As far as we can tell, nobody has published total cost estimates for Western Australia’s Wholesale Electricity Market (WEM).
This is most likely due to the significant (but opaque) role of bilateral contracting in the energy market, which makes up >80% of energy traded in the WEM. Because the details of these contracts are not made public, one can only estimate the cost of bilaterally traded energy.
However, there is enough public data available to make reasonable enough estimates, which is what we’ll attempt to do here.
WEM trading volumes
The table below provides an overview of the volumes of energy traded in the WEM over the past four calendar years, with a breakdown of the proportions traded via bilateral contracts, the day-ahead Short Term Energy Market (STEM) and the real-time balancing market (along with average balancing prices).
Of interest are the relative shares of bilateral, STEM and balancing volumes in the total amount of energy traded. While total consumption has remained fairly flat in the last 4 years (fluctuating between 17.2 and 17.8 TWh), the share of bilateral contracts in the total mix has fallen steadily from 95% in 2017 to 83% in 2020. In the meantime, the volume of energy traded on the balancing market has increased its share from 1.4% in 2017 to 12.9% in 2020.
What price to put on bilaterally traded energy?
Bilateral contracts are inherently private transactions where the terms are known only to the counterparties. However, a regulatory scheme imposed by the government after the re-merger of Synergy and Verve Energy in 2014 to mitigate the risk of anti-competitive behaviour (called the Electricity Generation and Retail Corporation Regulatory Scheme) sheds some light on the range of prices one could expect from bilaterally traded energy.
Synergy is required to publish information on its Standard Products, which since 2017 has applied the same Energy Forward Curve to set transfer prices between Synergy’s Generation and Retail arms. Using the Standard Product historical pricing data, the following average annual prices are calculated:
The average price for bilateral contracts will most likely fall somewhere between the FLAT BUY and PEAK SELL prices. Taking the midpoint as a rough estimate of the average bilateral price and comparing this estimate to average balancing prices indicate that the prices broadly line up:
Average bilateral price estimate
Average balancing price (volume-weighted)
Estimated average bilateral prices vs average balancing prices
Total wholesale market costs
A breakdown of the wholesale cost components in the WEM is shown in the table below:
Reserve Capacity Mechanism
Load Rejection Reserve
Breakdown of total WEM costs (Calendar years 2017-2020) [Sources: see Appendix below]
(*) Ancillary services costs for 2020 are estimated as the latest Ancillary Services Report has not been released at the time of writing.
Note that the table above excludes the following cost components that would also be included in an end-user’s electricity bill:
Western Power transmission and distribution network access fees (including metering)
Renewable and environmental government policy costs, such as the renewable energy target
Electricity retailer (e.g. Synergy) fees
Unit wholesale cost breakdown
Finally, the unit cost breakdown (in $/MWh) can be calculated by dividing total costs by total consumption in the WEM:
Reserve Capacity Mechanism
Load Rejection Reserve
Breakdown of unit wholesale costs in the WEM (Calendar years 2017-2020)
Western Power then deployed a SPS on two properties left without power after the Wooroloo bushfires, while being able to remove 2km of distribution lines to the properties. This marks the first time an SPS has been installed in the Greater Perth metro area.
PowerBank community battery trial extended
Western Power and Synergy’s community battery trial (PowerBank) has been extended to nine communities in metropolitan Perth and regional WA. The trial will allow household participants to access a community battery and “bank” excess solar from their rooftop PV systems for deferred use in the evenings (with a cost of $1.20 or $1.40 a day for 6 kWh or 8 kWh of storage respectively).
Launch of Virtual Power Plant trial Project Symphony
The WA government announced the launch of a $35m virtual power plant trial dubbed Project Symphony. The trial, which is being jointly implemented by AEMO, Energy Policy WA, Synergy and Western Power, will coordinate rooftop PV, batteries and loads from 500 homes and businesses in the Southern River area.
Merredin Solar Farm secures PPA with BHP Nickel West Kwinana
On 25 February 2021, the AEMC made a final determination on the establishment of DER Technical Standards to allow AEMO and DNSPs to better manage the increasing penetration of rooftop PV in the system. The DER Technical Standards will include the requirements set out in the newly updated AS 4777.2:2020. The rule will commence on 18 December 2021 and only apply to new or altered connections in the NEM. At this stage, it isn’t clear whether the same standards will be adopted in Western Australia.
Parmelia Gas Pipeline converted to be “hydrogen-ready”
Now that 2020 is finally behind us, let’s dig into the data and pull out the latest estimates for rooftop PV capacity in the South West Interconnected System (SWIS), Western Australia’s largest electricity grid.
