|NDT.net - July 2001, Vol. 6 No. 7|
To avoid unnecessary girth weld repairs during pipeline construction accurate flaw sizing becomes increasingly important. This paper discusses the inspection and evaluation of Gas Metal Arc Girth Welds with mechanized ultrasonic inspection using alternative girth weld acceptance criteria.
Flaw sizing techniques have been applied from national standards that make use of signal amplitude to determine flaw size and signal signature to establish through wall interaction of vertically adjacent weld flaws.1.1 Abbreviations and Key words
|GMAW||Gas Metal Arc Welding|
|Flaw||A weld discontinuity|
|Stacked flaws||Flaws lying above each other along the same weld bevel side|
|FSH||Full Screen Height|
|Stacked A scan presentation||Ultrasonic data presentation using color to image amplitudes|
|TOFD||Time of Flight Diffraction|
|LCP||Lack of Cross Penetration|
|ECA||Engineering Critical Assessment|
|Over-trace||The detection of planar flaws in adjacent inspection zones|
|Signal Signature||The echo dynamic trace traditionally viewed on a flaw detector|
|Focussed Probe||An ultrasonic probe which has a finely controlled beam size in its working range|
|Standard Probe||A ultrasonic probe with no specific modifications to the beam size|
The majority of flaws created during Gas Metal Arc Welding (GMAW) are planar in nature and oriented on the original weld bevel profile. Such flaws can be readily identified using ultrasonic inspection, Figure 1. The relative size of such flaws can be estimated by comparison of the reflected ultrasonic signal to that produced from flat bottomed calibration holes machined in the same orientation as the weld bevel.
|Fig 1: Weld Bevel Design.|
In the early 1980's mechanized ultrasonic systems were developed to inspect girths of production rates of up to100 welds per day. The first systems used conventional ultrasonic transducers to provide full coverage of the weld volume. Results were promising but better discrimination of flaws within the weld bevel was a concern.
NOVA, a gas transmission company in Canada and RTD BV of the Netherlands, developed the concept of using focussed transducers oriented to inspect discreet areas of the weld. Transit time measurement was introduced to locate the flaw within a specific "time gate" which could be directly related to a given location within the weld volume, Figure 2. This eliminated the occasional false calls where weld geometry reflections were misinterpreted as weld flaws.
|Fig 2: Geometry and Transit Time.|
Progressive developments of ultrasonic inspection shifted to the detection of non-planar volumetric flaws within the body of the weld. Historically flaws such as porosity have been difficult to detect and discriminate with ultrasonic inspection. A technique known as a "stacked A scan" or "raw data" presentation was developed to identify such flaws, Figure 3. More recently Time of Flight Diffraction (TOFD) has been applied to augment focussed probe inspection.
|Fig 3: Stacked 'A' Scan Showing Porosity & Geometry.|
Nowadays the TOFD technique, Figure 4 is used in mechanized ultrasonic systems on pipelines as a safety net. It assists the operator to detect the presence of center line cracks, mis-oriented flaws and, in the case of amplitude sizing, it gives information as to the severity of the flaw being sized. Additionally, it acts as a confirmation tool for flaws detected with "A" scan.
A = Lateral wave, B = top tip diffraction of defect, C = bottom tip diffraction of defect, D = Compression wave reflection of inside pipe,
E = Shear wave reflection of inside pipe|
Fig 4: The Time Of Flight Diffraction Technique.
The focussed probes inspection technique that became known as "zonal inspection". An array of focussed probes was organized in a manner to inspect all the zones of a GMAW weld in one inspection pass. Generally speaking the inspection zones mirrored the weld passes needed to complete the weld. A typical probe array is shown in Figure 5.
|Fig 5: Example probe array.|
Zonal inspection allows the application of alternative flaw acceptance criteria based on an Engineering Critical Assessment (ECA). An ECA enables the ability of a girth weld to accommodate flaws to be determined based on the strength and toughness of the weld together with a thorough knowledge of the loads that will be applied. This technique relies on an inspection method that can size, or at least categorize weld flaws into various size groups. Zonal inspection allows flaws to be categorized based on the size of the probe's focal spot. A typical inspection zone maybe in the order of 3mm with a corresponding focal spot size of 2 mm.
