Vous êtes sur la page 1sur 15

Short-term electricity price in Brazil: current limitations and the benefits of a

transition to a market price formation model


Andr Luiz Zanette*
Overview
Short-term electricity price is an important economic indicator for the use of resources
for electricity production and electricity use. Short-term price also represents a reference
for medium and long-term price, which are the main indicators for investment in new
capacity. The existence of an efficient mechanism for electricity price formation can
result in important incentives for agents responses that result in more efficient
decisions on investment, operation and electricity use [1]. The primary function of an
organized spot market for electricity is to maximize cost efficiency by supplying the
demand for power from the most economic source available. It is difficult to achieve
such optimization without a continuous price setting mechanism producing a
transparent equilibrium price [2].
In the Brazilian electricity market, the price that better represents a spot price is the
Settlement Price for the Differences (PLD), which is the price at which the differences
between the actual amount of energy produced or consumed and contracts are settled at
the Electric Energy Trading Chamber CCEE. This price, however, is not defined by
the market, but calculated through centralized models, which are the same models used
by the Brazilian Electric System Operator (ONS) for operation planning and system
dispatch. As every model, these also represent an approximation of the reality. Although
these models simulate the system operation considering the expectations of electricity
supply and demand, the prices they calculate are often substantially different from the
marginal cost of electricity production, thus resulting in an inefficient signal to
producers and customers.
In this context, the objective of this work is to analyze the main limitations of the
current model for short-term electricity pricing in Brazil, propose a transition to a
market price formation model and evaluate the main benefits of this transition.

Methods
The first part of the work analyzes the main limitations of the artificially calculated
short-term electricity price in Brazil, that are related to the quality of the data used by
*

Energy Planning Program COPPE/UFRJ

the models and the differences between the operation planned by the models and the
real operation, and its impacts on the operation and expansion of the electricity system
and on electricity trading. The differences between the prices and the real operational
costs are also evaluated.
The second part proposes a transition to a market price formation model, considering
the specific characteristics of the Brazilian electricity system. To compare the price
formation in Brazil with a market price formation model, the work also presents some
comparisons with the electricity market in the Nordic countries. Finally, the expected
results and potential benefits of this transition in terms of efficient operation and
investments are evaluated.

Results
Limitations of the short-term electricity price formation model in Brazil
Short-term electricity price in Brazil is calculated by the same modes used by the
Brazilian Electric System Operator (ONS) for operation planning and system dispatch.
The basic objective of the operation planning for a hydrothermal system is to determine,
for each stage of the planning period, the dispatch for each plant that meets the demand
at the lowest cost during the entire period. This cost comprises the variable cost of
thermal plants and the cost of supply interruptions [3]. According to hydrological
conditions, electricity demand, fuel prices, deficit costs, the existing generation and
transmission capacity and the schedule of new generation and transmission projects, the
models calculate the optimal dispatch. The process for calculating the Settlement Price
for the Differences (PLD) comprises the use of the models Newave and Decomp that
calculate the marginal cost of operation. PLD is calculated weekly for each demand
level and region based on the marginal cost of operation, limited by a minimum and a
maximum price1. The models consider three demand levels and four regions, according
to transmission constraints among them (Southeast/Middle-West, South, Northeast and
North). In 2012, the minimum and maximum prices are respectively USD 6 and
360/MWh. The main difference between the model simulations used by ONS is that for
PLD calculation the transmission constraints internal to the regions are not considered,
1

The medium-term operation model Newave represents the hydropower plants in an aggregated form and
the operation policy is based in stochastic dual dynamic programming. The short-term model Decomp
represents the hydropower plants individually and uses the future cost functions calculated by Newave
model to determine the operation policy and the marginal costs of operation.

