Market impact analysis of HVDC MWh charge

Market impact analysis of
HVDC MWh charge
October 2014
Objective
• Transpower has asked us to model the potential
impacts on the New Zealand wholesale electricity
market of changing the HVDC HAMI charge to a
MWh charge
•
•
HAMI charge = HVDC revenue requirement allocated to SI
injection customers on the basis of their historical anytime
maximum injection (HAMI). HAMI is the average of the 12 highest
injection quantities at each location for the current and previous
four years
MWh charge = HVDC revenue requirement allocated to SI
injection customers on the basis of annual energy injection. In this
analysis, the average annual energy injection over the last 5 years
is used.
Analysis summary I
• vSPD is used for the analysis to simulate the market dispatch
and prices over the 12 month post pole 3 period of Aug 13
to July 14.
• ~177MW of additional capacity was offered at Benmore,
Roxburgh and Clyde power stations to reflect potential
unoffered capacity that could be available in the market
as a consequence of removing the current HAMI charge.
• Considered impact on:
– System costs = estimate of fuel costs under the assumption of
cost-reflective offers (offer price x cleared offer quantity at
each injection location*)
– Load market costs = aggregate costs faced by loads in the
spot market (nodal price x quantity at each offtake location*)
– Generation revenue = aggregate revenue earned by
generators in the spot market (nodal price x quantity at each
injection location)
– Spot market nodal prices
*Offtake and injection quantities are based on the data in the final pricing case files.
Analysis summary II
• Two scenarios were considered in regard to the response of
SI generators to a MWh charge:
– Unchanged offer prices: Under this scenario it is assumed that
no changes are made to SI generator offer prices. The extra
capacity is offered at the highest tranche price.
– Adjusted offer prices: In this scenario it is assumed that SI
generators increase energy offer prices as a result of the MWh
charge.
• The analysis indicates a reduction in system costs of $17.4m
and $16.8m over the 12 month period for the unchanged
offer and adjusted offer scenarios respectively.
• The additional capacity displaces higher cost generation
which reduces system costs relative to the existing charging
approach. This is valid in both scenarios. The adjusted offer
price scenario does result in some higher cost generation
being dispatched, due to the increased offer prices,
however this is smaller than the reductions in system cost
due to the additional generation. Hence, the difference in
system costs between two scenarios is quite small.
Analysis summary III
• The analysis illustrates a reduction in peak spot
prices >10% (and average spot prices) for both
scenarios with an increase in off-peak spot prices
in the adjusted offer price scenario.
• With higher priced periods generally coinciding
with higher demand periods, the net impact is a
reduction of load costs in the spot market and
also generator revenue.
• The additional SI generation capacity and
reduced peak prices signal a reduced demand
for NI peaking generation. This would have flowon effects into the peak hedge market and peak
generation investment.
Analysis summary IV
•
•
The calculated reduction in system costs ~$17.4m is larger than the $3.5m over 3
years (~$1.2m per annum) estimated reduction calculated by the EA in its recent
work on HVDC problem definition. Potential reasons for the differences are:
– Differences in the assumptions on offering the additional SI generation
capacity (EA analysis assumes the additional capacity would be offered at
the maximum of an estimate of the long-term water value ($80/MWh) and the
maximum observed market offer price whereas we assumed the highest
energy offer tranche price at the generator was a reasonable estimate of the
generators’ willingness to sell given prevailing market conditions and portfolio
effects). This is applicable when comparing to the baseline scenario (B1).
– EA analysis covered both the pre and post pole 3 periods (Aug 11 to Jul 14)
whereas our analysis is concentrated on the post pole 3 period (Aug 13 to Jul
14). We suspect that the majority of the benefits are in the post pole 3 period
of the analysis.
– Treatment of frequency keeping constraints – in the EA analysis it appears
these constraints are removed whereas we retain these constraints (but adjust
with the additional capacity) in our analysis. Removing constraints would tend
to reduce system costs.
