Post on 20-Jan-2016
Winter 2008 – a Meridian Perspective
Aug 2009
Winter 2008 – What Happened?
• Winter 2008 was another stressful supply-side period for the NZ electricity industry, primarily driven by very low inflows over the Nov-07 to Jun-08 period:
– NZ received 80% of average inflows (a ‘loss’ of 3,700GWh) over an 8 month period.
• Following on from the events of winter 2008 and in particular in the wake of the EC’s Winter Review there has been a great deal of public scrutiny and associated commentary as to what occurred and why.
• However the general quality of the debate has been poor, often contradictory, and largely anecdotal in nature:
– Market prices should have been higher
– Reservoirs were mismanaged
– Commercial objectives were pursued ahead of national interests
– Costs were unfairly avoided by generators and passed-on to consumers
• But as long as the debate and associated policy deliberations are conducted in an informed and impartial manner Meridian encourages a quality public discussion in answering:
“Is the electricity market delivering a fair deal for NZ?”
1
Winter 2008 – A Hind-casting Exercise
• What is an appropriate metric to use in measuring the actions of the market?– This metric should judge behaviour from the perspective of what is best for NZ.
• To examine the events of winter 2008 we use the Spectra model configured to reflect the fundamental costs underlying electricity supply and demand in NZ:
– Spectra is a SDDP model that balances the costs of excess thermal fuel burn against the costs of excess shortage in the face of uncertain hydro inflows to minimise the overall (supply) cost to NZ.
– This approach is consistent with the short-run marginal cost (SRMC) of the system.– This is NOT a re-litigation of Meridian’s operational decisions on the day – rather we are
using Spectra in the role of a proxy ‘regulatory benchmark’.• We update the model for events that characterise the Jul2007-Dec2008 period:
– Start storage conditions as at Jul2007– Inflows and wind– Geothermal and co-generation output– Thermal outages (POCP)– NI and SI demand– Link configuration
• All the model is now solving for is unknown future hydrology
• Beginning in Jul2007:– First: optimise the use of water in storage in the face of the historical hydrological
uncertainty – ie all weekly flows over the 1931-2008 period.– Second: run the model to simulate the 19 month Jul-2007 to Jan-2009 period for all
hydrological sequences.– Thirdly: examine the inflow ‘sequence’ that corresponds to the Jul07-Dec08 period.
• Now examine how the modelled benchmark results compare to market outcomes.
2
Winter 2008 – A Comparison of Storage Levels
Pukaki & Tekapo Hydro Storage:Hindcasting versus Actual
-
500
1,000
1,500
2,000
2,500
Jun0
7
Jul0
7
Aug
07
Sep
07
Oct
07
Nov
07
Dec
07
Jan0
8
Feb
08
Mar
08
Apr
08
May
08
Jun0
8
Jul0
8
Aug
08
Sep
08
Oct
08
Nov
08
Dec
08
Jan0
9
week
Wai
taki
sto
rag
e [G
Wh
]
0
500
1,000
1,500
2,000
2,500
sto
rag
e [G
Wh
]
PKI+TEK Historical Average PKI+TEK Hindcast PKI+TEK actual
Lowest inflows on record over Nov07-Jun08:- 80% (3,500GWh) below average NZ inflows
- 77% (2,200GWh) below average MEL inflows
Lowest inflows on record over Nov07-Jun08:-20% (3,500GWh) below average NZ inflows-23% (2,200GWh) below average MEL inflows
3
Winter 2008 – A Comparison of Market Prices
Wholesale Market Prices:Hindcasting versus Actual
$-
$50
$100
$150
$200
$250
$300
$350
$400
Jun0
7
Jul0
7
Aug
07
Sep
07
Oct
07
Nov
07
Dec
07
Jan0
8
Feb
08
Mar
08
Apr
08
May
08
Jun0
8
Jul0
8
Aug
08
Sep
08
Oct
08
Nov
08
Dec
08
Jan0
9
week
bas
elo
ad p
rice
[$/
MW
h]
$-
$50
$100
$150
$200
$250
$300
$350
$400
NI Hindcast SI Hindcast HLY2201 actual BEN2201 actual
4
Winter 2008 – Comparison of Market Outcomes• Market outcomes over the Jul-2007 to Jan-2009
period are very similar to those suggested by the model:
– Market prices, generation, DC transfers, demand response, reservoir spill, and reservoir management all track very closely with what actually occurred.
• Answering some specific questions:– Should market prices have been higher?
• Marginally perhaps but in general no.– Did Meridian draw down its reservoirs to protect its
own commercial position?• No – lake levels largely followed inflows.
– Did the thermals behave themselves?• In aggregate yes.
– Was security of supply threatened?• No – or no more than it would have been under ECNZ.
– Did Meridian ‘pass on’ high prices to consumers?• No – MEL lost $125M over the winter 2008 period.
