Guest post by Mike O’Ceirin
This is not about “Climate Change.” It addresses the issue of whether wind as implemented is an effective replacement for fossil driven power stations. This is about Australia where we are according to the mainstream media in a transition to renewable energy even though after many years, we are far from it. We are closing coal-fired power stations, but the expansion of renewable energy is slow. We are approaching a crisis point. The reality of the Australian situation can be applied worldwide, and this report draws on data and nothing else. This
There are many offhand comments that we should just replace fossil as a source of electricity with renewables. What is the detail and how successful has it been? That is exceedingly difficult information to come by. This so-called transition started in 2000 with the implementation of the Renewable Energy Target.
For years, the Australian Energy Market Operator (AEMO) has published data on the amount of energy dispatched to the eastern grid by all generators registered with them. Using this data, a relational database was created for analysis.
Wind at the end of 2019 was 8.3% of the dispatched energy to the eastern grid. The infrastructure to achieve this is extremely large. There are 55 active wind power stations over a huge area. The plate capacity is 6973 MW and their capacity factor is 29%. On today’s prices that is a cost of $15.7 billion. The dispatched energy from wind was 16.9 TW hours in 2019. This exceeds the second-biggest coal-fired power station on the grid, Bayswater. After 20 years the biggest source of renewable energy can on average replace one large coal-fired power station. There are 16 coal-fired power stations and their lifespan is twice that of wind.
The supporters of wind energy ignore the actual performance. Plate Capacity is less important than energy dispatched. Critically the importance of variability is not understood. Faith that the large pumped storage facility of Snowy Mountains 2.0 will stabilise renewable energy dispatch is a delusion.
Our current wind infrastructure plus that large pumped storage theoretically could replace only one large coal-fired power station. To do that it has to be stable. That comes at a cost the current estimate is $5 million but more importantly the other cost is a diminishing of the available energy from wind. The data when modelled for 2019 shows that the amount of energy available drops by 3 TW hours to 14 TW hours. To put it another way stability with pumped storage means 17% less electricity.
Wind is a failure; every 2 GW of fossil generation is only with replacement by rebuilding the entire wind infrastructure plus a pumped storage facility of the size of snowy Mountains 2.0 again. It will not happen. Wind is growing very slowly and that will continue at that place. Over the last nine years wind has grown by 5.5%, and only 1.3% in 2019.
Bayswater Versus Wind
If wind energy generation is to replace coal, we must compare them. Bayswater is the second largest coal-fired power station on Australia’s eastern grid it has a plate capacity of 2.6 GW. A comparison can be made with the data produced by the AEMO.
Wind dispatched about the same amount of energy as one large coal-fired power station in 2019. For the comparison I choose Bayswater Black coal station in the Hunter Valley of New South Wales. That single station dispatched 16 TW hours during 2019. The 55 wind stations connected to the eastern grid spread over the entire east coast of Australia dispatched 17 TW hours. Advocates of renewables propose that that is the answer to replacing coal. Wind supplied 8.3% and fossil fuels 77.5% of electricity on the eastern grid. Can we then simply set about duplicating our current wind infrastructure ninefold and coal is gone?
First, we must consider how coal which is a controlled generator compares with wind that certainly is not.
The chart above shows the whole month of July 2019 in five-minute steps. The published data by the AEMO has that resolution but to show this in print is difficult. This chart stands for 9000 rows of data and is representative of all the other months.
The dispatched electricity from a coal-fired power station depends on the load and the number of turbines running. The low output in the beginning of the month for instance was as result of only three turbines running. The chart shows coal does vary, in this case from 112 MWh to 220 MWh. The variation is because of demand.
By government mandate the energy delivered by wind must be accepted and it is not driven by demand but by the amount of wind occurring at any one time. The chart shows that variation, it is from 12 MWh to 385 MWh. Since it is not controlled how can it be accommodated? In the electricity industry that the question is, how dispatchable is it? For a generator to be useful when run it must reliably produce a known amount of energy when needed. Would be it useful for instance to have a car which is only capable of more than 20 km an hour some days? If renewables are going to become practical as the sole source of electricity it must be able to satisfy the demand when required. The only way to achieve that is storage of the energy produced so that it is dispatchable. For this to happen there is a significant financial and efficiency cost.