– We use a slightly different threshold for adjustment of offers to detect the
potential HAMI impact on unoffered generation (1% of HAMI vs the EA usage
of fixed thresholds).
Both studies are estimates which are based on assumptions of market behaviour
neither of which would be exact. We have undertaken a sensitivity using the same
offer price adjustment assumption as the EA analysis. This indicates a reduced
system cost impact of $4.2m. This is larger than the EA calculated value, but given
the other differences described above, not dissimilar.
Analysis summary V
•
•
•
•
•
We have undertaken sensitivity analysis to illustrate a potential range for the system
cost impact.
An upper bound sensitivity where the additional capacity is offered at the lowest
price, indicates a potential reduction in system costs of $55.6m and $55.1m over
the 12 month study horizon for the unchanged offer and adjusted offer scenarios
respectively. These sensitivities result in excessive SI hydro generation (an infeasible
hydro storage trajectory), hence are only a loose upper bound.
A sensitivity with an $80/MWh de-minimus price on the additional SI generation
capacity indicates a reduction in system costs of $4.2m and $3.6m over the 12
month study horizon for the unchanged offer and adjusted offer scenarios
respectively. There are periods where the $80/MWh applied threshold is larger than
the offer price of the remaining generation at the station (e.g. Benmore in mid Feb
2014 where all capacity was offered at < $1/MWh). This sensitivity results in the
lowest additional energy from SI generators and provides a lower range on the
reduction in system costs.
All scenarios and sensitivities indicate reduced system costs by releasing additional
SI generation capacity. Even where SI generators adjust offer prices in response to
a MWh charge there is still be an overall saving in system costs.
We note this is based on assumed offer scenarios and against the prevailing
market and hydrological conditions observed between Aug 13 to July 14.
Approach
• We simulated the potential market outcomes (MW
and prices) given:
– Removal of the current HVDC HAMI charge
– Introduction of a MWh charge
• The analysis utilised vSPD  simulating dispatch and
prices post pole 3 (01 August 13 to 31 July 14)*
• Recognise that estimating the impact is uncertain
given the uncertainty in participant’s response
• Analysed scenarios to estimate the potential
magnitude of the impact with variants on some key
assumptions
*vSPD does not have a representation of hydro storage  post-process calculations were undertaken
to estimate the impact of the changes in generation on hydro storage.
ESTIMATING UNOFFERED
GENERATION CAPACITY
Estimating unoffered generation
capacity
• Current HVDC HAMI deters additional SI generation
capacity above the existing HAMI level
• We used the current HAMI level and market offers
from Aug 13 to July 14 to estimate the potential
“unoffered” generation as a result of HAMI only
• If offered generation was within 1% of HAMI level
then assume HAMI charge is restricting the offered
energy. In these instances energy offers were
increased to the full offered capacity of the
generating station*
• Otherwise we leave offered energy unchanged
*Over period 03 Dec 13 to 10 Feb 14 the offered capacity at Benmore was ~ HAMI level but no corresponding
Benmore generator outages were observed in POCP. In these instances it was assumed that the HAMI was
restricting the offered capacity and the Benmore offered capacity was increased from ~474MW to 534MW.
