Solar Power Station MLFs

Transcription

1 A.B.N Report () to

2 VERSION HISTORY Version History Revision Date Issued Prepared By Approved By Revision Type Melinda Buchanan Ben Vanderwaal Draft Melinda Buchanan Ben Vanderwaal Release Melinda Buchanan Ben Vanderwaal Appended pool price analysis (Appendix C) VERSION HISTORY

3 TABLE OF CONTENTS 1) BACKGROUND ) SCOPE ) KEY ASSUMPTIONS ) DEMAND ) Olympic Dam expansion ) CARBON PRICE ) RENEWABLE ENERGY TARGET ) COMMISSIONING AND RETIREMENT OF FOSSIL FUEL PLANT ) NETWORK ) METHODOLOGY ) RESULTS ) MARGINAL LOSS FACTORS ) POOL PRICES ) REVENUE STREAMS ) A NOTE ON TIME ZONES ) CONCLUSIONS APPENDIX A) APPENDIX B) MLF CALCULATION METHODOLOGY... I MODELLING OF RENEWABLE GENERATION... II B.1) SOLAR PHOTOVOLTAIC AND SOLAR THERMAL MODELLING... II B.1.1) Solar PV generation... III B.1.2) Solar thermal generation... IV B.2) WIND MODELLING... IV B.3) BIDDING OF RENEWABLE GENERATORS... VI APPENDIX C) SOME CONTEXT FOR POOL PRICE FORECASTS... VII C.1) C.2) HISTORICAL NEM WHOLESALE PRICE FLUCTUATIONS... VII A FORECAST TO VIII TABLES OF CONTENTS Page i of ii

4 LIST OF TABLES TABLE 3.1 MODELLED CAPACITY FACTORS... 4 TABLE 5.1 MARGINAL LOSS FACTORS IN WITHOUT OLYMPIC DAM EXPANSION... 5 TABLE 5.2 MARGINAL LOSS FACTORS IN WITH OLYMPIC DAM EXPANSION... 6 TABLE 5.3 REVENUES WITHOUT OLYMPIC DAM EXPANSION, INCORPORATING MLFS ($2010 MILLIONS)... 9 TABLE 5.4 REVENUES WITH OLYMPIC DAM EXPANSION, INCORPORATING MLFS ($2010 MILLIONS) FIGURE B.3 RENEWABLE GENERATOR BIDDING... VI LIST OF FIGURES FIGURE 5.1 AVERAGE NEM POOL PRICES IN WITHOUT OLYMPIC DAM EXPANSION... 7 FIGURE 5.2 AVERAGE NEM POOL PRICES IN WITH OLYMPIC DAM 400 MW EXPANSION... 7 FIGURE 5.3 THE IMPACT OF 400 MW OLYMPIC DAM EXPANSION ON WEIGHTED AVERAGE NEM POOL PRICES IN FIGURE 5.4 TOTAL REVENUE WITHOUT OLYMPIC DAM EXPANSION FIGURE 5.5 TOTAL REVENUE WITH OLYMPIC DAM EXPANSION FIGURE 6.1 AVERAGE TIME OF DAY GENERATION CONCENTRATING PV FIGURE B.1 RENEWABLE ENERGY PLANTING TO MEET THE RET... II FIGURE B.2 LOCATIONS OF BOM WEATHER STATIONS... V FIGURE C.1 HISTORICAL NEM POOL PRICES... VII FIGURE C.2 SOUTH AUSTRALIAN GENERATION AND VICTORIAN IMPORTS... VIII FIGURE C.3 HISTORICAL AND FORECAST NEM POOL PRICES... IX FIGURE C.4 RATIO OF REGIONAL POOL PRICES TO SOUTH AUSTRALIAN POOL PRICE... X FIGURE C.5 ASSUMED INTERMITTENT RENEWABLE ENERGY PRODUCTION AS A PERCENTAGE OF TOTAL REGIONAL GENERATION ENERGY PRODUCTION... XI FIGURE C.6 ASSUMED INTERMITTENT RENEWABLE GENERATION CAPACITY AS A PERCENTAGE OF TOTAL REGIONAL GENERATION CAPACITY... XI TABLES OF CONTENTS Page ii of ii

