All posts by pjb253

Second Submission to the Senate Inquiry into Wind Turbines

Second Submission to the

Senate Inquiry into Wind Turbines

Peter Bobroff, AM

Introduction

I lodged one submission before the initial deadline where I addressed wind related issues for which I already had analysed data.  Subsequent communication from the Committee staff raised further areas of interest. Those within my area of interest were:

This second submission addresses these issues. I don’t suggest that I can provide definitive answers to all of these complex questions, but I hope to broaden the readers’ understanding.

Contents

 1. Capital and operating costs of wind and comparison to conventional power generation
1.1. Short Run Marginal Cost
1.1.1. Fuel Cost
1.1.2. Thermal Efficiency
1.1.3. Variable Operating and Maintenance Cost
1.1.4. Renewable Energy Certificate
1.1.5. Final Short Run Marginal Cost
1.2.  Long Run Marginal Cost
1.2.1. Interest Costs
1.2.2. Fixed Operating and Maintenance Costs
1.2.3. Final Long Run Marginal Costs
1.3. Profitability under different marginal costs
1.3.1. Long Run Profitability
1.3.2. Short Run Profitability
1.3.3. Short Run Profitability without RECs
1.3.4. Long Run Profitability without RECs
1.4.  Summary – Capital and operating costs of wind
2. Actual power delivery by wind farms 2014
2.1. Plots
2.2. Histograms
2.3. Tabular
2.4. Summary – Actual power delivery by wind farms
3. Times at which wind power was delivered
3.1. MACARTH1 on 2014-09-17 one generator for one day
3.2. MACARTH1 during 2014-09 one generator for one month
3.3. All wind farms on 2014-09-17
3.4. All wind farms during the month of 2014-09 by unit
3.5. All wind farms during the month of 2014-09 by date
3.6. All wind farms during the month of 2014-03 by unit
3.7. All wind farms during the year 2014 by unit
3.8. All wind farms during the year 2014 by month
3.9. Summary – Times at which wind power was delivered
4. Projected impacts  on global temperatures
4.1. Impact of renewables on atmospheric CO2
4.1.1. The Pessimistic Scenario
4.1.2. The Optimistic Scenario
4.2. Impact of CO2 concentrations on global temperatures.
4.2.1. MAGICC outputs
4.3. Summary – Projected impacts  on global temperatures
5. Conclusions

1 Capital and operating costs of wind and comparison to conventional power generation

The actual costs are almost certainly Commercial in Confidence.

The Australian Energy Market Operator (AEMO) publishes generic generator parameters which can  be combined with actual generator dispatch data and actual state price data to allow estimated Marginal Costs to compared with actual Prices.

AEMO publishes multiple sources of these generic generator parameters which do not always agree. Parameters are missing for some generators that dispatched power in 2014.  AEMO does not assign unique generator and station identifiers so the matching of data between AEMO spreadsheets often requires considerable effort and cannot be relied upon to be 100% accurate.

In this discussion of capital and operating costs, only generators dispatching more than 40,000 MWh in 2014 have been considered. The 40 generators dispatching less than 40,000 MWh in 2014 have very high marginal costs for Interest on capital and Fixed Operating and Maintenance and this results in poor plot scales for viewing the larger more important generators.

1.1 Short Run Marginal Cost

This is the extra cost of generating an additional 1MWh of electricity. It neglects all annual costs such as: Interest on capital and operating/maintenance costs.

SRMC ($/MWh) is defined a:

  • Thermal Efficiency (GJ/MWh) x Fuel Cost ($/GJ)
  •  + Variable Operating and Maintenance cost ($/MWh)
  • Renewable Energy Certificate value (-$35.24 for renewables, +$3.52 on fossil fuels)
1.1.1 Fuel Cost

screenshot-localhost 2015-04-28 14-28-42The Fuel Costs were derived from the type of fuel and the geographic zone of the generator.

Wind and Hydro have zero fuel costs but there could be a monetary value assigned to the release of dam water for generation in competition to release for irrigation.

Gas is usually much more expensive than Coal.

These scatter plots show a point for each generator at the year of commissioning. The dot size represents the size of the generator in MW.

1.1.2 Thermal Efficiency

screenshot-localhost 2015-04-28 14-24-37This relates the theoretical heat energy of the fuel in GJ to the electrical energy produced from the fuel in MWh.

AEMO parameters give Hydro about 82% efficiency. Presumably that is the conversion of gravitational potential energy into electrical energy. The figure is unimportant as the Hydro fuel cost is assumed to be zero.

AEMO parameters give Wind 100% efficiency. It is unlikely that the kinetic energy of the wind in the blade circle is converted to electrical energy with anything like 100% efficiency but it is unimportant as the fuel cost is assumed to be zero.

The increasing efficiency of modern gas turbines is evident.

1.1.3 Variable Operating and Maintenance Cost

screenshot-localhost 2015-04-28 14-31-12These costs are related to the amount of power being generated. Open the throttle and the VOM costs go up.

Coal plants and some gas turbines have extremely low VOM costs, even lower than wind apparently.

1.1.4 Renewable Energy Certificate

Wind generators receive one REC for each MWh they generate. RECs can be traded on a market with average price in 2014 being about $35. Electricity retailers had to source about 10% of their energy from wind. This seems to equate to a 10% rise in the effective price of non-wind energy for the retailer.

1.1.5 Final – Short Run Marginal Cost

screenshot-localhost 2015-04-28 14-54-29This is the total of: Fuel Cost by Thermal Efficiency, VOM costs and the REC.