so the electricity traded is assumed as equally available in the entire region and,
consequently, there is only one price inside each region.
Although the models try to represent the actual conditions of electricity supply and
demand to calculate electricity prices, the data used by them have several limitations.
As discussed below, the demand and supply projections considered in the models may
be significantly different from the reality. Moreover, the system operator tends to use
several mechanisms to increase the security of electricity supply that are not considered
in the models, which may result in an operation that may be considerably different than
the indicated by the models.
As it represents the official expectations, the demand projected is frequently higher than
the observed. Figure 1 compares the demand projections used in the medium-term
model and the actual demand. As can be noted, the demand projected for 2009 was
considerably higher than the actual, as a result of the international economic crisis.
After a few revisions, the difference was reduced, but it increased again in 2011. These
demand projections can indicate the need of a higher thermal generation, which may not
be necessary. On the other hand, a demand higher than the projected during the summer
months, as observed in recent years, can result in an additional thermal generation while
the prices are low, thus resulting in an unsuitable signal to the market. These differences
between projected and realized demand are also significant in the short-term model. As
Figure 2 shows, the demand during peak hours in summer days can be substantially
higher than the considered in the model, thus requiring the dispatch of expensive
thermal plants, which is not represented in the electricity prices.
Figure 1 Differences between the demand projections considered in the medium-term
model and the actual demand.

Source: Based on data from ONS-EPE [4].

Figure 2 Demand considered in the short-term models and the actual demand in a
typical summer day.

Source: Based on data from ONS [5].

Demand is also assumed to be price-inelastic, despite an approximation of a demand


curve is considered in the models in the form of four deficit costs. As shown in Figure
3, the medium-term model considers that at an electricity production cost of about USD
600/MWh a 5% supply deficit is acceptable, at about USD 1200/MWh a 10% deficit,
and so on. Although this mechanism is similar to a demand curve, it is not possible to
ensure that it effectively represents the real propensity of the customers to reduce their
demand, once this deficit costs are not defined by the market, but by the official
institutions. Moreover, the deficit costs are only used in the medium-term model and are
not considered in the short-term model, which assumes an inelastic demand.
As a comparison, Figure 4 shows the aggregated demand curve built considering all the
customers bids for a one hour period at the electricity market in Nordic countries (Nord
Pool Spot) for the following day. As can be noted, even with the orders posted with
only one day of anticipation, customers accept to significantly reduce their demand at
prices much lower than the deficit costs considered in the medium-term model in Brazil.

Figure 3 Deficit costs used in the medium-term model for electricity price formation
in Brazil.

Source: Based on data from EPE [6].


Figure 4 Demand curve for a period of one hour in Nord Pool Spot.

Source: Based on data from Nord Pool Spot [7].

Moreover, the official supply projections used by the models are frequently
overestimated. This is particularly true when considering new capacity additions to the
system, which is especially relevant in the Brazilian system, where demand is expect to
grow at a 4.5% rate in the next decade [8]. The frequent revisions of capacity expansion
projections by the Brazilian Electricity Regulatory Agency (ANEEL) to represent a
more realistic supply in the models have led to sharp increases in prices and in thermal
generation. As an example, Figure 5 shows the differences between the power
generation from alternative renewable sources (small hydro, biomass and wind power)
considered in the models and the actual generation, which resulted in a revision of the
availability of these power plants in 2011 that represented a 3,000 MWa reduction in
power generation. These reviews in supply projections are also frequent with respect to
thermal power generation. In 2010, ANEEL reviewed the schedule of new projects to

consider a 2,000 MW delay in new thermal power capacity. A new revision occurred in
2012 to consider a 2,500 MW reduction in the future capacity of fuel oil fired thermal
plants.
Although these revisions contribute to improve the quality of the data used by the
models to operation planning and price formation, they are frequently abrupt and may
result in a significant increase in price and thermal generation volatility. In a market
price formation model, these revisions could occur gradually, representing the market
expectations related to electricity supply, in the same manner as it would occur with
demand projections if they represented market expectations.
Figure 5 Differences between the actual alternative renewable power generation and
the considered in the models.

Source: ANEEL [9].

