In extremely complex, high-volume industrial production processes – typified by automotive body shops – maintaining a high build quality is not the only critical concern. A smooth production flow is also of pivotal importance, since any interruptions or rejects are both time-consuming and costly. However, the in-line measurement systems that many manufacturers use in these areas provide much more than a mere OK/Not OK assessment for the manufactured product. Powerful analysis tools yield valuable information regarding process capability for individual features, which is highly valuable for quality assurance personnel and production stability.
One of the main tasks of in-line gaging technology is to provide 100% measurement data – reliably and in real-time – as a basis for statistical process control (SPC). Process capability (expressed by the value cp) is a statistical indicator of particular interest to the quality assurance team, as it provides a measure of the stability of a production process. During the launch phase of a new process and while optimizing running production, this indicator is useful to first identify areas where corrective action would be most effective; and second to verify whether implemented measures have led to a sustained improvement.
Metrology supplier Perceptron has implemented a consistent methodology for process capability evaluation within its reporting software suite. This integrated analysis tool was developed primarily via the close cooperation of the Munich-based company with the quality assurance experts of a major automotive manufacturer in Northern Germany, but it is also relevant to other plants.
Process stabilization at launch
During the launch phase, a typical approach for adjusting production processes is to first minimize process variation and then shift the stable process to meet the design intent. With parts manufactured by multiple suppliers, using temporary processes and materials, production personnel need to know the achieved process stability level, independent of whether specification limits have been exceeded. Car manufacturers have developed different approaches to manage this phase: for instance by adjusting their limit settings or temporarily disabling certain alarms. However, a common denominator is the need for an efficient means to assess production stability, as well as the quality of individual supplied batches.
Automotive companies tend to follow three basic approaches (see chart to right) for handling limits during model launch. If the limits defined for full production are applied during the launch phase, the process is reported as completely out of spec and also mostly out of reject. This process triggers limit alarms constantly, although it is basically stable. Maintaining limits produces results that are already within the design intent tolerance band, even though the process is not stable and indeed shows considerable variation.
Therefore, some plants opt for an approach where the complete tolerance band is temporarily centered on the mean of the actual data, and the tolerances themselves are widened, as necessary. In this way, only extreme outliers cause an alarm during the launch phase. Another approach relies on adjusting only the reject limits, until the process has reached its final position. Depending on the method pursued, this leads to different scenarios for process qualification in the production line. The results for a basically stable process may be perceived as worse than those for a still unstable process, while the latter may in turn be considered mostly okay.
In all these scenarios, it is not apparent whether a) the process is already stable but still too far away from nominal, or b) limits are exceeded because the process is not sufficiently stable yet. In both cases, it is difficult for the in-line team to prove to management how much progress has already been made in terms of process stabilization, and to derive actionable information for their ongoing process optimization efforts. In order to leverage the full potential of the in-line gaging systems and provide user-oriented support towards these two goals, structured process capability reporting has now been added to the SPC software package of the gaging systems.
Structured process evaluation
Starting with a process capability summary for a selected period of time – data typically used for management reporting – the top-to-bottom structure of the software allows the user to drill down to the details within a specific production area. This analysis enables problem areas to be highlighted instantly, providing the means to identify root causes quickly without investing significant time in data evaluation. This is particularly important when most of the production-critical features are already in control and capable. In these circumstances, batches of parts with larger dimensional variation will exhibit a higher percentage of critical process capability values, and therefore be instantly recognizable. A reliable distinction between process variability and shifts from the nominal position may be determined.
In addition, this management summary proves the progress made during a set time and documents continuous improvement of the cp indices. As soon as the required stability level has been reached, the launch team can start working on shifting the stable process to its nominal position.
Ongoing production monitoring
Throughout the life cycle of a car model, those responsible for production require the ability to monitor process reliability and stability using in-line gaging systems. Again, the main tasks of cp reporting are two-fold. As a management reporting tool, it provides an immediate and ongoing means of illustrating process capability, and in turn production quality. For the in-line team, the top-to-bottom structure and ease of navigating the reporting software are its most important assets, especially when sharing this data with the quality assurance team. If the quality indices show a problem, this methodology provides the reliable data needed for effective action within the busy production environment. In addition to this corrective action approach, statistical process capability reporting enables proactive process improvement. The possibility of rejects and associated rework is minimized, as the software highlights potential issues before a limit alarm is triggered.
Proactive process optimization
The automotive plant that supported the development of process monitoring capabilities illustrates the structured approach. The plant monitors cp, especially its development throughout time, in regular intervals. As a starting point, engineers can look at the entire plant, at a group of measurement cells selected according to plant-specific criteria, or at one particular measurement cell. Depending on the monitoring or reporting intervals, shifts, days, weeks, or even months can be combined in one summary chart.
The chart can show, for example, daily cp results throughout two weeks in February 2014. The color red represents cp values of less than 1.00; yellow indicates cp values between 1.00 and 1.33; while the green columns contain cp values that are higher than 1.33, indicating good process capability. In this example, the rightmost bar shows a clear shift towards a larger percentage of yellow and red, which means that process capability deteriorated on Feb. 25. At this point, it is important for the in-line team to isolate the relevant data whose process capability has deteriorated during that day as quickly as possible.
The reporting tool relieves the user of what can be a long and tortuous task. Selection of the right-most bar, provides the user with a color-coded Pareto chart, detailing the cp values for each individual inspection feature that together comprise the data set for Feb. 25.
Next, individual inspection point characteristics can be investigated in more detail. When comparing one inspection point (picked at random) from each of the three cp value categories (see charts above), the meaning of the process capability results becomes clearer. The left-most trend line refers to a green bar with a high cp value. The capable process is stable and close to nominal, and uses only a small segment of the tolerance band. The situation is quite different for the trend chart in the middle, which is associated with a red Pareto bar. Such a low cp value typically means that not only is the variation large, but also – as in this example – that reject or tolerance limits have already been exceeded. Accordingly, it is likely that the line operators have already been alerted to these points by the triggered limit alarms.
Therefore, the in-line team will focus its attention on those red bars belonging to points that have not caused any alarm conditions yet, and above all on the yellow bars. The right-most chart shows the graph for one of these yellow bars. Even though the process exhibits considerable variation in this area and uses almost the complete tolerance band, no line stop has occurred yet. As a result, there is no obvious problem for the line operators, and production still delivers the specified build quantity.
However, the low process stability makes it very likely that a process issue is about to arise in this particular area in the near future. Guenther Fisser, a quality assurance expert at the Volkswagen plant in Emden, Germany, explains why these areas are of particular interest for him and his colleagues.
“The process capability reports show us where proactive corrective action is advisable in order to prevent serious issues, which may even culminate in rejects. Here, the mechanisms of process management work perfectly. We can detect potential risks early on, narrow down the root cause of the problem, and move the process in question back to nominal, before any limits are exceeded,” Fisser says.
Thanks to the continuous process capability monitoring provided by the summary reports, it takes little time or effort to analyze the constant production data to confirm and prove that the processes still meet the stability requirements. Corrective action and optimization efforts in the monitored areas are only necessary if process capability deteriorates.