The Combined Cycle Gas Turbines appear to be steadily improving their SRMC. Coal and Hydro tend to be much lower than Gas.

The negative SRMC of Wind is due to the $35/MWh Renewable Energy Certificate.

 1.2 Long Run Marginal Cost

This adds the annual costs of Interest on Capital and Fixed O&M costs to the Short Run Marginal Cost. To avoid bankruptcy, a company must sell its product for at least the LRMC.

LRMC ($/MWh) is defined as:

  • + Short Run Marginal Cost ($/MWh
  • +Interest at 10% on Capital and Capital Works / Energy  dispatched in 2014 (MWh)
  • +Fixed Operating and Maintenance cost ( $/MW/year) / Energy  dispatched in 2014 (MWh)
1.2.1 Interest Costs

screenshot-localhost 2015-04-28 15-10-24No capital cost parameters were available for Hydro, so zero has been assumed. Perhaps all the debt has been repaid?

The 10% pa capital cost came from What is Normal Profit for power generation? Simhauser and Ariyaratnam p14 and covers debt or equity finance.

The annual interest cost has been reduced to a marginal cost by spreading it over the energy generated in 2014.

Marginal interest costs for Wind appear greater than those for Coal and some Gas

1.2.2 Fixed Operating and Maintenance Costs

screenshot-localhost 2015-04-28 15-19-24Hydro generators have quite a spread of FOM Costs, probably reflecting different amounts of energy generated in 2014.

Wind has higher FOM Costs than Coal and most of the CCGT.

1.2.3 Final – Long Run Marginal Costs

screenshot-localhost 2015-04-28 15-25-19Hydro LRMC is very low, due mostly to the assumed lack of Interest on Capital. This might not be correct, but no data could be found.

Most Coal, Wind and CCGT gas fall in the range 30-70 $/MWh.

Open Cycle Gas Turbines are often much higher.

1.3 Profitability under different marginal costs

  • Long Run Marginal Cost
  • Short Run Marginal Cost
  • Long Run Marginal Cost with no Renewable Energy Certificates
  • Short Run Marginal Cost with no Renewable Energy Certificates

All of these are covered by showing:

  • the Regional Reference Prices (state wholesale spot price)  as a histogram of occurrence against Price in $/MWh. Separated by state.
  • the Marginal Costs as a bar chart of Registered Capacity in MW against cost in $/MWh. Separated by generator type

The X axes are identical to facilitate comparison.

1.3.1 Long Run Profitability

screenshot-localhost 2015-04-28 15-27-31 If a company cannot consistently sell its generated power at above its Long Run Margin Cost, it will eventually  go bankrupt.

If AEMO’s generic parameters are approximately correct, most generators are unprofitable, except for Hydro where Interest costs have been assumed to be zero.

Why do the generation companies persist? Things are obviously not as simple as they seemed at first sight.

 1.3.2 Short Run Profitability

screenshot-localhost 2015-04-28 15-56-42The wholesale electricity market is a short term spot market. The lowest bids are accepted. If you can’t win contracts at your Long Run Marginal Cost you will go broke but if you can win bids at above your Short Run Marginal Cost you will go broke more slowly.

Most Short Run Marginal Costs are below the state wholesale prices so there is much scope for this short term death spiral.

This is known as the Missing Money Problem. The suggestion that infrequent price spikes allow some recompense seems inadequate.

The REC gives Wind a particular advantage in being to able to drive prices into the negative. This does occur.

1.3.3 Short Run Profitability without RECs

screenshot-localhost 2015-04-28 16-19-46Removing the RECs would lessen the ability of wind generators to drive the price into the negative region but do nothing to solve the Missing Money Problem and to restore the industry to long term profitability.

1.3.4 Long Run Profitability without RECs

screenshot-localhost 2015-04-28 16-11-48The main effect of withdrawing RECs is to make most Wind generators very unprofitable in the Long Run.

 1.4 Summary – Capital and operating costs of wind

Given the assumptions stated earlier and the incompleteness of data, these observations are offered:

  • The spread of wholesale prices is usually less than the Long Run Marginal Costs of most generators. Money is being steadily lost.
  • The Short Run Marginal Cost of most generators is below the spread of wholesale prices. Understandably, generators are using this and bidding low to go broke as slowly as possible.
  • Without the Renewable Energy Certificates, the Long Run Marginal Cost of Wind is far more than that of fossil fuels.
  • Renewable Energy Certificates allow wind generators to bid negative prices and still exceed their Short Run Marginal Cost.

The national electricity market does not seem to be developing sustainable prices.

2 Actual power delivery by wind farms – 2014

The following plots and histograms are based on AEMO 5 minute data for every generator attached to the grid for the whole of 2014 – about 20 million samples.

2.1 Plots

screenshot-localhost 2015-05-01 16-38-14Over the whole of 2014, Wind provided 4% of the energy to the grid. Coal provided 75%.

The daily plots show Coal averaging more than 15GW on most days with Wind averaging less than 2GW on most days.

There is a small change throughout the year.

2.2 Histograms

We are now interested in the amount of power generated, irrespective of the time of day, day of month or month of the year and irrespective of which unit generated it. These histograms show what proportion of the time a certain power level was generated. In the first (Coal), the maximum of about 17GW was being generated almost 16% of the time. 20GW was generated only about 1% of the time

screenshot-localhost 2015-05-01 16-36-54

Coal is dominant, delivering between 11GW and 21GW when the demand required it. There is no need to look at daily or monthly data as power is generated when required.

Gas provided about 1GW to 8GW of peaking power whenever it was required. There is no need to look at daily or monthly data as power is generated when required.

