1. Inefficiencies in Zonal Market Coupling due to Uncertainties in Generation Shift Keys
Björn Felten, Tim Felling, Christoph Weber
Chair for Management Science and Energy Economics
University of Duisburg-Essen
40th IAEE International Conference
Singapore, June 19th, 2017

2. Agenda Introduction Theoretical Background Application
Results & Conclusions

3. Agenda Introduction Theoretical Background Application
Results & Conclusions

4. Overview Market Coupling Process in Central Western Europe (CWE)
Two days in advance „D-2“
One day in advance „D-1“
Time of Delivery „D“
 Redispatch possible
D-7
D-6
Price
zone B
Price zone C
Price zone A €
D-5
D-4
D-3
TSOs calculate parameters for „Flow-Based“ market coupling algorithm
Market clearing (incl. market coupling computation)
Matching bids & asks with aim of welfare maximization
Dispatch, prices, net exchanges
Delivery of electricity
(Redispatch measures if necessary to prevent critical grid situations)
Sources: BNetzA, BKartA (2015), Monitoringbericht 2015; EPEX Spot et al. (2016), Euphemia Public Description

5. Overview Market Coupling Process in Central Western Europe (CWE)
Two days in advance „D-2“
One day in advance „D-1“
Time of Delivery „D“
 Redispatch possible
D-7
D-6
Price
zone B
Price zone C
Price zone A €
D-5
D-4
D-3
TSOs calculate parameters for „Flow-Based“ market coupling algorithm
Market clearing (incl. market coupling computation)
Matching bids & asks with aim of welfare maximization
Dispatch, prices, net exchanges
Delivery of electricity
(Redispatch measures if necessary to prevent critical grid situations)
Uncertainty of line flows due to infeed & load assumption
Sources: BNetzA, BKartA (2015), Monitoringbericht 2015; EPEX Spot et al. (2016), Euphemia Public Description

6. Overview Market Coupling Process in Central Western Europe (CWE)
Two days in advance „D-2“
One day in advance „D-1“
Time of Delivery „D“
 Redispatch possible
D-7
D-6
Price
zone B
Price zone C
Price zone A €
D-5
D-4
D-3
TSOs calculate parameters for „Flow-Based“ market coupling algorithm
Market clearing (incl. market coupling computation)
Matching bids & asks with aim of welfare maximization
Dispatch, prices, net exchanges
Delivery of electricity
(Redispatch measures if necessary to prevent critical grid situations)
Uncertainty of line flows due to infeed & load assumption
Include „security“
margins in the parameters =
Decrease effects
of uncertainty
Flow Reliability Margins (FRMs)
Sources: BNetzA, BKartA (2015), Monitoringbericht 2015; EPEX Spot et al. (2016), Euphemia Public Description

7. Overview Market Coupling Process in Central Western Europe (CWE)
Two days in advance „D-2“
One day in advance „D-1“
Time of Delivery „D“
 Redispatch possible
D-7
D-6
Price
zone B
Price zone C
Price zone A €
D-5
D-4
D-3
TSOs calculate parameters for „Flow-Based“ market coupling algorithm
Market clearing (incl. market coupling computation)
Matching bids & asks with aim of welfare maximization
Dispatch, prices, net exchanges
Delivery of electricity
(Redispatch measures if necessary to prevent critical grid situations)
Main Research Questions:
How to assess the Flow Reliability Margins (FRMs)?
How high do the FRMs need to be in order to prevent redispatch?
How do alternative Market Designs influence the FRMs?
Uncertainty of line flows due to infeed & load assumption
Include „security“
margins in the parameters =
Decrease effects
of uncertainty
Flow Reliability Margins (FRMs)
Sources: BNetzA, BKartA (2015), Monitoringbericht 2015; EPEX Spot et al. (2016), Euphemia Public Description

8. Agenda Introduction Theoretical Background Application
Results & Conclusions

9. 4 nodes example - introduction
Nodal system
Zonal system i c g
node index
variable costs
generation f
cap
A(i,f)
ISO
line index
(line) capacity
nodal power transfer distribution factor
Independent System Operator q d
A(z,f)
CWE
net export
demand
zonal power transfer distribution factor
Central Western Europe
GSK
gmax,i X
Generation Shift Key
nodal generation capacity
margin to account for
uncertainty

