In this month’s edition of the Pipeline Technology Podcast sponsored by Pipeline & Gas Journal, Guy Maes of Zetec, Inc. discusses various Non-Destructive Testing (NDT) methods that can be used to support weld inspections.
In this episode, you will learn about ultrasonic testing (UT) techniques, Phased Array Ultrasonic Testing (PAUT), Full Matrix Capture (FMC), Total Focusing Method (TFC), and Time of Flight Diffraction (TOFD). You will also learn about where the technology is heading to improve NDT techniques, improve data processing, and expedite data analysis.
Episode Show Notes, Links, and Insider Terms
- Guy Maes is the Sales Engineer Director UT for Zetec, Inc. Connect with Guy on LinkedIn or email him directly at email@example.com.
- Zetec is a global leader in nondestructive testing (NDT) solutions for the mission critical inspection needs of several industries, including oil and gas. Learn more about Zetec by visiting their website or following Zetec on LinkedIn.
- Read Guy’s Pipeline & Gas Journal article, “Using Phased Array Ultrasonic Testing in Lieu of Radiography for Weld Inspections,” in the September 2020 edition of the magazine.
- NDE (non-destructive evaluation) uses quantitative measurements to identify and characterize a defect in a pipe. Measurements focus on the size, shape, and orientation of the defect and take into account the physical characteristics of the pipe.
- NDT (non-destructive testing) is a group of noninvasive analysis techniques to determine the integrity of a material component or structure, without the need to take apart or destroy the test object. NDT includes several different types of tests and techniques to perform the assessment.
- Volumetric techniques can detect defects in the complete volume of the test object, and are used to examine the integrity of welds or materials where the joint quality is critical.
- Ultrasonic Testing (UT) is a type of NDT technique based on the propagation of ultrasonic waves in a tested object or material.
- Pulse-echo technique uses a single piezoelectric transducer to send and receive ultrasound waves in the test sample.
- Radiography is a NDT method that utilizes electromagnetic energy (ionizing radiation) from X-rays or gamma rays to detect both surface and internal discontinuities.
- RT (industrial radiography) is capable of testing welded joints that can be accessed from both sides. This NDT technique can be used to detect porosity, cracks, and other deformities in weld interiors.
- PA UT (Phased Array Ultrasonic Testing) is an advanced UT testing method that can generate multiple ultrasound beams to improve flaw detection and speed up inspections.
- Ultrasonic Testing (UT) is a type of NDT technique based on the propagation of ultrasonic waves in a tested object or material.
- Full Matrix Capture (FMC) is an advanced phased array data-collection technology in which each probe element is pulsed in sequence, and the signals from each possible pulser-receiver pair are stored in the instrument for processing and data analysis.
- Total Focusing Method (TFM) is how the captured FMC data is processed. TFM allows for the production of high-resolution images by creating perfectly focused signals in each pixel.
- Time of Flight Diffraction (TOFD) is an advanced UT method used for weld inspection to detect and characterize flaws. TOFD uses a pair of ultrasonic probes that sit on opposite sides of a weld-joint to perform the test.
- Pitch and Catch configuration is a set up where one probe emits the ultrasound wave and another probe receives the signal.
- Eddy-current testing (ECT) is a type of NDT testing method that uses electromagnetic induction to detect and characterize surface and sub-surface flaws in industrial materials.
Full Episode Transcript
Russel Treat: Welcome to the Pipeline Technology Podcast, episode 4. On this episode, our guest is Guy Maes, Sales Engineering Director for Ultrasonic Testing at Zetec. We’ll be talking to Guy about his September 2020 Pipeline & Gas Journal article entitled “Using Phased Array Ultrasonic Testing in Lieu of Radiography for Weld Inspections.”
Announcer: The Pipeline Technology Podcast brought to you by Pipeline & Gas Journal, the decision-making resource for pipeline and midstream professionals. Now your host, Russel Treat.
Russel: Guy, welcome to the Pipeline Technology Podcast.
Guy Maes: Thank you, Russel. I’m glad to be here.
Russel: I’m going to say right up front that this is another one of those podcasts where I’m talking to an expert. I may ask some dumb questions, because I don’t know a whole lot about this technology. That’s my caveat to lead us in. First thing I want to do is just ask you a little bit about, what is your background, and how did you come to be involved in pipelining?
