Industrial production process visualization with the Elastic Stack in real-time at MM Karton
1 like3,620 views
The document outlines a project at MM Karton focused on real-time industrial production process visualization using the Elastic Stack. It details the challenges, technology choices, and specific applications in the board-making process, emphasizing data management and visualization through dashboards. The document concludes with lessons learned and next steps for further implementation and integration of machine learning and data analytics across production locations.
![26
Effect:
Example: consumption of special chemical
Background: Same performance of the board but improved dosing strategy!
60
77
93
110
05 2018 06 2018 07 2018 08 2018
specific consumption [%]
- 20 %](https://image.slidesharecdn.com/elasticon-2018-frankfurt-mmkarton-181108041016/85/Industrial-production-process-visualization-with-the-Elastic-Stack-in-real-time-at-MM-Karton-26-320.jpg)
Industrial production process visualization with the Elastic Stack in real-time at MM Karton
- 1. 1 Industrial Production Process Visualization with the Elastic Stack in Real-time at MM Karton Stephan Hampe Jürgen Kerner October, 30, 2018
- 2. 2 Agenda • Introduction & context of the project • Finding the right technology • Use case presentation • Conclusion & next steps
- 3. 3 It would be really great to see the usage of all the relevant materials and chemicals from production in one dashboard, if possible in near real-time! Stephan Hampe, Technologist, MM Karton
- 4. 4 CHALLENGE ACCEPTED CANNOT BE THAT HARD ;)
- 5. 5 Let’s talk about machines What IT has in mind
- 6. 6 Let’s talk about machines What technologists have in mind – board making machine
- 7. 7 Let’s talk about machines What technologists have in mind – board winder
- 8. 8 Let’s talk about machines What technologists have in mind – cross cutting machine
- 9. 9 Let’s talk about machines What technologists have in mind – printing machine
- 10. 10 Let’s talk about machines What technologists have in mind – die cutting machine
- 11. 11 Let’s talk about machines What technologists have in mind – gluing machine
- 12. 12 Glossary Common standards in IT and OT field IT OT Cables Twisted pair copper, fiber Twisted pair copper, fiber Connectors RJ45, ST, LC, … RJ45, ST, LC, … Protocols TCP, UDP, ICMP, … ModbusTCP, Profinet, … Messaging protocols JSON, REST, SOAP, XMPP,… OPC-DA, OPC-UA, MQTT, Modbus, Profibus,.. Standard MTU 1500 bytes up to 255 bytes Standard latency < 3ms ~ 50µs Standard speeds 1-40 Gbit/s 2-100 Mbit/s
- 13. 13 WE JUST NEED TO BUY A SYSTEM WHERE ALL DATA COMES IN AND ML, AI AND DEEP LEARNING DOES THE REST
- 14. 14 Data heat cycle Predict Optimize Empower EmbedDiscover Trust Describe Find available data and check if there is value in my data? Can I trust my data? What happened and why? What might happen next? What is the right conclusion for your business? Is insight being delivered to the right people at the right time? How do you embed the new data analytics knowledge to your organization? Start Source: PWC How it should be
- 15. 15 Data heat cycle Predict Optimize Empower EmbedDiscover Trust Describe Find available data and check if there is value in my data? Can I trust my data? What happened and why? What might happen next? What is the right conclusion for your business? Is insight being delivered to the right people at the right time? How do you embed the new data analytics knowledge to your organization? Start Source: PWC Where most approaches stop Still valid and usefull
- 16. 16 Elastic Stack journey of MM Okt. 2016 Apr. 2017 Mai 2017 Nov 2017 Start with basic license features Switch to platinum license features Implementation of IT use cases. Mainly logging Start of Business Use Case – Production process visualization Mar. 2018 Finished Development of Business Use Case
- 17. 17 Why MM chose Elastic Stack? • Open code • Right to contribute • Speed of development from Elastic • Same stack across the group in different editions ‒ Production locations basic subscription license features ‒ Corporate IT platinum subscription license features ‒ No additional effort when migrating projects to Corporate IT • Kibana ‒ Ease of use for not yet IT affine people (like electrical-, automation-, maintenance engineers)
- 18. 18 Why MM chose Elastic Stack? • Specific trainings for target user groups ‒ IT Elastic Search Engineer ‒ BI Elastic Stack Data Administration ‒ Production Kibana • Sufficient feature set for production in one tool • Nice UX across user groups • Performance
- 19. 19 LET’S TALK ABOUT BOARD MAKING
- 20. 20 Board making Stock- Preparation Approach- Flow Board Machine • “treat the fiber” • Chemical pulp • Mechanical pulp • Other fiber qualities • “mix them together!” • Fibers/pulp • Additives • Chemicals • “build the board!” • 3 different fiber layers • 2 surface applications • 3 coating layers
- 21. 21 Sensors and relevant process data Stock- Preparation Approach- Flow Board Machine • Flows • Consistencies • Flows • Consistencies • Speeds • I/O´s (BM condition)
- 22. 22 From 4-20mA to hard facts DigitalTwin Mass balance & Virtual board structure Controlling/BI Recipes & Production data Visualization DashBoard with KIBANA
- 23. 23 Why? Responsible use of resources Recipe Grade change Process stability inventory 30 day´s 1 hour --- “traditional situation”
- 24. 24 Why? Responsible use of resources Recipe Grade change Process stability real-time real-time real-time “situation today”
- 26. 26 Effect: Example: consumption of special chemical Background: Same performance of the board but improved dosing strategy! 60 77 93 110 05 2018 06 2018 07 2018 08 2018 specific consumption [%] - 20 %
- 27. 27 Recipe System Calculation Engine Securables Database (Tags, Materials) API Gateway PLC Systems Datawarehouse Group reporting Process Information Systems DA / UA Our approach Combining tradition with Elastic Stack
- 28. 28 Example dashboard running in control rooms
- 29. 29 Real life view from dashboard in control rooms
- 30. 30 Lessons learned Visualization board making process Have a clear vision of what you want to achieve Kibana rulez ;) – people love it3 Best performance from Elastic is useless when sensors are not calibrated inside board machine 4 Once it is visualized it’s the most normal thing in the world5 Have the data heat cycle in mind – we often had to go back to the beginning and in the end bring missing sensors to the board machine 2 1
- 31. 31 Next steps Visualization board making process Roll out process monitor to all production locations Include data from other systems (lab, quality system, etc.)3 Machine learning for analysis of critical process conditions4 Canvas dashboards for mill management (mobile applications)5 Include other areas with potential (finishing, cutting, etc.)2 1
- 32. 32 This was just the beginning Visit us at the AMA booth Stephan Hampe stephan.hampe@mm-karton.com Jürgen Kerner juergen.kerner@mm-karton.com