Hydro provided about 400MW to 5GW of power whenever it was required. There is no need to look at daily or monthly data as power is generated when required.

Wind provided from a tiny 5MW to nearly  3GW controlled by the vagaries of  the wind.

Wind is of significance only in South Australia which has little coal and must use gas whose fuel price is increasing.

2.3 Tabular

screenshot-localhost 2015-05-01 16-48-54Here are the statistics as numbers.

2.4 Summary – Actual power delivery by wind farms

Wind is unreliable and other generators need to be online ready for it to fail.

From AEMO spreadsheet,  contribution factors specify the portion of installed capacity expected to be available to meet peak demand. The AEMO table has been pruned of some rows and the X for 100% column derived.

Untitled 1The table shows, that to meet an expected peak load of 1GW, 45GW of wind should be installed in NSW and Qld. South Australia could get by with installing 12GW for each 1GW of expected demand.

Without the subsidy of Renewable Energy Certificates, Wind’s Long Run Marginal Cost is already twice that of Coal. If NSW and QLD need to install 45 times the required capacity, then it becomes 90 times more expensive.

The proposal to replace Coal by Wind in the major states is absurd.

3 Times at which wind power was delivered

AEMO provides data covering the power dispatched by almost all generators connected to the grid for every 5 minute period. Data is available for over 30 wind farms. This analysis has been restricted to 2014, but still contains over 3 million data points for wind alone.

It is easy to lose track of the fine detail when attempting to comprehend this amount of data, so we will start simply and then expand the view out and up.

The Times at which can be interpreted as:

  • the hours of day – variations throughout the day
  • the days of the month – variations between days
  • the months of the year – variations between the seasons

When looking at one wind farm, MACARTH1 will be used as it is the biggest at 420MW.

Most of the following plots will be in terms of the Capacity Factor of the generator which is the actual power dispatched to the grid expressed as a percentage of the Registered Capacity (maximum). This facilitates the comparison of generators of different size.

3.1 MACARTH1 on 2014-09-17 – one generator for one day

screenshot-localhost 2015-04-30 15-57-56When looking at one generator for one day, a simple line plot clearly shows each 5 minute period with no statistics needed.

The great variation in output from one wind farm during the day is obvious.

3.2 MACARTH1 during 2014-09 – one generator for one month

screenshot-localhost 2015-05-01 11-26-55Now we have expanded to all the days in the month of September 2014. All 5 minute points are still visible but now encoded by colour. Colour is not as clear as the plot line in the previous plot, but the minute to minute variation can still be seen.

The date shown earlier, 2014-09-17, can be seen as unexceptional. There are windier days and calmer days.

MACARTH1 reached 100% capacity on occasions.

There appear to be 3 or 4 calm days followed by a few windy days, probably corresponding to the passage of high pressure systems from west to east over Australia.

 3.3 All wind farms on 2014-09-17

screenshot-localhost 2015-04-30 16-44-28Here we have returned to 2014-09-17 but are now showing all wind farms. All 5 minute points are still visible.

Some generators reached 100% capacity briefly. Others never reached 50%.

MACARTH1 is not exceptional with other units doing better and others doing worse.

3.4 All wind farms during the month of 2014-09 by unit

screenshot-localhost 2015-05-01 11-34-33For the first time, the basic 5 minute data points are no longer visible. Each point in time is now the average of such points for all days of the month. The day to day variation is now lost.

There appears to be a period of reduced wind  around 9am and another around 7pm (19:00 hours), but this is a very weak observation.

At no time of day did any generators have monthly averages of over 50% capacity.

Most wind farms did better than MACARTH1 during September 2014.

3.5 All wind farms during the month of 2014-09 by date

screenshot-localhost 2015-05-01 15-07-07 Each point in time is now the average of such points for all generators on that day. The unit to unit  variation is now lost.

In comparison with the previous plot, there is more variation from day to day than from generator to generator. Weather causes more variation than siting.

On some days wind gets up to 50-75% capacity for much of the day but on others is below 10% for much of the day.

3.6 All wind farms during the month of 2014-03 by unit

screenshot-localhost 2015-05-01 11-38-19March 2014-03 was calmer than September 2014-09.

SNOWTWN1 (99MW)  reached a monthly average of 54% capacity briefly between 8pm and 10pm but most generators did far worse at most times.

3.7 All wind farms during the year 2014 by unit

screenshot-localhost 2015-05-01 11-41-20SNOWTWN1 (99MW) reached a yearly average of almost 50% capacity briefly between 9pm and 10pm but most generators did far worse at most times of the day.

There is some suggestion of wind changes at about 9am and 7pm (19:00 hours).

3.8 All wind farms during 2014 by month

screenshot-localhost 2015-05-01 14-31-24Here each cell represents the average of all wind generators and the average of all days in the month. This process has suppressed all the variation between generators and the variation between days.

2014-03 March is the worst month and 2014-07 July was the best month.

3.9 Summary – Times at which wind power was delivered

The above plots demonstrate:

  • the output of a single wind farm can vary widely during a day , from very little to 100% capacity.
  • a single wind farm almost never produces 100% capacity all day
  • the output of a single wind farm can vary widely from day to day
  • the total output of all wind farms varies during a day, but much less than the variation of single wind farms
  • the total output of all wind farms varies from day to day, approximately between 10% to 75%
  • the total output of all wind farms varies throughout the months, from about 20% to 45%

There is almost no storage on the grid. Energy must be generated to meet the instantaneous demand. The minute to minute wind variations must be countered by matching variations on other generators which can no longer run at steady efficient settings. Other generators must be kept on standby in case of a rapid decline in wind. This is expensive and wasteful.

