Software-Defined Manufacturing Demands Better Hardware
Industry has run on gears and grease for decades, and the shift toward driving machines by code is not going to change that. Most people assume this kind of software-defined future means hardware becomes a secondary thought—a mere vessel for digital instructions. But software-defined systems will still rely on hardware to connect it to the physical world. They will actually require more hardware, more sensors, and a deeper understanding of physical physics than ever before to be able to encapsulate the processes at work in the virtual environment.
In the latest episode of the Future Ready Podcast from Siemens, Nick Finberg sat down with Mark Hindsbo, the Head of Operations Software at Siemens Digital Industries, to deconstruct this link – more software means more hardware. They dive deep into the intersection of AI, the Digital Twins and the physical shop floor to explore how the next generation of industrial operations will function. If you want to understand how the convergence of the virtual and physical is reshaping the factory of the future, you won’t want to miss this conversation.
Software-defined demands a more granular understanding of hardware to fully implement innovative and sustainable features. A pump, for example, is often highly tuned for flow characteristics with features on the order of nanometers to define fluid flow. Sticking a physical sensor in the fluid path to monitor the system would negate the engineering that went into the component. This is where the X-ray vision afforded by the Digital Twin becomes essential; it allows engineers to simulate internal flows and use that data to strategically place sensors beyond the bounds of the pump, bridging the gap between the microscopic and the macroscopic, the virtual and the physical.
At the same time, the relationship between AI and the Digital Twin is moving away from the idea of a hierarchy and toward a concept of synergy. While AI is an incredible reasoning tool, it can still “hallucinate” or make errors. By testing AI-driven decisions deterministically within the Digital Twin, companies can catch these errors before they result in real-world disasters, such as two robots colliding on a shop floor. It might also kick off simulations to check the viability of system updates on the shop floor – both physical and software.
But by focusing on historical data from a factory, businesses need to understand the impact of survivor’s bias on the decisions. In a real factory, most data represents normal operating conditions because engineers naturally avoid running machines to the point of failure. This leaves a gap in the data for training an AI to handle emergency or abnormal situations. Simulation can help fill this void by providing synthetic data; recreating the rare, dangerous and out-of-spec scenarios that a physical machine could never safely replicate, ensuring that AI is prepared for the unexpected. This idea is already used for some robotics systems to train as well as in the automotive industry to improve autonomous driving capabilities.
The shift toward digital-first also impacts how businesses try new ideas. Prototyping is fundamentally changing. Traditionally, engineers are taught to limit their what if questions because physical prototypes are prohibitively expensive and time-consuming. However, simulation allows for tens of thousands of iterations at a fraction of the cost. This empowers engineers to move from cautious, incremental changes to more radical optimizations that could lead to massive breakthroughs in production efficiency or may have seemed counterintuitive and too expensive to test.
There is a future where factory operators possess superhuman capabilities because of digitalization. Instead of being limited to the mastery of a single machine or process, operators will use digital tools to inspect and understand the cascading effects of entire production lines, factories and even global networks. This connectivity will enable a shift from mass production to extreme mass customization. Manufacturers will be able to transition from producing millions of identical items toward producing much smaller batches tailored to individual needs, such as personalized medicine or consumer products.
To learn more about software-defined systems as well as many other interesting topics, check out our other episodes of The Future-Ready Podcast from Siemens.