With zonal inspection a flaw is deemed to exist if it produces an ultrasonic signal above a pre-determined threshold amplitude e.g. 40% Full Screen Height (FSH). Once above this threshold amplitude, the flaw is conservatively assumed to be the full inspection zone in height. The circumferential length of the flaw, while producing a signal above the threshold amplitude is measured. The decision to repair the weld is made by comparison of this circumferential length with the calculated acceptable length from the ECA calculations. If transducers from two adjacent inspection zones find a flaw in the same circumferential location, the flaw has generally been considered to be a stacked flaw and assessed as having a through wall height of both inspection zones.
The conventional zonal inspection approach has three shortcomings:
Therefore it is possible that flaws be smaller than sized by the mechanized ultrasonic technique. A further complication is the reporting of "stacked" flaws. Stacked flawss are those that occur in two consecutive weld passes or inspection zones but in the same circumferential location. The acceptable length of such flaws is always significantly less than single zone flaws so a weld repair is generally required. Seldom are stacked flaws confirmed after making cross sections of the area although the ultrasonic signals obtained from the inspection probes involved clearly suggest that the flaw in question does indeed lie in two inspection zones.
This paper reviews the ultrasonic development work that was carried out in Canada in preparation for the Alliance Pipeline project by team members from the Alliance Pipeline Ltd. technical group and Weldsonix International Inc. The work has been aimed at providing a flaw sizing technique that measures flaw height more accurately but is nevertheless based on long established national standards. The concept of amplitude sizing has been used in the nuclear, fabrication, primary forging and foundry industries in Europe for approximately 30 years. This work also introduces the concept of signal signature to characterize a flaw. This characterization has been used to distinguish between stacked and interzonal flaws.
The work was based on the following principles:
The above methods of flaw sizing does not apply to volumetric flaws.3.2 Experimental Program
Traditionally the calibration of an ultrasonic system is based on responses from flat-bottomed holes (FBH). A review of typical flaws in girth welds indicated that a flat-bottomed notches (FBN) with heights of 1, 2 or 3mm and a length greater than the beam width would provide a much more realistic representation of weld flaws. Reference blocks were manufactured that contained both holes and notches so that a comparison of the ultrasonic responses could be made.
Three calibration blocks were manufactured using NPS 42 x 16mm (0.56") WT project pipe from the Alliance Pipeline system. Flat bottom holes and notches were prepared using electro-discharge machining. The calibration reflectors were based on the compound J bevel preparation used by the American CRC-Evans and Canadian RMS mechanized welding technique, Figure 1. Details of the reference block arrangements are given below.
|GMAW Weld preparation with typical reference reflector positions.|
|Schematic view GMAW weld preparation with typical reference reflector positions.|
The following responses were obtained from flat-bottomed holes and flat-bottomed notches using an 80% Full Screen Height reference response established on a 2-mm FBH. Responses were the same using both pulse-echo and pitch-catch techniques:
|1 mm fbh||1 mm notch||2 mm fbh||2 mm notch||3 mm fbh||3 mm notch|
This first test summarized above produced three responses in excess of 100% screen height. A second scan was thus carried out after lowering the calibration response by 12 dB:
|1 mm fbh||1 mm notch||2 mm fbh||2 mm notch||3 mm fbh||3 mm notch|
These results suggest that notches provide much higher responses than traditional flat-bottomed holes.4.1 Calculating response between flat-bottomed reflectors.
Ultrasonic responses can be calculated for FBH's of various diameters as follows:
|0.5 mm fbh||1 mm fbh||1.5 mm fbh||2 mm fbh||3 mm fbh||4 mm fbh|
|-24 dB||-12 dB||-5 dB||0 dB||7 dB||12 dB|
Doubling the diameter results in 12 dB increase in ultrasonic response. These responses can be compared to the values obtained from flat bottomed notches with notch heights comparable to flat bottomed hole diameters:
|0.5 mm notch||1 mm notch||1.5 mm notch||2 mm notch||3 mm notch||4 mm notch|
|-6 dB||0 dB||3 dB||6 dB||9 dB||12 dB|
From these results it can be seen that a 2mm FBH and a 1 mm flat bottomed notch would provide the same response at a 0 dB reference level.
