With many leak testing methods to choose from, which is the right one?

A dunk tester inspects an automotive radiator at a radiator repair shop. If it leaks, the bubbles show where and it can be repaired. Source: Stewart Ergonomics Inc.

Leak testing equipment is a type of nondestructive testing equipment used to measure the escape of liquids, vacuum or gases from sealed components or systems. Some configurations require a separate leak detector or sensor as an input. They are often equipped with various other components such as pumps, calibrators, gages and cases. A leak is a hole or porosity in an enclosure capable of passing a fluid from the higher pressure side to the lower pressure side.

There are many basic leak test methods and a few variations. The most familiar are dunk testing, pressure decay, mass flow, mass spectometer and ultrasonics. Dunk testing is still the most popular method, with pressure decay and mass flow rapidly gaining in use. Mass flow is the de facto test method of choice in automotive applications. An exception is pressure decay testing on a brazing fixture.

If there is a situation where a no leak application is required, there is only one choice- mass spectrometer-regardless of cost. This application uses helium and hydrogen gases as part of its inspection process. The good news is the leakage may be so microscopic it is rarely detrimental to product performance.

Dunk Testing

Dunk testing, sometimes called bubble testing, is used for applications that do not require high sensitivity. With dunk testing, the part under test is pressurized, submerged in a liquid-typically water-while the operator looks for bubbles. Bubbles form at the source of the leak as a result of air pressure, and the amount of bubbles per minute can signify the size of the leak. Automotive radiators often are checked for leaks this way. If a leak is present, the bubbles indicate where and the leak can be repaired. Leak testing works best when speed is not a factor. On a production line where test time is critical, leak testing is not the best choice.

While the initial cost of a dunk tank is low, in production it is expensive primarily because the water becomes contaminated, thus producing a hazardous toxic waste requiring special disposal. Tramp oil on the parts, as well as residual brazing flux are the main problems. Tramp and brazing flux from the parts leech into the water causing water contamination and costly special disposal. In one plant, aluminum condensers-AC radiators-are tested on four lines, six tanks total. Each operator has 19 seconds to determine if the bubbles represent air trapped in the fins or an actual leak. The operator is solely responsible for quality control prior to final mass spec testing. Failure of the operator to find a leak of 3.5 standard cubic centimeters per minute (sccm) or greater can shut the mass spec line down for hours. In addition to disposal, maintenance of the water for pH, bacteria and skimming the surface to control skin rashes is a big factor-possibly $30,000 per year for maintenance of each tank. In the above mentioned condenser plant, each 3- by 3- by 12-feet tank is said to cost $30,000 per year to maintain.

This is an example of a familiar process for low-volume applications and repairs but an inappropriate use in high-volume applications. High-speed leak testing in a production line situation hampers the operator's ability to accurately identify bubbles. However, dunk testing can be used on fuel tank filler assemblies and fuel tanks themselves.

One advantage of water dunking is temperature stability. The large volume does not change temperature, which affects most of the more sophisticated testers.

Dry testing a radiator, this pressure decay tester inspects with a test cycle time of 16 seconds at 95 psi. This is the same tester that tested the brazing fixture. Source: Stewart Ergonomics Inc.

Pressure Decay

The pressure decay testing method measures the decrease in pressure in an object. A test object is initially inflated and then a reference pressure is established. After a designated amount of time, the pressure is monitored again, and the initial and final measurements are compared. The change in pressure can be used to calculate the leak rate given the internal volume of the device. Pressure decay is able to detect minute changes in pressure. A drop in pressure signifies a leak; the greater the pressure drop, the larger the leak. This method is convenient in that it is easily automated and dry.

Years ago, the operator drew a grease pencil mark on a gage and came back a while later to see if the gage pointer had moved. If it had, this meant that pressure decayed. Today, electronics monitor to 0.00001 psi and the test is fast-the fastest on small-volume parts. During the past 20 years, electronics have progressed from a 12-bit analog to digital converters, which provided a resolution of 0.0012 to 5 psi. This reduced to 0.0244 psi at 100 psi, far too coarse to be practical.

One way around this problem was the differential pressure tester. Two sensors, having a 5-0-5 psi range, could be set up at 100 psi and provide a resolution of 0.0012, which was usable. Presently, 24-bit resolution is available which provides 16,777,215 steps or 0.0000002 psi at 5 psi and 0.000059 psi at 250 psi. Consequently, differential testing has practically been phased out.

The pressure decay method can be used to test small parts at high speeds. A 0.3-inch diameter by 1.2-inch long medical filter can be tested to see if the filter is in place, if it is plugged and if it allows for correct air flow. Test time from clamp to next clamp is 0.85 second at 15 psi.

Also, a molded vacuum tube connection for the automotive industry is 12 inches of 0.2-inch inner diameter tube with circular molded fitting can be tested. Its test time is 3.5 seconds, clamp-to-clamp at 15 to 30 psi, depending on part. Each test result is permanently recorded and traceable if required.

