“Predictive maintenance techniques are designed to help determine the condition of in-service equipment in order to estimate when maintenance should be performed. This approach promises cost savings over routine or time-based preventive maintenance, because tasks are performed only when warranted.”
Rafael Gouriveau, Kamal Medjaher and Noureddine Zerhouni
Prognostics and health systems management to predictive maintenance
HIGH PERFORMANCE BRAKE SYSTEMS PRODUCER
Use of Real-Time Data to improve the efficiency on production plants
An Italian company, globally recognition as a best-in-class manufacturer of braking systems and components, with a strong focus on R&D and a keen interest in the potential of innovation and new technologies chose Agile Lab as a partner to adopt a new “Industry 4.0” solution enabling its TPM (Total Productive Maintenance) manufacturing best practices and innovating the Preventive Maintenance strategies.
With a presence in 14 countries worldwide (and a workforce of over 10.000 employees) and the control of the entire production chain, including owning its own foundries, a decisive success factor proved to be the adoption of “EDGE Computing” concepts, computing done at or near the source of the data, running fewer processes in the cloud and moving those processes to local places, in the Plant itself. A solution enabled by Agile Lab’s witboost Data Streams (formerly WASP), a feature-rich, highly scalable framework, ideal for an IoT / IIoT context and resulting in a bidirectional coordination of Machine Learning modelling processes between HQ and the Plants increased both the efficiency of Data Scientists’ activities and the effectiveness of the AI models. The Implementation of data flow analysis directly at the Plant Edge – without streaming high frequency data to central HQ – allowed for a great effectiveness in predicting the correct life time of machinery assets, obtaining more reliable estimates of machinery RUL (Remaining Useful Life) and real time alerting / predictive alerting capabilities.
Our Client is an Italian manufacturer specialized in braking systems and components, with a strong focus on R&D. With an offer of over 1,300 products in the automotive aftermarket worldwide, including calipers, drums, rotors, and brake lines, it is globally recognized for its unmatched design and manufacturing quality of disc brakes, calipers, and complete braking systems. Its products fit a wide variety of applications, such as cars, industrial vehicles, motorbikes, machinery and motorsport racing.
The company’s HQ are in Italy, coordinating its presence in 14 countries worldwide and a workforce of over 10.000 employees Italy, Poland, China, Czech Republic, USA, Mexico, India, Brazil, UK, Spain, Japan and the rest of the world. With the only exception of the US, in all other markets the company controls the entire production chain, including owning its own foundries to produce the required specialized alloys and materials.
Our client enjoys worldwide recognition as a best-in-class manufacturer, a well-deserved accolade originating from having adopted for many years the principles, methodologies and techniques of Total Quality Management, Total Productivity Maintenance, JIT, and Lean Production.
One of the key objectives of TPM is to eliminate / minimize all losses attributable to the manufacturing system in order to improve the overall production, addressing the causes for machinery deterioration and creating the best possible environment for both operators and equipment. TPM’s main objective is to increase the Overall Equipment Effectiveness (OEE) of plant apparatus. OEE is a measure of how effectively manufacturing equipment is utilized (in terms of facilities, time and material) compared to its full potential; by measuring OEE and the underlying losses, important insights can be gained on how to systematically improve the manufacturing process.
Our Client strongly believes in the “Industry 4.0 revolution”, as testified by its investments in cutting-edge, interconnected machinery, both in its Italian plants and foreign facilities, which have further accelerated the pace of innovation – an essential component of the Company’s identity. It also realised very quickly the importance of engaging its People in the development process for the adoption of the new technologies, including in particular some Industry 4.0 “enabling pillars” such as Big Data & Analytics, exploitable to optimize products and production processes, Industrial IoT (multidirectional communication between manufacturing processes and products), Advanced Manufacturing Solutions, and Manufacturing Plant Simulation.
