Integration of RES in the Swiss electricity grid ...€¦ · Integration of RES in the Swiss electricity grid Challenges, solutions and methods Alexander Fuchs (Research Center for

Integration of RES in the Swiss electricity gridChallenges, solutions and methods

Alexander Fuchs (Research Center for Energy Networks, ETH Zurich)March 15, 2016

Project overview

Financed by CCEM, Swisselectric Research

Phase 1: Distribution grid aspects (2014-15)

Phase 2: Transmission grid aspects (2016-17)

Project partners

ETH Zurich (Research Center for Energy Networks, FEN)

Paul-Scherrer-Institut (Technology-Assessment-Group, TAG)

Paul-Scherrer-Institut (Energy-Economics-Group, EEG)

Hochschule Luzern (HSLU)

CKW, Axpo

Overall goal: Develop a general planning and strategy tool fordistribution grids .

Integration of RES in the Swiss electricity grid ‖ A. Fuchs ‖ March 15, 2016 2 / 48

Outline

Situation today

Grid operation and planningOptimal grid operationOptimal grid planning

ResultsResult structureResult interpretation

Summary


Outline

Situation today

Situation today

Grid operation and planning

Results

Summary


Grid operation today

Situation today

Transmission Grid

Distribution Grid

Medium and Low Voltage Grid

Small Consumer

Industry

Bulk Consumer

ConventionalControllable

RenewableControllable


RenewableControllable Conventional

Controllable



Predictable loads, pricesand production capacities

Sufficient stabilityreserves in the grid

Unidirectional load flow

Established operationalprocedures


Grid operation future

Situation today

Transmission Grid

Distribution Grid

Medium and Low Voltage Grid

Small ConsumerLoad Management

IndustryLoad Management

Bulk ConsumerLoad Management




RenewableVolatile

RenewableVolatile



RenewableVolatile

RenewableVolatile


ControllableControllable VolatileLoadControllable

Limited predictability ofloads, prices; volatileproduction

Reduced stabilityreserves in the grid

Bidirectional load flow,production on thedistribution grid level

New operationalprocedures required


Challenges for the distribution grid operator

Situation today

Especially uncertainty regarding:

the development of future network structure.

the development of future production capacity and demand profiles.

new regulatory framework (Swiss and Europe wide).

As a result uncertainty regarding future operating procedures andinvestment decisions.


Goals of the project

Situation today

Systematic investigation of existing approaches regarding theirapplicability for the Swiss distribution grids.

Quantification of costs, risks and chances of each approach.

Determine conclusions and recommendations for the Swissdistribution and transmission grid operators.

Main result is a road map for a concrete network example and a generalsoftware tool for the investigation of other networks.


Outline


Situation today


Results

Summary


Outline

Grid operation and planning ‖ Optimal grid operation

Situation today



Summary


Grid components


Trafos Loads PV-Sources Batteries

0 5 10 15 200

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

hour

MW

0 5 10 15 2025

30

35

40

45

50

55

60

65

70

hours

CH

F/M

Wh

Figure: Load profile (left, blue), PV profile (left , green), Price profile at the trafo (right)


Classical distribution grid


Aggregated loads at the trafo

Load and price profile are given

Controllable quantities: none


Classical distribution grid - example


Load and losses compensated by the trafo

0 5 10 15 20−0.4

−0.2

0

0.2

0.4

0.6

0.8

1

1.2

1.4

hours

MW

Figure: Load profile (red) and trafo power (blue).


Classical distribution grid with PV



Profiles of Load, PV and price are given

Controllable quantities: none


Classical distribution grid with PV - example


Load and losses compensated by the trafo and the PV source

0 5 10 15 20−0.4

−0.2

0

0.2

0.4

0.6

0.8

1

1.2

1.4

hours

MW

Figure: Load profile (red) and trafo power (blue), PV-Power (black).


Distribution grid with controllable PV




Controllable quantities: PV curtailment


Distribution grid with controllable PV - example


Load and losses compensated by the trafo and the PV source

PV peak power brings network to its limits (thermal or voltage).

0 5 10 15 20−2

−1

0

1

2

3

4

hours

MW

Figure: Load profile (red) and trafo power (blue), PV-Power (black), available PV power (blackdashed).


Distribution grid with controllable PV and batteries




Controllable quantities: PV curtailment, battery powerIntegration of RES in the Swiss electricity grid ‖ A. Fuchs ‖ March 15, 2016 18 / 48

Distribution grid with controllable PV and batteries -example


Load and losses compensated by the trafo and the PV sourcePV peak power brings network to its limits (thermal or voltage).Battery allows increased usage of PV and cheap electricity from thefeeder.

0 5 10 15 20−2

−1

0

1

2

3

4

hours

MW

Figure: Load profile (red) and trafo power (blue), PV-Power (black), available PV power (blackdashed), battery power (green).


Distribution grid with controllable PV and batteries -example


Planning algorithm considers state-of-charge, power limits, efficiencyof the batteries.Planning algorithm can be repeated hourly to include new predictions(weather, price, demand, ...).

