The draft 2022 Integrated System Plan provides a dramatic break between the National Electricity Market as it used to be, and the NEM as it’s going to be.
In this note, ITK takes a first look at the implications for pool electricity prices of the Step Change least cost model provided by the ISP. Our model does not account for reliability, capital costs or system services. We assume the ISP modelling has built a least cost system that meets the reliability requirement.
What we look at is average daily prices for flat load, wind and solar, and next time round for storage, under close to Step Change ISP capacity assumptions.
We build up utility wind and solar region by region exactly as provided in the ISP from FY24. We run down coal just a little slower than assumed in the ISP. We build storage and build marinus link.
We find flat load prices rise post 2030, but we see room in a second iteration of our model for those to fall, as we get better at building storage in our model. Equally, our wind production is below that assumed in the ISP despite having identical capacity, partly as a result of curtailment in our model and partly due to an issue with capacity factors that remains to be addressed.
Our modelled returns to wind are above LCOE (the levellised cost of energy) and so probably will have to be reduced in the future or new capacity will be built.
Our modelled solar prices are below LCOE, and so either less solar will be built, or more storage will be built, reducing gas and lifting middle of the day prices a bit, and/or some hydrogen capacity will be built.
Equally, we have increased our demand growth from 0.4% pa to 1.6% but we have not shifted the daily pattern of consumption. Over time, we can expect demand patterns to adjust to spot price opportunities: Not just storage but EV charging and so on. This, too, remains as a future exercise. In short, lots more to do.
Still, heading into 2022 this is where we sit.
Clients can access the year by year prices by region, for flat load, wind DWA and Solar DWA. Quarterly forecasts are unchanged as they are modelled separately to the annual numbers.
Implications for ITK price forecasts
The draft ISP was released in December, 2021, and adopts – based on industry consensus at the time – an “aggressive” pace of change for supply and demand in the NEM.
A couple of years back ITK regarded our forecasts calling for 50% VRE (variable renewable energy, or wind and solar) in the NEM by 2030 as likely, but equally well ahead of industry consensus. Today, our most recently published numbers are well behind the step-change scenario (the new central scenario) of the draft ISP.
The ISP is at its heart a transmission planning document, and except in the area of transmission has no prescriptive force or policy outcomes. And despite this year’s consensus on Step Change or the even more aggressive hydrogen boost to demand and supply, there are a range of potential outcomes.
As ITK is interested in the outlook for electricity prices and what is and is not priced into the market, we look at the potential outcomes from that perspective.
AEMO has released an excellent spreadsheet of supply. However, as yet the demand numbers haven’t been produced. Therefore, we can’t easily see what the transmission flows are and the export import data for each region. Those interstate transmission drivers are important to prices. With that caveat in mind….
The overall total supply forecast and hence demand including supply to storage across the NEM has a CAGR (compound aggregate growth rate) of 1.7% in the step-change scenario, at least through to 2035. That compares with ITK’s existing forecast of 0.5%.
Even more interestingly is the divergence between regions. The ISP model shows a significantly reduced share of total supply coming from Victoria falling from 26% in 2024 to 19% in 2035.
The growth in Queensland comes despite, in this scenario, any hydrogen production.
As Qld has the largest growth in supply we look at the composition. We do that in terms of energy (TWh) rather than power (GW), because we see little point in comparing technologies that have different capacity factors and accomplish the task in different fashions.
For the scenario, AEMO’s model results in a strong steady growth in solar at both utility and behind the meter level, but the real super charging comes from wind:
Again this is a somewhat surprising result. In this scenario 10GW of wind are added, 3.5GW of utility solar and 6.4GW of coal is retired by 2035.
We also observe that the outcome of the ISP is that the majority of new wind and utility and solar is built in the Darling Downs area. Considering that this region is also the Coal Seam Gas area or close to it, it may well be the place to own a house or a shop. But equally there will be a strong social license required.
What about Victoria?
An outcome of the ISP step change modelling is that all of Victoria’s brown coal is gone by 2032. This too will require a massive social adjustment. Only about 1/3 of the brown coal energy is replaced by wind and maybe another 1/5 from mainly residential solar.
In terms of keeping the lights on in Victoria, this shouldn’t be a worry as the AEMO models out the reliability requirement and ensures its met. However, it’s an open question whether politically Victoria wants to become the new NSW: Importing from South Australia and Tasmania both of which are modelled to significantly grow, and also perhaps taking a bit from NSW.
Initially surprising, the QLD and Victorian results make partial sense from the perspective that QLD solar will always be lower cost than Victorian solar and the inflexibility of coal sees it go away. The remaining element of surprise is that only 5 GW of wind are built in Victoria but 9 GW in Qld.
The main REZ (renewable energy zone) for new wind in Victoria that is identified in the ISP is South West Victoria. As compared with Darling Downs in Queensland, South West Victoria has a better wind resource, but it’s less correlated with demand. The build cost is fractionally higher but in the end the size of the resource is just smaller at a bit over 3 GW compared to Darling Downs at 5.5 GW by 2040.
It can be seen from the table below, which uses 2035 forecast by REZ of installed capacity, that Darling Downs, Orana, New England and South West Victoria are going to be the big sources of wind energy. The first three also get significant solar development. The listed REZs are about 75% of total solar and wind built in the model. The wind development is more diverse than solar.
It’s unlikely that development will follow exactly this profile, but we can expect that as REZs become established in each NEM region then the preponderance of wind and solar will be encouraged into those areas.