Offer behaviour at Clyde
Planned outage
Implied HAMI limit
Offer behaviour at Roxburgh
Implied HAMI limit
Offer behaviour at Benmore
Additional capacity offered but high price for
quantities below HAMI as well  could not
attribute these high priced offers to HAMI
Implied HAMI limit
Unoffered generation capacity
assumed due to current HAMI
• It is estimated that the three most affected gridconnected generators identified (Benmore, Clyde
and Roxburgh) have a combined capability of
~177MW that is currently not being offered
• It is estimated that the impact on other SI gridconnected generators is smaller with a potential
~10MW of further grid-connected generation not
being offered due to the current HAMI charge.*
*Manapouri installed capacity = 850MW. Although offers from Manapouri are at HAMI
(795MW) the additional potential capacity from Manapouri is restricted to 800MW (due to
resource consents)
MWH CHARGE CALCULATION
Estimating the MWh charge
• HVDC MWh charge was calculated using:
– HVDC revenue requirement for 2014/15 pricing
year ($145m)
– Annual average output of SI generators over the
last 5 years+ (17,224GWh)
– HVDC MWh charge = $145m/17,224GWh =
$8.4/MWh
+Average over 5 years was chosen to reduce inter-year variability which could occur as a result of
wet and dry years
*Total SI generation is used  which includes some embedded generation  this would tend to
overestimate the energy injection and underestimate the MWh charge  an upper bound for this
underestimation is $0.5/MWh
ESTIMATING THE MARKET
IMPACT
Potential market response
•
•
•
•
•
Analysed two scenarios to understand the potential market impact
In both scenarios it is assumed that an additional ~177MW of SI generation
capacity is offered at Benmore, Roxburgh and Clyde. This additional capacity is
offered during times when the observed offered capacity was within 1% of the
current HAMI limit. SI generators were assumed to follow the same offer strategy
with the additional capacity priced at the maximum offer price of that
generator.
Baseline scenario 1 (Unchanged offer price): It is assumed that SI generators
already facing the HVDC charge have adjusted offer prices in response to the
existing charge and no further adjustments to offer prices are observed when
moving to a MWh charge.
Baseline scenario 2 (Adjusted offer price): It is assumed that SI generators
respond to the HVDC MWh charge by increasing offer prices to reflect the
increased variable cost of SI grid-connected generation through the MWh
charge.
In addition to these scenarios, two sensitivities were undertaken:
– Sensitivity a: Assuming the additional generation capacity is offered as the
lowest offer price tranch. This is an attempt to obtain a potential upper
bound of the effect.
– Sensitivity b: Assuming the additional generation is offered at the maximum
of $80/MWh (LRMC of new generation) or the highest priced offer tranch.
This is similar to the assumptions used in the EA’s analysis.
RESULTS
Price impact – Benmore (SI)
500 highest price periods
Peak prices reduce
in both scenarios
Increase in
minimum price with
MWh uplift
500 lowest price periods
•
•
Additional SI capacity reduces BEN peak spot market prices (in both
scenarios)
If HVDC MWh charge added to SI offer prices then SI minimum price
increases  overall still see a reduction in average prices
Price impact – Otahuhu (NI)
500 highest price periods
Peak prices reduce
in both scenarios
Increase in
minimum price with
MWh uplift
500 lowest price periods
•
•
Additional SI capacity reduces OTA peak spot market prices (in both
scenarios)
If HVDC MWh charge added to SI offer prices then OTA minimum
price increases  overall still see a reduction in average prices
Load during high and low priced
periods
Benmore
Otahuhu
Average
3234MW
Average
4972MW
Average
3676MW
Average
4890MW
• Expected demand during highest price
periods > expected demand during
lowest price periods
Impact on high spot prices
Benmore
Otahuhu
12.3%
11%
10.5%
12.6%
• Additional SI generation capacity
reduces high spot prices
• This could have flow-on effects into peak
hedge prices
Impact on low spot prices
Benmore
Otahuhu
• During lower spot price periods there is an increase in
price if SI generators increase energy offer prices in
response to the HVDC MWh charge
• Can increase costs to loads during periods where
prices would otherwise have been lower than the
MWh charge (low load and high storage (spill)
periods)
Impact on average spot prices
Otahuhu
Benmore
10%
6.7%
2.8%
2.8%
• If offer prices are increased, there are periods of
lower spot prices (due to additional capacity)
and periods of increased spot prices (due to
MWh uplift charge) relative to the base case.
The net impact is still a reduction (but not by as
much as peak prices)
Price impact summary
• Peak price impact:
– In both scenarios, a reduction in peak (99th percentile) spot
prices in both islands is observed.
