)











SE + S















N + NE










(*)Includes all Itaipu

It should be noticed the fact that the multi-annual character of the system has been gradually reduced, as shown in Figure 6 for the SE region.

Figure 6: Storage/ (thermal generation load) ratio, expressed in months, along time that shows the multi-annual storage reduction of the Southeast system.

Source: ABRAGET: Lecture of Antônio Gama Rocha of UTE Norte Fluminense at the 1º Continuous Forum on Energy– Brazilian Energy Agenda – Rio de Janeiro 9-10/12/2003 – FGV and COOPEFURNAS

In the description that follows, the systems with “exact regulation for a year with normal affluence” and “run-of-river” were included; even though not corresponding to any region, they are important from the conceptual point of view.

The main results for the types of the studied systems, described in Annex 1, are presented in what follows.

System with Multi-annual Regulation

As an example, it is shown here the Southeast Region representation in year the 2003 as base[x]. The values used in the simulation are expressed relative to the average affluent energy (=100) and are shown in Table 2 that also shows data for the SE region that were used for the example case. The initial value of the stock was taken for simulating the 2001 “blackout”.

Table 2: Characteristics of the SE system and of the Simulation (Case 1)


SE Region


Average ANE

27,4 GW month


Maximum A NE

42,2 GW month



25,5 GW month


Installed Capacity

45,2 GW month



176,6 GW month


Minimum Spilling



(*)Reference value; the other values are relative to the average Affluent
Natural Energy (ANE) in the Southeast system (27,4 GW)

In Figure 7 are presented, month after month, the affluence (affluent natural energy), the accumulated stock, the spillover volume and the production. There are options in the program for the bi-annual and multi-annual representation shown in the figure. In the multi-annual graphic it is indicated the annual average affluence (=100 in a normal year), pointing out the “dry year”. It is shown the expected evolution for a situation similar to the one that occurred in the Southeast Region where a low stock and a decrease of the annual affluence caused a production deficit in 2001

In the simulation (Figure 7) it was considered the demand to be satisfied plus the minimum flow (94 + 3) lower than the average affluent natural energy. In this case, the stored energy would tend to grow and after a sufficient time, to be spilled. However, with a low (as shown) atmospheric precipitation (20% lower than usually) there would not be sufficient energy stock to maintain the necessary production.

Example Case  based on the Southeast Region

Average Affluence




Minimum Stock


Minimum Monthly Affluence


Accumulation Capacity


Maximum Stock


Monthly Production


Initial Stock




Maximum Monthly Affluence


Minimum Flow


Loss in dry year


Figure7: Expected evolution for a system with conditions analogous to those of the  Southeast Region in 2003. The presentation of this figure is similar to the screen  of the program, where it is possible to modify (white cells) the input data. Furthermore, it is possible to choose the type of graphic to be presented (bi- or multi-annual). The initial stock and the affluence decrease in the third year were considered so that they could simulate the “blackout” that occurred in 2001. Note: the % stock curves (scale on the right) practically coincide with those of the stock due to the scale adopted.

System with Regulation for a Typical Year

The represented system would be considered for the full use of the affluent energy in a year of normal precipitation (within the average value). It could include a storage considerably smaller than that of a system with multi-annual regulation.

For a system similar to the exemplified case (the same minimum and maximum flows relative to the average value), the stock could be twice the average monthly flow and only 30% higher than the month of largest affluence.

In this system, the stock of stored water would be annually “zeroed”, since the storage would coincide with that necessary to satisfy a normal year. All affluent energy can be used and it would be “the optimum system”, except for the predictable existence of years with precipitations lower than the average value, when supply is severely reduced. For a 20% reduction of affluence during a year, the electricity production would be affected for five months and in the most critical month it would drop to 40% of the demand.

Run-of-river System

It has also been simulated a system with no accumulation in which all generation would be made with the natural affluence. Depending on the expected rainfall regime for the region, an important fraction of the available energy would not be used. This fraction grows when the maximum natural flow / minimum flow ratio grows. As compensation, the intervention in the fluvial system would be minimal. It should be emphasized that the example case does not concern run-of-river plants using regulation with upstream dam but rather a system that was designed to operate entirely  with the minimum annual flow, a run-of-river one. Obviously, the system could have been designed to better use the affluent energy: it would suffice to have the installed capacity higher than the minimum one. In this case, its contribution to the generation would be larger and its contribution to the stability of the system would be smaller or negative.

Since the system is designed to operate at minimum affluence conditions in a normal year, its production is quite stable. In this case, it was designed to operate using the typical minimum monthly affluence, 46% of the annual affluent energy would be used.

The possible use (of the total annual affluent energy) for a plant of this type was evaluated for the different regions using the average curves of ONS. It would be 52% in the Southeast Region, 58% in the South Region, 32% in the Northeast Region and 21% in the North Region. Since it is in the last region that it is expected the largest generation expansion for supplying the integrated systems, the installation of this type of plant could limit the usable potential of the region. However, it should be remembered that in a system like this one the really usable potential should be re-evaluated since the energy use conditions could vary because of minor environmental problems due to the adopted flooding pattern where, for example, could be included uses that would be presently unlikely.

Still concerning the North Region, it should be remembered that the present affluence values along the year are based on the Tocantins River flow. However, for the two largest projects under study (Belo Monte plant and the Mamoré River) the flows present dry months, with low affluence relative to the average value, much similar to that of the present plants of the region, as shown in a note at the end of the present study.[1].

Systems with Partial  Regulation

The System with Partial Regulation is an intermediary type between that of regulation for one year and the run-of-river type. This type of system does not have the capacity of compensating the seasonal variations along the year but does not operate as a run-of-river system either. Water spillover is part of its normal procedure and only a fraction of energy is used. An example of this type of system is that operating in the North Region whose data, including those referring to the dry season, were the base for simulating a case studied in Annex 1.

Besides loosing production due to the uniform drop in the monthly affluence during the year, a new type of instability was detected in this type of system, caused by a variation of the monthly precipitation along the year (without reduction of the annual production), inducing an important production decrease. Therefore, this type of system presents a large instability regarding the rainfall regime, what indicates that the introduction of plants with strong seasonal character and low storage strongly requires complementation of other plants capable of sustaining the stability of the system.

Behavior of the Regional Systems  and Verification Test

As previously pointed out, the representation using the model should be the simplest possible and compatible with the correct description of the system. A good verification test of the equations used is to obtain by difference the value of the spillover volume + the evaporated volume. When coherent results are obtained it means that no important variable has been forgotten. Furthermore, the knowledge of the variables behavior of the model in the real situation is an important step for elaborating the scenario for the future. Comparing the results obtained is also a good intrinsic coherence test for the model. It was verified that in spite of the simplicity of the model, the reproduction of the real system is fairly good.

1. Southeast System

The simulation of the system with multi-regulation already shown (Figure 7) was made with characteristic data from the Southeast System. The situation previously simulated, as indicated in Figure 8, is much similar to the one that resulted in the 2001 “blackout”.