Applying these calculated responses to the range of flaw heights found in gas metal arc welds
results in the amplitude - flaw height relationship detailed in
|Amplitude range||Flaw height||Amplitude range||Flaw height|
|0-5%||0.2 mm||50%-55%||2.6 mm|
|5%-10%||0.5 mm||55%-60%||2.8 mm|
|10%-15%||0.7 mm||60%-65%||3.0 mm|
|15%-20%||1.0 mm||65%-70%||3.3 mm|
|25%-30%||1.5 mm||75%-80%||4.0 mm|
|30%-35%||1.7 mm||80%-85%||4.2 mm|
|35%-40%||2.0 mm||85%-90%||4.4 mm|
|40%-45%||2.2 mm||90%-95%||4.6 mm|
|45%-50%||2.4 mm||95%-100%||4.8 mm|
|Table 1: Amplitude flaw height relationship based on calculated amplitude response.|
Two ultrasonic Inspection procedures were evaluated to determine the best through wall coverage available with both "focussed" and "standard" inspection probes. The aim of procedure development was to produce a quantifiable and homogenous ultrasonic coverage throughout the weld thickness to guarantee the ability to detect and quantify flaws regardless of position on weld bevel. Details of the two probe arrays identified as "A" and "B" are given in Table 2. The probes used for the work reflect current production inspection probes from Krautkramer used by Weldsonix. The standard probes have a focal spot size of approximately 3 mm compared to the focussed probes that have a focal spot size of approximately 2 mm. Inspection angles were obtained by mounting the probes on PolypencoTM wedge material of the required angle.
|Inspection Zone||Probe Array "A"||Probe Array "B"|
|Fill 3, Cap||55° 5MHz, 3/8" standard probe operating in pulse/echo mode.||55° 5MHz, 3/8" standard probe operating in a pulse/echo mode|
|Fill 2||45 & 55°5MHz 3/8" standard probe operating in a pitch/catch mode||45 & 55°5MHz 1/2" focussed probe operating in a pitch/catch mode|
|Fill 1||45 & 55°5MHz 3/8" standard probe operating in a pitch/catch mode||45 & 55°5MHz 1/2" focussed probes operating in a pitch/catch mode|
|Hot pass 2||50° 5MHz 3/8" standard probe operating in a pulse/echo mode||50°5MHz 3/8" standard probe operating in pulse/echo mode|
|Hot pass 1||50°5MHz 3/8" standard probe operating in a pulse/echo mode||50°5MHz 3/8" standard probe operating in a pulse/echo mode|
|Table 2: Probe array details for the two inspection procedures evaluated in the study.|
The two probe arrays, A & B, were calibrated on 2mm Flat Bottomed Holes in order to provide an 80% full screen height (FSH) response. The focused probes and standard probes were then compared to determine the over-trace amplitude available in adjacent inspection zones.
Scanning the reference blocks with the two probe arrays produced significant results. The "standard" probes produced 40% FSH (-6dB) over-trace in the adjacent zones in the same plane. However the "focussed" probes only produced 20% FSH (-12dB) over-trace under the same conditions.
The reference block was scanned with the two probe arrays. The standard probes provided 80% FSH for the reflectors located in between inspection zones with focused probes providing 55% FSH under the same conditions. On the basis of these results it was determine that the standard probes and not the focussed probes provided the necessary homogeneous coverage.5.1 Applying the Ultrasonic Inspection Procedure to the Alliance Project Pipe.
|Pass Description||Nominal Height, mm|
|Root Pass (including land)||3*|
|Fill Pass 1, 2 & 3||2.8*, 2.8*, 2.8*|
|Table 3: Maximum through wall height of weld passes as measured on the weld bevel to produce
a nominal 14.2 mm wall thickness weld.|
In order to achieve this, the relationship between the flaw height and ultrasonic signal response presented in Table 1 was used for four groups. The groups reflect the flaw height categories used in the ECA calculations.
Traditionally mechanized ultrasonic inspection has had difficulties differentiating between "stacked" flaws that occur in two adjacent inspection zones and flaws that merely occur between two weld passes. Flaws in two inspection zones are detected by two or more transducers resulting in the conclusion that such flaws have occurred in two sequential weld passes and are therefore large. In reality such flaws tend to occur at the weld pass interface and are not nearly as large as the inspection result suggests.
|Fig 6: Differentiating between stacked and inter-zonal flaws.|
The next challenge was to develop a methodology that enabled such flaws to be differentiated so unnecessary repairs could be avoided. The solution was to use the amplitude "signature" or "echo dynamic". Every ultrasonic reflection has a unique signature that can be viewed on an "A" scan presentation. However when the reflection is above 100% Full Screen Height this signature is often lost, Figure 6.
