I don't understand how foresight helps, because in Australia we have to anticipate worst-case scenarios like two or more successive nights during wind droughts. In Europe we can expect several weeks on occasion with with little or no wind.
The cost and land use of the storage required to ride through those periods is out of the question.
To make the challenge even more impossible, how will the storage ever be fully charged without enough conventional power installed to do it?
In the context of Mr Osmond's posts, 16 or 24 hours of perfect foresight helps boost the %RE he can claim. His algorithm knows exactly what the conditions will be that far into the future, so it can time use of sources of RE (especially Hydro) across each day so as to maximise battery usage and hence minimise curtailment.
In the wider context, of course when the Wind doesn't blow at night / across several nights, then a solely-RE-powered territory is stuffed.
Which I think argues for a practicable target % RE far lower than Mr Osmond promotes.
... assuming that bill-payers & tax-payers continue to agree that 'decarbonisation' is a sensible path considering what the rest of the World is doing as well as the current state of various enabling technologies, the most obvious being energy storage.
On the other hand, ERCOT which has a much less diverse grid because it covers less than a third of the area of of the NEM, claims it can forecast day ahead wind and solar output more accurately than demand or gas and coal output. If ERCOT can do it why can't AEMO.
By the way AEMO has very detailed analysis of actual vs forecast performance of all its generators.
Then again spreadsheet jockeys on the other side of the world with no engineering or science training can do a better job I am sure.
And by the way what is the rest of the world doing expanding wind/solar/storage at an ever increasing pace while actually doing very little with other sources. In this decade nuclear capacity has declined even though output has increased slightly
Generator inertia is almost irrelevant in a high renewables grid. There is inertia in loads, hydro and syncons and even in the inductance of the transmission. In any case inertia is a just fast response but low magnitude store of energy that can easily be replaced at far lower cost with batteries, asynchronous flywheels, dry running hydro etc.
In the Australian context, all Victoria's ten brown coal units can supply 670 MWs of energy from inertia if frequency falls by an outrageous 0.4 Hz. If the fall occured over 1 second (this has never happened in Victoria) Victoria's 1.2 GW of batteries, can supply 1,100 MWs. The battery power is soon to be 4 GW. Further once the frequency excursion has been arrested, most of that energy has to be returned over the next 20 seconds to return frequency to normal so then energy is being deducted from the primary response, The battery just keeps on supplying for hours while the fault is corrected
1. Just because the maximum output of hydro was 6.1 GW, that is no reason to use that limit. There is 11GW of hydro on the NEM. The output of hydro in the Snowy is often limited by transmission constraints which are currently being addressed by Humelink. While small, Clover was out of action all of last year as was Barron Gorge for most of it, and Kidston will be coming on line sometime over the next 12 months. So with current hydro long before we have reached wind and solar capacity in David's models the upper limit on hydro will be around 8-9 GW. Even without transmission upgrades or Kidston it has exceeded 7GW in the past.
Then if Marinus is built another 700 MW can be added from Tasmania.
Following that and again long before we have reached David's W&S targets, Snowy II will be on line with another 2 GW followed closely by Borumba and Lake Lyell with another 2.7 GW. So by about 2033 16 GW of hydro/pumped hydro can easily supply 10-11 GW
2. David's model has enough wind and solar to supply 115% of annual demand. Twenty five years ago the NEM had almost 40 GW of thermal capacity to supply a peak of 24 GW (27 GW less hydro) and an average of 18.7 GW i.e. annual thermal generating capacity at 90% CF was 315 TWh but fossil fuels only supplied 165 TWh. Similar overcapacity can be shown in the US, UK and most of Europe. So if he is wrong and we actually have half the overcapacity as we had 25 years ago i.e. 45% rather than 15% then we easily exceed David's target and your claims are irrelevant.
3. As I have pointed out before David has ignored the 2-4 GW of V2G likely to be available by then, 2-3 GW of demand response which is being targeted and the 4-8 GW of customer batteries.
4. As for running gas and hydro while the sun is shining why not, if low wind or solar is forecast? ERCOT claims it can forecast wind and solar output better than it can forecast demand or FF output, so why can't AEMO. If there is sufficient storage to absorb the output of gas and hydro ahead of a shortage isn't that exactly what any sensible Grid operator would do. Isn't that why pumped hydro was built fifty years ago.
In summary another lot of sound and fury signifying nothing
You Said: <"David's model has enough wind and solar to supply 115% of annual demand. Twenty five years ago the NEM had almost 40 GW of thermal capacity to supply a peak of 24 GW (27 GW less hydro) and an average of 18.7 GW i.e. annual thermal generating capacity at 90% CF was 315 TWh but fossil fuels only supplied 165 TWh. Similar overcapacity can be shown in the US, UK and most of Europe. So if he is wrong and we actually have half the overcapacity as we had 25 years ago i.e. 45% rather than 15% then we easily exceed David's target and your claims are irrelevant.">
Using AVERAGE values is misdirection. The grid works on a second-to-second basis. The issue with NEM Wind and Solar is that it is highly intermittent. This causes 2 issues:
1) During low periods, it must be backed up with hydrocarbon generation because there is currently insufficient storage capacity to do more than make a token response.
