Centralised vs distributed generation: tailor-made energy models

Print Friendly

In Brief

This note looks at the Blakers and CSIRO/ENA high renewables penetration studies. One takes a centralised approach, the other – unsurprisingly, when you look at the sponsors – takes a distributed approach. We conclude that:

  1. On the numbers, pumped hydro is cheap, but it takes forever to deploy. Overseas estimates say 5-15 years. Its cheap enough that clearly more work is fully justified. The Blakers study also explicitly formalises the concept of the levelised cost of balancing [LCOB] and this makes it a useful contribution to the literature all on its own. The LCOB is modelled to be under $25/MWh for 100 per cent renewables.
  2. The CSIRO/ENA study is difficult to work through as it covers so many topics. However we think its estimate of the amount of rooptop PV is optimistic. The best part of this study was that it did focus on distributed energy, rather than the centralised-centric approach of most other formal modelling (such as the Blakers study). It did include the valuable suggestion of a stand-alone power system tariffs for households and businesses.
  3. Neither study provides a clear pathway from today’s system to the future.
  4. In our view, the definitive piece of modelling is yet to be done. Such a model would combine the Blakers and CSIRO/ENA points of view to come up with an integrated model that provides a pathway to a high renewables penetration future and identifies the institutional regulatory steps needed to achieve it.

Another day, another model, another answer, and who cares anyway

Over the past year there have been a series of  academic reports modelling out high (>50%) renewable penetration in the NEM . The key questions such research attempts to answer are:

  1. Will the lights stay on, and the aluminium smelter/data centre keep working?
  2. What will the price of electricity be?
  3. What is the generation technology mix.

Broadly speaking, our reading of these reports suggests that the studies are broadly in agreement on questions 1 & 2. There will be enough dispatchable power and the price in the wholesale market (before adding in distribution costs) will be below A$100/MWh.

However there is little agreement on question 3. Every study comes up with its own answer. Unfortunately, a side effect is to cast doubt on the answer to questions 1 and 2. If studies don’t agree on how it will be done, how can they be sure it will work at all?

Next we would say that, in our view, none of the studies are completely convincing, and yet each offers promise and potential.

Of all the recent studies the one that is of most interest to me is the CSIRO/ENA study. Not because it’s the best of the studies. Far from it, it’s convoluted, confused and comes to some (on the face of it) implausible conclusions. However its value lies in the fact that, from the beginning, it takes a distributed energy approach. Most other studies don’t model distributed energy in detail but concern themselves with utility-scale generation and transmission.

CEEM studies

Last year the focus was on UNSW CEEM which published a summary of the academic research into the topic at that time.

100% renewables in Australia – research summary

Essentially this research ended up doing its balancing (see below) via “bio energy.”

Already in 2017 we have had two further significant contributions:

Blakers study

Firstly there was the ANU study lead by Andrew Blakers

Blakers study focus on pumped hydro

This paper had two innovations. Firstly it made explicit use of the “levelised cost of balancing [LCOB}’. This is the additional cost, beyond the actual generation cost, in a high renewable penetration system required to make the electricity supplied on demand rather than when available. Secondly, and following on from point 1 Blakers provided the first large-scale estimate of the cost of doing the balancing in Australia via pumped hydro.


Next,we had last week’s CSIRO/ENA study, which concluded that rooftop PV could provide as much as 20 per cent of electricity energy by 2050 based on over 70GW of distributed PV compared to 5.5GW today.

Figure 1 Rooftop PV projections. Source: CSIRO ENA

Figure 1 Rooftop PV projections. Source: CSIRO ENA

In our view, that’s a big ask and would require PV on a lot of business rooftops as well as residential. If we assume that townhouses, flats and highrise apartments don’t have PV, there are about 7 million detached houses in Australia. Lets assume 75 per cent of them end up with an average 5kW of PV, that’s just 30GW in total, leaving around 40GW to be picked up from the commercial market. How realistic is that? By contrast, the Blakers report estimated 17GW of rooftop PV

Pumped Hydro

We added up the capex implied to take the NEM to 100 per cent renewable using the generation mix modelled in the Blakers report and got to $190 billion. Of that, about 25 per cent, or $49 billion is pumped hydro storage.

