Leak testing is often an art as much as it is a science. Manufacturing quality engineers and machine operators must contend with many external factors that can undermine accuracy and repeatability. On today’s factory floor, digital process signature (or waveform) analysis, coupled with powerful applications for big data analysis and visualization, can be used to eliminate much of the uncertainty and guesswork that plagued leak testing in the past.
The waveform, or digital process signature, of a manufacturing process or test cycle is like a complete video replay of a crucial event as it happens, versus just a few snapshots.
Take a hockey game with your favorite team on the ice. A contentious goal is scored and the referees go upstairs for a review of the play to decide if the goal will count. Now, what would you prefer as the basis for that decision – a video instant replay or snapshots of only a few isolated points in time as the puck passed through the goalie’s crease?
A digital process signature is that full replay, while scalar data offers only a few snapshots.
By collecting the signature from every test cycle into a central database, you now have an archive, traceable by serial number, that can be used to drive the reliability and repeatability of your leak test. Below are a few ways to lever this data.
Signatures can be converted into a histogram of leak rates to show the waveform for a good part and the range of acceptable deviation. This makes it easy to create and visualize a baseline against which to compare all parts. The more signatures you have, the easier it becomes to understand what waveform anomalies to watch for and what they signify. In addition to distinguishing good parts from bad, you can also spot problems with the test station itself.
Mapping involves taking a few known good parts and running them through the test station 20 times each to create a baseline. Specific defects, flaws or even test station setup problems are then introduced to see if and how they show up in the resulting test data. This takes the guesswork out of determining which limits to set to catch specific issues.
Apply mathematical techniques directly to the process signature in the form of a “feature.” Applying post-processing features such as Slope, Peak to Peak values, Mean and Standard Deviation will help to quantify sections of the signature for further analysis.
If a part comes back from the field with a leak-related warranty issue, you can draw on your archive of leak test data to trace root cause. Use the warranty claim as an opportunity to drive continuous improvement.
Check to see if there was anything off about the problem part’s original leak test signature, even within the range of standard deviation. And look at the signature data from the other processes upstream that touched the part – sometimes, a problem just doesn’t show up at the leak test station. Run the problem part through the test again to see how its signature compares to its original test. Lastly, create an algorithm to screen your archive of signature data for that type of part to see if any others have the same anomaly.
We explore these and other best practices to boost the performance of your leak test in our e-book, “7 practical steps to a better manufacturing leak test.” What you will learn in this e-book: