NDT.net • May 2006 • Vol. 11 No.5

Critical Comments on Detection and Sizing Linear Defects by Conventional Tip-Echo Diffraction and Mode-Converted Ultrasonic Techniques for Piping and Pressure Vessel Welds.

Ciorau, P.
Ontario Power Generation Inc.-Inspection and Maintenance Services - Pickering - CANADA


A Piping Pilot Study was performed to assess the capability of detection and sizing linear defects (EDM notches and fatigue cracks), as part of Performance Demonstration Initiative for CANDU Reactor Ultrasonic Inspection. The capability and limitations of tip-echo back-scattering technique and creeping waves-mode converted technique was assessed on a large variety of piping and pressure vessel samples with OD x t = 100 mm x 6 mm - 1,200 mm x 36 mm for carbon steel ASTM 106 Gr.B, austenitic steel ASTM 308 and dissimilar welds between carbon steel and stainless steel. The influence of defect orientation, height, location, piping counter-bore slope and length, weld geometric imperfection is presented for a large variety of probe frequency and size. The ligament assessment and the minimum crack height were also evaluated.


Sizing of inner surface-breaking defects is based on tip-echo back scattering. EPRI and other organizations include this technique in basic sizing for IGSCC in stainless steel piping[1,2]. Our published results for time-of-flight (TOF) diffraction[3] and for mode-converted techniques[4] trigger further investigation regarding the reliability of these techniques for piping welds of CANDU reactors. A Pilot Piping Study (PPS) project was performed during 2000-2001 period for both detection and sizing. The Technical Justification for sizing identified specific cases for ligament assessment. EPRI was the main consultant for this study. The recent R/D Tech books [5-6] use these techniques in the basic chapters for conventional ultrasonic inspection. Further OPG experimental work confirmed the limitation of conventional TOF techniques to distinguished between inner surface-breaking cracks and embedded defects (sidewall lack of fusion), if the ligament is smaller than 3 mm[6]. It is also a common practice to confirm the service-induced cracks located in the 1/3 of inner part of the weld by creeping waves probes. Most of the welds have counter-bores and in specific cases, the critical cracks are developed in the counter-bore transition area and/or in the weld root zone. Crack detection, confirmation and sizing with these above-mentioned techniques will be presented in the paper. A study of geometric irregularities on creeping waves propagation is also presented. More accurate crack tip measurements were performed, and the data are compared with TOF results, namely for conventional probes used in day-by-day inspection. The present paper is trying to answer the following questions:
  • What is the accuracy of TOF technique in sizing for EDM notches with inclined angle [10-40 degrees]?
  • What is the accuracy of TOF technique in sizing fatigue cracks in bars?
  • What is the accuracy of TOF technique in sizing cracks in the weld root area and in the counter-bores?
  • Is creeping waves technique capable to improve the ligament sizing / confirmation of surface-breaking defects?
  • Is creeping waves technique capable to locate and to estimate the crack depth near weld root, suck-back, mis-alignment?
  • What is the influence of counterbore on creeping wave detection capability?
  • Can creeping waves be used to detect the linear defects in K-welds of carbon steel shell to stainless steel set-in nozzle?


The present study was performed with regular mono-crystal probes of 12.7-mm diameter, used in combination with shear waves wedges of 45 and 60 degrees and longitudinal wave wedges of 70 degrees. Frequency range was between 2.25 MHz to 5 MHz. The population of samples with EDM notches, fatigue and SCC cracks ranges from OD x t = 100 mm x 6 mm to straight bars and piping coupons OD x t = 625 mm x 38 mm. A large variety of samples (see Table 1 to Table 3) were used in lab conditions (no time restrain, no limited access, ideal conditions for detection and sizing, a priori knowledge about crack/notch location, a priori knowledge about defect parameters).

