January 2009

Factory Automation

Seeing is believing

Spot welders use ultrasonic for high-strength steel measurements


  • Auto industry seeks spot-weld testing alternative for light, high-strength steel.
  • Welding institute teams with vision package for answers.
  • Nondestructive evaluation method uses ultrasonic with vision module.
By Ellen Fussell Policastro

As if the automotive industry did not have enough problems these days, rising energy costs, stringent emission regulations, and higher safety standards are also driving the industry to demand lighter-weight automobiles with better strength. That is why automakers are exploring the use of higher-strength steels, which have better mechanical properties at a reduced weight when compared to the conventional steels for automotive construction.

Yet traditional resistance spot weld destructive test methods, such as pry-bar or hammer and chisel, for tearing spot welds and measuring weld nugget size, are not the safest or the best tests, especially with advanced high-strength steel (AHSS). They have the potential to initiate cracks in the weld when applied to AHSS.


As an alternative for measuring weld quality, Edison Welding Institute (EWI) redesigned a probe to use ultrasonic waves. Similar to medical ultrasound, EWI relied on imaging the spot weld through data collected by the ultrasound. "We're looking at the fusion line between the two sheets of metal once they're spot welded together," said Roger Spencer, applications engineer at EWI in Columbus, Ohio. "We're transmitting sound energy through the spot weld, trying to determine if the weld is actually fused or not," he said.

EWI is using a vision software development tool to create a software routine that allows them to measure the spot weld nugget sizes without destroying the part. "We are measuring the size to determine if it's a good weld or not. That way they don't have to tear the weld apart to find that out," Spencer said.

Spencer and his team developed the routine and probe as a basic concept, and from there they procured one auto nondestructive evaluation manufacturer so far who is interested in pursuing it further. "They're funding a separate project to look at their specific weld. They see it as a benefit because it's easier to use and more reliable," Spencer said.

"We've looked at it for determining the width of a laser weld where you have two sheets of metal lapped together. But you can also use it in marine applications, where they're melding thin sheets of metal together either by spot weld or seam weld."

Spot weld sees use when two electrodes come together to create a spot (round weld). Seam weld is melding with a laser from outside, "where you would laser weld through one metal into another. It's more of a linear type weld. It would be joining two sheets together. In some cases, they want to know the width of that weld."

Testing probe

The company used a new custom-designed two dimensional (2D) matrix phased array ultrasonic testing (PAUT) probe and a commercially available PAUT system to develop a custom application with a graphical programming language vision development module. The module is really a library or set of algorithms for machine vision that nondestructively determines the quality of the AHSS resistance spot welds in real time. With this package, the developer can easily connect to the PAUT system while acquiring and analyzing the PAUT data.

The new nondestructive evaluation method resolves the issue of quality determination of AHSS resistance spot welds and reduces, or eliminates, any required user interpretation. The ultimate goal of the proposed nondestructive evaluation method is to use it on the shop floor to examine the quality of resistance spot welds immediately after robotic welders complete them and determine whether they need additional welds. The current solution is designed for advanced users and engineers to review the settings for configuring the system; thus it offers a flexibility that may not be required on the shop floor.

System hardware, software

The nondestructive evaluation method proposes using a custom-designed, 2D matrix phase array probe with specific frequency responses and pitches that are suitable for the thicknesses of the material. It also has an off-the-shelf phase array system with remote control and data access to the ActiveX libraries.

"The vision algorithms give you access to high-level and low-level machine vision functions, which includes things like filters, convolution, and thresholding," said Matt Slaughter, product engineer at National Instruments in Austin, Tex. Filtering sees use with an electronic signal (such as voltage coming in) when you want to filter out noise. "You can use a low-pass or high-pass filter in basic electronic signals. You can also do those filters on an image. If there's a lot of noise, you can pass through the filters to clean it up," Slaughter said. "It's just like you do filtering on an electronic signal; you'd do the same on an image." Convolution is another algorithm to clean up the image. You can also set a threshold with the image as well. "If, say, the intensity of a certain point is low or high enough, and you want it pulled out of the image to be used for processing later, you can disregard everything you're not concerned with and keep everything else," he said.

At EWI, "they've actually connected this for use in a nonconventional way," Slaughter said. Most users of the vision library will connect up to an actual camera with firewire (camera interface standard IEEE 1394) or Ethernet. "EWI chose to use something outside those standards. They're actually just using the library on images they're creating with a special tool," Slaughter. "Using a 2D matrix phased array ultrasonic test, the company is hooking up to an ultrasonic sensor and getting an image back and doing machine vision processing."

