Hydrology Days 2004

Applied Stochastic Hydrology Lessons Learned from the Brazilian

Electric Energy Crisis of 2001

Jerson KelmanPresident of ANA

(Brazilian Water Resources Agency)

Area: 8,574,761 km2 Average Temperature: over 20ºCFederative Republic: - 26 States + 01 Federal District - 5,561 Municipalities13 River basins

BRAZIL

Hydroelectric power accounts for more than 90% of the total

electric energy produced in Brazil

BRAZILIAN ELECTRIC SYSTEM

MAIN CONSTRUCTIVE SYSTEMS USED

Compacted rock fill with concrete faceCompacted rock fill with impervious coreEarth fillConventional concreteRolled compacted concrete (RCC)

SEGREDO HYDRO PLANTA compacted rock fill structure with an

upstream concrete face

SALTO SANTIAGO HYDRO PLANTRock fill dam with impervious clay core

FURNAS HYDRO PLANTA construction of zoned earth and rock fill

dam

SOBRADINHO HYDRO PLANTThe typical section is a zoned embankment type,

comprising a clay central impervious core

SALTO CAXIAS HYDRO PLANTRolled Compacted Concrete (RCC)

ITAIPU HYDRO PLANTA buttress concrete structure

Vast territorial extension and hydrological variability

Country Wide Integrated Electric System

BRAZILIAN ELECTRIC SYSTEM

INTEGRATED ELECTRIC

SYSTEM Installed Capacity = 72,299 MW

96 Hydropower plants > 30 MW

57 Regulating reservoirs

Country is interconnected by 44,000

miles of high-voltage lines

INTEGRATED ELECTRIC

SYSTEM Up to 1996, new hydroelectric power

plants were built almost exclusively

by the Federal and State

Governments

Expansion planning was based on

reliability criteria: probability of any

energy shortage along a year would

be 5%

INTEGRATED ELECTRIC

SYSTEM All power plants, hydro and thermal,

were centrally dispatched, taking

advantage of the hydrological

complementarities among river

basins

Stochastic Dynamic Stochastic

Programming was used to decide

how to split energy production

between hydro and thermal sources

When the use a mathematical model to dispatch power plants is necessary?

If the system is thermal, it isn’t. Example: Suppose demand = 20 and three generators

The dispatch would be G1=10; G2=5; G3 = 5

Marginal cost = spot price = 15

Generator Cost

($/energy unit)

Capacity

G1 8 10

G2 12 5

G3 15 20

When the use a mathematical model to dispatch power plants is necessary?

If the system is hydro, it is.

wet

dry

OK

Deficit dry

wet

Future inflows

Use stored water

Decision

Do not use stored water

OK

Consequências operativas

Spill

When the use a mathematical model to dispatch power plants is necessary?

If the system is hydro, it is. Z = Max E [ immediate cost + future cost ]

Marginal cost = spot price = Z/ demand

wet

dry

OK

Deficit dry

wet

Future inflows

Use stored water

Decision

Do not use stored water

OK

Consequências operativas

spilling

THE POWER SECTOR REFORM

The reform process started in 1996

Rationale:

– Public sector has no $ to invest

– To promote economical efficiency

Guidelines:

– Private investment and competition in energy generation and retailing

– Transmission and distribution to remain regulated, with provisions for open access

Reforms based on the same principles had worked in countries based on thermal production.

Would it work in a country like Brazil, based on hydro?

THE POWER SECTOR REFORM

Simulação do comportamento do reservatório de FURNAS(Vazão firme = 670 m³/s)

0

5000

10000

15000

20000

25000

Mês/ano

Volu

me

(hm

³)

Armazenamento final (hm3) Volume Morto (hm3)

Nov/56 = 5733 hm³

Volume morto = 5733 hm³

Fonte dos Dados: ONS

Reservoirs are full most of the time

Storage variability

SOUTH-SE SystemEnergy Wholesale Market Prices (US$ / MWh)

Sampling probability distribution of spot price (R$/MWh)

0.0

100.0

200.0

300.0

400.0

500.0

600.0

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64

spot mean

EVOLUTION OF THE STORED ENERGY IN THE EQUIVALENT RESERVOIR OF

THE NORTHEAST REGION

ENERGIA ARMAZENADA (% DO VALOR MÁXIMO)

0

10

20

30

40

50

60

70

80

90

100

jan

/97

jul/9

7

jan

/98

jul/9

8

jan

/99

jul/9

9

jan

/00

jul/0

0

jan

/01

jul/0

1

jan

/02

THE ENERGY CRISIS OF 2001

There was a market failure

As a consequence…

•20% of the energy demand had to be curtailed in 2001

•Population reacted better than expected:

consumption reduction remains until now

THE ENERGY CRISIS OF 2001

How to prevent market failures?

• Brazilian Government restored centralized expansion planning for new plants and transmission lines

• Reliability criteria is being modified. For example:

P (Curtailment < 0.05 Demand) < 0.1

P (Curtailment > 0.20 Demand) < 0.001

P (Curtailment | occurrence of worst drought) = 0

THE ENERGY CRISIS OF 2001

How to prevent market failures?

• Owner of a new hydroelectric plant will get, before construction, a long term contract with a pool of distribution companies (equivalent to an annual “rent”)

• All power plants will operate according to rules set by central dispatch, based on a multiple reservoirs Stochastic Dual Dynamic Programming Model (SDDP)

Should hydro be abandoned?

Cost of new energy (US$/MWh)

Hydro 30

Thermal 40

Alternative 50

Amazon Region

CHEBelo Monte

Rio

Xin

gu

Localization

Total Capacity (MW)

Energy (MW méd)

Xingu River(PA)

11.182

4.796

Reservoir Area (103 acres) 110

Belo Monte

Conclusions

The energy crisis of 2001 has shown that the electric energy business based on hydro requires more than regulation it requires Government planning

When the crisis occurs, planning criteria based on the probability concepts are difficult to understand. The concept of trade off between reliability and cost was ill perceived by the population the old concept of firm energy is better accepted

Conclusions

Brazil needs to ensure a sustainable growth of energy supply

Hydroelectricity cannot be spared

The strategy is to select, from the numerous sites technically and economically feasible for hydroelectric power plants, a subset that would cause minimum environmental and social impacts

However, we are not seeking for a subset that would cause no impact