Given the importance of hydropower in Brazil, which represents about 90% of total
electricity production, water inflow forecasting is particularly important for system
operation planning and electricity price formation. Despite the continuous improvement
of the official models used to forecast hydropower water inflow, the differences
between the inflows projected and the actual inflows are frequently significant. As a
comparison, in a market model the offers from hydropower producers may reflect better
their expectations of electricity supply.
The models also do not consider the need of spinning reserve for the system, i.e., an
additional capacity that is kept available to generate when necessary, of about 5% of the

total demand. This means that the models consider that the capacity available for power
generation is higher than the actual available capacity. With respect to this limitation,
ANEEL is planning to discuss the representation of the spinning reserve in the models
in the regulatory agenda for 2013-2014 [10].
Another limitation of using these models to calculate prices is that the real operation
may be significantly different from the operation calculated by the models. As the
system operator tends to adopt a more conservative operation, given its aversion to risk,
it may determine additional thermal generation to increase energy security, which is not
considered in price formation.
Besides the thermal generation to meet peak demand, which is not included in the
models used to calculate the PLD, the system operator also applies the so-called shortterm operational procedures, which define a target for a minimum level of water storage
in the reservoirs for the month of November (the end of the dry season), to guarantee
the supply in the next year in the case of a severe dry [11]. To achieve this target, the
operator can call thermal plants for an additional power generation during the dry
season, which is not represented in the models that calculate short-term prices. It means
that the prices can be maintained lower than the real operational costs during long
periods, without an efficient signaling to the market, given that the costs of this
additional generation are divided among all customers and, in the case of the regulated
customers, this will represent an increase in the tariff only for the following year.
Figure 6 presents the differences between the thermal generation calculated by the
models and the total thermal generation including the additional thermal generation and
the differences between the short-term prices and the real operational costs. As can be
noted, these differences were particularly significant during the dry seasons in 2010 and
2012 (2011 was a year with water inflows much higher than historical averages, thus
not requiring almost any thermal generation) and tend to continue if the criteria
considered for system operation are kept different from the used by the models for price
formation.

Figure 6 - Additional generation and differences between short-term prices and average
costs for additional thermal generation.

Average cos t of additional generation (USD/MWh)

PLD (USD/MWh)

jul/12

Additional generation (MWa)

0
ago/12

jun/12

abr/12

mai/12

mar/12

jan/12

fev/12

dez/11

nov/11

set/11

out/11

jul/11

ago/11

500
jun/11

25
abr/11

1000

mai/11

50

mar/11

1500

jan/11

75

fev/11

2000

dez/10

100

nov/10

2500

set/10

125

out/10

3000

jul/10

150

ago/10

3500

jun/10

175

mai/10

USD/MWh

Additional generation for energy security

Additional generation (MWa)

Source: Based on data from CCEE [12, 13, 14].

Once the same models used to operation planning are also used to project the system
expansion, the economic signaling to an efficient expansion has the same limitations as
those represented by the artificial prices to system operation. Given that the models
used for expansion planning and for the auctions for new power generation projects also
do not consider the risk-aversion curves2 used in the medium-term model and the shortterm operational procedures discussed above, the benefits of flexible thermal generation
are underestimated in these auctions, thus affecting the competitivity of these projects
[15]. This is also true for wind and biomass projects that are complementary to
hydropower generation, whose benefits of producing more energy during the dry season
are not well represented.

Proposal for a market price formation model and its potential benefits
Considering the limitations of these centrally calculated prices, this work presents a
proposal for a transition to a market price formation model for electricity in Brazil and
evaluates the potential benefits of this model. It is proposed that the transition occurs in
two stages. In the first stage, the system operator continues to determine the generation
of hydro and thermal plants, including the additional dispatch for energy security, but
2

The risk-aversion curves are used in the medium-term model to represent an additional cost in the case
of depleting the hydro reservoirs to under certain monthly levels.