Table 1: Blocks with EDM notches
Block ID Thickness
[ mm ]
Target type / dimensions [mm] Remarks
TED 1 6 Wire EDM 1.0 / 3.0 Straight bar/wire cut along the width
TED 2 12 Wire EDM 1.2 / 4.0 Straight bar/wire cut along the width
TED 3 18 Wire EDM 2.0 / 6.0 Straight bar/wire cut along the width
TED 4 25 Wire EDM 2.5 / 8.0 Straight bar/wire cut along the width
TED 5 36 Wire EDM Straight bar/wire cut along the width
TED 4-C 25 5 x 1, 5 x 2, 5 x 3, 5 x 4, 5 x 5 Straight bar / foil cut
TED 4-D 25 4 mm at 0-40 ; step 10° Straight bar/wire cut
TED 4-E 25 4 x 1.5 mm ; 0-40°, step 10° Straight plate with crack-like EDM notch
PPS-12 12 1.2 / 6 Straight bar/wire cut
PPS-18 18 1.8 / 9 Straight bar/wire cut
PPS-24 26 2.4 / 12 Straight bar/wire cut
PPS-38 38 3.8 / 19 Straight bar/wire cut
CB-T 27 A 25 2 ASTM 106 Gr.B PHT piping / 3 counter-bores at 10, 15 and 25 degrees. Roots, notches
CB-T 35A 35 2 ASTM 106 Gr.B PHT piping / 3 counter-bores at 10, 15 and 25 degrees. Roots, notches
Misalignment 17 5 x 2 / 37 Stainless steel 8" sch 100
Root 17 5 x 2 /37 Stainless steel 8" sch 100
Suck-back 17 5 x 2/37 Stainless steel 8" sch 100
Bleed Condenser 35 / 38 10 x 2 CS/SS K-weld

Table 2: Blocks used for ligament evaluation
Block ID Thickness
[ mm ]
Target type / dimensions [mm] Remarks
TED 4-A 25 EDM cut /4 mm Variable angles 0-40 , ligament 0 mm
TED 4-D 25 EDM cut /4 mm Variable angles 0-40 , ligament 3 mm
Lig 1.5 -25 25 EDM cut /4 mm Variable angles 0-40 , ligament 1.5 mm
Lig 6 -25 25 EDM cut /4 mm Variable angles 0-40 , ligament 3 mm
Lig 1.5 - 35 35 EDM cut /4 mm Variable angles 0-40 , ligament 1.5 mm
Lig 3 -35 35 EDM cut /4 mm Variable angles 0-40 , ligament 3 mm
Lig 6 -35 35 EDM cut /4 mm Variable angles 0-40 , ligament 6 mm

Table 3: Samples with cracks
Block ID Thickness
[mm ]
Crack height
#4 6.4 0.6 Forging bar; dynamic-induced fatigue crack
#6 6.4 1.6 Forging bar; dynamic-induced fatigue crack
#7 12.7 1.3 Forging bar; dynamic-induced fatigue crack
#8 12.7 5.5 Forging bar; dynamic-induced fatigue crack
#9 12.7 2.3 Forging bar; dynamic-induced fatigue crack
# 10 12.7 3.2 Forging bar; dynamic-induced fatigue crack
#12 19.1 1.2 Forging bar; dynamic-induced fatigue crack
OHR-20 20 7.8 ASTM 106-corrosion fatigue crack
#18 25.4 0.6 - 12.1 Forging bar; dynamic-induced fatigue crack
3E 39 15.5 / 3.2 Header; slope 25°; space 40 mm
9B 36-38 6.9 / 9.0 Feedwater piping 13" x 1.5", counter-bore cracks; slope 15°
3B 36 5.2 mm Feedwater piping 13" x 1.5", counter-bore cracks; slope 15°
7W 35-40 mm 4.2 / 4.7 / 5.3 Feedwater piping 13" x 1.5", counter-bore cracks, weld root cracks; 15°

Figure 1 to Figure 3 present some examples of the blocks used in this experiment.

Fig 1: Block CB-T 35A with counter-bores and roots.

Fig 2: Stainless steel blocks with artificial internal geometric irregularities and 5 x 2 EDM notches, oriented at 37.5°.

Fig 3: Examples of reference blocks 9B (left) and 7W (right) with service-induced cracks in the counter-bore and weld root area (feedwater piping).

Optical/metallographic, phased-array and magnetic particles additional techniques were used to assess the "bench-mark" crack height. Examples of crack morphology are presented in Figure 4.

Fig 4: Different types of cracks in bars and retired-for-cause piping welds.

Ultrasonic machines Sonatest Masterscan 335 and OMNISCAN 16/16-conventional UT were used in conjunction with GE Inspection technology probes MSW-QC of 2.25, 3.5 MHz and 5 MHz. Inter-change wedges W 211, W221, W212 and W 222 were used for 9.1 mm and 12.7 mm crystal-size bench mark piezo-composites. Measurements were averaged after three repeatable readings. Relative and absolute arrival time was used to size the crack height. Special charts for both UT path and projected distance were used to identify CE1 and CE2 signals. Measurements were taken on both sides of the crack, or through parent metal and weld (for retired-for-cause samples). Positive angles and negative angles are defined according to Figure 5.