5 1) BACKGROUND When electricity is transported, resistive heating in the transmission lines causes losses. To incentivize loads and generators to situate in locations in the National Electricity Market (NEM) with low transmission losses, marginal loss factors (MLFs) are applied to the sale and purchase of electricity. A loss factor greater than 1 in a certain location incentivizes generators, since it will proportionally increase the price at which they are paid for their electricity by the pool. A loss factor less than 1 means generators will receive proportionally less than the pool price for each megawatt hour of electricity. Furthermore, the marginal loss factor affects the total number of renewable energy certificates generated by an eligible renewable generator. In a previous study for RenewablesSA, Revenue Projections for Solar Installations (June 2010), ROAM Consulting provided generation and revenue forecasts for solar plants in a number of locations for the financial year The technologies investigated were: - tilted plate photovoltaic (PV); - concentrating PV; and - solar thermal parabolic trough without storage. The sites chosen were: - Olympic Dam in South Australia; - Mildura in Victoria; - Moree in New South Wales; and - Emerald in Queensland. It was noted in the report that the trade-off between solar resource and transmission strength will need to be carefully assessed by states or proponents when selecting sites. The impact of MLFs can potentially outweigh any benefit of higher solar resource in remote areas. The purpose of this study is to conduct a marginal loss factor assessment for the four sites listed above for the year , for each of the three solar technologies investigated in the earlier report. 2) SCOPE The derived values of MLFs, wholesale pool prices and generator revenues are strongly dependent on the generation pattern that applies throughout the NEM. Market forecasts are sensitive to the assumed supply-demand balance, the generation mix and generator trading behaviour (among many other factors). The simulations underpinning the results in the earlier ROAM report Revenue Projections for Solar Installations (June 2010) were conducted in August Since that time, there have been a number of public announcements which are expected to have a significant impact on wholesale 1 ROAM conducted these simulations to inform the Boston Consulting Group s recommendations to the Commonwealth Government on the structure of the Solar Flagships Program. MAIN REPORT Page 1 of 13

6 pool prices and generator dispatch in To provide the most accurate assessment of marginal loss factors, ROAM has incorporated these updated assumptions into a new set of simulations which form the basis for the results in this report. The resulting forecast solar plant wholesale pool revenues have also been included for completeness and comparison with earlier estimates. To provide context for these pool price forecasts, the factors driving historical NEM wholesale price fluctuations are discussed in Appendix C). The past and forecast future relativity between South Australian prices and those in other NEM regions is also explored. In addition to these market simulations, ROAM has provided marginal loss factor forecasts for 250 MW 2 solar generators in at each of the four sites; Emerald, Moree, Mildura and Olympic Dam. Furthermore, marginal loss factors have been calculated for each site with an expanded load at Olympic Dam. This yields a total of 24 marginal loss factor calculations. 3) KEY ASSUMPTIONS The key assumptions for the year made in the revised simulations and marginal loss factor assessment are described below. 3.1) DEMAND The simulations without an Olympic Dam expansion used the 2009 AEMO regional medium demand forecasts 3. The medium 50% probability of exceedence (M50) forecast represents the demand and total energy forecast under medium economic growth and moderate weather conditions exceeded once every two years. The medium 10% probability of exceedence (M10) forecast represents the demand and total energy forecast under medium economic growth and extreme weather conditions exceeded once every ten years. To calculate marginal loss factors, ROAM uses the dispatch obtained from M50 demand simulations. However, to obtain realistic pool price and revenue forecasts, ROAM carries out simulations with both the M50 and M10 demand forecasts. The final pool prices and revenues are calculated as a weighted average of the pool prices and revenues from the M50 and M10 simulations. Specifically, the weighting applied is 0.7 times the M50 results, plus 0.3 times the M10 results. The earlier study for Renewables SA, Revenue Projections for Solar Installations (June 2010), used the 2008 AEMO regional medium demand forecasts. This forecast, published before the Global Financial Crisis, had significantly higher energy and peak demands for in all regions except South Australia, compared to the 2009 AEMO regional medium demand forecasts. Consequently, pool price outcomes are quite different. 2 Nameplate capacity 3 These were published in the 2009 Electronic Statement of Opportunities, available from MAIN REPORT Page 2 of 13

