Here we give a summary of why it is important for manufacturers to adopt Industry 4.0. We base much of this on an interview two German executives gave to McKinsey.
The manufacturing executives were quite blunt in their assessment. They said that because of outdated thinking and a lack of knowledge that too many companies have not yet tied together physical and information flows in the factory and supply chain. Doing that is a key provision of Industry 4.0. So these companies are missing out on the opportunity to run their operations at least in the optimal manner.
There are various problems. First, information systems are not connected to physical inventory and machines. This is in spite of IoT making that easy and inexpensive to do. Second, in many cases companies have taken steps to gather data from IoT but they are not using that properly to streamline operations. The reasons for that is they do not have IT people who understand machine learning and other algorithms. And they do not have the IT architecture needed to support big data.
A Blunt Assessment of What is Wrong
People are skeptical when presented with a new, supposedly game-changing, concept. So we are not going to give you a lengthy definition of what Industry 4.0 means (at the end of this article we provide a short one.) Instead, we are going to jump right in and tell you what Siegfried Dais, Deputy Chairman of the Board of Management at German engineering company Bosch GmbH, and Heinz Derenbach, CEO of Bosch Software, said. They are quite clear in laying out what needs to be done to tie together the physical and virtual worlds to drive efficiencies in manufacturing under Industry 4.0.
In their view of the cyber-physical view of the factory and supply chain, every component in the supply chain and factory is attached to the network. Each transmits its status to the big data cloud, analytics algorithms, and manufacturing planning and operations systems. They call this "process2device." The executives says, “That means a physical device becomes an active part of a business process: delivering data, sending events, and processing rules.”
Not only does each component report its location, “Everything is interlinked with everything else. “ So each component in the subassembly knows to which customer order it belongs. Thus, the component reports variations as it moves along the supply chain and across assembly line and the factory planning and operation models change in real time. This eliminates the lag between planning cycles and actual events present in the traditional approach and the disconnected nature of that.
What makes all of this possible today is generally what we call IoT, big data, and analytics.
The earliest adopters of these ideas in the factory have been preventive maintenance. That is because it is the easiest to apply and easiest to understand there. For example, a machine that is overheating and vibrating because it needs a new filter can create a work order in the PM system automatically. But integrating such information into the logistic systems is far more difficult.
Siegfried Dais makes this criticism, “For example, take cyber-physical systems, which can tell us where every single unit is at any given time. Logistics players often use this tool, but with an old mind-set that fails to exploit the advancements the tool was designed to offer. So the first requirement is that logistics players truly use what's new.”
And Heinz Derenbach points out what IT architects and programmers would call the need for abstraction: “In the connected world, we cannot separate the physical world from business processes.”
He adds, “It is essential to translate the physical world into a format that can be handled by IT.”
A Heavy Lift for IT
Doing all of this requires more than just attaching a wireless card or chip and sensors to machines and components. Companies need programmers, data scientists, and big data architects to work with factory planners and product managers to plug that data into planning and operating models. And those models have to change to accommodate real-time information.
This is going to be a problem for many, if not most, IT shops because of a lack of skills. It’s often overlooked by big data proponents that is takes some understand of mathematics and statistics to be able write and use analytics algorithms. The average computer programmer does not understand that. So a company needs to add data scientists to the IT team. Algorithms coded into modern programming frameworks make much of that easier, as do some cloud platforms, like IBM Watson and Databricks, that can draw conclusions without programming in many cases.
What Industry 4.0 Means
Now, as we promised, here we give a short definition of Industry 4.0. There are several differing ones. Basically this is the idea that combination of IoT, big data, and analytics are the 4th industrial revolution. McKinsey and others also add virtual reality and 3D printing. Proponents say these tools will overhaul manufacturing in the same way that the invention of the steam engine, the invention of the assembly line, and earlier computing changed manufacturing.
The reason this is a genuine game changer is that advances in software and networking have made it inexpensive and not complicated to tie together virtual and physical worlds. As we mentions above, that is called abstracting the physical environment into the virtual one.
Here are the key pieces.
IoT lets companies connect components, machines, and even people to the cloud via low-cost sensors and wired and wireless networks. And the big IT companies Yahoo, LinkedIn, Google, Twitter, and Facebook have written and given away big data software, like Spark and Hadoop, which has turned convention data structures upside down. These free-form, distributed databases lets companies easily process data streaming from machines and software applications. A manufacturer using an old-fashioned data warehouse cannot do that.
Finally, data science has pushed the esoteric ideas of academics and operations researchers into solving real-world business problems. This is because their difficult algorithms have been coded into easy-to-use APIs. Distributed systems like Spark even let machine learning run across a distributed architecture that can scale without limit. All of this means that data coming from radically different sources, and in differing formats, can be understood and turned into actionable items.
So the framework is in place. What is needed now is education, studying what other manufacturers have done, and then applying those practices to daily operations and planning. The end result will be leaner manufacturing with higher profits, lower costs, and more satisfied customers.