Differentiating with RL in the Industrial Energy Efficiency Landscape

Energy efficiency is applied in different sectors and is implemented with different technologies, encompassing hardware and software. Recently, also with massive exploitation of data from IoT (Internet of Things) and AI.

The most commonly crowded application is in building, where HVAC (High Volume Air Conditioning), whether domestic and business, is the king with 30% of dissipated energy. According IEA (2020), USD 26bn out of 86bn belong to measures associated to buildings. The growth in AC adoption on domestic (70% peak in demand of cooling) and business has reached 60% Y-to-Y in certain markets (China), to protect humans beings from increasing temperature in the last years.

IEA, Global energy efficiency-related end-use investment in the Net Zero Scenario, 2019–2030, IEA, Paris https://www.iea.org/data-and-statistics/charts/global-energy-efficiency-related-end-use-investment-in-the-net-zero-scenario-2019-2030, IEA. Licence: CC BY 4.0

Energy efficiency grew 16% in 2022 as a Govern stimulus (up to USD 600bn). In buildings, efficiency is accountable for USD 255bn (light blue), while end uses and electrification USD 91bn (dark blue). Industry (green) is accountable for USD 25bn.

To reach the energy savings milestones in the NZE Scenario, efficiency will need to improve by almost 40% in residential buildings, by just under 35% in car travel and by just over 15% in industry by the end of the decade. This means that gains which took around two decades in the past will need to be achieved in just 8 years, highlighting the need for technological innovation and accelerated deployment. (IEA)

IEA, Groups of actions contributing to a doubling in the rate of annual primary energy intensity improvements in the Net Zero Emissions by 2050 Scenario, IEA, Paris https://www.iea.org/data-and-statistics/charts/groups-of-actions-contributing-to-a-doubling-in-the-rate-of-annual-primary-energy-intensity-improvements-in-the-net-zero-emissions-by-2050-scenario, IEA. Licence: CC BY 4.0

One of a series of interim steps to achieving this is to double the rate of average global primary energy intensity improvement from the 2022 level just over 2%, to slightly above 4% until 2030. This rate of improvement would mean that energy demand in 2030 is nearly 10% lower than in 2022, even as the global economy expands by almost 30%. (IEA)

So, 0.7% comes from dismanling less efficient fleet, while 0.8% from better energy use and consumer changes.

In industry, annual energy productivity grows by 2.3% per year (regardless the weather condition influencing the and electricity accounts for 30% of energy use by 2030.

However, there are more sectors where there is more room for efficiency, because of the magnitude of energy usage. The industrial processes, whether using electricity (i.e. metal accountable for 7% of the overall energy system emissions, paper, chemical or food processing) or natural gas (i.e. cement, glass, and more), still requires a lot of effort to reach that 20% estimated available to

Positioning in Energy Intensive Industries

Because of being capital-intensive, energy intensive sectors, like the steel industry, for instance, the long-lasting assets typically operate over 40 years. To ensure that the best available technologies are optimally used in new and existing production facilities, efficiency can be improved by:

• optimising processes through better controls and monitoring,
• upgrading process equipment to commercially available best available technologies, and integrating digitisation with SCADA or OPC-like systems,
• integrating artificial intelligence to enhance productivity,
• recovering waste heat, for example by using coke dry quenching or top-pressure recovery turbines. (IEA)

Hence, industrial processes stand as a pivotal niche where advancements can significantly impact sustainability goals and operational efficiency. Since 2020, the industrial energy efficiency market has undergone substantial segmentation, propelled by technological innovations and the imperative for resource optimisation. Let’s embark on a journey to explore this dynamic landscape, from key segments to the game-changing role of AI, particularly real-time AI technology, like Reinforcement Learning.

Segmenting the Industrial Energy Efficiency Market:

Segmentation in the industrial energy efficiency domain delineates various sectors, each with its distinctive energy consumption patterns and optimisation strategies. According to data sourced from the International Energy Agency (IEA), key segments include manufacturing, chemical processing, heavy machinery, and utilities. These segments represent diverse industries with unique energy demands and operational challenges.

IEA, Manufacturing energy efficiency, New Policies Scenario vs. Efficient World Scenario, 2017–2040, IEA, Paris https://www.iea.org/data-and-statistics/charts/manufacturing-energy-efficiency-new-policies-scenario-vs-efficient-world-scenario-2017-2040, IEA. Licence: CC BY 4.0

NPS represents New Policy Scenario while EWS, Efficient World Scenario. Although there are many policy from Governs, there is urgency to change the mindset and invest in operational energy efficiency.

IEA, Manufacturing energy intensity in IEA member countries and major emerging economies, 2000–2017, IEA, Paris https://www.iea.org/data-and-statistics/charts/manufacturing-energy-intensity-in-iea-member-countries-and-major-emerging-economies-2000-2017, IEA. Licence: CC BY 4.0

While an energy management system creates the ground-floor and processes for a company to monitor energy consumption and efficiency, industrial productivity gains by improving energy performance and overall system efficiency. In a more holistic view, this can include enhanced production and capacity utilisation, reduced resource use and pollution, and lower operation and maintenance costs — all of which result in increased value generation, and thus improved competitiveness.