Here are the headline numbers:
Rooftop PV Capacity (MWp)
Total small-scale capacity
Total large-scale (>100 kW) capacity
Total rooftop PV capacity
Despite COVID-19, there was robust growth in small-scale (mainly residential) rooftop PV installations in 2020, with over 295 MW installed compared with 240 MW in 2019 (a 23% year-over-year growth).
Meanwhile, large-scale installations slowed down significantly from 13.1 MW in 2019 to just 3.8 MW in 2020.
The aggregate rooftop PV capacity of 1,605.8 MW has now overtaken the combined capacity of the coal-fired power stations in the SWIS (Bluewaters, Collie and Muja C/D have a combined maximum capacity of 1,566.9 MW).
The cumulative rooftop PV capacity in the SWIS since 2007 is shown in the graph below:
As can be seen from the graph above, rooftop PV capacity growth has accelerated in the past few years. It remains to be seen whether installations will continue to grow at the same pace with the reduction in the feed-in tariff in November 2020. The answers to the following questions will likely determine whether there will be continued growth or saturation in the demand for new rooftop PV systems:
How sensitive are prospective buyers to the implied investment returns / payback period on new PV systems?
Will owners of older existing systems with expired grandfathered feed-in tariffs begin to upgrade to larger systems?
Will there be an uptake in battery energy storage systems to take advantage of falling prices and higher feed-in tariffs during peak times?
Will new housing developments / construction include rooftop PV as a standard feature?
Will an uptake in electric vehicle purchases incentivise owners to install larger rooftop PV systems for charging (and possible vehicle-to-grid) purposes?
Data sources and notes
All of the data used in this analysis is sourced from the Clean Energy Regulator (CER) website:
The postcode data for small- and large-scale installations is filtered for postcodes in the SWIS. Large-scale ground-mounted installations are also removed from the data set (e.g. Merredin solar farm or Greenough River solar farm). The coarseness of the postcodes at the edges of the SWIS may introduce some error into the statistics, but the volume of installations in these areas are small relative to the total number of installations.
The small-scale postcode data was extracted on 25 January 2021, and it appears that the data set for December 2020 is not yet fully complete (showing only 2,295 installations for December 2020 compared with a monthly average of 3,759 installations over the period from January 2020 to November 2020). That said, the reduced volumes could potentially reflect softening demand after the REBS scheme was replaced by the DEBS scheme and the tariff reduced to $0.03/kWh during the day after 6 November 2020 (though such a sharp drop-off seems unlikely).
Note that the quoted dates are actually accreditation dates, which may not necessarily line up with the actual installation dates. This is more of an issue with the large-scale installations (>100 kW). For example, a rooftop PV system is installed and it may be months before the proponent applies for accreditation.
Load flow studies form the basis for the steady-state analysis of power systems and are performed for a wide range of applications from planning to operations. The studies are typically conducted using computer software packages such as PSSE, DIgSILENT PowerFactory, ETAP, PowerWorld and SKM Power*Tools.
The load flow (or power flow) calculation solves for four steady-state electrical quantities (at each bus and branch element):
Voltage V at each bus
Phase angle Φ at each bus
Active power flow P through each branch element
Reactive power flow Q through each branch element
We can use these four basic electrical quantities to calculate other quantities of interest, for example:
Apparent power S through each branch element
Current I flowing through each branch element
Loading of branch elements (e.g. transformers, cables, etc) based on continuous and short-term equipment ratings
Voltage drop between buses
Total system load
Total amount of generation dispatched
These calculated quantities are of interest to electrical network operators and design engineers in order to ensure that the power system operates efficiently and within limits. The rest of this article will talk about the generic structure of a commercial power flow solver / algorithm and how to debug convergence problems.
Power flow solution algorithms are typically comprised of inner and outer loops:
The inner loop calculates the main iterations of the solver (e.g. Newton-Raphson, Gauss-Seidel, etc) in order to minimise the power (or current) mismatch equations.
The outer loop is used to check and satisfy control conditions (e.g. auto tap changer / shunt control) and limiting behaviour (e.g. generator reactive power limits).
Note that automatic tap or shunt control is sometimes included in the inner loop, but only when continuous tap/shunt steps are used. When automatic tap and shunt control is incorporated in the outer loop, the power flow algorithm will attempt to switch the positions of the taps and shunts at the end of every inner loop iteration to satisfy the control conditions (e.g. bus voltage or branch power flow setpoints).
Diagnosing Convergence Issues
When a load flow fails to converge, the first step to diagnosing the problem is to identify whether the algorithm has failed in the inner or outer loop.
An inner loop problem signifies that the Newton-Raphson equations could not be solved, which could be due to ill-conditioning, voltage stability issues, etc. This usually results in a program error message stating that the algorithm has either diverged or stagnated.
