Industrial leak testing of dangerous goods packagings

Correspondence Eva Schlick-Hasper, Federal Institute for Materials Research and Testing (BAM), Berlin, Germany. Email: eva.schlick-hasper@bam.de The Dangerous Goods Regulations currently do not include limit leakage rates or sensitivity requirements for industrial leak testing procedures that are equivalent to the bubble test, which is the prescribed test method for design type testing of dangerous goods packagings. During series production of such packagings, various methods are used, which often do not meet the requirements of the bubble test with regard to important criteria. Sensitivity, flow direction, pressure level and automatability are particularly important factors when selecting a suitable industrial leak testing method. The following methods are in principle both suitable and equally effective as the bubble test: pressure rise test (vacuum chamber), ultrasonic bubble leak detection and gas detection methods (pressure technique by accumulation and vacuum chamber technique). To ensure a uniform test level during design type testing and production line leak testing and therefore a comparable safety level as required by the Dangerous Goods Regulations, it is necessary to include a more precise specification in these regulations. This requires, on the one hand, information about the sensitivity of the bubble test and, on the other hand, the inclusion of a list of suitable, equally effective industrial test methods with their specific boundary conditions.

permissible limit leakage rates 4 nor guidance about suitable industrial test methods, which must be at least equally effective. During production (piece-by-piece testing), various leak testing methods are used whose sensitivities differ considerably. As a result, an unequal safety level and distortions of competition are possible. 5 For the bubble test as defined in UN 6.1.5.4, the test sample is exposed to an internal air pressure (gauge) of 30 or 20 kPa, depending on the degree of danger, and is restrained under water for 5 min. The result is a simple fail/pass statement depending on whether bubbles rise or not (qualitative approach). 6,7 The minimum detectable leak diameter d min of the bubble test under the test conditions stated in the UN Model Regulations is approximately 9.7 μm for 30-kPa gauge pressure and approximately 14.6 μm for 20-kPa gauge pressure. 7,8 But this diameter is not a suitable measure for a quantitative comparison with other leak testing methods. 7 The first question in this work is what test level can be assigned to the bubble test as a specification for the sensitivity of an industrial leakproofness test.
The following approaches are compared below to assess the sensitivity q min of the bubble test: • Specification from leak testing standards (q min,std ) The second main question is which industrial leak test method is equivalent in terms of sensitivity and other important parameters. An overview of common leak testing methods used in industrial series production is given. These procedures are compared and evaluated for their suitability for the series production of dangerous goods packagings.
The aim of this article is to establish a systematic way in choosing suitable and equally effective industrial leak testing methods for dangerous goods packagings. The results could serve as a basis for adding a list of such test methods into the UN Model Regulations.

| Leak test sensitivity
Absolute leakproofness does not exist. 9,10 Instead, leakproofness is a relative term. 11,12 One definition is the following: The test object is considered to be leakproof if the test method selected and its corresponding detection sensitivity cannot prove the passage of the test medium from one side to the other or to the outside. 13 The sensitivity of a leak testing procedure (detection limit) is the minimum detectable leakage rate q min the test procedure is capable of detecting. 6,7,12 The definition of the leakage rate is the pV-throughput of a specific fluid that flows through a leak under specified conditions. 4,14,15 The lower the numerical value of q min , the higher the sensitivity. For a proper definition of a leakage rate, it is also important to state the prevailing test conditions. 12 One possibility is to specify the leakage rate with reference to the driving force: the pressure differential. The leakage rates given in ISO 12807 are standardised leakage rates (SLRs), 6  Another possibility is to express the leakage rate as released volumetric flow into the surrounding atmosphere at standard ambient pressure 101.325 kPa. 16 The European standard DIN EN 1593 uses this type of indication to quantify the leakage rate of the bubble test on the basis of bubble diameter and bubble frequency. 17 This type of leakage rate information only considers the amount of gas released, regardless of the exact level of upstream and downstream pressure.