7 3.1.1) Olympic Dam expansion The BHP Billiton report Olympic Dam expansion EIS: Energy and Greenhouse Gases 4 states that the expected increase in load will be about 650 MW, consuming 4,400 GWh annually (a load factor of 77%). Options being considered for servicing this load are an onsite 600 MW combined cycle gas turbine or a duplication of the 275 kv transmission line from Port Augusta to Olympic Dam. The full expansion project is expected to take 11 years. In the simulations modelling an expansion of Olympic Dam, the Olympic Dam load has been increased by 400 MW in every period in The existing line from Port Augusta has been upgraded (this is necessary to support the new load overnight, when the solar plant is not generating) and the solar plant is the only onsite generator. This represents a reasonable best case for the solar plant. 3.2) CARBON PRICE In these simulations ROAM has applied a carbon price of $32.32/tonne CO 2 -e in (real June 2010 dollars). This corresponds to the price developed by Treasury for to achieve a reduction in CO 2 -e emissions of 5% below 2000 levels by In the earlier study for Renewables SA, Revenue Projections for Solar Installations (June 2010), ROAM applied a carbon price of $46.10/tonne CO 2 -e in , corresponding to a reduction in CO 2 -e emissions of 15% below 2000 levels by While there is great uncertainty surrounding Australia s carbon policy, ROAM considers it reasonable to incorporate a carbon price in medium and long term modelling. The cost of emissions permits (or a carbon tax) for each generator is passed through to the wholesale electricity price via an uplift in generator bids. 3.3) RENEWABLE ENERGY TARGET The Large-Scale Renewable Energy Target has been modelled as being met predominantly by wind generation, with some biomass and one gigawatt of solar plant (corresponding to four 250 MW plants, one at each of the four sites of interest). The assumed breakdown of RECs by year to 2030 is shown in Appendix B). The solar plant capacity factors modelled at each site for each technology are shown in Table Australia s Low Pollution Future, The Economics of Climate Change, Australian Government Treasury, MAIN REPORT Page 3 of 13

8 Table 3.1 Modelled capacity factors Emerald (QLD) Moree (NSW) Mildura (VIC) Olympic Dam (SA) Tilted plate PV 19% 19% 19% 20% Concentrating PV 23% 23% 24% 26% Solar thermal parabolic trough 20% 20% 20% 23% 3.4) COMMISSIONING AND RETIREMENT OF FOSSIL FUEL PLANT Sufficient new fossil fuel plant is installed in addition to new renewable generation to meet minimum reserve levels. Minimum reserve levels are set by AEMO at the minimum level needed to ensure that reliability of supply will meet the Reliability Standard for the NEM (that is, load shedding does not exceed 0.002% of total energy). New minimum reserve levels were published in July 2010 and these were used in the updated simulations. New entrant generators are planted in line with committed and likely proposed projects. CS Energy announced in March 2010 that it would progressively close the 480 MW coal-fired Swanbank B power station in Queensland, with all units offline by This retirement is included in the updated simulations and has a significant impact on the Queensland supplydemand balance and wholesale pool price. Munmorah power station in New South Wales is scheduled to close in late This was included in both the previous simulations and revised simulations. In South Australia and Victoria, all existing thermal power stations were modelled as operating in The Australian Government Energy White Paper and AEMO s NTNDP 6 input data assumes that Playford coal-fired power station will retire in , due to a shortage of coal supply from Leigh Creek. The retirement of the Playford or Northern generators will significantly change the transmission flows in northern South Australia and affect the marginal loss factors of plant in that region. 3.5) NETWORK The network is modelled using the 2010 National Transmission Network Development Plan constraints 7. These take account of existing and committed transmission network upgrades and are updated each year by AEMO. The only additional network augmentation modelled is an upgrade of the Davenport to Olympic Dam line, in the case with an expanded load at Olympic Dam. 6 Available from 7 The NTNDP constraint equations are a subset of those used in dispatching the NEM to ensure that transmission lines are not overloaded in the event of forced outage of any single transmission line. MAIN REPORT Page 4 of 13

