Optimize Equipment with Data-Driven Analytics
By Jessica Wachtel / Aug 02, 2023 / IIoT
We want machines in good working order, making products of superior quality. This isn’t news. But what is newsworthy is that routine maintenance can still lead to more downtime than necessary. Not all maintenance programs are created equally. Keeping capital equipment running doesn’t exist inside a vacuum of chance. Outside the fraction of unavoidable catastrophes, there’s much power in the decision-making process.
Overall Equipment Effectiveness (OEE) is the measurement, displayed as a numerical percentage value, of how well equipment performs. This score takes multiple factors, like performance speed, availability, the quality of output, and other characteristics into account. A higher score means the equipment is in better working order. The power of OEE is obvious: Hire skilled operators, use equipment as intended, perform diligent quality assurance (QA) programs on products to detect unseen machinery issues, and maintain equipment.
Maintaining equipment now means a few different things. Estimation-based routine scheduled maintenance, once the leading way to keep machinery running, is no longer sufficient. Data-driven predictive maintenance programs optimize capital equipment and lead to a better return on investment. Predictive maintenance uses machine and product data to determine when maintenance is necessary, but before a problem or breakdown occurs.
Shifting from preventive to predictive maintenance
The move from preventive to predictive maintenance has many benefits. AnalyticsPlus is a leading Chicago-based advanced analytics and predictive modeling company with strong domain knowledge in healthcare and other industries. The company built a platform to help organizations with predictive analytics and modeling that relies on InfluxDB. A few examples from AnalyticsPlus demonstrates the value of moving to a predictive model.
This first example focuses on a German welding and riveting shop that produces auto parts. Originally, they performed quality assurance investigations on a random sampling of the auto parts they produced. However, the company wanted to expand their quality assurance program to include all manufactured auto parts. This meant a transition from a manual process to a data-driven process using analytics and models to predict potential failures. The data powering these models and advanced analytics came from the machine sensors. The machine process data, a type of data known as time series data, includes reading for things like amperage, temperature, draw, watts, and pressure.
The shop used the AnalyticsPlus platform and machine data to analyze all their parts, looking for lesser quality ones. When the data revealed these items, factory teams immediately responded with necessary adjustments to the machinery to either get everything back in working order or take other necessary steps. Becoming a data-driven organization sped the QA process up from manually testing one car per day to inline predictive testing of all 20,000 welds for all cars. The data revealed that the shop originally overproduced the welds by 20%, which impacted the company’s bottom line costs, production time, and the vehicle weight.
Machine learning (ML) and artificial intelligence (AI) predictive modeling are part of the arsenal of tools that help an organization become data-driven. These cutting-edge technologies go hand-in-hand with time series data analysis. A US-based engine head manufacturing plant relied on data science and AI/ML during their transition from random sampling QA to their data-driven preventive maintenance program.
Similar to the German shop, this factory used their machine sensors’ process data to power the models and analytics. Rather than using QA to determine issues in the machinery, they used a team of data scientists to determine parts that had a high predictability of failure. The raw data came from five assembly lines and 25 CNC machines. Models aggregated the data to predict failure across all the lines. With the new process in place, the plant increased their QA from manually inspecting 1% of engine parts to automated inline testing 100% of them.
The IoT technology
There’s more than one way to become data-driven. These examples both use AnalyticsPlus’ IoT Edge Computing Platform, which is built on top of InfluxDB. The IoT Edge Computing Platform is an easy-to-use point-and-click edge computing platform that helps factories unlock the power of their machine’s data in real-time. Generating value from this data is the key to transitioning from preventive to predictive maintenance.
InfluxDB is a purpose-built time series database with edge and cloud offerings. This is valuable when it comes to preventive maintenance, and important for the IoT Edge Computing Platform, because factories can keep their data where they need it, on-site, and send aggregated versions to the cloud for longer term and/or centralized analysis. InfluxDB is the foundation for the platform’s real-time analysis and control. The platform uses open-source plugin-based Telegraf as the data collection agent. With over 300 plugins, including MQTT, Modbus, and OPC-UA, Telegraf eliminates compatibility issues between machines and the IoT Edge Computing Platform.
Join the time series data revolution
This is but a scratch on the surface when it comes to the benefits time series data brings to industrial organizations and how it can improve OEE. To learn more about the IoT Edge Computing Platform, read the full case study.