DEFECT CHARACTERISATION

Jim Houf

This article describes common defects found in angle beam UT weld inspections along
with examples of the transducer positions and resulting screen presentations related to those defects. Common false indications and the means of determining them are also discussed. The following terms are often used interchangeably and sometimes incorrectly to describe ultrasonic testing signals. Relevant and Nonrelevant Indications. A signal seen on the UT screen that results from sound reflecting from an internal reflector within the part being inspected or caused by a physical condition such as geometry is called an indication. A relevant indication is created by a physical reflector in the part. A nonrelevant indication is caused by part geometry or a physical condition other than an actual discontinuity or defect. Discontinuity. A discontinuity is any foreign material in the part (slag, porosity, etc. or an unintended disruption in the weld or base material such as a crac!, incomplete penetration or incomplete fusion. "f the discontinuity type and si#e are within the acceptable limits of the criteria of the go$erning documents, they may be acceptable. %mall amounts of slag or porosity may be acceptable depending on their si#e. Howe$er, certain discontinuity types (crac!s, incomplete penetration or lac! of fusion are often re&ectable regardless of si#e. Defect. A discontinuity that is re&ectable is a defect. To determine that a discontinuity is indeed a defect according to the criteria of the go$erning code or specification, the UT operator must first !now the type of discontinuity causing the indication, accurately locate it with respect to a !nown reference point on the weld, and determine its si#e.

Cracks
Crac!s are usually linear and irregular or &agged. "n normal indications, the damping in the transducer causes the right side of the signal to drop off 'uic!ly. Howe$er, because a crac! is &agged, multiple returns result, causing the screen signal to be much wider thus producing a distincti$e indication on the UT screen referred to as a (church steeple( ()ig. * . Another effecti$e way to determine that an indication is a crac! is in the way in which a crac! indication (wal!s( across the screen ()ig. + . As the sound beam begins to mo$e across the crac!, the inherent beam spread causes the leading edge of the sound cone to reflect sound bac! well before the signal is maximi#ed. This causes a short or low amplitude signal to appear on the far right side of the screen (signal A . As the transducer gets closer to the crac!, more of the sound beam reflects bac! causing the

signal height to increase and mo$e toward the left side of the screen as the sound path gets shorter. ,hen the centerline of the sound beam, where the sound strength is greatest, reaches the base of the crac!, the signal is usually at maximum strength (signal - . As the centerline clears the crac! and the trailing edge mo$es past the crac!, the signal height drops and it mo$es toward the left of the screen until it disappears (signal C . "n effect, we see a wide short signal that grows in height as it mo$es towards the left and then diminishes in height until it disappears. "f the width of the UT screen is set at one full s!ip distance, the crac! should start to appear &ust to the right of the midpoint and then drop off the screen near the 'uarter point of the screen between the main bang and the midpoint ()ig. + . "f the crac! starts from the scanning surface, the signal should wal! in from the far right of the screen and drop off &ust to the right of the midpoint. These are generali#ations and the exact location will depend on the material thic!ness and will $ary due to the different sound paths. -ecause of this, it is possible for the operator to o$erloo! a shallow same.side crac! in the first leg if the signal is so close to the scanning surface that it appears to be part of the main bang. Incom lete !enetration An incomplete penetration ("/ signal tends to beha$e much li!e the signal for a crac!, tra$eling from right to left on the screen0 increasing and then decreasing in height. Howe$er, because "/ is generally smooth.sided (the unfused edge of the weld groo$e rather than &agged li!e a crac!, the width of the screen signal is generally much narrower than that of a crac! signal. A problem that occurs fre'uently is that since the "/ is at the root of the weld, the screen signal may not ha$e been fully maximi#ed when the transducer hits the near edge of the weld crown. "n thinner materials, this problem is li!ely to occur. %canning from both sides of the weld may not catch this configuration. "f the signal continues to increase in height until the transducer hits the weld crown, it may be that the signal, which has not reached re&ectable si#e, is indeed re&ectable. "f this occurs, a smaller transducer may be needed to get close enough to fully maximi#e the signal. )ailure to do so may cause an operator to accept a re&ectable defect.

