15th November 2016
Mr. John C. Aldrin, from Computational Tools, kindly shared with us his thoughts on simulation and the future of NDT.
You are working at Computational Tools. Can you explain to us what it is you do?
Since graduating from Northwestern University in 2001, I’ve been doing engineering consulting in the field of nondestructive evaluation, under the name Computational Tools. I’ve always had an interest in the transition of research developments and emerging technologies to practical applications. Much of my work has been focused on the transition of technologies with the U.S. Air Force Research Laboratory. This technology transition process is often called the ‘valley of death’ in the research and development (R&D) community, because of the inherent difficulty of the task. Promising laboratory techniques can often fail when confronted by the true complexity of real-world inspections and economic considerations. My work has concentrated on the application of computational methods to help support NDT technology transition. This includes NDT modeling and simulation, NDT data analysis tools, inverse methods for flaw characterization, and NDT reliability assessment.
How did you start with NDT Simulation?
While working on my PhD at Northwestern, I had a great introduction by Prof. Achenbach and the students and post-docs there to a wide variety of analytical and numerical methods than can be used to study ultrasonic NDE. At the time, I had to write my own boundary element method (BEM) code including the graphical user interface, for the ultrasonic inspection of aircraft structural joints. There, I discovered both the great value of using NDT simulation to bring insight into an inspection problem, as well as how hard it is to develop quality NDT simulation software. Putting a high quality software package together like CIVA UT is an immense task.
In what sense can simulation help innovate in NDT?
There are so many opportunities for the use of NDT simulation to improve the NDT technique development process. First, models are very useful in helping to interpret raw measurement data for complex inspection problems. As well, the key parameters of any inspection technique including optimization of the ultrasonic transducer can greatly benefit from parametric studies using accurate measurement models. This is one great feature provided by CIVA, where any parameter in the model can be varied in a parametric study, saving the user much manual labor in re-running studies. There are a few emerging uses of models for advanced NDT too. For the development of inspection techniques using automated classifiers, models can also be used to expand the training data set, potentially reducing sample costs and improving reliability if done correctly. Another emerging approach is the use of models directly within data classifiers through inverse methods. During the technique validation process, the model-assisted probability of detection (MAPOD) methodology has the potential to evaluate the reliability of an NDT technique at a lower cost with respect to conventional POD studies. Finally, when transitioning a new technique, NDT simulation can be quite beneficial in displaying a fundamental understanding of the inspection problem to project sponsors and helpful during the instruction of the inspectors.
As the first user of CIVA FIDEL 2D, can you give us your feedback on this tool?
From my opinion, CIVA FIDEL 2D is a great addition to the CIVA software suite. Under certain circumstances, the base package of CIVA UT has a very difficult time trying to manage all of the many reflections, transmissions and mode conversions of waves propagating in a multi-ply composite stack-up. By replacing this semi-analytical scattering model with a numerical finite difference mesh, accurate simulations of the scattering field can be achieved in a reasonable amount of time. This has been very helpful in trying to study the ultrasonic inspection of certain damage scenarios in composites.
One very useful feature which I believe most CIVA users will appreciate is being able to view the scattered field at any time step in this gridded region. This goes well beyond just viewing the beam in the part, but one can view the fine details of the scattered waves from a defect, and even create a movie of it. This feature can greatly simplify the task of reviewing the simulated B-scan data and trying to ascertain where each signal came from. Lastly, because this is a full numerical model of the scattered field, such characteristics as surface waves and diffraction at corners are better addressed. Even though these simulations are numerical and somewhat slower than a typical CIVA UT run, because the domain is limited to 2D, most B-scan simulations I’ve run to date were solved in less than an hour, or at worst case overnight.
Which improvement of the software are you expecting?
My interests in improvements might not be typical. For example, I’d really like to work with CIVA results in Matlab without having to export results to text files, or being able to control the number of threads CIVA uses for a run (so I can still use the computer for other things during the simulation). Long-term, I’d be very interested in being able to import more complex defect representations into both CIVA UT and CIVA FIDEL 2D. Most of my requests I think are quite minor though. CIVA has come a long way over the roughly 10 years I’ve been working with it. I would like to say "Merci beaucoup!" to the development team for the great effort that has gone into this package.
What is, for you, the main challenge for NDT companies in the next years?
I’d guess market forces and cost pressures are the main challenge for NDT companies in the coming years, especially dealing with on-going changes in energy production, such as oil & gas and nuclear. Now, if I instead consider the main challenges for computational tools in NDT today, I think there is a great opportunity to better take advantage of the ever-growing digital NDT data that are being collected. The challenge is how to move beyond traditional NDT ("accept"/"reject" decisions) and extract more useful information to support "condition-based maintenance" programs, as well as potentially impact process controls in production, and eventually design decisions. Delivering this capability is a significant opportunity for NDT, which would provide more value to our customers and encourage long-term use.