Wind Energy Storage
The proposed pumped storage facility called Snowy Mountains 2.0 is the expected answer. It is designed to produce 2 GW for seven days. This means 336 GW hours to stabilise wind. The current installation of wind has a plate capacity of 7 GW. Can these two together supply electricity so that coal could be no longer needed?
A standard dispatch needs to be calculated and wind would supply that as the first order of business. Measuring the average power output of wind over the whole year gives a result of 1953 MW. Our data is in five-minute steps so that output will deliver 163 MWh for each five minutes. That will be achieved if there is no loss in the storage. As explained below those losses will bring the standard load down to 134 MWh for each five minutes.
The flowchart shows the process needed. If more is being generated than the standard load, then the load is dispatched, and the remainder used to charge the storage. Unless of course the storage is fully charged. If less than the standard load is being generated, then sufficient from storage is added and the standard load dispatched.
Charging means in a pumped storage system that water will be pumped to the upper dam. Discharging means water will be released from the upper dam to run turbines that generate the required electricity. This means a loss of 30% and must be included in the calculations.
Another of loss will occur if the maximum capacity has been reached. Any excess electricity generated at this point will be lost. These factors mean it is not possible to maintain electricity dispatched under the current system. Trial and error shows that in the case of 2019 hypothetically 83% of the average can be achieved. Stabilising using pumped storage means 17% less energy will be dispatched! That is 134 megawatt hours every five minutes which is 83% of 163 megawatt hours.
It is not possible to show these things in their entirety in a chart, so it is calculated on the five-minute steps for 2019. The following chart shows part of that, July in five-minute steps.
So here graphically is the process. The green shows when there is excess energy that can be used to charge storage and the red shows when energy must be drawn back from the storage. The black line represents a constant dispatch to the grid. Of course, if this infrastructure were built wind would then be under control and the dispatch would vary according to the demand.
Then the question is can this be applied to an entire year? It can be modelled, and the following chart shows the result of modelling the storage in the above example for 2019. Snowy Mountains 2.0 has a maximum capacity of 336 GW hours. As a starting point it is assumed that at the beginning of the year there is 225 GW hours available as carryover from the previous year. The modelling has been done in 30-minute steps because of the physical limitations on displaying the resultant chart. The amount of electricity dispatched in half-hour increments is 0.802 GW hours which is of course six times the above figure for five minutes.
So, step through the operation for each half-hour for the entire year. That is follow the flowchart above. This produces data that can be put into a chart so that it can be seen visually what will be happening.
At the end of January 112 GW hours is available. From there charges increase until the middle of February. The charge is maintained more or less until early March. Here is an example of what might happen from there until late March it is all discharge. On 24 March the status of the storage is 19 GW hours. That is a discharge of 200 GW hours in 17 days! There is a rise in charge from this point until it drops again to 12 GW hours 20 days later. There is doubt in practice dropping to such a low level would be an acceptable risk. In terms of percentage it means that the pumped storage is at 4.8% of the full charge. The expected stable output achieved would be 0.802 GW hours per half-hour. This is critical trying for any increase results in failure.
From there the wind comes back and there are constant rises until 11 June. At that point the pumped storage is fully charged. Notice the flat tops in the storage line. This represents the storage being fully charged. As shown in the flowchart this is where energy must be dumped because it has nowhere to be stored. It cannot just be dispatched because that would be destabilising the grid in a world where there is no other energy source.
So what is the end point of this? Hypothetically a large pumped storage in combination with the existing wind infrastructure could produce stable dispatchable power to replace one fossil power station. But there is a cost. According to the model this would produce 14 TW hours for 2019. This is not equivalent to Bayswater, being short by two terawatt hours. Financially the cost will be about $21 billion! $16 billion for the wind infrastructure and 5 billion for pumped storage.