Figure 8: Values for the Southeast of the storage, the affluent natural energy and of energy production that resulted in the 2001 “blackout”. It should be observed that the minimum level of the reservoirs reached 18% in December 1999, but a superior initial stock (22% at the end of 2000) lead to the 2001 “blackout”

The values of the spillover energy are calculated by difference and are rather reliable, showing that the adopted approximation, considering the Southeast systems as a single plant, supplies satisfactory results. The low value of the spillover energy relative to production reveals, on the other hand, that the system is well administered. It should mainly be considered that, taking into account the random characteristic of the rainfall and requirements imposed to the flows, it is not always possible to avoid, as it would be desirable, spilling water, concomitantly wasting generation from other sources (in thermal plants) or whenever its is still possible to accumulate water in other reservoirs of the region. The perfect administration of the system becomes more difficult when, as in 2003, the stocks approached the maximum level. It should be still pointed out the growing institutional complexity of the present system relative to the previous one that was almost exclusively state-owned. Until the system adapts itself to the new circumstances, it can be foreseen that the rigidity of the contracts prevents the optimal use of the available hydroelectric energy.

As in the simulation (Figure 7) previously shown, the year 2001 started with low water stocks in the reservoirs and it was known that an additional decrease in the average annual affluence could cause “blackout”. In fact, in the previous year this possibility already existed.[xi] . Therefore, for 2001 it was chosen the same tactics adopted in the previous year, namely not revealing the risk to the consumers. As the rainfall was below the normal level, the government was forced to adopt “blackout”, which could have been anticipated and perhaps partly attenuated.

It is shown in Annex 1 that should the stocks be in their maximum value it would be possible to face the 20% decrease of the affluent energy without problem and even a 35% decrease. Naturally it is not the aim of the system to reach every year the maximum stock since it would be anti-economic to use the thermal contribution to accumulate a water stock to be later possibly wasted by spilling the stored water. The procedure adopted consists in fixing a “risk aversion” curve and activate the thermal plants whenever the storage moves away from the desired level[xii].

For the approach of the next part of the present study, that will deal with the role of the thermal complementation (present and future), it is interesting to observe how the demand of each system was satisfied, including the inter-regional energy exchange and the thermal generation.

Figure 9: Generation and electric energy exchange in the Southeast System. Exchange has been represented  as a negative value for export and positive for import, which permits obtaining the electric energy offer in the system. The total energy generated by Itaipu is shown.

In Figure 9 it is shown the electric energy offer (generation + exchange) in the Southeast region where it is included (coherently with which it has be done with the stock) all the production of Itaipú.

2. NE System

From the storage point of view, the NE System presents a situation similar to that of the SE System but, due to the fact that its demand is larger than the offer, it operates normally, except for special circumstances, by importing energy from other regions. The behavior of the NE system is shown in Figure 10.

Figure 10: The NE system has storage characteristics similar to those of the SE system. However, its larger dependency on the energy generated in other regions causes a lack of stability. Additionally, it calls attention the large participation of spilled energy, indicating the use of water for other purposes. As can be seen, the spilled energy presents a larger seasonal character in the dry season.

Regarding the transfer of energy and the thermal complementation, the historical values of the last years are presented in Figure 11. It should be noticed the dependency on imports from other regions and the near absence of thermal generation.

Figure 11: The NE System is characterized by the fact that it is an energy importer (from 1999 on), depending on the interconnection of the systems. The thermal generation share is still very small.

3 . North System

As has been pointed out, the North System is characterized by partial regulation. Its characteristics were used for one of the case studies of Annex 1 (Case 4). In the represented period (1996 to 2003) there has been considerable export for satisfying demand in other regions, notably the NE one. The system’s behavior (Figure 12) is quite similar to the simulated one, showing the two types of deficit caused in 2001 by a decrease of affluence along almost the entire year and in 2002 by a shift of affluence from the dry months to those of larger rainfall.

Figure 12: The North System has a small storage capacity and this fact makes it quite unstable. It should be noted that advance of the rainfall season (as it happened in 2002) can cause a collapse in supply which did not occur due to the existing interconnection, since thermal generation (in the interconnected part of the region) does not exist.

In Figure 13 one can note the exporting character of the system, with some importing episodes as it occurred at the end of 2002.

Figure 13: Energy generation in the North Region has been partially used for export. It should be noted the production deficits corresponding to the “blackout ” (2001) and to the rainfall shift (2002). Imports have permitted to face the 2002 deficit; there is no thermal production because all plants that integrate the system are hydroelectric plants.

4 . South System

The South System has also low storage capacity. Its particularity is that it is situated in a region with a rainfall regime different from the other ones. Furthermore, the rainfall cycle is not regular as in the other regions.

The irregularity of the rainfall regime makes it less attractive for applying the type of simulation used in the other regions. It is interesting to point out that the expected occurrence in the peak month (October) is that verified in a dry month in the Southeast region (rainy season only in the beginning) and would favor (whenever it occurred) a certain complementation relative to other regions. It also seems that there is some coincidence of dry years in the NE region and intense rainfall in the South and vice-versa[2]. This complementation enhances the interconnecting roles of the systems.

Examining the historical values of the last years in the South Region (Figure 14), it can be noticed that the adopted policy for the electric energy exploitation has adapted itself to the climatic reality. It should be also pointed out the more significant presence of thermal plants (operating at the base) that helps stabilizing the system (Figure 15). There has also been an intense energy exchange (between South and Southeast) that has permitted taking advantage of the differentiated rainfall.

Figure 14: Electric energy production in the South System has followed the availability of water. The system was able to have during years (including the “blackout” period) storage close to100%. There are transmission limitations that have prevented the use of the pointed out complementation.

Figure 15: Energy offer in the South Region that shows, besides an important variation in the generated hydroelectric energy, a participation of the thermal generation (mainly coal) in the base and a significant exchange with other regions. In this “exchange” is included the energy from Itaipu that, even though it is generated in the South Region , has its production included in the Southeast Region and reaches the South Region to be consumed through the interconnection between the two regions.

5. The Integrated Systems

If the Integrated Systems were perfectly interconnected they could be treated as a single one. When the operating variables of the set of systems are observed (Figure 16) it is verified that they do not follow the logic of only spilling water when the maximum storage is attained. The limitations of a perfect use of the capacity of the set are due, on one hand, to the generation capacity limit which is designed to satisfy an assured average demand and, on the other hand, to transmission limitations.

In 2000 and 2001, for example, even with low stocks, the plants of the North and South Systems spilled a significant quantity of water. Besides those physical limitations, operation errors can occur in the system that can lead to shortage situations.

Figure 16: United operation of the integrated systems showing that when storage reaches the maximum value (or to satisfy other needs), the spillover logic is not followed by the set, as it was observed in each of the previously shown systems. This is fundamentally due to the limited capacity either of generation or of transmission among them.

