By Simon Vardy, Utilities Strategy Lead, Accenture Australia and New Zealand and Frank Ochel, Senior Principal, Utilities Strategy.
In an era of negative daytime prices, sophisticated data analytics is required to optimise the commercial position of renewables and improve the links between utilities trading floors and control rooms, says Simon Vardy and Frank Ochel from Accenture
Recent reports by the Clean Energy Regulator and others have once again highlighted the phenomenal growth in renewable generation in Australia. When we consider that the first grid scale solar installation in Australia was only built in 2011 – with a size of 1MW, quickly growing to 10MW one year later in 2012 – it’s obvious we have come a long way.
Today’s project sizes are something different altogether; 100MW is becoming quite common, and sizes of 400MW as well as a few Gigawatt projects, are also in the pipeline. Indeed, Australia installed three times as much renewable capacity in 2018 (3,455 MW) as we did in 2017 and the forecast is for the output to double again by 2020, to 40,000 GWh per annum. There is even a proposal for a 10GW solar farm in the Northern Territory designed to supply Singapore by underground cable!
This trajectory of grid scale solar has now made Australia the number one in per-capita renewable deployment, globally. Australia is definitely not short of a very large wind and solar development pipeline. But is this as good as it gets? What could possibly go wrong?
Dealing with negative wholesale prices
The existing levels of renewable generation already have a profound impact on the wholesale market. In the past, negative wholesale prices were only seen during the night i.e. they did not affect the wholesale revenue of early solar projects and were quite predictable.
With increased levels of both grid scale, and rooftop solar generation, wholesale prices can now be negative during the day. That, however, is the only time solar projects can generate. To illustrate, for a brief period on a weekend earlier this year, renewable generation from wind and solar was high (44 percent) and demand dropped strongly, resulting in the wholesale spot price falling to $0 in all National Electricity Market (NEM) states at the same time.
Up until recently wind and solar farms, having a near zero marginal cost of production, just generated when they could and that was it. But now, with negative prices a reality, there are many options such as redirecting output to battery storage (if available onsite), utilising maintenance windows, or even switching off altogether as happened at two solar plants in South Australia last May. From a trading perspective, what was once just a set of constraints now becomes a set of ‘physical options’ from which to inform and direct operations.
It has become increasingly more important to forecast, bid and operate with more foresight and accuracy – and do so constantly to avoid negative price periods and maximise the value from ‘physical options’ in each five-minute window. If traders get this wrong, bid and the load is dispatched into the market the generator is obligated to pay, instead of receiving payments from the market. A bad outcome!
Add to this that Corporate Power Purchase Agreements (PPAs) in Australia now tend to address zero-price conditions (these clauses have the effect that the off-taker will not be required to pay for the output if the market price is negative) the generator is often left out of pocket in these situations.
Forecasting is also getting increasingly more complex. Previously, only pricing was variable and volatile. Now volumes are too. From a trading perspective the forecasting of increasing volumes of Variable Renewable Energy (VRE) in one’s own portfolio and across the market is a step change from traditional generation fleets.
That’s on top of Australia’s energy only wholesale market already being one of the most volatile in the world. Add to that the introduction of 5-minute settlement and it becomes clear that bidding and risk management capabilities need to become a lot more sophisticated as there is no more opportunity to ‘course correct’ using subsequent bidding intervals in the same trading period.
Utilising applied intelligence for ongoing viability
Moving forward, grid scale renewable projects will need to significantly increase the use of data and applied intelligence to optimise their commercial positions and remain viable. The close interaction between operations, to confirm availability, and traders, to commercially optimise it, is crucial.
Utilising third party marketing agreements (e.g. offtake by a Gentailer) in these projects adds an extra layer of complexity as the controllers and traders are not even in the same company and can’t simply walk across to the other side of the building. They need to have immediate access to the same information and be able to exchange instructions constantly.
Utilities in Europe have achieved a 2-5% trading margin uplift from better forecasting the supply-demand balance and market prices. To do this they utilised operational data from assets and combined it with external data sets. Self-learning algorithms then increase the accuracy of forecasts.
Getting granular with data
The first capability to address these challenges is visibility and granularity in real time. Weekly or daily production meetings to discuss daily maintenance windows are no longer sufficient. Advanced utilities already measure multiple data points such as blade or yoke alignment constantly and do so for each individual asset, not simply for the overall plant or project.
One of the most powerful ways to illustrate the priority that needs to be given to specific maintenance activities is to link it directly to financial outcomes, e.g. a potential loss in the wholesale market in a five-minute period or the rest of the day. Translating an outage into the language of “these turbines not working is currently costing us $10,000 an hour” is easy to understand and spurs on operations teams while allowing planners to optimise using one common metric: revenue.
The ability to manage and combine large internal and external data sets is essential as there will be more granular data from more assets. The number of renewable assets, including consumer-owned Distributed Energy Resources (DER) is increasing, meaning that instead of one data point for a baseload generator turbine, hundreds, if not thousands, of assets per project need to be measured, monitored and controlled.
The granularity of these data points will also increase. For DER by the minute data is already in use. The data intervals required for true commercial optimisation will shrink further as value shifts from energy markets to flexibility markets such as Frequency Control and Ancillary Services (FCAS). Ultimately, we can envisage a combination of sub-second SCADA or Historian data being directly used in commercial modelling.
Weather data, too, is becoming increasingly granular. In an ARENA funded project, the Australian National University together with commercial partners has developed very near-term weather forecasts, sometimes called “nowcast” to predict wind or cloud movements.
This enormous volume of data can no longer be managed with many existing forecasting and scenario modelling tools. Commercial and operational decisions will have to be made with much more speed. Sophisticated models that can provide recommendations, not simply analysis, are required.
Utilities in Europe have already built Applied Intelligence (AI) tools that combine market and fleet data with predicted competitor bidding behaviours and provide recommended courses of action. This has seen them lift margins by over three percent. Bidding and re-bidding have already attracted strong criticism, and subsequent scrutiny, in Australia for allegations of market manipulation. Getting bids right the first time has never been more important to maximise the value in a window of only 5 minutes under the watchful eyes of market regulators and operators.
Volatility and variability have always provided opportunity for those who had the capability to manage and benefit from them. Australia’s massive take-up of grid scale solar and wind will now need a similar uplift in data analytics.
For Australia to continue the development of large-scale renewables and maintain a leading global position the use of sophisticated data analytics, innovative new data sources and operating models is a must.
Australia now has the opportunity to show the world how we can not only build wind and solar but how it can be managed, integrated and optimised to create new jobs, new skillsets and ultimately a better functioning electricity market.