Figure 2 Capex to rebuild NEM to 100% renewables. Source: derived from Blakers report

Figure 2 Capex to rebuild NEM to 100% renewables. Source: derived from Blakers report


Figure 3 Cost of total rebuilding NEM to 100% renewable. Source: ITKe derived from Blakers Report

Figure 3 Cost of total rebuilding NEM to 100% renewable. Source: ITKe derived from Blakers Report

We calculate that an electricity price of say $100/MWh is required to justify $190 billion of capex, assuming constant electricity volumes of 200TWh and all-up operating costs of $25/MWh. These numbers are right around those reported by Blakers.

Figure 4 IRR on $20 bn real annual revenue and $190 bn of capex. Source: ITKe

Figure 4 IRR on $20 bn real annual revenue and $190 bn of capex. Source: ITKe

This model is just a back of the envelope check.

Using the same format, but just looking at the $56 billion cost of the pumped hydro and transmission alone, produces something like $20/MWh as the LCOB – maybe even less.

The pumped hydro storage assumes 31 hours of storage for every MW of capacity; much higher than most existing projects.

The Blakers report used pumped hydro capex of around $A205/kWh ($0.8 m per MW in equipment + pipe costs + $70 KWh of storage (dams)).

These numbers are broadly comparable with the preliminary capex estimate for the Kidston pumped hydro project.  Kidston preliminary capex estimate is $300 million and using the Blakers  parameters gets us to around $305 million.

Figure 5 Pumped hydro capital cost estimates. Source: Blakers Report, Genex

Figure 5 Pumped hydro capital cost estimates. Source: Blakers Report, Genex

As an analyst who has seen many capex proposals, I’ll not be in the least bit surprised if the final cost is higher than that. Kidston had cheap land costs but only has $200 million of “head”. Blakers modelling is based on a “head” of $600 million.

Recent global pumped hydro capex costs are broadly in line with the numbers quoted above, at least based on the capex $/KWh of energy/electricity.

Figure 6 Recent pumped hydro storage projects. Source: Adapted from Canaccord Genuity, Genex research Mar 2017

Figure 6 Recent pumped hydro storage projects. Source: Adapted from Canaccord Genuity, Genex research Mar 2017

Total  pumped hydro capex estimate

The Blakers report broadly envisages 490GWh of storage required for balancing the wind investment. Unusually, this breaks down between 16GW of  power capacity and 31 hours or so of storage. 31 hours storage compares to Kidston at say 6 hours. On our calculations this means about $49 billion of investment in pumped hydro storage.

Figure 7 Total investment in pumped storage. Source: Calculated from Blakers data

Figure 7 Total investment in pumped storage. Source: Calculated from Blakers data

Blakers view of transmission cost

The Blakers report models a HVDC transmission backbone from Queensland to South Australian that more or less parallels the existing AC transmission backbone. To quote:

“We envisage sequential construction in the 2020-2030 timeframe of several independent overhead HVDC interconnectors, each 700-1500 km long, through flat country with good access (inland Australia). The interconnectors operate at a voltage of ± 800 kV and have nominal power capacity of about 6-7 GW. On the basis of similar systems constructed around the world we estimate a cost of $400/MW-km” Source: Blakers Report

The transmission backbone “notional” map shown in the Blakers report runs from North Queensland to Adelaide. Using 2500km of Brisbane to Adelaide and the Blakers estimates we end up with $8 billion of transmission costs.

Figure 8 HVDC transmission backbone. Source: derived from Blakers Report

Figure 8 HVDC transmission backbone. Source: derived from Blakers Report

Combined pumped hydro and transmission less than distributed battery cost

If we add the pumped hydro and transmission capex together its $56 billion for 496GWh of storage.