Wind without storage does not replace any other generators.

4 Projected impacts  on global temperatures

This can be approached in 2 steps:

  • the projected impact of renewables on atmospheric CO2 concentrations
  • the projected impact of CO2 concentrations on global temperatures.

4.1 Impact of renewables on atmospheric CO2

screenshot-en wikipedia org 2015-04-10 14-49-40There has been a massive growth in wind capacity since 1996 and in solar capacity since 2006. The claimed reductions in tonnes of CO2 emitted are very large.

CO2 concentrationsHowever there has been no detectable change in atmospheric CO2 trends over this time period. The trend remains at slightly less than 2 parts per million increase per year, so:

  • Either some unknown natural process has commenced and is cancelling the savings
  • or the CO2 concentration savings from current levels of renewables are non-existent.

4.1.1 The Pessimistic Scenario India, China, Japan, Germany and Russia are all building more coal-fired generators so a pessimist might suggest that the CO2 concentration trend might increase from 2  to 3 ppm per year. The 2050 concentration would then be 505 ppm.

4.1.2 The Optimistic Scenario The EU (and UK in particular) plans to increase renewable generators, so an optimist might suggest that the CO2 concentration trend might decrease from 2 to 1 ppm per year. The 2050 concentration would then be 435 ppm.

4.2 Impact of CO2 concentrations on global temperatures.

MAGICC is a Model for the Assessment of Greenhouse-gas Induced Climate Change. It is a set of coupled gas-cycle, climate and ice-melt models that allow the user to determine the global-mean temperature of user-specified greenhouse gas emissions. It was developed by the US based National Center for Atmospheric Research and represents the view of the IPCC Fourth Assessment Report.

screenshot-www sustainableorganizations org 2015-04-10 16-13-53The Pessimistic Scenario is the Reference Scenario in MAGICC and WRE750 has been chosen to give about 500 ppm in 2050. The Optimistic Scenario is the Policy Scenario in MAGICC and WRE450 has been chosen to give about 435 ppm in 2050.

The User Model was used with default values except that Climate Sensitivity was set to the AR4 most likely value of 3°C for a doubling of CO2. The time period selected was 2015-2050.

4.2.1 MAGICC Outputs

MAGICC Run 1

4.3 Summary – Projected impacts on global temperatures

The difference between the Optimistic and Pessimistic Scenarios at 2050 is the benefit from the world renewables policy and is about 0.25°C.  The cost of the the world renewables policy is hundreds of billions of dollars.

There have recently been several papers suggesting that the Climate Sensitivity may be lower than 3ºC. Some are suggesting as low as 1.5ºC. This would reduce the benefit to about 0.12ºC.

5 Conclusions

  1. When ample coal is available at current prices, wind is too expensive and unreliable to supply the grid.
  2. The fashionable proposal to completely replace Coal in the 3 major states by Wind is absurd.
  3. If ample coal is not available (South Australia) and gas prices rise, wind might cost effectively reduce gas consumption. An alternative would be the  building interconnectors to access more distant coal generators as has been done for Tasmania under Bass Strait.
  4. All the world’s current wind capacity has had no effect on CO2 concentrations and so can not have had any effect on temperature.
  5. MAGICC outputs shows that a massive world-wide effort to increase wind capacity might save 0.25ºC by 2050. A proportional Australian effort would have contributed virtually nothing.
Advertisements

Submission to Senate Select Committee on Wind Turbines

Submission to the Senate Inquiry into Wind Turbines

Peter Bobroff, AM

Contents

Introduction

This submission addresses the any related matter aspect of the Terms of Reference and offers information relating to the question Should more wind farms be built?

This submission is based upon a personal analysis of the 5 minute data from the Australian Energy Market Operator (AEMO) for every day of 2014 which comprises:

  • Demand power for each state
  • Regional reference price for each state which is assumed to be the wholesale spot price.
  • Dispatch power for every generator.
  • Dispatch power for each of the interconnectors between states.
  • Technical data on each generator including its registered capacity and data to allow a simplistic estimate of CO2 Emissions.

The Fuel Class: WindSolar currently contains only one solar generator – Royalla1, but no dispatched power was detected in 2014. Roof top solar is not dispatched by AEMO so is not included within analysis. Western Australia and the Northern Territory are not connected to the Grid. The Australian Capital Territory is considered part of NSW.

This submission was prepared as a blog posting and it was intended that the images should link into the interactive database that prepared the images. The database is not yet robust enough to allow public access.

Wind contribution to the whole grid during 2014

screenshot-gridyear publicknowledge com au 2015-02-25 12-35-20The pie chart is based on the 5 minute dispatch power of every generator for the Whole Grid in 2014.

As far as the current whole grid is concerned, wind is irrelevant. Other existing generators could instantly replace wind if wind’s special status were removed.

 Dispatched power into the Whole Grid for 2014

The histograms are all formatted with probability on the y axis. The 2 digits of fractional probability can be read as the percentage occurrence within one bar.

  • screenshot-localhost 2015-02-27 17-40-40Coal fired generators dispatched between about 12 and 20 GigaWatts (GW) with an average of 16.6 GW
  • Gas fired generators dispatched between about 2 to 4 GW with an average of 2.9 GW
  • Hydro generators dispatched about 1 to 3 GW with an average of 1.6 GW
  • Wind generators dispatched less than 3 GW with an average of 0.96 GW

The histograms show that coal dominates the grid. It provides the base load power, never less than 12GW. Gas and hydro provide the peak loads with their reliable quick responses. Sometimes only a little peaking is required, but their rapid responsive reserve is always needed for overall grid reliability. Wind, with all it’s special privileges,  has over 40% probability of producing almost nothing.