2) During periods of high generation then much must be curtailed. David shows this, but does not show the magnitude of that curtailment. He also does not make allowances for inertia and frequency response as discussed elsewhere.
And the reason that hydrocarbons were reduced is because Wind & Solar are given precedence on the grid. It is therefore up to the hydrocarbons to compensate for changes in W&S Output.
But this is were using AVERAGES really misinforms the reader. Not only do you need to track the Average generation in a period, but you must also look at the peak generation. When you have low average generation, but High Peak generation, this makes for very expensive gas generation. You can read about it here: https://wrjohn1.substack.com/p/capacity-factor-cf-vs-lcoe
You Said: <"4. As for running gas and hydro while the sun is shining why not, if low wind or solar is forecast? ERCOT claims it can forecast wind and solar output better than it can forecast demand or FF output, so why can't AEMO. If there is sufficient storage to absorb the output of gas and hydro ahead of a shortage isn't that exactly what any sensible Grid operator would do. Isn't that why pumped hydro was built fifty years ago.">
Using storage to time-shift hydrocarbon generation is detrimental. It has no impact on maintaining storage levels, but most importantly, it becomes very expensive due to storage losses. There is more information about this here: https://wrjohn1.substack.com/p/can-energy-storage-reduce-nat-gas
BTW, when I first started to work with David's model, I built "Foresight" into my simulator. However, what I found was that it made no measurable difference over time, except to increase costs. For that reason, I removed it.
Which brings up another failure in David's model:
- He does not account for maximum power rates of his storage. You can see that changes in his storage level are very rapid. This is not reasonable as all storage power has limitations set by inverter sizing and absorption rate of the batteries themselves. If you go back on look at my model output, you will see that the rate of change in storage level is not linear. As demand for storage discharge increases, it reaches a maximum power limit.
- He does not account for State of Charge Limitations.
The bottom line here is as Chris B says, David's model does not reflect reality and is effectively meaningless.
I don't understand how foresight helps, because in Australia we have to anticipate worst-case scenarios like two or more successive nights during wind droughts. In Europe we can expect several weeks on occasion with with little or no wind.
The cost and land use of the storage required to ride through those periods is out of the question.
To make the challenge even more impossible, how will the storage ever be fully charged without enough conventional power installed to do it?
https://rafechampion.substack.com/p/grid-scale-electricity-storage-why
https://rafechampion.substack.com/p/are-we-taking-wind-droughts-seriously
Hi Rafe,
In the context of Mr Osmond's posts, 16 or 24 hours of perfect foresight helps boost the %RE he can claim. His algorithm knows exactly what the conditions will be that far into the future, so it can time use of sources of RE (especially Hydro) across each day so as to maximise battery usage and hence minimise curtailment.
In the wider context, of course when the Wind doesn't blow at night / across several nights, then a solely-RE-powered territory is stuffed.
Which I think argues for a practicable target % RE far lower than Mr Osmond promotes.
... assuming that bill-payers & tax-payers continue to agree that 'decarbonisation' is a sensible path considering what the rest of the World is doing as well as the current state of various enabling technologies, the most obvious being energy storage.
On the other hand, ERCOT which has a much less diverse grid because it covers less than a third of the area of of the NEM, claims it can forecast day ahead wind and solar output more accurately than demand or gas and coal output. If ERCOT can do it why can't AEMO.
By the way AEMO has very detailed analysis of actual vs forecast performance of all its generators.
Then again spreadsheet jockeys on the other side of the world with no engineering or science training can do a better job I am sure.
And by the way what is the rest of the world doing expanding wind/solar/storage at an ever increasing pace while actually doing very little with other sources. In this decade nuclear capacity has declined even though output has increased slightly
"with no engineering or science training"
I have a Master of Arts degree in Chemical Engineering from Cambridge University, Peter, plus over 40 years of engineering experience.
What qualifications do you have?
You are correct about "Perfect Foresight". It is just smoke and mirrors.
David continues to ignore the need for dispatchable generation and Inertia. He has a long period every day where the Grid would collapse.
You can see more of my comments here:
https://wrjohn1.substack.com/p/reply-jul-12-2025-chris-b
Generator inertia is almost irrelevant in a high renewables grid. There is inertia in loads, hydro and syncons and even in the inductance of the transmission. In any case inertia is a just fast response but low magnitude store of energy that can easily be replaced at far lower cost with batteries, asynchronous flywheels, dry running hydro etc.
