Auctions are increasingly being applied as a mechanism to allocate support to renewable energy sources (RES). AURES (Auctions for Renewable Energy Support) is a H2020 European research project focused on auction designs for renewable energy support. The project addresses the important and urgent issue of improving current support policies for electricity from renewable energy sources through competitive market measures. The general objective of the project is to promote an effective use and efficient implementation of auctions for renewable energy support in the European Union Member States, especially regarding their cost-efficiency. In this new webinar series, the AURES team will share research results and provide guidance to policy makers on the best options to organize renewables support under the new rules of the Clean Energy Package.

1. Auctions for Renewable
Energy
Model-based analysis
Marijke Welisch (TU Wien)
Gustav Resch (TU Wien)
Webinar Leonardo Academy,
27/06/17

2. AURES: Who we are
 A coordination and support action under the EU Horizon2020 programme
 Project runs from January 2015 to December 2017
 Eight partners from seven EU countries
 Cooperation with policy makers, market participants and other stakeholders.

3. AURES project at a glance
AURES combines
 Target-oriented analysis
empirical analysis
interviews with stakeholders
lessons from other industries
auction experiments
simulations in energy models
 Capacity building activities
workshops
webinars
case cooperations
bi- and multilateral meetings
interactive website
…find more information on:
auresproject.eu

4. Auctions for Renewable Energy
Game Theoretic Modelling

5. Motivation
• Ongoing auction schemes in many European member states & even
more are planning to get started soon
• EU legislation (Directive 2009/28/EC on the promotion of the use of
energy from renewable sources and “Guidelines on state aid for
environmental protection and energy 2014-2020” )
• Currently we can observe lower prices in each auction round and for all
kinds of technologies (onshore & offshore wind in Germany, PV in
Denmark etc.)
• Long-term experience exists in energy auctions (e.g. in Brazil, in the UK –
we can also learn from past mistakes…)
• In AURES, we have the unique possibility of observing and consulting
auctioning entities for designing and improving their auction scheme

6. Research Question
• The question investigated is part of a larger study assessing different
auction designs and their influence on outcomes
• We assess different questions of interest for/in different partner countries
and show how changing certain auction design parameters changes
auction outcomes
• For the UK, we investigate the influence of penalties and
prequalification criteria on the CfD (renewables) auction outcome
• Further cases assessed are onshore wind & ground-mounted solar PV
auctions in Germany and ideas for multi-unit auction schemes in
Denmark

7. Modelling Framework
Potential auction participant (Agent)
Actor-specific characteristics
• Level of risk aversion (utility function) including the ability to bid strategically
• Budgetary constraints (interest rate, amount and limits)
• Available information about competitors
Auction-design specific characteristics
• Sunk costs (participation fees, contracting land
owners, penalties etc.)
• Expected level of competition
Technology-specific characteristics
• Available potentials (location, size, technology)
and corresponding costs (assuming risk free
interest rate, etc.)
• Point in time of auction (Repetitive auctions)
Auction design
• Amount and type of
auctioned goods (e.g.
energy vs. capacity)
• Prequalification criteria
(minimum bid size, etc.)
• Diversity targets?
Offered Bid
• Criteria based on location,
access priority etc.
• Learning effects
• Bid strategy according to theory• Determines available strategies

8. • Uniform pricing: bid truthfully, incentive compatible
• Pay-as-bid: bid at least their costs, bid shading; maximize expected profit
over all rounds (discount factor δ for future rounds)
for t=0,1,2,...,T
𝐸𝐸 𝜋𝜋 𝒃𝒃 = � 𝛿𝛿𝑖𝑖−𝑡𝑡
𝑇𝑇
𝑖𝑖=𝑡𝑡
⋅ (𝑏𝑏𝑖𝑖 − 𝑐𝑐𝑡𝑡 ⋅ 𝑃𝑃𝑃𝑃("𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑏𝑏𝑏𝑏𝑏𝑏 𝑖𝑖𝑖𝑖 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑖𝑖𝑖)
� 𝑃𝑃𝑃𝑃("𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 𝑏𝑏𝑏𝑏𝑏𝑏 𝑖𝑖𝑖𝑖 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑖𝑖 − 𝑥𝑥𝑥)
𝑖𝑖−𝑡𝑡
𝑥𝑥=1
)
Modelling Framework
𝑏𝑏𝑖𝑖 = 𝑐𝑐𝑡𝑡