Guy: Russel, I worked for Zetec since 2004. Currently, I’m the director of the global sales engineer team for our UT business. The company is based in Canada, the UT part. I have lived here since 2000, but I was born and raised in Belgium. I’m an engineer, so I have a technical background.
I have a degree in applied physics, and I majored in optics. I got into the NDE in UT in the late ’80s, because a large NDE inspection vendor in my home country, Belgium, was looking for engineers to develop new inspection techniques with acoustic focusing ultrasound.
I got involved in gas pipeline inspection because this company wanted to develop an inspection system automated for pipeline girth weld inspection. We built the system. We deployed for pipelines in Morocco and later in Belgium for Distrigas, which was, at that time, the national gas infrastructure company. That’s how I got into pipelines.
I got into Canada because after spending some time using equipment from a company called R/D Tech, they offered me a job. I decided to make the jump over the ocean with the family. Here I still am and still excited to do this.
Russel: You used the term UT several times. What does that acronym stand for?
Guy: Sorry to be so technical from the get-go. UT is ultrasonic testing. It’s testing of infrastructures with ultrasound.
Russel: Then likewise, I think most people in the pipeline world know what NDE is, but I’m going to ask you the same question. What does that acronym stand for?
Guy: Non-destructive evaluation. We also say NDT, non-destructive testing.
Russel: We didn’t talk about this before we got on the microphone, but what little bit I know about integrity management is prior to some of these more advanced techniques. The technique was to break stuff to determine how good it was. [laughs]
Guy: That’s what we call destructive testing, right?
Russel: Yeah, exactly. Obviously, there’s limited applications to destructive testing.
Guy: I also work for the nuclear industry. Destructive testing is not a good thing there.
Russel: You wrote an article for Pipeline & Gas Journal back in July of 2020 [published in September]. It was titled, “Using Phased Array Ultrasonic Testing in Lieu of Radiography.” Maybe we should talk a little bit about what is radiography and what’s the science around radiography.
Guy: I’m not really the expert in radiographic testing or radiography, but I know the technique enough to talk about it a bit and to compare with ultrasound, with UT.
First of all, both radiography and ultrasonic testing are what we call volumetric techniques. We use them to inspect the complete volume of a component or a weld as opposed to what we call the surface techniques, like dye penetrant and magnetic testing.
RT or industrial radiography, it’s very similar to medical X-ray, but it uses either X-rays or gamma rays in combination with a film. After exposure and when the film is developed, you can detect welding flaws, welding defects. For decades, it has been really the technique by choice to do volumetric inspection because it generates a permanent record of what the inspection saw. It’s the film.
In the old days, when ultrasound was done — I’ll get back to that, I think — it was manual. It was a guy with a scope. The only thing you had after the inspection was his observations written down or a statement that he saw nothing, which is less valuable than the film, right?
Russel: Yeah, exactly. What the film did at that time is it created a record that multiple people could look at. You could — some will say argue — discuss and build a determination around a physical record, where with the original ultrasonic, they basically looked at something through a scope.
Once they turned it off, the image was gone. It was basically what they directly observed and wrote down. It becomes much more dependent on the operator.
Guy: Absolutely. It’s exactly what you said, Russel. Several people could look at the film. We call that offline, whereas the manual UT guy, he just had to decide, on the spot, what he sees. Nobody could look at his scope together with him unless you recorded the whole thing on film, on movie, but that didn’t happen.
Russel: What is ultrasonic testing and how does it work?
Guy: Ultrasonic testing, it’s based on generation, propagation, and reflection of elastic waves in materials. Basically, ultrasound is a very high-frequency sound. We talk about frequencies mostly from 1 to 10 megahertz, which is about 1,000 times higher than what we can hear.
Once you get to the industrial process, what you do is you generate an acoustic wave in a component by using a piezoelectric transducer. The operator puts his transducer on the surface of the component. The scope, or the UT equipment, provides a pulse to the probe. It’s transformed in a vibration. It’s going to propagate in the material.
To make an analogy with optics, once the acoustic energy is in the material, it behaves like a sound beam, a bit like a beam of a flashlight. As soon as this energy encounters a discontinuity, for instance, a defect — a part of the energy — is reflected back to the probe. It again generates a pulse that comes up on the scope.