9 The earlier study for Renewables SA, Revenue Projections for Solar Installations (June 2010), used the 2008 Australian National Transmission Statement constraints and corresponding committed transmission augmentations known at that time. 4) METHODOLOGY As mentioned in Section 2), the derived values of MLFs are strongly dependent on the generation pattern that applies throughout the NEM. The generation and load patterns determine the flows on all transmission elements in the NEM, which in turn determine the loss factors. To establish generation patterns to allow MLFs to be calculated, ROAM uses its proprietary 2-4-C market forecasting software to generate the market forecasts. The results of the 2-4-C simulation studies, including generator dispatch and load, are then be imported into the PowerWorld load flow software package for the calculation of MLFs on a half-hourly basis throughout the year. Details of the MLF calculation methodology can be found in Appendix A). 5) RESULTS 5.1) MARGINAL LOSS FACTORS Table 5.1 lists the volume-weighted marginal loss factor in for each site and each technology without an expansion of the Olympic Dam load. Table 5.1 Marginal loss factors in without Olympic Dam expansion Emerald (Lilyvale) Moree (Dumaresq) Mildura (Red Cliffs) Olympic Dam (Olympic Dam West) Tilted plate PV Concentrating PV Solar thermal parabolic trough The Mildura site (connecting to Red Cliffs 220 kv substation) has the most favourable MLF in under the assumptions in these simulations. This site s MLF will be strongly affected by the pattern of wind development in north-west Victoria and south-west New South Wales. Emerald and Olympic Dam have similar MLFs. The slightly greater capacity factor for solar plants at the Olympic Dam site is offset by the slightly lower MLF. As noted in Section 3.4) the retirement of the Port Augusta Northern or Playford generators is expected to have a significant impact on the MLF at Olympic Dam. The Olympic Dam MLF will also be affected by the pattern of wind development in northern South Australia. Despite the high voltage and large export capability at the Dumaresq connection point, the MLF at Moree (Dumaresq) is relatively poor. The MLF here is predominantly determined by the pattern of flows along the Queensland New South Wales interconnector and is relatively insensitive to MAIN REPORT Page 5 of 13

10 generation from the modelled Moree plant connected at Dumaresq. In these simulations, the supply-demand balance and regional generation mix (particularly coal versus gas) in is such that New South Wales is importing heavily from Queensland. Consequently, the losses relative to the New South Wales reference node are high and the MLF relatively poor. Nevertheless, the Dumaresq MLF is significantly better than that at the Moree 132 kv connection point, where the MLF is between 0.68 and 0.75, dependent on the solar technology. Such an MLF makes this connection point unviable for a 250 MW generator. Table 5.2 lists the volume-weighted marginal loss factor in for each site and each technology with a 400 MW expansion of the Olympic Dam load and an upgrade of the Davenport to Olympic Dam line. The Olympic Dam MLF increases to over 1.01 with the expanded load, making this the most attractive site. The Mildura MLF is also positively affected by increased demand at Olympic Dam, increasing by around across all technologies. Table 5.2 Marginal loss factors in with Olympic Dam expansion Emerald (Lilyvale) Moree (Dumaresq) Mildura (Red Cliffs) Olympic Dam (Olympic Dam West) Tilted plate PV Concentrating PV Solar thermal parabolic trough ) POOL PRICES The pool prices outcomes for each region of the NEM without and with an expansion of Olympic Dam are shown in Figure 5.1 and Figure 5.2 respectively. The weighted average pool prices in both cases are compared in Figure 5.3. The 400 MW Olympic Dam expansion has the greatest impact on the average South Australian pool price, increasing it by $10.60/MWh. However, this increase also flows through to other regions of the NEM. Victoria s average pool price increases by $3.80/MWh, Tasmania and New South Wales increase by $2.30/MWh while Queensland (the most distant region from South Australia in terms of electrical connectivity) is relatively unaffected. MAIN REPORT Page 6 of 13

11 Figure 5.1 Average NEM pool prices in without Olympic Dam expansion Figure 5.2 Average NEM pool prices in with Olympic Dam 400 MW expansion MAIN REPORT Page 7 of 13

12 Figure 5.3 The impact of 400 MW Olympic Dam expansion on weighted average NEM pool prices in Historical pool prices and the relativity of prices between regions are discussed in Appendix C). 5.3) REVENUE STREAMS Table 5.3 summarises the wholesale pool revenue, REC revenue (assuming a REC price of $50) and total revenue for each site and each solar technology, without an Olympic Dam expansion. Total revenue incorporating the MLFs listed in Table 5.1 is graphed in Figure 5.4. MAIN REPORT Page 8 of 13

13 Table 5.3 Revenues without Olympic Dam expansion, incorporating MLFs ($2010 millions) Solar technology Location Wholesale pool revenue REC revenue (assuming $50/REC) Total revenue Emerald (QLD) Tilted plate PV Moree (NSW) Mildura (VIC) Olympic Dam (SA) Emerald (QLD) Concentrating PV Moree (NSW) Mildura (VIC) Olympic Dam (SA) Emerald (QLD) Solar thermal parabolic trough Moree (NSW) Mildura (VIC) Olympic Dam (SA) In terms of total revenue, the higher capacity factor of solar plant at the Olympic Dam site is offset by the lower pool prices in South Australia and the lower MLF. In contrast, despite having relatively poor capacity factors and MLFs, the Moree site generates the greatest total revenue, due to high New South Wales pool prices. MAIN REPORT Page 9 of 13

