By Phil Kreveld
Are the brightest minds enough to make Australia’s transition to an AI and data centre economy a national asset? It’s a tautologous question. However, the many seemingly independent parts of the ‘national energy pie’ provide serious challenges for those brightest minds when it comes to smoothly connecting data centres to the national electricity systems.
Tim Edwards, managing director of CutlerMerz—an engineering and economic consulting firm supporting the power and energy sector— closed the public discussion held on 16 September with the above phrase, expressed as a hope.
The discussion panel his company had assembled to bring its expertise to bear on the integration of data centres into the national electricity markets of the south east and west, represented leading thinkers and planners on Australia’s electrical energy networks and generation.
Related article: Spotlight on the data centre debate
Edwards opened the proceedings with the following:
“Australia’s trying to answer one of the most important questions around infrastructure of the coming decade. Can we become a major digital infrastructure economy without compromising affordability reliability and the energy transition and if so, what is the bargain that makes that possible. [It is] tipped to potentially increase electricity demand by more than 30% of the coming decade.”
The participants in the discussion panel were:
- Dominique van den Berg, Chief Executive of Energy Networks Australia
- Tim Jordan, Commissioner at the AEMC
- Eleanor Wood, Director of Networks and Pipeline Policy, DCCEEW (NSW)
- Mark Ragusa, General Manager Strategy, Ausgrid
- Kane Thornton, Head of Strategic Impact, CDC Data Centres.
In listening to the exchange of thoughts relating to generation and network requirements to accommodate an additional 54TWh of annual consumption by 2035, three critically important items emerged:
- Jurisdictional interests in conflict with those of the Commonwealth Government, impeding datacentre integration;
- Lack of a clearly-defined investment climate for additional generation capacity to meet datacentre future energy demand; and
- The conflicts between economic and engineering rationales.
The thoughtful presentations by Tim Jordan and Eleanor Wood provided not only very clear expositions of datacentre integration hurdles to clear but also exposed a Gordian knot—not specifically introduced by data centres but rather one that results from the market-driven philosophy governing national electricity generation and networks. Note that this is a conclusion of the reporter covering the meeting.
New generation to service additional datacentre demand was highlighted by Tim Jordan. This is in line with the Commonwealth Government’s requirement. Newly minted REGOs (Renewable Electricity Guarantee of Origin) certificates have to be acquired by datacentres. They replace the large-scale generation certificates which cease to exist after 2030. He opined that in the absence of new generation investment, there might be the possibility to ‘soak up’ REGOs from underutilised renewable generation, and also that Tim Nelson’s Electricity Services Entry Mechanism, designed to provide a ‘floor’ for long term generation investment, might birth genuine new REGOs.
Eleanor Wood spoke to the network capacity constraints and itemised the following:
- Shopping for connections in multiple network locations and therefore providing artificially inflated network capacity projections;
- Transgrid’s current 30 terawatt-hour annual NSW transmission forecast to increase to 80 terawatt-hours by 2036;
- Preventing datacentre connection costs flowing on to other consumers by way of additional network charges;
- Connection fees of the order of $100,000/Megawatt;
- Network constraints necessitating Sydney Ring South additional transmission from near Goulburn to south west Sydney, and
- To consider less congested corridors, for example Wagga Wagga.
Excerpts from the panel discussions
Eleanor Wood: “there are some incentives for data centres that are quite perverse at the moment around capacity holding and you know [that] speculative applications in multiple locations where the cost of entry is just not of enough consequence—there’s also things like non-disclosure agreements being signed the prevent the networks from sharing information with each other, which is just not good from the planning perspective—aggregate numbers are so uncertain. It is there [that] it’s going to be very difficult to pull the trigger on the major network projects.”
Tim Jordan: “Forty percent of data centres that were in AEMO’s tracker either dropped out or regressed in their connection process—that’s very hard to have that level of uncertainty.”
Mark Ragusa: “The customer wants certainty but also we don’t want to be sitting there with a potential ramp-up that may take longer than expected and latent capacity just sitting there that other customers could actually be taking advantage of.”
Dominique van den Berg: “I think it’s pretty clear for data centres in Australia to learn the lessons that [there] cannot be a price rise in household and business cost, and so getting this right from the outset makes it kind of sense that network cost in particular is borne by data centres.”
Kane Thornton: “[The] way to resolve that is to ensure data centres that [are] covering the full cost, covering the additional generation and covering the full cost at ‘connect’ and access the network [is] straightforward if that’s not the case at the moment then fix it and make it happen. You see companies today that have never built a data centre for multiple gigawatts. I think we need to accept plenty of these won’t eventually be realised.
“[Data centres are like] the renewable industry 10 years ago—hey it’s really windy here, let’s build a wind farm here and can someone wander down the road and tell that grid business we need to plug in—so I mean I think generally a lack of probably understanding of the complexity. We build our data centres expecting a lot of flexibility and understanding that in three, five or 20 years-time the expectations about customers the way in which they’ll use their facilities [will have changed significantly].”
Tim Edwards: “Can we actually split an upstream transformer augmentation into 20% for mums and dads, and 80% for data centre and is that where we’re headed if at a principle level we’re agreeing that data centres will bear that risk? If so, how do you distinguish upstream cost?”
Mark Ragusa: “The bigger challenge is actually how do you recover fairly across that entire customer base so ultimately it’s not punishing the first mover [data centre to connect].
“The difference between managing for peak versus managing for energy is hugely different and there’s plenty of energy availability in the system even if peak demand pressures exist so flexibility gives you gives you an avenue to actually maximise the throughput of that energy without having to reset demand. So, for a distribution business it’s highly valuable to make the most of that flexibility. I guess that the risk goes back to where the data centre sits in their need for high availability and how they continue to service that availability.”
Tim Jordan: “If it turned out that you didn’t have to build any additional renewables to meet data centre load then the price of those renewable REGOs would be very low so—the REGO price would be very low so yeah it should work with the natural flow the market I think we’re creating.”
Kane Thornton: “It’s a given that future data centres will bring with them new supply I think we need. There’s devil in the detail here and this is the work to be done. Minister Bowen put some markers down around from what date forward, data centres [would need] new and additional generation. All those things need to be worked through. We do need to remind ourselves that today data centres are only two or three percent of the energy system—maybe they’re going to six or 10%.”
Tim Jordan: “For not terribly flexible loads if there’s a switch off it can cause significant disruption to the grid and so giving the operator (AEMO) the inside into how they behave through disturbance is really important.”
Kane Thornton: “For full ride-through work that’s being done for example but our ability just to say right we’re going to dial it down for a few hours [is] deeply problematic and back-up generation now [takes care of this]—this is an inconvenient truth our customers and our data centres cannot go down under any circumstance. We have highly redundant network connections to multiple points to the strongest part of the grid we have UPS systems on site and the very last resort is diesel generation.
Related article: Grid weakness stuffing up data centre connections
“I don’t think it’s the right outcome to necessarily require a data centre to put for example a BESS on site. Now for some sites that may well be a sensible outcome and be effective but equally if there’s an obligation [that is different]. I think New South Wales uses the word ‘proximate’ which still I think, requires some level of clarity. I think if there’s an obligation in the requirement that a data centre has to contract with a certain amount of firming whatever and however that’s defined, approximate to the data centre then let us know what that is, let us know how it works, what the expectation is.”
In conclusion, a great deal of engineering as well as economic appraisals are necessary. Removing jurisdictional boundaries impeding whole-of-network appraisal would appear a good start.