• Market outcomes in winter 2008 were determined almost entirely by environmental factors and NOT by undesirable behaviours.
– It’s all about the inflows!– The market was a pretty good (but not perfect)
reflection of a centrally controlled benchmark• While we can successfully explain (and defend)
what happened in winter 2008 this does not answer the bigger question of whether or not this is an outcome that NZ is happy with in the future.
SPECTRA HindcastJul2007 to Jan2009
Spectra Market DiffPrices SI 99.5$ 99.4$ 0.1-$
NI 83.7$ 86.1$ 2.4$ Generation Waitaki 10,905 10,586 319-
Manapouri 7,720 7,536 184- MEL Wind 798 791 7- Taupo 6,702 6,652 50- Clutha 5,368 5,519 151 Geothermal 5,497 5,476 21- Co-Gen 2,400 2,387 13- Huntly 7,533 6,162 1,371- CCGTs 12,108 13,455 1,347 OCGTs 282 580 298
TOTAL Hydro 30,695 30,293 402- TOTAL Thermal 19,923 20,197 274
Dem Res NI 63 210 147 SI 194 183 11-
HVDC S->N 2,876 2,426 451- N->S 1,376- 1,250- 127
Spill Waitaki 402 337 64- Clutha 10 - 10- Taupo - - -
Storage Waitaki Min 526 469 57- Waitaki Avg 1,262 1,221 41- Hawea Min 83 5 78- Hawea Avg 143 107 36- Taupo Min 78 78 1 Taupo Avg 330 349 20
Fuel Burn Thermals 925.6$ 994.0$ 68.4$
5
2000-2008 The Wolak Years
2000-2008 Hindcasting: A Comparison of Waitaki Storage
Pukaki & Tekapo Hydro Storage:Hindcasting versus Actual
-
500
1,000
1,500
2,000
2,500
Jun00 Dec00 Jun01 Dec01 Jun02 Dec02 Jun03 Dec03 Jun04 Dec04 Jun05 Dec05 Jun06 Dec06 Jun07 Dec07 Jun08 Dec08
sto
rag
e [
GW
h]
0
500
1,000
1,500
2,000
2,500
sto
rag
e [
GW
h]
PKI+TEK Hindcast PKI+TEK actual PKI+TEK Historical Average
Hyd Seq Beginning: 2000-2001
6
2000-2008 Hindcasting: A Comparison of Market Prices
Wholesale Market Prices:Hindcasting versus Actual
$-
$50
$100
$150
$200
$250
$300
$350
$400
Jun00 Dec00 Jun01 Dec01 Jun02 Dec02 Jun03 Dec03 Jun04 Dec04 Jun05 Dec05 Jun06 Dec06 Jun07 Dec07 Jun08 Dec08
week
bas
elo
ad p
rice
[$/
MW
h]
$-
$50
$100
$150
$200
$250
$300
$350
$400
NI Hindcast SI Hindcast HLY2201 actual BEN2201 actual
Hyd Seq Beginning: 2000-2001
7
SRMC Sensitivity Results• The definition of thermal SRMC makes a
large difference to the conclusion that market prices have prevailingly been higher than they should have been.
• An alternative thermal SRMC definition can lead to entirely the opposite conclusion.
• The definition of hydro SRMC makes a lesser but still significant difference to the conclusion that market prices have prevailingly been higher than they should.
• Physical storage outcomes are robust to any reasonable definition of hydro and thermal SRMC.
• The lack of internal consistency in regards physical and pricing outcomes can be seen in Wolak’s approach when security of supply is severally compromised.
• These sensitivities should not be interpreted as “Wolak may be right” rather they should be seen as:
– A demonstration of “garbage in garbage out”.
– A demonstration of the inherent noise in the estimation of fair market outcomes.
Pukaki & Tekapo Hydro Storage: Hindcasting versus Actual
-
500
1,000
1,500
2,000
2,500
Jan07 Jul07 Jan08 Jul08 Jan09
sto
rag
e [
GW
h]
0
500
1,000
1,500
2,000
2,500
sto
rag
e [
GW
h]
Market Hindcast NPL Offer MWHindcast Naïve SRMCNaïve SRMC, NO VoLL Hindcast Coal StockpileNaïve SRMC Coal Stockpile Naïve SRMC Full Thermal Storage
Excess Market Rentals:Jan2001 to Jul2007
-$6,000
-$4,000
-$2,000
$-
$2,000
$4,000
$6,000
Wolak Hindcast HindcastNPL Offer
MW
HindcastCoal
Stockpile
NaïveSRMC
NaïveSRMCCoal
Stockpile
NaïveSRMC
FullThermalStorage
NaïveSRMC +No Voll
thermal fuel representation
ren
tal [
$M
]
Voll@5K Voll @10K Voll @20K
8
Additional Winter 2008 Material
Winter 2008 – Some Sensitivities• We examine a range of “what-if’ outcomes over the 19 month period via
sensitivities to the baseline hind-casting assumptions.
