What’s next for non-destructive testing?

Randolf Hanke talks about Lederhosen, artificial intelligence and cognitive sensor systems

Sensor systems used in non-destructive testing can do a lot more than is expected of them today – of that much Prof. Randolf Hanke is sure. The deputy director of Fraunhofer IIS, division director of Fraunhofer EZRT, and holder of the Chair of X-ray Microscopy at Julius-Maximilians-Universität Würzburg (JMU) regards sensor data as the raw material of the future and explains how it will change the modern world.

How would you define “non-destructive testing,” and what kind of tests does it refer to?

A few years ago, I could have simply given you the textbook answer without any hesitation: non-destruction testing, or NDT for short, simply meant inspecting components and structures for quality defects that prevent them from operating correctly, without having to take them apart or destroy them. That’s still the case today, but the definition is no longer quite as narrow or straightforward.

© Fraunhofer IIS/Udo Rink
Prof. Randolf Hanke is deputy director of Fraunhofer IIS and head of the division Development Center X-ray Technology in Fürth. In addition he is director of the Fraunhofer Institute for Nondestructive Testing IZFP in Saarbrücken and holds the chair for X-ray microscopy at the Julius-Maximilians-Universität Würzburg.

Why is that? What’s changed?

The current view of NDT is almost certainly too limited and incomplete, not only as regards solutions to NDT problems, but more especially concerning the range of possible applications for NDT methods. Let me illustrate this point with a topical example:

Imagine you’re in Munich on business and you’re invited to Oktoberfest. You’d look completely out of place in a suit and tie; what you need are authentic German Lederhosen. They’re not exactly cheap, but you go ahead and order a pair on the Internet, wear them to the beerfest, and impress everyone with your dress sense. So far, so good. At the end of the evening, you take off the Lederhosen and hang them in the closet, where they will remain unworn for 364 days. Some might hit upon the clever idea of sending the Lederhosen back to the store where they were bought. And this brings us back to non-destructive testing: when they receive returned goods, vendors need to inspect their condition to determine whether they are still salable. In my opinion, non-destructive testing methods can be easily adapted to such tasks. This is a huge market for sensor technology, but nobody would describe it as non-destructive testing in the usual sense.

 

In other words, the market for non-destructive testing has a bright future ahead?

Absolutely! You simply need to consider the phases of the product lifecycle. Each product’s life consists of phases before and phases after manufacturing or production. The cycle starts with the raw materials and ends when the product is recycled or put to another use. In between these two points, the product passes through a series of links in the value chain, including sales and marketing, shipping, and e-commerce. The question we need to ask is, where can we find other customers and potential users of NDT technologies and, above all, what issues are of importance to these customers and what solutions can we offer? One thing is sure: the answer doesn’t lie solely in the production sector – even bearing in mind the wide diversity of issues specific to different business sectors.

© Fraunhofer IIS/Udo Rink

You mention the wide diversity of issues in different business sectors. What do you think customers want?

As scientists, we always try to simplify the question, which means applying the exclusion principle. Hence, we start by asking ourselves what the customer doesn’t want.

I think it’s safe to assume that most customers aren’t very interested in hearing about test systems (in the conventional sense). In brief, what customers really need is an intelligent solution that creates added value and helps them optimize their processes. This solution might use, for example, cognitive sensors that are smart enough to determine what data needs to be collected and analyzed to deliver the information needed to make the right decisions.

This principle applies to every business sector, any process, and any imaginable task. A good example is the project we did with the startup company Mifitto. We were asked to find the most efficient and cost-effective way of extracting digital data on the internal dimensions of thousands upon thousands of different pairs of shoes so that online shoppers can be sure of choosing the right size. We were able to deliver the necessary information based on precise, high-speed computed tomography data, and we also created a further significant source of added value by combining highly accurate X-ray data with intelligent software. As a result, Mifitto could advise its customers not only on size alone, but also on what shoe would provide the optimum fit.

 

What do you mean by “smart” in the context of data analysis?

Intelligent monitoring is the watchword. In the future, it will be less a question of deciding between good and bad than of providing customers with a tool that enables them to optimize their processes by means of a monitoring system. And, as I mentioned earlier, “process” doesn’t refer only to production processes. It also includes things such as materials development, design, maintenance and recycling processes. Each of these can and must be optimized separately. Consequently, we have shifted the focus of our research efforts toward the development of cognitive and self-adapting sensor systems.

© Fraunhofer IIS
Non-destructive monitoring along the product life cycle.

Where do you see non-destructive monitoring in the world of Industrie 4.0?

What we are trying to do today is extract information from the huge quantities of available data using adaptive algorithms that will enable us to take action, i.e. better understand, observe and optimize processes.

Big data analysis plays a part in this, but until now it was almost exclusively reserved for fabrication, logistics, cost management, machine performance and the like. Little thought has been given to the role of big data in the development of smart materials. In the future, changes to materials and products will be monitored throughout the value chain, from raw materials to the use and recycling of the resulting products – everywhere where the action of people, machines or environmental factors has an impact. And in the future, instead of indiscriminately measuring all data, we will gather only the relevant data. Smart measurement systems and cognitive sensor systems will decide for themselves what data is considered relevant or intelligent. This is the next R&D challenge for non-destructive monitoring, an area of huge potential in which NDT experts like ourselves have a great deal of experience.

I can easily imagine a scenario happening someday in which a customer receives an intelligent monitoring system in the form of a black box. The customer doesn’t need to know exactly what’s inside the black box and doesn’t need any NDT know-how to use it. It might contain robots, for example, which have access to numerous different sensor systems and can use whichever method they decide is best in each case. When the robot activates an X-ray system, an air ultrasound system, or a thermography system, it does so to find the answer to a specific, precisely defined question, and not simply to test something.

© Fraunhofer IIS/Udo Rink

This all sounds like science fiction. Do you really believe that such visionary ideas will soon become reality?

Absolutely! After all, this mimics the way human beings work. Everyone has a body equipped with different embedded sensor systems and a brain to process the data and control the body’s responses. Each new task is processed with a moderate degree of attention. As soon as an event is registered indicating that something isn’t right, we become more alert. We activate more of our senses and try to focus harder on what our eyes and ears are telling us. In other words, we humans always apply a nuanced approach to sensing, using our intelligence. That’s the standard we ought to be aiming for in non-destructive testing, too.