The Five Rules of Effective Leak Testing
Designing a leak-testing system that meets appropriate reliability and reproducibility standards requires a thorough understanding of relevant test methods and technologies. That knowledge base is rarely available in-house.
Penny wise, pound foolish
It is a distressing fact that leak-testing missteps are all too common in the medical device industry. Unlike some industries that traditionally enlist experienced outside applications engineers to develop turnkey test systems, many device manufacturers try to keep this in-house under the misguided notion that they are controlling costs by not outsourcing the project. On the contrary, attempting to design a test system without an encyclopedic knowledge base of test methods and technology typically will add considerable cost to assembly and test operations. These are not trivial mistakes—they can increase production costs by 25%, sometimes more.
Here are five critical factors to bear in mind when developing cost-effective and accurate leak-testing systems. It’s my professional opinion that they are often overlooked by device manufacturers.
1. Leak test specifications should match real-world requirements
There are two ways that leak-testing specifications can be “off”— by being too loose or too tight. Always correctly identify the leak test specs. If they are incorrect, you will pay for it one way or another.
When the specs are too tight—this is widespread across the medical device industry—devices and parts that, in fact, meet requirements will be rejected. Unnecessary rejects add cost in terms of the time and materials used to fabricate them as well as the engineering resources that go into redesigning systems to attenuate defect rates, which could be significantly reduced simply by optimizing the specs. That’s the least of it actually. Leak testers designed according to overly tight specifications will almost always create unnecessary costs. For example, these systems run test cycles that are as much as 40% slower than they need to be.
At the other extreme, poorly understood specs that end up being too loose are the stuff of headlines, resulting in product recalls, lawsuits, negative branding and customer flight. The impact of loosely defined specs is difficult to quantify, but it should compel companies to heed the consequences of product failures. What might not be understood as clearly is that taking precautions to the extreme by setting leak testing specs too tightly also has an impact.
It cannot be emphasized enough that the first step should always be to correctly define leak test specs so that appropriate test methods can be designed. That is the cornerstone of CGMP for any leakproof medical device. If accuracy requirements are not too high and testing can be done at relatively low pressure—at 10 psi for a plastic moulded part, for example—an upstream test usually will suffice. In medical applications where the leakproof requirement is 1 sccm or less or where the part must function at higher pressures, more-accurate downstream test methods involving added tooling and fixturing costs may be necessary. Downstream leak-testing systems that deploy mass flow transducers or other sensing technologies are not exposed to the test part pressure and, therefore, will provide an accurate measurement that is independent of test pressure.
2. Software integration is key
Assemblies built by in-house manufacturing engineers or with the assistance of outside machine builders who are not testing experts, per se, will typically require additional test stands. This expense could have been avoided if the code for high-speed data acquisition had been written correctly and tightly integrated with machine controls. The idea that you can cut customization costs by using generic machine control software and stringing together modules is a widespread misconception. I can think of numerous instances where additional hardware costs could have been avoided by using custom software engineering at about 10% of the cost. For example, a valve that typically takes 50 milliseconds to open with traditional software controls opens in one millisecond or less when best-practices leak testing solutions (data acquisition software tightly integrated to machine controls) are employed.
For low-cost products such as medical check valves and similar applications that require high-speed testing, the problems caused by reliance on generic testing software tend to be magnified. Control engineering programs that allow users to string together several functional units on a screen are not cost-effective for high-speed testing. There is a trade-off in data acquisition speed. While it’s easy to add function upon function to the software, you end up with extremely bloated and sluggish applications.
As incredible as it may seem, some medical device manufacturers are still trying to use Windows XP software or comparable programs created for office environments in their production plants. Real-time operating systems for industrial environments have been the proven superior alternative for more than a decade.
3. Calibrate the system, not just the instruments
Calibration and traceability of leak-test measurements should be simple. Complicated calibration methods, at best, slow down production; at worst, they are unreliable or inaccurate. CGMP guidelines require that leak test instrumentation and system software be able to adequately store data. This includes logging parts by serial number for subsequent analysis.
The most common calibration mistake medical device manufacturers make is assuming that an accurately calibrated leak detector ensures accuracy of the test system as a whole. A perfectly calibrated leak detector is useless if it is hooked up to fixtures that have seal creep or other problems compromising the integrity of the leak testing system. Before being calibrated, instruments should be connected to the test fixtures and to the automated production line where the parts will be tested and assembled.
4. Fixtures must meet leak-test specifications
An assembly and test system with off-the-shelf test fixtures that have been cobbled together typically will not meet gauge repeatability and reproducibility (R&R) requirements. This is especially true for applications with demanding leak testing specs and tight tolerances. Companies that try to get away with spending less on fixtures may, in fact, be adding weeks or months of engineering time as they rework their systems to get the gauge R&R right.
An infusion pump, for example, cannot be leak tested without the use of very precise fixtures that typically have a surface roughness lower than the 16 Ra of common off-the-shelf fixtures. A 2-L lapped finish is not cheap to machine, but if your test specifications require fixturing tolerances of 10% or less, then you probably won’t be able to avoid this expense. Thinking about fixture requirements correctly in the first place will save money and ensure that CGMP-quality standards are consistently achieved.
5. Component traceability is essential for recalls and quality assurance
The basics of cost control in test-centric assemblies (i.e., test-intensive assembly processes, often including leak testing) at various stages of product development are not always well understood.1 Device manufacturers with an eye towards QA and cost controls recognize that parts traceability and logging test data by part number must be an integral part of the leak-testing process.
Traceability data is essential for process optimization to streamline production. Storage of test data at each step of the production process is needed for statistical process control, which underlies most QA systems and ISO 13485:2003 compliance.
There are numerous affordable methods that make traceability painless: bar code labeling systems are commonly used and laser-marking systems sometimes can be added on. Archiving test data for analysis of process improvements will boost yield, minimize defects and ensure that products that may jeopardize patient safety are never released to market.
Ask the experts
The intelligence, dedication and expertise of your in-house engineering team or the contract machine builders designing your assembly chassis may be peerless. But will they be able to achieve the cost controls, accuracy and gauge R&R standards of applications engineers whose sole focus is on testing and who have solved an array of leak-testing challenges? Think of it this way: if you need life-saving neurosurgery, would you select a neurosurgeon who does a dozen or so operations a year or one who does a dozen or so operations a week?
The aforementioned examples of leak testing–related shortcomings and failures in the medical device industry are just the tip of the iceberg. Nothing can replace the knowledge base of testing engineers who design, customize and calibrate leak detectors on a daily basis and who have created hundreds of test-centric assembly systems. Reputable testing applications engineers will provide no-cost evaluations of your test system requirements and make recommendations on test technology required to meet ISO 13485:2003 and CGMP standards at the lowest possible costs to stipulated gauge R&R. Such evaluations are available both for existing assemblies and leak testing systems that are still on the drawing board.