0 5 10 15 20−0.5

0

0.5

1

1.5

2

hours

MW

h

Figure: Battery state-of-charge (blue) and charge limits (blue dashed).


Comparison of energy costs


0 5 10 15 20 25−600

−400

−200

0

200

400

600

800

1000

1200

1400

hours

CH

F

Figure: Cumulative energy costs no PV (red), with PV (blue), with controlled PV (black), withcontrolled PV and storage (green).


Grid operation - Mathematical problem formulation


J ∗ =(

minu1,...,uN

N∑i=1

li(xi , ui))

s.t. ∀i = 1, ..., N : g(xi , ui) ≤ 0, h(xi , ui) = 0,

Nonlinear multi-period AC power flow problem as extension to thesoftware Matpower (no DC power flow approximation necessary)

Efficient Parallelization possible with MATPOWER / IPOPT /PARDISO (http://www.pardiso-project.org)

Further improvement of the solution speed possible by exploiting theproblem structure


http://www.pardiso-project.org

Outline

Grid operation and planning ‖ Optimal grid planning

Situation today



Summary


Overview


Investigation of CKW distribution grids near Luzern.

Assumption: High increase of PV injection in the distribution grid

Figure: Distribution grid example Trafos (black and blue), 20 kV lines (red), 400 V lines (green).


Solution approaches for PV integration


Several academic methods and general approaches with referencesystems available:

Grid extension, operation and business as usual

Usage of storage elements

Load management, smart metering

Production management and control


Grid planning - Mathematical problem formulation


J ∗ = minp

(min

u1,...,uN

N∑i=1

li(xi , ui , p))

(1)

s.t. ∀i = 1, ..., N : g(xi , ui , p) ≤ 0, h(xi , ui , p) = 0, (2)

Planning parameter p includes grid extension, degree of PVcurtailment and size/location of the storage units

Can be included in the basic multi-period OPF problem formulation

Is solved for a scenario of family of scenarios


Outline

Results

Situation today


Results

Summary


Outline

Results ‖ Result structure

Situation today



Summary


PV-injection without storage


0 5 10 15 200

5

10

15

hour

MW

Figure: Distribution grid example Sursee: Maximum grid capacity (black), maximum PV-Injectionwithout PV-curtailment (blue), increased PV-injection with PV-curtailment (green), stronglyincreased PV-injection with PV-control (red), available PV-capacity (dashed).


PV-injection without storage


0 5 10 15 200

1

2

3

4

5

6

7

8

9

10

hour

MW

Figure: Distribution grid example Sursee: Maximum PV injection without storage (blue),increased PV-injection with storage for 30 minutes PV (green), with storage for 60 minutes PV(red), available PV-capacity (dashed).


Placement and sizing of storage units


0 50 100 150 200 250 3000

0.5

1

1.5

2

2.5

3

3.5

Node number

MW

h

Figure: Distribution of the storage sizes at different nodes of the grid for a total storage capacityof Etotal = 33MWh.

→ Storages are placed in the entire grid.


Placement and sizing of storage units


0 5 10 15 20 25 30 35−0.2

0

0.2

0.4

0.6

0.8

1

1.2

MWh

corr

elat

ion

Figure: Correlation between optimal storage distribution and installed PV capacity PPV,rated atevery node, as a function of the total storage capacity Etotal.

→ Storages are sized according to the size of nearby PV units.


Comparison of PV curtailment, storage and networkextension


0 5 10 15 20 25 30 350

5

10

15

20

25

30

35

MWh

%

K

G = 1.00

KG

= 1.25

KG

= 1.50

KG

= 1.75

KG

= 2.00

KG

= 2.25

KG

= 2.50

Figure: Ratio of curtailed PV power as function of the total storage capacity Etotal for differentdegrees of grid expansion KG .

→ Grid extension results in strong reduction of PV curtailment.


Decision surface for PV integration strategies


0500

10001500

20002500

0500

10001500

20002500

30003500

0

500

1000

1500

2000

2500

Storage cost [CHF/day] Cable cost [CHF/day]

Cur

tailm

ent c

ost [

CH

F/d

ay]

Figure: Daily operational grid costs for different strategy combinations. Individual simulations(blue stars) and approximation by a decision surface. Assumed cost parameters are the averageelectricity cost of c = 55CHF/MWh, grid degradation costs of pG = 55CHF/(km · day) andstorage degradation costs of pS = 70CHF/(MWh · day) .


Comparison of PV integration strategies


Trade-off between storage costs pS, grid degradation costs pG andelectricity costs c.Minimum sum of all three costs is always an extreme point of thedecision surface (corresponds to a single pure strategy).

Linear criterion for the optimality of each strategy:pS < 1.29pG and pS < 1.16c: Use only storage units.pG < 0.77pS and pG < 0.90c: Use only grid extension.c < 1.12pG and c < 0.86PS: Use only PV curtailment.

Without the linear approximation the minimum total cost is also reachedat a combination of different strategies.


