Ambitions

State-of-the-art

The current state-of-the-art is that factories put enough solar on their roof to cover all or most of their operations, and then add batteries or grid-supplied power.

Some companies have maximised solar PV investment including on their car park. ABB and Neoom (Austria) for example use them also as trial sites for their own technologies.

This project goes beyond the state-of-the-art with finely-grained real-time data analysis (per machine or appliance), AI-based predictions and scheduling.

The ambition is to achieve a process and a system that makes it quite "normal" to aim for 100% self-sufficiency. There are fears to be overcome, as the system "somehow intervenes" in factory processes. Solar forecasts are deemed to be very accurate, but freak weather events do happen.

The process will be to look at each machine/appliance type to see how flexible it can be. This depends on the nature of the machine, appliance or production line, and on how it is integrated in the overall production process.

For example;

• Electric forklift trucks and delivery vans/trucks can be charged flexibly, provided drivers don't regularly have to drive long distances.

• Temperature-related non-production appliances (heating, cooling, ventilation, air con, refrigeration) have a measurable level of thermal inertia so can be used for energy flexibility

• Temperature-related production machinery also benefits from thermal inertia. They include for example blast chillers and heaters in food production, plastic extrusion machinery (melts plastic), various solutions and baths in the chemical and metals industry etc.

• One-off or weekly production processes or machines that are used rarely can more easily be allocated times when there is a lot of sun. This is more so when machines can be operated remotely (e.g. during sunny hours on weekends).

• Repeat production runs and entire production lines are naturally less flexible, especially with just-in-time production, where only as many product inputs are produced as needed right now.

So this project will have to create:

• a process/mechanism of connecting to machines
• a general framework to enable industrial energy consumption to be connected to renewable energy.

In order to forecast an effective energy consumption, we will combine approaches such as:

• Time series (autoregressive, moving average, autoregressive moving average, etc)

• ANNs (Artificial Neural Networks), in order to obtain algorithms suited for large datasets

• Ensemble methods, for enhanced accuracy. By integrating the outputs of various algorithms such as decision trees, support vector machines and neural networks, ensemble methods will mitigate individual model biases and errors. In addition, ensemble methods will enable the incorporation of multiple data sources (historical data, weather conditions, etc.)

• Incorporating external factors such as public holidays, industrial activities, events, policy changes, etc. into ML algorithms will enable the models to capture the intricate relationships between energy usage and external dynamics

• Transfer Learning for adaptability. By leveraging pre-trained models on similar energy consumption datasets, transfer learning will significantly reduce the time and resources required to train new forecasting models from scratch.

• Online Learning for Real-time Adjustments. The dynamic nature of energy consumption necessitates real-time adjustments in forecasting models. Online learning, a technique where models are updated continuously as new data becomes available, addresses this need by enabling models to adapt to changing consumption patterns on the fly. By incorporating the most recent data, online learning ensures that forecasting models remain up-to-date and capable of providing accurate predictions in rapidly changing environments.

• Explainable AI for Stakeholder Confidence. Interpretable models allow energy managers, policymakers, and other stakeholders to understand the underlying factors driving forecasts and make informed decisions based on the generated insights. This transparency is essential for facilitating the adoption of AI-powered forecasting tools across diverse industries and sectors.

Technology Readiness Level (TRL)

GridDuck overall system: at TRL 9 (commercially available and in market, 1,200 sensors and switches live).

GridDuck AI algorithms: TRL 7. GridDuck has put together an energy data prediction system based on open-source algorithms. This system has successfully been tested to predict the consumption curves of appliances (hospitality fridges in this case). GridDuck is currently testing these algorithms as part of a predictive maintenance trial with manufacturing clients.

GridDuck will progress the TRL from 7 to 8 as part of this project by collecting data immediately after project start, and by tracking and improving the prediction accuracy month by month for each appliance/machine.

Incorporating the advanced methods and strategies described in the section "state of the art" into energy consumption forecasting not only enhances the accuracy of predictions but also broadens the application of ML and AI techniques in the energy sector. By continuously improving the quality of forecasts and enabling adaptable models, the integration of advanced technologies offers the potential to revolutionize energy management practices and contribute to a more sustainable and efficient energy future.

GridDuck production system integration: TRL7. GridDuck has experience of data collection via feeds, software APIs, building communication standards (ModBus, MBus etc.) as well as by using IoT communication hardware (CT clamps, relays, sensors). ModBus has become a standard communications protocol in industry and is now the most commonly available means of connecting industrial electronic devices.

Plan to reach TRL 8: we will use ModBus and other means to connect with machinery at TEC Eurolab, Diversey and Metal Office. All three have given us a list of machines and appliances.

Weather forecasts: TRL 9, they are readily available from different sources.