We recently Googled the manufacturing use case “predictive maintenance” and was astonished by the results there were 82 MILLION results returned. Next, we Googled “process optimization” and it yielded even more results – 302 MILLION. Clearly, these use cases are top of mind in today’s manufacturing landscape, considering digital transformation will deliver $11 Trillion USD in economic value by 2025. Manufacturers are looking for ways to improve processes, operations, and asset maintenance and these use cases are the key to addressing the cost of unplanned downtime estimated at $100k/hour.
As Industry 4.0 accelerates, more and more use cases are being deployed that support the ultimate goal, maximization of overall equipment effectiveness (OEE). Computer vision is fundamentally changing quality and yield optimization, process optimization is changing manufacturing line and process performance and predictive maintenance have greatly impacted equipment and process availability.
What is required to implement these use cases in a real-time and connected factory? There are a couple of must-haves in order to deploy advanced use cases:
- The ability to collect and leverage all your factory data. Traditional OT data from sensors, PLCs, SCADA, MES systems and data historians; IT data from ERP, SCM, warranty, QMS and maintenance systems; and “new” data streams such as clickstreams, social or other customer-facing sources.
- The ability to learn and optimize. Discovering insights from large volumes of data and then optimizing processes via machine learning models.
- The ability to act in real-time. Deploying models at the Edge to minimize latency, allow for autonomous decisions, and conserve IT on-prem or cloud resources for high power app deployment.
But the fundamental question I hear when speaking to plant managers is, “where do I start?” To ensure success, consider these three fundamental and foundational principles:
Number 1 – Build the Industry 4.0 solution on a data platform that considers the entire data lifecycle. Leverage an open-source platform that can ingest, process, and analyze high volumes of real-time data from any source— production sensors, MES, SCADA, historians, databases, distributed assets, or worker wearables. Ensure it can enable predictive analytics or apply machine learning algorithms to petabytes of data while maintaining strict enterprise data security, governance, and audit trails across on-premise and cloud hybrid environments. Ensure it accepts structured and unstructured data sources originating from process sensors, computer vision, robotics, or acoustic sensors. It should also provide instant analytics to drive insights, intelligence, and action from data at the edge, on-premise in the data center, or in any public, private, or hybrid cloud.
Number 2 – Start at the Edge with no-code connectivity. Ensure it delivers the ability to support hundreds of industrial device protocols, out-of-the-box, with no programming and an intuitive interface. Ensure that it can remove the complexity of setup, configuration, and management, and is ready with edge analytics, common KPIs and built-in third-party integrations for rapid-time-to-value. In addition, the technology should be:
Purpose-built for industry: Instead of offering generic IoT solutions, industrial markets need tools that are ready-made solutions working from day one.
- Secure and easy to use. It needs to connect a variety of machines with no coding, and then start collecting data in minutes.
- Integrated. Once the data is collected and normalized it should be easily integrated to any cloud, third-party, or enterprise application.
Number 3 – Ensure your technology partners have a defined POC plan. The operational and intellectual maturity of the IIoT ecosystem including hardware, sensor, edge computing, data lifecycle, and the cloud is now well established. Each deployment is no longer a novel one-off adventure, but rather a proven endeavor delivering operational value for established use cases. Experienced partnerships such as Cloudera and Litmus offer complete connected manufacturing solutions that can be implemented for immediate value without disrupting current processes.
Cloudera has recently implemented a self-service digital customer journey for those interested in a POC of the Cloudera Data Platform, while Litmus has a structured 60-day program that delivers concrete results along the way.
Learn more about our perspective in a panel-style discussion on August 25th at 13:00EST, 10:00PST. Micahel Ger (Cloudera – Managing Director, Manufacturing and Automotive), Dinesh Chandrasekhar (Cloudera, Product Marketing Director) and Bill Kalogiros (Litmus – VP Marketing) will provide insight from their years of experience in this area that will deliver practical value producing results from edge to AI. Come join us!
The post Factory Edge to Cloud Analytics- Three Fundamental Steps to Success appeared first on Cloudera Blog.