The Event Mesh: Powering Industry 5.0's Real-Time Revolution
From 3D printing to AI, manufacturing processes are in the crosshairs of change. And now, for the first time, the technology is there to give manufacturers the power to view their processes not from the part to the product with all the intervening disconnects, but from the product to the part, in real time. Let's explore how this seismic shift is getting manufacturers, their workers, and their processes operating harmoniously.
There's been a dramatic evolution in manufacturing data exchange and it's coming close to realizing its full and transformative potential. Rewind to Industry 3.0 where it all started, with factories beginning to introduce connectivity, with centralized ERP monoliths connecting various aspects of the system. Manufacturing was taking the very first simple steps in understanding and integrating data movement.
Once this digitization got started, along came Industry 4.0 where the centralized ERP system evolved to become more modular in nature. This was a turning point in the journey to understanding data in movement, where the worlds of Information Technology (IT) and Operational Technology (OT) started to integrate with each other and exchange information—such as Manufacturing Execution Systems (MES), SCADA systems, and historical fault detection systems on the plant side.
The Human Resurgence in Manufacturing
From this Industry 4.0 foundation, we are now moving towards Industry 5.0. A European Commission Policy Brief defines the target of Industry 5.0 is to "aim beyond efficiency and productivity as the sole goals, and reinforce the role and contribution of industry to society." It should complement the existing Industry 4.0 approach by specifically putting research and innovation at the center of the transition to a sustainable, human-centric, and resilient manufacturing industry.
North American organizations too see human workers at the core of Industry 5.0 with company growth spearheaded through a combination of innovation and human capital. Alongside this, there are also pressing opportunities to capitalize on artificial intelligence (AI) all while keeping tabs on human centricity, sustainability, and resilience. To enable this requires a technical balance to ensure humans and machines work together—each learning from the other.
Achieving this harmony begins at the data level. Industry 5.0 triggers further convergence of IT and OT—aggregating shop floor data in real-time to help bridge the gap, continuously collecting and analyzing data, and offering a comprehensive view of manufacturing operations. This data can then be delivered to enable key stakeholders, especially workers on the factory frontline, to make critical decisions based on accurate, real-time data.
A New Approach to Data Exchange Required – A Different Set of Enablers
Imagine a world where business operations are no longer constrained by delays or bottlenecks. This applies not only on the factory floor, but across the rest of the enterprise, and each aspect of the business. What if you could connect everything, even if you were using multiple systems to manage manufacturing in the front and back office?
This could be systems in the OT layer with systems in the IT layer in a secure, guaranteed, throttled, multisite and cloud, and most importantly, real-time streaming manner. Specifically, operational systems such as MES, Programmable Logic Controllers (PLCs), SCADAs, OPC Unified Architecture (OPC UA) layers bi-sectionally connect with master data, Product Lifecycle Management (PLM) systems, Customer Relationship Management systems (CRMs), and incident management systems from SAP, Salesforce, ServiceNow, Microsoft, Oracle, and many others.
Any of these systems can be connected easily with each other across a wide variety of factory, inventory, and logistics use cases. To enable this, integration needs to reinvent itself, especially when it must bring in other technologies that are becoming prevalent in factory operations, such as AI. For this Industry 5.0 and AI-driven world to happen, we need to think differently.
The beginning of a manufacturing event mesh—find that single pane of glass!
Loosen up
This requires thinking about integration in a much more modern way. We need to stop using a traditional batch historian and update or an Enterprise Service Bus (ESB) as an architecture, where a centralized software component performs integrations between applications. We need to evolve to a much more loosely coupled event-driven approach to moving data. This is the foundation of Industry 5.0 as far as its enablers are concerned.
As You Sow, So Shall You Reap
One application publishes messages or events and other applications receive it. Newer applications or devices just come in—as they start to listen to data, new data starts to flow toward them. Adding more and more applications becomes super seamless.
Data Flows
This allows manufacturers to extend this data movement between devices and sensors, so all kinds of sensors can be publishing information or subscribing to information.
