By Randy B. Hecht
Additive manufacturing is an answer that raises many new questions. When we build components outside the traditional mass-production model, how do we control and standardize their quality? How do we ensure compliance with the regulatory requirements in sectors such as medicine or aviation? What do testing and quality control look like for Industry 4.0? Tim Horn’s job is to answer these and related questions.
“We have to bring in the expertise of many different disciplines … it’s about learning to speak one another’s languages and learning about the different processes.”
Tim Horn isn’t just developing quality control and testing protocols outside the parameters of traditional manufacturing. He’s also drawing on knowledge and experience gained while following an unconventional path to an additive manufacturing career.
After training as a cabinetmaker, he earned a degree in wood and paper science and engineering. As a researcher in wood machining and tooling, he worked within a program focused on cutting tool design and tool instabilities for the sawmill industry. A master’s degree in manufacturing engineering and a Ph.D. in Industrial Engineering set the stage for contributing to advances in testing and quality control for additive manufacturing. We spoke with him about what testing looks like today and where it’s headed.
When you began working in additive manufacturing, much of the focus was still on prototyping, and the cost-benefit ratio was not yet clear. What inspired your interest at that stage?
I was ready to go out and get a job at about the same time that NC State acquired the Arcam metal additive technology. This was nearly 18 years ago. I’m not sure how many people really took it seriously back then. But I was amazed at the tremendous potential in the idea that you could make a metal object directly from the CAD file without a mold or tool. From that point, I started working in materials and processes for metal additive machines.
One of the challenges in testing and quality control is repeatability in results. What are the biggest changes you’ve seen in results using electron beams and laser-based metal processes with metal powders?
Fundamentally, the powder bed additive manufacturing processes haven’t changed appreciably over the years. What has changed is our ability to predict process outcomes given a rather complex set of inputs. The processes have become more robust and more repeatable for a small subset of alloys – titanium aluminide, nickel superalloys, alpha beta titanium – maybe five or six proven commercial alloys. And there’s this enormous space out there with new materials that exist but aren’t used for this process and materials that don’t exist and are yet to be utilized. There’s a tremendous amount of work that can be done.
What is it about the nature of AM that makes it necessary to find a new approach to testing?
Additive in general lends itself to small-batch quantities of geometries that are highly complex and otherwise difficult to produce. As amazing as it is, AM can be incredibly slow. Build times can range from a few hours to several days or more. The real power and promise of additive manufacturing is in the elimination of part-specific tooling and the idea of producing just one of something – a custom patient-specific implant or a replacement part for an aircraft – but to do so in a digital, predictable and traceable way.
The main challenge then becomes one of quality control and assurance. These are almost all critical applications where the risk of failure may be low but the cost of failure may be exceedingly high.
How do you, as a process engineer, guarantee when you make a new custom geometry that it meets all of the quality requirements for a given application?
That becomes quite challenging with additive because there’s a geometric dependency on the localized heat inputs. For instance, thick and thin wall parts within a layer may exhibit different thermal inputs and therefore different solidification microstructures – and, of course, the properties.
On top of that, we have this difficult-to-predict set of stochastic boundary conditions around this weld pool that we’re creating. The powder bed itself is made up of very small particles that range in size, so we see local variations on the order of a few hundred microns. That significantly alters the packing characteristics and the bulk thermophysical properties, which necessarily requires us to change our inputs – power, velocity, focus – to maintain a constant set of solidification conditions.
The question becomes what fidelity of model is required to predict these properties and performance across multiple length scales. Ultimately, I would like this done in real time and integrated with many machine controls so that defects can be identified and eliminated in a single process step.
You describe your research as application-agnostic. But in practice, there are sector-specific challenges and instances of skepticism. How do you deal with those?
Additive really lends itself to specialized high-value-added applications, often in regulated environments. Qualifying these components and processes for non-standard or widely varying geometries becomes quite challenging.
Within specific sectors, there is a great deal of skepticism, but I think it’s generally been open-minded skepticism, and that is a good thing. The stakes are incredibly high in these environments. We have to show, as researchers and process engineers, what the processes are capable of and what the limitations are. It requires the combined expertise of many different disciplines.
If we imagine a day when we are making implants in the surgical suite for instance, we would need to compound all the expertise, information, material characteristics, simulation tools and process controls into a single operation. Today, at least in our lab, when we design 28 technology & innovation implants, we have the clinician, the surgeon, often the radiologist, the anesthesiologist, the engineers, the modelers, all together in the room, and we design this implant concurrently.
It’s a step in the right direction, but also a long way from where we need to be in order to make the “vision” a reality. The situation is the same whether it’s a particle accelerator, a reactor or a high-temperature aerospace application.
What does the testing process look like, and how do you test the testing?
The scientific question is, how do we affect quality control on a single component? If I make one implant for one patient, how do we predict its lifetime? How do we guarantee that failure will occur within an acceptable window of confidence? Which tools do I need to improve my confidence?
In traditional processing, we do that by sampling among a population of many other parts. In additive manufacturing, we don’t always have these flavors of data. I think that is where our research is today. It’s developing the materials, moving through qualifications and utilizing process monitoring, process control and post-build inspections.
Recently, we developed a suite of sensors that utilize artifacts of the electron beam melting process itself to generate in-situ imaging and data, in real time, during the melting process. Essentially, we have turned our Arcam production EBM systems into high power electron microscopes that also manufacture componentry. With these tools we can identify porosity, cracking and variations in material density and composition.
“We’re getting better and better at predicting. In two or three years, this will be a very different conversation.”
What advances should we anticipate in the evolution of nondestructive testing?
It is computationally expensive right now, and we’re not always entirely sure what to do with the massive quantities of data we generate. But all these things are advancing as I’m speaking. We’re getting better and better at predicting. In two or three years, this will be a very different conversation.
A great deal of work is going into the science of additive manufacturing and measurement. The modeling of the processes and our understanding of the underlying physics is continually improving. And all the while, the processes themselves are improving. The tolerances are getting tighter. Standards and best practices are developing. As these factors converge, we will be able to operate within a tighter set of manufacturing and design limitations.
What can you tell us about the Consortium on the Properties of Additively Manufactured Copper and its work?
We’ve observed a growing demand for sophisticated solutions in power electronics, radio frequency devices, accelerator components and thermal management using high purity, oxygen free electronic grade copper for years. But it’s difficult to process copper using welding-based additive manufacturing processes and maintain the quality, density and purity required of these applications.
We’ve done quite a bit of research in this area and made some viable demonstrations. Moving these results into a set of qualified components and processes requires us to leverage the growing interest and the support of the machine manufacturers, powder material suppliers, parts producers and end users. So, this consortium was established as a prelude to the qualification of additively manufactured copper, to give us a deep understanding of the material itself and the influence of external factors like oxygen content, orientation within the build, effective geometry. We’re trying to get the level of understanding for copper that we have for more mature materials like titanium today.
he group’s founding members are GE Additive, Siemens, Radiabeam Technologies and Calabazas Creek Research, and we’re actively recruiting others. The more members we have, the more resources we can commit to the robustness of the data set we produce.
With his varied background, tim Horn has no trouble managing a trio of titles. In addition to being Assistant Professor of Mechanical and Aerospace Engineering at North Carolina State University, he is Director of Research at the Center for Additive Manufacturing and Logistics and Director of the Consortium on the Properties of Additively Manufactured Copper.