In Figure 17 it is shown the generation in the integrated systems, including the participation of the nuclear generation and that of conventional plants. It should be observed the high predominance of hydroelectric generation and the low presence of import. Once the crisis is over, thermal generation tends to be reduced as long as hydroelectric energy is available (and the stocks are maintained to minimize the risk).

Figure 17: Service of the integrated system, showing the magnitude of the supply problem in 2001, only partially supplied by thermal generation . It should be noted the important participation of nuclear energy during the crisis, which can also be used to help restore the water stock.

The full integration of the existing systems and the increase of equipment in some plants in operation – within the economic limits of these investments – can favor the best use of the regulation capacity of the systems. However, it should be remembered that if the additional generation capacity becomes simply a part of normal operation, there would be a smaller generated energy/storage capacity ratio and, consequently, a smaller stability in the system.

Inclusion of the Thermal Plants in the Simulation.

In Annex 1 it is presented the computer simulation of a hydroelectric system and its behavior regarding different affluences and storage capacities. The regulation is carried out with the accumulation capacity of the reservoirs, either to face the predicted seasonal oscillations or to absorb annual variations of the rainfall regime.

The inclusion of thermal plants in the simulation of the systems is described in Annex 2 and aims at studying the role of these plants in the system regulation considering the expected reduction of the storage capacity/average affluent energy ratio.

In the simulation it was adopted the premise that the electric systems would be administered so that fuel consumption is minimized. This means that available hydroelectric energy would be used at its  maximum. This also means that the stock of stored energy in the reservoirs would be close to its maximum at the end of the rainy season, defined here as the beginning of the month in which – in a typical year – the affluent natural energy (ANE)[xiii]  has a value below the annual average value. In a system where the electricity production is equal to the average affluence value, this means the beginning of the month when the reservoirs’ level stops rising because the water volume used in production normally exceeds what naturally flows to the reservoir. Electricity generation, hydraulic or thermal, is handled so that, at the beginning of the dry season, the passing stock permits an adequate safety for future generation but prevents the frequent waste of the stored energy. For this purpose, it is chosen a value for that stock that is a fixed fraction of the capacity, slightly lower than 100%.[xiv]

As the stocks would be restored by substituting hydroelectric generation by the thermal one, the determination of this maximum value would optimize the use of fuel of the latter, avoiding its use in order to spill water.

The procedure adopted in the program for each month considers that the affluence of the months that follow would be normal and determines the necessary thermal generation in order to reach the passage stock searched for. Practically, the system used in the program is similar to the one presently adopted by the administration of the electrical systems where it is established a “risk aversion” curve that helps programming the production in the plants.

The regulation using thermal plants was studied for various typical cases in Annex 2. The main studied systems (relative to storage capacity) were:

·          systems with multi-annual regulation ,

·          systems with regulation for a  normal or typical year (monthly affluence within the historical average value),

·          systems with partial regulation (less than one year) and

·          systems without storage (run-of-river)

In all cases it was assumed that the system aimed at using the average affluent natural energy of the power plants.

From the conceptual point of view, the electric systems should be able to face, as mentioned, the seasonal variations of a typical year and variations of the years with atypical precipitation. They should also be prepared to absorb unexpected demand variations caused by, for example, economical growth above the predicted one.

In Figure 18 it is illustrated the regulation of a system where there is a deficit situation because of thermal regulation and/or water stock absence. In this case, the problem could be solved either by increasing the stock or increasing the thermal capacity.. It is shown in the figure both types of solutions. In this specific case, it would be also possible to solve the deficit by increasing storage even more. Even then, some thermal capacity would be necessary for the dry years when the affluent energy would be below the normal value.

The objective of the regulation is to fulfill the period of smaller affluence. In an integrated system like the Brazilian one, even the hydroelectric regulation can be carried out using the storage capacity of other power plants. There already exist, and probably they would be more frequent in the future, power plants that operate as run-of–river with useful accumulation practically null. 

Annex 2 describes the results and conditions for the Reference Scenario. As in previous occasions, GDP projections are lower than the official ones. The GDP growth rates used were 



Average Rate

























The results were extrapolated for 2035 considering an GDP average growth rate of 5.1%, like the one projected between 2015 and 2025.


      Stock Increase Solution                   Thermal Capacity Increase Solution

                               month                                                         month

Figure 18: On top it is represented  a deficit situation manifested by total production decrease. The stock (limited to 120) is not sufficient to supply the affluence decrease ;on the other hand, during the months 4 to 6 the excess water is spilled. The system regulation can be carried out : by increasing the stock (center left) or increasing the thermal capacity (center right). The risk aversion curve serves the purpose of indicating (when the green curve is below it) the need of thermal production. The objective of regulation (thermal or hydraulic) is to fulfill the affluence decrease  of the last six months , as shown below for the two cases.

Economic Growth in the Reference Scenario

One of the greatest difficulties of energy planning in the governmental area is to handle projections that frequently reflect rather a growth wish than a real possibility. In order to disconnect the present study from this planning vice, it was developed a model that takes into account limitations to growth which was successfully applied in several studies [xv].

The GDP projections are shown in Figure 18. The main parameters are shown for selected years in Table 3.

Figure 19: Projected values for the GDP  expressed in constant 2003 dollars.

Table 3:  Reference Scenario: Main Values of the Projection 












US$bi 2003










% aa



















Capital/Product Ratio











US$bi 2003










% aa










External Trade

 [(X + M)/2]











US$bi 2003




















US$bi 2003



















Commercial Balance

US$bi 2003









Non Factors

Goods and Services


BSNF Exports

US$bi 2003









BSNF Imports

US$bi 2003









Transfers Abroad

US$bi 2003









 Apparent Consumption

US$bi 2003










Million inhab









Per capita Consumption










Per capita Consumption


% year









Per capita GDP










Per capita GDP


% year









It seems interesting to mention some of the premises and pre-conditions considered:

A positive data regarding the evolution in the last years – with expected effects on growth in the next ones – was the recovery of internal saving. In fact, this parameter, that corresponds to the non-consumed share of the GDP, has significantly recovered from 2000 on, as indicated in Figure 18, and it can retake the pace it had before the nineties.

Figure 20: The internal saving had a significant recovery in the last four years and it is essential for maintaining the growth process.

The internal saving recovery did not have up to now significant effect on investments yet because of the increase of transfers abroad that correspond to a negative external saving. In order to transform the internal saving increase into investment it is necessary, as shown in Figure 21, that this transfer should be gradually reduced in the next years. It was also considered that the saving would slightly increase, tending to 27% of the GDP. It should be noticed in this figure that the seven-point drop (in percent of the GDP) of the internal saving has not been compensated in the nineties by the afflux of external saving that was limited to 2% of the GDP after the Real Plan.

Figure 21: In order to sustain growth it is necessary to maintain internal saving and reduce the level of transfers abroad (balance of  services non factors of the debt and goods) .