Think about sonar. This type of manual UT, we call that the pulse-echo technique. I’ll build on that later on. The simplest of transducers is what we call the straight beam transducer. It sends a beam straight into the material. Typically, you’re going to use this for thickness measurements to detect corrosion in pipelines.
There’s other kinds of transducers. It can be more sophisticated. There’s a way to generate angle beams. Typically, when you inspect a weld, you want to use transducers with different angles because different angles will see different defects.
Basically, it’s this guy who moves the probe along the weld and looks at his scope. That’s the UT of the old days, manual UT. It’s still used a lot, by the way.
Russel: I’m going to try to do a couple of visual analogies because it always helps me to do that to integrate something like this into my thinking or understanding. If I can visualize it in my mind’s eye, then I can understand it.
If you think about a guitar and I pluck a string on a guitar, the vibration of that string is dependent on the tension and the length. As I move my finger up and down the string, as I change the tension of the string, I hear a different sound.
Russel: That sound, if I’m playing an acoustic, it’s echoing out of the sound chamber. It’s echoing off the wood, basically. If you think about the ear as the sensor and the guitar as the generator, I’m changing what I hear by manipulating the tension of the string and the length of the string. It’s similar to that.
Probably a better analogy would be a light prism, where I take the light. I shine it into a piece of grass. It breaks the light out into its various colors. It’s kind of that, but I’m only seeing the part that reflects back at me.
Basically, I’m sending a signal. I’m looking at its reflection or listening to its reflection. The nature of that reflection is telling me the thickness of the material and are there any features in that material that I need to be aware of.
Guy: That is correct. That’s absolutely correct.
Russel: In particular, with UT, it gives me a way to visualize that, to see it.
Guy: Exactly. Especially with the evolutions that we had with the technology and with the computers and then the visualization of the UT signals. Basically, now you talk about visualization. How the visualization happens for manual UT, this is when the operator would get indications, would get echoes on a scope.
While he makes his report, he would have to record the arrival times — we call it the flight time — actually, the distance between the probe and the discontinuity, where the echo comes from. He knows that. He knows what the wave velocity is in the material. He knows the angle of his probe. At that time, he has to do some trigonometry. He has to make sketches. That was the imaging 50 years ago.
Since we do automated UT, we’ve got computers to do all the calculations for us. We get these nice 2D images and even 3D recently.
Russel: The technology of how I visualize the results is as important, if not more important, than how I actually generate the signal and capture the results.
Guy: There have been many breakthroughs and many evolutions in this technology, but especially, I would say, since the early ’90s, equipment became digitized instead of just analogic scopes. Then, with the digitization, the computing, and the imaging, there were great advancements between the early ’90s and the late ’90s.
Russel: Let’s walk through that history. Let’s start by talking about what is phased array and how does phased array relate to UT.
Guy: Actually, Russel, phased array works exactly the same way as conventional UT, at least from the moment that you get the acoustic energy into the material. Once the beam is into the material, it’s the same thing.
The difference lies in the transducer, in the probe. A conventional probe is what we call a single crystal, a monocrystal. As I said, it generates and receives back acoustic energy, but it’s all fixed. It’s the fixed angle and a fixed focal range where the probe works well. You need to use several probes to do a good inspection.
A phased array probe, in this probe the piezoelectric element is cut into multiple individual elements, 16, 32, 64. These individual elements can be individually pulsed. If we have the equipment to apply the pulses to these individual elements with very small time delays, you can electronically change the characteristics of your probe and of your beam.
You can change the angle of the probe. You can change the range where the probe is focused, where it works well. You can even change the size of the probe. That’s a huge advantage because, all of a sudden, you can use the same physical probe to mimic lots of conventional UT probes. It simplifies a lot of stuff.
First of all, you can generate different acoustic beams at the same time. Second, you can always shape your acoustic beam to be optimized for a certain type of defect. Those are two very important benefits. They were both applied when pipeline girth weld inspections were automated and then from automated to phased array. That happened about 20 years ago.
Russel: If I’m again trying to visualize what this might look like in my mind’s eye, I think about the transducers prior to phased array as shining a single beam, like I’m looking through a loop at jewelry.
I’m looking at a very small thing and probably monochrome, like I’m looking at it in a green screen kind of thing, shades of green. Then when I go to phased array, I’m actually looking at more material. I’m seeing a broader spectrum of colors.