– The increased HVDC capacity (post pole 3) does not
restrict the additional SI generation capacity and hence
both islands experience a reduction in peak prices
– This reduction in peak spot prices signals reduced demand
for peaking generation with flow-on effects into peaking
generation investment and peak hedge prices
• Off-peak price impact:
– If SI generators increase their energy offer prices in
response to the HVDC MWh charge, we observe an
increase in lower spot prices as the HVDC MWh charge
acts as a floor on SI spot prices.
– This also affects NI spot prices during times when the low
priced SI generators were setting the NI price.
– This can increase the costs to loads during lower priced
periods
Price impact summary
• Net price impact:
– We observed a greater reduction in high spot
prices than increase in low spot prices and
therefore observe an overall net reduction in
spot prices
• Net market impact:
– The reduction in higher spot prices is more likely
to occur during high demand periods and the
increase in lower spot prices more likely to occur
during lower demand periods.
– Hence in terms of overall market impact, we
observe an overall reduction in load market costs
and also a reduction in generation revenue in
both scenarios.
Impact on generation spot market
revenue and load spot market costs
If offer prices are not If offer prices are
changed
adjusted
•
•
•
Generator revenue ($m)
-217
-81.2
Load market costs ($m)
-213.7
-81.8
Net change ($m)
(Loss and constraint excess)
3.3
-0.6
The difference between load market costs and generation revenue is the loss and
constraint excess (LCE). An increase in the LCE implies a greater difference
between load and generation nodal prices which could be due to transmission
constraints and/or losses.
In the unchanged offer price scenario, the increased SI generation and
predominantly northward HVDC flow results in larger transmission losses and larger
nodal price differences between loads and generators, relative to the base case.
In the adjusted offer price scenario, there are periods of reduced LCE (lower HVDC
flow and smaller price differences) and periods of increased LCE (larger HVDC flow
and larger price differences) relative to the base case. The net impact over the 12
month study periods was a slight reduction in the LCE, relative to the base case.
Impact on system costs
If offer prices are not If offer prices are
changed
adjusted
System cost impact ($m)
-17.4
-16.8
• Assuming the observed market offers are cost reflective, the
change in cleared generation priced at the current market
offer provides an estimate of the impact on system costs.
• In both scenarios, an overall reduction in system costs is
observed as the additional capacity reduces the dispatch of
more expensive generation during high-priced periods. In the
case of the adjusted offer price scenario (where SI generator
offer prices were increased), this reduction in costs exceeds
the increased system cost due to using higher priced
generators during off-peak periods.
• The adjusted offer prices scenario has a lower reduction due
to more expensive generation being used in the NI as a result
of the uplift added on SI generators in this scenario.
High priced period analysis –
29 Jan 14 (trading period 28)
•
•
Scenario
OTA2201 ($/MWh)
BEN2201 ($/MWh)
Actual
1,864
1,512
Unchanged offer price
1,042
845
Adjusted offer price
1,042
845
In trading period 28 on 29 Jan 2014, actual spot price in NI and SI
exceeded $1,500/MWh
Estimated unoffered capacity:
–
–
–
•
•
Benmore = 60MW
Roxburgh = 0MW
Clyde = 0MW
Roxburgh and Clyde energy offers were > than 1% away from the HAMI
level  HAMI is not restricting the energy offer
Benmore offered energy was within 1% of HAMI  unoffered capacity =
offered capacity-offered energy (534MW – 474MW = 60MW)
29 Jan 14 – energy offers
Offered energy within 1% of HAMI 
additional energy offered = offered capacity
– offered energy (60MW) at highest-priced
offer tranche ($0.03/MWh)
Offered energy greater that 1% away from
HAMI  HAMI is not restricting the energy