the thermal generation represents only a reference to the market. After that, producers
and customers post their orders to buy and sell electricity for the different periods of the
next day. The bids and offers will form the aggregated supply and demand curves. The
market price and the electricity production and use for each agent result from the
balance of supply and demand. At this stage, thermal generation is settled by the market
and customers may buy and sell electricity according to the prices, which is not allowed
under the current model, although the possibility of customers to sell the excess
amounts of electricity they purchased is being studied [16]. Also, there was an
experience of allowing customers to sell electricity during the supply shortage in 2001,
when the energy certificates were created [17].
At this stage, thermal generators will be able to buy electricity from customers or other
generators to meet their contracts when prices are lower than their production costs or
produce additional amounts of electricity when the prices are higher. Figure 7-a shows
an example of a bid from a thermal generator with a production cost of USD 50/MWh.
In this example, the generator makes a bid for prices lower than this value and offers an
additional amount of electricity when the prices are higher. Electricity customers will be
able to buy additional amounts of electricity or sell part of the energy they purchased
according to market prices, thus becoming active participants in the process of price
formation, as shown in Figure 7-b.
Figure 7 Examples of bids and offers from thermal generator (a) and customer (b).
(a)

(b)

Source: Own work.

Another difference is that the secondary energy, i.e., the additional amount of
hydropower generation that is distributed among hydropower producers is traded at
market prices. Many benefits of market price formation can already be verified at this

stage. An important signaling to energy efficiency arises when thermal generation is


defined by the market and consumers can increase or reduce the amount of energy they
use according to market prices.
In the second stage of the proposed model, the system operator becomes an independent
operator that coordinates the generation and transmission system, and hydropower is
also included in market price formation, completing the transition to a market price
formation model. From this stage on, the bids and offers from all producers and
customers will determine the system dispatch and market prices.
The main characteristics of the market model and the transition period proposed, as well
as the current model for price formation, are presented in Table 1. At this stage the
supervision of market regulators to avoid the exercise of market power is crucial. There
are several benefits of this transition to a market price model, as discussed below.
Table 1 Main characteristics of the current model for price formation and the proposed
transition period and market model.
Current model

Transition Period

Market Model

ONS defines hydro


thermal generation

and ONS defines hydro generation ONS coordinates operation


and the reference thermal and evaluates the security of
generation
supply

Prices are calculated


computational models

by Price resulting from producers Price resulting from producers


and customers bids and offers and customers bids and offers

Additional
dispatch
for Additional
dispatch
are Hydro and thermal generation
security of supply (peak included in price formation
defined according to market
demand
and
short-term
prices
operational procedures) are
not
included
in
price
formation
Customers not allowed to sell Customers allowed to sell Customers allowed to sell
electricity
electricity
electricity
Thermal plants only allowed Thermal plants can buy
to substitute other plants electricity from customers to
dispatch
meet the demand and their
contracts

Thermal plants can buy


electricity from customers and
hydro producers to meet the
demand and their contracts

Hydro producers are paid Hydro


producers
sell Hydro producers free to
PLD for additional generation additional
generation
at negotiate electricity among
market prices
them and with other agents

Inefficient
signaling
for
energy efficiency and system
expansion, especially for
flexible thermal generation,
wind and biomass

More efficient economical


signaling to energy efficiency
and
system
expansion,
especially for flexible thermal
generation, wind and biomass

More efficient economical


signaling to energy efficiency,
hydropower generation and
system expansion, especially
for
flexible
thermal
generation, wind and biomass

Source: Own work.

This model of market price formation can contribute to a more efficient use of energy
resources. A short-term price that represents the real operational costs allows the market
to decide the better combination of flexible generation and energy efficiency measures
to meet the peak demand and to compensate the lower hydropower availability during
the dry season. As discussed before, in the current model the system operator defines
the thermal generation to meet peak demand and to increase the security of supply and
transfer these costs to all customers, thus not allowing producers and customers to
choose a more efficient alternative.
The transition to a market price formation would allow a more effective customer
response, especially with the introduction of smart metering and distributed generation
systems, both recently regulated by ANEEL [18, 19]. Although only the customers with
demand higher than 0,5 MW are allowed to choose their electricity supplier and thus
would be able to sell the amount of electricity that exceeds their demand, due to a lower
level of activity or energy efficiency measures, distribution companies or retailers
would be able to develop and offer customized products to customers that accept to
reduce their demand during peak hours or in the dry season, for example. The
introduction of the electricity retailer trader, which will represent smaller customers in
the CCEE [20] and the white tariff for electricity distribution companies, which results
in higher prices during peak hours and smaller prices during hours with low demand for
regulated customers [21], are two good examples of measures that can have their
benefits enhanced with the transition to a market price formation model.
In the long-term, the maintenance of long periods of higher prices during peak time
would represent an important signal to evaluate the need and viability of investments to
increase flexible thermal generation, the power generation capacity in existing
hydropower, energy efficiency projects or even wind and power or solar photovoltaic,
whose peak in electricity production is close to the peak of demand. Figure 8-a shows
an example of an hourly price signal in the Nordic countries market.