| General criteria for selecting a suitable leak test method
When choosing a suitable leak test method for the respective application, various aspects must be taken into consideration in advance.
These criteria can be divided into the following areas [18][19][20] : • Sensitivity of the test method: The permissible limit leakage rate of the test object determines the selection of the method. 14 The sensitivity and the measuring range are the most important parameters for choosing a test procedure. 3 The requirement for a test method is usually a sensitivity that is higher by a factor of 10 than the limit leakage rate to be detected. 21,22 • Objective and extent of the investigation: In general, two types of leak test methods can be distinguished: methods for leakage rate measurement (quantitative methods) and those for leak localisation (qualitative methods). 6,7,14,23 Therefore, the objective can be a leakage rate measurement or a leak localisation. The range can be the total area of the test object or a single local area. 14,23 • Operation conditions and testing conditions: Test media are liquid or gaseous substances that must be specifically detectable after they have passed through a leak. 13 Generally, a test fluid other than the operational fluid is used for the leak test to ensure better handling and higher sensitivity. 14 Relevant parameters are also the test pressure conditions (upstream pressure, downstream pressure and direction of flow) and the testing temperature 14,23 For simplicity, the leak test is usually carried out at ambient temperature. The test pressure should be in the order of the operating pressure, and the flow direction should be the same as under operating conditions. 14,24 • Test object design: The dimensions of the test object, openings for test gas supply and limit values concerning pressure and vacuum also have an influence on choosing a test procedure. The test medium must be compatible with the object material. 14 • Safety and environmental requirements: Many test methods involve the application of a pressure differential, either an overpressure or a vacuum. This must not endanger the test personnel or the system. Some test gases are either toxic or harmful to the environment. Consequently, measures must be taken so that they cannot escape. 14 If a leak test method is not to be used in the laboratory, but in industrial series production, there are further requirements. These are listed in the next section.

| Special requirements for industrial leak test methods
Important requirements in the selection of a leak test method for series production are the ability for automation, the reliability of the method and the available economic opportunities (acquisition costs and operating costs). 21,24,25 A central point is the compliance to a given production cycle. 26 By deducting the handling time, the total test time, that is, the time available for the leak test, results from the production cycle time. 27 Because of these additional requirements, sensitivity of one and the same leak test method is often lower under industrial conditions than under ideal laboratory conditions. 24

| Resulting requirements for a leak test method for industrial series production of dangerous goods packagings
From the requirements listed above, the following criteria are of particular importance for the industrial leak testing of dangerous goods packagings: • Sensitivity of the test method: Limit leakage rates for dangerous goods packagings do not currently exist in the UN Model Regulations. 4 The question is fundamental of how sensitive a method equivalent to the bubble test should be. Therefore, in Section 3, different approaches on how to assess the sensitivity of the bubble test, which is a requirement for an industrial process, are presented.
• Objective and extent of the investigation: Because the bubble test is performed by completely submerging the packaging, the industrial method must also allow a leak test on the total area of the packaging. However, the original closures may be removed (UN 6.1.1.3). 1 The bubble test is primarily used as a method for leak localisation, even though a quantitative evaluation is possible by bubble counting or measuring of the escaped gas quantity. 6,14,17,23 To select an equivalent method, a quantitative analysis of the bubble test is essential. Quantitative methods have an advantage over qualitative methods because they are more objective.
• Operation conditions and testing conditions: Depending on the degree of danger, either 20 or 30 kPa is specified as test overpressure in the bubble test. Even with an alternative method, this pressure level should be selected, otherwise irreversible deformation of the test sample may result. The flow direction in the bubble test corresponds to the direction of flow, which is also present during the release of a dangerous substance under transport conditions, namely, from the inside to the outside. An equivalent industrial process should also ensure this flow direction.
• Test object design: In series production, a leak test procedure should be nondestructive, so that the tested objects can be distributed. Therefore, the specification in the UN Model Regulations makes sense that the original closures do not have to be in place.
In this way, it is possible to pressurise the test sample, for example, with compressed air or test gas via a test connection or an attached adapter, without destroying the packaging.
• Suitability for automated series production: In this work, the automatability and the production cycle time are considered in particular. Because the focus is on the technical suitability of the individual procedures here, cost aspects are not considered.
Safety and environmental requirements are also not considered in the following. Of course, individual specifications of the industrial test method to be selected can only be made with reference to the specific application with its specific boundary conditions in the production line.
The question of how sensitive a method, equivalent to the bubble test, should be is fundamental. Therefore, in Section 3, different approaches on how to assess the sensitivity of the bubble test, which is a requirement for an industrial process, are presented.

