U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-HRT-04-124
Date: April 2005
Lab & Field Testing of AUT Systems for Steel Highway Bridges
The fundamental role of NDE for the quality control of welding practices is to detect internal defects, determine the severity of any defects, and identify the appropriate defect disposition (i.e., accept or reject). The requirements governing the inspection process are preeminent, and these are provided by the AASHTO/AWS D1.5M/D1.5: 2002 Bridge Welding Code.(1) The following sections provide background on the development of weld inspection and code requirements that influence the detection, evaluation, and disposition of defective welds.
AWS D1.5 WELDING CODE
AWS is a professional association concerned with the fabrication and welding industries. AWS produces and supports a wide range of technical resources regarding welding technology, including a series of consensus specifications regarding welding practices. AWS produced its first welding code in 1928 for fusion welding and gas cutting in building construction. A committee was formed in 1934 to prepare specifications for welding practices for highway and railway bridges. The first bridge specification was published in 1936 and was adopted for use by AASHTO in 1941. Two separate AWS committees, responsible for building and bridge specifications, respectively, were joined in 1963 to form the Structural Welding Committee.(1) In 1969, the first ultrasonic inspection procedures were issued by AWS as an appendix in both the Code for Welding in Building Construction (D1.0-69) and the Specification for Welded Highway and Railway Bridges (D2.0-69).(7) The appendixes were identical in each code, with the exception of the minimum acceptance levels for weld discontinuities, with the bridge code being slightly more restrictive than the building code.(7)
Between 1962 and 1974, a series of Federal Highway Administration (FHWA) circular memorandums required alterations or additions to the existing AWS codes to meet specific needs related to highway bridges. This resulted in AASHTO publishing another welding code in 1974, separate from the AWS code, entitled Standard Specification for Welding of Structural Steel Highway Bridges. Finally, in 1982, a joint committee was formed from AWS and AASHTO representatives to develop a single document to meet the needs of the bridge community. The Bridge Welding Code D1.5(1) is the result of an agreement between AWS and AASHTO to produce a joint document that addresses AASHTO's needs and makes the AASHTO revisions mandatory. This code is similar to the present AWS D1.1: Structural Welding Code--Steel,(8) but has several significant differences. Among these differences are the addition of a fracture control plan for highway bridges and the elimination of sections on statically loaded and tubular structures.
Chapter 6 of the D1.5 code,(1) "Inspection," addresses the requirements for quality control of welding practices for bridges. This chapter is supplemented by requirements in chapter 12, "AASHTO/AWS Fracture Control Plan (FCP) for Nonredundant Members." Commentaries on the requirements of chapter 6 are included in the code. Additional sections of the code that are relevant to chapter 6 include annex VII, which addresses alternate calibration procedures and various inspection forms.
Chapter 6 of the D1.5 code(1) includes provisions for magnetic particle testing (MT), RT, and UT. Section 6.7.1 of the code requires the use of RT for all CJP welds subject to calculated tension or stress reversal. CJP welds in corner joints require UT testing, while either RT or UT is allowed for CJP welds subject to compression or shear. Section 6.7.2 of the code requires MT for fillet welds and partial joint penetration (PJP) welds.
ACCEPTANCE-REJECTION CRITERIA FOR UT VERSUS RT
The AASHTO/AWS D1.5M/D1.5: 2002 Bridge Welding Code(1) includes defect acceptance criteria for both UT and RT. These criteria are an essential part of the effectiveness of the code in ensuring the quality of workmanship. This section will describe the defect acceptance criteria and issues surrounding classifying defects according to the code. For the application of RT, defects appearing on a radiographic film are typically termed discontinuities. For the application of UT, signals resulting from internal reflectors are termed indications. This terminology allows the internal features of welds to be identified without being considered acceptable or rejectable.
For RT, the defect acceptance criterion consists of two parts: First, for a weld to be acceptable, no cracks may be evident on the radiograph. Second, a combination of length, weld size, and spacing is used to determine if volumetric discontinuities are acceptable, and can remain in the weld, or are rejectable, thereby requiring the weld to be repaired.
The acceptance criteria for UT are based on two factors: (1) the amplitude of the ultrasound reflected from the surface of the indication, and (2) the indication length. The amplitude of the waveform is determined in decibels (dB) relative to a standard 1.5-millimeter (mm) (0.06-inch) side-drilled hole in a reference/calibration block. An indication is rated based on the amplitude of the reflected ultrasound and the indication length, and is placed in one of four classes: A, B, C, or D. Figure 1 illustrates the threshold acceptance levels for tension welds as defined in table 6.3 in the AASHTO/AWS D1.5M/D1.5: 2002 Bridge Welding Code.(1) From figure 1, assuming a 25.4-mm- (1-inch-) thick steel plate, a class A indication would have a rating of +8 dB or less. Class A indications are rejectable under all conditions. Class B indications having a rating of +9 dB in the same plate would be rejectable only if the length was greater than 20 millimeters (mm) (0.75 inch). Class C indications having a rating of +10 dB in this plate would be rejectable only if the indication length was greater than 50 mm (2 inches) and the indication was located in the middle half of the weld or the indication length was greater than 20 mm (0.75 inch) and located in the top or bottom quarter of the weld thickness. An indication of +11 dB or greater would be classified as class D and would be acceptable. There are two important issues to be recognized in this figure: First, the scale is such that the smaller the indication rating, the larger the reflection amplitude. Second, the difference between an indication that is acceptable and one that is unacceptable is very small. For example, the difference between an acceptable class D and rejectable class C indication may be as small as 1 dB.