On the other hand, when the problem is in the outer loop, the Newton-Raphson equations can be solved, but control conditions can’t be met (e.g. voltage setpoint could not be reached with transformer taps at maximum / minimum settings or are hunting between two tap steps). Typically, a maximum number of outer loop iterations must be reached before an error is reported by the program.
Tackling Inner Loop Problems
As mentioned earlier, inner loop problems suggest that the system is ill-conditioned and the Newton-Raphson algorithm either diverges or is stuck in a local minimum (i.e. stagnated). Mismatches in generation, poor distribution of generator dispatch and local voltage stability issues are among the common causes for an inner loop problem:
Mismatches in generation and load: when there is a large mismatch in generation and load (i.e. over- or under-generation), the discrepancy in active power has to be supplied or absorbed by the slack element. Depending on the location of the slack element, this can lead to large power flows from one part of the network to another, which can cause imbalances in the voltage profile and lead to convergence problems.
Distribution of generator dispatch: similar to mismatches in generation, poor distribution of dispatched generation can lead to large unwanted power flows from one part of the network to another.
Local voltage stability: relates to busbars reaching the nose point of the PV curve (voltage collapse). This could be very localised due to large loads placed at weak radial parts of the network without sufficient reactive compensation, or could affect larger areas where multiple buses have depressed voltages.
Some approaches to resolving inner loop problems are as follows:
Turn off enforce reactive power limits option: most programs have an option to limit the reactive power output of PV buses (generators, SVCs, etc) so that they are within their reactive power capability curves. When a PV bus reaches limits, it is converted to a PQ bus in the outer loop with the reactive power set to its limits and the inner loop is run again, which can then fail due to voltage stability reasons (i.e. the PV bus is supporting the local voltage, but with limits enforced, voltages can go wild). Turning this option off can help diagnose which generators, SVCs, etc are supplying / absorbing large amounts of reactive power and their associated problem buses.
Change the location of the slack element and distribute generator dispatch: so that there are more balanced flows through the network. Some programs have a distributed slack bus option that automatically caters for mismatches in generation by distributing the slack across all in-service generators according to their nominal ratings.
Switch in or add shunt compensation: in order to support weak buses that may have voltage stability problems.
Tackling Outer Loop Problems
Outer loop problems are normally associated with voltage control issues from the following elements:
Generators – reaching their reactive power capability limits
SVCs – reaching their reactive power capability limits
Transformer automatic tap changers – reaching maximum tap settings without satisfying voltage setpoint tolerances
Shunt controllers – reaching maximum shunt steps without satisfying voltage setpoint tolerances
Some approaches to resolving outer loop problems are as follows:
Increase the maximum number of outer loop iterations: particularly for larger systems, the outer loop solution may take many iterations.
Change voltage control setpoints and tolerances: the bus voltage setpoints and tolerances can be adjusted to help the solution along.
Lock taps and/or shunts: automatic transformer tap and shunt adjustment options can be de-selected to lock the taps and shunts at their initial positions.
Reset taps and/or shunts: the initial tap and shunt positions may not be well configured for the particular scenario. Reset tap positions by setting all transformers to neutral tap and re-configure initial shunt positions to better control system voltages (e.g. if you know that a bus has a low voltage, put some capacitors in service).
The accurate modelling of inertia in an islanded power system is crucial when attempting to integrate low or inertia-less generation into the network, particularly those from intermittent sources, e.g. solar PV, converter-fed wind turbines, etc.
In the context of synchronous generators, the term “inertia” generally refers to the kinetic energy in the rotating mass of a generator shaft. The inertia depends on the speed of rotation and mass of the shaft, i.e. the heavier the shaft and faster the rotational speed, the higher the inertia. Under normal operation, the instant a new load is applied to the system, rotational energy (inertia) is converted into electrical energy to supply the load. As energy is removed from the system to supply the load, the speed of rotation (or frequency) decreases. This is commonly referred to as the inertial response and occurs before primary frequency control actions (e.g. from governors) take place. As a result, inertia has a direct impact on transient frequency deviations resulting from sudden changes in generation and load.
Effects of varying inertia on system frequency
In large interconnected systems, these frequency deviations are minor since the instantaneous mismatches in generation and load are very small relative to the amount of synchronous generation dispatched. However, in small island systems, these frequency deviations are much more significant and can even lead to frequency collapse.
By virtue of their remoteness and lack of other resources, the small island power systems that are normally found in island nations and archipelagos typically have high penetration of diesel engine generation. As diesel engines generally have low inertia, the inertia constants selected during system modelling can have a large effect on frequency swings. For example, consider the load acceptance (25% load step event) of a 1.5MVA diesel generator with varying inertia constants:
The simulations show that maximum frequency swings can vary from 1.235Hz up to 3.548Hz depending on the inertia constant selected.