Whole grid on calm days

screenshot-localhost 2015-02-26 11-02-19There were 45 days when the average wind power throughout the day was less than 2% of the grid demand and 7 days when less than 1%.

2014-03-29 is an example of a particularly calm day across the whole grid.

The whole grid does not require any wind power for its reliable operation.

Whole grid on windy days

screenshot-localhost 2015-02-26 11-12-55There were 18 days when the average Wind power throughout the day exceeded 10%.

2014-09-28 is an example of a particularly windy day across the whole grid.

South Australia as a portent?

screenshot-localhost 2015-02-25 15-18-14

The present conditions of the SA grid might give some indication of the future of the eastern states’ grids if wind capacity is greatly increased and fossil fuel capacity  greatly reduced. The pie chart is for dispatched power separated by Fuel Class.

In 2014 South Australia generated about 45% of its power by Gas, 33% by Wind and 22% by Coal. Some of the wind power was exported, sometimes up to 8%. Often power was imported from Victoria, sometimes up to 20% of SA requirements.

How did South Australian Wind perform in 2014?

In terms of power dispatched

screenshot-localhost 2015-02-27 17-48-17Over the year 2014 South Australia’s wind farms produced anywhere from 0 MW to about 1200 MW with an average of 447 MW. Below 100 MW was the most common output, with 1200 MW being very rare.

In terms of whole state capacity factor

Capacity Factor is the power produced compared to the maximum possible (Registered Capacity).
screenshot-localhost 2015-02-27 17-55-56As would be expected, all the SA wind farms never reached full output simultaneously. They did occasionally exceed 80%. Far more commonly they produced less than 10%. Wind turbines can only produce their maximum power at a particular wind speed. If the wind gets stronger, the turbine must feather it’s blades to prevent damage.

In terms of individual farm capacity factor

screenshot-localhost 2015-02-27 18-03-25Looked at individually, the wind farms more frequently approached 100% capacity. However four times more frequently, they had to draw power from the grid to prevent damage to their main bearings during calms.

Windy days in South Australia

screenshot-localhost 2015-02-26 11-19-30There were 4 days when the average Wind power throughout the day exceeded 100% of the state demand. Power is often exported on windy days.

2014-09-28 is an example of a windy day in South Australia.

When the wind blows strongly and consistently, the dispatched power data looks good.

Calm days in South Australia

screenshot-localhost 2015-02-26 11-22-19There were 37 days when the average Wind power throughout the day was less than 10% of the state demand. Power is imported from Victoria on calm days.

2014-08-02 is an example of a calm day in South Australia.

Wind was almost non-existent except for a slight breeze during the night.

Contrasting windy and calm conditions in SA

Coal output in SA is relatively unaffected by wind conditions. However as the wind increases, the gas turbines and import interconnectors start to shut down.screenshot-localhost 2015-02-28 12-44-12 With no wind, the gas is almost at maximum. As the wind increases, gas decreases but never much below 20% of demand. There are times when gas is around 20% of demand when wind is also very low. The effects of imports and exports are not included in this plot.

The gas and imports are needed on wind-free days. There are no gas-free days when wind is need. This submission did not investigate whether the extra capital costs of wind generators are justified by the savings in fuel cost of gas generators on windy days.

Prices on a windy day

screenshot-localhost 2015-02-28 13-31-02On a windy day without any drama, the Regional Reference Price tends to fall with increasing wind power dispatched.

On 2014-10-06 prices varied from $45/MWh down to $6/MWh. With their special status and extremely low marginal costs, there seems little to prevent wind companies undercutting themselves in a downward spiral. Better to get a low price than no price.

screenshot-localhost 2015-03-01 16-40-282014-10-27 provides an example of this spiral resulting in negative prices.

Prices on calm days

screenshot-localhost 2015-02-28 13-34-30On calm days the price changes can be more dramatic. 2014-07-01 shows prices rising to 150 $/MWh during very calm periods. During this period there was a price spike to 11,000 $/MWh, which as been omitted so the lower price changes can be seen.

In this document the term PriceDeSpiked refers to the Regional Reference Price limited to the range -50 to 150 $/MWh

Price vs Wind % South Australia 2014-01-15

screenshot-localhost 2015-02-28 19-44-08A more extreme day was 2014-01-15 when spot prices rose to 13,000 $/MWh and fell to -1,000 $/MWh. This sort of price would seem to be an abberation.

Negative prices

screenshot-localhost 2015-02-28 19-51-56It is difficult to understand a market with negative prices. How can a seller pay customers to take a product? This seems to be a characteristic of electricity markets that include significant amounts of wind and solar. It is apparently quite common in Germany.

The two scatter plots to the right have their prices limited to the range +150 to -50 $/MWh with spikes being limited. Otherwise the detail in this range is lost if the spikes are shown linearly. Negatives are a bit of a problem with log scales.

In South Australia the negative prices can occur at any wind percentage but seem more common in higher winds. Above about 40% wind the floor of the scatter plot goes negative.

In New South Wales negative prices are far less common and the floor of the scatter plot remains positive.

It would take a fast talking Keynesian economist or a merchant banker to explain how such a market can be healthy in the long term.

An Austrian economist would probably say that such aberrations are inevitable when governments interfere in markets.

The Germans are apparently considering a scheme where generation companies must supply contracted power whether the wind blows or not. This moves the brown-out risk from the market operator to the contracting generation companies. This may improve the market health.