Side"all #ack of Fusion

Sidewall lack of fusion (12) is a particularly hard defect to identify. Common be$el angles in weld &oints do not often complement the angles used for ultrasonic testing. As a result, the orientation of the unfused sidewall of the weld may not return a signal amplitude large enough to be re&ectable. Additionally, sidewall 12) may not show at all in the first leg when the weld is scanned from the side where the defect is located, or may show as an acceptable indication when seen in the second leg. Thus, the weld must be inspected from both sides when the part configuration permits. "f signals occur that are determined to be along the side of the weld, e$en if they do appear acceptable, it may be necessary to switch wedge angles and interrogate the area again to determine whether or not sidewall 12) is present.

Sla$ 3uring the welding process, powdered or granular flux is melted to pro$ide shielding gasses for the welding arc. As the residue cools, the solids form a glassli!e substance called slag which may be trapped in the weld creating a slag inclusion. An o$erall irregular shape is a clue that a slag inclusion is the indication being seen. -ecause the cone.shaped sound beam hits the irregularly shaped slag at se$eral points at once, the signal tends to gi$e a presentation with multiple pea!s ()ig. 4 . As the transducer mo$es forward, the centerline of the beam hits the $arious contours with more strength, causing the height or amplitude of each signal to increase and then drop off as the sound beam crosses each reflection point. As a result, the multiple pea!s of the signal will go up and down as well as mo$e slightly to the right or left as the scan progresses. This signal with multiple pea!s alternating up and down is typical of slag inclusions. The alternating pea!s can be effecti$ely demonstrated by stopping the transducer when a multipea! signal is maximi#ed then angulating the transducer bac! and forth sideways at that position.

!orosity ,eld porosity forms in a manner similar to slag, when gas is trapped in the cooling weld metal before it can escape to the surface of the weld. Howe$er, rather than ha$ing solids in the inclusion and being irregularly shaped, pores are usually gas.filled and rounded in shape. Reflectors. A flat (planar surface perpendicular to the sound beam will reflect a large amount of sound resulting in a $ery high screen signal. A side.drilled hole is a line reflector, reflecting sound from the circumference only where the sound beam stri!es the hole perpendicular to the circumference with a length e'ual to the depth of the hole. As a result, the signal from a side.drilled hole is smaller in screen amplitude (height than a planar reflector. A rounded reflector, such as a gas pore, is a point source and sound only reflects from it at a single point where the sound beam stri!es the hole perpendicular to the circumference. -ecause a pore has no appreciable length and the diameter is usually a $ery small percentage of the cross.sectional area of the sound beam, the signal from a pore returns a $ery small amount of sound, resulting in a $ery discrete screen signal often of low amplitude. The effect of the sound beam reflecting from that circular shape gi$es the screen signal for porosities distincti$e characteristics. ,hen the sound beam first hits a circular pore, it does so at a slight angle, causing the sound to reflect away from the transducer ()ig. 5a . As the centerline of the sound beam hits the pore directly, sound is reflected bac! to the transducer ()ig. 5b . Howe$er, as the centerline passes o$er the pore, sound in the lower portion of the sound cone is again reflected away from the transducer ()ig. 5c . The o$erall result is a single, narrow, sharp signal that pops up on the screen at only one location and then disappears immediately. There are some exceptions to this0 if the pore is large or is close to the scanning surface, the signal may mo$e slightly to the left, usually * to + minor graticules, before disappearing. Again, the pore is a point source and once the signal is maximi#ed, the operator will notice that a slight mo$ement of the transducer forward or bac!ward, or oscillation of the transducer slightly to the right or left, will cause the signal to disappear. "n the case of cluster porosity or closely aligned pores, the screen display may demonstrate se$eral signals $ery close together that can be confused with a slag inclusion. Howe$er, it is often possible to isolate the indi$idual signals, which will show the $ery tight or narrow trace and the location of each signal at one spot without significant lateral mo$ement, which generally does not occur with a slag inclusion. Nonrelevant and False Indications As mentioned earlier, nonrele$ant indications are those that will not affect the use of the part or are created by geometry or other physical characteristics. %ome of the more