Guy: The color — I wouldn’t go there in this specific case — because there’s one thing in a phased array probe that you cannot change, at least not up to now. That’s its frequency.
Russel: I see.
Guy: Frequency has a link with color in optics. Phased array probe for a fixed frequency, you can change a lot of the characteristics but not the frequency. You would still need to use different frequencies for different materials.
What I think a very good example, if we think about medical ultrasound — by the way, the medical ultrasound is phased array — when we have a medical ultrasound, we see this pie-shaped. The pie shape is actually built by sending beams at different angles with very small increments. We call that a sectorial scan in NDT, but it generates a pie-shaped image. You’ve got a section of a weld, which you can image in this pie shape.
Whereas conventional UT, you would just have one line. You would just have one ray, and now, you have 40 or 50 rays.
Russel: Guy, that’s a great example, because I can visualize that quite clearly. When you start talking about medical ultrasound, everybody’s going to think about looking at their babies in the womb. We’re all familiar with what that looks like.
The difference between phased array is I can actually see an image in two dimensions, whereas if it were the old school, I’d just see a single line.
Guy: Yup. The thing is, Russel, even when you do a manual phased array, when it’s not recorded, the operator, instead of seeing just an echo on his scope, he now has an image on his screen, and he can take a screenshot.
Somehow, you have now moved to an inspection technique that has a permanent record, albeit partial, because you don’t record everything, but you still have images.
Russel: One of the things that you talked about in your article is the idea of full matrix capture. Is this what we’re talking about here, is I’m capturing all these lines and all the information about the signals at a single moment in time?
Guy: No, full matrix capture goes way farther than that. This is really what we call…we call it now standard phased array, because the technique is now 20 years old. It was a big hype 20 years ago. Full matrix capture is taking it more than one step further.
Let me explain. As I said, this phased array probe has individual elements. Let’s say 64. Now, when we’re doing regular phased array, what we want those elements to do is work together and generate a beam in the material, and then have the beam come back, etc.
Now, with full matrix capture, we’re doing it differently. What we will do, we will pulse one of the elements at a time and then receive with all the other elements. We start with number one, we pulse it, and then we listen with the other 64.
We record all these individual signals. We call that A scans. Now, and then when that’s done, we pulse the second one, and we listen again, another 64 signals. That, up to element number 64. Now, at the end, we got more than 4,000 signals.
That is for one probe position, and that’s what we call a full matrix capture for one probe position. Now, I know it may sound like this is a huge amount of data, and to be honest with you, it is, but the reward when you have done this is huge.
When all this is done, then you can start processing these signals. Basically, you could decide after your inspection which probe you’re going to use. What if I used a 45 degree angle? What if I used another wave type, as long as you stick with the same frequency?
You can actually optimize the whole inspection after the fact. That’s pretty powerful.
Russel: It is pretty powerful. Again, Guy, it’s a really good explanation, because again, I can visualize this. The idea of sending from one of the elements in a phased array and having all the elements in the phased array receive the information back, basically, I get the square.
My signal concentration is whatever my number of instruments are, squared. That means that, well, for one pass through those 64 instruments, I’m getting 4,000 individual things to look at. Then I can then take back and evaluate by changing how I’m evaluating all those signals, like taking a lens, focusing it, changing the light, and all that kind of stuff, and seeing what you see.
Guy: Yeah, and all that after the fact. Make no mistake about this. All these calculations are complicated. There’s a lot of processing to do, but now, fortunately, the computers that we have nowadays, this is all pretty transparent for the user.
You just set your parameters. Most of the advanced systems on the market have easy-to-use software, and they help the operators to do the right thing. It generates, like you said, images that are almost perfectly focused in each pixel. That helps, of course, to have a better idea about the type of flaw, and also for accurate sizing.
Russel: One of the other things you talked about in your article is this idea of a bipolar pulse. What is a bipolar pulse, and what value does that add to this process?
Guy: One of the points is that, because this FMC [Full Matrix Capture] relies on signals from single elements, emitted by a single element, and received by a single element, those signals are weak. It’s very important to have a good quality of your pulser and your receiver hardware.
Now, quality is important, but you also want a very strong excitation pulse. Now, it’s complicated to put a high number of volts in a small instrument. You have to manage the heat inside. Now, a bipolar pulse gives you two half oscillations.