Take, for example, a production line going down. This is an event on the OT side of the house. Which customers and what elements of the supply chain are affected can only be determined by correlating this OT event with all those other IT systems—ERP, CRM, transport management, logistics, etc. Imagine streaming the event "production line is down" in real time to the IT systems in a one-to-many manner and the visibility and real-time action this brings.
An Event Mesh Is Born
An event mesh is the data fabric that connects systems—IT, OT, AI—effectively as a universal connectivity fabric. It enables us to have many devices integrated together, connecting plants and logistics and then being able to govern all of this, understanding all events in real time.
In real time is key because events are instantaneous "occurrences" whether on the IT side or OT side. Using an event mesh, we can look astutely at any event and find that they can be depicted as objects + actions + properties using "topic routing." Take the example of "production line" (object) + "failed" (action) + "in region X, at plant Y, line Z" (properties).
An event mesh routes these occurrences in real time, guaranteed to all applications and systems that are interested and "tuned" in. Applications are natively connected to the event mesh, such as SAP and OPC UA using MQTT, or they connect via micro-integrations, which are lightweight connectors to allow legacy applications to connect to the mesh to produce or consume events.
AI and the Metaverse on the Future Factory Floor
The industrial metaverse, as highlighted by EY, has the potential to revolutionize production floor processes and management optimization. It's become possible through data-driven digital twins, model simulations, and data analytics—paving a path that will transform how work is performed. Nvidia Omniverse is an example of this.
But the metaverse needs to connect to the physical universe in real time. It's cost and energy prohibitive to keep training AI models in real time. So how do we integrate AI with real-time data? Agentic AI and Retrieval Augmented Generation (RAG) are being used to integrate Generative AI capabilities with real-time data. This real-time data essentially acts as event streams on the event mesh, flowing into Agents or RAG vector databases, keeping them up to date with real-time context.
Event-driven integration can set up manufacturers to adapt to all these new changes on the factory floor, better capitalize on new technologies such as AI models, or connect to growing industrial metaverse applications in manufacturing. The enabler for AI is an event-driven integration mechanism, providing real-time context to AI in autonomous systems, through something called a "context mesh."
A context mesh brings together the entire context of an enterprise, which can serve as the bedrock of sophisticated AI-based applications—from intelligent assistants using GenAI models and complex analytical tools, to machine language-based recommendation models and purpose-built AI applications using deep or reinforcement learning. This becomes the fabric that pushes that real-time context onto these digital twins.
Industry 5.0 Manufacturing Real-Time Use Cases
Take the example of one leading multinational engineering and technology company that uses event-driven integration to connect its global manufacturing facilities. Specifically, the manufacturer has deployed event brokers in each of its 160 plants, and also in its IT layers on-premise and in the cloud, forming a large interconnected event mesh.
Master Data Management at Scale
With an event mesh, the company has been able to streamline the distribution of huge volumes of product master data across its enterprise.
For instance, its global operations produce 7,000 parts per minute. To achieve this, the company needed to send the production master data, such as a master Bill of Material, to multiple production facilities. This involved sending 6,000,000 messages, equivalent to three terabytes of information per day, to different plants worldwide.
As these products are rolled out to be manufactured at more plants or master data changes, the data requirements become more vast and more complex, creating a huge end-to-end scenario.
But with an event mesh, the manufacturer has been able to confidently share production master data to all of its required environments in real time. This ability to connect factories around the world is underpinned by a stable asynchronous connection between core data centers around the world and their different plants.
AI Deployment to Minimize Errors
This company has also been able to deploy AI across production lines to concentrate error levels to lower than one part per 10 million produced parts. The company is now looking to speed up production lines with these different models and reduce the error rate.
By relying on an event-driven integration backbone to underpin the data requirements of new AI models, the company aims to further increase its success rate within production lines. In fact, the manufacturer estimates it can scale to minimize errors to one part per 100 million produced parts in the future.
Manufacturing Integration Requires a Rethink
Traditional integration methods are no longer sufficient in a manufacturing sector moving towards Industry 5.0. The foundation of Industry 5.0 demands seamless human-machine collaboration, facilitated by real-time data exchange.
Event-driven integration is a key enabler to unlock the full potential of real-time data in manufacturing on a global scale.
The technology is here and the humans are back!