common indications in this category are bac!ing bar indications, mode con$ersion and signals from sources outside of the weld. %ackin$ %ars. Backing bars are commonly used in butt welds on structural steel. Their purpose is to pro$ide a surface under the weld groo$e on which to lay the first bead or root pass of the weld. A properly welded bac!ed butt &oint will result in full penetration between the sidewalls of the weld and the bac!ing bar. As a result, sound may enter the bac!ing bar through this weld &unction, ricochet around in the bac!ing bar, reflect from one of the corners and then return bac! to the transducer. )igure 6a shows this &oint configuration and how the sound beam may reflect bac! from a corner. The signal from the bac!ing bar reflector shown will appear &ust a bit into the second leg and may be misidentified as a sidewall discontinuity because of the apparent location of the signal on the screen ()ig. 6b . "f the bac!ing bar is accessible, this type of false indication can usually be damped with a finger wet with couplant as shown. &ode Conversion. Mode conversion results when a shear wa$e reflects from a surface that causes the wa$e to con$ert to a longitudinal wa$e. This occurs occasionally on bac!ed welds when the shear wa$e hits the gap created by a poorly fit up bac!ing bar, con$erts to a longitudinal wa$e and tra$els up to the weld crown ()ig. 7 . The 1. wa$e then returns to the root, con$erts bac! to a shear wa$e and returns to the transducer. -ecause the 1.wa$e $elocity is approximately twice that of a shear wa$e, the resulting signal appears about halfway out in the second leg. The weld crown is a good reflector and the signal is $ery strong, often exceeding *88 percent of full screen height. The location and height of the mode con$ersion signal easily identify this type of false indication. )irst, the height of the signal is extremely strong for a discontinuity at the apparent location. %econdly, if the sound path is calculated correctly, the operator can see that the apparent location is outside of the weld and heat.affected #one. ,hen this occurs, damping the surface of the weld with a wet finger shows that the sound has con$erted and can be damped on the weld crown. 9ode con$ersion li!e this can also occur in open root pipe welds and double : plate welds if the root or opposite side weld crowns are a shape that permit it.

E'traneous Reflectors
2ccasionally, screen signals show up where no signal should logically be. An example can be shown in beam.column connections. "f a gusset plate has been welded between the column flanges to carry the structural load across the column and the plate is in line with the upper and lower beam flanges, the gusset fillet welds may reflect sound bac! to the transducer as shown in transducer positions A and - in )ig. ;. These signals can be confusing because they appear to be outside the weld area. ,hen these occur, it is necessary to $isually inspect the part to see if another part or weld may be causing the errant signal.

Another nonrele$ant indication may occur in beam.column connections if cope holes ha$e been cut in the web to permit access to the full width of the beam flange. This signal usually appears strong in the first leg when the transducer is near the center of the beam (o$er the web but is too far away from the weld to be a rele$ant indication as in transducer position C in )ig. ;. "f this occurs, the operator should wet a finger with couplant and damp along the edges of the cope hole to confirm that the signal is coming from the hole. "t should be noted that the discontinuity descriptions discussed in this article are general in nature and the actual discontinuities in production welds may $ary depending on wedge angle, material thic!ness, weld configuration and discontinuity si#e. Howe$er, if operators are familiar with the general screen presentations for different discontinuity types, they are better e'uipped to categori#e a weld defect or correctly identify false indications that would be costly if misidentified and called out as defects. T<T

=Jim Houf is %enior 9anager of A%<T>s Technical %er$ices 3epartment and administers all A%<T certification programs. (?88 +++.+;7? @+*+, (7*5 +;5.7?AA fax, B&houfCasnt.orgD.