Presented at the ASNT Annual Conference 2018

Copyright © 2018 by The American Society for Nondestructive Testing, Inc.

ASNT Annual Conference 2018

How To Fully Utilize Your Phased Array System

Jesse Groom, Zack Groom

Excalibur NDT
2763 Buckhannon Pike, Mt Clare, WV 26408More Address
(724) 650-3696; email JGroom@Excalnow.com

ABSTRACT
When asked to do 80 welds in a day a phased array inspector might feel a little over whelmed. But more over
whelming than the inspection and field evaluation is the code compliant reporting and documentation of the 80
welds. Phased Array inspection is one of the most challenging NDE methods to date. The amount of information
and documentation can slow down total inspection times and fatigue inspectors. But there is some relief in the fact
that PAUT data is in a digital format and because of that we can automate processes. Over the years I and others
have developed advanced tools and workarounds to PAUT software and operating system limitations. This paper
will explore some of the advanced solutions and exploitations that can decrease inspection and reporting times while
reducing errors. This paper will also explore the use of optically track PAUT scanning and an available free opens
source software called 3Dslicer that can be configured to accept PAUT data from any unit with digital display ports
that can then model the data in 3D space as it was scanned.

Keywords: PAUT, Corrosion Mapping, Optical Tracking, 3D Scanning, Optically Tracked Scanning, Productivity

INTRODUCTION
In the past couple of years, we have seen some big advancements in phased array technology as well as the
instruments themselves. But as things have raced forward many gaps and issues have arisen. Report generation,
project tracking, and file storage continue to be a burden on the technician. The industry is still primarily stuck in a
two-dimensional world. Though there have been advancements in 3d evaluation, scanning is for the most part still
limited to a X/Y plane for data collection. Software that can manipulate and display 3D data is expensive and
difficult to learn. Technicians and level IIIs can make do and wait for more features, or they can develop their own
solutions. I have chosen the later. In this paper will discuss the challenges, issues, and techniques as well as what
would be the most beneficial steps forward.

Issues
REPORTING
Most NDT companies have a general template report of some sort: a blank Microsoft Word or Excel sheet, a PDF
form with fields, or even a cloud-based mobile reporting system. But the generation of a single document is not an
efficient system. If one has 100 PDF reports and needs to find the status of a report, or which inspections have a
particular version of a scan plan, the task can take hours. To overcome this, one would create a log or a ledger of the
inspections so that variables can be tracked, searched, and filtered. A system can be developed and created to
perform this tracking, and may work well until a customer’s variables change. Perhaps the introduction of a national
board number, or a drawling number. As the simple logging system is expanded to meet the needs of more
customers, the complexity grows and the likelihood of a breakdown in the system increases. All this large, complex,
flexible system work is very costly, and the learning curve is steep. Often times this results in failures or breakdowns
in the system.

Copyright 2018. This paper is intended for the sole use of registered attendees. No part of this publication or its contents may be copied, uploaded
to the internet, or stored in any shared retrieval system.
Recording
Most PAUT units offer the same basic functionality, with slight variations in features from manufacturer to
manufacturer. In many cases the technician, when approaching an inspection, will have information that needs to be
recorded. Most of the time this will become the data file title, weld number “-“ pipe size “-“ pipe schedule “-“
welder stencil “-“ 90 and or 270. This maintain a trackable record of the inspection as well as identification for the
scan data. After a day of scanning, a technician would go back to a laptop and create folders and start extracting the
information from the titles. There are other alternatives such as custom report fields, but this tends to bury the data
in a report file, PDF, or HTML. Often times the report would not be considered a complete report without a copy of
the scan plan and easily retrievable disposition of the weld, as well as notes, calibration blocks, SN, and so on. A
manual report by report system has redundant steps and introduces more opportunities for operator errors. A more
elegant solution is needed that does not depend on such high attention to detail and perfect performance at multiple
levels.

3D Data
As one looks at the scanners available on the market, they can see the extraordinary implementation of human
ingenuity in the mechanical field. Impressive accessories and automatic drive systems have come a long way since
the early 2000s. Scanners can be found for every common application and can be designed, modified, or
reconfigured to extend their range of use. But to this day most scanners are limited to mechanical means of data
encoding, and are planar in nature. If one would want to corrosion map a vessel, they would find that most scanners
want to collect data in a straight line, would not travel straight on the surface of a head without intervention, and will
have points of contact broken between wheels or probe/wedge and interfaces due to the rigid body of the scanner. If
a device was made to collect the data on the head, the resulting model would still be inaccurate due to the flat data
recorded vs the curvature of the head. Software may exist such that the data can be placed of a 3d model of the
head, but again accuracy would be lost due to out of roundness, data over lapping due to the curved inner and outer
surface. Areas like the knuckle radius or straight side of the flange would still require additional scans, or would not
be included in the model. This is just one example of the inability of modern scanners to accurately capture data on
complex shapes, and more problems exist in the following:
• Pipe Elbows

• Reducers

• Composites with plane changes

• Castings

• Fan Blades

• Tapered shafts

• 3D printed parts

For the most part, our current capabilities are weak in these areas.