Fig 5: Definition of positive and negative angles.

The ultrasonic data were evaluated for the last significant crack tip and for the best fit of TOF signals for CE1 and CE2. However, the amplitudes of crack tips were very small compared to EDM notches tip signals (see Figure 6). More than 20-dB difference were found between crack tips and notch tips amplitude. The sizing was performed by experts and by technicians, in lab conditions mentioned above.

Fig 6: Comparison of crack tips signals (left - crack height = 4.3 mm, middles-crack height = 6.3 mm) with EDM notch tip amplitude (h = 2 mm).

All A-scans were stored in Masterscan 335 and analyzed in a zoomed mode for a higher accuracy. Recent data were acquired with OMNISCAN 16/16 for better display, recording and reporting. However, OMNISCAN presented enhanced features versus Masterscan 335, such as optimization of the receiver filters, high-definition, smoothing, averaging, and larger sceen. Crack tips are much better displayed with OMNISCAN compared to Masterscan 335. Even with these enhancements some experimental results acquired in 1995, 1997, 2000, 2001,2004 cannot be improved. Details about these results are presented in the next section.


Taken into account the large number of samples and variables (probe size, frequency, refracted angle, defect size, defect orientation), the results were limited to the probes namely used in daily inspection (procedures). Studies of frequency effect on notch orientation and crack sizing using different types of probes (high-damped, FAST-1, SLIC 40, RTD creeper, SIGMA creeper, OPG creeper dual side, OPG creeper dual front - back, different refracted angles -35, 38, 40, 42, 52, degrees) will be published later on. Based on ALARA principle, the sizing is performed with the same probe used for detection. The ultrasonic results are grouped in the following categories:
  • tip-echo diffraction on inclined notches
  • tip-echo diffraction in sizing cracks
  • tip echo diffraction and creeping waves for ligament evaluation
  • tip echo diffraction and creeping waves for defects located in counterbore, root, and misalignment.


The results on TED 4A were grouped based on technique, refracted angle and probe frequency. Figure 7 to Figure 8.

Fig 7: The influence of notch angle and beam direction on sizing for 45 and 60 absolute arrival time technique (AATT).

Fig 8: The influence of notch angle and beam direction on sizing for 45 and 60 relative arrival time technique (RATT).

The data from these figures concluded:

  • the AATT is a better technique than RATT; disadvantage is related to probe movement towards defect for a signal optimization;
  • positive slope at larger angles (30-40 degrees)- more specular - is under-sizing the notch by a significant amount; more affected is 45 refracted angle;
  • negative slope is over-sizing the defect due to technician inability to distinguish the real corner trap from the skip echoes with higher amplitude; this might be the real-case in the field;
  • there is no major difference between 2.25 , 3.5 and 5.0 MHz probe performances for large defects (h >3 mm)


As mentioned above, there is no comparison between sizing EDM notches and sizing the last significant tip of the inclined cracks. The notch tip is round and wide (50-100 m), versus the crack tip of 0.5-2 m. There is trend to undersize the large cracks and an impossible task to size cracks with height < 1.5 mm. Sizing performance depends also on probe pulse duration (frequency). Figure 9 to Figure 10 presented data for this experiment.

Fig 9: Capability of crack sizing with a 5-MHz/45 12.7-mm diameter probe.

Fig 10: Minimum crack height dependence on frequency and thickness for 45 angle.

Fig 11: TOF sizing capability dependence on thickness and refracted angle for notches and cracks.

The results presented in the figures above concluded:

  • cracks with height < 1.5 mm cannot be sized with 5 MHz probes-bench mark; high-damped probes and higher frequency (7.5 MHz / 90% BW) are required to perform this task;
  • the detection probes could size cracks if their height is 15-25% of the pipe thickness;
  • sizing capability for cracks is almost double height-wise compared to EDM notches.


The ligament TOF (TED) measurements were performed with 5 MHz probe of 12.7-mm diameter, in combinations with 45 and 60 wedges. The results presented in Figure 12 concluded:
  • ligament is reliable assessed if the L > 3 mm
  • ligament is undersized, even for L=6 mm
  • ligament is better sized by 60 degrees technique

Fig 12: Ligament sizing with shear waves at 45 degrees (left) and 60 degrees (right).

