Improving the bottom line in automotive applications

Total Cost of Quality

Most people think only about the costs associated with internal failures, such as rejects or scrap, and external failures, such as shipped product that needs to be repaired, scrapped, or returned. Other costs associated with quality that can be incurred include appraisal and prevention. These are the areas in which companies can spend money and work smarter, not harder. Money spent wisely upfront can save much more later (see Figure 1).

Figure 1 Appraisal and prevention are proactive areas in which money spent wisely upfront can save much more later.

Prevention of Weld Defects

Companies can take many proactive steps to help prevent weld defects before any welding of parts begins:

1. Make sure the detail parts (stamped, formed, cast, extruded) have an acceptable process capability, for example, CpK >1.67, which means all critical dimensions and features are repeatable.
2. Design and build fixturing to ensure joint location is repeatable and gaps are within required process tolerances.
3. Make sure the complete robot system has been designed and built to be robust and repeatable.
4. Establish and validate welding procedures during the prototype stage so all performance criteria are met. Weld sequence and location are two critical areas.
5. Make sure the welding process is made as robust as possible to handle the normal factory floor variations associated with employees, equipment, and material.
6. Verify that employees are trained properly to operate automated welding equipment so high performance can be maintained.
7. Use poke-yoke (errorproofing) to eliminate unusual occurrences such as loose clamps causing parts to move and gaps to form.
8. Use the robot or programmable logic controller to count individual weld segments or the time an arc is on to try to ensure all the welds are made and that they are long enough.

Figure 2 A Pareto chart should be the result of an effective visual inspection program.

Appraisal of Welds

The next place to look for savings is in the inspection area. Ineffective inspection leads either to bad parts passing through the system or good parts being improperly marked as bad and repaired. The nature of high-volume production means that only a few techniques are practical enough to evaluate quality correctly in the factory.

Weld Cross Sectioning. This evaluation technique can be used several ways. The first way is as a first-piece-setup method, in which a few parts are welded, their cross sections are checked, and if the quality is acceptable, the welding proceeds. The second method is go/no-go, in which a batch of welded parts may be quarantined if they are not acceptable. A third weld cross sectioning method is overall trending over time. For a typical automotive truck chassis, the customer requires 30 to 250 cross sections to be taken at a frequency ranging from one to five times per week.

Visual Inspection. This technique can be quite effective if it's done correctly. The ideal situation is if the robot operator is a certified welding inspector who can immediately make a process change in response to a problem. Unfortunately, the reality in most plants is that the inspection is done at the end of the line by unskilled workers who cannot make an immediate process change. A Pareto chart should be the result of an effective visual inspection program (see Figure 2).

Figure 3 Laser vision technology now is being used to inspect and evaluate the surface quality of a finished weld.

Arc Data Monitors (ADM). ADM programs have been implemented in several plants for the purpose of both accept/reject and trending. These monitors typically measure current and voltage and reject welds based on real values exceeding set tolerances. ADMs can work well for gross defects such as too much porosity or burn-through.

ADMs also can catch smaller defects if an effective design of experiments (DOE) has been run, in which defects are purposely created and correlated to the arc data monitoring signals. Once a strong correlation has been proven, the feedback will be fairly reliable. However, the biggest problem in real-world usage has been that this upfront experimental work was not done.

Laser Vision Technology. A newer technology available to inspect welds and monitor the process involves laser vision technology. The laser vision camera itself is not new -it has been used for many years for seam tracking-but what is new is the ability to use it for evaluating the surface quality of a finished weld (see Figure 3).

A typical laser vision system either can be mounted right behind the welding head or it can be placed in a station after the welding is completed. This system can provide simple accept/reject results with a summary of what defects have occurred in a Pareto chart. In addition, it can produce detailed statistical process control (SPC) analysis of each individual defect. The consistency achieved with laser vision technology can reduce the percentage of false positives or missed negatives, reducing expensive repairs on perfectly good parts.

Figure 4 To help make sure a repaired weld is within specification, an operator can use a hand-held laser vision gauge to check the reworked area.

Internal Failure

Once a part has been rejected for unacceptable welding, it should be determined if it can be repaired or if it needs to be scrapped. Repairs typically are done manually and then inspected a second time. The report from an automatic inspection system will tell the technician assigned to do the repair exactly where the defects are located and how bad they are.

To ensure a proper repair is done, approved procedures must be used, with qualified welders doing the work. To help make sure the repaired weld is in fact now within specification, the operator can use a hand-held laser vision gauge (see Figure 4) to check the reworked area.

External Failure

When customers find parts with defective welds on their assembly lines, the parts can be extremely expensive to correct. Charges of thousands of dollars to the supplier are not unusual. While it is true that defective welded parts sometimes make it through a supplier's quality system, other charges come from disputes about whether a welded part really meets the applicable standard. Human visual inspection is variable, which can lead to disagreements.

Using an objective calibrated system such as a hand-held laser gauge can eliminate this subjectivity, so that everyone can agree on what is good and bad. The best way to establish an acceptance standard is by developing workmanship samples representing typical defective welds. Measurements should be recorded with the laser vision sensor and also by the cross-section method to establish a good correlation. This will be helpful in future discussions about what automatic surface inspection tells about the overall weld quality.