Quantitative Guidelines for Ultrasonic Inspection of Spot Welds

Spot welds can be difficult to test. This paper describes a study in which 600 welds were inspected using UT. The waveforms were recorded and analyzed in a spreadsheet form using algorithms that raised the "conditionally acceptable" classification rate from 60 to 87 percent G.P. Singh Associate Technical Editor

INTRODUCTION
This article came about as a result of a multi-client project undertaken by the Iowa Demonstration Laboratory (IDL), an outreach division of the Center for Nondestructive Evaluation at Iowa State University. The IDL primarily works with clients in the manufacturing sector, attempting to guide them toward suitable implementation of nondestructive evaluation in their shops. When this interaction is successful, the IDL effectively assists the manufacturer in becoming an educated consumer for nondestructive goods and services.

Many clients have approached the IDL since its inception with similar concerns regarding the nondestructive inspection of spot welds. The clients represented a variety of industries, from automotive assembly to office furniture to household appliance manufacturers. A project was initiated wherein all of these clients submitted samples for study, and various nondestructive techniques were assessed for their ability to discern between good, undersize, and stick (sometimes called "cold") weld conditions. The evaluations were conducted to determine how well a conscientious application of a given technique could discriminate among the various weld conditions.

---

Ultrasonic testing of spot welds has been successfully implemented in many applications.

---

This article presents some of the conclusions we made at the end of this project. Most important, perhaps, is that we present a method of setting up instrumentation in a way that should be easily duplicated by inspectors in different inspection environments. In essence, we feel that we present a quantitative guideline for beginning to perform ultrasonic tests on spot welds.

A note to the reader is warranted here. This work was not intended to be a fully developed research effort, but rather an exercise in technology transfer and industrial assistance. Ultrasonic testing of spot welds has been successfully implemented in many applications. Yet, manufacturers new to using this technique (in our experience) seem to have difficulty in performing successful inspections. With this thought in mind, we pursued a most generic "recipe" for using UT in spot weld applications.

The application of ultrasonic inspection for spot weld evaluation has been discussed in varying detail (Mansour, 1988; Welding Design & Fabrication, July 1988; Welding Design & Fabrication, March 1987). The principle behind this technique lies in discerning weld quality through the interpretation of different ultrasonic signals, or echo patterns, generated on the welds being tested. The ease of applying this technique is a debatable point (Hain, 1988), and proponents of various techniques evaluate the utility of ultrasonic testing for spot weld inspection to varying degrees. As mentioned above, this phase of the current project is not aimed at settling this debate. Rather, the intent was to gather information concerning how successful a conscientious, methodical, application of ultrasonic inspection would be in a broad context.

It was the philosophy of this investigation that what appears to be missing from published literature was useful rules guiding the hands-on application of ultrasonic inspection in the weld shop. The goal was to develop some trial guidelines for using ultrasound, apply them to a diverse group of test welds, and see how well they worked for interpreting weld condition. In this manner, the universality of the projected guidelines could be evaluated. In this sense, the "right answer" was known from the start: testing would likely be successful if applied to clients' samples in an appropriate manner. The object was to determine how much fine-tuning of a generic approach would be necessary until this technique differentiated between the various weld states.

 

TEST PROCEDURES
The testing discussed in this paper was performed at the Iowa Demonstration Laboratory at Iowa State University, using solid delay lines attached to high frequency transducers. This was done to address the issue of using probes that are easier to handle than captive water column transducers, which can sometimes be difficult to assemble properly.

In the current study, three different transducer diameters were used: 3.175 mm (0.125 in.), 5 mm (0.197 in.), and 6.35 mm (0.25 in.). These are conventional transducer diameters available from various manufacturers. Solid delay lines typically come in two sizes, 3.175 mm (0.125 in.) and 6.35 mm (0.25 in.). This means that the intermediate diameter transducer was used with a slightly larger solid delay line attached to it. With respect to transducer hardware, the in-house tests at the IDL evaluated a further variable. Solid delay lines are typically cylindrical in shape, ending in a sharp corner at the plane of contact. This type of delay line was evaluated against similar-sized delays that had either a small chamfer or a radius 0.8 mm (0.032 in.) machined to break that sharp edge.

Another facet of this study was the evaluation of a systematic means of setting the gain level on the ultrasonic equipment. The conventional wisdom guiding the proper choice of settings is somewhat subjective. Essentially, the test operator adjusts the gain setting based on signals from a calibration set of welds. Once confident that the echo patterns from known good, undersize, and stick welds can be distinguished from one another, that gain setting is judged as appropriate. Use of more specific guidelines for obtaining this setting would presumably avoid possible discrepancies between the results of two different operators performing the same test. In addition, it was felt that this could conceivably lead to shorter setup times prior to testing.

Equipment gain settings used in the IDL tests were based on experience gained in a prior project. In that project, it was noted that an optimized gain setting appeared to exist for the most accurate weld characterization. At too high of a gain setting, false rejects were seen to increase; too low of a gain setting resulted in a greater number of false accepts. Further, it seemed that this gain setting could be linked to the appearance of the echo pattern of unwelded material.

It was concluded that the optimum gain setting for weld discrimination occurred at a setting 12 dB higher than the setting that saturated the display for the second backwall reflection of unwelded material. That is, the signal gain was adjusted until the second backwall reflection through unwelded material reached about 100 percent full screen height (FSH). An increase in gain of 12 dB above that level was then used for weld testing. This approach was employed in the current study for all weld testing. While it is the same approach used conceptually by many inspectors in this field, it is possible that this project marked the first time various welds were studied using this as a strict quantitative rule.