offer  leave unchanged
29 Jan 14 (trading period 28) –
adjusted offers and island
scheduled generation
Benmore
With MWh
charge
45MW
60MW
40MW
•
Additional 60MW offered in SI @ $0.03/MWh  45MW of this additional
offered SI capacity was scheduled  displaced 40MW of more
expensive NI generation  reduces system costs and spot prices
Low-priced period analysis –
05 Jan 14 (trading period 7)
•
•
Scenario
OTA2201 ($/MWh)
BEN2201 ($/MWh)
Actual
0.01
0.01
Unchanged offer price
0.01
0.01
Adjusted offer price
8.73
8.66
In trading period 7 on 05 Jan 2014, actual spot price in NI and SI
was $0.01/MWh
Estimated unoffered capacity:
– Benmore = 60MW @ $975/MWh
– Roxburgh = 45MW (32MW @ $50/MWh and 13MW @ $33/MWh)
– Clyde = 69MW @ $60/MWh
•
•
Additional capacity offered at prices > clearing price 
additional capacity not scheduled
In adjusted offer price scenario adjusted offer prices sets floor
price in the SI  price in NI is greater due to transmission losses
05 Jan 14 – energy offers
Offered capacity at BEN = HAMI level but no
POCP outages and storage >> mean 
increased energy offers to capacity (534MW
– 474MW = 60MW)
Offered energy within 1% of HAMI 
additional energy offered = offered capacity
– offered energy (69MW at CYD and 45MW
at ROX) at highest-priced offer tranches at
each station respectively
05 Jan 14 (trading period7) – adjusted offers
Benmore
Clyde
69MW
60MW
With MWh
charge
With MWh
charge
Roxburgh 110kV
32MW
Roxburgh 220kV
13MW
05 Jan 14 (trading period 7) – island
scheduled generation
40MW
44MW
•
•
Additional MW offered in SI is not scheduled since offer price > market clearing price
Increase of SI energy offer prices (adjusted offer price) results in reduction in cheaper SI
generation (43MW) and increase in more expensive NI generation (40MW)  net
increase in system costs and system price
Generation impact summary
Change
relative
to actual
Reduction in aggregate SI generation
in adjusted offer price scenario –
displaced by lower-priced NI
generators
•
•
•
The additional generation capacity in the SI displaces NI generation 
includes NI thermal and NI hydro
If SI generators adjust offer prices to include an uplift to cover the HVDC
MWh charge, then we observe a smaller change in SI generation
A reduction in NI thermal generation is observed in both scenarios 
further illustrates that NI thermal peaking usage is being eroded with the
additional SI generation  during high prices SI generator uplift of energy
prices is less of an issue
Island generation impact results
• Additional SI generation capacity can displace
NI generation however this would be
dependent on the adjustment of SI generator
offer prices in response to the HVDC MWh
charge
• The percentage change in energy output
(relative to actual energy) is shown below 
energy impact of less than 1%
NI
SI
Unchanged offer prices
Adjusted offer prices
-0.74%
0.95%
-0.04%
0.08%
SI generation impact
• Largest increases are seen on the Waitaki
(Benmore) and Clutha (Roxburgh and
Clyde)  additional generation capacity
NI generation impact
•
•
•
Reduction in NI thermal generation in both scenarios  even with a
MWh charge uplift on SI generator offer prices
Waikato (WTO) hydro generation increases under scenarios with
increased SI hydro offer prices
Total SI generation increase > total NI generation reduction due to
increased transmission losses with larger proportion of SI generation
HVDC impact results
Corresponds to
a 4.4% increase
in HVDC
northward flow
over the year
•
•
Additional SI generation increases South-to-North (Northward) energy
transfer on the HVDC link under both scenarios, relative to the base
case.
Under the adjusted offer scenario  the reduced South-to-North and
increased North-to-South HVDC transfer, relative to the unchanged
offer price scenario is a result of higher-priced SI generation (with the
MWh charge) being displaced by some lower-priced NI generation.
APPENDIX
Sensitivity analysis
• To understand the potential impact range,
we considered variations in the adjustment
of energy offers.
