Hydrothermal scheduling A case study - Comillas · 2013. 1. 27. · Hydrothermal scheduling. A case study b) Discharge decision taken from guiding curves (usually for small reservoirs)

Massachusetts Institute of Technology (MIT). January 2013

ESD.S30Electric Power System Modeling for a Low Carbon Economy

Hydrothermal scheduling. A case study

Prof. Andres Ramos

http://www.iit.upcomillas.es/aramos/

[email protected]@mit.edu

1. Introduction

2. Long-term stochastic hydrothermal scheduling model

3. Medium-term simulation model

Index

2Instituto de Investigación Tecnológica

Escuela Técnica Superior de Ingeniería ICAIHydrothermal scheduling. A case study

3. Medium-term simulation model

1

Introduction

Long-term stochastic hydrothermal scheduling model

Medium-term simulation model

Introduction

Introduction

• Relevance of hydroelectric power

– Reduces electric system operation cost

– High flexibility to integrate intermittent generation

– Important role in the generation mix

• Objectives of hydro scheduling

– To optimize the water value in an stochastic environment,



– To optimize the water value in an stochastic environment, fulfilling all the operation constraints (to minimize the operation cost)

– To analyze and test different scheduling strategies of storage hydro and pumped storage hydro plants

Procedure in two steps

1. Stochastic Hydrothermal Coordination Model (MHE) to deal with a large-scale electric system and with the dimensionality of stochastic hydro inflows

– Based on Stochastic Dual Dynamic Programming (SDDP)

– Scope: 2 years. Time step: week. Load levels: peak and off-peak

– Some hydro details must be simplified in order to reduce the size of the problem

2. Simulation-based model to deal with hydro cascades in a



2. Simulation-based model to deal with hydro cascades in a more detailed way

– Receives the main outputs from the optimization model and performs daily simulations

Simulation

toolMHE

production /

pumping tables inflows, outages,

detailed criteria

2

Introduction




MHE model creates DISCHARGE, PUMPING and PRICE TABLES

•Hydrothermal model (Spanish or Iberian Electricity Market)

•Detailed modeling of the hydro system

•Less detailed modeling of demand and the rest of the generation mix

•Representation of inflows through weighted scenario trees

Azután Data for generating the

Stochastic Hydrothermal Coordination Model (MHE)



Azután

0

200

400

600

800

1000

1200

01-e

ne

10-e

ne

20-e

ne

31-e

ne

10-fe

b

20-fe

b

28-fe

b

10-m

ar

20-m

ar

31-m

ar

10-a

br

20-a

br

30-a

br

10-m

ay

20-m

ay

31-m

ay

10-ju

n

20-ju

n

30-ju

n

10-ju

l

20-ju

l

31-ju

l

10-a

go

20-a

go

31-a

go

10-s

ep

20-s

ep

30-s

ep

10-o

ct

20-o

ct

31-o

ct

10-n

ov

20-n

ov

30-n

ov

10-d

ic

20-d

ic

31-d

ic

Apo

rtac

ione

s [m

3/s]

Torrejón-Tiétar

0

100

200

300

400

500

600

700

800

900

01-e

ne

10-e

ne

20-e

ne

31-e

ne

10-fe

b

20-fe

b

28-fe

b

10-m

ar

20-m

ar

31-m

ar

10-a

br

20-a

br

30-a

br

10-m

ay

20-m

ay

31-m

ay

10-ju

n

20-ju

n

30-ju

n

10-ju

l

20-ju

l

31-ju

l

10-a

go

20-a

go

31-a

go

10-s

ep

20-s

ep

30-s

ep

10-o

ct

20-o

ct

31-o

ct

10-n

ov

20-n

ov

30-n

ov

10-d

ic

20-d

ic

31-d

ic

Apo

rtac

ione

s [m

3/s]

Gabriel y Galán

0

100

200

300

400

500

600

01-e

ne

10-e

ne

20-e

ne

31-e

ne

10-fe

b

20-fe

b

28-fe

b

10-m

ar

20-m

ar

31-m

ar

10-a

br

20-a

br

30-a

br

10-m

ay

20-m

ay

31-m

ay

10-ju

n

20-ju

n

30-ju

n

10-ju

l

20-ju

l

31-ju

l

10-a

go

20-a

go

31-a

go

10-s

ep

20-s

ep

30-s

ep

10-o

ct

20-o

ct

31-o

ct

10-n

ov

20-n

ov

30-n

ov

10-d

ic

20-d

ic

31-d

ic

Apo

rtac

ione

s [m

3/s]

... ...

