The new plant was financed with a massive US$737 million loan guarantee2http://www.energy.gov/lpo/crescent-dunes. With a design output expected to average just 55 MW (or 482,000 MWhours per year), it confirms that the levelised cost (LCOE) of CSP is very much more than that of other mainstream renewable energy technologies such as wind and solar photovoltaics (PV). Depending on the wind resource, a similar output from wind would require an installed capacity of between 130-200 MW.
The EIA3Annual Energy Outlook 2015 With Projections to 2040, U.S. Energy Information Administration estimates total system LCOE for CSP, PV and wind as US$240 per MWhour, US$120 per MWhour and US$74 per MWhour, respectively (in $2013 terms). At these prices, the equivalent output from wind would require an investment of $200-300 million, meaning deployment of CSP plant will be expected to continue to lag PV and wind significantly4 in installed capacity terms CSP lags PV and wind by one and two decades, respectively, as per Keith Lovegrove presentation. With CSP learnings commensurately slower than the competing renewable technologies, the concern amongst advocates is that CSP will continue to lose ground in the renewables push, and the opportunity to develop sufficient scale to drive down costs will be missed.
However, LCOE may not be the best comparator, since compared to wind and PV, CSP with storage, affords dispatchability and the opportunity to exploit daily variability in wholesale prices and/or price-cap contracting.
In exploiting such variability, the costs cited above suggest CSP dispatch would need to garner around $100 per MWhour more than wind to be competitive.
In energy only markets, like the NEM, price volatility is expected to correlate with penetration of intermittent renewables. The effect is two sided, with high intermittent output depressing prices and prices rising at times of diminished intermittent output. This second effect can be exacerbated if dispatchable plant is withdrawn to accommodate the growing renewable capacity. A consequence is the average price taken by intermittent generation fall as deployment grows, while the prices taken by plant able to take advantage of price fluctation will tend to increase.
The impact for CSP is dependant on the inherent volatility of the market, itself a reflection of the generation capacity and mix, and demand profile. Relevant questions for CSP and other proponents seeking to exploit market price volatility include
what is the optimal plant design (storage versus power) and capacity for a given market structure?
the best opportunities for large-scale storage in the NEM are in the South Australian region (SA1) and
historical prices indicate little the optimals capacity is less than 6 hours of storag.
This contribution summarises some of the sepcific characteristics of the South Australian market (SA1), noting that imminent changes in generation mix, including the withdrawal of the last remaining coal plant in SA1, is likely to substantially increase volatility.
South Australian (SA1) region
The figures below show a recent week-long period of dispatch on SA1, coloured by the three contributing generator fuel types; namely wind, natural gas and brown coal.
For the period shown, SA1 brown coal accounted for
34% of the total dispatch, at average output of 443 megawatts, while wind accounted for 37% and gas 29%.
(above) Dispatch on SA1 for a recent week in 2016 by fuel type and power station. Note while stations that averaged less than 30 megawatts for the period are not shown, their dispatch is included in the fuel type totals.
(left) Breakdown of SA1 dispatch fuel type in megawatts by 5 minute intervals over a week in February, 2016.
SA1 is connected to the NEM via the Murraylink and Heywood interconnectors, both linking into VIC1. Murraylink has a nominal transfer capability of 220 MW, while Heywood currently has a nominal 480 MW transfer capability, with an upgrade to 650 MW scheduled to be completed by 31 July 2016.6http://www.aemo.com.au/Electricity/Planning/South-Australian-Advisory-Functions/~/media/Files/Other/planning/The20Heywood20Interconnector20UpgradeUpdate202015.ashx The upgrade is motivated partly by the recent (Pelican Point Power Station Unit 2) and proposed (Torrens Island Power Station A, Northern Power Station and Playford B Power Station all by 2017) plant withdrawals that will reduce SA1 capacity in coal and gas generators by more than 1500 MW. On average, power flows from VIC1 to SA1 at a rate of about 100-200 MW, though flow direction oscillates, largely depending on the output of SA1 wind power generators. Over the week shown in the figure below, the average flow into SA1 was 188 MW, implying around 12.6% of local SA1 demand was sourced from VIC1. 7 With VIC1 dispatch dominated by Latrobe Valley brown coal generators, the imports imply the CO2 intensity of SA1 power consumption is in effect rather higher than indicated by local dipsatch alone.