9. Modelling Case: Penalties and pre-qualifications in the UK
CfD auction scheme

10. • The original policy objective of the Contract for Difference (CfD) auctions
was to increase competition within technology groups to bring down
support costs and limit producer surplus
• The CfD auctions are multi-unit, sealed-bid, uniform price auctions and
they have technology-specific ceiling prices
• The auction takes place for several bidding years at one point in time;
each bidding year is capped by a certain budget
• The auctions are technology diverse, but divided into two pots: mature and
less mature technologies (we look at the mature technology pot)
Source:
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/307993/uk_national_
energy_efficiency_action_plan.pdf
Background: The UK auction scheme

11. • Two technologies participate: onshore wind and solar PV; four types of
bidders (weak and strong for each technology); 21.5 % for solar PV and
78.5 % for wind onshore to represent the sector in the year 2014
• Auctioned capacity is translated from the budget by using the official
“budget impact” equations by DECC
• Pricing rule is uniform pricing with the highest awarded bid determining the
final strike price for all bidders
• Competition in total is assumed to be rather low (as has proven to be the
case in the actual auctions)
• Bidders’ costs are derived from several sources on current and future cost
developments, as bidding years are up to 2020
Model set-up

12. • We compare two cases: Uniform pricing with or without a functioning
penalty scheme in place
• Risk behaviour changes depending on whether or not bidders expect to be
penalized for a bid that does not cover their costs:
The bidder receives a signal x with an uncertainty factor :
The bidding function resulting for the non-penalty case is:
• As the bidder is able to default, she can submit a bid in the lower bound of
the range of her signal, even though it might result in a loss
Model set-up
[ , ]wherey x ε ε= + ∂ ∂∈ −
2
( )b x x= + ∂

13. Agent distribution
Agents
PV
(strong
bidder)
PV
(weak
bidder)
Wind
(strong
bidder)
Wind
(weak
bidder)
Average number of
bidders (first delivery
year)
10 10 15 15
New random draw of
bidders (per delivery
year)
0-2 0-2 0-2 0-2
Range of capacity bid
[MW]
5-15 5-15 5-50 5-50
Cost distribution
[p/kWh]
4.7-6.2 6.2-7.6 7.1 - 8 8 - 9.4
Cost digression
piecewise: 7.5% first year,
then 2.5% annually
1.95% per year

14. Bidding process
2015/2015
(£50 M)
2016/2017
(£65 M)
2017/2018
(£65 M)
2018/2019
(£65 M)
2019/2020
(£65 M)Bidder
• One shot
auction
• Bidder can bid
into one of five
bidding years
• Each bidding
year is capped
by a budget

15. Bidding process
Auction participant (Agent)
Actor-specific characteristics
• Level of risk aversion (utility function)
• Budgetary constraints
Auction-design specific characteristics
• Sunk costs (participation fees, penalties etc.)
• Expected level of competition in certain delivery
year
Technology-specific characteristics
• Available potentials (location, size, technology)
and corresponding costs (assuming risk free
interest rate, etc.)
• Delivery year
Auction design
• Amount (budget constraint
per year)
• Prequalification criteria
• Penalty (functioning?)
Offered Bid
• Bid strategy according to theory• Determines available strategies

16. Model results
UK Contract for Difference (CfD) auction prices (modelled results)
*Comparison of the two modelled cases; one can see that the prices do not differ
substantially between the case with and without penalty

17. Model results
*Dropout is shown as number of withdrawals; this is on average 45 MW per round,
depending on whether large or small bidders drop out, up to 100 MW of drop out is possible

18. Findings and policy implications
• Auctions without penalties and/or pre-qualification criteria have a higher
likelihood of strategic bidder behaviour which leads to a certain amount of
default (insecurity)
• This does not come at the benefit of lower prices but does increase
slightly the average profit of bidders
• In terms of policy implications, if the auctioning authority wants to achieve
certain capacity goals, it should consider functioning penalties and/or pre-
qualification criteria