Small power systems are usually equipped with under-frequency load shedding (UFLS) systems to prevent network collapse during frequency swings and active power deficit events (e.g. trip of generator). Integrating inertia-less sources such as solar PV plants would displace synchronous generation and thus reduce the total inertia in the system. Grid simulation studies are performed to predict whether or not the UFLS system is at risk of operating during normal day-to-day fluctuations of the solar PV system, and this requires fairly accurate modelling of the system inertia.
Selecting appropriate inertia values
Generator inertia values are usually found in vendor / manufacturer data sheets and are often expressed as a moment of inertia quantity (e.g. in kg.m2 or slug.ft2). While some software packages can accept these quantities directly as inputs, other programs require that they are converted into per-unit inertia constants (H).
In some cases, particularly in older systems or networks with temporary generators, the inertia data is not available and must be estimated. Based on our database of actual equipment data, the following inertia values can be used as guidance:
Transmission and distribution networks are natural monopolies and in Australia (as in most other countries), they were once part of vertically integrated corporations comprising either the complete or significant portions of the electricity supply chain (e.g. the state electricity commissions). During the industry restructure of the mid-1990s, these corporations were broken up so that in each state, there was a single monopoly transmission company (e.g. TransGrid, Western Power, PowerLink, Electranet, etc) and one or more geographically-focused distribution companies (e.g. Ausgrid, Ergon, Citipower, etc).
Thus a natural question to ask is – can present-day distribution network service providers (DNSPs) be further unbundled?
Local distribution cooperatives
One possibility is for cooperatives in a localised geographical area (e.g. a suburb or housing estate) to acquire ownership of the low voltage distribution assets in the area, connect to the distribution network at one or two points as a major customer (e.g. at kiosk substations) and participate directly in the wholesale electricity market as a single coordinated entity.
Such a local distribution cooperative could be a microgrid with a mix of its own distributed generation and storage, or it could simply be a consumer aggregating the households in the area (e.g. Virtual Power Plant). However, to get the most out of participating in the electricity market, the cooperative needs to have some level of controllability. What this could mean in practice is an energy management system that can control discretionary household loads (such as hot water heaters, air conditioner thermostat settings, swimming pool pumps, washing machines, etc), as well as generation and storage resources, in order to optimise the kWh price on the spot market for the whole cooperative.
For example, when the spot price is high due to peak loading or a binding network constraint, the energy management system can control discretionary loads, generation and storage to either limit energy purchases or sell energy back to the market (and vice versa when spot prices are low).
The cooperative itself could be run like a body corporate or owners corporation, with a professional management company appointed to maintain and manage the local distribution assets (e.g. poles, lines, energy management system, etc) on behalf of the owners. The cooperative acts like a micro-retailer and the owners simply pay an equivalent electricity tariff to maintain the system. In some circumstances, the cooperative could even pay out a dividend from profits made trading on the wholesale electricity market.
A looming obstacle for local distribution cooperatives will be convincing (or coercing) DNSPs to sell / transfer their assets to a cooperative. Under the current regulatory scheme, networks are paid a guaranteed return on their regulated asset base (RAB), and this of course gives DNSPs very little incentive to reduce the size of their RAB.
In the United States, the recent track record for municipalising electric utilities has been mixed. In some cases, local municipalities were able to negotiate the purchase of utility assets without much pushback from the incumbent utility business, typically when the incumbent saw strategic gains from pulling out of the area. In other cases, there was significant resistance from the incumbent to municipalisation proposals and negotiations were difficult and acrimonious.
However, there is a case to be made that micro-scale municipalisation proposals like local distribution cooperatives could face less resistance since the value of the assets are relatively low.
A new role for DNSPs
If local distribution cooperatives were to emerge as viable businesses, this could signal a trend towards a widely distributed ownership of electricity network assets. Micro-scale networks owned and operated by the locals who live there would change the landscape of the electricity supply industry. Where present day DNSPs have to plan and optimise a network across large geographic areas, the local distribution cooperatives only need to worry about their own small sections of the grid. The role of DNSPs would be more akin to what transmission networks are now – a backbone for the bulk supply of power to the local networks. Two or more cooperatives could potentially build private interconnections between each other’s networks, bypassing the DNSP altogether.
Moreover, there is the possibility that local distribution cooperatives can negotiate their own terms of connection to the DNSP, for example, customising the trade-off between network reliability and use of service charges. Of course, such provisions would necessitate a fairly significant change in the rules by the AEMC, but it has the potential to give more choice and responsibility to end-users regarding their own electricity supply and how they choose to manage risks and resources.