In the Brazilian system, with increasing hydro production in plants without reservoirs
that generate most of their energy during the wet season, this may also represent an
important incentive to investment in flexible thermal plants and wind and biomass
plants that are complementary to hydroelectric plants and produce more electricity
during the dry season, as shown in Figure 9. Also, the existence of an efficient market
price and the possibility of customers to sell electricity would represent an incentive to
energy efficiency projects, probably as effective as energy efficiency auctions. Fig 8-b
illustrates the seasonal price signal in the Nordic market, where the higher demand and
lower availability of hydropower results in higher prices in the winter months.
Figure 8 Electricity prices in the Nord Pool Spot for one day (a) and in the forward
and futures Nordic electricity market (b).
(a)

(b)

Source: Based on data from Nord Pool Spot [7] and Nasdaq OMX Commodities [22].
Figure 9 Seasonal complementarity between hydropower and biomass and wind
power.

Source: ONS [23].

The existence of a short-term market price would also contribute to the development of
financial markets for electricity, with products like futures, forwards and options, thus
also stimulating long-term contracts in the deregulated market, which would incentive
the investment in new power generation projects especially to this market.
An efficient price signal, which represents the differences between the marginal costs in
each region, is also important to indicate the real benefits of investments to increase the
transmission capacity between the regions and in the increase of generation capacity or
energy efficiency in each region. Moreover, this price signal is important to evaluate the
viability of the projects for energy integration that are being developed in South
America, which include new transmission capacity between the countries and binational
hydropower.
Finally, the participation of all players in market price formation may result in a
significant evolution from the centrally calculated prices. The bids and offers of all
agents represent the expectations of the whole market, which contains a much larger
amount of information than the currently used to calculate prices and can thus lead to
more efficient and representative prices. The bids and offers also reflect the risk profile
of the agents, which is probably a better representation than the risk criteria centrally
defined. Although the models are being constantly improved, it is difficult to imagine
that the centralized models will be able to keep pace with developments of the industry,
especially with the introduction of smart grids, intermittent renewables and distributed
generation.

Conclusions
The main conclusions of this work are that the short-term electricity price in Brazil has
several limitations that may lead to an inefficient operation and investment. These
limitations arise both from the quality of the information used by the computational
models to calculate the prices as from the differences between the planned operation
and real-time operation, and can result in substantial differences between the short-term
prices and the real operational costs. These differences are becoming especially relevant
when considering the increasingly role of more expensive thermal generation in Brazil
and the lower participation of hydropower in new capacity addition. Also, it is
important to note that the economic signal to investments in projects that aim to meet
peak demand and compensate the lower availability of hydropower during the dry

season is inefficient, given that the models used for expansion planning are the same
used for operation planning and thus have the same limitations of these.
Considering the importance of an adequate price signaling to the efficient system
operation and expansion, the work proposes a transition to a market price formation
model for electricity in Brazil.
The transition to a market price model is proposed in this work to occur in two stages.
In the first stage, the Brazilian Electric System Operator keeps its current function of
determining the hydrothermal dispatch, but short-term price is defined by the balance
between market supply and demand. The main change in this first stage is the effective
participation of customers in the price formation process, which would result in a more
efficient signaling to the entire market. Moreover, the dispatch to meet peak demand
and for security reasons due to the lower availability of hydro power during the dry
season are included in the price formation mechanism.
In the second stage, the transition to a market model would be completed with the
effective inclusion of hydropower producers in the price formation and system dispatch.
The system operator would cease to determine the dispatch and the price formation
would involve the effective participation of all producers and customers. In this stage,
the role of regulators is of special importance.
The main benefit of this transition to a market price formation model is the increase in
the efficiency of the electricity industry, which would benefit both electricity customers
and producers. In the market model, shot-term prices would provide an economic
signaling to a more efficient electricity production and use. In addition to the benefits to
system operation efficiency, the short-term market prices, associated with an electricity
exchange with futures, forwards and options traded in a transparent way, would also
contribute to more efficient decisions of investment in capacity expansion, resulting in
an important incentive to energy efficiency and the investment in a more efficient
system expansion, including flexible thermal generation and wind, biomass and solar
power.

