Endava : How to Improve Intelligent Energy Storage Systems Using AI

AI | Uros Bajec |
05 October 2021

The generation and availability of electricity has to be perfectly matched with our consumption at all times to ensure a stable and safe supply. Energy storage can help stabilise fluctuations in demand and supply by storing excess electricity and releasing it when the demand is high, thus improving energy efficiency. This is particularly evident when it comes to renewable energy generation, such as solar and wind energy, because of their inherent intermittency: power is generated only when there is plenty of sun and wind, but this does not always coincide with the demand in the electric grid.

ENERGY STORAGE TECHNOLOGIES

A variety of technologies to store electricity are being developed at a fast pace and allow us to save energy in large quantities over different time periods, from fast storage for only seconds to longer-term storage over days:

• Batteries - either flow (such as lithium-ion) or solid-state batteries. Each battery type has its advantages and drawbacks, with the price always being an important factor for investors.
• Flywheels - an electric motor is used to spin up a wheel or rotor to store energy. The energy is then discharged by an electric generator, thus spinning down the flywheel.
• Compressed air energy storage - compressed air is utilised to create a potent energy reserve.
• Thermal - capturing heat and cold to create energy on demand. Storage media include water or ice-slush tanks, masses of native earth or bedrock accessed with heat exchangers through boreholes, deep aquifers contained between impermeable strata, shallow, lined pits filled with gravel and water and insulated at the top, as well as eutectic solutions and phase-change materials (latent heat storage (LHS) units).
• Pumped hydropower - Water is pumped from a lower-elevation reservoir to a higher elevation using low-cost surplus off-peak electric power. During periods of high electrical demand, the stored water is released through turbines to produce electric power. Although the losses of the pumping process make this kind of plant a net consumer of energy, the system increases revenue by selling more electricity during periods of peak demand when electricity prices are highest.

WHAT IS INTELLIGENT ENERGY STORAGE?

What all of these technologies have in common is that - although they are highly advanced pieces of engineering - they are still just "dumb" machines that store and release energy, but they can't:

• optimise their operations by controlling when they are charged and discharged depending on the electricity price,
• control how and when they are charged every day to achieve a longer cycle life,
• coordinate their operations with other energy storage solutions, power generation capacities, and consumers,
• predict when a failure will occur and act before it happens. For example, if a lithium-ion storage system suffers a thermal runaway or other degradation-inducing event, the batteries will not recover, and the system may become a permanently low-performing or even stranded asset.

WHERE ARTIFICIAL INTELLIGENCE COMES INTO PLAY

Artificial intelligence (AI) is an approach to computing that uses large amounts of data in order to accomplish a task. AI is particularly powerful where large volumes of data exist that can be harnessed to train computers to "think and act as a human". Energy is one such field where a lot of data exists, is readily available, well-structured, and accurate, and is therefore particularly suited for various kinds of AI solutions. We can use this data to feed it into an AI system to deliver results that can hugely benefit energy storage solutions.

Demand and generation forecasting: AI is particularly good at forecasting electricity generation and demand, and consequently the price at a particular point in time. It does so by taking into account various data sources, such as years of historical electricity data and weather data - particularly hot or cold days require more air conditioning or heating, driving the electricity demand.

Accurate forecasts not only support the safe and reliable operation of the grid by balancing electricity supply and demand; they can also substantially improve the operational efficiency of energy storage solutions. If an energy provider knows in advance what electricity will cost at each hour of the day, they can store electricity when it is cheapest and release it during the peak hours, thus maximising their assets.

Wind and solar forecasts are key to reducing the uncertainty associated with variable renewable energy generation; scheduled delivery of energy output is more valuable to the grid than standard, non-time-based delivery. Making use of that energy can be difficult because knowing how much a given farm will generate and how best to store and then deliver that energy to the grid changes every day.

To develop wind and solar generation forecasts, AI utilises a combination of weather and satellite data, numerical weather prediction models, and statistical analysis to produce estimates. A lot of energy storage solutions are deployed to balance the uncertainty of wind and solar energy generation, so accurate forecasts are crucial for these systems.

Predictive maintenance: AI can detect anomalies across a range of electrical, electro-mechanical, chemical, and thermal subsystems before they cause any damage to the system, enabling the operator to act in time. AI does so by gathering data from different sensors and the environment and comparing it to historical data that was used to teach AI what set of conditions typically leads to the failure of a certain component. Predictive maintenance can reduce downtime, improve the operational life of a storage system, prevent damage, and increase profits for the operator.

Intelligent building control: AI can be used for super-detailed modelling of energy use across buildings, including predictions of the passive solar capacity, wind speed, and building energy load. This allows us to optimally use energy storage in a building and reduce its overall energy consumption.

AI is already shaping smart grids to become truly intelligent and enables them to meet the energy demands of the future. Energy storage is becoming an increasingly important part of the smart grid, and AI will revolutionise our understanding of consumption patterns and tweak the operation of these devices to drive revenue for its operators.