The custom software application uses the graphical programming module and a vision development module to analyze the data from the PAUT system in real time. With the connectivity of the programming module, the developer can easily connect to the PAUT system while acquiring and analyzing the PAUT data as they acquire it.

Some specific goals in developing the custom algorithms included:

  • Artificial enhancement of data resolution and dynamic gating of PAUT data for reduced processing time
  • Image processing routines for binary image processing, border generation, and particle examination
  • Measurement functions such as the sizing of welds both in area and long/short axis distance, amplitude thresholding, sizing deviation from nominal value, and final weld-quality determination (pass/fail determination).

This application can also see use as an offline analysis tool to review all previously saved data files. With the flexible user interface built into the application, advanced users and engineers can verify the proper settings for the system to make further adjustments to improve accuracy.

With the user interface, raw and processed data are both displayed on the front screen as well as the final determination of quality (a green indicator for a good weld and red for a bad weld). You can also export the analysis data into an Excel-compatible text file format.

Verification tests indicated a more than 95% confidence level for reporting good/bad weld determination, together with sizing from a group of more than 200 sample welds. The team tested all samples for destructive test verification, and only two welds were found to deviate from the quality determination. More information showed the two welds were significantly distorted such that valid measurements were difficult to reach meaningful quality determination.

The team at EWI sees other applications with the proposed nondestructive evaluation method on shop floors where spot welds are made in automotive spot welding applications-as a quality assurance tool or a feedback mechanism to determine weld quality immediately after making welds.

Other industry uses

The graphical library is also seeing use in the food industry, using color cameras to check out fruit to see if there is any bruising on apples or if pears are the right color, Slaughter said. One group is using the library to look at pecans coming into their factory to see if husks have come off correctly. If not, they push them back into air jets to get the husks off, he said.

The packaging industry also benefits from using the library by being able to verify if the right number of a product was put into the box or even the right product.  "For those you don't even need color cameras, you can use standard black and white to see if the text is right or whether the label is on straight," Slaughter said.

The ultrasonic testing used in automotive for spot weld checking can also see use in other functions with this same library. Reading text and reading 1D and 2D barcodes is also handy in the automotive industry, Slaughter said. "Almost every component of your car has a small 2D code printed on it. So to verify you're putting the right part in the right place, you read codes. You can track those codes throughout the factory. Someone can use machine vision to do that. The same library the people at EWI are using to check these spot welds, others can use the same library to read the codes to verify the right pieces are going in the right place, as well as verifying nuts are put on bolts correctly."


Ellen Fussell Policastro is associate editor at InTech. Email her at efussellpolicastro@isa.org.

Ultrasonic sense

Ultrasonic sensors are designed for contactless and wear-free detection of a variety of targets by means of sonic waves. It is not important whether the target is transparent or colored, metallic or non-metallic, firm, liquid, or powdery. Environmental conditions such as spray, dust, or rain seldom affect their function.

Ultrasonic sensors are mainly used in the diffuse mode. An object in front of the sensor is detected by its reflection of a part of the emitted sound wave. It is also possible to use ultrasonic sensors in the opposed and reflective mode. An ultrasonic opposed mode sensor consists of an emitter and a receiver that "listen" to each other permanently. The ultrasonic sound is interrupted by an object between the emitter and receiver, and the sensor generates a switching signal.

With almost all ultrasonic sensors, it is possible to adjust the lower and the upper limit of the switching or measuring range. Objects outside this range may be detected, but they do not initiate the output to change state.

Among other factors such as the wave length, the accuracy of ultrasonic sensors is mainly limited by speed fluctuations of the sound during temperature changes.


Ultrasonic transmitters send a sound wave from a piezoelectric transducer to the contents of the vessel. The device measures the length of time it takes for the reflected sound wave to return to the transducer. A successful measurement depends on reflection from the process material in a straight line back to the transducer. Ultrasonic's appeal is the transducer does not come in contact with the process material and does not contain any moving parts.

Today's ultrasonic devices typically require no calibration and provide high accuracy level measurements in both liquid and solids applications. However, excessive process temperatures and pressure can be a limiting factor. And since ultrasonic technology is based on a traveling sound pressure wave, a constant velocity via its media (air) is required to assure a high degree of accuracy. Material such as dust, heavy vapors, surface turbulence, foam, and even ambient noise can affect the returning signal. Because sound travels at a constant known velocity at a given temperature, the time between the transmit burst and detection for the return echo will be proportional to the distance between the sensor and the reflecting object.

SOURCES: Tektron, an Irish company specializing in explosion-protected instrumentation for the pharmaceutical, chemical, and offshore industries (www.tektron.ie); and ISA Handbook of Measurement Equations and Tables, 2nd Edition, edited by Jim Strothman.