ACQUISITION OF TIME OF FLIGHT DIFFRACTION DATA BY
MOBILE ROBOTS
S. Mondal, T.P. Sattar, B. Bridge
Center for Automated and Robotic NDT, School of Engineering, South Bank University,
103 Borough Road, London, SE1 OAA
ABSTRACT. It is shown that Time of Flight Diffraction (TOFD) data acquired with 6 and 7 axes
robot scanning arms (that are carried by mobile wall climbing vehicles which can travel on curved
surface) is of a much better quality than that acquired by manual deployment. The robotic devices are
able to operate in remote and hazardous locations while maintaining more constant contact forces and
obtaining repeatable data unaffected by fatigue which manual operators find very difficult to match.
INTRODUCTION
The Time of Flight Diffraction (TOFD) technique has gained a higher profile over the
last ten years with much interest focussing on whether or not it can be used to replace more
established NDT methods [1]. A survey [2] shows that the annual average growth rate
(AAGR) of the TOFD market is 10-20% higher than other NDT techniques. The TOFD
method is gaining in popularity because of its high probability of defect detection, low false
call rates, portability and most importantly its higher flaw sizing accuracy of defects at
greater depth. To increase the application of the TOFD technique it will be necessary to
incorporate the technology into automated devices such as robot assisted scanning devices
and remotely operated vehicles. Then the technique can be employed in hazardous
environments e.g. for the inspection of nuclear power plant, oil rigs, and for the inspection
of large structures such as storage tanks, the hull of a ship, etc. Many industrial companies
such as Phoenix ISL, OIS pic, TransCanada and AEA, have developed automated
equipment to implement the TOFD technique. For example, the MAG-scan mobile robot
[3] is designed for automated TOFD inspection. It is driven by powerful magnetic wheels
and is applicable to pipe work, pressure vessels, and structural components. The Sub-sea
Pipe Scanner (SPS) [4] is an automated underwater TOFD scanner that was designed for
corrosion mapping of sub-sea pipe work. Automated TOFD weld inspection methods are
now used on girth welds in pipeline construction [5]. The TORUS system [6] has been
designed to carry out mobile automated TOFD weld inspection to the highest standards for
the nuclear industry and to cope with the rigorous conditions to be found in fabrication
yards of the offshore oil industry.
To further the automated deployment of NDT, the authors have developed a range of
robotic devices with the aim of building multi-tasking and cost-effective inspection systems
CP657, Review of Quantitative Nondestructive Evaluation Vol. 22, ed. by D. O. Thompson and D. E. Chimenti
© 2003 American Institute of Physics 0-7354-0117-9/03/$20.00
899
that can provide access to large vertical structures and remote test sites that constitute a
hazardous environment for manual operators. These systems comprise of wall-climbing
robots that utilize vacuum suction cups to adhere to large types of surface and perform a
variety of scanning routines with dexterous robotic scanning arms. This paper describes
TOFD inspection results obtained with such systems. It shows that the quality of data is
much improved due to the ability of the robotic devices to maintain more constant probe
contact forces and to obtain repeatable data unaffected by fatigue which manual operators
find very difficult to match when operating in remote and hazardous locations.
THE PRINCIPLE OF THE TOFD TECHNIQUE
Although the TOFD method is well known, we repeat it here for the purposes of later
convenient reference to certain aspects. The principle of the TOFD technique is briefly
illustrated in Figure (1). The transmitting transducer T emits a short burst of ultrasound into
the steel plate whose thickness is H mm. This energy spreads out as it propagates into a
beam with some definite angular variation. Some of the energy is incident on the crack tips
(O & O') and is scattered at these points. Scattering from the edges of the crack (called
diffraction and normally 20dB less than the reflected beam) causes some fraction of the
incident energy to travel towards the receiving transducer R.
If the defect is midway between the transmitter and receiver (a position that can be
found by minimizing the time of flight by adjusting the two probes), the depth is D below
the surface, and ti, t2 are the top and bottom diffraction times, S is the probe separation, L is
the defect length, H is the plate thickness and C is the propagation speed of sound in the
test material, then the arrival times of the top and bottom tip diffracted signals are:
C
(2)
C
From the above equations (1) & (2) we obtain:
the depth D from the surface:
(3)
the size of the actual defect:
(4)
Amplitude
Time
FIGURE 1. The principle of the TOFD technique.
