Monitor and improve the efficiency of production processes (challenge retrieved from the literature).
Challenge:
Energy prices are continuing to rise higher and higher. It is at the point where energy costs are comparable to the cost of raw materials. The company is a medium-sized enterprise consisting of several small food product manufacturers. They are faced with the challenge of needing to reduce their energy costs while maintaining production results.
Main Requirements:
- Monitor the production state.
- Identify the different sources of energy consumption.
- Recognize patterns in energy consumption.
- Automatically produce insights and analytics that can inform production decisions for reducing energy consumption.
Key Performance Indicators:
- Units produced.
- Energy saved (in kWh).
Industry Sector:
Manufacturing
Challenge classification:
Production optimization; Digital monitoring of production states; Green IIoT; Improve the cost-effectiveness and eco-friendliness of manufacturing processes.
Time for Project Completion:
N/A
____
Preliminary Analysis:
To meet the need of reducing energy consumption for a medium-sized enterprise made up of smaller food product manufacturers, there are two current approaches in the existing literature. The first uses machine learning and optimization algorithms to find the optimal input parameters and state of production equipment. This is scalable and uses a simple architecture of only one model, but the savings opportunities are limited. The second uses some degree of IoT data collection and then revolves around manual expert data analysis to find the most meaningful and tailored energy-saving opportunities. This is the ideal in terms of the quality of results, but is expensive, time-consuming and does not scale well across the subsidiaries without the higher cost.
We think the best approach to meeting the needs of the company would be leveraging some of the benefits of each current method. That is, a method based on multiple models for understanding the state of the production, finding patterns in consumption and highlighting that information for decision-makers to act on. This solution can be encapsulated on an edge device to increase scalability and convenience of deployment in new environments.
____
Solution Summary:
The proposed solution seeks to balance scalability and quality of suggestions. It, therefore, comprises two major components: 1) a layer of ML models for converting machine data and industry management information into a holistic digital representation of the production state. 2) an analytics layer that deduces points of interest for decision-makers.
Additional Benefits:
- In scaling the solution across subsidiaries, the company can consider providing the solution to other companies as well.
- The data collection set in place can act as a foundation for further innovation.
- Unlike a tailored solution which is a one-off high-quality consultancy, the solution would remain continuously monitoring the company as they continue to change and evolve.
Possible Difficulties:
- There can be a lack of data or means of collection available which would prolong the first stage.
- It may be challenging to decouple individual products in a machine’s usage.
- The resource constraints of an edge device could limit the solution’s performance.
Sources:
[1] Diogo A.C. Narciso, F.G. Martins (2020), “Application of machine learning tools for energy efficiency in industry: A review”, Energy Reports, Volume6, Pages 1181-1199, ISSN 2352-4847.
[2] Zhiqiang Geng, Rongfu Zeng, Yongming Han, Yanhua Zhong, Hua Fu(2019), “Energy efficiency evaluation and energy saving based on DEA integrated affinity propagation clustering: Case study of complex petro-chemical industries”, Energy, Volume 179, Pages 863-875, ISSN 0360-5442.
[3] Benedikt Beisheim, Keivan Rahimi-Adli, Stefan Krämer, Sebastian En-gell (2019), “Energy performance analysis of continuous processes using surrogate models”, Energy, Volume 183, Pages 776-787, ISSN 0360-5442.
[4] Feifei Shen, Liang Zhao, Wenli Du, Weimin Zhong, Feng Qian (2020),”Large-scale industrial energy systems optimization under uncertainty: A data-driven robust optimization approach”, Applied Energy, Volume259, 114199, ISSN 0306-2619.
[5] Jian Qin, Ying Liu, Roger Grosvenor, Franck Lacan, Zhigang Jiang (2020),”Deep learning-driven particle swarm optimisation for additive manu-facturing energy optimisation”, Journal of Cleaner Production, Volume245, 118702, ISSN 0959-6526.
[6] Renzhi Lu, Yi-Chang Li, Yuting Li, Junhui Jiang, Yuemin Ding (2020),”Multi-agent deep reinforcement learning based demand response for discrete manufacturing systems energy management”, Applied Energy,Volume 276, 115473, ISSN 0306-2619.
[7] Da-sheng Lee, Yan-Tang Chen, Shih-Lung Chao (2022), “Universal work-flow of artificial intelligence for energy saving”, Energy Reports, Volume8, Pages 1602-1633, ISSN 2352-4847.
[8] Zerynth, «Real-Time Production Performance Monitoring», Accessed 21 July 2023. https://zerynth.com/customers/case-studies/real-time-production-performance-monitoring/.
[9] Jagtap, S, Rahimifard, S, Duong, LNK. (2022). Real-time data collection to improve energy efficiency: A case study of food manufacturer. J Food Process Preserv, 46:e14338.
[10] Jiahao Yang, Yingfeng Zhang, Yun Huang, Jingxiang Lv & Kai Wang(2023) Multi-objective optimization of milling process: exploring trade-off among energy consumption, time consumption and surface roughness, International Journal of Computer Integrated Manufacturing, 36:2, 219-238.