Electricity market modelling: The case of VPP in the Netherlands

1
Electricity market modelling The case of VPP
in the Netherlands

• François Boisseleau Paul Giesbertz
• ETE Modelling Workshop
• 15-16 September 2005
• Leuven, Belgium

2
Content

• Background
• Liberalisation, competition, liquidity and VPP
• Why VPP?
• Ex-ante regulatory measures analysis
• Modelling VPP
• Model description
• Assumption
• Results
• Consultation
• Why?
• Consultation questions
• Conclusion

3
Background
4
Liberalisation, competition, liquidity and VPP

• Since the start of the liberalisation process in
  NL there have been concerns and discussions
  about
• The extent of competition (e.g. price spikes on
  APX)
• Integration of the Dutch market with neighbouring
  markets (e.g. Belgium)
• Potential regulatory measures to increase the
  level of competition and mitigate the exercise of
  market power (e.g. VPP).
• March 2004 DTe published a report providing an
  overview of the degree of competition on the
  Dutch electricity market for the year 2003
• Compared with 2002 liquidity has declined

5
Liberalisation, competition, liquidity and VPP

• DTe made 61 recommendations
• Promote co-ordination between TenneT and foreign
  grid operators, (e.g market Coupling?)
• Stagger the times at which import capacity is
  auctioned
• Balancing market for gas is introduced in the
  near future
• Facilitate the reduction of market players
  credit risks
• Enhancement of monitoring activities
• Improve transparency (info about OTC market and
  realised production)
• And
• Provide production capacity to the market through
  virtual power plants.
• This measure was considered as a last resort
  measure if the other above-mentioned
• measures fail to improve liquidity.

6
Why VPP?

• Objectives of VPP studied by DTe Increasing
  liquidity
• Indirect lowering market concentration --gt
  enhancement of market confidence
• Direct Distribution of supply/suppliers
• What is VPP?
• Options to buy electricity at a fixed price
• Premium is the result of an auction
• Experience in Europe Reduce dominant position
• France (EDF/ENBW case)
• Belgium (Electrabel/default supplier for the
  customers of several intermunicipal distribution
  companies)
• Netherlands (Nuon/Reliant case)

7
Ex-ante regulatory measures analysis

• Objective
• Ensure that relevant issues are examined
• Document the likely impact of each option
• Determining the most appropriate regulatory
  option
• Approach
• Study international experience
• Identify Critical key factors in designing VPP
• Quantify impact of different options (modelling)
• Consultation with those affected by proposed
  regulation

8
Modelling VPP
9
Market Power and Strategic Bidding - Different
Approaches in Theory
Advantages
Disadvantages
Cournot (Quantity)

• Flexible/Tractable
• Transmission constraint easy to consider
• Important existing literature

• Little descriptive power
• Not very realistic (pure quantity bids)
• Weak predictive power

Kahn (1998), Andersson and Bergman (1995), Hogan
(1997) Borenstein and Bushnell (1999), Smeers and
Wei (1999), Hobbs et al. (2002, 2003), Younes and
Ilic (1997), Berry et al. (1998), Stoft (1999),
Willems (2002)

• Flexible
• Tractable
• non-storability aspects

Bertrand (Price)

• Not very realistic (pure price bids)
• Producers have limited capacity
• Weak predictive power

Hobbs (1986), Aghion and Bolton (1987)

• More realistic (reflects actual bidding rules)
• Allow better understanding of companies
• bidding behaviors
• Better predictions

• Little computational tractability
• Multiple equilibrium
• Transmission constraint difficult to consider

SFE (Quantityprice)
Klemperer and Meyer (1989), Green and Newbery
(1992), Bolle (1992), Bohn et al. (1999),
Rudkevich et al. (1999), Day et al. (2001),
Baldick and Hogan (2001)

• Realistic (firms do not sell all their output to
  the
• spot market )
• Impact on market power?

• Requires additionnal assumptions e.g.
• Competition type in forward?
• chicken and egg

Forward Contracts
Allaz and Vila (1993), Bolle (1993), Powell
(1993),Batstone (2000), Newbery (1998) , Green
(1999)
10
SYMBAD Underlying Assumptions

• Approach based on linear Supply Function
  Equilibrium
• Results achieved by solving differential
  equations
• Generation
• Non-symmetric generation portfolios
• Piecewise linear marginal cost function
• Capacity availability per demand segment
• Several demand segments, each defined by
• Attributed load values
• Constant slope (elasticity per segment)
• Capacity availability
• Forward contracts can be included
• To take into account the fact that generators
  sell a large fraction of their output in advance
• Each generator sell in the forward market an
  exogenous quantity which is publicly known before
  bidding into the spot market