 Is Wind power cheaper?

The plot below compares the average price of power per day in each state for 2014.

South Australian prices don’t seem noticeably lower than other states who have less wind power. Wind power does not seem to be free.

A generation company cannot survive in the long term if the short term price it receives for its power is insufficient to cover its long term capital, operations and maintenance costs. The German proposal for contracting companies to supply power at a particular price irrespective of wind conditions, might help.

This submission does not address the long term viability of power generation companies.
screenshot-localhost 2015-02-25 16-29-09

Does wind deliver its registered capacity?

screenshot-localhost 2015-02-27 18-11-25Actual Dispatched Power divided by the Registered Capacity is known as the Capacity Factor and is usually expressed as a percentage.

This set of histograms averages all the units of a Fuel Class for each 5 minute period. When a generator is not delivering any power, it is assumed not to be required and that period is excluded from the samples, except for Wind which is assumed to be delivering zero power.

  • The most probable Capacity Factor of the whole SA wind class is only a few percent of Registered Capacity. The probabilities decrease to only a small probability of exceeding 80%.
  • SA usually relies on imports to some extent. The interconnectors operate when required, usually at between 10-60% of their Nominal Capacity.
  • SA has a lot of Gas generators which run a Capacity Factors from 5-60%
  • SA occasionally exports power, rarely getting up to 40% of the available interconnector capacity.
  • SA coal fired generators run at between 20-70% capacity with 40% being the most probable.

Capacity factors – treating units individually

screenshot-localhost 2015-02-27 18-06-12This set of histograms treats each generator separately. Each histogram therefore shows the probability of an individual generator of that Fuel Class operating at  the particular Capacity Factor.

  •  The interconnectors used for imports operate when required at desired factors of 0-100%
  • The Gas turbines operate when required at desired factors of 0-110%
  • The interconnectors used for exports are not often required to run at high capacity.
  • Coal fired generators are almost always in use, running at 50-100% capacity.
  • Individual wind turbines sometimes consume power to prevent damage to bearings in a calm but also occasionally deliver their rated capacity.  The most common output is almost zero.

Comparing windy and calm days.

screenshot-localhost 2015-03-01 16-47-17Here we are comparing a windy day with a calm day that had a similar demand curve. Sets are from South Australia.

Prices are cheaper on the windy day as wind generators have low marginal costs.

Prices generally appear to rise as the demand increases.

The low negative price spikes seemed to occur on windy days. Perhaps these are attempts by the market operator to shed excess capacity.

Variability of individual wind farms

screenshot-localhost 2015-02-27 10-42-07The output of an individual wind farm throughout the day is often quite variable. The short term variation (between 5 minute periods) as measured by the Variate Difference method is no greater than other Fuel Classes, however the dispatched power can disappear over 15 minutes.

Variability of wind across South Australia

screenshot-localhost 2015-02-27 10-55-03The plot shows standard deviations by the Variate Difference Method for each Fuel Class for each day in 2014-Q1. Wind is more variable than coal. Gas variability is probably due to balancing wind variability.

The wind variability is seldom over 2% of the dispatched power.

CO2 Emissions

The plots below are of daily averages for the year 2014. The weekly ripple is evident. Import and Export of power via the interconnectors between states is not being considered. South Australia often imports Brown Coal power from Victoria.

screenshot-localhost 2015-02-27 09-46-17The CO2 Emissions ( tonnes/hour )  don’t vary much with the change of demand throughout the year. Perhaps the seasonal demand increases are handled by gas turbines which emit less than coal fired ones.

The CO2 Intensities (tonnes/MWh) of the states reflect their dominant Fuel Class

  • VIC – brown coal
  • NSW and QLD – black coal
  • SA – wind
  • TAS – hydro

Could unilateral action to reduce emissions by South Australia significantly reduce Australian emissions?

VIC, NSW and QLD are the big  emitters and they determine Australia’s emissions. Similarly China, India and USA are the big emitters and they determine the Earth’s emissions.

Summary of the advantages of wind power

  • On windy days the wholesale price of electricity may fall. This might have some influence on the long term price to consumers;
  • Less CO2 is emitted, if you consider this an advantage.

Summary of the disadvantages of wind power

  • Wind generators do not replace existing generators as calm periods often occur. Even enormous numbers of wind farms can only reduce the probability of no wind output – not eliminate it. The big states (NSW, VIC, QLD) rely almost entirely on fossil fuels and have large reserves. It would be totally unaffordable to build enough wind farms to supply most of the demand most of the time. Even then the existing generators would still be required on occasions.
  • Increasing use of wind appears to result in large positive price spikes and periods of negative prices. These don’t seem to indicate a healthy market.
  • Less CO2 is emitted, if you consider this to be a disadvantage.

Final Observation

This inquiry has an unstated assumption:The world is in danger of Catastrophic Anthropogenic Global Warming and the most cost effective solution is to build wind and solar farms.

If the assumption fails, this inquiry is irrelevant. Another inquiry is needed into:

  • Have the climate models been accurately predicting the future?
  • Is the predicted future so bad that CAGW out ranks all other human problems?
  • Will China and India stop building fossil fuel powered power plants and replace all existing ones with nuclear, hydro, wind or solar?

Queensland price spikes become plateaus 2014-12-17

screenshot-grid publicknowledge com au 2014-12-18 07-54-53

Huge price increases

The Queensland average “wholesale” price  (Regional Reference Price) for electricity on 17 December 2014 was $1,965 per MegaWatt Hour. Usually it is in the region of $30-40. At a typical price of $30 per MWh consumers might be paying 15c per kWh, which is 5 times the wholesale price. At this new price of nearly $2000 per MWh and assuming the same markup, the consumer would be paying $10 per kWh – an enormous increase.