14 Figure 5.4 Total revenue without Olympic Dam expansion Table 5.4 summarises the wholesale pool revenue, REC revenue (assuming a REC price of $50) and total revenue for each site and each solar technology, with an Olympic Dam expansion. Total revenue incorporating the MLFs listed in Table 5.2 is graphed in Figure 5.5. The significantly improved MLF and higher South Australian pool prices result in Olympic Dam being the site which generates the greatest revenue in this case. The additional 400 MW load and upgraded transmission increase the total revenue for the Olympic Dam site by around $10 million in MAIN REPORT Page 10 of 13

15 Table 5.4 Revenues with Olympic Dam expansion, incorporating MLFs ($2010 millions) Solar technology Location Wholesale pool revenue REC revenue (assuming $50/REC) Total revenue Emerald (QLD) Tilted plate PV Moree (NSW) Mildura (VIC) Olympic Dam (SA) Emerald (QLD) Concentrating PV Moree (NSW) Mildura (VIC) Olympic Dam (SA) Emerald (QLD) Solar thermal parabolic trough Moree (NSW) Mildura (VIC) Olympic Dam (SA) Figure 5.5 Total revenue with Olympic Dam expansion MAIN REPORT Page 11 of 13

16 6) A NOTE ON TIME ZONES The NEM is dispatched on Eastern Standard Time (or market time 8 ). Load and generation data in ROAM s model of the NEM are set to market time. ROAM s solar model uses the Bird and Hulstrom model for clear sky radiation, with the output in GMT+10 (market time). This model uses the longitude and latitude of the site to calculate sunrise, sunset and clear sky expected radiation at each half hour period, relative to market time. Bureau of Meteorology data, collated at the local standard time, is then used to adjust the clear sky trace to incorporate simulated periods of cloud or rain outages into the solar insolation pattern 9. The time of day generation of solar plants at lesser longitudes (further west) is shifted later in the day relative to sites at greater longitudes (further east), when measured against market time. Figure 6.1 shows the average time of generation from modelled concentrating PV generators at each of the four sites. The Mildura and Olympic Dam plants are generating for an hour later in the day (market time) compared to the Emerald and Moree generators. This is accounted for in the revenues and pool prices described in the previous sections. Figure 6.1 Average time of day generation Concentrating PV 8 See the glossary entry under time and day in the National Electricity Rules, available from 9 Further description of ROAM s solar modelling is provided in Appendix B) MAIN REPORT Page 12 of 13

17 7) CONCLUSIONS The differences in revenue across the four sites for the single year investigated are sufficiently small that the ranking of sites by total revenue could change with a change in the forecast demand, supply-demand balance or generation mix. It may be that site-specific attributes not related to the electricity market (such as land availability and cost, site access and infrastructure, planning approvals etc.) provide a greater incentive for a generator to locate in a particular jurisdiction rather than favourable regional electricity market conditions alone. The forecasts indicate that the marginal loss factor for a 250 MW solar generator at the Olympic Dam site will be highly dependent on the load profile at Olympic Dam mine and competing generation developments in northern South Australia. By , it is expected that the penetration of intermittent renewable generation in South Australia (particularly wind generation) will lower the South Australian pool price relative to prices in other NEM regions. This factor and the relatively poor MLF at Olympic Dam adversely affect the pool and REC revenue streams compared with the other regions. On the other hand, the addition of a significant base load and transmission upgrades to Olympic Dam makes this site the most attractive in terms of total revenue. MAIN REPORT Page 13 of 13

