In conjunction with my initial analysis of AEMO’s Electricity Statement of Opportunities 2017 report today, I have prepared this brief explanation of the meaning of “USE”.
What is “USE”?
AEMO’s assessment of NEM reliability is bound by the National Electricity Rules to quantify something called “Expected Unserved Energy”. This typically gets abbreviated as USE which strictly means just “Unserved Energy”, but the “Expected” qualifier is normally read in, and is very important.
Unserved energy means the amount of end-customer demand (measured in megawatt hours) that cannot be supplied within a NEM region due to a deficiency of generation or interconnector capacity. In other words forced load shedding.
1) For examples we only have to look back to last summer when AEMO initiated load shedding in Adelaide for about an hour on Feb 8, and in NSW on Feb 10 when a potline at the Tomago smelter was shed for just over an hour. Both of these actions were taken to maintain overall supply-demand balance and system security.
2) In the NEM reliability context, localised blackouts which occur due to failures of distribution or local transmission equipment do not count towards “unserved energy”.
The “Expected” qualifier on USE can be confusing because here it means statistical expectation – an average across a range of future scenarios, weighted for probability – not what AEMO thinks will definitely, or is even most likely to happen. For an analogy, the statistically expected outcome of betting $100 on red at the roulette table is a return of just under $100 – but the returns from an actual bet will be either zero (occurring just over half the time) or $200 (just under half the time). So when AEMO assesses Expected Unserved Energy for a future period to be, say, 700 MWh, this really means that in some scenarios – hopefully most – they project no load shedding, but in others, a significantly larger amount than 700 MWh would be shed.
Just to make the measure even more opaque, the NEM Rules target and most of AEMO’s data express USE as a percentage of annual regional consumption. So if that 700 MWh of USE were being projected for Victoria, which has an annual energy usage of around 45,000,000 MWh, it would be expressed as USE of 0.0016% – which sounds tiny, but if you were one of the customers making up 350 MW of load being shed for two hours, might not seem quite so small.
The Reliability Standard and Measuring USE
The NEM Reliability Standard requires that projected USE should not exceed 0.002% in any region. That’s easy to say and very difficult to assess. Load shedding becomes necessary when there is not enough supply to meet demand, evaluated down at half hourly granularity. Both demand and supply are highly variable with time of day, season, weather, generation outages, and numerous other factors. As implied above, measuring “Expected” unserved energy requires evaluating a wide range of forward supply-demand cases involving different levels and combinations of these variables for each NEM region.
There’s documentation on AEMO’s website about how this is done using detailed power system models, but the key variables that affect results are the amount of generation and interconnector capacity available (determined by how much is built and available in each region, but also subject to unexpected breakdowns / outages), the maximum level of demand projected, and now with significant amounts of renewable generation in the system, levels of wind and sunshine and how these align with demand.
To keep all the modelling manageable, AEMO run projections of demand calibrated to just two different maximum demand levels in each region, denoted as 50% POE and 10% POE. These refer to maximum demand levels forecast to be reached one year in two, and one year in ten respectively. Obviously the 10% POE value is higher – often substantially, reflecting the wide variation in maximum summer temperatures that is possible in south eastern Australia. Given these two demand benchmarks other uncertain factors like weather / wind / sunshine patterns and possible generator outages are statistically combined using historical patterns or failure rates to build up well over a hundred distinct supply-demand simulations for any given region and year and assessing which of these might lead to load shedding, and how much.
The results from these individual simulations are then statistically combined back into a single weighted average (ie Expected) Unserved Energy value for each NEM region – the USE measure.