Communication has become a significant quality tool. Sometimes, only a permanent record of key test parameters and results are sufficient. Other times, a label is printed out and permanently affixed to the part. Still others apply a permanent paint spot or stamp a mark. On certain safety-critical parts, a guillotine destroys a bad part.

Temperature variation is a concern. It is widely believed that pressure testing a gas tank filler assembly while red hot on the braze fixture is impossible. The part in the brazing fixture is approximately 1200 F in the brazing area. Room temperature air fills the part to test pressure, then the fill valve closes. A 30 psi test would normally be considered incorrect at 29.9 psi. However, the pressure is rising, not decaying. The pressure will climb to 34 to 35 psi before starting to decay. It will require at least 15 minutes to reach 29.9 psi and will not be repeatable enough to be practical. The required test time is 20 seconds maximum.

Mass Flow

With the mass flow method, the part is pressurized throughout the test. Any pressure change measured by a pressure sensor is compensated for by inputting air into the test part, therefore exact pressure control is critical. The amount of air entering a part is measured by a flow sensor, directly determining the leak rate of the part. Leakage flow is directed across a heating element. The temperature change across a temperature transducer bridge results in an output voltage proportional to mass flow.

If there is a leak, air will flow into the part. This flow is monitored electronically and processed directly in sccm. Because the flow of air into the part is equal to the leak loss, large-volume parts may be tested with relative ease. The problem with testing large parts is long settle or stabilize time. Because precise pressure control is critical, any oscillation during testing will compromise the test.

Several types of electronic sensors are available. One has a fine tube through which air flows to the part. An electric heater at the input heats the air slightly and that temperature is precisely measured just beyond the heater. The change in temperature is proportional to the flow of air.

Another application measures the milliampere (mA) of current needed to maintain temperature. Ambient incoming air temperature will affect the test. Most production systems are recalibrated every morning, noon and late afternoon. Large parts, such as truck radiators, are commonly placed in a wooden "coffin" to protect against air currents from fans and open doors. The slow test with long-settling time requires protection from thermal changes for repeatability. Mass flow is the process of choice in many automotive applications such as air conditioning condensers, radiators and some fuel lines.

A wand testing system, while not as accurate as placing the part in a vacuum chamber, is superior to any other kind of test, and substantially less expensive than the vacuum chamber system. Source: Galileo TP Inc.

Mass Spectrometer

The mass spectrometer method involves pressurizing the test object with a helium mixture and placing it in a snug-fitting vacuum chamber. The air is then evacuated from the chamber, creating a pressure gradient between the internal volume of the part and the vacuum. The helium molecules move out of the part through any porosity, holes and cracks. A mass spectrometer then samples the air inside the chamber and finds individual atoms of helium.

This is the most sensitive test presently available. These testers are capable of detecting a leak of R600a refrigerant as small as 0.0028 ounce per year. In use, the helium is usually a disposable item adding cost, although recovery systems are now available. Electronic parts often are tested by placing them in a chamber and pressurizing with helium, then placed in the vacuum chamber of the mass spectrometer to see if helium is drawn out. This method also will find porosity.

The initial cost of a snug-fitting chamber is high, requiring precise machining and extremely tight seals, limiting the practical part size. A basic operation limitation is that a large leak-more than 4 sccm-will saturate the vacuum container with helium. This requires several hours of flushing to lower the helium background to a workable level. A pre-test can be employed to weed out the gross leakers. As with dunk testing, a 19-second inspection has proven troublesome and is slowly giving way to pressure decay testing, with or without permanent recording. Even a condenser 16- to 18-inches high, and 24-inches long can be easily tested to 3 sccm in 12 to 15 seconds at 200 psi. Many air conditioner condensers (radiators) are quite large, but they still must not leak for 8 to 10 years. For these applications, a wand testing system has been developed which, while not as accurate as placing the part in a vacuum chamber, is superior to any other kind of test, and substantially less expensive than the vacuum chamber system.


Ultrasonic leak testers, sometimes called sonic, choked flow or turbulent flow, analyze the turbulent flow of a fluid across a pressure boundary that creates acoustic waves. These waves can be transmitted through the medium of the fluid itself, through the containment structure or through the air surrounding the containment structure.

Because gas escaping through small holes generates ultrasonic sound, an array of ultrasonic sensors can be placed around the part. Computer control permits leak detection and, in certain circumstances, exact leak position may also be obtained, similar to dunk testing. Ultrasonic testing has become so widely used to test bearings, gearboxes and general mechanical inspection wear trends over time that it is often overlooked in leak detection. Part size is usually not a problem, nor is thermal variation. Arrays of sensors are used to pinpoint a leak. In some instances, pressure decay testing and mass flow testers have been integrated with ultrasonics to provide increased sensitivity. While this method has not been widely used, it has potential for larger parts.

In the future, leak testing will experience faster cycle times, improved reliability and validation. Leak testing will continue to be integrated into the assembly process and its accuracy and sensitivity will become higher. NDT