05th July 2022
Today, we have the unparalleled honor of interviewing Ed Ginzel, whose career has brought to work for many a prestigious NDT organization, and who still brings his expertise and passion to the NDE community in his retirement years!
Such a long and rich experience in NDE for you! It might not be possible in a single interview, but could you give us an overview of your past experiences in the NDE community?
Indeed, it now seems like a long career. It started in 1974 when I was looking for work after completing my degree at the University of Waterloo. One of the courses in my last year of study was radiochemistry (isotopes). Radiography sounded an interesting application of radioisotopes, so I found a job in a local NDT company and started as a radiographer's helper. The little lab provided services in the four main methods (Eddy Current testing did not come into Canada until the late 1970s), and I soon decided that ultrasonic testing was the most interesting for me. Over the next few years, I gained experience in NDT, mostly manual UT, by participating in the periodic inspection programme for Ontario Hydro.
In the early 1980s, I was given an opportunity to work in the NDT lab at Ontario Hydro Research. It was not until then that I was exposed to a more academic side of NDT. Apple computers were used to facilitate mechanised scanning, and we used HP 7475a multi-pen colour plotters to make B-scans and C-scans for our reports. One of the chaps in the lab was a newly graduated PhD. He was using his computer to predict wave forms in the ultrasonic setups I was using to inspect pressure tubes in the CANDU reactors. Another chap was trying to configure a xenon-arc strobe lamp to reproduce the photoelastic images we were seeing in the NDT International Journal. This was the start of my fascination with computerisation of NDT and modelling to see the real events occurring in ultrasonic testing, instead of the simple ray-tracing using straight lines drawn from a probe.
At the end of the 1980s, I was contacted by TransCanada Pipelines Limited to help with their development of the new approach to girth weld inspections. Zonal discrimination had just been pioneered by RTD in the Netherlands. The concept of zonal discrimination was a good idea, but the parameters in the field were being poorly controlled. Critical parameters were identified and a strict specification was issued that made the process much more reliable. In 1996, I became a member of ASTM (the American Society for Testing and Materials), and began a long association there as a volunteer on the NDT committee. My first task was the development of a new standard for zonal discrimination fashioned after the TCPL specification. ASTM E1961 was published in 1998. This became the first of several documents I was active in developing. Others included E2192 (Planar flaw sizing), E2373 (TOFD), E2491 (Phased-Array equipment evaluation), E2700 (Phased-Array testing of welds), E2904 (Characterising Phased-Array probes) and E3044 (UT of polyethylene butt-fusion joints). As part of my interest in Phased-Array inspection development, I also participated in the IIW committee that developed the ISO standard 19675 for the Phased-Array calibration block.
In the mid-1990s, I met a person in the UK that connected me with Dr. Phineas David Hanstead. For those unaware, Dr. Hanstead was the first to make a working photoelastic imaging system that was used to image ultrasonic pulses. Dr. Hanstead's system eventually turned up in the British Rail NDT lab where Ken Hall used it to make the images that so inspired me in the mid-1980s when I was in Ontario Hydro Research. After corresponding with Dr. Hanstead (in the old-fashioned way using pen and paper and the Royal Mail service), he sent me his PhD thesis with instructions on how to make the strobed light source. I had this duplicated at the University of Waterloo physics department and made my first presentation at the WCNDT in Montreal where I provided the first photoelastic images of Phased-Array pulses.
Work on my photoelastic system was perhaps the costliest set of experiments with the lowest financial reward in my career; however, this work has been some of my most enjoyable. As a result of the imaging I could achieve, I was asked to attend a conference in Pretoria, South Africa in 2007. There, I met a pair of researchers working with M2M. Their presentation was about simulation using CIVA. Sitting behind one of the presenters, I watched him set up simulations and run calculations on his laptop. I decided then that I too wanted to be able to do this! I eventually saved my money and have never regretted the investment, in money and training, that I made to learn to use CIVA.