Reporting Solutions
Reporting solutions require a system that is predefined, user controlled, semi-automated, and flexible. To my
knowledge, there is no such system one the market that will fulfill everyone’s needs. Software dose exist that allows
users to define their own process and output. These systems include Utex InspectionWare, eclipsescientifics
InspectionBank, and Microsoft Office products. For the purposes of this paper we will focus on Microsoft Office
products, but most database or data systems will operate with similar fundamental concepts and methods.
Development of these systems are difficult and time consuming. Often times they change as requirements change
from customer to customer. One of the most efficient way to develop systems is to use modular systems that cover a
type of inspection, a specific customer, and/or specific projects. The main goal is to create a system that has a
functionality that will work with the majority of the work to be covered, and then change variables on an as needed
basis from use-case to use-case.

Data Logging
The general method for doing this is creating tables that hold information or records in rows. Each row is an
inspection or a weld to be inspected. Data integrity techniques can be employed, such as database normalization,
where standard rules ensure the highest efficiency and error-proofing. We will not get into database normalization
in this paper, as resources are widely available and standardized. I will provide the following example. If one was to
create a simple data system for general PAUT weld inspection, they would need at a minimum two tables. One
table would store and record welds. Variables would include index number, size, schedule, material, joint geometry,
construction code, and welder stencil. This would be the basic information that would be expected. In another
table, weld inspections would be tracked. The variables would include index number for the inspection, index
number of the weld (linking to the weld table), date of inspection, technician, scan plan, number of indications, and
disposition of the weld. If we play out the common scenario of a rejected weld, we have a record in the weld table
and all of the information pertaining to the weld. We would have two records in the inspection table: an initial
inspection and an R1 inspection. We can see in the data a rejection of the weld, and then a reinspection of the same
weld. If a customer has an unnormalized or poorly structured database where they create a new record for the weld
repair designating an new weld id or weld name, the data can be easily converted from the inspection table but the
record is maintained as is should be in the two tabled database: a record of a weld and two records of inspection on
the weld, with two different results at different points in time.

Macros
The use of macros or Visual Basic for Application (VBA) command codes will allow for automated row indexing,
retrieval of the data, modification of data, printing to PDFs, file folder creation, and coping or directing created
PDFs to its respective folder. All of these decrease reporting time while reducing errors. Macros operate based on
the fact that a computer can be directed to perform simple and sometimes complex tasks that are very specific and
repeated frequently. The computer can do these tasks many times faster than we can with no chance of creating a
typo or error, as long as the base data is error free. A common activity a technician would perform would be to
organizing large sets of data; let’s say 100 scan data files, along with their respective 100 scan plans and reports. A
sheet would be created for the input of data specific to the weld. With some VBA coding, one could create a button
to take that data and log it in a table the weld information, and in a separate table information regarding the
inspection of that weld. Then, more VBA code can be added to create folders based on the name of the index
number and name of the weld. Two subfolders, one called data and one called reports, can be coded in. A file
picker can pop up and open to an inserted SD card, directing the user to select the scan data from the weld. Once
selected, the macro would move the selected data into the folder it had created. Finally, it could be moved to a
different data sheet where other variables could be updated into the record, such as number of indication and
disposition, while opening the data file in the respective software in anticipation of evaluation. The macro can do all
of these activities without error in a matter of seconds, where a human would take a considerably longer time and
potentially insert mistakes. All of the steps are easy and consistent, making them ideal for automation with the use
of macros. The step by step additive prosses allows individuals to make macros as simple or complex as they desire.
In the above example, records cannot be entered without getting a file folder and the user being prompted to move
scan data into that folder. By design, this system would increase organization and reduce the number and/or
possibility of errors while making the technician’s task easier. This would allow them to focus more in that moment
on data evaluation, rather than file storage systems.