Outline

Results ‖ Result interpretation

Situation today



Summary


Simulation example for fixed cost parameters


0 5 10 15 20 25 30 35 40 45 50-4

-2

2

4

Hour

MW

DemandImport

0 5 10 15 20 25 30 35 40 45 50-2000

-1000

0

3000

4000Cumulative costs

Power

CH

F

ImportBattery SOCBattery PBattery constantTotal costsObjective

Two-day simulation exampleof the CKW grid Ettiswil withan increased PV productionand high demand: Powerdemand and import (top).Cumulative electricity andoperational costs (bottom).


Fixed simulation parameter


Parameter are fixed before the simulation.

Simulation of the Sursee region (summer and winter day, 48 hours)

PV integration potential of today’s grid is determined and increased

Load profile of the households are maximized and simulated withcoincident factor (25% base load and random distribution )


Variable simulation parameter


Only a range of simulation parameters has to be selected.

Electricity price (20 - 80 CHF/MWh)

Battery cost model: J = bE(SOC − a)2 + cP + dEstorage size EState-of-charge SOCLoad power P

Parameter {a, b, c, d} can be fitted to multiple battery models andtechnologies.

Grid extension is a constant increase of the trafos, lines and cables .

PV curtailment is not remunerated, but enters the planning decisionthrough opportunity costs.


Full operational costs


0 50 100 150 200 250 300 350 400−4000

−2000

0

2000

4000

6000

8000

Batterie SOC Kosten Parameter b [CHF/(MWh h)]

Ges

amtk

oste

n S

trom

und

Bet

rieb

[CH

F]

c = 0 CHF/(MW h), Strompreis = 20 CHF/MWh, Ausbaugrad = 1, hohe/niedrige Nachfrage

d = 2.5 CHF/(MWh h)d = 5 CHF/(MWh h)d = 10 CHF/(MWh h)d = 20 CHF/(MWh h)

0 50 100 150 200 250 300 350 400−4000

−2000

0

2000

4000

6000

8000

Batterie SOC Kosten Parameter b [CHF/(MWh h)]G

esam

tkos

ten

Str

om u

nd B

etrie

b [C

HF

]



Figure: Optimal grid planning with variable battery parameters. High demand (solid lines) andlow demand (dashed lines). Low electricity price (left) and high electricity price (right). No gridextension.


Sizing of installed storage units


0 50 100 150 200 250 300 350 4000

1

2

3

4

5

6

7

8

9

10


Ges

amte

nerg

ie B

atte

riesp

eich

er [M

Wh]



0 50 100 150 200 250 300 350 4000

1

2

3

4

5

6

7

8

9

10


Ges

amte

nerg

ie B

atte

riesp

eich

er [M

Wh]





Curtailed PV power


0 50 100 150 200 250 300 350 4000

0.05

0.1

0.15

0.2

0.25

0.3


PV

Abr

egel

ungs

quot

e [%

]



0 50 100 150 200 250 300 350 4000

0.05

0.1

0.15

0.2

0.25

0.3


PV

Abr

egel

ungs

quot

e [%

]





Results


Simulation results form a decision surface, depending on multiplesimulation parameter.

Based on the decision surface, the grid owner can quickly makeplanning decisions for selected parameter values.

0

100

200

300

400

5

10

15

20

0

2000

4000

6000

8000

Batterie SOC Kosten [CHF/(MWh h)]Batterie konstante Kosten [CHF/(MWh h)]

Ge

sa

mtk

oste

n S

tro

m u

nd

Be

trie

b [

CH

F] Location and size of the storages

are determined by theoptimization.

Control of the controllablequantities are part of theoptimization.


Selected observations


Storage technologies are only interesting for further decrease ofstorage costs or higher electricity costs.

Evaluation of the grid extension depends on the costs of theindividual grid components.

High potential for controlled loads and PV components (currentlysimulated without additional costs).


Outline

Summary

Situation today


Results

Summary


Summary

Summary

Unified problem formulation for the grid operation and planning.

Optimal usage of available assets to compute a Pareto decisionsurface

AC power flow extension for planning and multi period scenarios .

Large multi-period simulations of AC power flow problems (1000nodes, 1000 time steps) become tractable with dedicated solvers.


Ongoing work

Summary

Quantitative:

Comparison of economic and CO2 objective

Qualitative:

Robust plan with scenario uncertainty

Long term transfer of the methods to general distribution andtransmission grids


Thank you for your attention

Summary

FEN - Research Center for Energy Networks

ETH Zentrum SOI C 1CH-8092 Zurich

Dr. Alexander FuchsSonneggstrasse 28Tel: +41 44 632 28 60Fax: +41 44 632 13 30E-mail: [email protected]: www.fen.ethz.ch

Support through the Competence Center for Energy and MobilityProject ISCHESS: Integration of stochastic renewables in the Swiss electricity supply system .


Documents

Integration of RES in the Swiss electricity grid ...€¦ · Integration of RES in the Swiss electricity grid Challenges, solutions and methods Alexander Fuchs (Research Center for