Recommended
Recommended
More Related Content
Similar to How Machine Learning and Cloud Technology turn Plastics Manufacturers into Data-Driven Powerhouses.pptx
Similar to How Machine Learning and Cloud Technology turn Plastics Manufacturers into Data-Driven Powerhouses.pptx (20)
Recently uploaded
Recently uploaded (20)
How Machine Learning and Cloud Technology turn Plastics Manufacturers into Data-Driven Powerhouses.pptx
- 1. How Machine Learning and Cloud Technology turn Plastics Manufacturers into Data-Driven Powerhouses automatica 2023 | 30.06.2023
- 2. WHO IS BEHIND sensXPERT? Parent Company: NETZSCH Family-owned global technology leader with 4100+ employees present in 36 countries Corporate Venture: NETZSCH Process Intelligence GmbH Enhancing productivity through advanced process analysis technology for the industry 4.0 As a recognised industry expert with 50+ years of experience in material science and sensor technology, it is the NETZSCH Group that transforms industries with next-level intelligence for increased efficiency, quality assurance and process reliability for the plastics industry. &
- 11. How does it work?
- 12. MATERIAL KNOWLEDGE AT THE HEART OF PROCESS ENHANCEMENT • Measuring material behavior • resin viscosity, degree of cure, glass-transition, material condition (mixing ratio, ageing, shrinkage, contamination) • Combined with third-party measurement devices (pressure transducers, thermocouples, and more) • Thermosets, rubbers and thermoplastics • Fiber reinforced polymers • Sands or natural stones bonded with resin • (Reaction) Injection Molding • Thermoforming & Compression Molding • Transfer Moulding Processes • Vacuum infusion & Autoclave Curing MATERIALS PROCESSES Real-Time Material Characterization with Dielectric Sensors
- 13. Dielectric Sensor Information Introduction of the material Minimum resin viscosity Progression of cure / gelation / crystallization Completion of cure / crystallization sensXPERT® process data
- 14. MACHINE LEARNING AND PROCESS OPTIMIZATION Dielectric Measurements Kinetic Model Degree of Cure Glass Transition Temperature Data Preprocessing ... ML Model Data Generation Prediction & Optimization Training & Testing +
- 15. Simulation and Optimization A data driven solution. • AI model calculates and predicts material properties • Dynamically control and adapt the process to ensure constant quality Real-Time Process Optimization
- 16. CLOUD SERVICE • Process data transparency • 24/7 access on any device • Customizable dashboards • OpenAPI: flexible data handling
- 17. of existing scrap reduction 50% Up to energy savings 23% Up to cycle time reduction 30% Up to (re)- commissioning time Reduce RESULTS OF sensXPERT®
- 18. Let us dive deep into the material behavior during production
- 19. Small series production in collaboration with Schwarz Plastic Solutions Influence of shearing on the curing
- 20. Variation in injection time affects the cure behavior
- 21. Variation in gate and injection time
- 22. Consolidated summary of process variations and cure behavior
- 23. Small series production in collaboration with Schwarz Plastic Solutions Influence of batch-to-batch-variations on the curing
- 24. Batch-to-batch variations affect the cure behavior
- 25. sensXPERT lab trials Influence of batch-to-batch-variations on the curing
- 26. Aging affects the cure behavior
- 28. sensXPERT USE CASE: OPTIMIZING ELECTRONICS ENCAPSULATION FOR E-MOBILITY Component: High-power electronics circuit board Material: Epoxy molding compound Quality criterion: Degree of cure > 90% Initial cycle time: 3 minutes
- 29. Variation in the curing time vs. DoC theoretical relation Theoretical cure behavior vs reality
- 30. Time interval for 90% DoC • Prediction of DoC after 30 s of curing • Prediction matches target DoC almost perfectly
- 31. - single machine
- 32. • Coming soon: • Thermoplastic materials • sensXPERT in your production? Outlook sensXPERT digital mold
- 33. We look forward to welcoming you into the sensXPERT community! Contact us NETZSCH Process Intelligence GmbH Dr. Nicholas Ecke Gebrüder-Netzsch-Str. 19 nicholas.ecke@sensxpert.com 95100 Selb, Germany
Editor's Notes
- As a recognized industry expert with almost 50 years of experience in material science and sensor technology, it is the NETZSCH Group that transforms industries with next-level intelligence for increased efficiency, quality assurance and process reliability.
- Dielctric analysis is the heart of the sensXPERT solution. It can measure and monitor important material behavior including viscosity, cure, glass transition temperature, and more. It is capable of measuring numerous types of materials used in a variety of plastic processing technologies.
- With dielectric analysis, sensXPERT can measure and record valuable process data. The ion viscosity data (blue) is an analog to the mechanical viscosity of the material (the ability for the material to flow). Point 1 shows when the material was introduced to the mold and reaches the dielectric sensor Point 2 shows where the minimum viscosity of the material occurs. The viscosity is important because it affects the ability of the resin to fill the mold. Lower viscosity allows for quicker mold filling. Point 3 shows the increase of viscosity as the material begins to cure at the elevated temperature. The inflection point of the ion viscosity curve indicates the gel point, which is where the resin can no longer flow. Point 4 shows the ion viscosity plateau marking the end of the curing reaction. It is clear the reaction completes around 120 minutes, and there is opportunity to optimize this curing cycle to reduce the cycle time.
- sensXPERT is more than just dielectric analysis. We first use sensor data to create a kinetic model, which can predict the degree of cure, glass transition temperature, or other relevant thermal/mechanical properties of your material. Once this kinetic model is created, additional data is collected and used to train and test the machine learning model. With that machine learning model, we can accurately predict the process information and dynamically control the process to ensure constant part quality. We can also implement information from accessory sensors, including mold temperature, pressure transducers, ambient temperature, humidity, and more. The machine learning model will determine which information is most relevant to the manufacturing process.
- [animated version of previous figure] Example: The curing (solidification) of an epoxy resin. The in-mold sensor is collecting data (light blue dashed line). This collected data is used in combination with the machine learning model to create a prediction (green dashed line). If the prediction does not match the target value (blue solid line), the process will be dynamically adjusted. In this case, the temperature (pink line) is increased to make sure the target value of degree of cure is reached.
- The Cloud Service stores all of the process data and calculations. From here, you can create customized dashboard to observe process trends by serial number, operator, machine, relative humidity, or whatever is important to you. The data is fully accessible to you and built with OpenAPI so it can be easily transferred to other software.