Creeping wave technique is a complex task for every technician, namely due to the following factors:

  • multiple signals displayed based on probe position versus defect location
  • dynamic evaluation could produce confusion between CE2, CE1 and SW 30
  • defect orientation and ligament may be accurate evaluated only for parallel surfaces (within 5 tolerances);
  • crack signals are different from notch signals due to the crack shape, tightness and orientation
Technicians were instructed to follow the procedure (CE2 signal at 80% FSH + 8 dB), and report signals based on position in the gate, on gain and on probe position (projected distance) versus defect. Position and amplitude of CE1 were also recorded, to allow a decision for ID-connected based on combination of events (only CE1, only CE2, both CE1 and CE2, TOF between CE2 and CE1). Based on this matrix of events, the indications, which were out of the range, were called acceptable; same exclusion principle was applied for amplitude value (only for CE2 register/reported or acceptable). The results are presented in Table 4. Figure 13 illustrates some of the most significant OMNISCAN screen captures.

The conclusions of this experiment are:

  • confirmation of ID-connected or ligament value cannot be made by mode-converted technique, even for ligament value of 6 mm;
  • defect tilt angle can produce stronger amplitudes due to the skip of 30 SW component, which has a similar TOF and amplitude like CE2;
  • the amplitude from ID defects at +10 and -30 are below recording level
  • static evaluation based on pattern CE1-CE2 and 5 mm errors of CE2 in the gate is not reliable
  • dynamic evaluation moves the pattern CE1-SW 30 either to the left (by 6-10 mm) of to the right by 5 mm-10 mm, technician having difficulty in decision-making process;
  • frequency of 3.5 MHz and 2.25 MHz perform better for ligament L=6 mm
  • once the defect is tilted, there are 6 to 12 signals to be analyzed within 10 mm UT range of CE2;
  • pattern recognition (decay of CE2 and CE1 - see Figure 13) drastically improved the decision, but more data are required, namely on piping;
  • if CE2-CE1-SW30 pattern is combined with SW-60 calls for ligament, the rate of false calls for ligament L 3 mm dropped to less than 10%.

Table 4: Results of creeping wave probe for ligament assessment (thickness = 25 mm).

Fig 13: Example of CE1-CE2 OMNISCAN screen captures for different ligament /angle values ; probe frequency 2.25MHz, 3.5 MHz and 5.0 MHz.


Table 5 and Figure 14summerized the notch detection capability of shear waves and creeping waves located in counterbore region, root, misalignment and suck-back.

Fig 14: Detection and sizing capability on stainless steel reference blocks with variable geometry.

Table 5: Summary of ultrasonic results for shear waves and creeping waves in detection and
sizing notches located in test pieces with variable shapes.
Block / target SW Creeping Remarks
ROOT 17 mm /
EDM notch 5 x 2 mm / 37
Detected, sizing problem due to specular andf root Detected from target side; missed from root side 60 shear performed better
SUCK BACK/ 5 x 2 mm Missed from defect side (shadow effect) side Missed from both side Similar TOF and height
MISALIGNMENT 5 x 2 mm Detected and sized Missed from one side Two notches at 37 degrees
CB 27A; CB 35 / long notch 25 mm x 2 mm, vertical Geometric echoes and specular limited detection and sizing Missed in counterbore; missed when weld is in front Creeping waves not performing on uneven surfaces; detected from thinner part

Once the geometric irregularities are present, the POD for detection and sizing of EDM notches decreases. The worst case scenario is the suck-back, when geometric irregularity is equal with defect height. A reliable detection is performed for defect height twice greater than geometric defect. Specular reflection hampers the TED technique, while root, suck back and misalignment (counter bore slope) limited the detection/confirmation of ID EDM notch.

The field welds with uneven surfaces are far away from parallel surfaces (see Figure 15). Crack is located in the counter bore area, with very tight tip, and at 10-15 . Inner surface angles are between 11 to 15 , and as-welded root is in place, with start cracks of 1-2 mm height. Could TOF and creeping waves reliable detect and size these cracks of major concern for the FEA and piping life management in feedwater system?

Fig 15: Example of cracks in counterbore and weld root area in feedwater piping welds; thickness 35-40 mm.

On top of these adversative factors, the outside surface is uneven, with a weld crown. Most of the welds are pipe to T and pipe to elbow/flange, with extra curvature effect and different structure.

Our experiments on retired-for-cause samples concluded:

TOF undersized the cracks by 2-3 mm; last-significant tip is easy to be evaluated only if the inspection is made with the probe on the weld, beam crossing the weld structure. Scanning from component side is not reliable, and cracks are undersized by 3-5 mm.
Creeping waves failed to detect/confirm these cracks due to the factors mentioned above. Sizing the crack to the last significant tip is a challenging task. Only phased array with special set-up and high-frequency probes could produce a reliable database for life-assessment team (see Figure 16).