900
The accurate size measurement of defects depends on the transit time and factors such as
couplant thickness, probe separation and surface irregularities. Changes in couplant
thickness cause errors in the transit times t\ and h. If a transducer contacts on the surface
with couplant and the thickness of the couplant varies by Ay, then the error in timing St can
be obtained by using the formula given by Lidington et al, [7]:
(5)
Where Co is the ultrasonic velocity in the coupling media, Cp is the wave velocity in the
component under investigation. If glycerin gel is used to a 30mm thickness steel plate
surface and the probe separation isTOmm, then for O.lmm couplant variation the timing
error is 61.8ns/mm. If we add this timing error to equation (4) we will obtain 10% error in
the defect sizing. It will be shown later that due to couplant variation false surface breaking
defects can be indicated if the contact forces are not maintained constant e.g. by manual
deployment of the TOFD probes. These false readings can be prevented by using robotic
devices that apply and maintain proper contact forces with the test surface.
COMPARISON OF TOFD INSPECTION PERFORMED BY MANUAL AND
ROBOTIC ARMS
During manual TOFD, the transmitter and receiver probes are usually deployed
independently and it is very difficult to maintain a constant contact force and speed for both
probes while they are moving on the surface. A rigid robotic scanning arm on the other
hand can provide a more constant contact force and a steadier speed on both probes. In
addition the use of force sensors incorporated into the wrist of the arm can maintain a
desired optimal contact force between the probes and the test surface. As a result, the defect
detection capability and the quality of TOFD scanning can be substantially improved.
The other advantage of using a robotic scanning arm is the ability to obtain repeatable
TOFD data unaffected by fatigue which manual operators find very difficult to match when
operating in remote and hazardous locations. The requirements of a scanning arm for
remote NDT in hazardous environments are discussed in [8,9].
Tests were performed to assess the quality of TOFD data obtained by deploying the
probes with two different robotic scanning arms. A probe holder was designed to carry a
zero degree roller probe and TOFD transmitter and receiver probes (Figure 2). The probes
were deployed by a PUMA 260 robot arm. A test piece was constructed from a 18 mm
End effector
;e holder
2.5 MHz wheel probe
Flat spri
'ring
Receiver
Transmitter
FIGURE 2. TOFD and pulse echo combined probe holder.
901
TABLE 1. SOFRA test data acquisition board setup.
Type of TOFD probes
Angle of the wedges
Type of Metal
Size of the plate
Thickness of the plate
Depth of holes
Diameter of holes
Flaw Detector
2xlOMHz probes
60 degree
steel
440 x 65 mm
18mm
13 mm
5 &6mm
SOFRA test
Pulse Amplitude
Pulse width
Gain
Pulse repetition freq.
Trig-out-delay
Damping
Blanking time
Digitizing sampling rate
200 volts
70 nsec
50 dB
IkHz
22 jisec
50 ohms
lOjisec
100MHz
thick steel bar into which artificial defects were created by drilling 5 & 6mm diameter and
12mm deep holes from the bottom of the bar. The specification of the test piece and the
parameter set up of a SOFRA Test data acquisition boards are shown in Table 1.
An automated B-scan obtained with the PUMA 260 robot arm is shown in Figure 3. The
first dark line is from the lateral wave and the second dark line is from the back-wall echo.
Spot-1 is from the 5mm diameter hole and spot-2 is from the 6mm diameter hole. The
manual B-scan (Figure 4) has spotted the two back wall holes but the difference is that the
scan has also indicated two false near surface breaking defects which were not present in
the steel bar. This is due to variation in couplant thickness caused by variation of the
contact force.