Report Generation
Data that has been structured in rows can every easily be extracted into sheets pre-formatted to look like reports.
The user would call for a record, and the information would populate from the table into its respective location on
the report sheet. This allows for a central location of all the data as well as a single report form that, when modified,
updates all reports in the future. In the case of a new customer, the data system could be copied, cleared, and
variables could be changed to suit the reporting requirements of the new customer. A macro can be written to save a
specified area of an excel sheet to a PDF in a specific folder; all reports would be correct if the data was entered
correctly. Reports can easily be reprinted after an error in the data has been corrected. More advanced print-all
macros could be written so the reports could batch print themselves, thereby creating both customer and internal
copies. As you can imagine, the options are endless.

Recoding Solutions
While the system described in the above section covers some of the recording issues, there is an even more efficient
way to perform such functions. In fact, the more efficient system would make the system described above to be as
difficult as its less desirable manual alternative. Many PAUT field units are running on a version of windows or
Linux, and are modified in some way to display as if they are just running a PAUT operating system. In many cases
this means that Office products could run simultaneously on the units while they are preforming their PAUT
functions. One could now preform all the data input functions on a weld by weld basis on location in the field.
Rather than type in the size, schedule, material, and welder stencils, user forms could be created that have selectable
options per project and pre-populated welder stencils. Technicians would simply select grouped variables as they go
from weld to weld. The same folder generation could take place, only in this case the user could save the data back
to the default folder. The data system could move the data automatically into the folders. Access to the procedure
and scan plan could be a button click away. Setup files could be loaded or placed into the default load setup folder
where the desired setup would be the only one present, allowing the technician to just focus on collection and live
evaluation of data (of course depending on complexity of the inspection being performed). Though touch screens
and on-screen keyboards have come along way, it is still time consuming and difficult to type on the units. Utilizing
buttons and prepopulated data is the most ideal method of entering information.

Indication Reporting
A large movement in post-evaluation software has been centered around algorithms detecting defects within weld
data. This post-process automated defect detection is a new innovative development welcomed by all in the
industry. The downside of this technique is computation and data opening time. If sixty welds were scanned, and it
would take one minute just to load and then save the data back, then that would add an hour of time. A more
efficient data evaluation (again dependent on application and code) would be to evaluate data as the scan progresses.
In conjunction with the on-unit database, an indication table could be built in said database using mouse macros and
on-screen optical character recognition (OCR) to gather start/end location, through-wall height, amplitude, and
index location. This was simulated with PAUT PC software where OCR in conjunction with a mouse macro
obtained information off of the upper data display of gate and scan readouts. It was theorized that it would be
tedious when using a touch screen, particularly with amplitude value, due to angle select movement as the
technician’s finger leaves the screen. Without the use of a scroll wheel, live indication recording is difficult and
prone to errors. But this could be overcome with the angle selector/data readout being set to max peak. This would
alleviate the angle selector issue, but as many could imagine, the root reflectors could easily overpower indications
when maximum amplitude is the criteria for the region of interest. A work around would be the use of gates that
could be manipulated to into a polygon on a sectoral view so that the root reflector could be ignored. An additional
method would be to incorporate areas or shapes on the sectoral view where indications in said zones would not
become a region of interest.
But in the future, we could see the onboard software select indications in real time during inspection, then prompting
the user to evaluate the indication on the spot and log the indication in the indication table. This, coupled with a
with an onboard database software, would allow for a completed inspection (including report) as the technician
walks away from the weld. Such a system would save time with respects to indication computation, data loading,
and saving while providing for a real time weld disposition.

