Industry 4.0: Digital intelligence applied to energy management

From artificial intelligence to digital intelligence

In the digital era, artificial intelligence (AI) is used in all spheres of our society. Whether in the fields of advertising, telecommunications, or the auto industry, AI promises analytical prowess that goes far beyond human abilities. But what is the potential of AI when it comes to buildings and other industries? Our thoughts on this subject have brought us to reconsider the use of the term artificial intelligence and to replace it with digital intelligence (DI).

Digital intelligence includes all fields related to making data useful, such as data acquisition, data mining, machine learning, deep learning, and many others. According to the Institut de valorisation des données (IVADO), digital intelligence is a “... set of tools and methodologies that, in combining the collection and harnessing of data with the design and use of models and algorithms, facilitates, enriches and supports decision-making.”

Level of digital maturity

Digital intelligence can be seen as a way of creating value from raw data. There are four distinct levels of digital maturity, however, an organization or a project may be engaged at different levels at the same time. For example, the complex algorithms associated with Level 3 may be used to facilitate the acquisition, organization and transformation of data associated with Level 1.

Level 1: Measuring (Data acquisition)

This is the level where the structure for the acquisition of data is set up. The data can be consulted in real time using interface tools. The history of the data can be accessed to better understand past events and avert potential problems or anomalies. However, only basic analyses are carried out at this level. This is the most important level of maturity to handle well, since the quality, reliability and accessibility of the data remain the principal issue in digital intelligence today.

Level 2: Contextualizing (Descriptive)

The data are contextualized using classic data science techniques (e.g., regressions, statistical tools). Data contextualization facilitates conducting analyses to understand events that have happened, but not visualizing and adjusting models in real time. Concrete examples of this level of maturity are implementing an energy management information system (EMIS) or using the results from measuring and verification plans according to the International Performance Measurement and Verification Protocol (IMVP).

Level 3: Learning (Predictive)

The predictive models that are established at this stage anticipate what will happen based on a multitude of factors. Usually, machine learning algorithms are called on to apply various predictive methods aimed at anticipating the evolution of a process based on historic and current data.

Level 4: Optimizing (Prescriptive)

At this level of maturity, complex algorithms implement corrective measures aimed at optimizing the functioning of a process. These optimization techniques, without any human intervention, are associated with operational research.

Digital intelligence applied to buildings and industry

At first sight, digital intelligence may give the impression of being a “black box.” Collecting and analyzing data, and then letting a process run based on complex algorithms may not be reassuring for some people. That is why the integration of digital intelligence to a structured energy management approach is essential. As mentioned in a recent article on energy management systems, Énergir wants to support its customers in this process and to begin work on ensuring that digital intelligence is fully integrated into its energy efficiency programs.

Data acquisition

Take a concrete example – imagine a building that already has a centralized control system. It will collect and archive data which will provide operators with high performance tools to supervise and adapt the system operations. Performance drifts can be detected and analyzed following complaints, major outages, or a significant increase in energy bills.

Data contextualization

To detect energy drifts faster, an energy management system would be installed. The system automatically collects, consolidates and records data in an archive that can also store control system histories, data from energy suppliers, weather data, as well as any other pertinent factors. Controlling and pooling data helps to better define the building’s performance indicators and to develop more user-friendly dashboards that can benefit everyone, from technical personnel to the building manager.

Learning

With this archive of contextualized data, it is now possible to proactively respond when drifts are detected. Machine learning algorithms based on well-organized months of historic data mean that reference models can be developed in real time to predict the behaviour of the building, as well as identify potential improvements to its functioning.

Optimization

The final step uses the predictions to optimize systems in real time. For example, it could delay the start or limit the power demand of a system based on any variety of factors such as temperature, sunshine, occupation or productivity.

Integrating digital intelligence to a more comprehensive approach also fosters longer-term interaction between processes, building systems or production services to encourage energy exchanges. The need to manage energy performance and GHG emissions is an excellent reason for improving the level of digital maturity.

Digital intelligence applied to the industrial sector

Here are some concrete examples to clarify the applicability of artificial intelligence to an industrial setting for optimizing the energy consumption of a process.

Management of ladle preheating in the metallurgy sector.