This effort would produce 8.3% of the dispatched electricity on the eastern grid. Fossil energy is still 77.5%. It does not seem likely that wind will replace it. Page Break
Wind Energy Infrastructure
There are a large number of wind power stations, 55 in all. The most northerly being Mount Emerald in Queensland and the most southerly Musselroe in Tasmania. These are 2627 km apart. On the other dimension the furthest west is Cathedral Rocks in South Australia and the furthest east on the same latitude is Gullon Range in New South Wales. They are 1269 km apart. This should suffice to attain stability if what is required is a sufficiently large area. If we estimate the number of wind turbines by dividing the plate capacity by 2.5 MW there are approximately 2789 actual turbines. Actual land area can be found from the work done by David Mackay. The area is nearly 3500 km².
In 2019 wind dispatched 16.94 TW hours to the grid. Take the plate capacity and multiply it by the number of hours in the year gives the maximum energy possible of 57 TW hours. Dividing that into the actual dispatched terawatt hours results in a figure of 29%. That is known as the capacity factor and is a measure to compare with other forms of energy generation. In the case of renewables by government mandate all energy must be taken by the retailers. From this I assume in the case of wind it is an accurate measure.
The cost of this large infrastructure is approximately $15.7 billion. A recent power station Silverton near Broken Hill has been taken as a base for this cost its size is 200 MW and was built at a cost of 450 million. It also has 25 km of transmission line which is assumed to be a common requirement. The expected dispatch was 780,000 GW hours per annum but in 2019 only 424,000 MW hours was achieved. Possibly it was not fully operational, but it was fully commissioned by May 2018.
The dispatched energy from an electricity generator will vary. This also applies to wind. The chart shows this variation for all the wind stations above combined.
On average wind dispatched 46.4 GW hours per day. The lowest day was 11.53 GW hours and the highest 99.11 GW hours. That’s a range of 88 GW hours.
Coal Energy Infrastructure
In the case of wind all stations connected to the eastern grid were included. In the case of coal comparing all stations with wind would create a large mismatch. There are 16 coal-fired power stations which dispatched 140 TW hours of energy in 2019 against 17 TW hours for wind. I will choose our second largest coal-fired power station Bayswater in the Hunter Valley. In 2019 this one station dispatched 16 TW hours and was not running at full capacity. It has four turbines each of 660 MW so a joint plate capacity of 2640 MW. In 2019 the capacity factor for this station was 70%. What this means if it is to be applied overall to black coal power stations is important. A black coal power station is obviously under the control of its operators and this can be seen graphically.
This chart shows the dispatched energy in gigawatt hours of each turbine on daily rests. There are 237 days of a turbine off-line. If it is assumed that it was not necessary to switch those turbines off, then the capacity factor is 83%. Fully operational Bayswater Power Station can produce 19 TW hours of electricity annually. A 2 GW HELE coal-fired power station has a cost of $4 billion. That figure is based on one built in Germany in 2016 and the proposed station in Queensland at Collinsville, so an estimate to replace Bayswater would be $5 billion.
How variable is the output? This is more difficult than an estimate for wind because because turbines may deliberately be off-line also it appears there are variations which are according to demand rather than an intrinsic variation.
All data in this work comes from the official source publicly published on the National Energy Market website. The files published there are zip files which have a 13-month lifetime. This has been accumulated into a relational database on a powerful PC. The sources in their raw form now occupy about 300 GB. When you unzip these files, they are in CSV form. That is, they are text files that have the data values separated by commas. Those CSV files were imported into a relational database. The data used is the reports on electricity generators registered with the eastern grid of Australia. There are about 300 generators registered. For each of those power output at five-minute intervals is recorded. This is a considerable amount of information and not at all easy to make sense of. This is hundreds of gigabytes and from this all the information above was extracted. Being an analyst/programmer with expertise in the SQL query language allows me to do data mining of this for information. For instance, it would not be that hard to determine what a particular generator was outputting at a particular time to the nearest five minutes over the past nine years.