Consequently a logarithmic amplifier was incorporated into the system to allow high amplitude reflected data to be re-processed. Amplitudes over 100% Full Screen Height can be re-processed without re-inspecting the weld. This re-processing allows the operator to determine the shape or "signature" of the reflected signal regardless of the initial amplitude. Because the ultrasonic signal from a given flaw is unique, an ultrasonic technician can determine if a flaw is truly stacked by examining and comparing the response signatures from the two transducers involved. If the signatures are the same the flaw is simply being picked up by both transducers and can be regarded as an inter-zonal flaw. Flaws having different signatures can be regarded as being stacked. This technique provides an accurate differentiation between such flaws especially when the amplitude of the flaws is also taken into account.A set of provisional "rules" was developed to enable field technicians to quickly categorize flaws into the four size groups. This requires the field technician to carry out six discrete flaw evaluation steps aided with computer software:
Common examples of this sizing methodology are given in Table 4.
|Flaw Type||Step 1||Step 2||Step 3||Step 4||Step 5||Step 6|
|Amplitude Response||Flaw size||Amplitude - 12 dB||Calculated % FSH||Same Signature||Flaw size MM|
|Zone 1||Zone 2||Zone 3||Zone 1||Zone 2||Zone 3||Zone 1||Zone 2||Zone 3|
|LOF||0-40%||~40%||0-40%||~0.5 mm||~0.5 mm|
|LOF||0-40%||< 100%||0-40%||< 1.5 mm||0-40%||~120%||0-40%||n/a||0.5 - 1.5|
|LOF||0-40%||> 100%||0-40%||_||0-10%||~30%||0-10%||0-40%||~120%||0-40%||n/a||0.5 - 1.5|
|LOF||0-40%||> 100%||> 100%||_||0-10%||~30%||~30%||0-40%||~120%||~120%||yes||0.5 - 1.5|
|LOF||0-40%||> 100%||> 100%||_||0-10%||~30%||~30%||0-40%||~120%||~120%||no||1.5 - 3.0|
|LOF||0-40%||> 100%||0-40%||_||0-10%||~65%||0-10%||0-40%||~260%||0-40%||n/a||1.5 - 3.0|
|LOF||> 100%||> 100%||0-40%||_||~65%||~65%||0-10%||~260%||~260%||0-40%||yes||1.5 - 3.0|
|LOF||> 100%||> 100%||0-40%||_||~75%||~65%||0-10%||~300%||~260%||0-40%||yes or no||3.0 - 6.0|
|LOF||> 100%||> 100%||0-40%||_||~65%||~65%||0-10%||~260%||~260%||0-40%||no||3.0 - 6.0|
|LOF||> 100%||> 100%||> 100%||_||~65%||~65%||~65%||~260%||~260%||~260%||yes or no||>6.0|
|Table 4: Categorization instructions for ultrasonic operators.|
Two mechanized GMA welds were made on project pipe having a 36" (914mm) diameter and 0.56" (14.2mm) wall thickness using the CRC-Evans welding process. During the welding process various lack of side-wall fusion flaws were created in the fill passes by adjusting the weld tracking during welding. Poor weld tracking is a common cause of zonal and inter-zonal lack of side wall fusion flaws and is representative of common flaws that occur during welding. In total nine suitable flaws were created.
The nine flaws created in the two seeded test welds were sized using the six step sizing method discussed in section 6 and additionally along with the traditional zonal sizing method.
Following the scanning, ultrasonic evaluation and sizing, the flaws were removed from the seeded girth welds. Circumferential sections containing the flaw plus 10mm of sound weld on either side were cut out. The sections were notched around the location of the flaw and broken open after cooling below the ductile-brittle transition temperature by immersion in liquid nitrogen. This technique presented the full length, shape and height of the flaws created during welding.