The assumption here presented regarding the internal saving means that the decrease of resources transferred abroad would occur together with the corresponding increase of investment to sustain growth.

Another interesting point of the analysis made is the control of the capital productivity drop that restrained growth in the eighties. Considering the present Brazilian development phase, the capital productivity is quite low and could be increased.

A good part of the 2004 growth was due to a better use of the installed capacity, which was 3% higher than the historical average value of the last years, even though it is eleven percent points below the value observed in the year of maximum use (1973). On one hand, this has indicated the need of replacing investments, on the other hand, it can mean a trend to a larger use of the installed capacity that, if sustained, could result in definite gains in capital productivity.

Note: this program is a result of the model described in the book “Brasil: o Crescimento Possível”, published in 1996 based on data up to 1993 then available. At that time, contradicting the general optimism, it was assumed a limitation of 3.5% to the growth rate between 1993 and 2003. At that period, this growth limit was considered a terribly pessimistic scenario because it was believed that Brazil was entering a virtuous growth circle with the economic liberalization.[xvi]

The result of the policy in the nineties, stressed by the Real Plan, was (in terms of the real economy) a hiatus regarding internal saving. Even with the input of external resources it was not possible to rebuild the previous investment rates that dropped about 4 to 5% of the GDP. It is the first time that the reference scenario (inertial) of the program presents the possibility of growth higher than 4%. In the next years, it is still foreseen lack of investment due to the unprecedented transfers of resources abroad. From 2007 on, the internal saving, recovered in the last years, would permit a sustained growth above 4%.

Growth of Total Energy and of Electric Energy Demand, in Equivalent Energy

Energy demand is a function of the economic growth in its quantitative and qualitative aspects[xvii]. The different forms of energy have different efficiencies according to their use and this should be taken into account in the global demand. For this purpose, it was basically considered the use efficiency for different ends by using the equivalent energy concept. The representation of energy demand demand in equivalent energy was developed by the e&e staff and it is described in an article published in the e&e periodical [xviii].

The evolution of the equivalent energy demand by product is calculated for Brazil between 1970 and 2003 and estimated for 1996 for other countries. The observed trend for Brazil is extrapolated considering the value of this structural parameter in several developed countries.

Different from the primary energy /GDP parameter, that decreases with development (by substituting some energy forms by others more efficient ones), the equivalent energy /GDP ratio depends little on the development level of the country, as shown in Annex 4. A program coupled with the macroeconomic model - projetar_e – permits to project the equivalent energy and the electric energy demand considering the hypothesis adopted for the economic growth.

           The projection of the equivalent energy/GDP parameter is shown in Figure 22.

Figure 22: Evolution of  equivalent energy parameter  by dollar of the GDP (1994 prices) and its  extrapolation.

The electric energy demand is calculated from a projection of its share in the equivalent energy demand in Brazil and in different countries. The electric energy share in the consumed energy grows according to the development of the country. It should be noted that this share is already relatively high for its development level and it is higher than those of Spain, Germany and United Kingdom (1996 data). Even so, it is projected for the end of the period an increase of 14% of the electric energy share in the total consumed energy (values given in equivalent energy). The evolution of the electric energy share in the total energy is shown in Figure 23.

In Figure 24 it is shown the expected evolution of the GDP, the total equivalent energy and electric energy in values relative to 1998.

The growth rates from 1995 to 2000 and those projected for the period are:




















 Equivalent Energy









Electric Energy









Figure 23: Projection of the Electric Energy Share.

Figure 24: Projections of the GDP, total equivalent energy and electric energy up to 2035.

For this evaluation it is mainly important the electricity demand satisfied by the public service power plants[xix]. Therefore, it is necessary to subtract from the global demand the transmission and distribution losses and the demand satisfied by the auto-producers and by imports. The historical and extrapolated values regarding the participation in demand of these three items are shown in Figure 25.

Figure 25: Values of  losses, auto-producers production and imports used for calculating the  the public service plants demand.

Table 4: Values Estimated for Public Service Plants in TWh/year

















































Public Service Power Plants








Storage Capacity

The Brazilian electric system was built based on large reservoirs that are able to favor annual and multi-annual regulation of the energy produced.

As previously shown, this regulation capacity has been gradually reduced. The question now is not if Brazil will use the existing hydroelectric potential but how this use could be regulated through reservoirs.

The acknowledgment in the official discourse that the Brazilian energy system is not an essentially hydroelectric system anymore but a mixed system, predominantly hydroelectric, that considers an essential thermal participation, implies that the thermal plants will partially have this regulating function.

The fact that the complementary characteristic of the thermal energy in the system has not been explicitly declared has lead, in the opinion of the authors, to errors regarding the strategy of its introduction.

Evolution of the Storage Capacity

Figure 6 has shown how the storage capacity (in energy) has developed relative to the hydroelectric production.

One can also evaluate the storage capacity decrease by examining the behavior of the reservoir capacity increase as a function of the power increase. According to ONS (Annual Planning of the Energy Operation, Year 2004), the average value of the plants that are part of the integrated system was 2.99 hm3 by MW of installed power (effective power) in 2003. In the programmed additional power until 2006, the foreseen increase was 1.24 hm3 by MW (41% of the existing one).

In Figure 26 it is shown the accumulation capacity of water volume that can be stored (in hm3) of power plants that represent 75% of the total existing accumulation with the dates when the reservoirs started to be used

Figure 26: Evolution of the accumulation capacity of plants that represent 75% of total generation. (ONS list of main reservoirs)

The data shown in Figure 26 are based on the ONS list of the main reservoirs. Even though it does not include 100% of the storage capacity, it calls attention to the fact that in this set, the storage capacity growth was only 3% in 20 years. On the other hand, the installed capacity (total of the country) grew almost linearly between 1970 and 2003 (five times in the period) and, as observed in Figure 27, doubled in the last twenty years.

Installed Electric Power

Figure 27: Growth of the installed power; the paths between 1950 and 1968 and between this year and 2003 are almost linear.

The construction of large reservoirs, considering the present environmental restrictions, is a challenge for the expansion of the generation system that the new institutional structure is trying to solve. Anyway, one cannot foresee in the medium term a significant increase of reservoirs with multi-annual regulation capacity.

The Hydroelectric Preference

In any system it is natural, when thermal and hydroelectric energies are available, to use the one with lower operational cost which, in the case of thermal plants, is closely linked to the used fuel price. In a system administered by the state it is easy to establish the preference and to determine the remuneration of the thermoelectric utilities considering the fuel availability, as has been done during decades concerning the Brazilian thermal plants using fuel oil. It is interesting to know what happens in a free market system where the contracts could inhibit the logical choice.

In the USA the question of offer variation due to the occurrence of rainfall practically does not exist. On the other hand, there is a strong seasonal demand oscillation with well marked summer and winter peaks.