Rough Surface Inspection
Experiment results obtained by automated and manual deployment of TOFD sensors on
a rough surface. The test specimen was constructed of two layers (steel and rubber bonded
together) with no defects present in either layer. The steel surface was rough due to mild
corrosion and the dimension of the test specimen was 300 x 300 x 10 mm. The set up of the
data acquisition board was exactly the same as in Table 1. A SOFRA-Xscanner was
attached to the end-effector of the robot arm. The arm was programmed to move in a
straight line on the rough surface while applying a suitable contact force. A speed alarm on
the flaw detector was turned on to indicate when the speed limit of the data acquisition
exceed 3mm/sec. The Automated B-scan obtained by a PC controlled IBM 7545 robot arm
is shown in Figure. 5.
Time (pj)
Bottom
Time (|i§)
.Backwtl!
eetio
Ftlse
bfeakmg deffej
indicated
FIGURE 3. Flat bottom hole steel plate
bar scanned by PUMA 260 robot arm.
Two bottom holes detected.
FIGURE 4. Manual scan of the flat
bottom hole steel plate bar. Two holes and
false surface breaking defects detected.
902
This spot
occurred due to
variable hand
movement and
leaded to a
deviation from
the straight line.
FIGURE 5. Two layers (steel and rubber)
rough surface plate scanned with an IBM
7545 robot arm deploying a SOFRA
TOFD X-scanner.
FIGURE 6. Two layers (steel and rubber)
rough surface plate scanned by manually
deploying a SOFRA TOFD X-scanner.
A manual B scan was carried out on the same rough surface and the result is shown in
Figure. 6. During the experiments the speed alarm sounded many times due to variable
hand speed. In the automated data acquisition system the speed alarm did not go off even
once due to the ability of the robot arm to maintain a constant speed. The results show that
the scanning arm has produced a better B-scan image than possible with manual
deployment of the X-scanner because of its ability to maintain a constant contact force and
steadier speed on the surface.
TOFD DATA ACQUISITION BY A DEXTEROUS 7 DEGREE OF FREEDOM
(DOF) ROBOT ARM IN HAZARDOUS ENVIRONMENT
Figure 7 shows a mobile inspection system developed by the authors that comprises of a
climbing vehicle that can crawl on smooth surfaces of any material composition and surface
curvatures down to a minimum diameter of 860 mm. Adhesion is via suction cups. The
vehicle carries a 7 DOF scanning arm specially designed for NDT. The arm is a sphericalTwo
tif
will
FIGURE 7. A 7 DOF robot arm
mounted on the pneumatic wall climbing
vehicle for the inspection of pipe nozzle
welds.
FIGURES. A seven DOF robot arm
mounted on the magnetic wall climbing
vehicle for the seam weld inspection.
903
revolute-spherical (3R-1R-3R) anthropomorphic arm. The extra degrees of freedom of this
redundant scanning arm can be exploited for a variety of applications- e.g. collision
avoidance, manipulability enhancement, and greater dexterity. All sensor signals for
inspection and the couplant feed are routed internally through the arm. A force sensor
mounted in the wrist enables desired probe/surface contact forces and orientation to be
controlled. This inspection system is designed to find nozzle weld defects on feeder pipes
in the primary circuit and the pressure vessel in a nuclear power plant. Figure 8 shows the
arm mounted on a climbing vehicle that uses permanent magnets to adhere to ferrous
surfaces. This vehicle is designed to operate on the hull of large cargo container ships to
perform weld inspection.
Tests were performed on the 860 mm diameter feeder pipe shown in Figure 7. The
length of the test pipe is 4.1m and the wall thickness is 2mm. For TOFD inspection we set
up the parameter for the robot arm as follows: scan length - 90mm, scanning velocity 5mm/sec, contact force - 15N and acquisition of data samples every 1mm during the
motion trajectory. The set up of the data acquisition board was exactly the same as in Table
1. The automated B scan result is shown in Figure (9). A manual scan, Figure 10, was
performed by pushing a hand held flat SOFRA Xscanner to get data and its position coordinates from the scanner's encoders. Less internal flaws were detected in the joint weld
line than the data obtained by the robot arm. The robot arm maintained a more constant
contact force with the surface on both probes and as a result spotted more internal flaws in
the joint weld line.