18 Appendix A) MLF CALCULATION METHODOLOGY Step 1: Conduct a 2-4-C market forecast 2-4-C is ROAM s flagship product, a complete proprietary electricity market forecasting package. It was built to match as closely as possible the operation of the AEMO Market Dispatch Engine (NEMDE) used for real day-to-day dispatch in the NEM. However, it is capable of modelling any electricity network, and is in use to model small systems such as the North-West Interconnected System (NWIS) of Western Australia, and the enormous 4000 bus CalISO system of California. 2-4-C implements the highest level of detail, and bases dispatch decisions on generator bidding patterns and availabilities in the same way that the real NEM operates. The model includes modelling of forced full and partial and planned outages for each generator and inter-regional transmission capabilities and constraints. ROAM continually monitors real generator bid profiles and operational behaviours, and with this information constructs realistic market bids for all generators of the NEM. Then any known factors that may influence existing or new generation are taken into account. These might include for example water availability, changes in regulatory measures, or fuel availability. The process of doing this is central to delivering high quality, realistic operational profiles that translate into sound wholesale price forecasts and accurate marginal loss factor assessments. Step 2: Build and modify load flow into a form suitable for study ROAM has maintained a number of system snapshots of the NEM and included major transmission augmentations as published by the transmission network service providers from time to time. This information has been modified to suit the conditions of the study. This information includes all data on: New lines, line connection points and substations; Dates for commissioning/decommissioning of transmission lines; Line parameters; and Loads and load distribution factors. Step 3: Export simulation data to load flow and solve the load flow cases The half-hourly generator and load data for each connection point is fed into the load flow program. The flows in each half-hour period are solved using the PowerWorld simulator. Step 6: Calculate the average annual MLFs Each half-hour of the forecast has a unique MLF due the variability of demand and generator dispatch patterns. Generator connection points are assigned an annual average volume-weighted MLF which is calculated as: MLF VW each half hour ( MLF * Generation) each half hour Generation APPENDICES Page I of XI

19 Report to: Appendix B) MODELLING OF RENEWABLE GENERATION Sufficient renewable generation was planted to meet the Enhanced Renewable Energy Target (comprising the Large-scale Renewable Energy Target, or LRET, and the Small-scale Renewable Energy Scheme, or SRES), as shown in the figure below. The structure of the scheme, which allows for banking of renewable energy certificates (RECs), means that the shortfall in annual generation in later years is covered by banked RECs created in earlier years. An allowance for 5% growth in GreenPower sales has been included in the planting Figure B.1 Renewable energy planting to meet the RET Geothermal Biomass Solar thermal Large-scale solar PV Wind Hydro Landfill / Waste Solar water heating SGU SRECs SRES LRET + SRES B.1) SOLAR PHOTOVOLTAIC AND SOLAR THERMAL MODELLING ROAM s modelling uses a detailed meteorological model to produce solar availability traces that vary by time of day, by time of year and by location. The clear sky solar radiation incident on a location in the absence of any atmospheric effects is modelled by a solar model used by the National Oceanic and Atmospheric Administration (NOAA), part of the United States Department of Commerce. This models the position of the sun and incident radiation on Earth s atmosphere for any given date and location, and takes into account the local elevation. APPENDICES Page II of XI

20 ROAM then uses an atmospheric model developed by Bird and Hulstrom 10 that estimates the incident solar radiation, both direct (line of sight to the sun) and diffuse (sunlight reflected from the ground or from clouds), based on a number of atmospheric parameters (including type of local terrain, ozone thickness, water vapour present in the atmosphere and atmospheric pollutants). This produces a clear sky (no cloud) solar insolation trace, at the half hour level, that takes into account local atmospheric effects. However, solar plant is significantly affected by cloud cover and this must be taken into account in the final model. Ideally, data at the hourly level (at least) would be obtained for each specific site of interest for a full calendar year. Unfortunately, relatively few sites have to date been monitored in Australia, and those that have been are not located ideally for solar plant. Instead, ROAM has obtained data from the Bureau of Meteorology (BOM) on the total daily (global) solar radiation received each day at weather monitoring stations around Australia. This data is obtained from cloud cover satellite imagery and uses a sophisticated computer model estimate daily solar exposure. Calibration tests by BOM have shown it to be accurate to within 7% on sunny days and within 20% on cloudy days. The BOM data is used to calibrate ROAM s model by introducing periods of partial or total cloud cover during each half hourly period of each day until the reported daily total global incident radiation is reached. From this method, ROAM produces a half hourly global solar radiation trace. An empirical model of diffuse solar radiation is employed to separate out the diffuse and direct beam components, calibrated by sites where detailed half hourly data is available. B.1.1) Solar PV generation A detailed geometric model is employed to calculate the portion of the direct and global solar insolation on the PV plate. Only the direct component is used by concentrating solar PV plant, while both the direct and diffuse components are utilised by flat panel solar PV. The name plate capacity of the cells is assumed to correspond to Standard Testing Conditions (STC) which correspond to 1000W/m2 incident radiation (either beam or global as appropriate) and an operating temperature of 25 C. A derating factor of 78% (in the form of a reduction in output energy) is applied to solar PV to account for the losses in conversion from DC to AC current. Solar PV cells display a generally linear response to incident radiation. However efficiency decreases at high temperatures. A simplified model 11 is used to estimate the cell temperature based on incident radiation and ambient temperature (obtained from BOM), and a further derating factor of 0.44%/ C is applied. 10 A Simplified Clear Sky model for Direct and Diffuse Insolation on Horizontal Surfaces, R.E. Bird and R.L Hulstrom, SERI Technical Report SERI/TR , Feb Solar Energy Research Institute, Golden, CO 11 Photovoltaic Array Performance Model, David L. King, William E. Boyson, Jay A. Kratochvil, Sandia National Laboratories 2004 APPENDICES Page III of XI