19. Thank you!
Any questions?

20. Auctions for Renewable Energy
Energy System Modelling

21. Model-based analysis of 27% RES by 2030
Aim: Prospective (renewable) energy system modelling in the
2030 context at European (EU) level:
• Conduct a comparative assessment of the performance of auctions
to other instruments used for incentivising of renewable energy
deployment
• Shed light on the required RES uptake for meeting 27% RES by 2030,
and on RES-related costs & benefits
– in focus here: policy costs (i.e. support expenditures)

22. Simulation model for energy policy instruments
in the European energy market
•RES-E, RES-H, RES-T and CHP, conventional power
•Based on the concept of dynamic cost-resource curves
•Allowing forecasts up to 2030(2050) on national / EU level
Reference clients: European Commission (DG RESEARCH, DG TREN, DG ENV, DG
ENER), Sustainable Energy Ireland, German Ministry for Environment, European
Environmental Agency, Consultation to Ministries in Serbia, Luxembourg, Morocco, etc.
Base input
information
Scenario
Information
Power
generation
(Access Database)
Policy
strategies
selection
Social behaviour
Investor/consumer
Externalities
Framework
Conditions
(Access Database)
Results Costs and Benefits on a yearly basis (2005-2020 )
Country
selection
Electricity
demand reduction
(Access Database)
Technology
selection
Economic
market and policy
assessment
potential, costs,
offer prices
Simulation of
market interactions
RES-E, CHP, DSM
power market, EUAs
Base input
information
Scenario
Information
Power
generation
(Access Database)
Policy
strategies
selection
Social behaviour
Investor/consumer
Externalities
Framework
Conditions
(Access Database)
Results Costs and Benefits on a yearly basis (2005-2020 )
Country
selection
Electricity
demand reduction
(Access Database)
Technology
selection
Economic
market and policy
assessment
potential, costs,
offer prices
Simulation of
market interactions
RES-E, CHP, DSM
power market, EUAs
(2006-2050)
The applied modelling system:
The Green-X model (complemented by HiREPs)

23. Model coupling: Green-X & HiREPS for a
detailed assessment of RES developments in the electricity sector
Solar PV Wind Locational power plant database Transmission Grid
RES policy
Non-economic barriers
Dynamic cost-potential curves,
Policy interaction, Investment
decision
RES LRMC [€/MWh]
[MWh]
RES support expenditures
RES investments
Supply Demand
Storage
Common electricity
market model
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
1 25 49 73 97 121 145 169 193 217 241 265 289 313 337 361 385 409 433 457 481 505 529 553 577 601
Power plant dispatch & -
commitment
Transmission grid
expansion
Electricity prices
Total system costs
• Energy/CO2-price development
• Default (2010) technology costs
• Energy demand development
• RES deployment
• Dynamic cost development
(technological learning)
• Feedback on RES market values
• Feedback on curtailment of RES
• Feedback on electricity prices
ITERATION
Yearly time resolution (2006 -2050),
years modelled: 2010 to 2030
Hourly time resolution (8760h),
years modelled: 2020, 2030, (2050)
Financing conditions
RES deployment
(electricity, heat, transport)
RES system costs
Assessment of benefits
(CO2 & fossil fuel avoidance)
Green-X HiREPS

24. Key parameter
• To ensure maximum consistency with existing EU scenarios and
projections the key input parameters of the Green-X scenarios are based
on PRIMES modelling and the Green-X database
Based on PRIMES* Based on Green-X database Defined for this assessment
Primary energy prices Renewable energy technology
cost (investment, fuel, O&M)
Renewable energy policy
framework
Conventional supply portfolio
and conversion efficiencies
Renewable energy potentials Reference electricity prices
CO2 intensity of sectors Biomass trade specification
Energy demand by sector Technology diffusion / Non-
economic barriers
Learning rates
Market values for variable
renewables
Main input sources for scenario parameters
*PRIMES scenario used: PRIMES euco27 … with 27% RES by 2030,
and an enhanced use of energy efficiency (27%) compared to
reference conditions. [cf. (SWD (2016) 410 final)].