Instead of just putting a negative square pulse at, let’s say, 75 volts, you follow up the negative square with a positive square for 75 volts. It’s like, if you have a swing, and you push the swing from one side. Then when it comes back, you push it from the other side. You almost double your energy.
Russel: Right, yeah. By doubling the effective energy going into the swing, another way to do that is you could have the same amount of swing, but you’re inputting half the energy.
Russel: Which, in the electronics, is a big deal.
Guy: Yep. From that, you get more signal and, therefore, a better signal to noise ratio on your raw data and, of course, a better signal to noise ratio on your resulting processed data.
Russel: There’s a couple of other concepts you talk about in your paper that I want to unpack for the listeners. One is this idea of total focusing. We’ve talked about single-element phased array, and we’ve talked about having the energy requirement.
We alluded to this idea of being able to deal with all these signals and analyze them. How does total focusing apply to all of this?
Guy: What I have explained up to now is the FMC, the full matrix capture, which is how you take the data. The total focusing method, the TFM, is how you process the data. The TFM is specifically well known for producing high-resolution images.
What it does is generate images where each individual pixel of the image has been computed using a virtual sound field that is perfectly focused. You said it before, Russel. You talked about a magnifying glass.
You would focus your light energy in each pixel of your image and make sure that each pixel has received a focused light beam. That’s exactly what we do. Basically, we’re generating an improved version of this phased array sector scan image, but it’s better focused. It gives a better opportunity to look at your flaws and to measure them accurately.
Russel: That’s really interesting. That’s causing me to have a whole ‘nother sidebar conversation. I’m recalling one of the conversations I had with one of the satellite vendors and what they were doing with their image processing, doing multispectral, where they look at each element of reflection, and they look at multiple spectrums of light.
I think this is the reverse of that, where I look at one reflection, and I focus it perfectly.
Russel: I look at the spectrum that matters, would be another way to say that.
Guy: Yeah, the big advantage of using every element separately is that you basically are going to look at a certain defect from many, many different angles. If you go back to the beginning of our discussion, where we said, “Well, you can have three or four angles from different probes,” and in traditional UT, you always rely on as much energy that comes back to the transducer. We call that specular reflection.
It’s like a mirror image, but FMC and the TFM, you’re going to put sound from all these different directions, so your image from a complicated flaw is going to be more complete. That’s a huge benefit of this technology.
Russel: Let’s move on, if we could, to time of flight diffraction. Tell me a little bit about that technique, and how does it apply to this process of analyzing welds?
Guy: Actually, I have to say time of flight diffraction, or TOFD, it’s one of my favorite techniques, because it’s so simple and so powerful. It’s not phased array. It’s conventional UT. It was developed in the early ’80s, but it only became popular around the ’90s, once again, because of advancements in computer processing.
In TOFD, you’ve got two probes. One is sitting on one side of the weld, and another is sitting on the other side of the weld. One is the transmitter, and the other is the receiver.
We call it pitch and catch configuration. You have got the pitcher and the catcher.
One of the differences, the main difference actually, with conventional UT, like I said, pulse-echo where you’re just using one probe, is that now instead of looking mainly for these high-amplitude sound waves that you get from a specular reflection, the TOFD looks for low amplitude waves also that are generated by the tips of cracks and that’s what we call the diffracted signal.
By measuring the arrival times of these diffracted signals, we get a much better accuracy on sizing of these defects. That’s what makes the TOFD technique so special. It’s very complementary with standard pulse-echo UT and also with phased array. Combining them is the best of both worlds.
Russel: How do you do that? [laughs]
Guy: That’s very simple, Russel. Actually, most of the advanced phased array systems are capable of doing phased array with two different probes, at least, one probe on each side of the weld. You ensonify and you look at it. Either you do phased array, or either you do TFM, FMC/TFM. You just add two other probes, which are the TOFD probes.
Your equipment and your software, at least the advanced ones, they are capable of managing this. You put all these probes on a scanning mechanism. You just move it along the weld. After a pretty short time, you have recorded all your signals at the same time, a single scanning sequence instead of having to pass each probe separately. That’s a huge win in productivity, right?
Guy: Now, after the inspection, you have one guy sitting and looking at his computer and looking at the data. Because you combine techniques with different characteristics, you always have one, two, or three good pieces of information about the defect. That’s what NDE is about. You confirm your findings.