In industrial processes, energy efficiency is paramount for enhancing productivity, while minimising environmental impact. Key players in this arena encompass a spectrum of stakeholders, ranging from multinational corporations to specialised technology providers.

Companies like Siemens, ABB, Schneider Electric, ABB and Rockwell Automation have established themselves as leaders in delivering solutions for industrial energy efficiency through a combination of hardware, software, and data-driven insights.

[energy efficiency] … can include enhanced production and capacity utilisation, reduced resource use and pollution, and lower operation and maintenance costs — all of which result in increased value generation, and thus improved competitiveness.

Machine Learning and AI Revolutionising Industrial Energy Efficiency:

Beyond the key technological dimensions of efficiency improvements, efficiency opportunities will also come from digitalisation, which is opening up a new generation of efficiency thinking and increasing its value to energy systems. Digitalisation includes not only enhanced data collection but also analytics that produce actionable insights by drawing on computation of vast amounts of data, as well as greater connectivity that increases the exchange of data between machines and humans. (IEA)

The advent of ML (Machine Learning) and AI (Artificial Intelligence) has revolutionised the approach to industrial energy efficiency, ushering in an new era of:

• predictive maintenance,
• optimised operations,
• and adaptive control systems.

By analysing vast volumes of data generated by industrial processes, machine learning algorithms can identify inefficiencies, predict equipment failures, and recommend optimisation strategies in real-time.

Digital technologies also allow for a more modernised, rounded concept of energy efficiency that considers system optimisation. The greatest potential efficiencies lie in large industrial processes that can provide demand-side flexibility for grids. This means new revenue stream and opportunities for industrial manufacturers to trade in energy markets.

In the United States, for example, the combined wholesale demand response capacity of all regional system operators grew to around 27 GW by 2020 (around 6% of peak demand), with an additional 5 GW offered through retail programmes. Load control, interruptibility services and reserves markets also expanded elsewhere in the country.

Accordingly IEA, efficiency improvements in industrial manufacturing are a mainstay of long-term decarbonisation strategies. Its potential improvement in lowering energy intensity of industrial activity (on global average) is 25% to 30%, particularly in the aluminium, paper and cement sub-sectors.

Reinforcement Learning (RL), used for self-driving cars, involves the use of rewards (positive reinforcement) and penalties (negative reinforcement) to incentivize the model to adjust its outputs. Value-based RL — often executed with algorithms such as Q-learning and SARSA, which are included in the outermost layer of the Palette — is a common approach by which the model learns to associate certain actions with rewards and adapt to different situations. This approach is useful in scenarios where it is impractical to provide detailed supervision for every individual step or state change. (Glasswing.vc)

Real-time optimisation

Real-time AI, particularly Reinforcement Learning, stands out as paradigm shift in how industrial processes are optimised for energy efficiency. Unlike traditional rule-based systems, reinforcement learning enables autonomous decision-making based on real-time feedback and dynamic environmental conditions. This capability is particularly valuable in industries where operational parameters fluctuate rapidly, such as manufacturing and chemical processing.

Real-time AI, particularly Reinforcement Learning, stands out as paradigm shift in how industrial processes are optimised for energy efficiency.

Reinforcement Learning (RL) has been key to enhancing efficiency: by optimising parameters, i.e. temperature, pressure and more, RL significantly reduces energy use and improves operations, on multiple machines in real time.

Its capability for real-time monitoring and executing allows for early defect detection, ensuring superior quality. RL also excels in resource scheduling, minimising costs and GHG emissions, while fulfilling customer demands swiftly. Additionally, its adaptability enables quick product customisation to meet changing preferences, elevating manufacturing innovation and sustainability. This approach not only boosts efficiency, but also sets new industry standards.

Pioneering the Frontier of Real-time RL in Industrial Energy Efficiency:

BeChained specialises in eliminating wasted energy in manufacturing processes, by leveraging RL algorithms to

• continuously optimise operations in real-time,
• learning from on-the-run data,
• and reacting to environmental stimulus, guaranteeing real-time adaptability,
• with 92–95% in increase of performance of the system (i.e. resource management, energy efficiency, and more).

Although several key players in industrial energy efficiency landscape are championing data-driven solutions, only few players focus on:

• hardware-agnostic approaches,
• seamlessly integrating with existing industrial infrastructure through protocols like SCADA (Supervisory Control and Data Acquisition) or OPC-UA (Open Platform Communications Unified Architecture)
• and executing improvements in production programming on MES (Manufacturing Execution Systems),
• meanwhile, the performance measurements are stored in a blockchain infrastructure for carbon accountability. This approach ensures a future transparency for demand flexibility and CO2 credits opportunities.

Few business cases have actually mentioned to use this technology in manufacturing production to enhance energy efficiency in industry and control physical systems (machines).

The start-up soon will file a patent to protect his unique approach

By combining data analytics, RL, and advanced control systems, these companies enable industries to achieve unprecedented levels of energy efficiency and operational excellence.

Conclusion

The industrial energy efficiency market has witnessed remarkable segmentation and technological advancements since 2020, driven by the imperative for sustainability and operational optimisation.

With the rise of ML and real-time RL, the future of industrial energy efficiency holds immense promise for unlocking new levels of efficiency, productivity, and environmental stewardship.