[The] learning rates incorporated into the AETA are primarily based on the Global Local Learning Model (GALLM) model developed by CSIRO’s Energy Transformed Flagship Group". )
It then also states that the GALLM model "assesses a number of factors to establish the learning rate for each technology based on: technology maturity (i.e. its progression on the learning curve); expected rate of technology deployment; and rate of cost reduction (with deployment)."
Using the rate of deployment means cost projections are dependent on the cumulative installed capacity of the technology.
BZE also use the same method for our assumptions on future costs of solar thermal. Deployment is what drives cost reductions, not the passage of time.
Unfortunately in the AETA, the “expected rate of technology deployment” for CST is not specified anywhere. This rate would obviously be affected by a decision to go to 100 per cent renewable energy in Australia. The AETA’s cost reduction is only presented on the basis of time, not deployment.
BREE and its research partners have presumably picked a deployment scenario as an assumption to determine these cost reductions – but they don’t say what it is.
If BREE’s projections were to be used by AEMO for its 100 per cent renewables modelling, it will be a false outcome. BREE will have effectively provided an answer based on a non-100 per cent renewables scenario for AEMO’s 100 per cent renewable modelling.
It’s possible to guess at what assumptions BREE have made. We know the GALLM methodology assumes a “learning rate” of cost reduction for CST at 14.6 per cent.
Assuming this 14.6 per cent learning rate, the cost reductions in the AETA (about 40 per cent by 2020) imply that by 2020, the worldwide installed CST capacity would be in the vicinity of 14 gigawatts.
This estimate may not be unreasonable based on business-as-usual now – but it would be significantly altered if a country (such as Australia) built a 100 per cent renewable energy grid, with a large amount of CST. This would have a substantial impact on global cost reductions (as outlined in BZE’s recent report Laggard to Leader: How Australia Can Lead the World to Zero Carbon Prosperity).
Equally, accelerated and substantial deployment worldwide would have the same effect.
Using the same extrapolation from GALLM methodology, BREE’s cost reductions for CST in 2050 would appear to be based on a total of 25GW of global capacity – which seems unreasonably low. For example, Saudi Arabia alone has announced a goal to install 25GW of CST over the next two decades.
If a BZE-style plan was adopted, with 42.5GW of CST being rapidly deployed in Australia, then the cost reductions would be substantially higher and faster. If the global installed capacity reached 80GW, then the cost reductions would be approaching 60 per cent, or about $120 per megawatt hour – and we hope this happens long before 2050!
Various estimates of learning rates for CST vary between 8 and 15 per cent. Some organisations have suggested 18-20 per cent may be achievable based on "related industry precedents" (such as for solar photovoltaics).
The US National Energy Modelling System uses a learning rate of 20 per cent over the first three doublings of cumulative capacity. For the next five doublings the rate is 10 per cent, and then for the rest of the cumulative capacity the rate is 1 per cent per doubling. If you were to assume this steeper learning rate, then even greater cost reductions are possible.
The AETA is more realistic than previous government estimates such as the Energy White Paper, in that it sees a significant role for some renewable technologies. However, it is seriously flawed on exactly the technology that is most crucial for a 100 per cent renewable energy future, CST.
When AEMO models a 100 per cent renewable energy scenario for 2030 and 2050, we hope it goes back to the drawing board for its CST modelling.