Electric energy generation in the USA is predominantly made using mineral coal, responsible for 51% of the 2002 and 2003 production. The second source is the nuclear one (20%), followed by natural gas (17%). Hydroelectric energy represents 7% of the total generation[xx]

The seasonal variation regulation is carried out using a group of energy sources used for electricity generation. As can be observed in Figure 28, the largest contribution is from coal and natural gas (summer peaks). The latter, strongly used in domestic heating, is an alternative generation source in summer. Nuclear energy and “Others” also contribute to this regulation.

Figure 28: Variation of electric energy production in the USA (values above the monthly minimum) show that regulation is mainly made by mineral coal and natural gas (summer peak); hydroelectric energy production follows it own pace in opposition to other sources.

Regarding hydroelectric generation, it practically does not contribute to regulation. In opposition to that, the other sources adapt themselves to its seasonal behavior. Even though its annual generation curve shows that there is a regulation for the system (50% oscillation amplitude, on the average), the priority given to water generation causes a shift in the other sources, whenever necessary. That is, in the American system, where the market economy is dominant, the hydroelectric generation priority is established, following the economic logic.

Thermal Complement

The smaller storage capacity in the new undertakings implies that the thermal complement has a regulation role in the new Brazilian hydro-thermal system. This implies the use of thermal plants for complementing the hydroelectric ones whenever there is low precipitation or whenever demand has grown above expectations.

The consequences of this complementing character of the thermal plants has not been well assimilated by the system that considers (or have considered) capacity factors around 85% for their viability.

It is assumed here that a rational system will not spill water while it burns fuel for thermal generation. In a free market system, energy sale among utilities would solve this problem and in a state-controlled system, it is hoped that the administrators will avoid irrationality.

The consequence is that the use of thermal plants in a predominantly hydroelectric system will obligatorily be low. In Annex 4 different situations of mixed thermal and hydroelectric regulation were studied and the expected use of thermal plants was low in almost all of them.

Figure 29 shows an example where demand is 20% higher than the average affluent natural energy and the stock relative to this parameter is 3 months. In this situation, the thermal plants must satisfy on the average 20% of the demand. For an affluence of 100, the thermal installed capacity should be 66.[xxi] There are situations where the thermal plants do not operate for one year, as that of the second year in the example of Figure 29.

Figure 29: Example of a hydro-thermoelectric system with demand 20% higher than the average affluent natural energy and storage/affluence of 3 months.

The situation presented illustrates the operational and financial problem that could occur (and is already occurring) for thermal utilities if there is no adequate compensating remuneration mechanism when there is no demand. It should be noted that the problem shown can occur both due to excess of rainfall and to frustrated expectation regarding growth of demand. Considering the hypothesis that thermal generation is the only consumption alternative for a fuel (as associated natural gas), one runs the risk of spilling more water in order to burn gas. The inadequate situation of using associated natural gas or gas under the take-or-pay type contract for thermal generation is inherent to a predominantly hydroelectric system and not a circumstance caused by demand forecast error, even though this can also be one of the factors that has reduced the average use of thermal plants.

Participation of Thermal Plants in Electricity Generation

The evolution of the electricity production installed capacity is shown in Figure 30. In the last years, as a consequence of the 2001 supply crisis and the projected availability of imported natural gas, there has been a strong increase of thermal plants.

Figure 30: The installed capacity in Brazil is predominantly hydroelectric, and from the 2001 supply crisis on thermal plants participation was resumed.

The share of thermal plants in the Brazilian generation park is shown in Figure 31. The generation capacity has slightly exceeded 16% of the installed capacity share existing at the beginning of the seventies.

Figure 31: Resuming thermal plants share in the generation capacity is an important fact after the 2001 crisis.

Figure 32: Participation in electricity generation by type of plant. In the detail below it can be observed that the largest thermal electricity generation was associated with the increase of economic growth (beginning of the 1970s and Cruzado Plan in 1986) or to difficulties regarding hydroelectric supply (2001).

In terms of generated capacity share, Brazil is essentially a hydroelectric country, as shown in Figure 32. As the thermal plants compete with a primary source of practically zero cost in the already installed capacity, the offer and demand oscillations act directly on the use of these power plants..[xxii]

Due to reasons already mentioned, the capacity factor of thermal plants is historically low in Brazil, as shown in Figure 32.

Figure 33: Capacity Factors for different types of plants

Projection of Thermal Plants Participation to Serve Regulation

For a long time it has been assumed in Brazil that its system would continue to be essentially hydroelectric and that thermal energy would play its role when the hydroelectric potential would be exhausted. This idea is still prevalent since only recently the regulating function has been emphasized. The low use inherent to this role regarding thermal plants practically has not been mentioned.

As the projection of demand and of the share corresponding to the public service plants has been made in a previous item, the next step is to determine the necessary generation park and its composition in terms of plant type. Modules have been added to the program for determining energy demand and the projection of the necessary thermal capacity. These modules are coupled with the macroeconomic module, allowing for considering in the energy part and in the electric plants planning other hypothesis regarding growth or energy policy.

In the simulations regarding the thermal energy need, it has been used a minimum value for its average generation share; this has been carried out assuming a demand above the average affluence. In the example of Figure 29 this minimum value is 20%.

In the past period studied here (from 1970 to 2003), this share varied between 2.6% and 12.8% with an average value of 5.6%. Maintaining an average percent use of thermal plants has operational reasons (the main one is to respond quickly to the demand growth above the forecast value) and political reasons (maintenance of mineral coal demand and of technology in the different types of plants). The 10% value (in generation) was adopted in the present study in order to initially indicate the need of thermal generation for the new plants. This percent value can be considered as a thermal base independent of the regulation needs.

The growth of thermal plants share in the future would be conditioned to this regulation need and to the eventual exhaust of the hydroelectric potential due to physical reasons and cost questions.

The stored energy/average affluent energy ratio of the existing plants is 5.8, as shown in Table 5. For the plants under construction, this data is not available. However, it is known that the stored water volume/installed power ratio has a value 59% lower than the present ones. Assuming the proportionality between these two ratios[xxiii] , the stored energy/average affluent energy ratio would be 2.4 months.

Figure 34 shows the distribution of the remaining electric potential calculated until 2003 (inventory + estimation). Data regarding the North Region potential represent 56% of the non used potential. The storage/average affluence ratio in the plants of this region that are part of the integrated system is 2.1 months. It should be remembered that the potential used in the North Region (fundamentally Tucuruí and Serra da Mesa) were much objected for environmental reasons. That is, in spite of the resistance, these plants already forecast an insufficient storage and were conceived to use the regulation capacity already existing in the integrated system. Even Itaipú, that can be considered a project-symbol of the large undertakings era, has a storage/affluence ratio of only 0.7 month[xxiv]

Figure 34: Remaining potential distribution by region for the year 2003 (estimate + inventory potential)

Table 5: Values of the storage capacity relative to the natural affluent energy


Storage capacity (Gw month

Monthly production (Gw month) / month

Storage/ production (months)


Monthly average(Gw)

Storage/ NAE (months)













se + s


















n + ne






Integrated Systems






For the hydroelectric plants, four hypotheses have been considered for the stored energy/affluent energy ratio: 2.5, 2.0, 1.5 and 1.0. From these values, one has the thermal plants share in the future and the utilization factor corresponding to their regulating role.