21 B.1.2) Solar thermal generation Solar thermal parabolic trough power stations are modelled as having a single axis tracking system. A minimum incident radiation of 200 W/m2 is assumed to be required for operation (in the absence of storage) and the start up time is assumed to be minutes to reach full capacity. Both these quantities can vary from plant to plant, but are representative parameters for near term plant 12. An auxiliary load of 15 MW (out of 250 MW nameplate capacity) and solar multiple of 1.1 is assumed in this study. B.2) WIND MODELLING A description of the method for determining wind generation is included as wind generation makes up a large proportion of the modelled new renewables. Individual announced wind farm projects are planted in their announced locations around the grid to make up the target, and are included in transmission congestion calculations on a half hourly basis. To model the output of wind farms, the average wind speed at the wind farm site is required for each half hourly period, which can then be converted into generator output using turbine power curves. Historical data was sourced from automatic weather stations around Australia from the BOM. The locations of the weather stations in eastern Australia are shown in the figure below. 12 See for example Potential for Renewable Energy in the San Diego Region, Appendix E, prepared by National Renewable Energy Laboratory ( APPENDICES Page IV of XI

22 Figure B.2 Locations of BOM weather stations The wind data from the BOM weather stations was taken at a variety of elevations (from 1m off the ground to 70m above the ground), and elevation strongly affects wind speeds. The wind at the height of a turbine hub (from 50m to 80m) will be much faster than the wind at ground level, and the amount of the increase in speed is strongly dependent upon many factors, including the type of ground cover (rock, grass, shrubs, trees) and the nature of the weather pattern causing the wind. In addition, the local topography affects wind speeds very strongly (winds tend to be focused by flowing up hillsides, for example). Therefore, the wind speed at a weather station perhaps 30km distant from a wind farm is likely to be correlated strongly in time with the wind at the site of the turbines, but the absolute scaling of the speeds is highly uncertain. However, it is reasonable to assume that the wind speeds at the weather station will be very highly correlated in time with the wind speeds at the turbine site (analysis of existing wind farm generation profiles compared with the BOM weather station data has shown this to be the case). APPENDICES Page V of XI

23 To provide the absolute scaling, ROAM uses data from the Renewable Energy Atlas of Australia 13. The Atlas contains modelling data provided by Windlab Systems giving the mean annual wind speeds, at a typical turbine height of 80m, at 3km resolution for most of Australia. The mean wind speed at the wind farm site is used to scale the data from the closest weather station to provide an estimate of the wind speed time series at turbine height. Finally, the wind speeds are adjusted (reduced) to account for turbulence and shading across the wind farm (the park effect ), calibrated by historic data from existing wind farms. A turbine power curve is then applied to convert the wind speeds into actual generation (this accounts for the fact that the efficiency of turbines varies strongly with wind speed). As a final check, the annual time of day average generation is compared to historic data, and the output adjusted if necessary to achieve an appropriate time of day average generation curve. This accounts for qualitative differences between time of day wind speed distributions at hub height versus the BOM stations. This method captures the daily and seasonal variation of wind at different sites, and also the likely correlation in the operation of nearby wind farms, which is highly material for assessing likely transmission congestion. B.3) BIDDING OF RENEWABLE GENERATORS Schedulable renewable generation (geothermal and biomass/bagasse) were bid into the market at prices which reflect their fuel and variable operation and maintenance costs, while intermittent generators were bid at $0/MWh. Consequently, intermittent generators are dispatched at their full capability in each period, unless limited by transmission constraints. Figure B.3 Renewable generator bidding Plant type Biomass / Bagasse Geothermal Solar PV and solar thermal Wind Bid price $29.77/MWh $2.05/MWh $0/MWh $0/MWh 13 APPENDICES Page VI of XI