References
[1] IEA, 2011. Empowering Customer Choice in Electricity Markets. OECD/IEA.
[2] Nord Pool Spot, 2012. Price formation in Nord Pool Spot. Disponvel em:
<www.nordpoolspot.com>
[3] CCEE, 2010. Viso Geral das Operaes na CCEE. Disponvel em:
<www.ccee.org.br>

[4] ONS-EPE, 2012. Previso da Carga dos Sistemas Interligados. Disponvel em:
<www.ons.org.br>
[5] ONS, 2010. Avaliao das Condies de Atendimento demanda horria do
SIN. NT 160/2010.
[6] EPE, 2011. Atualizao do valor para patamar nico de custo de dficit - 2011.
N EPE-DEE-RE-021 /2011.
[7] Nord Pool Spot, 2012. Elspot Prices. Disponvel em: <www.nordpoolspot.com>
[8] EPE, 2012. Plano Decenal de Energia 2012-2021. Empresa de Pesquisa
Energtica.
[9] ANEEL, 2011. Nota Tcnica n 023/2011 - SRG/ANEEL.
[10]
ANEEL,
2012.
Nota
Tcnica
Conjunta
n
001/2012
SPG/SCG/SRE/SEM/SRG/SRT/SRD/SRC/SFF/SMA/CEL/ANEEL.
[11] ONS, 2008. Procedimentos operativos de curto prazo para aumento da
segurana energtica do sistema interligado nacional. NT 059/2008.
[12] CCEE, 2011. Relatrio de Informaes ao Pblico Anlise Anual 2010.
Disponvel em: <www.ccee.org.br>
[13] CCEE, 2012. Relatrio de Informaes ao Pblico Anlise Anual 2011.
Disponvel em: <www.ccee.org.br>
[14] CCEE, 2012. InfoMercado CCEE n 62 Outubro/2012. Disponvel em:
<www.ccee.org.br>
[15] FERREIRA, L. S., ZANETTE, A. L., 2012. Impacto da incorporao dos
critrios de segurana utilizados no planejamento da operao sobre a
competitividade das usinas termeltricas nos leiles de energia nova. Anais do XII
SEPOPE 2012.
[16] MME, 2010. Portaria n. 73, de 1 de maro de 2010. Ministrio de Minas e
Energia.
[17] GCE, 2001. Resoluo n. 13, de 1 de junho de 2001. Cmara de Gesto da
Crise de Energia Eltrica.
[18] ANEEL, 2012. Resoluo Normativa n. 502, de 7 de agosto de 2012. Agncia
Nacional de Energia Eltrica.
[19] ANEEL, 2012. Resoluo Normativa n. 482, de 17 de abril de 2012. Agncia
Nacional de Energia Eltrica.
[20] ANEEL, 2012. Nota Tcnica n 55/2012-SEM/ANEEL.
[21] ANEEL, 2010. Estrutura Tarifria para o Servio de Distribuio de Energia
Eltrica Sinal Econmico para a Baixa Tenso. Nota Tcnica n 362/2010SRESRD/ANEEL.
[22] NASDAQ OMX COMMODITIES, 2012. Market Prices. Disponvel em:
<www.nasdaqomxcommodities.com>
[23] ONS, 2011. Plano anual da operao energtica PEN 2011. Vol. I Relatrio
Executivo. RE 3/116/2011.