Real-time analysis of factory equipment data! Predictive maintenance solution.

basic information

- Capturing data that cannot be measured in reality - Virtual sensing using CAE - Analysis tools that do not require data scientists and can be analyzed by the field - Real-time visualization of operational rates and production status of each site on the dashboard

Applications/Examples of results

1) Abnormality precursor detection of large-scale plant belt conveyors: Achieving continuous monitoring through sensor data collection. 2) Visualization of component fatigue inside the boiler of a power generation plant: Lifespan prediction through stress simulation.

Detailed information

• 

  1) Abnormality Prediction Detection for Large-scale Plant Belt Conveyors: Achieving Continuous Monitoring through Sensor Data Collection Challenge: The belt conveyors in large-scale plants, spanning several dozen kilometers, have numerous components that need inspection, many of which are in hazardous locations. Additionally, since this equipment is responsible for transporting raw materials, any unexpected stoppage poses a risk of halting the entire operation. Solution: By utilizing vibration sensors and the IoT platform "ThingWorx," we achieved continuous monitoring of the condition of the extensive belt conveyors. We established a system to evaluate the collected data using Cybernet's analytical model, enabling early detection of abnormal signs. Effect: We reduced access to hazardous areas, significantly improving the safety of inspection work. This minimized the risk of unexpected stoppages and achieved a balance between stable operations and efficient maintenance planning.

• 

  2) Visualization of Component Fatigue Levels Inside Boilers of Power Generation Plants: Life Prediction through Stress Simulation Challenge: The thermal piping in power generation plants is composed of nonlinear materials, making it difficult to directly measure stress and fatigue conditions with actual sensors. Additionally, increasing the number of heat cycle tests is not practical, and there has been a lack of means to quantitatively grasp component reliability. Solution: A 3D-FEM model was reduced to a Reduced Order Model (ROM), and real-time CAE was used to output stress and creep strain amounts at arbitrary locations as virtual sensors. This was combined with 1D-CAE simulation to establish a system that can numerically predict accumulated fatigue levels and remaining life from stress history data. Effect: Stress conditions and life predictions, which were difficult to obtain with actual sensors, were visualized in real-time as virtual sensor outputs. Component reliability was quantified based on physical phenomena, achieving optimization of safety factors and efficiency in maintenance planning.