11
SYMBAD Approximation of MC Curves- Single Line
and Piecewise Linear Approximation
12
SYMBAD Input and Results

• Cost Parameters
• Generation assets and ownership
• Heat rate, fuel price, OM cost, other cost
• Demand Parameters
• Load curve (up to 8760 values)
• Definition of demand segments
• Slope (elasticity) per demand segment
• Capacity availability per generation asset (per
  demand segment)

• Numerical output (tables)
• Equilibrium price, equilibrium quantity
• Price mark-ups
• Supply mark-ups
• Numeric values are provided for each load value
• Graphical output (charts)
• Equilibrium price quantity and price supply
  mark-ups vs. demand and equilibrium quantity

13
Assumptions

• Major assumptions
• Elasticity 0,1
• Import always competitive distributed amongst
  player based on historical data
• Availability unit specific (/- 90)
• VPP (4 large producers)
• 10 base-load to 1 virtual player (K5a)
• 10 base-load to 4 virtual players (K5b)
• 5 base-load, 5 peak-load to 1 virtual player
  (K6a)
• 5 base-load, 5 peak-load to 4 virtual players
  (K6b)
• Long term contracts (4 large producers)
• 10 of capacity contracted (K7)
• 20 of capacity contracted (K8)

14
First screening market concentration HHI
calculations

• Some producers have dominant position in alI
  demand segments/ Market concentration in middle
  demand-segment (risk for strategic behaviour).
• Segment 2 2 large producers 95 market shares
• Segment 3 3 large producers 92 market shares
• Segment 4 many IPPs

15
Results 10 base - 4 virtual players
markups and prices lower 1 virtual player higher
markups in lower demand-segments (VPP player has
market power)
16
Results 5 base, 5 peak - 4 virtual players
Compared to 10 base lower markups/prices in low
segments
17
Conclusions VPP
Vertrouwelijk

• VPP has direct effect on prices and mark-ups
• Larger effect in concentrated demand segments
• Selection of specific units for VPP is more
  effective than portfolio approach
• In practice possibly less effect because
• VPP is defined on portfolio basis
• VPP can be used for export
• VPP can be bought directly by large consumer
• The modelling allows to identify additional
  questions rather than providing definitive
  answers and constitutes a basis for discussion
  with stakeholders (unit Vs portfolio? Restriction
  on export? Base load/Mid merit/peak load?Which
  player/units?)

Consumption in NL 100 TWh-gt1 Euro
average100,000,000 Euros/year
18
Consultation
19
Approach

• Several issues cannot be directly addressed
  through modelling and need additional analysis
• Auction design.
• Contract Duration?
• Nomination time (buyers) of the VPP contracts?
• Buying back by large producers?
• Which parties have to sell VPPs?
• Complexity VPP-prices?
• Impact on investment climate
• Strategies for implementation

Consultation
20
Issues to be addressed through consultation-exampl
es

• Contract Duration?
• Long term lowers incentives to influence short
  term market price
• Short term contracting is sufficient for
  increasing liquidity
• Long term contracting increases the price risk
  (need to be consistent with OTC)
• Complexity VPP-prices?
• A complex structure (based on fuel-indices) can
  reduce the attractiveness for smaller players and
  thus reduce demand for VPP-contract
• Take-or-pay conditions also represented a
  constraint for buyer and limit the attractiveness
  of the product.
• The size of the contract should be rather small
  (e.g 10 MW) to allow small players to
  participates in the auction

21
Four questions for consultation

1. Can VPP auctions be regarded as good instruments
   to improve liquidity and reduce market dominance?
2. What are the factors for successful
   implementation? contract duration, nomination
   rules, number of VPP sellers, complexity of VPP
   contracts and auction design
3. Will the impact on the investment climate be
   moderate if VPPs are clearly defined and well
   designed so that regulatory risk is limited?
4. When should VPPs be implemented? In current
   situation or only as measure of last resort?

22
Consultation process

• DTe published consultation document on July, 5
• Market parties can respond until October, 1
• DTe will write advice to Ministry of Economic
  Affairs

23
Conclusion
24
Conclusion

• Virtual Power Plants are a relatively novel type
  of behavioural remedy and have become rather
  popular to competition authorities and Regulators
  in European electricity markets
• Careful analysis of impacts and critical factors
  is required ex-ante to ensure efficient
  implementation
• Modelling is a useful tool as a starting point
  for analysis by providing quantitative
  assessments of different options and need to be
  completed with other type of analysis
• The consultation approach ensure that all
  relevant issues are examined to determining the
  most appropriate regulatory option and allow
  market participants to react (and get prepared!)
• Next step results of the consultation

25
Thanks for your attention

• Contact
• Francois.Boisseleau_at_KEMA.com
• P.G.M.Giesbertz_at_nmanet.nl
• Info on the consultation (in Dutch)
• http//www.dte.nl/nederlands/actueel/nieuwsbericht
  en/dte_verlengt_consultatie_virtual_power_plants.a
  sp