I have worked with TWI in the UK to help develop their CSWIP training courses on AUT (the now-familiar term for Zonal Discrimination of girth welds) and TOFD. To augment the training, I wrote handbooks on AUT, Phased-Array UT and TOFD. The TOFD book was translated into Russian and published by Olympus in Moscow, 2021.
In 2015 I told everyone that I retired. Being retired without a hobby can be disastrous. And since I was never a sports enthusiast, nor much interested in birdwatching or gardening, it seemed that I had no hobbies. I was left with a selection of NDT equipment that I had accumulated over the years, so it just seems natural to call ultrasonics my hobby. With this hobby I have tried to provide informative images from CIVA and photoelastic videos on www.NDT.net. It is my intent that others should find these useful as explanations to the way ultrasound works. These tools have also been helpful as I attempt to mentor students in the engineering faculty at the University of Waterloo. CIVA has allowed me to explain to students the origins of surface displacements that they had observed using doppler laser. In many cases, photoelastic imaging can be much faster at illustrating wave characteristics than the Finite Elements models that the students assemble.
You are still active in your company Materials Research Institute. On which projects are you currently focusing on? I think Reliability and MAPOD aspects are one part of this; could you also share a word on these topics?
In the past few months, I have been working on a few projects that take advantage of the information that CIVA provides. I mentioned that I worked on the development of ASTM Standard E3044 (UT of polyethylene butt-fusion joints). A few months ago, I learned that a technique had been used in Russia to inspect butt joints using a form of same-side TOFD where the Tx and Rx probes are mounted on the same side of the joint. This eliminates the lateral wave dead zone. I have been using CIVA to establish the optimum positions and skew angles to detect flaws in butt-fusion joints. Having identified the optimum probe positions in CIVA, I then design probe holders and 3D print them to validate performance of the configuration predicted by CIVA.
Probability of Detection is the most popular quantitative method of establishing the reliability of an inspection procedure. My first exposure to a qualification using POD was in the year 2000, for an RD Tech PipeWIZARD system being qualified by Saipem in accordance with the newly revised DNV OS-F101. That was the first year it required qualification by POD. Had CIVA been available then, the process could have been much easier. CIVA tools for predicting inspection reliability are relatively new. Since the introduction of the POD module in CIVA, I have been able to take data from Excel tables and import them to the CIVA POD module and run PODs on actual datasets. Toggling off any outliers and adjusting thresholds makes the qualification analysis very simple. More recently, I attended training on the even newer reliability tool, Metamodels. This allows us to establish a MAPOD (Model Assisted POD). Whereas a POD computes reliability based on a collection of dimensions and amplitudes, Metamodelling is like making a model of a model. We can run dozens of scenarios varying multiple parameters, simulating conditions in the real world. The CIVA Metamodel module then allows us to determine what parameters have the greatest influence on sensitivity, and we can also extract data from the metamodel and add real lab data to establish PODs. The ability to blend modelled and lab data has always been my idea of what a MAPOD is supposed to be. With our associate, Mariana Burrowes, we used data from her field analysis results with a metamodel and came up with PODs essentially identical to her field results and were, at the same time, able to identify the most critical parameters. Some were parameters that could be controlled in the ultrasonic setup, whilst others were random features of the flaws.
Another recent interest has been TFM. I think that TFM has a better potential for improving sizing accuracies than using traditional UT methods. I have been working with interested parties to see if we can formulate some guidance on procedures to use TFM to achieve flaw-sizing accuracies suitable for POD qualifications. CIVA has the ability to export simulation data in an instrument-neutral TXT format. This should make it possible to generate files from virtual flaws that can be assessed by the TFM software of most instrument manufacturers. Such a virtual round-robin approach could provide a statistical foundation to indicate accuracies that can be achieved using TFM for sizing in POD applications.
According to you, what are the main challenges for NDT companies in the upcoming years?
Perhaps the main challenges for NDT companies in the upcoming years will be the same as they were when I first started nearly 50 years ago: "staying current with the advances of technology". When I started in NDT, all of the NDT technology was based on analogue instrumentation. We had to learn that digital was the future. It was costly for established companies to get rid of instruments, that seemed to work fine, and replace them with new and expensive digital units.