10. 4 nodes example - commonalities
Nodal system
Zonal system
well-functioning
competition:
market outcome =
outcome of a
central optimization i c g
node index
variable costs
generation f
cap
A(i,f)
ISO
line index
(line) capacity
nodal power transfer distribution factor
Independent System Operator q d
A(z,f)
CWE
net export
demand
zonal power transfer distribution factor
Central Western Europe
GSK
gmax,i X
Generation Shift Key
nodal generation capacity
margin to account for
uncertainty

11. 4 nodes example - differences
Nodal system
Zonal system
well-functioning
competition:
market outcome =
outcome of a
central optimization
ISO: Integrated planning
(generation and grid)
 Less uncertainty due to grid
CWE:
D-2 expectation
used for D-1 i c g
node index
variable costs
generation f
cap
A(i,f)
ISO
line index
(line) capacity
nodal power transfer distribution factor
Independent System Operator q d
A(z,f)
CWE
net export
demand
zonal power transfer distribution factor
Central Western Europe
GSK
gmax,i X
Generation Shift Key
nodal generation capacity
margin to account for
uncertainty

12. 4 node example – perfect foresight
= 100 % Certainty
(GSKr = GSKe)
Solution space results from convex hull (polyhedron) of zonal line constraints

13. 4 node example – one-sided uncertainty
Potentially realized
GSK(r) ≠ GSKe
Line loads can differ from expectation
Solution space is smaller (if redispatch is to be prevented)

14. 4 node example – two-sided uncertainty
Uncertainties in both directions make it more complicated

15. 4 nodes example – critical situations
Critical exchange situations are the corners of the convex hull of the potentially realized line constraints (family of hyper planes)

16. 4 nodes example – ex. critical situation
Let’s pick an example…
Prevention of potential
redispatch

17. Prevention of potential
4 nodes example – FRM
original zonal
line capacity
constraint
required zonal
constraint in
worst case
Prevention of potential
redispatch
FRM = Flow
Reliability
Margin

18. 4 nodes example – FRM FRM = Flow Reliability Margin
original zonal
line capacity
constraint
required zonal
constraint in
worst case
Prevention of potential
redispatch
FRM = Flow
Reliability
Margin
Potential welfare reduction

19. 4 nodes example – FRM FRM = Flow Reliability Margin
How to determine the FRMs required to prevent redispatch?
How do FRMs develop with diminishing price zone size?
original zonal
line capacity
constraint
required zonal
constraint in
worst case
FRM = Flow
Reliability
Margin

20. Derivation of „worst case“ FRMs
Linear supply function
GSK formulation
Derivative of zonal PTDF
Taylor approximation of load flow (1st order)
Formulating linear optimization problem
s.t.

21. Agenda Introduction Theoretical Background Application
Results & Conclusions

22. Application – Used Case
Current CWE+
Methodology from: Felling, Weber (2016), Identifying price zones using
Nodal prices and supply & demand weighted nodes

23. Application - Algorithm
Determine Price Zones
Determine nodal PTDFs
Matpower: R. D. Zimmermann, C. E. Murillo-Sanchez, and R. J. Thomas, “Matpower: Steady-State Operations, Planning and Analysis Tools for Power Systems Research and Education", IEEE Transactions on Power Systems, vol. 26, no. 1, pp , Feb. 2011
Operational Handbook Entso-E:
Calculate zonal PTDFs
Solve zonal market clearing (for representative exchange situations)
As explained in the 4 nodes example…
Solve maximization problem for FRMs
As shown before…

24. Agenda Introduction Theoretical Background Application
Results & Conclusions

25. First results – relative worst-case FRMs
preliminary

26. Conclusions Methods: Observations: Next steps:
A theoretical deviation of Worst-Case Flow Reliability Margins (FRMs) has been accomplished
A practical methodology for estimating FRMs in real-world systems has been proposed and tested
Observations:
FRMs tend to decrease with increasing number of price zones
However the decrease is not monotonic (i.e. eventually adding one zone may increase the Worst-Case FRM slightly)
Particularities of the grid, the price zone configurations and the considered exchange situations effect the Worst-Case FRMs (e.g. country  5 zones)
Next steps:
Further analyses (e.g.: driving factors for uncertainty impact, more restrictions, interdependencies of FRM choices)

28. Backup

29. Backup – Average FRM contribution of node within zone

30. Backup – further FRMs
preliminary