Leak Detection Methods:A Comparative Study of Technologies and TechniquesShort version

Table of Contents1Introduction . 32Leak testing methods . 32.1Water immersion bubble test method . 42.2Soap solution bubble test . 62.3Pressure decay test . 62.4Vacuum decay test or Pressure rise test . 82.5Tracer gas leak testing . 92.5.1Sniffing . 102.5.2Accumulation leak testing. 102.5.3Vacuum chamber inside-out leak testing . 112.5.4Outside-in leak testing . 112.6Applications . 122.6.1Halogen leak detectors . 122.6.2Inside-out helium sniffer detectors . 132.6.3Outside-in helium spraying . 142.6.4Outside-in helium leak testing. 152.6.5Inside-out helium vacuum chamber leak testing. 162.6.6Inside-out hydrogen sniffer detectors . 173How to Choose the Test Method . 184Conclusion . 20

1IntroductionIn the refrigeration industry, components and systems must be leak tested to ensurethat refrigerant leakages are below specified limits. The three basic functions of leaktesting are 1) determining if there is leakage or not (detection), 2) measurement ofleak rate and 3) leakage location. There are many methods and types of testequipment for solving these problems, but unfortunately there is no single techniquethat fits every situation. Each test method is suitable only for a specific leak rate orfor fixed forms and technologies. In most instances where leak detection is used,explicit leak rate measurement is not required, but the system must be able torecognize if the leak rate is above or below a specified level. This reference limitdepends on the maximum acceptable leak rate, consistent with the reasonableworking life expectation for final products, and, especially in certain countries, onrules and regulations constraints. The acceptable leak rate, depending on refrigeranttype and application, usually spans from 15 g/y (0.5 oz) of refrigerant for large airconditioning systems and/or automotive applications to 0.5 g/y (0.01 oz) fordomestic refrigerators.This acceptance level is the main parameter to consider when selecting theappropriate method or combination of testing methods. Several other factors mustbe taken into account as well. In particular, system costs, complexities,environmental impact, reliability, influence of external conditions, operatordependence and user-friendless should all be considered.There is a lot of literature available about leak-testing, leak detection and leaklocation methods; provided here in Annex C are some references. This articlepresents some leak detection techniques and compares their performance withspecial attention paid to refrigerant leakages.2Leak testing methodsA leak can be defined as an unintended crack, hole or porosity in an enveloping wallor joint, which must contain or exclude different fluids and gases allowing the escapeof closed medium. Critical leak spots in closed systems are usually connections,gaskets, welded and brazed joints, defects in material, etc. A leak test procedure isusually a quality control step to assure device integrity, and should preferably be aone-time non-destructive test, without impact on the environment and operators.Several leak-testing techniques are available, spanning from very simple approachesto systems that are more complex. The most commonly used leak test methods areunderwater bubble test, bubble soap paint, pressure and vacuum decay, and tracergas detectors (halogen, helium and hydrogen). The first three techniques, due totheir characteristics and sensitivity, can be used only for gross leak detection (300g/y (10.5 oz) or more refrigerant leakages). Tracer gas leak testing methods aremuch more precise than the previous group but, in many cases, their theoreticalsensitivity is more than is required. In a practical sense, however, this is limited byenvironmental and working conditions.Each method mentioned above and each its advantages and drawbacks arediscussed briefly in the following.

In annex A, a conversion chart for the most commonly used vacuum and leak ratemeasurement units is provided.In the diagram below, the performance of various leak-test techniques aresummarized.LEAK DETECTION SENSITIVITYUltrasonicBubble test (soap painting)Bubble test (He, alcohol)Water immersion (bubble test)Leak Detection TecniquesAcusticalPressure decay (without pressuredifferential)Pressure decay (with pressuredifferential)Vacuum decayThermoconductivityHalogen snifferHelium sniffer (inside-out)Hydrogen detectorVacuum chamber helium leak test(inside-out)Helium leak test outside-in1.E 021.E 011.E 001.E-011.E-021.E-031.E-041.E-051.E-06Leak Rate (mbar l/s or atm cc/s)2.1Water immersion bubble test methodThe water-immersion bubble test, also called "bubble testing" or "dunking", is atraditional and relatively primitive technique of leak detection. It consists ofimmersing a charged or pressurized part, usually with high-pressure dry air ornitrogen, in a water tank and watching for escaping bubbles. The larger and more1.E-07