Figure 9 Pumped hydro compared to batteries. Source Blakers, Avicienne, ITKe

Figure 9 Pumped hydro compared to batteries. Source: Blakers, Avicienne, ITKe

However we don’t think that’s the end of the story. Lithium storage is very fast to deploy, can provide functions such as frequency control that are more difficult for pumped hydro, and the pumped hydro approach needs to be proved up.

That is to say there is no doubt pumped hydro works, but there is doubt about whether 16 GW can be built as easily as that. Speed to market is a massive advantage for batteries. In our view the modular nature of batteries and the fact that they can be deployed next to the load are also significant advantages. Its not just transmission costs but also distribution costs that can be minimised with batteries. Pumped hydro can take 5-15 years to deploy.

That said. the numbers for pumped hydro are interesting and clearly worth further investigation.

A tariff to encourage distributed energy

There seems little doubt that before considering cost, an intelligent grid with resilient distributed generation and storage is going to be superior on any metric, to a centralised one-way grid dependent on the function of a few central generation units and limited high voltage transmission markets.

Such a grid will be fault resilient, and many suppliers and many consumers typically make for fairer prices and customers are likely to feel more in control.

That’s why the Energeia/ENA proposal to design a “stand alone power system tariff [SAPST]Energeia – Roles and incentives for microgrids as previously reported by reneweconomy is of such interest. In this scenario, networks would offer a discounted tariff for the right to disconnect a customer for a few hours each year.

Once again, though, this points to the ringfencing between networks and generators as a fundamental problem to reform. Any such tariff would be far better if it incorporated a generation as well as a network element. The other problem is that, at the moment, networks have plenty of spare capacity. They just don’t have the incentive to offer such a tariff.

David Leitch is principal of ITK. He was formerly a Utility Analyst for leading investment banks over the past 30 years. The views expressed are his own. Please note our new section, Energy Markets, which will include analysis from Leitch on the energy markets and broader energy issues. And also note our live generation widget, and the APVI solar contribution.  

RenewEconomy Free Daily Newsletter

Share this:

  • trackdaze

    Storage ought to include a proportion electric vehicle batteries being available to the grid.

    We need to both. With hydros long lead times the sooner they get started the better.

  • MikeH

    >there are about 7 million detached houses in Australia

    The issue here David is that there are likely to be a lot more than that by 2050 at the pace that we are building suburbs and solar system are getting bigger. That would make the number a bit more feasible.

    Good article.

    • I’d also argue that not all apartment buildings are the same.

      A 20, 30, 40, 50 level tower, sure, very little roofspace compared to the number of people (number of apartments) under the roof, but these are, and always will be the exception to the rule.

      There’s approximately 1300 medium/high-density apartment projects in the pipeline in Melbourne which translates into about 130,000-140,000 dwellings.

      50% of them are projects with 6 or less floors. 30% of them are projects with 4 or less floors.

      In this ‘goldilocks’ range of 3-6 levels (goldilocks because it’s not too high to frighten the locals and it’s not too small (underdevelopment – like detached houses) and this is the scale many many many more existing Melbourne suburbs are going to see apartment development in future), the projects generally have flatter roofs, the footprint of the building will be bigger (less setback requirements for shorter buildings) resulting in a greater amount of space/potential for whacking panels on the roofs. Yep, it’s now going to be the same KW to person-in-dwelling ratio as a detached house, but it’s not an insignificant amount of space either.

      • George Darroch

        Let’s not forget the large covered car-parks that many of these have (such as my own unit). Once local energy becomes a priority many of these will be utilised, and some more roofs may even be constructed.