Optically Tracked Probes

The industry has made a big push to produce 3D images from scan data in recent years. Taking 2D data and
converting it into a 3D flat plane is easy, and has been achievable for some time now. This can be done with many
imaging software such as Image J, most CAD software, and it can even be done in excel. The new trend is to wrap
data around shapes to display the data in 3D. We are collecting 2D data, for the most part an X and Y encoded scan,
and then converting that and wrapping it into 3D space. This works, but does not offer us the versatility that one
would desire. While it is a nice visual to offer clients, this does not decrease our current limitations of complexity of
parts that could be scanned and evaluated.
The mechanical nature of our scanners is limited. While we have come up with some very inventive ways to scan
complex parts, baring FARO arms or robotic arms we find it very difficult to encode this data. Looking over to our
medical sister industry we can see that they are very heavily involved with optical and electromagnetic tracking of
instruments and devices. A very common practice is to track needles as they are moved through the body toward
areas of interest while imaging the needle tip with a sonogram. Hospitals at this point are developing their own
integrated image therapy (IGT) systems that perform these functions. For a software to be used to treat individuals in
medicine, rigorous documentation and a testing requirement is required. To aid in IGT system development, a
group developed 3D Slicer. 3D Slicer is an open source medical software that allows individuals to prototype
evaluation tools allowing individuals to evaluate their effectiveness for research, and prove out ideas before
integration into their own systems. The software accepts many different types of data: X ray, CT scans, still images,
and sonograms. Developers have made addons and smaller applications that work in conjunction with 3D Slicer.
One such program is Public software Library for UltraSound (PLUS). PLUS is an open-source software package for
tracked ultrasound image acquisition, calibration, and processing. It has been developed to accept data from many
different pieces of ultrasonic equipment, as well as tracking systems. There are direct data feeds available from
various Ultrasonic scopes, but the software was also designed to accept data from recorded frame images. This
could be done with a specialized camera (not advised) or any digital display output by use of an electronic frame
grabber. In this case, pixels’ color and spacing become data points. This means there is no issue getting PAUT data
into the software. Aside from the robust evaluation tools, it can take the 2D frames and link them to tracked position
and orientation data from probe tracking systems. It can then transform the 2D fames into a 3D space.

With the investment of a $600 USD frame grabber, a webcam, a 3D printed calibration tool, and access to a
conventional printer, one can optically track a probe and scan in any direction or orientation with any PAUT unit on
the market possessing a digital video out port (VGA, HDMI, DP, Mini DP). This of course is not the most ideal
setup, but can produce reasonable results. Some easy and relatively inexpensive upgrades can be made to achieve
positional accuracy of 0.011” RMS over a ~9.8’ X 6’ pyramid for roughly 16,000 USD. The camera upgrades also
come with the added benefit of lower minimum requirements for computer components.
Calibration
A very inventive calibration technique has been developed to obtain the transformation and time delay values. A 3d
printed model known as a phantom is scanned in a bath of water and then interpreted to figure out the correct
scaling, position of the pixels in relationship to the marker or tracking device, as well as a time-sync between
recorded frames to positional data. The calibration is fast, consisting of a series of steps that take about 5 minutes.
This calibration would take place when a marker was moved or when points on the screen change position in
relationship to the probe. This would mean that a change in focal law or screen range would require recalibration,
but if the marker position dose not move, calibrations could be loaded that were performed previously with setup
files.
It should be noted that with the submerged water calibration, shear waves could not be calibrated as they are not
supported in water. This issue could be overcome by use of a Steel phantom block or a redesigned block, again
made out of steel.

Please note the next three sections are based on speculation due to lack of time. Ideas presented are not claimed to
be fact, but theoretical speculation.

Ideal Applications
The obvious and easiest application that comes to mind is corrosion mapping. Zero degree or shallow angle sweeps
(-15 to 15 degrees) would easily calibrate and display on screen. Another application that can be employed is the
use of a gate or onscreen indicator to mark the outer surface when the entry surface dos not produce an interface
response. This would provide an artificial surface for small probes. Surface adaptive TFM would be the best
application, as the resulting model would have an entry surface and back wall surface profile. While shear wave
inspection in possible, and the software can model this data, review of the effectiveness of the calibration and
resulting model would have to be explored further.

Evaluation
3D Slicer has many tools for recognizing surfaces, measuring distances, and measuring volumes. Slicer 3d has tools
designed to measure thickness, but these are used with MRI scans where the bone has a specific density (in our case
amplitude) that the software then measures. The software takes the outermost point of a recognized surface and
calculates the distance through similar density, until it reaches a large change in density. There is hope however,
based on the way that the image is compiled. The position data and image data are located in two separate files.
There is a file of probe orientation and position, and a file (single binary file) of consecutive upright images. When
the model is compiled the software takes pixels on the upright images and transforms them into the position and
orientation where they were collected. This storage system would allow for conventional 0-degree, surface-down
evaluation that we are used to, while still providing for accurate modeling. We could evaluate in both methods, or
sweep through the data in a conventional fashion while having the model and volume tools available. Another
technique that could be employed would be to convert the first pixel in an image to a color scale that would
represent the nearest backwall pixel response representing thickness. This response would be a value above the
range at which normal amplitude responses could be detected. It could be filtered on, off, or made transparent.
Within the configuration files, setting can be manipulated such as data overwrite method, interpolation, and
sampling. The most useful of these would be the data overwrite setting where manipulation of the probe orientation
or rocking the probe would not mask high amplitude responses with lower, poorly coupled, or off-axis weak signals.
Another promising feature is the ability to export models in to CAD or .stl files. This could allow for accurate CAD
models to perform finite element analyses, extending the life of damaged components that could potentially be left
in service until replacements or scheduled shutdowns could take place.

