Digitalisation and the energy sector

Digitalisation and the energy sector

Andy Compton, director at Compton Energy Associates and Matthew Brown, energy lawyer at UUÂãÁÄÖ±²¥, examine the opportunities and challenges presented to the energy sector by digitalisation.

The advance of digitalisation is evidenced by our increased use of buzzwords, such as data analytics, machine learning, Industry 4.0 and the internet of things (IoT). This has been accompanied by an explosion in the quantity of commercially useful data, and the introduction of artificial intelligence (AI) derived tools to analyse it in great depth. This is presenting ground-breaking business opportunities across the economy. Few doubt the transformative powers of digitalisation, not least in the energy sector. However, in common with all profound technological changes, there are risks and concerns.

An indicator of the growing importance of digitalisation in the energy arena is demonstrated by the International Energy Association’s (IEA’s) assessment that in 2016, energy companies invested almost 40% more in digital electricity infrastructure ($47bn) than in gas-fired generation ($34bn) worldwide [1]. Whilst the energy sector has utilised digital technology for decades, the sector has historically been risk-averse in its approach to adopting truly cutting-edge technology, governed more by its responsibilities to maintain energy security and ensure safety. Yet, as the energy market becomes increasingly fragmented, it is clear that digitalisation is presenting opportunities and threats at a number of points in the value chain, from smart meters to tools for managing the entire system.

Digitalisation arrives at a critical time. The energy system, designed and built around different technologies, and for different purposes, requires transformative change to respond to a highly distributed network of largely renewable energy projects, with an increasing number of smaller generators and/or storage providers, including ‘prosumers’ (consumers who also produce power). This change means increased complexity, with more interactions between energy vectors, increasingly variable supply and new sources of demand.

At present, perhaps the greatest changes are occurring in utility-scale power generation, where combined cycle gas turbine (CCGT) power plants are progressively becoming unable to operate as mid-merit plant; instead being used to fill in for renewables.  Open cycle gas turbines (OCGTs) are correspondingly attractive alternatives to investors, as plants are forced to generate their revenue over decreasing numbers of running hours.  This increasing variability in the market means short periods of highly inflated prices.  OCGT and CCGT operators can capitalise on these short-lived opportunities - if they can correctly predict them.

However, digitalisation should mean increased competition for this business from energy storage and demand side response (DSR) from aggregated groups of prosumers.  Machine learning based predictive algorithms can offer significant advantages to traders, given their sophisticated power to draw out relevant patterns in market drivers and behaviour, and their ability to handle larger datasets, enabling a broader range of parameters to be considered.  It is also clear that government, Ofgem and National Grid are keen not to stymie this opportunity. The government and Ofgem’s 2017 Smart Systems Plan  [2] was accompanied by a series of balancing services reforms from National Grid, most recently in the form of confirmed plans to further open up access to the Balancing Mechanism to ‘virtual power plants’, made-up of distributed generation and demand assets [3].

Digitalisation is also meeting challenges in energy infrastructure operation and maintenance.  The ability to deploy increasingly cheap sensors, tied to the ability to collect and analyse the resulting datasets, allows the construction of ‘digital twins’ for major assets.  Digital twins are a highly accurate digital representation of a complex physical asset that uses continual machine learning to model the performance of the asset throughout its lifetime.  Groups of digital twins can be modelled collectively to predict and optimise complete sections of energy systems.  Hence, machine learning can facilitate a transition from planned or condition-based maintenance practices to predictive maintenance regimes.  The prediction of failures allows the avoidance of unscheduled maintenance. When integrated with market predicting algorithms previously described, corrective maintenance can be scheduled to avoid high-revenue opportunity periods. Further, it can optimise operating regimes to ensure the asset reaches the intended maintenance period without failure.  Digital twins allow the opportunity to simulate performance under extreme scenarios without risk to the physical asset.  Better knowledge of the asset’s performance could allow operators to release additional performance, previously considered risky under necessarily conservative standard operating parameters.

Systems incorporating very high volumes of data, that need to be analysed and acted upon in near real-time, are ideal for machine learning based decision making. Deep analysis, using probabilistic decision algorithms, allows optimised decisions to be made at speeds not humanly possible. Digitalisation therefore allows operators to make better use of existing infrastructure, avoiding billions in capital upgrades.  Despite these advantages, handing control of the nation’s power to AI would undoubtedly make some feel uneasy. In the event of a failure, those machine-based decisions may be hard to reconstruct or explain to those responsible for oversight.

At the consumer level, the (at times controversial) implementation of smart metering represents a tangible example of the reach of digitalisation, with the potential for a truly two-way real-time relationship between supplier and customer. The benefits for the consumer have been heavily promoted but inevitably there is a certain amount of public disquiet about what happens to the data they produce, and the fact that suppliers have been tasked with leading on roll-out.

The reality is that the mind-boggling amount of data (35k data points per year for each household) is simply too huge to be analysed effectively – by humans. But with AI, deep-seated patterns of consumption behaviour can be teased out of huge datasets, almost in real time, through the use of analytics and this is where the valuable information starts to appear. Amazon and Netflix have been successfully harnessing machine learning for some time to recommend product and film choices and most consumers seem relatively relaxed about this. However, that insouciance may not equally apply to our energy use. Legitimate concerns around cybersecurity and data privacy will require robust safeguarding measures. Similarly, the question of what we allow AI to control will need to be addressed in a fashion that ensures continued public confidence in the system.

As regulation looks to keep pace with digitalisation in the energy sector (and more generally), most notably via the Network and Information Systems Regulations 2018 (NIS) and the implementation of the EU General Data Protection Regulation (GDPR), and energy suppliers face maintaining public confidence in the security of their data, the skills required within supply side companies are changing. Alongside the already complex energy specific regulation faced by the sector, there is a growing body of data and cybersecurity regulation, which energy companies and their advisors must be resourced to navigate and to apply. On smart metering specifically, the provisions of GDPR, coupled with the security requirements imposed by NIS and the Smart Energy Code (SEC), represent a subset of regulation quite distinct in nature from the transactional and regulatory understanding required to ensure the physical infrastructure is installed and operated appropriately.

There is no denying that cybersecurity is challenging, but the banking industry has demonstrated that with the right regulation and industry approach, consumer confidence can be gained.  A reminder of the benefits of digitalisation and the profound changes underway in our energy system means that hitting the stop button simply isn’t an option. If the energy industry can harness the power of digitalisation, it will bring enormous efficiency savings for the nation and for individuals, not to mention greatly facilitating the journey to decarbonisation.

Footnotes:

[1]

[2]

[3]


Related Articles:
Latest Articles:
About the author: