The energy sector is already using data in sophisticated ways to meet a wide range of challenges, from fault prediction in grid networks to the delivery of personalised energy to households. As more data becomes “big data” and smart devices proliferate along the value chain, Matt Brown and Ravi Mahendra of Pöyry Management Consulting make their predictions for more digitalisation in the coming year.
Over our working lives, change from digitalisation has always been there. Laptops, email, the internet, vast improvements in computing power; plus and an increasing reliance on algorithms and access to more data than we can manage. Hence the extremely high interest – and some would say hype – in energy digitalisation.
The IEA’s Digitalisation and Energy 2017 report says that digitalisation could save 5% of annual power generation costs. Given the fall in renewable and battery technology costs we’ve witnessed, this doesn’t sound very exciting; nor in the light of increasing volatility of commodity prices. Perhaps those technology cost falls have themselves resulted from greater digitalisation, but does that make the energy sector largely a recipient rather than a participant in the digitalisation journey?
The digitalisation of energy
The step-change that we see is two-fold: on the one-hand, computing power increasing and enabling greater levels of artificial intelligence (AI) and machine learning; and secondly, the amount of data available for those computers to analyse. These two combine to mean that better decisions are made and are also automated.
These are the key areas we expect to see further progress in 2019 in digitalisation in energy:
Fault prediction and dynamic maintenance: This is one of the clearest uses of AI which enables operators to predict equipment failures by using sensor data from various units to significantly reduce their costs of downtime and maintenance. Our consultancy Pöyry has an offering for this called KRTI4.0. On the retail side, a startup Verv is offering a meter device which identifies individual home appliances and tries to predict faults or a device being accidentally left on by building up individual profiles from the meter data.
Investment optimisation: BP’s venture arm invested in an AI startup called Beyond Limits to dig through seismic images and geological models to increase the chances of success when drilling wells. Another example of longer-term investment decisions is the US Department of Energy project where machine learning is being used on satellite imagery and operation data to prioritise reinforcement at vulnerable points of the grid to improve resiliency.
Energy efficiency: Deepmind, which is a part of Google, has championed the use of Reinforcement Learning to reduce energy use in its data centres by a claimed 15%. The model learnt by looking at years of operational data and then issued changes to individual units within the operating constraints of the plant.
Better prediction: Deepmind is also currently in talks with National Grid of the UK to better forecast demand of the system with the stated goal of reducing the entire country’s energy usage by 10%. Another example is improved prediction of wind power production to reduce imbalance costs by 50% which was achieved by a company called Swhere.
Retail: retailers are using machine learning to understand patterns of customer behaviour, to attract and retain customers and even to predict bill (non)-payment. Customer call centres are being fronted by algorithms which chat to customers (verbally or online) and deal with queries.
Customers: For customers, AI solutions are also gaining traction, and many retailers are offering these systems as part of an integrated package. Devices such as Amazon’s Alexa enable the customer to seamlessly interact with their thermostat (such as Centrica’s Hive). This increasing customer interaction with the device leads to the development of a more personalised usage profile, which reduces bills for the consumer and also helps the energy provider to accurately forecast demand.
Trading: According to the FT, systematic and algorithmic trading now account for nearly 60% of the traded volume on just the CME energy product group – the highest level of any commodity group. Anecdotal evidence from mid-2018 is that over 50% of trades on the EPEX Spot intraday market are algo-trades (although the total volumes are still smaller than trades executed by human). Sophisticated machine learning models are also being deployed by speculators which are relying on large streams of diverse data to respond to the market changes quickly.
A more commercial example is Origami Energy using machine learning to predict asset availability and balancing mechanism market prices in near real time to successfully bid into the Frequency Response markets. Pöyry is exploring a deep learning algorithm to support trading and dispatch decisions for generation assets in the prompt trading markets, focusing on the issue ‘when should I commit a trade’ (to maximise the option value of flexible capacity).
It would seem then that the digitalisation opportunities in energy are large. It will be a vital enabler of decarbonisation is some areas in the future such as flexible demand shifting to meet supply. The opportunities available rely heavily though on sufficient volumes of good quality data being available. So, expect more sensors and more data acquisition throughout the energy sector in 2019. And in time with growing autonomy expect the focus to switch to the appropriate monitoring, alerts and controls.
Matt Brown is Head of Western Europe, Middle East and Americas, Pöyry Management Consulting
Ravi Mahendra is an Analyst, Pöyry Management Consulting
This article is published with their permission.