material gains
Smart Manufacturing is Coming, but So Much More Could be Possible
We can make more (money) by making less (product).

Arthur C. Clarke once said, “Before you become too entranced with gorgeous gadgets and mesmerizing video displays, let me remind you that information is not knowledge, knowledge is not wisdom, and wisdom is not foresight. Each grows out of the other, and we need them all.”

Today, we’re all familiar with gorgeous gadgets, and not only those we carry in our pockets, wear on our wrists or help us drive our cars. The factories we work in are dripping with sensors and automation, which is increasingly robotized, bringing a level of dexterity, efficiency, and reprogrammable flexibility that previous generations could only dream of.

We are fortunate to live in this period we now call the fourth industrial revolution, although we should recognize our predecessors have been working toward this for generations. It’s simply human nature. Since the beginning of industrialization, people have been making analyses – of processes, end-products, and how things are done – to achieve some improvement. Often, the goal is to increase productivity and quality but also to ensure safety and reduce environmental impacts. Recently, of course, reducing pollution and energy consumption, while addressing issues like recyclability, has become increasingly important.

In the past, the sensors have been human eyes, sometimes hands or ears, and the storage has been a clipboard and paper notes or a report. The desire for smarter industry and smarter factories is not new.

Today, we are better equipped to realize it than at any time in history. Enabling all this is the incredibly diverse array of electronic sensors now available, which are affordable, known to be accurate, and provide digital outputs that are easy to record and store. We also benefit from low-cost mass storage that allows us to retain, organize and quickly access the data collected. And, of course, we have powerful computers to manipulate and analyze the vast repositories of data.

With all this at our disposal, how can we fail? Let’s not forget the second part of Mr. Clarke’s observation. Our powerful tools have allowed us to turn traditional data into Big Data. But this remains only the first step on the path to knowledge. And then we must turn that knowledge into wisdom and foresight – the ability to make better plans based on accurate predictions. For this, we need new skills like data engineering to be sure we capture the right data and prepare it correctly for analysis, followed by the data science and analytics needed to generate actionable insights.

These, the critical skills of the fourth industrial revolution, will take us toward truly smart manufacturing. As human workers are replaced by flexible, intelligent automation – robots are intrinsic to smart manufacturing – our future roles lie in learning how to handle the data and developing the AI algorithms that will enable smart management of equipment and processes.

We already know process capabilities must improve significantly to reach our future goals. The roadmap to advanced 5G services provides an excellent example. I remember hearing Ericsson’s Stig Källman describe forthcoming demands on PCB manufacturing in a presentation to the EIPC, including the reductions in process variability needed to ensure signal performance for 5G data rates of 112Gbit/s and beyond. By 2023, tolerances for key parameters must reduce by about 50%, such as those for line width, which must reduce from ±30µm to ±12.5µm, substrate thickness (±10% to ±5%), layer-to-layer registration (±150µm to ±100µm), and impedance (±10% to ±5%).

PCB fabrication is typically batch-oriented, so the tighter tolerances required could be achieved by adopting single-piece flow manufacturing techniques, which some companies are already using. Aided by RFID technology that uniquely identifies each unit and its individual measurements and characteristics, successive fabrication processes can apply any necessary compensations and thus effectively optimize settings on a one-by-one basis to meet tighter tolerances at the end of line. It can raise yield, save corrective processes such as laser trimming, and reduce scrap.

Of course, smart processes and factories represent only one aspect of a bigger picture. We need to smart-scale the entire supply chain and throughout the enterprise to maximize the potential gains in business performance. It can also help us toward our environmental targets. It’s not only about manufacturers sharing planning data with suppliers, a point I’ve made before. We need better tools to predict real customer demands. With this information we can use the world’s resources to make the products that customers will buy and cut the waste associated with making unwanted products that do nothing more than sit on shelves. The dotcom/telecom infrastructure implosion of the early 2000s showed us how things can go badly wrong without a proper understanding of market demands and usage patterns.

Accurately predicting end-user market demands is probably the most difficult challenge of all, however. Consumers themselves often don’t know what they want or need, and diverse interests – philosophical, commercial, political – are at work seeking to influence consumer behavior and buying patterns. Also, with the upheaval of the past two years, customer demand is currently a huge unknown and likely to remain that way for some time before settling down.

AI can probably give us the power to cut through the noise to capture an accurate picture, but the data science directing those AIs will be critical. By mastering it, however, we might significantly reduce our impact on the planet, making fewer things by focusing on making only the things we need.

Alun Morgan smiling
Alun Morgan
is technology ambassador at Ventec International Group (;