The ultrasonic method for spot-weld inspection detects multiple reflections from the backwall and any intermediate interfaces of the welded structure. Extracting various parameters from these echo sequences, such as the extent of the echo sequence, the signal attenuation, and the amplitude and position of the intermediate echoes, allows the differentiation between good and defective spot-welds and the classification of defective spots. Initial work with plain steel, showed a clear distinction between good and defective welds: good welds generated essentially no intermediate echoes. This distinction does not hold for more advanced materials that include thick corrosion resistance coatings and high-tensile steels, where intermediate echoes could be present even in "good" welds.
We have studied these phenomena and developed a classification tool for spot-welds on high-tensile steels or thickly coated plates. We describe the classification methods and report on the favorable results of a series of qualification tests performed to evaluate the performance of these tools in an industrial environment.
Keywords: NDT inspection, ultrasonic evaluation, spotweld
Spot welding is a mechanical bonding method, which lends itself to high production volumes. For example it is used extensively in the automotive industry; there are between 3,000 and 5,000 spot welds in a typical car.
Figure 1 shows a simplified diagram of the welding process and its parameters. In many factories the entire welding process is automated with welding robots. Nevertheless, the erosion of the welding tips, malfunction, setup errors, the quality of the electrical contact and variation in surface quality introduce some uncertainty of the quality of the resulting welds. In the process of inspecting the welds it is common to perform sample testing of products on the production line. Any defective welds found initiate a corrective processes. In the short term a defective welding station is attended to, and any parts that it may have produced between tested samples are mended. In the longer term the statistical information obtained from the sample tests is fed back to improve the welding setup and the production processes.
Fig 1: The spotweld welding process|
2. Motivation For Ultrasonic Spot Weld Inspection
The integrity of spot welds plays a critical role in the reliability of cars. Three methods are currently used for inspecting spot welds:
- The 'chisel and hammer' method. This is a quasi-nondestructive method that has a limited application - it does not provide relevant results for three-layer components and for coated materials, and cannot be used for painted parts. The method is time consuming and the tested part often requires rework or must be discarded altogether.
- Destructive inspection. The destructive method is capable of detecting almost all of the defective flaws, but is extremely expensive - both in terms of labor and manpower, and the loss of value-added inventory.
- The ultrasonic (UT) method. A non-destructive technique, and if automated is fast, accurate, and easy to use. It is also facilitates the generation of test data statistics - an important feature for the feedback to improve the spot-welding process in a plant.
The ultrasonic method for spot weld inspection is based on the detection of multiple reflections from the backwall of the welded structure, together with intermediate echoes, if any, generated by the interface between plates or any flaws that might be present. The weld can be classified from measurements of the waveform parameters: the signal attenuation (amplitude drop), and the amplitude, number and position of the intermediate echoes. For example, Figures 2a and 2b show a metallurgical section of a defective weld and its ultrasonic signature.
Fig 2a: Three-sheet welded structure showing a cold weld condition between the second and third plate.
Fig 2b: Ultrasonic signature of the defective weld shows signals reflected from the second plate backwall.|
3. The SpotWeld Inspector
We have developed and implemented the SpotWeld Inspector which is a dedicated, automated ultrasonic instrument for spotweld testing. Figure 2 shows a UT test station with the Spotweld Inspector as deployed in a car production environment.
The Spotweld Inspector stores a pre-programmed test sequence that covers the spot-welds in a certain tested part and guides the operator as to spot testing sequence. For each spot the program provides a suitable parameter setup. The user then applies the test transducer to the selected spot and once the amplitude of detected builds up, it is automatically captured and analyzed. The system also provides for various configurations: for example the system can be setup to display defect classification, or merely indicate that the spot is defective.
Fig 3: ScanMaster's Spotweld Inspector in the production line|
Figures 3 and 4 show typical signals that are captured for a good and defective spot-welds. In addition to the waveform the screen indicates the number of the spot tested in the part (top right corner of the display), the detected thickness of the tested structure (overall thickness if the weld is "Good", top plate thickness in the extreme of "No Weld"), and various other parameters of the detected ultrasound signature. For the operator's convenience, the reflections of the backwall are marked on screen with green crosses, and intermediate echoes with red crosses.
Fig 4: A "Good" Spot waveform as automatically classified by the Spotweld Inspector.|
Fig 5: A "Bad" Spot waveform automatically classified by the Spotweld Inspector as "Cladding".
The system also offers a database for storing all the parameters detected for each tested part. This database is invaluable for generating statistics on the defects found in the parts tested with a variety of sections: for example all defects found in a specific weld over a given period.