Russel: That’s one of the things that people that do this kind of work are always looking for is I want to have different mechanisms of gathering the data and compare the results because I’m always trying to weed through the noise to find the thing that matters.
Guy: Absolutely. Look, I’ve been a data analyst for somewhere between 5 and 10 years, looked at a lot of welds. Basically, the thing is you either look for things that shouldn’t be there, or you look for things that should be there and that are missing. That’s, in a nutshell, data analysis. It looks simple, but [laughs] it’s not that simple.
Russel: Exactly. Where do you think this technology is headed? What’s coming next?
Guy: We’ve made a lot of progress in data visualization. We still do a lot of analysis on 2D views, but the 3D visualization of your UT data in a CAD file, that’s a huge advantage if you want to communicate your results with third-party inspectors or end customers.
Of course, if the geometry is not a pipe but if it’s complex like a nozzle or — it’s not in the pipeline business — a turbine blade attachment, then the 3D view becomes your big rescue to locate an indication. That’s one thing.
In the future, the future development is going to be around automated analysis. Inspection data becomes huge, a huge amount of data. Even if you have a good analyst, it takes a lot of time. This automated analysis could be a big thing.
Russel: I have no idea what the time is from the time you gather the data until a qualified data analyst can say that weld is good, right?
Guy: It depends, Russel. If you have a very complicated weld, like in the main coolant piping of a nuclear power plant, I would say analyzing data could take somewhere between 4 and 12 hours for one weld.
For pipeline girth weld inspection, it’s going to happen in three minutes, at least when it’s onshore, because you’re going to drive to the next weld. They’re going to install this scanner. In the meantime, the analyst has to say go or no go.
It’s all in the way you do the analysis. Especially in pipeline girth weld inspection, there’s already some kind of automated analysis. We call it assisted analysis.
The thing is, with assisted analysis, the software itself will automatically create a list of indications from the data. These indications will satisfy a group of rules that the user can set. The analyst only has to review this and say yes or no because the preparation work has been done for him. Going to the full automated analysis, that’s still a big step.
Russel: The thing about an operator, an individual person, looking at data and analyzing the data and the error rate in that process versus the error rate in a fully automated process once it’s matured, those can be materially different.
Guy: Absolutely. They can. For instance, my Zetec colleagues in eddy-current testing typically work with two qualified analysts. When they agree on something, then it’s okay. Otherwise, you get a third analyst.
With UT, we typically don’t do that, but we get a second opinion for difficult indications. Yes, errors happen. If you can automate this process, that’s going to be a big thing. In my personal opinion, it’s going to involve neural networks and deep learning, but that’s not really my expertise.
Russel: It’s interesting. I actually worked with a company, gosh, back in the early ’90s, that was doing chromosome karyotyping and using image analysis to identify malformed chromosomes. The efficiencies and quality improvements they were able to get to were very compelling.
I think this is very much like that except that getting to the image you were analyzing was much easier in that case. In this case, getting to the image you’re going to analyze gets tougher just because of the nature of the complexity of the image.
Guy, this has been awesome. I have learned a ton. You’ve actually made my head hurt a little bit, which is always a good thing. I appreciate it. I appreciate it.
Guy, if somebody wanted to get in touch with you, what would be the best way to do that?
Guy: The best way is my email address at Zetec. That would be firstname.lastname@example.org.
Russel: For any listener that’s interested, we will put that information in the show notes. Then you’ll also be able to find Guy’s profile on the pipelinepodcastnetwork.com website. Guy, thanks again. Best wishes to you.
Guy: Thank you very much, Russel. Thank you for indulging my technical explanation.
Russel: [laughs] It was awesome.
Russel: I hope you enjoyed this week’s episode of the Pipeline Technology Podcast and our conversation with Guy Maes. If you’d like to support the podcast, the best way to do it is to leave us a review on iTunes Apple Podcast, Google Play, Stitcher, or wherever you listen to this podcast. You can find instructions at pipelinepodcastnetwork.com.
If there’s a Pipeline & Gas Journal article that you’d like to hear from the author, please let me know either on the pipelinepodcastnetwork.com website on the Contact Us page, or you could reach out to me directly on LinkedIn. Thanks for listening. I’ll talk to you next week.
Transcription by CastingWords