This cold notched breaking technique was adopted after consultation with the University of Ghent, Ref 8, which determined that multiple cross sectioning of flawed welds can result in an incomplete and inaccurate assessment of flaw shape. Details of the nine flaws from the two welds are presented in Appendix A to J [Pdf of 390 KB]. Inspection results for the two welds are given in table 5.
|Flaw #||Weld # 2|
|Circumference location in mm (12 Oclock),||1438-1495||1443-2900||1928-1950|
|Max height using amplitude sizing, mm||3.0-6.0||3.0-6.0||3.0-6.0|
|Max height using zonal sizing, mm||5.6||8.4||7.6|
|Maximum measured height, mm||~6.4||~ 5.8||~3.5|
|Flaw position||Fill 1&2||Fill 1&2||Between Fill 1&2|
|Flaw #||Weld # 8|
|Circumference location in mm (12 Oclock),||2620-2730||2640-2750||2165-2226|
|Max height using amplitude sizing, mm||2.2||2.6||3.0-6.0|
|Max height using zonal sizing, mm||2.8||5.6||5.6|
|Maximum measured height, mm||not visible||~ 2.8||~4.3|
|Flaw position||Fill 1||Fill 1||Between Fill 1&2|
|Circumference location in mm (12 Oclock),||1670-1740||1115-1240||309-560|
|Max height using amplitude sizing, mm||2.6||2.4||3.0-6.0|
|Max height using zonal sizing, mm||5.6||5.6||5.6|
|Maximum measured height, mm||~ 2.7||~ 2.3||~3.4|
|Flaw position||Fill 1||Fill 2||Between Fill 1&2|
|Table 5: Ultrasonic and measured Inspection Results from the two test welds|
A Review of the fracture surfaces of the nine planar flaws in Appendix A to J [Pdf of 390 KB] shows that they are irregular in shape but have a much longer length than height. Hence a calibration notch as opposed to a calibration hole would appear to provide a better basis for reference calibration.
The results of the two scans indicate that zonal sizing tends to over-estimate the weld flaws. This is apparent on the zonal and inter-zonal flaws and is a result of the zonal sizing technique itself and the irregular nature of planar flaws. Over-trace amplitudes above the threshold level are regarded as occupying the full inspection zone in the zonal sizing technique and result in the conservative assessment. Raising the threshold level would eliminate such over-tracing but risk missing the inter-zonal flaws that produce relatively small ultrasonic amplitudes.
Amplitude sizing on the other hand provides a much more accurate determination of flaw height. There are two results however that require further explanation; flaw 1 and flaw 7. In these two examples the actual measured height of the flaw is slightly larger than the determined height with amplitude sizing. At first sight this would appear to break the cardinal rule requiring the ultrasonic determination to be conservative. A review of the actual fracture surfaces in Appendix A to J [Pdf of 390 KB] reveals that the maximum height occurs for an extremely short distance and is a result of the irregular nature of the flaw. This dimension then is not the critical one that should be used in determining the acceptable length of such flaws. Indeed the failure mode of all current GMAW girth welds would be plastic collapse as opposed to brittle fracture. Under such circumstances the area or the average height of the flaw would be a more accurate assessment of critical dimensions. Based on the average heights of flaws 1 and 7 it can be seen that amplitude sizing does still provide a conservative assessment.
Amplitude sizing and signal signature analysis allows the technician to determine the difference between inter-zonal flaws and stacked flaws. The use of standard non-focussed transducers provides homogeneous ultrasound coverage throughout the weld wall thickness. Specifically an 80% FSH response is available for the detection of inter-zonal flaws. This high response level assists the field technician in assessing signal signature of inter-zonal flaws coupled with the ability to reduce signal response by 12 dB for those in excess of 100% Full Screen Height. Signal signatures regardless of initial amplitude can now be interpreted and used to determine whether flaws are inter-zonal or clearly take up two zones and need to be treated as such. The sizing accuracy of such stacked flaws is generally grouped in the 3 to 6mm category. Further refinement of sizing ability is clearly available but becomes increasingly time consuming for the field technician who is required to inspect and evaluate up to 130 welds per day on a typical Canadian mechanized mainline welding gang.
Sizing and location of flaws in girth welds can be enhanced by:
Currently the three mainline welding contractors constructing the Alliance Pipeline System in Canada are enjoying the lowest repair rate (under 1%) and the highest production rates (up to 290 welds/day) of any land based Mechanized Gas Metal Arc pipeline construction project constructed to date.
In some cases up to 900 consecutive w elds have been made without one rejectable defect. At least 80% of the flaws detected in 600 km of pipeline constructed to date have been between 0.5mm and 3.0mm in height. Stacked defect calls have virtually been eliminated and the pipeline contractors are now in agreement with the dimensions of flaws reported.
|© NDT.net - firstname.lastname@example.org|||Top||