By Phil Kreveld
‘How long is a piece of string?’ is an equally open question. Rooftop solar capacity limits ultimately depend on investment in network upgrading.
Upgrading investment gets the attention of the Australian Energy Regulator (AER). Every extra dollar spent on upgrading will add something like 5 cents or more on network charges—if and only if the AER approves. The AER is the cop on the beat making sure that the public is not ripped off—and came into being because privatised network service providers are monopolies.
Following the gold-plating accusations in 2012 (Julia Gillard’s reign), getting approval for network charge increases is nigh on impossible. It is easy to understand as network charges are 30 to 40% of domestic and commercial tariffs. Distribution networks have huge, historical financial assets which are the basis of network charges. Acquisition costs of the assets purchased from state governments have been sunk, causing Rod Sims, the erstwhile Australian Competition and Consumer Commission tsar, to recommend that networks write down their assets. Guess what? The networks paid no attention and they continue as contented rent seekers—but for the growth of rooftop solar.
It is fair to state that prior to solar panel popularity, network asset changes were limited to increases in the geographical spread of connected dwellings, and almost entrenched diversity factors. And the AER still believes this to be a fair enough basis for network charges. Its view of ‘hosting capacity of rooftop solar’ is limited to zero incremental investment being necessary! The Australian Energy Market Operator (AEMO) wears the consequences; AEMO’s manager of systems planning raised the eyebrows of attendees at the international CIGRE conference in Cairns in 2023 when referring to ‘shaking off solar inverters’ (disconnection) to preserve grid stability (see Fig 1).
The Australian Renewable Energy Agency (ARENA) reflects this in its rooftop solar hosting capacity definition, i.e., that transmission voltage and frequency stability not be affected. Contrast this definition against AEMO’S Integrated System Plan of 2024 projecting one third of 300GW capacity being consumer energy resources (solar photovoltaic CER) by 2050. Something has got to give!
Related article: Experts say AEMO’s ‘solar switch-off’ should be a last resort
The contrast between the south-east transmission grid with some 2,500 generation busbars and distribution networks with three orders of magnitude more (think of MV feeders and hundreds of thousands of kiosk and pole mounted transformers and nodes), is mind boggling. Were you to apply the kind of power flow calculations used for transmission lines to test hosting capacity limits, the computation time would be impracticably time and resource consuming—and even more so when accounting for variations in insolation, consumption patterns, etc. However, at its fundamental level, figuring out solar PV hosting capacity—in theory—would require that! Calling a spade a shovel, distribution networks experience solar PV as—well, as a nuisance. Take a look at an extract from an earlier AEMO ISP below (Fig 2): this has resulted in the easy way out of the overall problem, i.e., stop excessive power flow from solar PV inverters.
So, here is a fundamental question: should we, with an eye to the future, do something to increase hosting capacity? Prospects for more renewable energy are suitably hazy, given the political efforts to grab the attention of the punters with nuclear generation. Even so there is a clear issue: where do you stick renewables—in transmission? Or in distribution?
It is an ‘each way bet’ as householders and business are tempted with subsidies for solar PV and batteries on the one hand. And on the other hand, new large scale generation investors are tempted with the Capacity Investment Scheme. The way it is looking is that ‘large scale’ is in for ‘the win’, and householders and business are in for ‘a place’. That leaves the constraints in hosting capacity of distribution networks open to open-ended capacity limits.
The essence of ‘gold plating’ is the investment in upgraded transformers, on load tap changers, wiring, voltage regulation equipment, protection and control so ‘that nothing can ever go wrong’. We can rest assured that will not be allowed to happen again. Therefore, hosting capacity for solar PV CER will have to be subject to restrictions. Here are just some:
- Maximum rating of panels, single, and three-phase
- Phase distribution imbalance limitation
- Topology-related maximum rating for connection
- Absorbing load requirements for CER maximum ratings
- Insolation dependent rating
- Voltage and power restrictions imposed by transformers and circuit impedance
- Programmed diversity for EV chargers.
The list can easily be expanded. Thus, limits have to be imposed—sensible limits based on commercial requirements and agreed physical limitations.
The standout in Australia is that no permission is needed for connections of household solar. In Brazil, another country with high CER penetration, a permit to connect is required and that is based on meeting available headroom in power and maintenance of prescribed voltage low and high values. It amounts to dynamic hosting capacity.
The technical issues behind hosting capacity ideally require:
- Transparent network layout, including impedance between buses and nodes, for MV and LV networks
- Transformer power limitations
- Voltage control location (capacitors and inductors)
- Location of CER and ratings
- EV charging and V2G.
The fact is that no distribution network has all this data neatly stored. And even if it had, the equation, solving a Jacobean matrix comprising of plus 10,000 x 10,000 entries, basically to solve power flows would be foolhardy. Linearising it based on small phase angle differences (and subject to sequence analysis because of imbalance in three-phase feeders) would be an order of magnitude simpler but would still raise stalagmites on the skulls of participating mathematicians.
Statistical analysis based on Monte Carlo methods is being practised and can provide useful predictions of headroom in hosting capacity. There are also semi-analytical methods including one devised by the Electric Power Research Institute (USA). It is illustrated in fig 3 and is based on radial, multi-feeder analysis using superposition based on the n-feeder node voltage, Vnn, the n-node voltage without generation, and with generation using mnn, a power-voltage sensitivity constant. However, the hosting capacity is as much limited at the edge-of-network.
Related article: We have the bull by the tail
How long is a piece of string?
We are avoiding the equivalent PV hosting question. But that is stupid, because we are piling on more and more large-scale generation to connect to HV transmission—and—doing the same thing in distribution networks with rooftop PV. The growth in domestic and business batteries sees to low but highly variable power flow. What is ignored in the emotive, highly coloured, vehement and partisan volleys that pass for ‘energy debate’, is the increasingly onerous task for AEMO to maintain voltage and frequency stability.
In a previous article, a method which would render distribution networks as ‘quasi-microgrids was discussed. Let’s assume for the sake of argument that is never going to occur, being a ‘bridge too far’, thus preserving the electrical energy market for the ‘big end of town’. The ‘goss’ is that NSW and Queensland are going to have to switch over to smart meters—finally! That would give AEMO something to monitor across the south-east corridor, not only short time increment power and energy, but also voltage and phase angles (voltage and power factor, in effect).
In essence, that could provide AEMO with a tool (based on very accurate information) to force distribution companies to dynamically control their solar PV (assuming controllable inverters everywhere). Hosting capacity comes into play if CER owners are not to become totally browned off with networks and governments, having purchased gear in good faith, only to find that is hardly ever on full power. Therefore, hosting capacity for distribution networks will have to be based on:
- Granular, time based (voltage, power, insolation parameters) heatmaps extending into the LV edge of networks
- Reassigning time-based diversity factors based on regularly reviewed load profiles on the same time basis as AEMO’s ISP reviews (initially based on assumptions of domestic electrification and EV charging)
- Only allowing authorised CER connections which are based on granular heatmaps, i.e. this will not necessarily be uniform, and last one to connect to the transformer may well find their neighbours upstream having higher ratings
- Authorised battery connection based on storage capacity and existing or proposed CER
- And, although not directly related to hosting capacity, distribution networks being forced to maintain minimum authorised power inflow.
The length of piece of string will no longer be of interest and hosting capacity will be time variant, or ‘distribution-based ISP’.
