Seán McLoone

Queen’s University Belfast.
sean-mcloone-queenslandscapelogocolour

Topic:

Energy monitoring and reduction – the role of manufacturing informatics.

Synopsis:

Manufacturing informatics, which spans several scientific fields, including statistical data analysis, machine learning, modelling, optimisation and control, is a key enabler for low-cost energy monitoring, and for the identification and delivery of a range of energy cost reduction opportunities. These are briefly introduced in this article and the need for strong academic-industry partnerships to overcome the barriers to realising these opportunities highlighted.

Takeaways:

  • Comprehensive energy monitoring infrastructure is needed in order to identify and fully exploit energy reduction opportunities.
  • To be effective, monitoring of energy usage needs to be at the component level rather than machine level, and be at appropriate spatial and temporal resolutions.
  • Manufacturing informatics, which is at the heart of the Industry 4.0, has a key role to play in energy monitoring and energy reduction.
  • Strong academic-industry partnerships targeting innovative bespoke solutions, and the development of undergraduate and graduate training programmes in manufacturing informatics are needed to overcome the barriers to realising energy reduction opportunities.

Submission:

There are many opportunities to reduce energy costs in manufacturing from redesigning production systems and processes, to more efficient operation and management of existing systems. Roland Berger Strategy Consultants estimate that energy intensive industries can get a 4-fold return on investment in energy efficiency measures and technology through to 2050 [1]. ZVEI [2] estimates that exploitation of automation technology can deliver energy saving of between 10 and 25 % in machines and plants deployed in Germany. Similar opportunities exist in Ireland. The total turnover of Irish industry was €102.4 billion in 2014. A recent survey of 40 manufacturing companies in Ireland identified energy costs as ranging from 2-7% of total turnover. If we assume a base level of 5%, measures that would yield a 20% reduction in energy usage would benefit manufacturing in Ireland to the tune of €1 billion. Hence investing in energy reduction technologies and operating strategies is not just an environmental imperative – it is also good business.

The starting point is putting in place a comprehensive energy monitoring infrastructure in order to collect detailed energy usage data, which can then be analysed to determine where the best energy reduction opportunities lie. To be effective, monitoring of energy usage needs to be at the component level rather than machine level, and be at appropriate spatial and temporal resolutions, to capture the dynamics of energy use and flow throughout the production cycle. This is a challenging task due to the diversity of standard and non-standard communication protocols and interfaces that exist on industrial machines, inaccessible propriety systems, and the absence of the necessary localised sensing, data acquisition and external transmission functionality on legacy systems. QUB [3] and IMR [4] are actively working on developing low-cost non-obstructive retro-fit monitoring solutions in this space leveraging recent advances in wireless, internet of things (IoT) and cloud technologies.

Manufacturing informatics, which can be loosely defined as the use of ICT and data analytics in manufacturing, is a multi-disciplinary field spanning, engineering, computer science and mathematics. It draws on a range of scientific fields, including statistical data analysis, machine learning, modelling, optimisation and control, to identify opportunities for, and deliver enhanced performance from manufacturing systems. It is at the heart of the Industry 4.0. As such, it has an important role to play in energy monitoring and reduction. We have a variety of possibilities, including:

  • Data driven modelling for low-cost energy monitoring/inference of energy usage (a.k.a. soft sensing) – by relating more readily available process variable and control input data to component energy usage, soft sensing models can be developed to predict energy usage. These models can then be used across multiple production machines reducing the need for comprehensive energy monitoring infrastructure on all machines.
  • Data analysis to identify energy usage patterns and causalities (temporal and spatial) – potentially leading to actionable energy reduction insights. For example, analysing the ramp-up of oven temperature profiles at start-up can highlight cost savings in relation to reduced ramp-up and stabilisation periods, and reduction in energy use during idling periods.
  • Optimisation of production and maintenance scheduling to minimise energy costs. Optimisation can be employed across all aspects of process design, operation and scheduling to minimise cost and maximise performance. In the context of energy reduction, where there is flexibility in scheduling production and maintenance activities, scheduling can be optimised to align peak energy usage with periods of low energy cost/high levels of renewable generation (as reflected in time-of-use (ToU) energy tariffs).
  • Generation of a demand-response revenue stream – If short-term interruptions in production can be tolerated, and there is sufficient flexibility in scheduling production, a particularly attractive proposition is to provide demand-response auxiliary services to the electricity market [5]. This has the twin benefits of providing an additional revenue stream (offsetting energy costs) and supporting the decarbonisation of electricity generation.
  • The development of advanced control solutions for energy efficient process operation – Poorly tuned control loops can lead to increased energy losses as well as accelerated degradation of valves and actuators. Data analysis can help identify control loop performance issues and highlight where more advanced control methodologies such as self-tuning/adaptive control [6] and Model Predictive Control (MPC) [7] may be of benefit. Self-tuning controllers adapt control parameters to remain within the desired performance envelope as operating conditions change, while MPC can directly consider energy consumption as part of the control performance objective.
  • Enhanced equipment health monitoring – Monitoring energy usage variation can provide valuable data for, and enhance the performance of, equipment health monitoring systems. This in turn facilitates the development of predictive maintenance strategies, enabling more timely and cost effective maintenance interventions, and ultimately more energy efficient plant operation.

Realising these possibilities will become increasingly important as we strive to meet the demanding climate change targets agreed at COP21. The challenges for many manufactures, especially SMEs, are the up-front investment in energy monitoring infrastructure and accessing the informatics expertise needed to exploit energy reduction opportunities. These challenges can be addressed by strong academic-industry partnerships targeting innovative bespoke solutions, and the development of undergraduate and graduate training programmes in this highly sought after skill set.

References

[1] Roland Berger Strategy Consultants: Study: Boosting efficiency in electricity-intensive industries, outlook and strategies for action through to 2050, Munich, August 2011

[2] ZVETI, More energy efficiency through process automation, www.zvei.org/Publikationen/ ZVEI_Energienutzung-englisch.pdf, April 2012

[3] Point Energy, http://nispconnect.org/techwatch/pointenergy/

[4] Irish Manufacturing Research, http://imr.ie/research/cost-energy-reduction/

[5] SEM Demand Side Vision for 2020, SEM-11-022, May 2011. https://www.semcommittee.com/ news-centre/demand-side-vision-2020

[6] J. Deng et al., Energy monitoring and quality control of a single screw extruder, Applied Energy, vol. 113, pp. 1775-1785, January 2014.

[7] C. Garcia, D. Prett, and M. Morari (1989), Model predictive control: theory and practice, Automatica, vol. 25 (3): pp. 335–348., 1989.

About Seán McLoone:

Prof Seán McLoone is Professor of Applied Computational Intelligence, Director of the Energy Power and Intelligent Control (EPIC) Research Cluster, and Principal Investigator for the Pioneer Research Programme on Intelligent Autonomous Manufacturing Systems (i-AMS) at Queen’s University Belfast. His research interests are in computational intelligence techniques and data analytics with applications in smart-grid and advanced manufacturing informatics. Specialist areas include predictive modelling, unsupervised sparse feature selection, clustering and blind identification. His research activities to date have been supported by funding from a range of sources including Science Foundation Ireland, Enterprise Ireland, EPSRC (UK), FP7 (Europe) and industry. His research has a strong application focus with many projects undertaken in collaboration with companies in the manufacturing and power sectors. At a professional level, Prof McLoone is a Chartered Engineer, a Fellow of the Institute of Engineering Technology (IET), a Senior Member of the Institute of Electrical and Electronic Engineers (IEEE) and a non-executive Director on the Board of Directors of Irish Manufacturing Research. He also severs on the editorial boards of the international peer reviewed journals ‘Engineering Applicators of Artificial Intelligence’ and ‘Transactions of the Institute of Measurement and Control’.

Contacting Sean McLoon:

Energy, Power and Intelligent Control Research Cluster, School of Electronics, Electrical Engineering and Computer Science, Queen’s University Belfast.

See www.qub.ac.uk/epic

 

Leave a Reply

Your e-mail address will not be published. Required fields are marked *