Several industries in the metallurgy sector (aluminum, steel, etc.) use ladles or crucibles (see illustration) to transport molten metal. During molten metal transfer, to protect the refractory material lining the interior of the ladles, it is essential to maintain a high and uniform temperature to eliminate the risk of thermal shock. Natural gas burners may be used for this purpose. Since the refractory lining gets worn with each casting, it is difficult to predict when it will need to be repaired. Consequently, several standby ladles are kept at high temperature for long periods of time, leading to very high costs for natural gas consumption without any production output.

To economize on the natural gas consumed for preheating the ladles, machine learning and digital intelligence can be called on. High-definition thermal cameras could be installed to capture and analyze the temperature gradients of the steel shell in real time to detect the development of hot spots. By also integrating other production factors (using dedicated programmable logic controllers), such as temperature measurements, alloys to be produced, pace of production, number of stand-by ladles, etc. A digital model could act as a virtual operator and precisely predict the moment when a stand-by ladle needs to be preheated to replace the ladle in operation.

Predicting the precise moment when the ladle should be serviced eliminates the continuous heating of ladles on stand by and avoids operational redundancy and the high consumption of natural gas.

Advanced combustion controls for boilers furnaces and kilns

Flame and purge monitoring technologies and control over the rate of oxygen in flue gases and combustion air, are also part of the basic components of implementation toward Industry 4.0. The automatic start and stop of burners with dynamic adjustment to the flow of natural gas in a boiler, based on gas pressure and the number of burners in service, can be integrated to improve energy efficiency. In plants where energy is consumed simultaneously from different equipment (heat-recovery, biomass or natural gas boilers), it may be even better to optimize production using advanced controls and devices that can make the equipment function at its full potential, at all times, without any human intervention.

Also, when the combustion of some kilns is controlled manually, an operator may over-use the burner to ensure that the process temperature does not drop below the desired set point. Advanced automated controls would prove effective in such cases for continuously regulating the consumption of natural gas to ensure optimal combustion.

Operational safety

Given significant employee turnover, some new employees may be unaware of the safety rules regarding industrial combustion equipment. For example, in a non-automated combustion system, the rods adjusting the combustion air from a burner may be inoperable. If the operator does not rigorously monitor the equipment, the inputs of highly inefficient air/gas mixture from an energy point of view could also pose safety risks. In this scenario, recourse to a secured process control system as an Industry 4.0 component would ensure the protection of both workers and equipment.

Digital intelligence is an integral part of Industry 4.0. As shown by many applications in building and plants, here is a tool that can exploit data to improve equipment energy optimization measures. Digital intelligence can also be used to improve the operational safety of combustion equipment, even for the purpose of implementing an EMS. To take advantage of digital intelligence, Énergir customers have access to several grant programs aimed at optimizing the energy efficiency of natural gas equipment. (See boxed text)

Grant programs offered by Énergir

Énergir offers several grants to encourage its customers to progressively migrate toward Industry 4.0 and to deploy digital intelligence for energy efficiency purposes. The implementation of Industry 4.0 systems and equipment is eligible for the following programs: diagnosis, implementation, and energy management systems (EMS).

Feasibility studies grant helps companies carry out feasibility studies with an engineering consulting firm to evaluate various scenarios aimed at reducing energy consumption. The grant covers part of the cost of a feasibility study on energy efficiency measures, carried out by an engineering consulting firm accredited by Énergir. The grant offered major industry customers is the lesser of:
- 50% of the cost of the study before taxes; or
- $50,000 by account number per fiscal year.
Implementation grant facilitates the implementation of the energy efficiency measures identified in the feasibility studies. The two examples shown in this article are projects that are eligible for these grants. The maximum annual amount of the grant cannot exceed $1,000,000 or 50% of eligible expenses before taxes (estimated project additional costs), including the cost of appliances, installation and engineering. The grant under this program is awarded according to the activity sector, as well as the return on investment (ROI), calculated for each energy efficiency measure before grants.
EMS grant subsidizes the work involved in implementing an energy management system for operational improvements in energy use. The grant offers total financial assistance of $350,000, which includes an amount for each stage: diagnosis, implementation, and monitoring of measures. The EMS grant can thus contribute to financing the deployment of digital intelligence systems aimed at storing and analyzing energy data.