• These adjustments were:
– Low price quantity: Under this sensitivity it was
assumed that the additional generation was
offered at the lowest observed offer price. The
intention of this sensitivity was to understand a
potential upper bound on the impact.
– Threshold price quantity: Under this sensitivity it
was assumed that the additional generation
was offered at the greater of $80/MWh* and
the highest offer price at the generator.
*$80/MWh is an estimate the long-run marginal cost of generation
Sensitivities
• Baseline scenario 1 (B1: Unchanged
offer price):
– B1a: Low price quantity
– B1b: Threshold price quantity
• Baseline scenario 2 (B2: Adjusted offer
price):
– B2a: Low price quantity
– B2b: Threshold price quantity
Impact on generation spot market revenue,
load spot market costs and system costs
Assessed but
Infeasible hydro
storage
CLU hydro storage feasibility
Low price quantity (a) sensitivities result
in excessive Clutha hydro usage 
infeasible hydro storage trajectory
(inc. Hawea, Wakatipu, Wanaka)
WTK hydro storage feasibility
• WTK storage within feasible range for all
the scenarios (Tekapo, Pukaki, Ohau)
Sensitivity summary
Low price quantity (a) sensitivities result
in excessive Clutha hydro usage 
infeasible hydro storage trajectory
•
•
•
The low price quantity resulted in excessive utilisation of the SI hydro resources  resulting in an
infeasible hydro storage trajectory  the results from these sensitivities, while potentially providing an
upper bound, is of limited use given its hydrological infeasibility.
The sensitivity with the $80/MWh de-minimus price on the additional SI generation capacity is on the
lower range of estimated impacts on system cost and market costs. Under this sensitivity there are
periods where the $80/MWh threshold is larger than the offer price of the remaining generation (e.g.
Benmore in mid February 2014 where all capacity was being offered at below $1/MWh). This
sensitivity results in lowest additional energy and at the lower range of the estimated impact.
Reduction in system dispatch costs is observed in all of the sensitivity scenarios even with an
adjustment of offer prices in response to the MWh charge. Under this base line scenario the
estimated reductions in system costs are with a lower range of -$3.6m, upper range range of $17.3m* and a mid-range of -$16.8m. (* system costs reductions could be higher however we have
not been able to robustly estimate an upper bound at this point. The upper bound is likely to be
between B1 ($17.3m) and B1a ($55.1m), which resulted in an infeasible hydro trajectory)
Results summary – system costs
Scenario
B1
B1a
B1b
B2
B2a
B2b
Metric
Reduction
Increase
Net effect
System costs ($m)
-17.4
0.0
-17.4
Half hours impacted (%)
57%
0%
System costs ($m)
-55.6
0.0
Half hours impacted (%)
100%
0%
System costs ($m)
-4.2
0.0
Half hours impacted (%)
26%
0%
System costs ($m)
-17.1
0.3
Half hours impacted (%)
59%
41%
System costs ($m)
-55.1
0.0
Half hours impacted (%)
95%
5%
System costs ($m)
-4.1
0.5
Half hours impacted (%)
32%
67%
-55.6
-4.2
-16.8
-55.1
-3.6
Results summary – load market
costs
Scenario
B1
B1a
B1b
B2
B2a
B2b
Metric
Reduction
Increase
Net effect
Load market costs ($m)
-213.8
0.1
-213.7
Half hours impacted (%)
52.0%
0.6%
Load market costs ($m)
-626.5
0.3
Half hours impacted (%)
92.3%
0.7%
Load market costs ($m)
-139.0
0.1
Half hours impacted (%)
24.6%
0.2%
Load market costs ($m)
-164.9
83.1
Half hours impacted (%)
34.9%
53.3%
Load market costs ($m)
-511.7
18.2
Half hours impacted (%)
80.4%
12.8%
Load market costs ($m)
-110.1
104.0
Half hours impacted (%)
16.7%
66.8%
-626.2
-138.9
-81.8
-493.5
-6.0