  • Malcolm M

    Some of these investments are synergistic. For example Snowy 2 is useless without ~$2 billion in additional transmission capacity, because the current Snowy 1 already has the generation capacity to max out transmission capacity to both the Sydney and Melbourne markets. To be useful, an additional transmission line with at least 2 GW capacity would be needed from the new Tumut 4 pumped storage hydro to the NSW 500 kV network at Bannaby (near Goulburn), and to join the Victorian 500 kV network just north of Melbourne. This additional transmission line would do some of its own balancing even without Tumut 4 operating, because it would take excess daytime solar from NSW south into southern Victoria and Tasmania, and take excess wind power from SA and southern Vic into NSW.

    • George Darroch

      It’s well placed to balance Australia’s renewable energy generation, and the transmission will be needed at some point soon anyway.

      Let’s get on and build it.

      And what’s with the timeline of 15 years for pumped hydro? The entire Snowy Scheme was completed in 25 years, about one dam every 18 months.

    • Jonathan Prendergast

      I prefer an approach of optimising our current inter-connectors, but building local energy and limiting inter-state flows to when entirely necessary (peak demand, min-local-gen), rather than further consolidating the centralised system.

      Don’t have time to analyse and back it up with numbers though

  • Kevan Daly

    Yes, I like the idea of an LCOB that the Blakers et al study includes. Too often the extra transmission costs associated with intermittent, distributed sources is ignored.
    However the net result for grid consumers isn’t very encouraging with the cost being about 100 (wholesale) + 25 (distribution) + 25 (LCOB) = 150 $/MWh before retailer margin. Off-grid would be looking very attractive.

    • Peter F

      $100 wholesale isn’t going to last long. $55-75 for wind and solar today means $40-55 in 5-6 years time, existing hydro can do $30 so by 2025 you can have a weighted average including legacy PPA’s around $60 plus the LCOB. As you say at $175-180 retail, off grid looks very attractive therefore if the large generators and networks want the business they will have to figure out a way to get it to customers below $100-120 or 10-12c/kWhr for medium customers, lower for very large and not much more than 20c for domestic

  • Helen

    Is it also worth pointing out that you don’t have the risk of storms levelling the long
    transmission lines with distributed energy.

  • Peter F

    I agree that every model is incomplete and there are always unexplored options. It is complicated by the fact that the facts are changing so fast that initial assumptions are wrong by the time you complete a study. My particular criticism of both studies are that neither of them has given sufficient attention to the following opportunities.
    1. Energy efficiency. If we were to match Spain, California or Germany we could cut energy use by 30-40%
    2. Low wind wind turbines and tracking solar generate for a much longer fraction of the time and over the years can be spread more widely across the nation so that both the gap between demand and low wind condition will be smaller and the duration of storage needed will be less. You can make a good case that you can get by with about half the energy storage suggested by Blakers.
    3. Power to heat and ice at the premises has multiple benefits, Lower peak demand, lower line losses and higher plant efficiency among others and typically costs about 1/5th to 1/10th of batteries.

    If all these options are pursued with widely distributed renewables not only will we need less grid storage but probably very little extra transmission infrastructure.

    Just as we can pick holes in the detailed studies you referenced but the good news is that almost every time we find an error, there is a way to solve the problem at a lower price than used in whichever study we are criticicising

    • Mark Diesendorf

      Although low-wind wind turbines generate for a larger fraction of time, they generate very little power in low winds, because wind power is proportional to the cube of the wind speed.

  • ShadowBoxer

    I’m curious as to what work – if any – has been done on the risk of Pumped (fresh water) hydro in Aus conditions – esp with Climate change projections included. Over the decades I’ve hiked around the area I’ve seen the Snowy get very low a few times. Pumped hydro still requires some water. Given lows often correlates to Late summer, heat – power usage peaks – it seems to me that there is a major issue with security of supply (given the high capital cost) … and at a time when irrigation water supplies could become extremely critical. Whilst obviously we wont live in continual drought – we are seeing significant multi year droughts. Remember when NZ lost 25% of their power supply to drought.