The challenge for our Client was to identify and adopt a new “Industry 4.0” solution enabling its TPM (Total Productive Maintenance) manufacturing best practices and innovating the Preventive Maintenance strategies. One of the main goals was to exploit the new value potential of Real Time Data gathered “on the Edge” from specific Plant / Machinery through SCADA and multiple Industrial IoT (IIoT) protocols, together with AI / Machine Learning, in order to:
- support a continuous improvement of production processes, gathering, measuring and processing Real-Time Data directly at Plant level though Edge Analytics;
- gather the machinery data feed from SCADA and all further Industrial IoT subnets in the Plant;
- create a real-time Dashboard for TPM Analysis at Plant / Line level of OEE (Overall Effectiveness) and OPE (Overall Plant Effectiveness) KPIs, for real-time Drill-Down of Performance, Availability and Quality OEE / OPE component factors, and for further Root-Cause Analysis on the underlying Loss drivers;
- enable the real-time analysis of operational activity-level indicators such as TBF (Time Between failure), TTR (Time To Repair) related to breakdowns at plant / component level, and a deeper Event Analysis of each EWO (Emergency Work Order) to discover the Time Stratification of the single micro-activity timeframe composing the EWO / TTR Events
- extend the plant data domain to include human labor variables, as well (e.g. shifts, presences, regulated pauses, and so on) and not only machinery / asset variables;
- exploit AI / Machine Learning in order to discover the OEE / OPE Loss factors and the correlation with the operational activity-level data, in order to identify and eliminate the causes of the losses
- exploit AI / Machine Learning and Edge Analytics together in order to innovate the Preventive Maintenance of Plant Assets (e.g. through predictive measures like RUL – Remaining Useful Life) and the Predictive Alerting in order to optimize the related Machinery Supply Chain processes;
- set-up and coordinate a collaborative / bidirectional model between the single Plants and the HQ, in order to centrally manage the Business Logic and the Machine Learning models (in terms of production, tuning and deployment) to be deployed at Plant level to homogeneously enable the new Edge Analytics.
The Solution provided by Agile Lab is based on the adoption of its witboost Data Streams (formerly WASP) platform in compliance to the “EDGE Computing” concepts, in particular:
- Edge Computing is computing done at or near the source of the data, instead of only relying on the Cloud.
- In simple terms, Edge Computing means running fewer processes in the cloud and moving those processes to local places, such as on a user’s workstation, an IoT device, or a server on the Edge of the system, in the Plant itself.
witboost Data Streams is a feature-rich, highly scalable framework, ideal for an IoT / IIoT context: it performs batch processing with ease, but its forte is real time streaming data analysis at scale.
In order to enable all the Analytical and Dashboard capabilities directly on the Plant Edge, the data from SCADA and the further IIoT components are managed directly through the witboost Data Streams (formerly WASP) IoT Gateway. This allows for the adoption of a single and secure IoT Gateway, able to interface with both IoT devices and SoT (Subnet of Things) gateways, typically governed within an internal LAN without device management.
The following images illustrate the decoupling between the Plants’ Edge and the Central HQ: the Machine Learning Models and the Business Logic are produced at HQ and deployed to the Plant Edge level, where the Analytical capabilities are directly used in processing the MQTT data on the Edge, in order to maximize both the efficiency and the effectiveness of the solution at the same time. All Big Data flows are, however, gathered in a HQ Data Lake for further centralized analysis and tuning of the ML models.
Achieved results and benefits
The bidirectional coordination of Machine Learning modeling processes between HQ and the Plants increased both the efficiency of Data Scientists’ activities and the effectiveness of the AI models.
The Implementation of data flow analysis directly at the Plant Edge – without streaming high frequency data to central HQ – allowed for a great effectiveness in predicting the correct life time of machinery assets, obtaining more reliable estimates of machinery RUL (Remaining Useful Life) and real time alerting / predictive alerting capabilities, thus enabling the optimization of Supply Chain processes and related savings.
The TPM Analytical & Dashboard capabilities enabled the monitoring and improvement of related OEE / OPE KPIs:
- maximizing equipment effectiveness through optimization of manufacturing equipment;
- availability, performance, efficiency and product quality;
- establishing a preventive maintenance strategy for the entire life cycle of equipment;
- covering all involved departments (such as Planning, Maintenance, and so on) and involving all stakeholders from management to Plant-floor workers;
- promoting improved maintenance through Plant-level autonomous activities;
- increasing Employees’ performance – having taken in scope the human labor measures.
Clients (i.e. a maximum of 15 minutes of latency between data generation and its reception on the Client’s Cloud, including cross-time on both mobile networks and the Internet) and it has become the solid foundation for developing services that the company wouldn’t otherwise have been able to offer.