By Michael Ford, Aegis Software
Two words, with similar, yet distinct meanings. Putting aside the fact that we can’t always agree on how to spell them (“esses and zees”), the use of these terms and others, such as “Digital Twin”, “Digital Shadow”, “Digital Thread” etc. are all designed to inspire the application of digital technologies as part of manufacturing, but are so often used out of context or beyond their intended scope. Rather than argue the semantics, let’s focus on the important digitalization issues that affect manufacturing.
The Real Digital Divide
Today, we see digitalization everywhere through the manufacturing process, though at times it may seem like there is a conspiracy to hide it. Take the design of circuit board layouts. Digitally created layouts are still delivered to the majority of manufacturing in crude “1980s” data formats. With a more limited success than would be expected, at least we now have IPC-2581 as a more intelligent way to transfer design data in a single intelligent format such that New Product Introduction (NPI) systems in manufacturing can more easily make sense of how to best select, configure and prepare manufacturing processes used to create the product. Raw material packages are surely also designed digitally, yet we continue to see most manufacturing companies using calipers and optical systems to measure components on the shop-floor. Our approach to digitalization has been locked away in the stone-age, with seemingly great potential benefits, but in reality huge gaps in the flow. For both commercial reasons as well as technical, this has been allowed to persist. Restricting access to data has been suppressing large-scale rollout of “digital factory” innovation for a generation. The perceived cost-value proposition in retaining “control” or “IP protection” through the restriction of shared data is now at last being overcome by the cost-value proposition of the use of such data in a truly singular digital manufacturing environment, such as that suggested by Industry 4.0. This is perhaps the most significant step forward that we are, as a manufacturing society, to take. Forget the technical details, the fundamental will and business case here are always the real drivers. At the recent Productronica show in Munich, discussion was firmly turned in the direction of how companies can work together with data to reduce cost and create more value for customers rather than ring-fencing their domains against intrusion. Let’s hope the balance has tipped at last.
The Digital Story Going Forward
Ironically, this is a story of “human versus machines”, terminating in a manufacturing environment in the future that will look quite alien to what we see today. Consider the creation life-cycle of a popular electronic product, such as a mobile phone or tablet. There are two aspects of product design that must work together; the electrical, to make sure that the product works as intended, and the mechanical, to make sure that it all fits together and works reliably in the intended casing. All of this design is of course done digitally, from which there are two roads to go down. It is easily possible to feed all of the data in to a 3D gaming engine that allows people to have a “virtual reality” look around the product, inside and out to see whether there are any potential issues in the design or manufacturability. The digital world is therefore still dependent on people. The other road is to do the same, but using computerizations, which are algorithms up to and including artificial intelligence (AI). Today, the state-of-the-art in manufacturing software is a rather clever set of algorithms working together that have been evolving for many years.
Computerizations calculate effects of any potential issues, considering billions of inter-relating factors, not dependent on human interpretation. Actually, this is not quite true, as the digital model can only be as good as the rules that it was given as part of the algorithm. Rules are set up which govern a pass, fail, or “fuzzy warning” for each of the design rule constraints. As algorithms progress, rules can be altered by software to achieve a better result. The critical issue is the measurement by the algorithm of what is “better”. This in turn is defined by a set of parameters, again decided by people. We are not yet then truly in the digital world. As software algorithms progress, even to the levels of true AI, the goal will always have been defined. Given that we find that almost every person in existence has a different set of personal goals, so too are we likely to find that AIs will also have different parameters that define “success”, which will become a key factor when comparing “smart” software systems. There is a forever lingering dependency on human vision, experience and innovation, perhaps this is the reason why the best robot-based science fiction plays on the interaction of a person and an android, or that the world is destroyed by some misguided interpretation of what is best for mankind. Going forward then, the details of production technology are increasingly built into the machines and associated software. We are losing manufacturing engineering skills from our work-force, but many of these people are the designers of the machines, processes and software that now incorporate that knowledge. Electronic designers once designed using transistors one by one. Now, standard circuits in the form of integrated devices with billions of transistors are “dropped into” the design of any popular device today. This is the way of evolution of most technologies, we continually create generations of building blocks which become the materials of the next generation of building blocks. Software technology and Smart algorithms are no different. Factories of the future will be run by engineers and managers who care little for industrial engineering practices of today, rather they will be focused on the selection and fine-tuning of manufacturing technology in use.
Perceptions Of Digitalization & Computerization
The way towards our digital factories of the future continues to be aggravated by terminology. I guess I am as guilty as the next marketing person, trying to find the next magic word or phrase that completely describes evolving digital solutions. With similar terms being “invented” with slightly different original definitions, it has created a complex environment in which people gain differing perceptions and end up not believing anything. The phrase “digital twin” can be interpreted as being an exact digitized replica of a physical product. In fact, it is most likely that the digital form came first, as part of the design process, visible to humans through a cool 3D virtual reality environment. Software algorithms designed to create value from the information don’t need the flashy graphics, they simply get on with it, following pre-defined rules. There is then the term “digital shadow” which could mean the digital record of all of the events and transitions that occur during the various production processes, from raw materials to production completion and beyond. Both the terms “digital twin” and “digital shadow” are used interchangeably quite frequently. There is then the “digital thread”, which I guess is the combination of the “digital twin” and “digital shadow”, which for me includes the connection of the engineering and manufacturing data. Others may see it differently. We then get to the word “computerization”, as mentioned many times in relation to Industry 4.0. The word “computerization” for me implies the automation of decision-making, which goes beyond “simple” digitalization. For others, I expect the meaning of these words is inter-changed, with computerization being a lower level technology as compared to digitalization. It is no wonder there are so many questions.
The Three Laws Of Digitalization
However we like to talk about digitalization and computerization, the more important aspect is to ensure that it works for us, to deliver expected values and benefits. There are some fundamental “laws” about data, on which any digitalization is based, that we need to remember:
Data must be complete and accurate, otherwise incorrect action could be taken as a result. “Garbage in, garbage out” will always be true.
Data must be accessible. The potential of multiple benefits derived from multiple uses of each piece of data far outstrips any legacy benefit of trying to conceal information. Clear standard mechanisms over which data can be transferred and shared are essential
Data must be understandable. The language and content definition of data needs to be such that the interpreted meaning of the data is clear beyond the scope of the process that created the data.
Implementing these three “laws of digitalization” enables the creation of a digital platform on which the most advanced software algorithms and even those based on AI can exist. It is time to prepare for digitalization, no matter what we end up calling it. We have to select machines and software that we can trust that include the finer details of industrial engineering, ensuring that the most high-performance factories can be created and sustained.
If the science fiction writer Isaac Asimov were a scientist today, on the cusp of achieving his envisioned “positronic” artificial intelligence brains, he would certainly incorporate his famous three laws of robotics, and in so doing save the world. For the many companies working together with IPC to create the Connected Factory Exchange (CFX) industrial IoT communication standard, within the scope of manufacturing at least, there is a similar feeling. CFX is unique in that is it precisely targeted to provide the foundation of digitalization for manufacturing based on these three laws. CFX should be a part of all Smart factory and Industry 4.0 implementations going forward, in order to ensure the lowest cost of ownership, maximum value and sustainability. The future is here sooner than we think.