As everybody working in electricity knows, generation and consumption must be permanently balanced. In this respect, it’s not like any other sector. If a brewery makes too much beer, you can keep it for a while, you can drink it, or even pour it away.
Electricity doesn’t work like that – and even if storage is just beginning to feature in our thinking, for now, the old rules apply: you need to master the balancing act in a world of continuous production.
Utilities have learned to do this in a way that matches their established operating models and, until now, this has invariably been centralized. Even with centralized models, balancing production and consumption is a complex, if well-established activity.
Historically, when utilities were more monolithic, they had teams dedicated to the task of balancing inputs and outputs with real precision over widely differing periods: ten-year, one-year, six-month, daily and even ten second time windows.
The rules just changed
The skills and tools needed to balance production and consumption, however, are no longer fit for purpose in a world increasingly influenced by renewables and local production.
Now that wind turbines, large-scale solar farms, and even locally-managed hydro have become a permanent and growing part of the mix, the challenge of producing accurate and actionable models becomes massively more complicated.
There’s nothing academic about the need to cover this extended power landscape in predictive modelling. There are two principal challenges, and they both have a direct impact on operations, planning and profitability.
- Firstly, it is in everybody’s interests to consume locally energy which is produced locally. In doing so, you minimize losses caused by power-line attenuation.
- Secondly, unless you can predict how much electricity is going to be produced and consumed locally, you cannot determine how much needs to be re-absorbed back into the grid.
This second point is particularly important. Unless the utility, and specifically the Distribution Service Operator, can accurately forecast the local surplus, the only way to manage it is to radically over-engineer the local grid or risk damage to transformers and sub-stations by periodic overload.
The good news is that the challenge of modelling local power consumption and generation coincides with the availability of new tools for practical data analytics.
According to a commissioned study conducted by Forrester Consulting on behalf of Atos, 31% of utilities are already using data analytics to forecast demand and production, with a further 47% planning to do so over the next twelve months.
The success of these new analytics will depend on speed, agility and ease-of-adoption and application. The industry needs reliable intelligence on how best to integrate local renewables into production and consumption forecast now – you cannot wait five years for analytics practices to mature.
At Atos, with our strategic investments in the Atos Codex analytics framework, we have been actively building use-cases for our utility clients, and are already achieving forecast accuracy exceeding 95%. This is not perfect, but it is possible now – and 95% now is certainly better than 99% at some point in the future.
It’s not just the availability that’s impressive. We are using techniques that analyze both utility and third-party data using real-time data flows. This brings important benefits.
Not only does it mean we can take, for example, local weather conditions into account for greater accuracy and granularity. It also means that the active models exist as real-time assets for immediate business and operational advantage: we are not carrying out the analysis in remote and academic isolation.
Collaborate to innovate
For the Atos utility and data analytics specialists, these are early days. We are deliberately focusing on use-cases – on ways to deliver practical and immediate benefit to genuine and pressing industry challenges.
Examples include fraud detection and new customer smart home services. For the DSO, the immediate business challenge of forecasting and balancing in a new hybrid landscape is a clear priority – and if it’s a priority for clients, it’s a priority for us.
If you’d like to explore the challenge of real-time balancing across local and centralized networks, taking a closer look at the impact of analytics, contact us to book an innovation workshop.