7 Steps to Predictive Leak Testing
Leak testing can be a complicated process, particularly for high-speed processes, such as those found in the pharmaceutical industry. The main challenge is the time it takes for the part under test to stabilize under pressure (or in vacuum); as cycle time decreases, the difficulty in performing reliable leak testing increases.
Leak testing can be a complicated process, particularly for high-speed processes, such as those found in the pharmaceutical industry. The main challenge is the time it takes for the part under test to stabilize under pressure (or in vacuum); as cycle time decreases, the difficulty in performing reliable leak testing increases. As a result, for many processes, the leak testing stage becomes the bottleneck. To avoid the excessive equipment costs of multiple leak test stations, predictive testing is increasingly used to reduce the cycle time of leak tests.
Essentially, predictive leak testing takes a leak-rate measurement before the part has completely stabilized and scales that value to a final leak rate from a fully stabilized test. This method of testing can be very effective in reducing cycle time, but has the drawbacks of reduced accuracy and repeatability. However, when care is taken to select the equipment, develop the prediction scheme, and prove its validity, predictive testing can be reliable.
The following is a step-by-step method to assist the process engineer in the definition, validation, and maintenance of a reliable predictive leak test.
1. Identify the need. In many situations, the need for predictive leak testing may be known right from the start. If the part, or one similar to it, has been in production, then it may have been leak tested in the past. In this case, the fill and stabilization characteristics may be known. If not, then estimation and calculation can be performed based on volume to be filled, test pressure, and leak rate limit. Often, a leak test equipment manufacturer or system integrator can run tests or draw comparisons from a knowledge base of applications.
Once the fully stabilized test time has been determined, the cycle time and equipment cost requirements will dictate whether predictive leak testing is required.
2. Fixturing, sealing, volume fill. The physics of pressurizing a part with a test gas does not change by applying predictive testing; therefore, it is important that the part under test is properly fixtured and sealed. Where possible, the volume to be tested should also be minimized by using volume fillers. If not done properly, no amount of prediction will yield a reliable test system. This step, independent of the leak testing equipment, is typically performed by system integrators.
3. Equipment selection. The process or quality engineer responsible for the validity of the leak testing process should ensure that the equipment selected offers the tools to do predictive leak testing and validates that it is accurate and repeatable. The following features should be present:
Online data collection, storage, and display of a large number of leak rate and pressure signatures.
View and overlay multiple signatures—selectable from the history of stored profiles.
Adjustable leak rate and pressure limits: upper and lower limits, which can be arbitrarily complex, either manually entered point by point or based on a master profile. These limits include: specification limits—fail the part when exceeded; gross failure limits—abort the test when exceeded and catch situations such as a damaged sensor, a damaged or missing seal, a missing part, and a grossly leaking part.
Multiple part types—separate limits, sequences, and master profiles to allow for the running of multiple part types; ability to easily add new part types.
Configurable leak test sequencing: define sequences with an unlimited number of steps; define specialized test sequences for calibration, validation, and mastering; ability to change key parameters, such as fill time, stabilization time, test time; perform advanced logic on sensor values, such as fill until the test time >3 seconds and rate of change of pressure is &1 psi/sec; add controls and measurements for valves and sensors.
Predictive analysis tools analyze accuracy of predictive testing by plotting multiple full-length tests and calculating the error between the predicted final leak rate and the actual leak rate.
4. Optimize fill rates and times. Efforts should be made to reduce the fill and stabilization time before applying prediction techniques to the analysis. Predicting the leak rate closer to the point of actually achieving a stabilized leak will improve repeatability.
A potential area of optimization is to decrease the fill and flow dynamics stabilization time by filling the part from multiple fill points. Furthermore, decreasing the volume in the part and in the plumbing connected to the part will reduce stabilization time drastically. This can be achieved by using volume fillers whenever possible and reducing the distance from the test instrument to the part under test.
Another fill technique that reduces fill and stabilization time is to charge the part from a higher-pressure source during the fill phase. The higher-pressure source fills the part faster and the cooling of the high-pressure air reduces the temperature rise in the part under test as it is being pressurized.
5. Run parts with varying leak rates; do a preliminary signature analysis. Take a spectrum of parts, from a perfect part to a part that leaks about 20% greater than the leak rate limit, including one part that has a leak value equal to the test limit. Also include a grossly leaking part. After determining adequate fill times for the parts, perform leak tests to ensure that the post-fill stabilization and test phases are long enough that each part fully stabilizes.
It is important to characterize parts with varying leak rates. If a collection of such parts doesn't exist, consider adding a variable orifice to a known good part and running the modified part over and over, while varying the orifice size.
6. Run repeated, prolonged tests; pick a cutoff time; set limits. Using overlaid curves of the various leak rates, draw a vertical line at the required cycle time to examine the data at the point in the test when the tester will be making a decision. Assessing repeatability of various leak rate curves at this point of prediction is important.
7. Continuous verification and maintenance. Running a verification part at the beginning of every shift or once a day will ensure that the predictive leak test system is functioning correctly. Running a master part with a known leak rate through the tester will verify that the measured leak rate is within a required tolerance. This verification procedure can be set up in the leak test system so that it will automatically respond with confirmation that the test system is operating within specifications.
Routine maintenance of a leak test system will ensure it continues to operate reliably. Regular maintenance should include cleaning sealing surfaces, replacing worn seals, performing internal checks of the leak test instrument to ensure it is leak free, verifying the entire system is leak free, and repairing any leaks.
Keys to success
The two keys to successful leak testing in general, and predictive leak testing in particular, are equipment selection and knowledge. With the appropriate tools, the process engineer and the quality engineer can initially develop, and continuously validate, a repeatable and trustworthy predictive leak test.
In the end, the process and quality engineer are responsible for confirming and ensuring the initial and continued validity of all test equipment. Regardless of whether equipment is purchased directly or supplied by an integrator, proper selection of leak test equipment will make this validation possible.
Ben Zimmer is a software developer and Roydyn Clayton is a project engineering in training at Meikle Automation;
Leak testing methods
Linear scaling of the final leak rate value: This method uses a single scaling factor to scale the measured leak rate so that at the end of the abbreviated test, the value is the same as a full-length test. When the test is run, only the final scaled value is compared to the leak limits. The limitation of this method is that it ignores the shape of the leak rate and pressure curves until the end of the test. Part or process anomalies may cause the test to end up at a valid leak rate, while in fact there was a problem.
Non-linear scaling: Over a given range of leak rates, there is not always a linear correlation between abbreviated leak rate values and fully stabilized leak rate values. Consequently, using linear scaling to predict the final leak rate can lead to inaccuracies, which, for those parts which take the longest to fill and stabilize, get worse as the test gets shorter. The advantage to this method is that non-linear scaling considers multiple correlation points to more accurately predict the final leak rate, where linear scaling takes just two points into account.
Continuous verification against a trained curve: It is important to detect anomalies in the part or the process causing irregularly shaped pressure and leak rate curves. To achieve this, the signals must be monitored continuously throughout the test and checked against pre-set limits. It is important to point out that a test system can be set up to reliably pass or fail a part based purely on these limits. When an actual leak rate value is required, one of the above two predictive methods should also be employed.