The application of creeping waves to detect / confirm linear defects in K-welds between carbon steel and stainless steel forging nozzle (see Figure 17) is not possible due to the following factors:

  • shear waves are not penetrated the dendritic structure of the weld
  • target is located too far-away from interface CS-SS
  • re-enforcement weld is obstructing the probe movement towards nozzle
  • target was chosen as 5% t, ratio 5:1;
Phased array L-waves of 3.5 MHz detected and sized the EDM notch using the back-tip (right of Figure 17).

Fig 16: Comparison between optical, magnetic particles, conventional UT and phased-array data for sizing cracks in counterbore-feedwater piping welds.

Fig 17: Limitation of creeping waves applications on dissimilar K-welds (left) and detection /sizing of target by 3.5 MHz L-waves direct-contact linear array probe (right).


Our conclusions are:
  • ligament is reliable assessed by shear waves for value greater than 3 mm;
  • fatigue cracks are reliable sized if their heights is greater than 2.5 mm and their orientation is within ± 15° versus vertical position.
  • absolute arrival time technique (AATT) is more reliable than relative time technique (RATT) for defect orientation between 30°-40°;
  • creeping-wave technique (CE2) inner-surface defect detection/confirmation is very dependent of counter-bore slope and length and suck-back/excessive root penetration
  • weld root echoes and mis-alignment prohibited a reliable detection
  • if the defect is attacked only from one side of the weld, there is little chance for detection away from center-line and there is a large error of under-sizing by TOF due to specular effect;
  • for K-weld (set-in SS nozzle of CS shell) creeping-wave technique (CE2) is not defected elliptical notches on far-side in stainless steel of the dissimilar weld, when inspected from carbon steel side.
  • tip-echo diffraction amplitude from fatigue cracks tips is 6-10 dB over the noise, and 24-28 dB below the amplitude of corner-trap detection signal.
  • comparison between sizing capability on EDM notches and fatigue cracks is misleading; EDM notches as low as 0.6 mm could be sized by high-damped probes, versus fatigue cracks with minimum height of 1.6 mm; signal-to-noise ratio of EDM notch tip is 14-24 dB
  • creeping waves cannot confirm the ID/ligament of inclined defects for L < 6 mm.
  • Further investigation is required to analyze the root-cause of ID-connecting confirmation by creeping waves


  • capability / performance of detection and sizing must be made on representative blocks, very specific to inspection problem;
  • general course of sizing based on RATT/AATT and creeping waves on straight bars is far-away from field reality; simple application of these principles could produce negative results;
  • if TOF and creeping waves are to be applied to piping welds, weld profile and weld crown dressing will help to improve the performance
  • a-priori knowledge of the field weld profile/defect based on RT films may improve the technique
  • it is not recommended to cut corners and use only one-side examination of the weld, even for phased array technology[7]
  • advanced simulation is required to explain the experimental results, namely the contribution of 30 SW on false calls for ligament assessment


The authors wish to thanks the following:
  • OPG-IMS Management for supporting PDI-PPS program and allowing the publication/presentation of the present paper;
  • OPG-IMS technicians and engineers, for their contribution to this program.


  1. EPRI Advanced UT Sizing of IGSSC Cracks Course - Charllote-USA-1988
  2. Panametrics / Mark Davis NDE : Advanced UT Sizing Course - 1994
  3. Ciorau, P., et.al.:" Contribution to detection and sizing fatigue cracks in pipe welds"- Canadian Journal of NDT, part 1: time-of-flight diffraction techniques, vol.19, no.1, Jan. 1997, pp.18-22.
  4. Ciorau, P, et.al." " A contribution to ultrasonic detection and sizing of linear discontinuities in welded joints". Part 2: Mode-converted techniques, Canadian Journal of NDT, vol.19, no.3, pp 8-15.
  5. R/D Tech - Introduction to Phased Array Ultrasonic Technology Applications - pp.76-84-Quebec-Aug.2004
  6. R/D Tech -Phased Array Technical Guideline-Useful Formulas, Graphs, and Examples, pp. 142-152
  7. Ciorau, P. "Contribution to Detection and Sizing Linear Defects by Conventional and Phased Array Ultrasonic Techniques" - 16-th WCNDT-paper 233, Montreal, Aug.-Sept. 2004

© NDT.net |Top|