25. Key parameter: Energy price trends
Energy price trends based on PRIMES modelling
(EU reference scenario as of 2016)
0
10
20
30
40
50
60
2010 2015 2020 2025 2030
Coal
(PRIMES
2016)
Oil
(PRIMES
2016)
Gas
(PRIMES
2016)
Fossilenergyprices[€/MWh]
0
5
10
15
20
25
30
35
40
2010 2015 2020 2025 2030
PRIMES
reference case
PRIMES euco27
(moderate
efficiency)
PRIMES euco30
(strong
efficiency)
Carbonprices(ETS(&Non-ETSpost2020))[€/tCO2]
Assumed future trends for fossil fuel (left) and carbon prices (EU-ETS) (right)

26. Scenarios assessed
Overview on assessed cases
Harmonised
Quota
Harmonised (RES) support post 2020:
EU-wide quotas with certificate trading for RES-E
Stringent State Aid
Guidelines
Stringent implementation of State Aid Guidelines:
National auctions for RES-E support through sliding
premiums with
- partial, or
- full market opening
National Policies
with common
Guidelines
National Policies with common guidelines:
- National quotas with certificate trading for RES-E, or
- national auctions for RES-E support through sliding
premiums without market opening

27. Source: Green-X (2017)
Results: RES(-e) deployment up to 2030
27.3%
27.1%
27.2%
18%
19%
20%
21%
22%
23%
24%
25%
26%
27%
28%
2020 2025 2030
Harmonised
Quota
Stringent State Aid
Guidelines
National Policies
with Common
Guidelines
RESshare(shareingrossfinalenergydemand)[%]
50.4%
50.2%
51.0%
36%
38%
40%
42%
44%
46%
48%
50%
52%
2020 2025 2030
RES-Eshare(shareingrossfinalelectricitydemand)[%]
Comparison of the resulting RES (left) and RES-E deployment (right) in relative terms
(i.e. as share in gross final energy/electricity demand) over time in the EU 28
 27% RES by 2030 implies an increase of the RES-E share to ~50% within
the electricity sector

28. Source: Green-X (2017)
Results: How ambitious is the
required RES-e uptake?
Net and gross increase of renewable
generation at EU level
by decade (2010-2020 vs. 2020-2030)
in the electricity sector
in accordance with 27% RES by 2030
 Net and gross (i.e. incl.
replacement of existing RES
installations) increase of RES
generation strongly depend on
energy efficiency / future
demand growth
 Under reference developments
the ambition level can be
maintained when comparing this
and the next decade
 With strong energy efficiency
(30% demand reduction) the
required RES-E uptake is
however modest
0
100
200
300
400
500
600
700
800
IncreaseofRES-Egenerationbydecade[TWh/a]
Net
increase
Gross
increase
2010 to 2020 2020 to 2030

29. Results: The need for dedicated RES support
0
25
50
75
100
125
2015 2020 2025 2030
Levelised(15years)weightedaverage
remunerationforyearlynewRES-Einstallations
andcorrespondingmarketvalues[€/MWhRES]
Total remuneration
of RES-electricity*
Market value of
RES-electricity*
Wholesale price*
Net support
for RES-E
(on average)
Note: *Dotted lines show
the average values (at EU
level) of different scenarios
in line with 27% RES by 2030.
The shaded areas indicate the
ranges that occur across these
scenarios.
Future development of remuneration levels and corresponding market values of
renewable energy technologies (on average) at EU-28 level
 Moderate dedicated support for renewables
is required to reach the 2030 target of 27% RES
Source: Green-X (2017)