More investigation is needed to determine the tools available and the existing capabilities.

Limitations and Future Progression

Currently, there are two main ways to track ultrasonic probes: optically (which we have discussed) and
electromagnetically. Both have limitation. Electromagnetic tends to be less accurate as well as susceptible to
interface with metallic objects, but has the added benefit of being much cheaper. Optical tracking has limitations as
well. It requires a line-of-sight, and requires less than 90 degrees of off-axis view in relation to the camera. There is
however a third technology that is currently undergoing technological advancement, and could overtake the two
previous mention methods. The technology is internal measurement units (IMUs). IMUs rely on accelerometers and
gyroscopes, and detect changes in three directions. Equations utilize these readings to determine position and
orientation. Currently, IMUs are not accurate enough on their own to conduct larger inspections. Small amounts of
electrical noise and/or error compound over time, which can lead to large errors. There are techniques to correct its
position over time. One could have a point to which the probe will periodically return to, and effectively reset the
driver back to a zero-error state. A hydride system could work in a similar way where one would periodically show
the IMU as well as a target to a camera so that it could reset the error. Despite the issues with IUMs, there is
promise in the future due to a recent intensified level of interest, mainly because of its use in virtual reality, cell
phone technology, and camera stabilization. IMUs are cheaper than electromagnet and optical tracking, and would
reduce setup parameters while overcoming the limitations of the other two methods.
The other limitation is a classic one. When inspecting curved or contorted parts, issues can arise with the probe-to-
surface interface. Larger 64 element probes, being the most ideal for scan time, may produce a substand wave
propagation within the part or areas where no sound entry occurs. Ideally one would want to use a surface adaptive
probe (would require TFM or a Time reversal technique) to obtain better sound transmission on curved surfaces.
The alternative would be the use of smaller probes, but this would increase inspection time. In the future, a flexible
array may be able to be integrated with optical tracking.

Conclusion
While the industry continues to race forward with many technological advancements, challenges still remain. It
appears that many of the challenges proposed in this paper are not primary focuses within the industry or do not
have a quick manufacturer solution around the corner. In some cases, such as the databases’ productivity tools,
manufacturer solutions will undoubtedly have issues and user modification on a customer by customer basis will be
required. In other cases (such as 3d optical tracking) manufacturer solutions could provide tailored systems, thus
reducing the demand on the computational power, and potentially operate directly on PAUT units. Although it is
nice to have a 600 USD alternative.
Realistically, not all companies will be able or capable of developing and deploying 3d optical scanning method or
customer productivity tools. The work that this paper is based off of, in terms of productivity tools, happened over a
5 year time period. During this time the capabilities of the productivity tools increased from project to project, and
each tool developed utilized learned techniques and code from the tools before it. For a new company to attempt
what has been done, considerable time and resources would be required. But there is a quicker way. The most
difficult part of designing productivity systems for new developers is to know what a software can do. After that,
what is the best way to utilize the functions to get the desired result. To aid in this prosses, my work will be
uploaded to a website with descriptions and tutorials so that others can download files and use, modify, or evaluate
so that they too can have or develop productivity tools. All will be encouraged to upload their content as well, so
that as a community we can grow and eliminate growing pains by learning from the work of others. Along with the
productivity tools, techniques, procedures, and methods will be discussed and shared again with the same intent.
Instructions on how to download 3d Slicer and the equipment necessary will be provided along with setup files and
instructions. In doing this I hope to share the burden of development with others and continue to toy with future
technology not yet available.

REFERENCES
(1) Fedorov A., Beichel R., Kalpathy-Cramer J., Finet J., Fillion-Robin J-C., Pujol S., Bauer C., Jennings D.,
Fennessy F., Sonka M., Buatti J., Aylward S.R., Miller J.V., Pieper S., Kikinis R. 3D Slicer as an Image
Computing Platform for the Quantitative Imaging Network. Magnetic Resonance Imaging. 2012
Nov;30(9):1323-41. PMID: 22770690.
(2) 3DSlicer. (2018, September 10). Website. Retrieved from https://www.slicer.org/
(3) Tang, Kevin. “Optical Tracking for NDE Sensor Scanning”, 15 July 2016.
Retrieved from https://www.nasa.gov/sites/default/files/files/K_Tang-D_Haanpaa-
Optical_Tracking_For_NDE_Sensor_Scanning.pdf