(left) power exchange in megawatts between SA1 and VIC1, for the week in 2016 shown in the previous figure. Negative values indicate flow direction is from VIC1 into SA1, while positive values indicate flows from SA1 into VIC1.
Several features distinguish SA1 from the four other regions paricipatin on the NEM.
SA1 has the highest proportional daily variability in demand, due to relatively low industrial load.
SA1 has the highest proportion of wind generation on the NEM, and one of the highest for any competitive energy-only markets in developed economies.
SA1 typically has the highest wholesale prices.
Of particular relevance, is the imminent withdrawal of the Northern Power station, the one remaining SA1 coal generator8 Alinta energy initially scheduled the closure of its Northern Power station for January 31st, 2016, but is reported to have extended this to May 8th.. Northern’ has’s a registered capacity of 530 MW (maximum capacity is 546 MW) represents more than 40% of typical SA1 demand.
(left) Volume-weighted wholesale prices (VWP) in $ per MWhour, list as annual means (for calendar years 2006-2015) as well as the decadal mean (MEAN) for the five participating regions on the NEM.
REGION
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
MEAN
NSW
34.8
76.0
42.1
51.7
33.9
45.4
43.7
54.8
43.7
40.9
46.8
QLD
28.3
72.7
48.8
37.4
28.1
38.0
44.0
71.8
56.2
58.0
48.2
SA
44.7
64.9
92.7
89.9
55.1
46.9
46.5
78.2
54.3
55.6
63.1
TAS
37.7
59.0
50.7
53.8
32.9
30.0
42.3
46.6
39.2
47.5
44.0
VIC
37.7
69.6
43.4
43.7
39.9
31.5
46.8
54.9
44.9
35.8
44.9
SA1 has had the highest annual volume weighted prices in 6 out of the past 10 years, and second highest for 3 others. At $63 per MWhour, the SA1 decadal average is 40% (or $18 per MWhour) higher than the interconnected region of VIC1, $16 per MWhour higher than NSW1 and $15 per MWhour higher than QLD1. In recent years, QLD prcies have increased signficantly, and now more or less match SA1.
Daily price variations
For plant with limited storage, revenue is maximised by dispatch across the optimal set of interval(s) ordered in terms of price, until available storage is depleted. Dispatch then resumes after the next charging. For CSP, the charging occurs on a daily basis and do the relevant optimal dispatch set is a fixed proportion of each day. For a CSP device with X hours storage, the optimal dispatch set set maximises revenue across the best X hours intervals ordered by price. Of course, dispatch into these intervals requires perfect pricing foresight.
The figures below show the average SA1 prices for the optimal dispatch set for the best X hours of each day, for X = 6 and 3 hours, respectively) for the years 2006-2015 inclusive. Averages are across the full calendar year.
SA pool price for the averaged, best 6 hours of the day, for each of the last 10 years. Triangles represent average yearly prices, bars represent median yearly prices,small dots represent daily price event outliers. Averages are taken over the full calendar year. Note y-axis is log10 scaling
as for above but for best 3 hours of the day averaged over full calendar year.
The figures below show the average SA1 prices for the optimal dispatch set for the best 6 hours of each day for the years 2006-2015 inclusive, averaged by summer and winter season, respectively.
as for above but for three summer months only
as for above but for three winter months only
The figures below show the average price for the optimal dispatch set as a function of the best X daily hour- increments, for various multi-year time periods, for each of the four mainland regions participating in the NEM.
hours
1
2
3
4
5
6
8
10
12
15
18
24
NSW
136.3
108.7
93.5
83.3
76.0
70.7
63.4
58.5
54.9
51.0
48.0
43.0
QLD
184.0
138.9
115.9
101.1
90.7
83.0
72.5
65.6
60.6
55.2
51.1
44.6
SA
206.3
164.7
141.0
124.1
111.4
101.7
87.9
78.9
72.6
65.6
60.3
51.4
VIC
120.7
98.4
85.8
77.4
71.4
67.0
60.7
56.4
53.2
49.5
46.5
41.1
The figures show that over the decade 2006-2015, SA1 has had the higher prices across all best X daily hour-increment levels compared to otehr regions, averaging above $100 per MWhour for the best 6 daily hour-increments, and $140 MWhours for the best 3 daily hour-increments.