Looking back at my first computer with its 1.02 MHz processing speed and 64 kB RAM with dual 5¼" floppy drives, I had moved a long way from the Krautkramer USIP 10 that I leaned ultrasonics on. When I selected a computer to allow efficient use of CIVA, I found a unit with an Intel i7-7700 CPU quad core with 8 threads operating at 3.6 GHz, with 32 GB of RAM and 2 external hard drives with a TB of memory on each. A few months later, my grandson assembled his own computer for gaming and it was out-performing my CIVA-dedicated computer for speed.
Technology is changing fast. For NDT companies (or any technology-based company), trying to keep up with the changes is a delicate balance. When is the old equipment no longer able to provide a competitive service? Then, when you decide it is cost effective to get the latest and greatest replacement instrument, how many extra features will you need to adequately service your clients, and how long will you be able to use that instrument before it becomes obsolete?
Moore's Law was based on transiter density. This has been used to estimate technology growth. Moore's Law suggests technology is doubling at a rate of 18-24 months. It would not be cost-effective for NDT companies to throw out their equipment and replace it every 2 years. But I am sure NDT technology-based companies will no longer be able to use their old equipment for 20 years or more as was done in the little NDT company I started in in the 1970s, where even in the latter 1970s I was still being sent out to clients with the Krautkramer USIP 10, made in 1960.
You are a "sharp" user of simulation software, including CIVA. Could you describe how CIVA UT helps you in your different activities? How often do you use CIVA?
I use CIVA regularly (multiple times per week). You will have seen me posting explanations of signals on the technical discussion forum on www.NDT.net. CIVA has been a great tool to augment my illustrations of waveform interactions and beam paths in my photoelastic work. I have often used CIVA illustrations as overlays on photoelastic images that I have published. One that I was particularly pleased with was the prediction CIVA made of the birefringent beam paths in quartz crystal.
CIVA allows me to import 3D CAD (STP) files of the probe-wedges that I design. With this feature, I can identify locations of internal wedge reflections and make modifications to the wedge design.
Another feature of CIVA that I have been using is the ability to duplicate my immersion setup for materials acoustic characterisations. CIVA has allowed me to simulate the immersion through-transmission technique so that I can verify the rotation angle of the goniometer when the shear mode is maximised.
Being able to position probes in "pitch-catch" arrangements with skews on curved surfaces has been extremely helpful. I could use equations to plot centre-of-beam crossing points, but this does not provide enough information about beam coverage. CIVA not only allows me to locate the crossing point of the Tx and Rx beam-paths, it can also provide a sensitivity map to indicate the non-symmetric characteristics that occur in beam spread on a curved entry surface. This is what I have been using as an aid to the design of probe holders for the same-side TOFD work.
What main improvements would you expect from the software?
The only CIVA simulation software module I have is the Ultrasonic module. It is extremely user-friendly now, and the only time I have seen it give me trouble is when I make an error in the setup parameters.
CIVA has incorporated a Probe and Wedge library. This is convenient to call on for off-the-shelf probes and wedges. You can then get the exact critical dimensions of wedges and the correct parameters for the elements. With probe manufacturers regularly adding new probes and wedges to their catalogues, it must be very difficult for the CEA to maintain that library. I would suggest that they make the library similar to the Materials library and allow users to enter values in the parameters tables. To preserve the original library (so it does not get corrupted when we users make a mistake), perhaps the original listed probes and wedges could be locked and any changes made to those values require that the modified probe or wedge be tagged as User Defined.
When in the probe library, I cannot read the full description of the columns.
A callout window that identifies the contents of the cell, when the mouse-pointer is on the name-cell, would allow us to read the parameter in that column. The ability to add parameters listed is nice, but adding more parameters makes it even more difficult to read what the column is listing.
When using Phased-Array probes, the bandwidth should always start around 80%. The default 50% is more suitable for monoelement probes. Even the 10 MHz Phased-Array probes from several of the manufacturers is identified as 50% in the library.