Event-Driven Manufacturing: A New Architecture for Connected and Flexible Manufacturing
This is the second post in a series about how new software technologies are accelerating how we build physical products. The first post was about APIs.
Manufacturers have recently accelerated a trend that has been going on for many years: digitizing manufacturing systems and processes for more reliable production and quality. Even though there are still paper-based operations out there, manufacturers are opening up to the opportunity that digitalization can bring. And some are becoming increasingly aware of the existential threat to businesses if they choose to do nothing.
The manufacturing software systems that are generally in place today are what I consider transactional systems. As companies move faster to meet customer demand, deal with increased competition, and care more about the lifecycle of their product beyond initial production, our industry needs to consider a new type of system architecture. I refer to this new architecture event-based manufacturing, and it is significantly more flexible and adaptable to what manufacturing companies need to meet this century’s new goals.
The systems that currently power manufacturing digitalization can generally be summed up by a few acronyms, including ERP, QMS, MES, SCADA, and so on. With a few exceptions, most of these systems are transactional, and companies have adapted their business systems around functions like Sales, Finance, Accounting. Unfortunately, the technical Engineering and Manufacturing get overlooked in these systems because their processes are too specific for a generic, large system like ERP to manage. For example, it is unlikely that Design Engineer interfaces with an ERP system very much, if at all. For the most part, these systems are valuable in that they keep track of a lot of information for a large organization. In other words, these systems are Systems of Record. It is usually up to individuals or managers to update the information in these systems so that everyone at their company is up to date. Once the information is in, the system has served its purpose: The record is stored and is accessible for compliance purposes and in case it is needed for future troubleshooting. With more advanced operations, inputting information may automate other calculations and processes within the large system of record. For example, if a company is out of a certain part, an ERP system may automatically open a Purchase Order to restock that part. These types of automation are typically custom-built and only work within the runtime of the transactional system. In other words, companies that want these automations typically hire external consultants to customize their ERP implementation to support what they need.
These transactional workflows keep track of information so that large organizations can operate while having a level of control and information passing that keeps the organization on the same page. They require individual users to input the information rather than collecting that information automatically. The interactions with these systems are shown in the following diagram.
As companies move faster to adapt to shifting customer demand and increased competition, traditional transactional systems and workflows break down. New information is constantly changing previous decisions, connected sensors (IIoT) and automated workflows are generating a vast stream of continuous data, and design and production are becoming increasingly intertwined as hardware companies’ value propositions become increasingly based on innovation rather than just output. The transactional upkeep of this information gets completely out of hand, so individuals and teams take matters into their own hands with their own sub-processes that are more accurate. This has the cost of being disconnected from the rest of the operation.
Technologies and patterns like service-oriented architecture (SOA) and publish/subscribe message queues are increasingly proving themselves capable of handling mission-critical workflows, and they are a great fit for the dynamic and distributed nature of manufacturing. Rather than using a central system to manage the latest source of truth manually, event-driven manufacturing allows different individuals, teams, and departments to set up automated workflows based on new incoming information. In addition, event-driven manufacturing allows people and machines to share new information with the broader organization. They can do this as soon as the information is available rather than waiting for the nightly sync or for the appointed configuration management stakeholders to update the system.
Event-driven manufacturing is necessary for manufacturing companies in the 21st century because fast information is not only useful to deliver faster but because quick decision-making based on new information is critical to avoiding expensive and dangerous recalls. New information can halt faulty product in its tracks and save significant dollars that are spent reworking parts today.
With event-driven manufacturing, rules replace routines.