24 Appendix C) SOME CONTEXT FOR POOL PRICE FORECASTS C.1) HISTORICAL NEM WHOLESALE PRICE FLUCTUATIONS Figure C.1 shows the annual average pool price for each NEM region since Figure C.1 Historical NEM pool prices All regions of the NEM were affected by the severe drought in 2007 and 2008 and this is reflected in the high average pool prices in , and All-time record demands were recorded in Victoria, New South Wales, South Australia and Tasmania during The average weekly pool price in South Australia exceeded $120/MWh for three weeks in December 2007 March 2008 (including ten days where the average pool price was above $200/MWh) due to extreme high temperatures. While the weather in Queensland was mild in , prices were still high compared to longterm averages. Tarong Power Station reduced its output to 30-75% of full output in 2007, due to water shortages in Queensland s South-East. As Tarong Power Station typically provides up to a quarter of Queensland s power, this reduction had a significant upward impact on Queensland prices in and Moreover, New South Wales is a net importer from Queensland 14 Tasmania joined the NEM on 16 May The year is the first full financial year in which Basslink was operating and as such is the first year for which Tasmania has annual average pool price data. APPENDICES Page VII of XI

25 and the effects of the decline in Tarong s generation also flowed through to the New South Wales price. Hydro generation also dropped significantly in 2007 and 2008 compared to long-term averages. Hydro Tasmania s total production in 2008 was 6,646 GWh compared to their stated long term average of 9,500 GWh. As a consequence, annual average Basslink flows into Tasmania from Victoria almost doubled in and relative to flows in preceding and subsequent years. Tasmanian pool prices reflected the generation shortages in those years. Yallourn power station typically provides around 20% of Victoria s electricity. In late 2007, its coal supply was interrupted for several weeks and its annual generation dropped by around 1,200 GWh (compared to typical annual generation of around 11.5 TWh). This reduction and the effects of the drought on Snowy and Tasmanian hydro plant meant that interconnector flows from Victoria into South Australia dropped markedly in More expensive thermal generators in South Australia increased their production, pushing up pool prices. Annual average South Australian interconnector flows and South Australian generation over the period of interest is shown in Figure C.2. Figure C.2 South Australian generation and Victorian imports C.2) A FORECAST TO Figure C.3 shows historical NEM pool prices (discussed in the previous section) and ROAM s forecast pool prices to the year APPENDICES Page VIII of XI

26 Figure C.3 Historical and forecast NEM pool prices This forecast was conducted under the assumptions described in Section 3), with the assumption that a carbon price will be imposed from 1 July The introduction of a carbon price (at a starting price of $28.38/tCO 2 -e, rising to $32.32/ tco 2 -e by ) causes a jump in prices in as carbon costs are passed through to the wholesale price. It should be noted that these forecasts are a weighted average of the results under 50% and 10% probability of exceedence demands, in order to capture an expected price. Extreme weather conditions (such as those experienced in South Australia in 2007) will result in much higher average pool prices. Significant new thermal and renewable capacity has been commissioned in the past 6 months. In the absence of extreme weather or lengthy forced outages at large generators, this additional capacity is expected to alleviate the high prices observed over the past four years. Proposed new renewable generation to meet the Renewable Energy Target will also assist in keeping wholesale prices at or below average. South Australia is expected to receive a large proportion of renewable generation investment (mainly intermittent wind generation) to meet the RET. As intermittent renewable generators are price-takers, it is expected that South Australian wholesale prices will rise relatively less than 15 These prices correspond to the Treasury-modelled emissions permit prices under the Carbon Pollution Reduction Scheme, to deliver a 5% reduction in emissions by 2020 relative to 2000 levels (Australia s Low Pollution Future, The Economics of Climate Change, Australian Government Treasury, 2008.) The stated prices have been converted to 2010 dollars. APPENDICES Page IX of XI

27 prices in other mainland NEM regions. Figure C.4 shows the ratio of regional pool prices to the South Australian pool price since , and projected to (under the assumptions described in Section 3). Figure C.4 Ratio of regional pool prices to South Australian pool price Figure C.5 shows the intermittent renewable generation in each region in this forecast as a percentage of total regional generation, while Figure C.6 shows the installed intermittent renewable capacity in each region in this forecast as a percentage of total installed regional capacity. Under either measure, South Australian renewable investment far exceeds that in any other NEM region, and this underpins the relativity between prices shown in Figure C.4. APPENDICES Page X of XI

28 Figure C.5 Assumed intermittent renewable energy production as a percentage of total regional generation energy production Figure C.6 Assumed intermittent renewable generation capacity as a percentage of total regional generation capacity APPENDICES Page XI of XI