Data for generating the

scenario tree:

•Starting date

•Current inflows

•Tree structure

Two-year scope case study

• Spanish electric system

– 118 thermal units

– 3 main basins with 57 hydro plants and 7 pumped storage hydro plants that operate 39 hydro reservoirs

• Inflows uncertainty

– Recombining tree

with 5 branchesSIL



with 5 branches

every 4 weeks:

525 scenariosDUERO

TAGUS

Historical natural hydro inflows in an small reservoir



Historical natural hydro inflows in a medium reservoir



Historical natural hydro inflows in a large reservoir



Optimization process of MHE

Real Inflow

Inflow1

Probability 1

Inflow11

Probability 11

Inflow111

Probability 111

Inflow112

Probability 112Inflow12

Probability 12

Inflow2

Probability 2

Inflow21

Probability 21

Inflow211

Probability 211

Inflow212

Probability 212

Input data

� Demand forecast

� Non dispatchable renewable energy

� Thermal units and their variable costs

� Hydro plants and their cascaded topologies

� Scenario tree with probabilities of inflows

Period 4Period 3Period 2Period 1



Optimal output

Optimal output 1

Optimal output 11

Optimal output 111

Optimal output 112

Optimal output 12

Optimal output 2

Optimal output 21

Optimal output 211

Optimal output 212

STOCHASTIC OPTIMIZATION

Output results

� For each node it obtains the optimal solution

� Through interpolation methods it obtains

multidimensional tables that show the optimal

output of hydro power plants depending on

- Week of the year

- Level of the reservoir

- Natural inflows

- Water reserve in the basin

Volume and water head in an small reservoir

645

650

655

150

200

250

Co

ta [

m]

Vo

lum

en

últ

il o

Ve

rtid

o [

hm

3]

Perc 95 %

Perc 5 %

Mínimo

Media

RMn



625

630

635

640

0

50

100

s08

40

s08

43

s08

46

s08

49

s08

52

s09

03

s09

06

s09

09

s09

12

s09

15

s09

18

s09

21

s09

24

s09

27

s09

30

s09

33

s09

36

s09

39

s09

42

s09

45

s09

48

s09

51

s10

02

s10

05

s10

08

s10

11

s10

14

s10

17

s10

20

s10

23

s10

26

s10

29

s10

32

s10

35

s10

38

Co

ta [

m]

Vo

lum

en

últ

il o

Ve

rtid

o [

hm

RMn

RMx

Vertido

TunelInf

TunelSup

TnlMn

TnlMx

Cota

Volume and water head in a medium reservoir

670

675

680

685

600

700

800

900

1000

Co

ta [

m]

Vo

lum

en

últ

il o

Ve

rtid

o [

hm

3]

Perc 95 %

Perc 5 %

Mínimo

Media

RMn



650

655

660

665

0

100

200

300

400

500

s08

40

s08

43

s08

46

s08

49

s08

52

s09

03

s09

06

s09

09

s09

12

s09

15

s09

18

s09

21

s09

24

s09

27

s09

30

s09

33

s09

36

s09

39

s09

42

s09

45

s09

48

s09

51

s10

02

s10

05

s10

08

s10

11

s10

14

s10

17

s10

20

s10

23

s10

26

s10

29

s10

32

s10

35

s10

38

Co

ta [

m]

Vo

lum

en

últ

il o

Ve

rtid

o [

hm

RMn

RMx

Vertido

TunelInf

TunelSup

TnlMn

TnlMx

Cota

Volume and water head in a large reservoir

210

215

220

2000

2500

3000

Co

ta [

m]

Vo

lum

en

últ

il o

Ve

rtid

o [

hm

3]

Perc 95 %

Perc 5 %

Mínimo

Media

RMn



190

195

200

205

0

500

1000

1500

s08

40

s08

43

s08

46

s08

49

s08

52

s09

03

s09

06

s09

09

s09

12

s09

15

s09

18

s09

21

s09

24

s09

27

s09

30

s09

33

s09

36

s09

39

s09

42

s09

45

s09

48

s09

51

s10

02

s10

05

s10

08

s10

11

s10

14

s10

17

s10

20

s10

23

s10

26

s10

29

s10

32

s10

35

s10

38

Co

ta [

m]

Vo

lum

en

últ

il o

Ve

rtid

o [

hm

RMn

RMx

Vertido

TunelInf

TunelSup

TnlMn

TnlMx

Cota

Volume in an small reservoir



Volume in a medium reservoir



Volume in a large reservoir



Optimal discharge tables

Q(m 3/s)

2 181 360 540 719 898 1077

37 0.0 104.8 125.8 146.7 167.7 188.6 262.0

47 0.0 104.8 125.8 146.7 167.7 188.6 262.0

Volumen útil del propio embalseInput: volume of reservoir1

Water release table

Inp

ut:

na

tura

l in

flo

w

Result: water release

for a volume of reservoir2 = 1609 hm3

for week February 9-15.