frequent the bubbles, the bigger the leakage. Relatively small leaks are possible, butvery difficult, to detect.The main limitation of this method is sensitivity, which is the minimum detectableleak rate. Considering a spherical bubble of radius R, its internal volume V will be:V 4/3 · π · R3Let p the pressure inside the bubble and t the time required to form the first bubble,the leak rate Q will be:Q (p · V) / tThe two key parameters determining the sensitivity of this method are the smallestbubble detectable by the operator and the waiting time for bubble generation. Thistime must be compatible with the production rate and with operator attention.It is reasonable to consider that the smallest bubble an operator could detect has 1mm radius and that the waiting time is 30 seconds. Assuming that the pressureinside the bubble is at atmospheric pressure, it can be stated from the previousequations that the bubble volume is V 4.2·10-3 cm3 and therefore the minimumdetectable leak rate is:Q (p · V) / t 1000 * 4.2·10-6 / 30 1·10-4 mbar · l/sThis is a theoretical value. The real sensitivity is strongly influenced by manyexternal factors, such as illumination conditions, water turbidity, unit location andplacement, and water movement. All these issues, together with operatordependency, limit the useful sensitivity to 5·10-4 mbar · l/s, although 1·10-3 mbar ·l/s is usually considered.Some tricks to can be used improve to this method. Increasing the internal pressure in increments may increase the probability offinding a leak and can be less time-consuming in pinpointing the leak. A detergent can be added to the water to decrease surface tension, whichhelps to prevent the leaking gas from clinging to the side of the component. Using different gases (e.g. helium) and/or liquids may give some advantagesin system performance, but at a cost disadvantage. Hot water in the tank sometimes helps to increase the pressure inside thecomponent or piping system. If dry nitrogen is used, this does not helpbecause nitrogen does not increase its pressure significantly. If refrigerant iscontained in the system or component, it may help considerably to increasethe pressure and, therefore, increase the chance of finding the leak.In conclusion, this technique does offer leak detection accuracy in the 10 -3 mbar · l/srange in high volume production applications and, in most cases, leak location and isvery economical. However, the disadvantages range from a relatively lowsensitivity, high operator dependency and possible part contamination, to fluid wasteand the likelihood of having to dry the parts after testing. Moreover, especially when

dealing with big coils, excessive unit handling, putting parts in and out of tanks, addsto the complexity of production and results in higher part damages. There are alsosome more hidden costs. In fact, this process requires use of a large amount ofspace and produces a certain amount of wastewater. This is especially true for bigunits, such as large heat exchangers; the tank could be very large and require a lotof water. Dryers cost money to operate and maintain as well.2.2Soap solution bubble testInstead of submersing the part in water, the pressurized unit to be tested is sprayedwith a soap solution and the operator is able to see the bubbles formed by gasescaping from where the leak is.Soap solutions are available in many different types. Some have a brush applicatorand others have a dabber (an absorbent ball attached to a stiff wire inside of thecap.) Some brands may even have a spray applicator to quickly cover large areas oftubing in a short amount of time. This is an advantage but is also messy and timeconsuming to clean up.Some soap solutions even have an antifreeze base to prevent them from freezing inthe winter time. Others may have a lower density to make them even more sensitiveto very tiny leaks.This method has a higher sensitivity than water immersion. It allows detection ofleaks up to 10-5 mbar · l/s and is suitable for very large systems.This soap solution method is best used when the approximate area where a leak mayexist is known. In this case, the soap solution is only used in that specific area totest for and pinpoint a leak. It is the simplest and least expensive method, materialwise, known today. However, if the operator does not know where the leak might be,it can be more expensive because of labor costs.Increasing the gas pressure raises the probability of pinpointing the leak and is lesstime-consuming. However, for operator safety, the pressure must be limited to 1700kPa (250 psi).The soap-solution bubble test is limited by some drawbacks. The area to be sprayedmust be a simple and easily accessible surface. On finned pipes or the bottom part ofa large heat exchanger, it could be extremely difficult, if not impossible, for theoperator to spray the part and watch for a bubble. Moreover, the application is notwell suited for high productivity lines.2.3Pressure decay testThis method consists of pressurizing the system with a high pressure gas, usually dryair or nitrogen. Then the part is isolated from the gas supply and, after a stabilizingperiod, its internal pressure is monitored over time. The pressure drop p ismeasured in the time t. If the pressure in the system drops fast, there is a largeleak present in that component or section of the system. If the system’s pressuredrops slowly, there is a small leak present. If the pressure remains the same, thatcomponent is leak-free. The leak rate Q can easily be computed considering thevolume V of the component. That is:

Q ( p · V) / tLeak detection sensitivity is related to the testing time, the pressure transducerresolution and the volume. The most advanced systems allow for measuring pressurevariation up to 70 Pa (0.010 psig) at test pressure and, depending on the volume ofthe units to be tested, the leak detection cycle can be as short as 30 seconds andguarantees high resolution. Considering V 1.5 dm3 (0.4 gal) internal volumecomponent with a p 70 Pa (0.010 psig) of pressure decay at 3450 kPa (500 psig)test pressure in t 60 seconds, the leak rate is:Q ( p · V) / t 0.7 · 1.5/60 1.7·10-2 mbar · l/sSeveral external factors, such as temperature variations and mechanicaldeformations, affect this test. The internal pressure, in fact, depends ontemperature, and thermal fluctuations may cause changes in pressure, altering theresults. Fortunately, dry nitrogen experiences very little pressure changes when it isexposed to small temperature changes.The sensitivity of this testing technique depends on pressure measurementresolution, test time and pressure values. Longer test times allow for a moresensitive check but, in this way, the test can be very time-consuming because somelow-level leaks may require a very long holding period, some even hours. The higherthe pressure, the faster you can determine if a leak is present. However, operatorsafety concerns limit the maximum admissible pressure value without safetyprecautions. Components can be leak tested at low pressures, less than 2 MPa (290psig) without protection, and higher pressures, 7 MPa (1000 psig), may be usedadopting safety interlocked protection hoods. Using the proper pressure, this testmethod also allows compliance with technical specifications, such as AmericanUnderwriters Laboratory (UL) as well as burst test and the European EN378 rules.Burst test is designed to test the mechanical strength of the refrigeration tubingcircuit, to find ruptured tubing, and badly brazed joints with material separation.Pressure ranges for the test vary depending on if the test unit is a component of therefrigeration circuit, or if it is a complete refrigeration circuit wit