Another major aspect of the work performed in this study was assessing a set of simple criteria for evaluating waveforms obtained on the welds. The ultrasonic signals obtained on spot welds have been characterized as falling into as few as 4 and as many as 16 major categories (Mansour, 1988; Welding Design & Fabrication, July 1988; and Hain, 1988). Analysis techniques suggested for interpreting data range from simple pattern recognition on flaw detectors in real time to the use of neural network classification trees to analyze digitized waveforms (Ersil et al., 1988).

The data obtained in this project were evaluated using interpretive techniques which, ideally, will seem straightforward to inspectors. These interpretive techniques require some calibration data to be obtained prior to production testing, but allow relatively easy analysis of echo patterns. This analysis could be performed using electronic gates incorporated into the design of commercial discontinuity detectors, with the aid of a transparent overlay or erasable marking on the detector screen. The analysis technique used in this study focused on certain characteristics that distinguish waveforms obtained on welds of different quality. Typical waveforms for good and undersize welds are shown in Figure 1. The two most important features in evaluating a waveform obtained on a spot weld are the rate of decay of through-weld echoes and the presence of interfacial echoes. In referring to Figure 1, these features are, respectively, the slope of lines drawn through peaks 3, 3', and 3" and the amplitude of any single thickness echoes, shown as 2, 2', and 2".

The significance of these features is as follows. A good weld is assumed to have a nugget of finite thickness; coarse grains in this nugget act to attenuate the signal amplitude more than an undersize nugget. An undersize nugget not only has a smaller diameter, but will also have a smaller thickness to attenuate echoes to a lesser degree. This results in a slower rate of decay or shallower slope of the through-weld peaks.

The presence of interfacial echoes may indicate an undersize weld, with some of the incident sound energy being reflected from the unwelded surface of the single thickness metal. It may also mean that the transducer used was larger than the desired nugget size.

The nearly 600 welds assembled for this study were tested using the solid delay lines mentioned, the "+12 dB" method of setting pulser gain level, and the following method of waveform interpretation:

• Perform a linear fit through the peaks of the through-weld echoes; determine if the good welds exhibit a steeper slope (greater rate of decay) for the through-weld echoes than the undersize weld.
• Look at the presence/amplitude of single thickness echoes; such reflections should be stronger in the stick welds than in the undersize welds, and stronger in the undersize welds than in the good welds.
• If the average amplitude data for the welds of different conditions within a given material combination comply with these rules, then those welds are judged to be readily discriminated one from another in testing.

The above questions were interpreted by applying logic tests to values entered into a data spreadsheet. Test data were acquired using a digital anomaly detector with suitable software for printing out and evaluating waveform features, such as peak value for the various echoes.

 

RESULTS AND DISCUSSION
The results were very encouraging for using methods proposed in this study for setting gain levels for ultrasonic inspection and evaluating the subsequent waveforms produced. The IDL looked at 30 sets of test welds using the methods described. A total of 18 sets of samples were correctly classified using the alternative evaluation methods described here, giving a 60 percent success rate. An additional 8 sets of samples exhibited some behavior that warranted identifying them as "conditionally acceptable" raising the success rate to 87 percent of the welds tested. No obvious characteristic was found to predict which samples would not conform to our evaluation criteria.

The means of determining optimum gain setting for spot weld inspection is submitted here as a quantitative guideline for inspectors to follow. This guideline suggests adjusting the gain on an anomaly detector to saturate the second backwall echo from unwelded base material, and then adding 12 dB additional gain. Our work indicates that this provides the best energy level for subsequent weld identification.

Further, this study looked at simple algorithms for evaluating waveforms obtained from spot welds. Readily interpreted characteristics, such as the through weld peak amplitudes and interfacial reflections, showed good compliance with intuitive rules governing the influence of different weld conditions on ultrasonic response.

Of course, implementation of the approach presented in this report is predicated on acquiring a meaningful set of calibration data. Such data were available only with the cooperation and expertise of the industrial clients who participated in this project.

 

REFERENCES
Ersil, A., Y. Denizhan, A. Oksular, and G. Zora, "Classification Trees Prove Useful in Nondestructive Testing of Spot Weld Quality," Welding Journal, September 1993.

Hain, R. "Resistivity Testing of Spot Welds Challenges Ultrasonics," Welding Journal, May 1988.

Mansour, T.M., "Ultrasonic Inspection of Spot Welds in Thin-Gage Steel," Materials Evaluation, Vol. 46, No. 4, April 1988, pp. 650-658.

"RW Spot Welds Sound Good," Welding Design & Fabrication, Vol. 60, March 1987, pp. 67-68.

"UT Measures Spot-Weld Quality," Welding Design & Fabrication, Vol. 61, July 1988, p 24.

 

* Iowa Demonstration Laboratory, Center for Nondestructive Testing, Iowa State University, Ames, IA 50011; (515) 294-6095; fax (515) 294-6368; e-mail heydave@cnde.iastate.edu.

Copyright © 1998  by the American Society for Nondestructive Testing, Inc. All rights reserved.

[ Materials Evaluation ]

IRRSP, NDT Handbook, The NDT Technician and www.asnt.org are trademarks of the American Society for Nondestructive Testing, Inc. ACCP, ASNT, Level III Study Guide, Materials Evaluation, Nondestructive Testing Handbook, Research in Nondestructive Evaluation and RNDE are registered trademarks of the American Society for Nondestructive Testing, Inc. ASNT exists to create a safer world by promoting the profession and technologies of nondestructive testing.