So, in ITK’s opinion this is the where the generation will be built, even if the pace and technology mix varies from the ISP assumptions.
As an aside it it “nice” to see far north Qld wind resource does get exploited.The largest three zones account for about 40% of total development to 2035 and the top 5 zones 50%.
ITK’s price forecasts with the ISP scenario
In our opinion it’s great to have the ISP information available, notwithstanding that it’s unlikely that actual development will follow the ISP, since it is a model of a least cost optimisation and takes no account of domestic politics, social license or many other issues.
The ISP still represents a great guide as to where new, but as yet unannounced generation will be located. Equally, at least for the next decade or so the cost modelling, and the capacity factors and correlation with demand indicates that more wind than utility solar is likely to be developed. Again we choose to accept this guidance from the modelling and incorporate it into our forecasts, region by region.
Also we have reduced minimum generation by thermal plant and closed coal plants early, broadly in line with the ISP step change scenario. The ISP does not explicitly model “system services” mingen requirements but other developments in technology, particularly grid forming inverters, provide, in our our opinion, enough confidence to suggest that within a year or two this wont be an issue.
Equally, and to reiterate, the ISP modelling meets the reliability (unserved energy) standard as modelled by PLEXOS on a statistical half hourly basis. So that relieves ITK of our need to take care of that.
ITK also thinks it likely that there will be some hydrogen development, enough to shift the demand and supply needles a bit.
We expect that hydrogen demand to initially be in QLD and point to the Stanwell consortium and that it has already “agreed” I guess subject to hydrogen FID to buy the output from the 600 MW Aldoga solar farm, and should phase 1 go well there is an order of magnitude scale up.
The price outcomes from our modelling are, fortunately, kind of what we imagined they would be.
What does it all mean?
The flat load forecasts look not unreasonable, to our eyes. There are higher prices around the early 2030s as the system adjusts to essentially no coal.
The wind prices are above LCOE and this will incentives more wind to be built than modelled. However, that will also require an even larger transmission and connection effort. We expect to revisit this again at a later stage during 2022.
The solar prices are too low. In the case of QLD we expect some hydrogen production to improve midday demand and bump solar prices up towards LCOE. It seems the only way.
Previous versions of our “merit” based model have set the dispatch order of generation on the basis of variable cost, lowest variable cost first to highest variable cost.
The price that a fuel bids in is set to be just below the variable cost of the next highest cost fuel. However, in the case where there is excess capacity we model that margin to change with capacity utilizations.
The classic example is coal and gas. Gas has higher costs than coal and so the most the coal bid can be is $1/MWh less than gas cost. The lowest the coal bid can be is its variable cost (although we also include an assumed maintenance cost).
This provides a high for a coal bid of say $100/MWh and a low of $40/MWh (depending on fuel cost assumptions). Where the coal bid price sits in our model relative to this range depends on our estimate of coal capacity utilisation.
However, much of that goes out the window when there is little or no coal. Wind has higher operating costs than solar but both have a zero marginal cost.
When there are no mingen limits then there is no coal in the middle of the day and at times solar is supplying all or most of the demand and prices are low. As the solar goes away the wind, on average remains generating but is not enough to supply demand.
Therefore, the price is set by the various forms of firming capacity. Hydro is the lowest cost form of firming but also has limited and fixed supply. Traditionally its been the highest price form of energy in the NEM and we have left it that way, save that we build Marinus link and so that allows Tasmania to export significant hydro and on average it can do some of that each day.
The rest of the firming market is split between storage and gas. The gas costs have a long history of being modelled, and are unlikely to change significantly unless there is either a carbon cost or a fundamental shift in the cost of gas, but the storage costs and required margin are still subject to a high level of uncertainty.
We have modelled in lots of storage and assumed its bought out of the midday market at a close to zero average cost and then dispatched in the evening at a bit less than gas particularly post 2025, but this is still very much a working assumption.
In short, whereas in the past much of electricity price modelling reduces to a “calculation”/”guesstimate” of how much pricing power coal generation has, in the future much of the modelling will be about “calculation”/”guesstimate” of storage costs, margins and bid prices.
We advise clients, that although this is not a “back of the envelope” calculation but neither is it likely to be our last go at it. This note represents the first pass of adjusting our model to the new paradigm in Australia. Nor do we look at all at quarterly pricing in this note. Its about the medium term.
2035 as a sample year
Assumptions
Demand growth has been changed across the board from 0.4% per year to 1.6%. This reflects “electrification”, including electrification of heat processes, electrification within the residential sector, some EV growth and some hydrogen.
We don’t break any of this out separately, just incorporate it into a black box assumption. One we can change later. Of course the lower the demand growth the lower the price given that new supply in our model does not respond to changes in demand. That is the key weakness of ITKs modelling we fix supply and let price adjust to demand. Most other models build supply to satisfy a modelled level of demand.
We did not look at gas and coal cost assumptions for this note, so they are unchanged from prior. Minor adjustments have been made to rooftop but we don’t show those. ITK’s capacity factor assumptions, particularly for wind appear to be a lower than those apparent in the ISP presumably raising costs, but will have to be reexamined later.
State inter-connector capacity is lifted in line with ISP limits including the building of Marinus Link. However more work needs to be done on looking at medium term effective capacity between NSW and Victoria and Queensland and NSW. From 2030 VNI effective capacity South rises to 3148 MW and QNI South to 2148. Given Victoria’s new status as a net importer the balance of North South from NSW will need to be re-examined.