With event-driven manufacturing, rules replace routines. What once was a daily check to make sure everything is going okay, people can set up automated rules that watch for certain events and trigger subsequent actions. If you have used IFTTT, then you’re familiar with setting up automations like “If I receive an email, blink my lights blue”. While that’s only somewhat useful, the possibilities are huge in manufacturing. For example, engineers can set up rules such as “If we have less 50 cubic meters of insulation, create a purchase order to buy 500 cubic meters more.” As I mentioned above, this would otherwise have been a custom-built thing built right into the ERP system, but with event-based manufacturing, this is democratized. Rules can be updated regularly to adapt to your company’s needs rather than waiting for a software project scope to kickoff, requirements collected, developed, and implemented 6 months later. For those more technical professions left out of the large systems, they are no longer left out and can set up some pretty amazing rules, such as “When we get the latest flight test data, compare the maximum load against all of the flight computers we have tested and have in inventory. Alert me if there are any that do not meet the requirements.”
Event-based manufacturing not only makes manufacturing operations more flexible and efficient – it also makes them safer. By replacing routines with rules, event-based manufacturing helps individuals develop critical logic and only be alerted in critical circumstances. Humans are known to make big mistakes when routine and muscle memory precedes their judgment of a situation. There are plenty of unfortunate examples in history of power plant accidents, cybersecurity threats, and of course, manufacturing defects caused by people repeatedly performing a task and misjudging when they see something that is actually outside of normal conditions.
Event-based manufacturing not only makes manufacturing operations more flexible and efficient – it also makes them safer.
Machines, IIoT, and how to enable event-driven manufacturing
As more machines and sensors become connected in manufacturing operations, there is an opportunity to integrate those systems more closely to the business priorities of the company. It’s not enough for a sensor to have an internet connection. With event-based manufacturing, a temperature sensor on a composites oven could signal overheating to the materials department. When those temperatures are run against their rules, they can make sure that composite parts overexposed to thermal treatment are contained before they are installed on the vehicle. With transactional manufacturing, the best-case scenario would be if a quality inspector analyzed the data or found the defect after the part had already been installed. Event-based manufacturing allows technical groups to subscribe to the conditions in production and save the entire company time, money, and risk by halting the faulty part before it’s installed. Event-based manufacturing is the paradigm shift needed to accelerate machines and sensors to really get connected into manufacturing operations. By putting decision-making into the system or even down at the machine-level, engineers are freer to innovate and research the product design rather than spending time administering the system that is supposed to help them.
If you look at the software landscape, there are many technologies to choose from to enable real-time, event-driven workflow. We are far from having clear winners in manufacturing. That said, there are technologies, patterns, and protocols that we are looking at First Resonance to help our customers set up their event-driven manufacturing processes. For example, Apache Kafka is a unified, high-throughput, low-latency platform for handling real-time data feeds. It is a highly reliable system for managing multiple data streams coming from producers and sending the correct information to consumers that can use that information to make automated decisions or alerts. Similarly, GraphQL subscriptions provide real-time visual updates across the factory with real-time information. This scales from mobile notifications to factory heads-up displays, and more.
While digital transformation and Industry 4.0 in the COVID era are accelerating manufacturers to bring in systems to deliver new, customized, and sustainable products, many manufacturers are left wondering how to bring in new systems without the overhead of adapting to a large, monolithic, and opinionated system that might slow down their people. Event-driven manufacturing is the architecture for modern manufacturers to iteratively build a dynamic system and support an innovative workforce that is able to adapt to new information. The right type of event-based manufacturing system will bring the security, control, and centralized management that traditional systems do while empowering an increasingly technical and capable workforce to make faster and more informed decisions. At First Resonance, we believe that while exciting new technologies like IIoT, AI/ML, and robotics are critical pieces of the Industry 4.0 movement, the real drivers of the fourth industrial revolution are going to be those companies that democratize information access and change the way people work. Rather than transacting routine information to centralized databases, this new workforce will set up dynamic and automated decision-making that results in greater productivity, reliability, and ingenuity in the products we will see in the coming years.
In the next post, I will talk about how event-based manufacturing powers event-based analytics, a new type of data analysis that provides powerful insights to innovative companies and allows AI/ML to be deployed into manufacturing operations reliably.
If you are interested in seeing this topic in application, check out what we are building at First Resonance.