Considerable variability is evident over shorter time periods. For example, over the last five years (2011-2015), QLD1 and SA1 had similar averaged prices, in the mid $80’s per MWhour for the best 6 hour daily increments, rising to $140 MWhours for the best 3 hours.
hours
1
2
3
4
5
6
8
10
12
15
18
24
NSW
92.1
77.5
70.0
64.8
61.1
58.4
54.7
52.2
50.4
48.3
46.6
43.6
QLD
183.2
139.9
118.5
104.2
94.2
86.8
76.7
70.0
65.2
60.0
56.1
49.8
SA
169.9
126.5
107.7
97.1
89.3
83.3
75.0
69.4
65.3
60.7
57.0
50.2
VIC
87.2
73.1
65.1
60.5
57.4
55.1
51.8
49.6
47.9
45.8
44.0
40.4
Over the last 2 years (2014-2015), QLD1 has averaged consistently higher prices than SA1.
hours
1
2
3
4
5
6
8
10
12
15
18
24
NSW
84.2
68.3
61.7
57.9
55.2
53.3
50.5
48.6
47.1
45.4
43.9
41.0
QLD
238.8
178.9
148.8
128.0
113.7
102.9
88.1
78.4
71.6
64.4
59.2
51.4
SA
193.3
135.5
111.1
97.8
89.1
83.0
74.5
68.9
64.6
59.8
55.8
48.9
VIC
76.8
66.1
60.3
56.7
54.1
52.2
49.4
47.3
45.6
43.5
41.7
37.7
Across the calendar years 2008-2009, SA1 prices averaged $167 per MWhour for the best 6 hours, and over $250 per MWhour for the best 3 hours. This period coincides with the final part of the Millenium Drought. Druing the dorught, supply was constrained across the NEM in part because of below average hydro output and in part because of restricted availability of thermal plant cooling water. While the specific causes of the high 2008-2009 prices in SA1 are not investigated here, the anomalies are noted to the extsnt that they may provide an insight into the price impact of the increases volatility expected to accompany SA1 brown coal withdrawal.
hours
1
2
3
4
5
6
8
10
12
15
18
24
NSW
167.8
137.4
118.5
104.1
92.9
84.5
72.8
64.9
59.3
53.2
48.6
41.5
QLD
184.3
139.6
116.0
101.1
90.0
81.6
69.6
61.9
56.4
50.4
46.0
39.0
SA
333.2
295.7
255.6
217.5
188.4
167.5
136.2
116.4
102.9
88.6
78.5
63.4
VIC
112.5
93.1
83.8
77.5
72.5
68.5
62.0
56.9
53.1
48.6
45.0
38.4
Seasonal variation in prices are also evident as a consequence of the seasonal variations in demand, which show peaks in both summer and winter seasons.
hours
1
2
3
4
5
6
8
10
12
15
18
24
NSW
195.8
160.1
136.7
118.3
104.6
94.5
80.9
71.9
65.6
58.8
53.8
46.5
QLD
339.8
257.2
212.7
180.6
156.7
139.0
115.0
99.6
88.8
77.4
69.3
57.5
SA
310.4
258.7
225.8
199.5
176.4
157.6
129.6
111.8
99.3
86.0
76.5
62.8
VIC
195.3
154.5
130.8
114.0
101.9
93.0
80.0
71.1
64.8
57.9
52.7
44.8
hours
1
2
3
4
5
6
8
10
12
15
18
24
NSW
173.6
130.0
106.4
93.1
84.5
78.4
70.1
64.6
60.7
56.2
52.9
47.2
QLD
168.5
119.9
97.7
85.4
77.5
71.9
64.3
59.2
55.5
51.4
48.3
42.8
SA
174.3
127.7
107.9
96.6
89.1
83.7
75.8
70.4
66.4
61.8
58.1
50.4
VIC
122.0
97.0
83.9
76.3
71.2
67.4
62.1
58.5
55.7
52.5
49.7
44.2
(left) the number of half hour best price intervals - designated in the legend as count - over the year, by the time of day, that fall into the best daily X hour price events, where is X is either 6 or 3 hours.