TOFD DATA ACQUISITION BY A 6 DEGREE OF FREEDOM (DOF) ROBOT
ARM MOUNTED ON THE WALL CLIMBING VEHICLE
A wall-climbing vehicle has been designed for climbing surfaces of different
compositions such as concrete, brick and metal. It also incorporates the ability to avoid or
surmount small obstacles such as studs, welded seams, etc. The elements of the climbing
Lateral wive
Weld internal
flaws
Baekwalt
echo
FIGURE 9. Pipe nozzle weld inspection
scanned by a 7 DOF robot arm mounted on the
wall climbing vehicle. Automated defect
detection trial shows that it has detected more
internals flaws than the manual system.
Lateral wave
Weld
•flaws
BickwaJ!
echo
FIGURE 10. Manual scan of the pipe nozzle
spotted fewer internal flaws in the weld.
Surface and backwall signals were stronger
due to a smaller probe separation than for the
automated scan on the curved surface.
904
TOFD
wttib
60 dog. Migfe
FIGURE 11. A six DOF arm mounted on the wall-climbing vehicle for the TOFD inspection of a section of
ferrous oil storage tank.
vehicle are: actuating cylinders, a rotary cylinder and suction feet. A new servomotor has
been specially designed and developed in-house to replace the need for an electric motor. It
uses solenoid-operated control valves. This approach increases safety of the system, and
enables it to be used in potentially explosive environments.
A PUMA 260 six DOF robot arm was mounted on the vehicle to inspect the corrosion
thinning of a ferrous oil storage tank steel plate 2m high and 1.8m wide (Figure 11). The
robot arm has been programmed to move vertically along the Z -axis of the steel plate. An
artificially defect line has been inserted from back of the steel plate. The robot arm end
effector has been moved along this defect line. The automated B scan picture Figure (12)
indicates the defect line between the lateral wave and the back wall echo. The manual and
automated parameter setup of the data acquisition board was exactly same as in Table (1).
The manual B-scan has been carried out by the SOFRA Xscanner shown in Figure (13). It
has obtained poor resolution of defect line data because the inspection area was vertically
1.5m above from the ground level and the gravitation of the scanner weight and cables were
pulling downward, in results, there was no good contact on the surface.
Time
Weak
of
FIGURE 12. An artificial linear defect
machined into a ferrous oil storage
tank plate. Defect scanned with a
robot arm and detected strongly by
TOFD
FIGURE 13. Manual scan of the ferrous
oil storage tank plate. Weak defect signal
detected.
905
CONCLUSION
TOFD with a robotic scanning arm will in general produce more consistent results and
repeatable data than manual deployment due to the ability of the arm controller to maintain
a steady speed and constant force contact on the surface. The quality of data, acquired
manually in inspection tasks where large area scanning is required or where the tests have
to be performed in remote, harsh, hazardous or constrained areas, suffers from operator
tedium and fatigue and of course cannot be acquired at all in areas where access cannot be
gained either due to remoteness or due to a environment that is hazardous to human health
and safety. In these cases the quality of data is much improved over manual deployment by
gaining access via mobile robots and deploying the TOFD probes with a dexterous
scanning arm.
REFERENCES
1. Verkooijen, J., "TOFD used to replace radiography ", INSIGHT, vol. 37(6), pp. 433435, June 1995.
2. Khalid A., NDT markets: A brief review. NDT an Eastern-Western European
Perspective, September 1998, page 16-19.
3. Phoenix Inspection Systems Limited , Warrington, UK, TOFD Products information
guide, 2000.
4. "www.phoenixisl.co.uk/subsea/subsea.htmL"
5. Edward Ginzel, " TOFD enhancement to pipeline Girth Weld Inspection", NDT net
1998 April, Vol.3 No.4
6. "www.phoenixisl.co.uk/Torus/Torus.html"
7. Lidington B.H., Silk M.G., Montgomery P., Hammond G.F. (1976). Ultrasonic
measurements of the depth of fatigue cracks. Br. J. Non-Destr. Test., 18(6), 165-170.
8. Bridge B., " On the design of multi-task, compact, climbing robotic NDT systems for
remote operation on large surfaces and in hazardous environments." NDE 1997, vol. 13,
pp. 85-111.
9. Hollerbach, J. M. 1984, "Optimum kinematic design for a seven DOF manipulator.",
Proc 2nd Int. symp., Robotic research, Kyoto, Japan, pp. 215-222.
906