AI for Energy Provisioning and Management

Prices of natural gas are now almost treble their level at the start of year and up 70% since early August alone. This is also leading to record electricity prices, as gas is currently vital for power generation in the UK. This has raised fears of a severe economic hit to industry, similar to the petroleum crisis in 1970, which led to a decade of limited economic growth. The causes of this post pandemic crunch are manifold.

The majority of our energy mix in the UK is made up of natural gas, with 80% of which respectively imported from Russia. Bloomberg calculations show that Russia’s state-owned company Gazprom will need to store nearly as much natural gas at home to keep their citizens warm this winter as it currently ships to western Europe every day.

According to Tomas Marzec-Manser, lead European gas analyst at ICIS, Gazprom has to prioritise discretionary supply away from Europe in order to reach that goal. It has just two months to build its depleted inventories to the record levels. This will require pumping into their underground storage sites, supplies equal to about 80% of daily exports.

The recent phasing out of coal plants in the UK has limited the opportunity to switch fuels when prices rise. The contribution of wind turbines to the grid has also been low due to very still summer weather. The problem for the UK is that we are arguably more exposed than the rest of Europe, due to the extensive reliance on gas importation and “just-in-time” approach to its supply, which makes us dependent on EU pipelines. “We’re effectively at the end of the pipe – both physically and now politically”, as Niall Trimble of the Energy Contract Company puts it.

Energy provisioning is of course a constant pan-national economic challenge, with various levels of market autonomy and regulation involved. Energy UK’s chief executive Emma Pinchbeck points to part of the current problem in the UK being a gap in government policy to ease the transition from fossil fuels to renewables.

Focussing on just hitting renewable generation targets and not long-term investments in necessary storage technologies, means the country is highly susceptible to supply fluctuations. Even in gas, long seen as the bridge between fossil fuels and renewables, the UK is no longer able to store large volumes of energy to deploy when prices spike and supply dries up. The UK’s largest offshore gas storage facility, Rough, closed in 2017, which hasn’t been replaced. Furthermore, market economics aren’t responding sufficiently to net-zero targets and modern energy requirements, where profit incentives remain fastened to fossil fuel monopolies and legacy infrastructure.

Already, the size of a median power plant in has fallen from 800 megawatts in 2012 to 562 megawatts in 2020, and Bloomberg NEF projects a decentralised market will lead to average storage capacities of just 32 megawatts by 2050.

Investment in centralised grids, with their system of long wires and transformers, is the wrong way forward. Instead, governments need to plan for a decentralised grid, where various generation points can transition more quickly, supported by communities and buildings generating their own electricity, all managed in real-time.

In the USA, where deadly winter storms left millions of Texans without power, energy experts, politicians and pundits have called for massive spending on grid-related infrastructure upgrades. This landmark vision has been drawn up as the Green New Deal, which combines Roosevelt's historic economic approach, with modern innovations of renewable energy investment and resource efficiency.

The global transition to renewable energy can be supported with artificial intelligence (AI) technology to manage decentralised grids. AI can balance electricity supply and demand needs in real-time, optimize energy use and storage to reduce rates. Public and technology governance will be needed to democratize access, encourage innovation and ensure resilient electricity sources. This can ensure interoperability, transparency and equal access across the energy landscape.

Renewable energy increases complexity

As we move toward an increasingly electric world, more energy will be produced by decentralised, renewable sources. Welcome as they are from a sustainability point of view, they typically add complexity to energy grids across the globe.

Over the next 10-15 years, the growing adoption of electric vehicles, the electrification of heating systems, and the proliferation of distributed energy resources (DERs) like wind turbines and solar panels will require a balancing act to match supply with demand without collapsing respective power grids.

A similar scenario is playing out in many other parts of the world as businesses, government and residential consumers increasingly produce their own energy through solar panels, storing that energy in batteries and electric vehicles, or feeding it back to the grid. According to forecasts, approximately 36 million assets such as solar panels, electric vehicles and energy storage will be added to the grid in Europe in 2025, and 89 million by 2030.

In other words, reliance on a central utility to produce and transmit electricity is fading, and utility companies will need to shift their business models.

The power grid

Electricity is not like other commodities and it is extremely expensive to store in large quantities. There has always therefore been an interest in balancing its use and generation. Finding an equilibrium between the two remains a challenge, where there are several actors who participate in managing the generation, flow and storage of electricity.

The mixed-market system operates in which several participants sell and buy electricity at a competitive price. Ideally, the market provides reliable, consistent electricity to consumers at a minimal cost.