Note that the distribution of best price intervals has become more diffuse with time, especially for Summer months when it has shifted slightly later in the afternoon, from 3.30-4:30 pm (NEM-time) in 2005-2006 to 4:30-5:30 pm in 2014-2015. In winter the best time interval is between 6:30 - 7:30 pm (NEM-time), with a subordinate peak at 8:30-10:00 am. There has been a substantial hollowing out in middle of the day best price intervals from 2011 on, attributable to the impact of domestic PV on demand. Note that NEM-times are equivalent to AEST.
Prices for wind generation
Of relevance to the relative costs of non-dispatchable plant is the comparison of the wholesale prices taken by SA1 wind with SA1 gas and other fuels. In 2015, the volume-weighted price taken by SA1 wind was $38 per MWhour averged by month, some $16 lower than the average for all SA1 dispatch, and $26 lower than SA1 gas.
(above) 2015 Monthly volume-weighted prices in SA1, as taken by generation categorised by fuel-type. (left) 2015 calendar year averages of monthly volume-weighted prices taken by generation categorised by fuel-type, across the five partcipating NEM-regions. VWP represents the average volume-weighted price for the region.
NSW
QLD
SA
VIC
TAS
Black Coal
40.7
53.4
-
-
-
Brown Coal
-
-
49.7
34.1
-
Hydro
58.6
74.3
-
44.5
49
Natural Gas
44.2
59.6
63.5
49.5
68.1
Wind
38.3
-
37.6
31.4
45.9
VWP
41.1
56.8
53.8
35.8
48.1
The impact of intermittent generation in the SA1 marketis readily shown by comparing the price wind generation takes compared to that of natural gas. the prices are expected to diverge as widn penetration icnreases, because in the SA1 market natural gas effectively provides the complimemntary firming capacity when wind dispatch is limited. The plots below shows monthly volume weighted prices for SA1 and for NSW1 for the period 2011-2015, for individual fuels relative to the Regional Reference Price (RRP), and as a ratio of the natural gas and wind prices. For SA1, where wind, penetration can account for well over 50% of local dispatch, the price taken by gas averaged 1.6 times that of wind in 2015, equateing to a difference of about $25 per meagwatt hour. The ratio has steadily increased over the 5 years 2011-2015, though seasonal variation is evident in the high ratios in peak summer and peak months. Note the effect is much more subdued in NSW1 because there the penetration of wind is much smaller, and so the ability of wind to depress wholesale prices is subtantially diminshed relative to SA1.
Price-cap contracts
Price-caps area a common contracting mechanism to limit liability to extreme price events. The figures below show how a price caps impacts the best # hourly increment prices across the full calendar year for the last decade. The effect is significant.9 note, these figures show average prices, not volume-weighted prices. They are relavant to the price taken by a plant dispatching at a specified constant power into the best X-hour daily increments across the time period, for the given price cap. With a price cap of $500 per megawatt hour, the value of the wholesale revenue for the best 6 hour daily prices in SA1 is reduced by around $40 per MWhour (or 40%). That could be effectively offset by a contract of $10 per MWhour applying 24 hours a day.