As a starting point one considers the electric energy demand and the fraction corresponding to the already estimated public service plants. The capacity factor projection was carried out for the already existing thermal plants and for the hydraulic plants (existing and future ones) based on the historical behavior. For the new thermal plants, it was taken into account their limitation when used as system regulators.

Figure 35: Projection of the capacity factors to be used for the existing hydro and thermoelectric plants. For the new thermoelectric plants it will be taken into account their capacity factors as system regulators.

From the existing storage capacity in the last year and the calculated capacity factors one can project the additional demand to be satisfied by the new power plants.[xxv]

In order to determine the capacity factors and the thermal regulation need the results presented in Annex 2 were used, which consider historical values of 8 years for the Southeast Region. The affluence oscillations around the average value observed for the Southeast Region were the base for simulations adapted to seasonal parameters of the other regions.

In Figure 36 these results are summarized and are represented as functions used for interpolations in the program developed for the calculation. One expects that the stored energy/ affluence ratio will keep decreasing along the next years. For this reason the storage capacity is represented as decreasing values in the figure. It should be pointed out that in the Brazilian case the minimum value has been fixed by the existing storage capacity. If all the present potential estimation and inventory value (about 140 GW) were used without increase of the regulation reservoirs, the stored energy/affluence ratio would still be 1.8 months when all the potential is used.

What is indicated in the graphic of the thermal capacity as a function of the accumulated energy is that the relative thermal capacity relative to the demand (or affluent energy) grows when the storage capacity is reduced.

The thermal regulation capacity is planned as the minimum one that guarantees the fixed production. In these conditions, when the thermal plants participation in regulation grows, it is foreseen an increase of spilled water, mainly in the more rainy years..

Figure 36: Fitting of values regarding the participation of thermoelectric plants and of the installed thermoelectric capacity for complementing by reservoirs, as the storage capacity is reduced (values relative to a monthly affluence = 100).

To obtain the hydro and thermoelectric capacity values to be installed in order to satisfy demand, the procedure described in what follows was adopted:


Calculation Procedure

  • The foreseen demand for the future years was fixed;
  • The demand satisfied by the existing plants at the end of 2003 was estimated using extrapolated capacity factors from the historical values;
  • It was considered that the non satisfied demand would cause an increase in the corresponding (thermal and hydraulic) installed power;
  • The storage value of the previous year was considered for deducing the thermal capacity necessary for regulation and the expected capacity factor, using the curves shown in Figure 39;
  • A (exogenous) fraction of this capacity is considered to be nuclear, added in 1.3GW modules;
  • From the thermal regulation capacity and the expected capacity factor, the expected generation was determined;
  • It is determined the demand not satisfied by the previous capacity and by the new thermal plants that must be satisfied by the new hydroelectric plants;
  • Using the projected capacity factor for the hydraulic plants, the necessary hydraulic capacity is determined (*);
  • An exogenous value of the maximum stored energy/ affluent energy ratio (az) was assigned to this new capacity;
  • The new global value of the maximum stored energy/affluent energy is determined from the proportional average value of the data relative to the existing plants and of the additional capacity;
  • The procedure is repeated for the following year.

(*) The use of the utilization factor of the hydroelectric plants for the future ones, derived from the historical behavior, can be objected since the presence of plants operating with a small water regulating reservoir or none at all can be increased. That would require a larger installed hydroelectric capacity and would decrease the capacity factor but it would not modify the need of thermal regulation because the reduction of the storage/ affluence ratio has already been considered. However, it should be remembered that the present quantity of machines in the system exceeds the generation needs for the next years, according to ONS' operation planning.


Figure 37 shows the expected values for the installed capacity in the 2035 horizon for az=2.0 for the new plants. As first approximation, the limit for hydraulic plants has not been considered, what will be carried out later.

Figure 37: Projection of the installed capacity considering that the thermal plants will supply 10% of demand and the regulation needs. The physical and economical limits of the hydroelectric plants have not been considered.

Figure 38: Participation of thermal plants in electricity production and in capacity generation for az= 2 and for the reference scenario adopted here.

Figure 39 shows the results of thermal demand for different values of az.

Figure 39: Thermal capacity necessary for system regulation (with no limitation for hydroelectric plants) in different hypotheses regarding storage capacity of the new plants.

It can be observed that until 2010 the need of new thermal plants is slightly affected by the az ratio in the new plants. That is due not only to the existing storage capacity in the system but also to the thermal capacity that the reaction to the 2001 supply crisis (and additional offer of natural gas) has accumulated. However, in any case, the need of regulation will cause a considerable increase of the additional thermal capacity.

Figure 41 shows the participation of the thermal plants in the total electricity generation in the public service plants for different values of az.

Figure 40: Participation of thermal plants in the generation capacity for system regulation.

Other Limits for the Hydroelectric Plants

Besides the regulation problem that was examined, other limitations, such as the physical potential and the growing economic cost of the hydroelectric plants, should be considered.[xxvi]

Concerning the physical potential, the Ministry of Mines and Energy publishes periodical evaluations of the inventory and estimated potential whose evolution is shown in Figure 41. Past data were used for projecting the potential in order to include probable additions.

Figure 41: The evolution of the inventory or estimated hydroelectric potential and its extrapolation show the exhaustion limit (for 80% of the total potential) to be reached between 2030 and 2035.

The total potential (inventory + estimation) published in BEN 2004 was (for 2003) estimated as 143Gw of firm energy. Eletrobrás gives a value of 260GW for the total potential in its site. The values are coherent if we admit a 0.55 capacity factor (fc). If 80% of the potential is used, we have a limit of 114 GW of firm energy or (considering that fc=0.52, used here for the end of the period) 219GW of installed energy.

Extrapolating data from Figure 41, one gets a potential of 175 GW of firm energy or 337 GW of installed capacity. In an optimistic hypothesis one can reach 80% of that potential, what would correspond to a limit of 270 GW of installed capacity producing 140 GW of firm energy.

On the thermal plants side there is a strong dependence on fuel costs except for nuclear energy. In fact, studies carried out by e&e, based on fuel costs and investments of the nineties, indicate that natural gas cost represents 2/3 of the generation cost in a simple cycle and more than half in a combined cycle plant. The cost of the Brazilian coal represents little more than 1/3 of the generation cost. In what concerns nuclear plants fuel represents a little more than 10% of the electricity cost. That fact makes them a good reference for examining the limits of hydroelectricity generation. A value higher than that obtained in the study (65 US$/MWh) would be an acceptable threshold. This means a limit of about installed 140 GW.

That is, according to one or the other criterion, the limit for the hydroelectric capacity would be between installed 140 GW and 270GW. [xxvii].

An important remark about fuel is that it should admit discontinuous use along time, alternating months of intense operation with low operation. One of the conditions for the fuel is the ease of storage. The complementary character of the thermal plants means that most of the time they will be remunerated just for their availability. In order to favor competition in normal generation (outside the regulation role), one of the models is that where the state (through state-owned utilities or contracting private ones) would control regulation, that would be charged from the production pool.

Projection of the Generation Capacity to be Installed

In order to project the installed capacity necessary for satisfying an identified demand, it is essential to adopt some premises that will be explained in what follows, based on different simulations made, and to examine existent experience in Brazil and abroad.

·          Maximum stored energy / affluence ratio az = 2,0 for the next power plants,

·          10% minimum thermal capacity relative to the average demand (approximately 5% of the total installed capacity),

·          270 GW limit of the Hydroelectric Capacity of ,

·          30 % of nuclear share in the new thermal capacity.

The first parameter is based on the present plants of the North region. The second one can be considered quite conservative since the present thermal generation capacity relative to the average demand is 36%. The limit for using the hydroelectric capacity corresponds practically to using fully the hydroelectric potential. The nuclear energy share in thermal generation is close to that of the OECD countries (28%) and lower that that of the European Union (35%). In this particular, it should be considered that all these countries already had a thermal base installed before the petroleum crisis and restrictions regarding the increase of carbon emissions. Brazil, that is establishing a thermal park, will tend to install it by minimizing fuels that present supply risks or sudden price variations. In a conservative way, participation has been maintained close to the present one.

These premises, the economic growth and the energy consumption projections permit to establish a panorama of demand and  of generation capacity evolution.

To take into account the exhaustion of the hydroelectric potential, it was constructed in the first place a scenario without limit regarding its associated generation, as previously shown (Figure 37). Next, a limitation on the hydroelectric capacity at the end of the period was introduced and it has been transferred to the thermal plants the installed capacity growth to satisfy the imposed limitation (an exogenous time constant is used so that the exchange between the two types of plants is not abrupt).

Figure 42 shows the expected evolution of the energy production in public service plants and Figure 43, the share of electric energy generation by type of plant. The scale amplification in the lower part of the figure shows that the hypothesis regarding the plants’ share is quite conservative until 2030 and it would help, until that year, regulating the system.

Figure 42: Projection of energy production by public service plants with the indication of the foreseen generation type.

Figure 43: Participation of plants in the projected energy generation showing that the distribution among the different plants , until 2025, remains very close, in terms of participation , to that value in 2001 and 2002.The system would continue to be predominantly hydroelectric (below is shown the detailed shares of the thermal plants)

The historical installed capacity and its projection by type of plant are shown in Figure 44.

Figure 44: Evolution and projection of capacity by type of plant.

Figure 45 shows the annual increase of the projected installed capacity to satisfy demand.

Figure 45: Additional capacity necessary for generation; it should be noticed that the nuclear potential addition was considered as units of 1.3 GW

In Table 6 it is shown, for chosen years, the foreseen evolution of the installed capacity and of electricity production by type of plant.

Table 6: Installed Capacity and Electricity Production by Type of Plant










Installed Capacity GW (Public Service Plants)










Conventional Thermal


















Nuclear Plants*


















Electricity Production TWh (Public Service Plants)










Conventional Thermal



























* of 1.3 GW each

In Figure 45 it can be observed that the introduction of nuclear plants was considered as units of 1.3 GW. The projection of new thermal plants needed was considered as a whole while for the nuclear plants it was considered the introduction of each plant. The period of its introduction corresponds to the situation where 30% of thermal demand exceeds 50% of the capacity of the next plant. The introduction of the first additional plant (corresponding to Angra 3) is situated in 2011 and the following ones for 2016 and 2019. For the next decade, it was estimated the start of operation of ten power plants and for the first half of the thirties, thirteen more plants (a total of 26 additional plants)

The projection of the capacity factor for the different types of plants is shown in Figure 46. As expected, the capacity factor is low for the conventional thermal plants due to their complementary character relative to hydroelectric plants. Only from 2030 on the larger use of thermal plants on the base of the system will improve their utilization rate.

It has also be considered the hypothesis of limiting the hydroelectric potential to installed 140 GW.[xxviii]  In this case, the capacity of the installed thermal plants would reach 214 GW of which 62 GW are nuclear energy.

Figure 46: Capacity factor by type of plant, showing that the complementary character of thermal generation implies a low utilization index of the installed thermal capacity. In the case of nuclear plants the low cost of their fuel favors their use.


The study has shown that the need of regulating the hydroelectric system is the predominant factor that would lead to the need of implementing a thermal park in Brazil until the end of the second decade. The need for this thermal regulation is due to the difficulty of constructing large reservoirs as it has been done in the past. The presence of thermal plants became a necessary condition for expanding the hydroelectric system.

In fact, the regulating capacity of the systems has been reduced to one third of the value existing in 1970 and the new hydroelectric plants, with smaller storage capacity relative to production, may increase the instability.

Another important aspect that arises when the systems are simulated, except for the South Region, is that there is no complementation between the rainfall regimes of the rivers of the regions and the introduction of generation of the North Region's rivers into the interconnected system should worsen the seasonal problem, at least until it can be incorporated into the generation set (if at all) the tributaries of the left margin of the Amazon. In fact, at least the tributaries of the right margin have dry months that are coincident with those of the SE, NE and CO regions. That is, even though the diversity of the basins may lessen, due to statistic questions, the effects of occurrence of abnormal dry periods, it should not been assumed that its incorporation in the system will contribute to reduce the seasonal character of a typical year.

In the considered reference scenario, quite optimistic hypotheses relative to the hydroelectric potential were considered, including what concerns the evolution of the storage capacity. Even then, the need of additional regulation will be felt from 2010 on. Until then, the existing storage capacity and the thermal plants that have already been installed would be used in the regulation.

                The thermal plants used in this function will operate in a discontinuous system, both due to do the seasonal behavior of the hydroelectric offer and to the need of using thermal plants to compensate for precipitation's negative variations and the need of halting them when there is excessive rainfall in one or more years. Equating the investment involved in all types of electric plants is a challenge to be confronted.

                The larger interconnection and the increase of the generation capacity of the systems (if not used for regular production), due to the installation of new machines, can also result in an improvement of the stability of the whole system.

                Power plants fueled by associated natural gas or by take-or-pay type contract are not easy to be used regarding prolonged demand variations. Some plants with low utilization rate could also operate with petroleum products.

                The most probable option is that the thermal generation park would have a set of plants with different fuels where nuclear plants would have a central role because, once they are installed, they have the lowest fuel cost.

                From 2025 on the hydroelectric potential would be almost exhausted; this would require the installation of thermal power plants operating at the base and would continue to be expanded after 2035.

                In the considered horizon, it is not expected any significant change in electric energy generation at the world level. In effect, the centenary experience of the sector shows that changes in the generation profile are slow and not in phase in countries of the third world. In practical terms, the inexistence of expected technological solutions for the next fifteen years in the developed countries implies that they will hot happen in Brazil in the 30-year horizon studied here.

