FMC/TFM Experimental Comparisons

Roger Spencer 1, a), Ruth Sunderman 2, b), and Evgueni Todorov 2, c)

1
EWI 1250 Arthur E. Adams Dr. Columbus, Ohio
2
EWI 815 14th Street SW Loveland, Colorado
a)
[email protected]
b)
[email protected]
c)
[email protected]

Abstract. Ultrasonic full matrix capture/total focusing method (FMC/TFM) technology has progressed significantly over
the past few years and has seen increased use in industry. The technology has the potential to provide better detection and
measurement capabilities for weld flaws, as well as, many other applications including additive manufacturing. This
project looked at the effectiveness of FMC/TFM for detection and sizing of both planar and volumetric flaw types.
FMC/TFM experimental data was collected and processed using multiple combinations of probe types and wave
propagation modes. The data was then compared to typical ultrasonic phased-array results, as well as FMC/TFM
inspection simulations.

INTRODUCTION
Industry is looking for improved and more reliable nondestructive evaluation (NDE) methods. A relatively new
phased-array technology is now available that utilizes the FMC/TFM signal measurement and processing
algorithms1. This approach has the potential to provide better detection and measurement capabilities for weld flaws,
as well as other applications such as additive manufacturing. The primary objectives of this project were to
investigate the capabilities of FMC/TFM for butt weld discontinuity and standard reflector detection and to evaluate
a modeling approach to optimize inspection.

FMC is performed using standard phased-array ultrasonic probes to acquire data from every possible pulse-
receive element combination of the probe array. For example, a single element in the array is pulsed and the signal
amplitude versus time data is received for each element in the array and stored for later processing. Then the next
element in the array is pulsed and each element again receives data from the pulsing of that element. This sequence
continues until all elements in the array have been pulsed. Following data capture, the data is post-processed using a
signal processing routine such as TFM to reconstruct the information into an image for interpretation purposes.
Signal processing occurs automatically using user defined settings prior to starting the data collection. The TFM
process divides the inspection area, within the part being tested, into a grid or mesh at the probe position. For each
point of the grid, the software calculates the time from each element to that point and sums the waveforms received
from all elements. This results in a sound beam that is focused at every point of the grid. The processed data can
then be displayed in a B-Scan (side view) image. This entire signal collection and processing procedure occurs at a
speed permitting real-time data visualization using relatively slow scanning speeds.

During this project an alternative technique known as advanced focusing method (AFM)2 was used in addition to
the full FMC/TFM technique. AFM closely matches the focusing quality of FMC/TFM but uses fewer data
acquisitions which leads to faster data acquisition speeds and smaller data files. With AFM all elements transmit and
all elements receive for multiple beam angles. Data from multiple beam angles (plane waves) are then used for
reconstruction. Typically, five to seven beam angles are used in the data acquisition. By using less data, the
computation speeds for AFM are greater than the standard FMC/TFM technique but with slightly reduced image

44th Annual Review of Progress in Quantitative Nondestructive Evaluation, Volume 37

AIP Conf. Proc. 1949, 020015-1–020015-5; https://doi.org/10.1063/1.5031512
Published by AIP Publishing. 978-0-7354-1644-4/$30.00

020015-1
quality. AFM allows for faster scanning speeds than standard FMC/TFM. In addition, another advantage to AFM is
a better signal-to-noise ratio as compared to a standard FMC/TFM technique.

EXPERIMENTAL PROCEDURE
For this project, two sample types were evaluated. The first sample type consisted of flat carbon steel plates in
which flaws were intentionally implanted during welding. The second sample type was flat un-welded carbon steel
plate containing electrical discharge machined (EDM) notches. The welded samples contained typical weld flaw
types and sizes that could be of interest to industry. Each sample was fabricated from flat carbon steel plate
approximately 12 in. (305 mm) in length by 0.75 in. (19 mm) thick with a 45-degree included bevel angle. The
samples were radiographed after completion of welding and flaw implantation to verify flaw presence. The EDM
notch samples were fabricated from un-welded carbon steel 0.75 in. (19 mm) thick with vertically oriented (through
wall), planar-machined EDM notches having the following dimensions:

x 0.016 in. (0.41 mm) Wide by 0.031 in. (0.8 mm) Deep
x 0.016 in. (0.41 mm) Wide by 0.067 in. (1.7 mm) Deep
x 0.016 in. (0.41 mm) Wide by 0.142 in. (3.6 mm) Deep
x 0.016 in. (0.41 mm) Wide by 0.299 in. (7.6 mm) Deep

Samples were scanned using phased-array ultrasonic testing (PAUT), standard FMC/TFM, and the AFM
technique. Scans were conducted using the same phased-array probes. Probes having different frequencies and
element dimensions were evaluated during the project with each of the techniques. The resulting data output from
each of the ultrasonic testing techniques was compared. FMC/TFM modeling and inspection simulations were also
performed using CIVA modeling software. The modeling results were compared to the actual FMC/TFM results
obtained experimentally. Examples of scan data collected from a welded plate sample are shown in Fig. 1 and Fig. 2.
All data in Fig. 1 and Fig. 2 were collected with angle beam shear waves using a 5 MHz 60 element probe having a
pitch of 0.039 in. (1.0 mm). On all FMC and AFM data and modeling plots, the reference to “Mode” indicates what
sound wave modes are used in the reconstruction and imaging algorithms. For example, TTdTT mode in the Fig. 1
FMC plot means that a refracted shear (or transverse—T) wave is generated in the specimen. Then, the transverse
wave is reflected from the back wall (TT) and encounters the flaw or defect (TTd). The transverse wave is then
reflected back from the flaw to the back-wall surface and is received by the phased-array probe (TTdTT). Modes
such as LdL indicate generation of a longitudinal wave and receiving of a longitudinal wave directly reflected from
the flaw (defect).