47 0.0 104.8 125.8 146.7 167.7 188.6 262.0

62 0.0 104.8 125.8 146.7 167.7 209.6 262.0

86 0.0 115.3 136.2 146.7 188.6 209.6 262.0

108 0.0 115.3 136.2 157.2 188.6 241.0 262.0

216 19.4 125.8 157.2 178.2 230.6 262.0 262.0

227 21.6 125.8 146.7 178.2 230.6 262.0 262.0

258 28.2 125.8 146.7 188.6 251.5 262.0 262.0

375 53.1 125.8 178.2 209.6 262.0 262.0 262.0

704 78.3 125.8 167.7 241.0 262.0 262.0 262.0

Inp

ut:

na

tura

l in

flo

w

Water release table (for week 52)



Volume of the reservoir

Natural inflows

Water release

Water release table (for week 7)



Volume of the reservoir

Natural inflows

Water release

System marginal cost

Peak (in red) and off-peak (in blue)



3

Introduction




• Simulation allows full detail modeling of hydro plant operation

– Nonlinearities in the production function

– Specific behavior of river basin elements

• Simulation can produce scheduling plans

– Closer to real operation

– With lower computational requirements

Introduction



– Reservoir

– Canal

– Plant

Hydro topology is represented by a graph of

nodes where each node is an element

Five types of nodes

(objects) are needed:

Object oriented programming simulation



– Plant

– Inflow point

– Special objects

Natural inflow

Reservoir

Hydro plant

Two kinds of strategies

a) Discharge decision taken from a pre-calculated optimal water

release table depending on:• Week of the simulated day

• Natural inflow

• Volume of the own reservoir

• Volume of a coordinated reservoir, if needed

• Table calculated by MHE stochastic hydrothermal model (usually

for the main reservoirs of the basin)



for the main reservoirs of the basin)

b) Discharge decision taken from guiding curves (usually for

small reservoirs) depending on:• Week of the simulated day

• Volume of the own reservoir

• Guiding curves & associated discharges

Optimal discharge tables

Q(m 3/s)

2 181 360 540 719 898 1077

37 0.0 104.8 125.8 146.7 167.7 188.6 262.0

47 0.0 104.8 125.8 146.7 167.7 188.6 262.0

Volumen útil del propio embalseInput: volume of reservoir1

Water release table

Inp

ut:

na

tura

l in

flo

w

Result: water release

for a volume of reservoir2 = 1609 hm3

for week February 9-15.



47 0.0 104.8 125.8 146.7 167.7 188.6 262.0

62 0.0 104.8 125.8 146.7 167.7 209.6 262.0

86 0.0 115.3 136.2 146.7 188.6 209.6 262.0

108 0.0 115.3 136.2 157.2 188.6 241.0 262.0

216 19.4 125.8 157.2 178.2 230.6 262.0 262.0

227 21.6 125.8 146.7 178.2 230.6 262.0 262.0

258 28.2 125.8 146.7 188.6 251.5 262.0 262.0

375 53.1 125.8 178.2 209.6 262.0 262.0 262.0

704 78.3 125.8 167.7 241.0 262.0 262.0 262.0

Inp

ut:

na

tura

l in

flo

w

Guiding curves



• Two effects studied:

– Variation of peak and off-peak hourly prices spread

– Variation of installed thermal capacity

• Realistic case of 9 reservoirs

• Simulation for 24 yearly series

– Previous generation of production/pumping water release tables for each case

Application case to hydro scheduling (i)



tables for each case

• Results for reservoir yearly operation

• Effect of the increased price spread among peak and off-peak hours:

– Narrower reservoir volume evolutions

Application case to hydro scheduling (ii)

