When we take a casual walk through the woods like most human beings, we tend to pay attention to the immediate surroundings, such as that rock, stream or tree that we must negotiate around. Hence the expression, “Not seeing the forest for the trees.” The forest, it turns out, is a complex ecosystem of trees, fungus, animals and their droppings, moisture, microclimates, ferns, decaying vegetation and much more. Without this totality of interaction, that tree in your way could not exist.
By the same token, when we use the term “industrial automation,” we tend to think of computer controlled machines and robots connected via HMI and SCADA systems that are merrily turning out products and that can be reconfigured with software to manufacture variations or completely different products. While that concept is not incorrect, it is focusing on trees rather than the greater forest that is growing around us and evolving into the digital factory, and will become a truly integrated technical ecosystem.
There is an evolving entity known as Product Lifecycle Management (PLM) that is increasingly integrating conception and specification with product design and simulation and in turn with the actual design and simulation of the manufacturing process. This, in turn, is being linked with product service, support, documentation and maintenance all the way to product disposal. In addition, it becomes involved with procurement, inventory management, sales, marketing and corporate management.
This explains why the modern automated manufacturing operation will continue to hunger for ever more distributed compute power at all phases of the product life cycle. The digital factory depends not only on automated machines running real-time control software to produce products, but also upon the use and exchange of massive amounts of data.
For example, sophisticated 3D computer-aided industrial design (CAID) software on desktop systems is used to specify and model the higher functions of products to be developed as well as to check conformance with regulatory requirements. That data must be seamlessly ported to more detailed CAD systems for 2D drawing, 3D modeling, mechanical and electrical development, finite element analysis, thermal modeling, validation and optimization and more. Often, manufacturing managers will want to simulate a planned operation on the same systems that are currently running an existing operation. This can be well addressed by taking advantage of the partitioning and virtualization capabilities of multicore processors. But all the design data does not fully address the issue of how the product is actually to be manufactured.
That design data must in turn be used by computer aided manufacturing (CAM) systems to develop things like machining instructions for the products parts, instructions and specifications for casting, molding and the further specifications for the machines that are to assemble the parts. Production planning is then used to optimize the work flow, specify the procurement of materials and perhaps even modify the plant layout for optimal operation. A small but vital part of all this is the machine code for real-time operations of the actual manufacturing machinery.
Such an integrated operation will depend on high-speed networking throughout the plant and beyond, which is why industrial Ethernet and Internet protocols are vital, and why open software and hardware interface specifications are indispensible. For one thing the process is not limited to one physical location but can be geographically distributed and coordinated. For example, if another division of the company produces subsystems for the main product, it could take the needed detailed specifications and deliver finished subsystems.
By the same token—which makes open solutions imperative—a vendor who supplies needed components must be able to receive and execute the specifications of the customer yet be able to adapt local and proprietary facilities to meet the customer’s demands. This all implies a close interplay between design, development, process planning and real-time control, and that means embedded computing power everywhere in the factory environment. It also increasingly makes use of dual, quad and more-core processors. Some machines will need to be able to simultaneously run real-time control software, HMI, networking and data management software and simulation software for processes under development—all of which must not interfere with each other. The appetite for such computing power with minimal consumption of energy as such intelligence proliferates, will continue to drive innovation in terms of processors and the modules that make their integration into the digital factory easier and more flexible.