FIGURE 1. Example comparison of PAUT, FMC/TFM, and AFM techniques on carbon steel weld.

020015-2
 FIGURE 2. Results from LOF flaw 2-232 in carbon steel weld.

Following scanning metallographic cross sections were taken at selected locations in the welded plate samples to
better characterize the actual flaw sizes. Metallographic cross-section information was then compared to the
ultrasonic scan data. Figure 3 shows an example of a lack of fusion (LOF) flaw in one of the welded plate samples.
Ultrasonic data for the LOF flaw in Fig. 3 is shown in Fig. 2.

FIGURE 3. Metallographic cross section of LOF flaw 2-232.

Technique comparison scans were also conducted on the EDM notch samples to determine how each technique
would perform on planar vertical flaws having a smooth face. The probe selected for the notch sample scans was a
10 MHz 128 element probe having a pitch of 0.012 in. (0.31 mm). Multiple element aperture size combinations,
using refracted angle beam longitudinal and shear wave modes, were evaluated. Results revealed that the shear wave
modes provided the best overall imaging for the notches, and the AFM technique provided better imaging of the
notch tip than the standard FMC/TFM technique. An example comparison of PAUT and AFM data from a 0.142 in.
(3.6 mm) deep notch is provided in Fig. 4. The notch tip and notch corner trap can be clearly seen in the results;
however, the notch face is not imaged by either PAUT or AFM. Figure 4 also clearly shows the superior notch tip
resolution of the AFM scan versus the PAUT scan.

FIGURE 4. Comparison of AFM and PAUT techniques on a 3.6 mm deep EDM notch.

020015-3
 FMC/TFM Modeling and Inspection Simulation
During the project FMC/TFM inspection simulations were performed using CIVA modeling software. The actual
sample plate weld flaw geometries were used for the models, as well as, the notch dimensions from the EDM notch
samples. The purpose of the modeling and inspection simulation was twofold. First, it was necessary to determine if
the available PAUT probe was suitable for obtaining FMC/TFM data. Second, the simulations would be compared
to actual experimental results to determine how well the model compared to the actual experimental results. For
modeling purposes, the 5 MHz 60 element probe was selected to model the interactions with weld flaws while the
10 MHz 128 element probe was selected for modeling notch flaw interactions. Figure 5 shows a comparison of the
actual scan data and modeling data for the LOF flaw shown in Fig. 3. As can be seen, the results compared very
well.

Simulations of the notch samples showed that detection capabilities depend on the wave modes used by the
reconstruction software. For example, the LdLT and LdL-direct modes show the tip and corner trap areas of the
notch very similar to the actual scan data shown in Fig. 4. Based on these results it appears that incorporating other
wave modes in the FMC/TFM reconstruction could help image the smooth vertical face of planar flaws. Other
modes could not be used during this project due to software limitations.

FIGURE 5. Comparison of simulated FMC/TFM data to actual scan data.

FIGURE 6. Comparison of simulated wave modes on notch detection.

020015-4
 CONCLUSIONS
The following conclusions can be made from this experimental and modeling study:

x Along with EDM notches, all seeded and some natural flaws in the welds were detected and imaged
with PAUT, FMC/TFM, and AFM.
x Neither FMC/TFM, AFM, nor PAUT detected the smooth face of notches with the reconstruction
software used during this project.
x The 10 MHz, 128 element probe was better for FMC/TFM based on the material thickness used during
this project. The smaller pitch of 0.31 mm provided more beam spread which improved reconstruction.
x Computed simulation appeared to accurately predict FMC/TFM performance.
x Excellent focusing of TFM and AFM techniques provided better opportunity for accurate flaw sizing
but detection of small flaws was more challenging due to the flaws being displayed at actual size (large
background area with small indications—few pixels).
x PAUT provided good flaw detection due to a larger beam; however, the larger beam made flaws appear
larger than actual size reducing resolution.
x For many applications, it may be useful to perform PAUT for flaw detection and localized FMC/TFM
to obtain accurate sizing.
x The smooth fracture face of vertical flaws such as EDM notches was not detected with the TFM and
AFM. Modeling suggested inclusion of other wave modes in data output could provide better imaging
of vertical flaws.

REFERENCES
1. R. Long, J. Russell, and P. Cawley, “Ultrasonic phased array inspection using full matrix capture,” Insight, 54,
No 7, pp. 380–385, July 2012.
2. E. Carcreff, G. Dao, and D. Braconnier, “Total focusing method for flaw characterization in homogeneous
media,” Proceedings of the 14th International Symposium on Nondestructive Characterization of Materials,
NDCM2015, USA, Paper 75, 4 pp. (June 2015).

020015-5