2400

2800

Reservoir 1 Real Maximum

Upper limit curve Lower limit curve

Upper guide Lower guide

2400

2800






400

800

1200

1600

2000

2400

01

-jan

10

-jan

20

-jan

31

-jan

10

-feb

20

-feb

28

-feb

10

-mar

20

-mar

31

-mar

10

-apr

20

-apr

30

-apr

10

-may

20

-may

31

-may

10

-ju

n

20

-ju

n

30

-ju

n

10

-ju

l

20

-ju

l

31

-ju

l

10

-au

g

20

-au

g

31

-au

g

10

-sep

20

-sep

30

-sep

10

-oct

20

-oct

31

-oct

10

-nov

20

-nov

30

-nov

10

-dec

20

-dec

31

-dec

Vo

lum

e [

hm

3]

400

800

1200

1600

2000

2400

01

-jan

10

-jan

20

-jan

31

-jan

10

-feb

20

-feb

28

-feb

10

-mar

20

-mar

31

-mar

10

-apr

20

-apr

30

-apr

10

-may

20

-may

31

-may

10

-ju

n

20

-ju

n

30

-ju

n

10

-ju

l

20

-ju

l

31

-ju

l

10

-au

g

20

-au

g

31

-au

g

10

-sep

20

-sep

30

-sep

10

-oct

20

-oct

31

-oct

10

-nov

20

-nov

30

-nov

10

-dec

20

-dec

31

-dec

Vo

lum

e [

hm

3]

• Effect of the increased installed thermal capacity

– Allows free allocation of hydro production

– Does not need to keep a reservoir volume during summer

Application case to hydro scheduling (iii)

2400

2800




2400

2800






400

800

1200

1600

2000

2400

01

-jan

10

-jan

20

-jan

31

-jan

10

-feb

20

-feb

28

-feb

10

-mar

20

-mar

31

-mar

10

-apr

20

-apr

30

-apr

10

-may

20

-may

31

-may

10

-ju

n

20

-ju

n

30

-ju

n

10

-ju

l

20

-ju

l

31

-ju

l

10

-au

g

20

-au

g

31

-au

g

10

-sep

20

-sep

30

-sep

10

-oct

20

-oct

31

-oct

10

-nov

20

-nov

30

-nov

10

-dec

20

-dec

31

-dec

Vo

lum

e [

hm

3]

400

800

1200

1600

2000

2400

01

-jan

10

-jan

20

-jan

31

-jan

10

-feb

20

-feb

28

-feb

10

-mar

20

-mar

31

-mar

10

-apr

20

-apr

30

-apr

10

-may

20

-may

31

-may

10

-ju

n

20

-ju

n

30

-ju

n

10

-ju

l

20

-ju

l

31

-ju

l

10

-au

g

20

-au

g

31

-au

g

10

-sep

20

-sep

30

-sep

10

-oct

20

-oct

31

-oct

10

-nov

20

-nov

30

-nov

10

-dec

20

-dec

31

-dec

Vo

lum

e [

hm

3]

• Analysis of investment in new power plants in Douro river

• Example case:

– Simulation of 24 historical series

– Unplanned outage rate of 5%

• Assessment of the maximum outflow

– Power plant with up to 4 units of 200 m3/s and 48 MW

– Analysis of generation and spilled outflows

Application case to hydroelectric scheme design (i)



– Analysis of generation and spilled outflows

• Assessment of the maximum outflow

– Power plant with up to 4 units of 200 m3/s and 48 MW

– Analysis of results:

• Generation increase and spillage reduction

• Allocation of more energy in peak hours

Application case to hydroelectric scheme design (ii)



• Assessment of the number of units (1 to 4):

– For a fixed outflow of 600 m3/s

– Should be combined with the economic valuation of investment costs

– The increase from 1 to 2 units is more significant than the rest of new units installation

Application case to hydroelectric scheme design (iii)



• Medium term simulation model

– Connected to long-term stochastic hydrothermal model

– Considers detailed operation

– Stochastic inflows and outages

• Application case to hydro scheduling

– Provides feasible operation

– Different operation criteria

Conclusions



– Different operation criteria

• Application case to hydro scheme design

– Considering several options about installed units

– Unplanned outage sampling

Prof. Andres Ramos

Instituto de Investigación TecnológicaSanta Cruz de Marcenado, 2628015 MadridTel +34 91 542 28 00Fax + 34 91 542 31 [email protected]

www.upcomillas.es

Prof. Andres Ramos

http://www.iit.upcomillas.es/aramos/

[email protected]

[email protected]

Documents

Hydrothermal scheduling A case study - Comillas · 2013. 1. 27. · Hydrothermal scheduling. A case study b) Discharge decision taken from guiding curves (usually for small reservoirs)