AI-powered forecasting

By looking at these conditions mentioned above, we can see that smooth management of supply and demand from all market participants is vital in forecasting energy demand and generation.

Artificial intelligence is a tool that can fill this information gap with different methods. Time series is an example method of how AI can predict energy consumption and generation. Time series is a sequence of data in consecutive order and is used for forecasting future instances based on that observational data. In this case it can be used to predict power generation potential based on meteorological data.

By analysing historical meteorological data related to humidity, wind speed, cloud cover, tidal patterns, it can be possible to forecast potential electricity production of renewable energy devices. Time series are also used to predict what the electricity demand will be based on the specific time period such as weekends, holidays, weekdays, etc. as consumption tends to vary in these periods.

Artificial neural networks (ANN’s) are machine learning algorithms with interconnected nodes, recognising patterns and correlations of raw data and producing the output data. The neural network is a widely used method of forecasting time series. Depending on the weather or period data, the neural network will recognise patterns and understand during which days the demand for electricity is highest or how much energy can be produced in a particular season.

With a wealth of accumulated data (and unstructured data), and modern artificial intelligence methods, participants can predict electricity generation potential or demand, which will in turn help avoid disrupted and overpriced supply.

AI will balance millions of assets on the grid

With the help of AI software, decentralized energy sources can send any excess electricity they produce to the grid, while utilities direct that power to where it’s needed. Similarly, energy storage in industrial facilities, office buildings, homes, and cars can hold excess energy when demand is low, while AI deploys that power when generation is inadequate or impossible.

This makes an AI-centric system a game-changer. Shifting from an infrastructure heavy system to one cantered on AI enables forecasting and control in seconds, not days, resulting in a grid that is more resilient and flexible when unforeseen events occur.

This therefore means that utilities, policy makers and regulatory bodies need to plan what role they will play when in a decentralised energy network. The patchwork of distributed energy producers will depend on coordination and management. Utilities going bankrupt, could be supported to take this lead as they face a shrinking pool of customers purchasing direct grid electricity.

Utility companies will have to decide if they are to work with software companies, or if they want to become software companies in their own right. These advancements require a shift in thinking from legacy models of capital investment in a few large energy generation assets to demand management of an exponentially growing number of privately owned assets – all while protecting customer data and privacy and ensuring cybersecurity of grid management.

Governments also will have to both accelerate and evaluate their approaches to infrastructure projects, energy generation and transition infrastructure in particular. Sustainable infrastructure based solutions to maintain grid stability require years of planning and construction.

A move away from gas boilers also has to be part of this mix, especially for domestic heating use, towards electric air and ground dual-heat pump technologies. Both solutions are currently very expensive and therefore provides reason why we should have invested in the sector to push costs down. Investment, environmental and regulatory policy has to be sufficient enough to push the system towards negated climate targets.

This has been achieved with electric cars, which are vastly more capable and cheaper than they were a decade ago, with considerable investment now being made into supporting infrastructure. The same needs to be assessed with new domestic heating systems. They use little electricity to run and conveniently make most of their energy from the surrounding environment.

With renewable and nuclear electricity being able to supply some of the demand, the strategic switch to local, home- and business-based energy generation via air and ground heat pumps would transform both national energy security and households’ costs. These costs can be covered through a national switchover programme, similar to 1969, when we transitioned from town gas to transporting and installing natural gas into every home and workplace in the country. If market forces prove unable to implement a transition, then state funding will support a consumer scheme in fair and manageable way.

Installing electricity and phone lines, creating a national grid, building roads and motorways, pipelines for natural gas, laying fibre optic cables – such infrastructure projects have been done in the past, and we can do it again now. Governments are now committing to national programmes. It will pay for itself in the long run.

Recent events lay bare how vulnerable our current electric grids can be. Just as we shouldn’t be trying to build a better internal combustion engine, we can’t afford to rebuild the grid of the past. AI and software are key to a more sustainable, all-electric world, and provide a solution for power grids of the future.

The silver lining of the natural gas price spike is that it reminds us just how dangerously reliant we are on finite resources and may finally provide a catalyst for governments to reassess energy policy ahead of COP26 and to make investments into sustainable and resilient energy provisioning, along with making informed decisions supported by AI.

If you also want to take advantage of innovative AI tools for energy management, we have experts who can provide advice and free consultation. At BSI, we supply AI technology solutions and offer options for hosting systems in sustainable 100% renewable powered data centres.