4. Characteristics of UT Signals from Spot-Welds
The following lists the main weld parameters that are considered in the UT method.
- Type of material
- Type of coating (if applicable)
- Number of sheets (plates)
- Thickness of each sheet
In general the UT method can classify the weld defect into the following main categories. The characteristics and significance of each class is outlined in the following for standard carbon steel plates:
- No weld
- Undersized weld (small nugget)
- Thin weld
- Cladding or "stick" weld
- Flaws (porosity) in weld nugget
- Burnt spot
- Mixed defects (e.g. small nugget and stick weld)
"Good" Spot Weld
- Successively attenuated echo signals at intervals commensurate with the combined plate thickness.
- The attenuation of the successive echoes derives from the larger grain found in the weld nugget as compared to the parent material.
- The intermediate signals are, typically less than 10% screen height when the first backwall echo is set to 80% of screen height.
Undersized Spot Weld (small nugget)
- In setting up the test the ultrasonic test beam diameter is matched to the diameter of the weld; therefore undersized weld nuggets are characterized by partial reflections from the bond interface.
- The envelope of these intermediate echoes exhibits a maximum; this differentiates a small spot from a stick weld or porosity.
"Stick" or "Cladding" Spot Weld
- The "Stick" or "Cladding" type spot weld has a nugget that does not extend through the thickness of the welded structure.
- The lower attenuation of a cladding weld (as compared to a good weld) results in less signal attenuation.
- Often the bond interface echo is also evident.
"Thin" Spot Weld
- Incorrect current, contact time and pressure cause thin spot welds.
- The appearance of a characteristically "Thin" spot is similar to that of a good weld, often confusing even experienced operators.
- The thin weld characteristic is determined by measuring the total thickness of the welded structure from the intervals between repeated echoes from the combined plate thickness.
Pores in Spot
- The A-scan signal consists of echoes generated by the back wall of the part and intermediate echoes generated by pores or flaws in the spot.
- The sequence of intermediate echoes shows a decreasing amplitude and practically vanishes after several cycles.
- The A-scan signal generated by a burnt spot is characterized by irregular backwall echoes of low amplitude.
- The most obvious feature of no weld is the repeated backwall echo separation that corresponds to the thickness of the top plate.
- Attenuation of the bond interface echo is a characteristic of the grain structure of the parent plate material; it is unmodified due to absence of a weld nugget with its larger grain structure.
5. Results For Thickly Coated Material
The above data is common to the UT method as developed in the 60's, using analog instruments. When applied to material with thick coating, some of the basic assumptions break down - specifically the fact that Good Welds do not exhibit large intermediate echoes. When using plates with thick coating the intermediate echoes could be significant even for Good Welds. This requires the modification of the classification algorithm to account for this effect. Here the algorithm is bifurcated into classification of welds of equal thickness and unequal thickness plates. For the latter case, the location of the intermediate echoes differs significantly between "Cladding" and Good Welds. Whereas in Cladding Welds the intermediate echoes are located off-center of the backwall reflections, the intermediate echoes for Good and Undersized (Small) Welds are centered between multiple backwall reflection. The differentiation between Small and Good welds is more elaborate and required careful calibration and analysis of the relative amplitudes of the central reflections to those of the backwall multiples.
fig 7shows the situation of a Good Weld for thickly coated steel. In this case the intermediate echoes are large presumably due to the formation of and Cladding welds in deeply coated unequal plates. As noted above the location of the intermediate echoes is distinct between the two classes.
Fig 7: A "Good" Spot waveform automatically classified by the Spotweld Inspector - note the intermediate echoes that are taken into account by the advanced algorithm for thick corrosion resistant coatings.
6. Test results
Our modified algorithms that account for thickly coated material perform well on coated steel and steel alloys. The classification algorithm was qualified with a series of welded samples that were torn down to verify the UT classification. The system detected Good Welds with a certainty exceeding 99%. The classification of Bad Welds is better than 95% and can be further improved by fine-tuning the parameters for automatic decision making.
The UT method developed in the 60's for analog instruments required visual inspection; decisions were taken by the operator and depended largely on his expertise and concentration. These limitations, especially the difficulty of interpretation and the consequential non-reliability, rendered the UT method impractical.
The advent of the digital ultrasonic instruments provided the possibility of computer-assisted decision making with on-line UT signal analysis. The UT method is currently seeing wide application in the automotive industry and is proving the most appropriate method for spot weld inspection, both in the production line and in the lab. In addition to its non-destructive quality, the automated UT method offers high throughput, high reliability of detection and full documentation capability. With recent advances in the classification methods we are able to demonstrate high-reliability testing on modern material, including thickly coated steel and alloys.