                The perspectives in the medium term indicate an important role for nuclear generation. For the other developing countries, the nuclear option may be denied under non-proliferation allegations. Brazil – that already commercially dominates the PWR reactor fuel cycle, including the sensitive enrichment step - has the opportunity to maintain access to this type of energy. The termination of the Angra 3 power plant - that is perfectly inserted in the energy needs in the near future, is an important step in the consolidation of access to this energy form.

List of Annexes

(Available only in Portuguese)


[i] Obviously this means –except for a rough error in the original project – either  considerable cost increase or a larger seasonal aspect introduced in the system

[ii] ANE –Affluent Natural Energy is, according to ONS, the sum of the  natural energy that flows toward all plants of that region. Affluent natural energy of one plant is the multiplication of  the natural flow that flows to that plant  by its productivity considering that the volume of the reservoir is up to 65% of its maximum value.

[iii] ONS supplies flow statistics between 1931 and 2001 of 318 points in the Integrated System. Together with installed power data, one can construct historical curve to be used in the annual projections.

[iv] Co-sine function with annual period plus a  constant.

[v] The gross demand (that includes losses) is supplied in almost its totality by generation, since the import of electric energy except for that of Itaipu already included – has not been  relevant  and direct storage of electricity is negligible relative to consumption.

[vi] Source: Energy Information Administration (EIA)/ Department of Energy USA

[vii] The amplitude considered here is lower than the so-called seasonal index (amplitude of  4.8%) used in the Brazilian electric planning that has a component associated with the demand growth.

[viii] In the USA the hydroelectric generation corresponds to only 8.5% of generation and, according to EIA data from 1990 to 2001, it is used according to the available offer  (maximum in May) even when separated from  the demand curve  (maximum in August).

[ix] José Paulo Vieira et al. Sistema de Caracterização da Carga e Dimensionamento da Ponta do Sistema Elétrico do Estado de São Paulo. XVI Seminário Nacional de Produção e Transmissão de Energia Elétrica.  Campinas - São Paulo 21 a 26 de Outubro de 2001

[x] According to ONS, in December 2003 the total storage capacity was 76.6 GW month. The installed capacity in the Southeast region was 38,9 GW (ABINEE) to which, for the purpose of the present study, it must be added the Paraguayan part of Itaipu, reaching  45.2 GW, coherently with the considered storage that is the total volume. The affluent natural energy / month (average) is 27.4 GW and the monthly electricity generation in the SE/CO system is16.3 GW plus 9.5 of Itaipu (ONS data), totaling a production of  25.8 GW of electricity/month.

[xi] The minimum level reached at the end of 2000 was higher than that of the end of 1999 (when specialists warned about the risk of energy shortage). At the time, it was chosen not to take any measure regarding consumption restriction and not to warn the consumer. The 2000 (election year) rainfall permitted the year to pass calmly, in spite of the preoccupation of the system’s administrators.

si[xii] In a system regulated by the market, where there is anticipated energy sale,, this logic will only be followed as long the regulation and the buy and sale mechanisms are sufficiently agile to draw near the optimum production  configuration.

[xiii] In the present study the term affluence is used  for referring to the affluent natural energy (ANE)

[xiv] If the limit was fixed as 100% of the capacity, at the end of the rainy season of a year with normal affluence in many years (about half) there would be excess water to be spilled.

[xv]  Projetar_e, a  program originally developed by Carlos Feu Alvim and  Eduardo Marques for the “Brasil 2020” studies of the Secretariat of Strategic Studies of the Presidency  and later  used (in association with Macroplan) in the electric planning under the responsibility of Eletrobrás, in energy scenarios for CETEPETRO and  (by e&e) in the energy matrix and emissions for  MCT.

[xvi] The real average growth was 2,5% per year and the reasons pointed out  as limiting growth were realistic.

[xvii] The energy demand also grows due to other factors such as urbanization, but this can be included in the qualitative aspects of growth.

[xviii] The equivalent energy concept is based on the equivalence, for each sector  and for the different uses, of each form of energy relative to a “reference” one. A similar approach was developed for the energy matrix prepared by the National Energy Commission and in the Ministry of Mines and Energy in the Sarney and Collor administrations.

[xix] More specifically, it would be of interest the part of these plants in the interconnected system; since the trend is already to increase the interconnected systems that already are responsible for most of the satisfied demand, we have considered the demand satisfied by the public service plants as a proxy of the integrated system.

[xx] January, 2002 to July, 2004

[xxi] The units are relative to affluence. For example, if the average monthly affluence were 100 Gw, the storage capacity would be 300 GW month and the installed thermal capacity, 66 GW.

[xxii] The capital cost, both for hydro and thermoelectric plants, exists independently of its use; in the case of hydroelectric plants the cost is predominantly capital, in the case of thermal plants what predominates is the fuel cost.

[xxiii] The ratio also depends on the downstream productive values.

[xxiv] It is natural for a plant that does not have downstream installed capacity to have a smaller NAE. Even then, the value is low.

[xxv] The use of the parameter from the previous year for determining the values fitted to the historical value behavior prevent circular calculations.

[xxvi] The costs evolution foreseen for the new hydroelectric plant should be re-evaluated when the presence of dams of smaller or no regulation, since the machine cost would be increased and the dam cost, decreased. For this reason the economic limit was not considered in the reference scenario adopted here.

[xxvii] What corresponds to ignoring the rising cost limits of the hydroelectric plants

[xxviii] This limit is based on extrapolation of the cost growth of hydroelectric energy and considering that costs above  65 US$/ MWh are not economic.

Final Notes

[1] Seasonal character in the North Region with new plants

Flows in Belo Monte (1931/2001 Period)

Figure NF1.1: Maximum and average values das of flows in Belo Monte. The value of the minimum flow/maximum flow ratio (curve of average values) is 5%. The “minimum of the minimum values” is 2% of the maximum value of the average. It should be noted that the curve of the minimum (or maximum) values does not represent one year of minimum (or maximum) flow but the extreme values observed for each month.

Figure NF1-2:  Flow of the Madeira River in Porto Velho. The ratio between the minimum and maximum (average) flows taken from the graphic is about 13%. It should be noted that the months of maximum and minimum values are not much different from the values used for the North region, based on data of the Tocantins River.

[2] Complementation of dry years and abundant rainfall between NE and South:.
The graphic of the movable average value of 12 months of affluent natural energy for the South and NE Regions (Figure NF2.1) shows an apparent complementation between dry years and years of larger precipitation in the NE and South Regions. The period shown is not sufficient for definitive conclusions.

Figure NF2.1: The affluent natural energy data show a (negative) correlation between NE and South rainfall regime.


Graphic Edition/Edição Gráfica:
Editoração Eletrônic

Tuesday, 11 November 2008

Contador de visitas