The differences between RT and UT are important to this study. In RT, a crack has a different appearance than a volumetric defect. However, the through-thickness extent of a crack cannot be known from the radiograph alone. The physics behind RT dictate a reduced sensitivity to cracks that lie parallel to the direction of photon travel. As a result, RT has reduced sensitivity to vertical cracks that may be critical defects from a fracture perspective. Based on RT's reduced sensitivity to vertical cracks, UT can produce significant safety advantages.(3)
There are several limitations in the ability of UT to determine the size and shape of defects. These limitations stem from theoretically invalid assumptions made in defining the threshold criteria used to rate defects. The first assumption is that the amplitude of a reflection is directly related to the through-wall extent, or height, of the indication. While the response amplitude is an important factor to be considered, it is affected by many test parameters (e.g., defect type, orientation, surface texture, reflectivity, and size).(2,5) The second assumption is that reflection amplitude alone defines the severity of the defect. In this regard, the code assumes that all indications are cracks (the most severe type of defect from a fracture perspective). The code does not consider that defects may be slag inclusions or porosity, which can be less susceptible to fracture. The third assumption is that a unique relationship exists between the angle of incidence of the ultrasonic beam and the reflection amplitude from a given indication. The code assumes an angle of incidence equal of 90 degrees for reference. Furthermore, the code assumes that as the incidence angle decreases, the reflection amplitude decreases. Therefore, for a 70-degree incidence angle, the reflection amplitude would decrease by 6 decibels (dB) relative to the reference reflection. Similarly, 60-degree and 45-degree incidence angles result in 9-dB and 11 dB reductions, respectively. This results in threshold values for defect categorization that vary according to the incidence angle of the transducer. For example, table 6.3 in the AASHTO/AWS D1.5M/D1.5: 2002 Bridge Welding Code(1) indicates that for plates greater than 38 mm (1.5 inches) through 60 mm (2.5 inches), a class A defect amplitude is +4 dB for a 70-degree angle, +7 dB for a 60 degree angle, and +9 dB for a 45 degree angle. However, the interaction of response amplitude, incidence angle, and defect size is complex. The amplitude of response for a given incidence angle will vary significantly based on the size of the defect, with smaller defects being less sensitive to the incident angle than larger defects.(9)
The scanning pattern prescribed in the code requires articulation of the transducer ±10 degrees from normal to the weld length. This requirement recognizes the sensitivity of the reflection amplitude to the orientation of the defect in a horizontal plane. The articulation motion is intended to address defects that are misaligned (i.e., not parallel to the weld axis, and/or having a maximum response at incidence angles other than perpendicular to the weld axis). Articulation allows the inspector to achieve an improved alignment with a given defect, which can result in significantly larger ultrasonic responses. Research conducted by Serabian(9) indicates that the reduction in amplitude for certain discontinuities skewed as little as 5 degrees can be greater than 40 dB. Recent research conducted by Thavasimuthu, et al.,(10) indicates that for the 2.25-megahertz (MHz) transducer typically used for code inspections, a skew angle of 5 degrees can result in an amplitude reduction greater than 5 dB, while a skew angle of 10 degrees can result in an amplitude reduction greater than 10 dB. Given the relatively small difference between rating an indication as acceptable or rejectable (see figure 1), this effect could have a significant impact on the inspection results. The code provides for an optimum incidence angle to be obtained provided that a given defect is not skewed more than 10 degrees on a horizontal plane; however, it does not generally allow for variations in the inclination of a defect. The exception to this is the evaluation of fusion-type defects, for which the code requires transducer incidence angles be adjusted to maximize reflection amplitude. The available transducer incidence angles are limited to 45 degrees, 60 degrees, and 70 degrees.
Because of response amplitude variability, it is difficult to determine the size of the defect relative to the code. The threshold criteria set out in the code is modeled, to some extent, on the threshold criteria for RT.(2) Generally, the amplitude thresholds are intended to mimic the capabilities of RT by setting threshold values at an assumed percentage of the plate thickness. For example, RT may provide a sensitivity of 2 percent of the material thickness, allowing for larger indications in thicker material. For UT, the code mimics this RT requirement by allowing larger amplitude reflections to be acceptable as the material thickness increases (see figure 1).
Despite some shortcomings in the present code, it has been used with relative success since it was introduced in 1969. A study by Jonas and Scharosch(3) indicated significant variability in the results obtained by applying the code requirements to weld inspection using different operators. The research concluded that volumetric defects were more likely to be detected than planar defects; that vertical cracks as large as 40 percent of the plate thickness could be accepted in nearly 33 percent of the inspections; and that when using three operators, disagreement existed in the acceptance or rejection of 35 percent of the discontinuities examined. Therefore, operator variability was identified as a significant factor in poor UT results. The reduction of operator variability is an important advantage of AUT.