• Leaving aside issues related to the pricing of SI reserves, the HVDC has only mild impacts:
– Storage trajectories over 2007-2008 are largely unaffected by changes in either assumed link size or additional water in storage at the start of the period leading up to the HVDC decommissioning in Spring 2007.
– However a larger link would have driven lower market prices and reduced levels of spill in early 2009.
• National shortage costs can have large impacts on all aspects of system operation:– 2008 appeared to operate to somewhere between a $2,500 and $5,000/MWh VoLL.– A higher VoLL could have significantly increased minimum reservoir levels (by up to
500GWh) BUT at the expensive of an enormous increase in marginal market prices as well as large increases in reservoir spill.
– Given that no actual shortage occurred in 2008 it is questionable whether these additional costs are in NZ’s best interests.
• Thermal behaviours have large pricing impacts:– More benevolent thermal behaviour over the 2007-2008 period would not have significantly
changed any of the physical outcomes, but could have made the delivery of the same physical outcome significantly cheaper.
• Dry inflow assumptions mimic higher VoLL assumptions:– Always assuming drier operating conditions for reservoir management has clear impacts for
2007-2008 and would have raised minimum reservoir levels.– However taken to extremes, while large increases in storage levels could be achieved this
comes at the expense of enormous increases in marginal market prices and levels of reservoir spill.
9
Change in Hindcast Market Prices
-$50
-$40
-$30
-$20
-$10
$-
$10
$20
$30
$40
$50
Big
Lin
k (5
00S
N, 1
000N
S)
+200
GW
h E
xtra
Sta
rt S
tora
ge
+400
GW
h E
xtra
Sta
rt S
tora
ge
Vol
l $2,
500
Vol
l $10
,000
Vol
l $20
,000
Vol
l $10
0,00
0
Naï
ve S
RM
C
300M
W N
PL
P.S
.
No
The
rmal
Out
ages
Neg
ativ
e IP
O
Pla
n fo
r D
ry O
nly
sensitivity
chan
ge
in m
ean
pri
ce [
$/M
Wh
]
Prices SI
Prices NI
Jul2007 to Jan2009SI: $104/MWh
NI: $86/MWh
Change in Hindcast Minimum Storage Levels
-200
-100
-
100
200
300
400
500
600
700
Big
Lin
k (5
00S
N, 1
000N
S)
+200
GW
h E
xtra
Sta
rt S
tora
ge
+400
GW
h E
xtra
Sta
rt S
tora
ge
Vol
l $2,
500
Vol
l $10
,000
Vol
l $20
,000
Vol
l $10
0,00
0
Naï
ve S
RM
C
300M
W N
PL
P.S
.
No
The
rmal
Out
ages
Neg
ativ
e IP
O
Pla
n fo
r D
ry O
nly
sensitivity
chan
ge
in m
in s
tora
ge
[GW
h]
Waitaki Min
Taupo Min
Jul2007 to Jan2009
Waitaki Min: 628GWh
Taupo Min: 77GWh
Change in Hindcast Hydro Spill
-400
-200
-
200
400
600
800
Big
Lin
k (5
00S
N, 1
000N
S)
+200
GW
h E
xtra
Sta
rt S
tora
ge
+400
GW
h E
xtra
Sta
rt S
tora
ge
Vol
l $2,
500
Vol
l $10
,000
Vol
l $20
,000
Vol
l $10
0,00
0
Naï
ve S
RM
C
300M
W N
PL
P.S
.
No
The
rmal
Out
ages
Neg
ativ
e IP
O
Pla
n fo
r D
ry O
nly
sensitivity
ch
an
ge
in s
pill
[G
Wh
]
Waitaki
Clutha
Taupo
Jul2007 to Jan2009
Waitaki: 487GWh
Clutha: 8GWh
Taupo: 0GWh
Change in Hindcast Thermal/Hydro Generation
-1,000
-800
-600
-400
-200
-
200
400
600
800
Big
Lin
k (5
00S
N, 1
000N
S)
+200
GW
h E
xtra
Sta
rt S
tora
ge
+400
GW
h E
xtra
Sta
rt S
tora
ge
Vol
l $2,
500
Vol
l $10
,000
Vol
l $20
,000
Vol
l $10
0,00
0
Naï
ve S
RM
C
300M
W N
PL
P.S
.
No
The
rmal
Out
ages
Neg
ativ
e IP
O
Pla
n fo
r D
ry O
nly
sensitivity
ch
an
ge
in g
en
era
tio
n [
GW
h]
-$75
-$55
-$35
-$15
$5
$25
$45
ch
an
ge
in f
ue
l-b
urn
[$
M]
TOTAL Hydro
TOTAL Thermal
Thermals
Jul2007 to Jan2009
TOTAL Hydro: 30,604GWh
TOTAL Thermal: 19,977GWh
Thermals: $929M
10