In the Australian context, all Victoria's ten brown coal units can supply 670 MWs of energy from inertia if frequency falls by an outrageous 0.4 Hz. If the fall occured over 1 second (this has never happened in Victoria) Victoria's 1.2 GW of batteries, can supply 1,100 MWs. The battery power is soon to be 4 GW. Further once the frequency excursion has been arrested, most of that energy has to be returned over the next 20 seconds to return frequency to normal so then energy is being deducted from the primary response, The battery just keeps on supplying for hours while the fault is corrected
I admire and appreciate your persistence dealing with the junk science. thank you
1. Just because the maximum output of hydro was 6.1 GW, that is no reason to use that limit. There is 11GW of hydro on the NEM. The output of hydro in the Snowy is often limited by transmission constraints which are currently being addressed by Humelink. While small, Clover was out of action all of last year as was Barron Gorge for most of it, and Kidston will be coming on line sometime over the next 12 months. So with current hydro long before we have reached wind and solar capacity in David's models the upper limit on hydro will be around 8-9 GW. Even without transmission upgrades or Kidston it has exceeded 7GW in the past.
Then if Marinus is built another 700 MW can be added from Tasmania.
Following that and again long before we have reached David's W&S targets, Snowy II will be on line with another 2 GW followed closely by Borumba and Lake Lyell with another 2.7 GW. So by about 2033 16 GW of hydro/pumped hydro can easily supply 10-11 GW
2. David's model has enough wind and solar to supply 115% of annual demand. Twenty five years ago the NEM had almost 40 GW of thermal capacity to supply a peak of 24 GW (27 GW less hydro) and an average of 18.7 GW i.e. annual thermal generating capacity at 90% CF was 315 TWh but fossil fuels only supplied 165 TWh. Similar overcapacity can be shown in the US, UK and most of Europe. So if he is wrong and we actually have half the overcapacity as we had 25 years ago i.e. 45% rather than 15% then we easily exceed David's target and your claims are irrelevant.
3. As I have pointed out before David has ignored the 2-4 GW of V2G likely to be available by then, 2-3 GW of demand response which is being targeted and the 4-8 GW of customer batteries.
4. As for running gas and hydro while the sun is shining why not, if low wind or solar is forecast? ERCOT claims it can forecast wind and solar output better than it can forecast demand or FF output, so why can't AEMO. If there is sufficient storage to absorb the output of gas and hydro ahead of a shortage isn't that exactly what any sensible Grid operator would do. Isn't that why pumped hydro was built fifty years ago.
In summary another lot of sound and fury signifying nothing
You Said: <"David's model has enough wind and solar to supply 115% of annual demand. Twenty five years ago the NEM had almost 40 GW of thermal capacity to supply a peak of 24 GW (27 GW less hydro) and an average of 18.7 GW i.e. annual thermal generating capacity at 90% CF was 315 TWh but fossil fuels only supplied 165 TWh. Similar overcapacity can be shown in the US, UK and most of Europe. So if he is wrong and we actually have half the overcapacity as we had 25 years ago i.e. 45% rather than 15% then we easily exceed David's target and your claims are irrelevant.">
------------------------------------ REPLY ----------------------------------------------------
Using AVERAGE values is misdirection. The grid works on a second-to-second basis. The issue with NEM Wind and Solar is that it is highly intermittent. This causes 2 issues:
1) During low periods, it must be backed up with hydrocarbon generation because there is currently insufficient storage capacity to do more than make a token response.
2) During periods of high generation then much must be curtailed. David shows this, but does not show the magnitude of that curtailment. He also does not make allowances for inertia and frequency response as discussed elsewhere.
And the reason that hydrocarbons were reduced is because Wind & Solar are given precedence on the grid. It is therefore up to the hydrocarbons to compensate for changes in W&S Output.
But this is were using AVERAGES really misinforms the reader. Not only do you need to track the Average generation in a period, but you must also look at the peak generation. When you have low average generation, but High Peak generation, this makes for very expensive gas generation. You can read about it here: https://wrjohn1.substack.com/p/capacity-factor-cf-vs-lcoe
----------------------------------------------------------------------------------------------
You Said: <"4. As for running gas and hydro while the sun is shining why not, if low wind or solar is forecast? ERCOT claims it can forecast wind and solar output better than it can forecast demand or FF output, so why can't AEMO. If there is sufficient storage to absorb the output of gas and hydro ahead of a shortage isn't that exactly what any sensible Grid operator would do. Isn't that why pumped hydro was built fifty years ago.">
--------------------------------------- Reply ----------------------------------------------------
Using storage to time-shift hydrocarbon generation is detrimental. It has no impact on maintaining storage levels, but most importantly, it becomes very expensive due to storage losses. There is more information about this here: https://wrjohn1.substack.com/p/can-energy-storage-reduce-nat-gas
BTW, when I first started to work with David's model, I built "Foresight" into my simulator. However, what I found was that it made no measurable difference over time, except to increase costs. For that reason, I removed it.
Which brings up another failure in David's model:
- He does not account for maximum power rates of his storage. You can see that changes in his storage level are very rapid. This is not reasonable as all storage power has limitations set by inverter sizing and absorption rate of the batteries themselves. If you go back on look at my model output, you will see that the rate of change in storage level is not linear. As demand for storage discharge increases, it reaches a maximum power limit.
- He does not account for State of Charge Limitations.
The bottom line here is as Chris B says, David's model does not reflect reality and is effectively meaningless.
---------------------------------------------------------------------------------------------------