Fulvio Fontini
DSE – University of Padua, Italy Eric Guerci
GREQAM – Université Aix-Marseille, France Angelica Gianfreda
´ Nuclear Power production in Italy: a “new” topic (?) fostered by
the recent government proposal to build new Nuclear Power Plants (l. 23/7/09 n. 99 and ss.) and “strengthened” by ENEL-EDF agreement.
´ Most of the debate focused on a) “traditional” pros and cons
(risk and security issues, costs, needs of incentives) and b) the “effective” possibility of building NPP in Italy.
´ A common argument brought forward in favor of NPP: It will
reduce the price of electricity (the “typical” reply: depends on the amount of the “extra” costs and the possibility to transfer them to buyers: electricity cost might lower, final price might be higher).
´ There seems to be a “supporters” attitude towards NPP.
´ But … no study yet evaluating the real impact of NPP on the
´ Evaluating in a “realistic” way the impact of a plausible
scenario about NPP on the electric price, taking into account the effective Italian power production structure: a liberalized market with significant transmission constraints and a large (former monopolistic) operator.
´ Realistic day-ahead market clearing mechanism (zonal market
splitting mechanism)
´ Equivalent 11 zones Italian transmission grid
´ Detailed agent-based model of the Italian thermal production
pool (158 power-plants).
´ Historical values for maximum grid transmission capacity,
price-inelastic zonal loads and fuel costs.
´ NPP assumptions:
A) 4 NP Plants (1.25GW each), 2 locations: 1) One Plant per zone in NO, CN, CS, S; 2) Two plants in NO, 2 in CN.
B) no strategic bidding (and get rid of the debate on the effective cost of combustibles): NPPs bid at zero price, maximum capacity.
( )
(
)
( )
[Euro/h]
[Euro/MWh]
[Euro/h]
i i l i i i i i l i i l iTC Q
FP
a Q
b
MC Q
FP a
QFC
FP b
=
⋅
⋅ +
=
⋅
=
⋅
Producer i bids:
( )
ˆ ,
ˆ
[Euro/MWh], [MWh]
(
)
i iP Q
subject to:
Q
i<
Q
ˆ
i<
Q
iand
P
ˆ
i<
P
*It is characterized by the following total cost function:
where FPl [Euro/GJ]
ˆ [MW] i i i Q < Q < Q , , , , [MW] [MW] l st l st l ts l ts Q Q Q Q ≤ ≤
Active power generation limits:
Active power balance equations for each zone k:
Real power flow limits of lines:
min P ˆ iQi i=1 N
∑
[MW], s.t. Qi i∈Z∑
- Qk,load = Qk, inject [MW]The solution consists of:
the set of active powers for producer i
the set of zonal prices for zones
*
i
Q
k
ZP
k
=
1,...,
K
The profit per hour for power plant i belonging to zone k is obtained as follows:
( )
* *
[Euro/h]
i k i i i
SELLERS
´ 5 thermal technologies (74% of the gross national production):
Coal-Fired (CF), Oil-Coal-Fired (OF), Combined Cycle (CC), Turbogas (TG) and Repower (RP).
´ 53 representative sellers (aggregating power-plants for each operator –
technology - zone). The profits are computed as the sum of the profits earned by each power-plant. They always commit themselves (no unit decommitment) Î no strategic behavior (different technologies within zone behaves as different agents even if belong to same owner)
´ Future extension (see later): allow for strategic bidding (capacity
´ 60 actions per seller (mark-ups, maximum capacity). Simple offers
submitted by sellers (almost 75% of the real market offers are simple)
´ Sellers learn according to a reinforcement learning algorithm (Roth and
Erev 98)
TRANSMISSION NETWORK
´ Equivalent 11 Italian zones (we do not consider foreign and virtual
zones) LOADS
´ Price-inelastic zonal loads faced by thermal power-plants (removed non
dispatchable, export and import and renewables-bilateral contracts of demand side)
, ( ) (1 ) , 1( ) , ( ) i t i i t i i t i f a = − ⋅r f − a + E a Propensities: Experimentation function: Profits:
( r: recency effect parameter)
( e: experimentation parameter) , , , 1 ( ) (1 ) ( ) ( ) 1 i t i i i t i i t i a e if i plays a E a e f a otherwise M − Π ⋅ − ⎧ ⎪ = ⎨ ⋅ ⎪⎩ − , , ( ) ( ) i t i i t i i R a a R Π =
Update probabilities:
Random draw, according to mixed strategy distribution, selects the action at
time (t+1) r e c d 0.6 0.97 0.04 0.035 0.05 where , i t S ( ) ( ) , , , ( ) i t i t i t i t i S a i t i S a a e a e λ λ σ =
∑
d t ct λ = −´ Updating it !
« Use 2009 data to take into account new power plants, grid extensions
(and reduced market power ?).
´ Assuming more realistic agents:
« The agent can own different technologies in the same zone « The agent can own power-plants in different zones
´ Increasing strategies’ space, allowing for capacity
withholding:
« bidimensional strategy space (price, quantity).
« Evaluating if NP increases MP of “dominant” player
P
SMP2 P