30. 0
5
10
15
20
25
30
35
40
45
50
55
60
2020 2025 2030
EU-wide harmonised quota scheme
National tendering schemes with full market opening
National tendering schemes with partial market opening
National tendering schemes w/o market opening
National quota schemes
Levelised(15years)weightedaveragefinancialsupport
foryearlynewRES-Einstallations[€/MWhRES]
Results: Financial incentives …
Comparison of assessed policy options
Comparison of financial support
(premium to power price) for new
RES-E installations at EU 28 level over
time (2020 to 2030)
 Strong decline of the required
financial support for new RES
installations, but differences
between the policy variants can
be observed.
 Generally, the average support
is higher under a technology-
neutral scheme compared to
policy approaches that offer
incentives tailored to the
specific needs (… technology-
specific auctions).
Source: Green-X (2017)
National quotas
(without cross-
border trade)
Auctions with
partial market
opening

31. Results: Support expenditures for RES-e (1)
Future development of the resulting yearly
support expenditures for RES-E
over time in the EU 28
 As a general trend, one can see a
strong decline of support
expenditures over time: by 2030
support expenditures are slightly more
than one third of the starting value as
of 2020.
 This strong decline is caused by the
expected strong increases in fossil
fuel and carbon prices as well as by
the ongoing decline in cost for
renewables.
Source: Green-X (2017)
0
10
20
30
40
50
60
70
80
2020 2025 2030
EU-wide harmonised quota scheme
National tendering schemes with full market opening
National tendering schemes with partial market opening
National tendering schemes w/o market opening
National quota schemes
YearlysupportexpendituresforRES-electricity
[billion€]

32. Comparison of the required average (2021-2030) yearly
support expenditures for RES-E in the EU 28
 Bulk of support expenditures
in the forthcoming decade is
dedicated to existing RES
installations (built up to 2020):
only 5% to 11% of total RES-E
support in the forthcoming
decade will be for new
installations that will built in the
years 2021 to 2030.
 Clear preferences for feed-in
premium schemes where
support levels are determined
in an auction procedure in
comparison to quota schemes
with certificate trading.
 Best performing are auctions
where the allocation of RES
investments is done at a
multinational level (e.g. via
market opening) rather than a
pure nationally oriented one.Source: Green-X (2017)
Results: Support expenditures for RES-e (2)
47.5 47.5 47.5 47.5 47.5
4.6 2.7 3.0
6.1 3.7
0
10
20
30
40
50
60
EU-wide
harmonised quota
scheme
National tendering
schemes with full
market opening
National tendering
schemes with
partial market
opening
National quota
schemes
National tendering
schemes w/o
market opening
Average(2021-2030)yearlysupport
expendituresforRES-electricity[billion€]
Existing plants (installed until 2020) New plants (installed post 2020)
HarmQuo NatPol-GuideStateAid
Policy variants
using auctions

33. Conclusions (1)
Aiming for 27% RES by 2030:
• Half of the electricity produced will stem from renewables by 2030
• Required increase of RES strongly depends on energy efficiency /
future demand growth: with strong energy efficiency (30% demand
reduction) the required RES-E uptake is modest
• Moderate dedicated support for renewables is needed to reach the
2030 RES target

34. Conclusions (2)
A closer look at the required policy cost (support expenditures):
• Generally, strong decline of the required financial support for new RES
installations …
• … but differences between policy options can be observed: average
support is higher under a technology-neutral scheme compared to policy
approaches that offer incentives tailored to the specific needs
(e.g. technology-specific auctions).
In other words: Clear preferences for feed-in premium schemes where
support levels are determined in an auction procedure in comparison to
quota schemes with (technology-neutral) certificate trading
• Bulk of support expenditures in the forthcoming decade is dedicated to
existing RES installations: only 5% to 11% will be for new installations

35. Thank you!
Any questions?

36. Thank you!
Communication and web:
Michael Minter
Head of Secretariat
Kattesundet 4, 3rd floor
DK-1458 København K
mm@concito.dk
+ 45 26 16 64 14
Contact
Marijke Welisch
Researcher
Gusshausstr. 25-29
1040 Wien
welisch@eeg.tuwien.ac.at
+43-1-58801 370365
Project coordination:
Lena Kitzing
Researcher
Frederiksborgvej 399
DK-4000 Roskilde
lkit@dtu.dk
+45 24 65 90 64
Visit our website:
www.auresproject.eu
Follow AURES on Twitter
@AURESproject