The price spikes that have been occurring in Queensland recently have been for isolated 5 minute periods only. Today the “spike” to $13,499 lasted from 3:45pm to 5:20pm and was preceded by a 15 minute spike and followed by a 10 minute spike.

Not caused by generator failure

The Dispatch plots for the individual Queensland generator don’t show any obvious generator failures. The gas generators that normally handle the peak demands can be seen coming online between 10am and 10pm as usual.

screenshot-grid publicknowledge com au 2014-12-18 07-58-47

Lack of reserve power

The reserve power available just by pressing on the accelerator pedals of online generators dropped to almost 20% at 4:30pm.

screenshot-grid publicknowledge com au 2014-12-18 08-37-39

I don’t yet have a long term plot of the minimum reserve power on all days but picking 15th April 2014 at random gave the plot below. Perhaps 40% reserve is considered a safe value.

screenshot-grid publicknowledge com au 2014-12-18 08-41-59

High prices shutting down industry?

In the chart below, demand for power starts to dip after it becomes apparent that the high price is persisting. Demand recovers when prices return to normal.

UPDATE 2014-12-19 The demand dip mentioned here also occurred on the next day when it was not preceded by any high prices, so the speculation that high prices shutting down industry was false. In fact the dip in demand at about 6pm occurs almost every day in Queensland. [Take 100 demerits for not checking before writing]

screenshot-grid publicknowledge com au 2014-12-18 08-47-03

OAKEY1 Gas Generator – Local price

There were some early price spikes between 1:30pm and 2:30pm which preceded a spike in the “local price” of OAKEY1 down to -$14,000 per MegaWatt Hour. The generator appeared to be running normally at full power from 10am to after 8pm, so this spike remains a mystery.

screenshot-grid publicknowledge com au 2014-12-18 08-59-30

Let the reader beware

This site is not my day job. It was started only to see the power generated by various generator types/fuels against the background of power demand. I have a Bsc( Elec Eng) from 1969, but that was before the invention of electricity. I don’t have an insiders knowledge of the power industry. All the base data comes from the AEMO site but there is a certain amount of fallible human programming before the charts appear. The plots seem plausible to me but that could be because of my white, male, capitalist, military background. Do some independent checking before readjusting your super or starting a riot on the contents of this site.

South Australian prices blown away 2014-12-16

2014-12-16 was windy in South Australia. Renewables (entirely wind) provided 7.5% of the Whole Grid energy and 62% of South Australia’s electrical energy.

screenshot-grid publicknowledge com au 2014-12-17 07-37-01

Unfortunately the peak of wind power 9am-7pm did not coincide with the peaks of demand at 8am and 8pm and there was 80-100% of reserve power available from other operating generators. This amount of instant reserve power seems untypically large and implies some fuel wastage.

screenshot-grid publicknowledge com au 2014-12-17 07-44-53

The Regional Reference Price in South Australia dropped to almost 10 $/MWh for much of the day. While good news for electricity retailers, it may be below the survival price for other types of generators which are need on calm days

The Lake Bonney Stage 3 Wind Farm (LKBONNY3) raises some interesting questions which I don’t have time to investigate at present.

For most of the time that the state price was down near 10 $/MWh, the LKBONNY3 Local price was -50 $/MWh. Yes MINUS $50. For most of this -$50 period LKBONNY3 produced no power, which seems sensible if the Local price actually means anything. However for some of the -$50 period LKBONNY3 was running at 80% Capacity factor, which is interesting.

screenshot-grid publicknowledge com au 2014-12-17 07-55-50

At about 12:40pm the state price spiked down to less than -30 $/MWh at the same time as LKBONNY3 Local price spiked down to -300 $/MWh. As LKBONNY3 was producing no power at the time of the spike, it can hardly have been a signal to produce less power. Was it a threat?

Data to support submission to Senate

From Peter Lang:

Here’s a first cut at describing charts I think are needed.

For each generator unit (eventually covering a full year)

  1. Power v time
  2. Capacity factor v time (I’d suggest to have both lines on on the same chart with power on main Y axis and capacity factor on secondary Y axis).
  3. Length of time not generating v date, AND number of starts v date (both lines on same chart)

Frequency histogram and cumulative frequency plot (eventually for the full year):

  1. Proportion of time v power output
  2. Proportion of time v capacity factor  (3 and 4 may be on the same chart)
  3. Proportion of time v ramp rate
  4. Proportion of time v length of time not generating

Filter and sort by state, fuel / technology, month.

Make it easy to move between charts of the same type for the different generators rapidly, e.g. by easy click on a legend and with a forward and back arrows.

INITIAL IMPLEMENTATION:

  • Power v time Usually called Dispatch.  Capacity Factor might be more generally useful and is currently provided. Easy to add Dispatch
  • Capacity factor v time (I’d suggest to have both lines on on the same chart with power on main Y axis and capacity factor on secondary Y axis).
    Have found the twin YAxis approach a bit potentially confusing. Can do it but it adds complexity and might lose a few readers.
  • Length of time not generating v date, AND number of starts v date (both lines on same chart) This can be seen on the 30 day chart. Length of stops and Length of runs histograms might show it well particularly when more days result in more samples.
  • Proportion of time v ramp rate. The Rate of Change Histogram might suffice.
  • Proportion of time v length of time not generating Is this covered by CapFactor 0% on the Cap Factor Histogram. Also the Length of stops histogram?