hours
1
2
3
4
5
6
8
10
12
15
18
24
NSW
78.1
69.7
64.3
60.6
57.8
55.5
52.0
49.4
47.4
44.9
42.9
39.2
QLD
103.7
85.8
76.1
69.9
65.3
61.7
56.5
52.8
50.0
46.7
44.0
39.2
SA
105.7
89.0
80.1
74.3
70.0
66.7
61.6
57.9
55.0
51.6
48.6
42.7
VIC
78.4
69.6
64.0
60.1
57.2
55.0
51.6
49.1
47.1
44.6
42.4
38.0
hours
1
2
3
4
5
6
8
10
12
15
18
24
NSW
70.3
64.1
60.1
57.2
54.9
53.1
50.2
47.9
46.1
44.0
42.1
38.6
QLD
85.5
74.0
67.4
62.9
59.5
56.8
52.8
49.8
47.5
44.7
42.4
38.0
SA
87.3
77.0
71.2
67.1
64.1
61.6
57.8
54.8
52.5
49.5
46.9
41.4
VIC
70.1
63.9
59.7
56.7
54.5
52.7
49.8
47.7
45.9
43.7
41.7
37.4
hours
1
2
3
4
5
6
8
10
12
15
18
24
NSW
66.2
61.1
57.7
55.3
53.3
51.7
49.1
47.1
45.4
43.4
41.7
38.2
QLD
76.8
68.3
63.1
59.5
56.7
54.4
50.9
48.3
46.2
43.7
41.5
37.3
SA
78.8
71.5
67.0
63.8
61.3
59.2
56.0
53.4
51.3
48.6
46.1
40.8
VIC
66.1
60.9
57.5
55.0
53.1
51.5
48.9
46.9
45.3
43.2
41.3
37.1
Summary and other considerations
In summary, the experience over the last decade suggests that in SA1 a device with 6 hours of storage with a dispatch strategy of perfect foresight dispatching on a daily basis could expect to achieve wholesale prices of ~$100 per MWhour. For the best 3 hours the averaged wholesale price in SA1 rises to ~$140 per MWhour. Dispacth into the best 6 hours yields prices twice the average regional volume-weighted price (~$50 per MWhour).
Over the past decade, average SA1 prices for the best 6 hour daily intervals have been about 25% above QLD1 and 40% above NSW1 and VIC1. While QLD1 has had the highest prices in the pat 2 years, increases in SA1 price volatility will likely substantially icnrease with the Northern Power Station decomissioning in the next few months. If years 2007-2008 are any guide, then SA1 best 6 hours prices could rise by as much as $50-100 per MWhour, to above $150 per MWhour.
With relevant Renewable Energy Certificate prices10http://greenenergytrading.com.au/resources/certificates-prices currently at ~$80 per MWhour attaining average revenues of around $200 per MWhour for renewable dispatch into the best 6 hours a day in SA1 would seem plausible. In wholesale market terms, that would yield around $40 million for a 100 MW plant. In addition, such a plant could earn up to $10 million per year with capacity contracts of above $10 per MWhour, though depending on the contract, this would be largely offset by the reduction in wholesale revenue (by about ~40% for the best 6 hours daily dispatch into SA1).
CSP also gains relative price advantage over wind, because the price taken by wind is now almost $20 per MWhour below the SA1 average. As a consequence, CSP with 6 hours storage and perfect foresight could be expected to take as much as $70 per MWhour more than wind. Given the current costs estimates, the differential afforded by dispatchability is probably not yet sufficient to warrant deployment of CSP over wind and PV on fincnacinal terms, though it is not far short. The financial vability of CSP would be improved by
increases in volatilty, or
other price incentives that specifically advantage dispacthable renewables, such as time differentiated RECs11 mooted by Keith Lovegrove.
With increases in volatility anticipated once the Northern Power station is decomissioned, CSP is likley to be viable sooner rather than later.
Some other considerations for further analysis.
Seasonal variations in dispatch need to be better modelled. For example, additional revenue can be gained by CSP device if full charge can be attained prior to sundown.
Need to estimate how much extra volatility when coal exits the SA market, noting mooted upgrades on SA1-VIC1 interconnectors?
With perfect pricing foresight unrealistic, need to estimate the price impact of near to perfect foresight.
Finally, how much storage could SA1 withstand before volatility is significantly suppressed.
Data Sources
Primary data from the market operator, AEMO, are used to reconstruct prices, demand and dispatch. Specifically we use aggregated 30 minute price and demand data sets for the regions (REGIONIDs). Dispatch data are from 5-minute dispatch tables at the generation unit level (DUIDs), published every 5 minutes.