These are now available on all units as detailed below.

 One day sequences

The One Day Aspect of a unit shows these together with rate of change and prices. This is probably the finest detail required.

screenshot-grid publicknowledge com au 2014-12-12 11-24-23

30 day sequences

The Last30 days – Rate of Change Aspect shows that series longer than 1 day can be handled, although the X axis labeling needs work. The multi-day series lose the fine detail of the one day series but show the longer pattern of starts and stops better.  A year long series might be too compressed to show much. Histograms might be better.

screenshot-grid publicknowledge com au 2014-12-12 11-27-37

Histograms and scatter plots

The histograms will handle the year series with better precision.

screenshot-grid publicknowledge com au 2014-12-12 11-29-27

screenshot-grid publicknowledge com au 2014-12-12 11-31-57

screenshot-grid publicknowledge com au 2014-12-12 11-46-20

I have increases the marker size since this screen capture.

 

 

 

Reading Grid.PublicKnowledge.com.au home page

The main site is at http://grid.publicknowledge.com.au

Table 1

screenshot-grid publicknowledge com au 2014-11-25 16-48-21

On 2014-11-24 the grid produced a maximum power of 26.5 gigawatts at 14:00 in the afternoon – 2pm.

The Standard Deviation of demanded power was 53.8 megawatts. This is a measure of the short term variation of the demand.

The cost is determined using the Regional Reference Price for each 5 minute period. The “wholesale” price for this day for the grid was $24,000,000.

Coal fired generators provided 76.5% of the energy on this day.
Renewables is only wind at this stage as there don’t seem to be any solar generators contributing to the grid. Perhaps landfill gas and similar should be included in renewables as seems to be done in some countries.

This summary is intended to be suitable for use a radio broadcaster to use daily.

Figure 2

screenshot-grid publicknowledge com au 2014-12-02 16-27-17

This chart shows the percentage of energy generated by different broad fuel classes. Most legend items (ARROW 1) are links to lower level data.

In the lower figure  the demand for each fuel class is shown for each 5 minute period of the day.

The ARROW 2 points to the icon of a larger version of the chart.

Figure 3

screenshot-grid publicknowledge com au 2014-12-02 16-33-19

This shows how the demand for each state changes throughout the day.

Browsing to the state in the legend (ARROW 1) will make that state the subject of the next page.

The lower figure  shows the Regional Reference Price for each state during the day. Sometimes the states remain in fairly close agreement but often some state prices will have extreme values. 2014-01-14 is an example of very high price spikes while 2014-01-15 has both low and high spikes.

The price spikes warrant further investigation and perhaps a few dedicated posts.

Navigation bar

screenshot-grid publicknowledge com au 2014-11-25 17-17-12

The indicated icons will bring more content on to the page. These settings will be remembered if you have cookies enabled in your browser.

Navigation bar

screenshot-grid publicknowledge com au 2014-11-25 17-33-30

ARROW 1 allows for the selection of a past date. Currently the 1st and 15th of each month back to 2013-11 have been loaded. Since about 2014-11-12 each day has been loaded. Any day within the last 12 months can be typed into the box but will take a minute or two to respond.

ARROW 2 allows all loaded dates to be plotted and compared.

ARROW 3 shows where to change from dark on light to light on dark which is very vivid. Currently part of the plot background gets stuck in black, but this will eventually be fixed.

Australian Electricity Grid

The National Energy Market (NEM) is run by Australian Energy Market Operator (AEMO). Western Australia and Northern Territory are not included and the ACT is considered part of NSW.

The Grid

The grid supplies the power to satisfy the demand of consumers. In a grid the supply must always equal demand. If the demand increases suddenly, the grid frequency will reduce slightly causing the demand to be slightly lowered and indicating that the online generators should increase power to restore the frequency.

Variation of Demand

The demand changes throughout the day and through the seasons.

screenshot-grid publicknowledge com au 2014-11-21 15-18-06
Generators use different fuels.
screenshot-grid publicknowledge com au 2014-11-21 17-08-27

The generators that are actually used depends on the wind and prices amongst other factor. Here is the breakup by fuel class on a very windy day and for a calm day.

screenshot-grid publicknowledge com au 2014-11-21 17-59-09 screenshot-grid publicknowledge com au 2014-11-21 18-04-02

Generator Classes and Types

The generators that use the same fuel may differ in technology.

screenshot-grid publicknowledge com au 2014-11-21 17-32-12

Interconnectors between states

The State grids export and import power via Interconnectors. This site treats the forward and reverse directions as separate Interconnections. The interconnectors are not of unlimited capacity so the market may sometimes be fragmented and this may contribute to differences in price between states.

screenshot-grid publicknowledge com au 2014-11-21 17-27-26

Companies

All the generators are now owned by private companies.
screenshot-grid publicknowledge com au 2014-11-21 17-41-37

National Energy Market

The grid is managed as a market with companies bidding their generators and prices fluctuating. Occasionally the prices are quite stable and uniform across the whole grid, usually otherwise. Prices are an area for further investigation.

screenshot-grid publicknowledge com au 2014-11-21 18-19-18 screenshot-localhost 2014-11-21 18-29-36

Premiere of Grid.PublicKnowledge.com.au

I have  a new website at Grid.PublicKnowledge.com.au that presents some data from the Australian Energy Market Operator (AEMO). Only the Electricity grid is covered.  The focus is on individual days but some broader views are available.

Please comment here on site issues that urgently need  fixing or on enhancements that you can’t possibly live without.

There is a good French site that is more advanced but with a similar approach.