| Specification from leak testing standards (q min,std )
Although the bubble test is classified as a method for leak localisation, information about its sensitivity is provided in various leak testing standards. 6,14 ISO 12807 states the nominal sensitivity of gas bubble techniques under industrial conditions as 10 −4 Pa m 3 /s (SLR). 6 Higher sensitivities could be achieved using other test fluids than water that have lower surface tension. 6,12 An equivalent value of 10 −4 Pa m 3 /s for the sensitivity of the bubble test under industrial conditions is also mentioned in DIN EN 1779 for comparable flow conditions. 14,23 The value q min,std,SLR represents the sensitivity under the practical aspects of industrial leak testing.
Leakage rates can be converted from one condition to another assuming a constant leak geometry and a specific flow regime. 6 For a given leak and one and the same type of gas, the relationship between the leakage rates at two different pressure levels is 14 A conversion of the air leakage rate q I = 10 −4 Pa m 3 In this equation, p b is the effective absolute pressure inside the bubbles, and V b,tot is the total bubble volume and t is the test time. In this case, the leakage rate refers to the amount of air released from the component to the surrounding atmosphere. 16 This value for the sensitivity of the bubble test in the order of 10 −7 Pa m 3 /s is not feasible under industrial conditions from an economic and practical point of view. 6,23 Therefore, the sensitivity value obtained in this way differs from the value of q min,std obtained in Section 3.1 for industrial conditions in several orders of magnitude.

| Calculation of q min,UN,cap
The basic steps of this calculation are described in previous studies. 7

| Pressure rise test
When applying the pressure rise test, a lower pressure is generated inside the test object across the object boundary. For this purpose, the test object is either connected to a vacuum pump system F I G U R E 1 Pressure decay test ( Figure 2) or placed in a pressurised chamber. After reaching the specified pressure difference, the test object is isolated, and the internal pressure is recorded at regular intervals. 32 For the pressure rise test (technique D.2 in previous publications 14,32 ), the same values for the sensitivities are given in previous studies 6,14,32 as for the pressure decay test. In the practice of this process, the outgassing of water or other volatile components can be problematic. By this effect, an initial pressure increase can be generated that is above the pressure rise due to leakage. 14,32 Because of the flow direction, which is directed from outside to inside in this method, this does not meet the requirements based on the bubble test.

| Pressure rise test (vacuum chamber)
A third kind of pressure change method is the bell pressure change technique (technique D.3 in previous publications 14,32 ).
The test object is enclosed by a rigid chamber (bell chamber), and a pressure difference is created between the test object and the chamber. If the test object is pressurised or the chamber is evacuated, any leakage across the boundary wall of the test object enclosed by the chamber will cause a pressure rise in the chamber (Figure 3). If the test object is evacuated or the chamber is pressurised, any leakage across the boundary wall of the test object enclosed by the chamber will effect a pressure decay in it. 32 The sensitivity of the bell pressure change technique can theoretically reach values to 10 −6 Pa m 3 /s, depending on free chamber volume, test time and equipment. 14 Because of the requirements of the bubble test that the flow direction should be from the inside to the outside of the packaging, only the implementation is suitable in which the test object is pressurised or the chamber is evacuated (vacuum chamber).
In principle, the test pressure level specified for the bubble test (flow from 30 or 20 kPa gauge pressure level to atmospheric pressure level) can be applied for all these methods. All three methods can be automated.
Another test method that is formally classified as a pressure change method but is not mentioned previously 2 16 Because the nominal sensitivity of this method is smaller than for the three methods just mentioned, it is no longer listed below. Table 3 lists the individual aspects of the three methods.
Because the flow direction is exactly opposite to that of the bubble test in the pressure rise method, this method is in principle not considered as the bubble test equivalent method. When comparing the pressure decay method and the pressure rise method using an external vacuum chamber, the latter method is found to be more appropriate. On the one hand, it has a higher sensitivity (SLR); on the other hand, the use of a vacuum reduces the temperature sensitivity of the measurement. 34 On the basis of the parameters considered, the third method is best suited for the industrial leak testing of dangerous goods packagings. If in practice a test pressure difference of more than 30 kPa is applied, the leakage rates determined in this way should be converted to the test pressure level of the bubble test using the correlations in ISO 12807. 6

| Ultrasonic leak detection
In the application of ultrasound for leak detection, there are two different variants: in the field of airborne ultrasonic leak testing and for the detection of air bubbles in the water bath in the bubble test.

| Airborne ultrasonic leak detection
Airborne ultrasound detection is an inspection technique applied to locate leaks in pressurised systems. Gas leakage generates sound waves when the gas flow through the leaks is accompanied by turbulence. Airborne ultrasound can be detected at a distance from its source with directional scanning microphones or acoustic probes ( Figure 4). For applicability, it is necessary that the leakage rate is sufficiently large (10 −4 Pa m 3 /s or higher for air at ambient pressure) and thus the flow regime is in the turbulent range. 12 ASTM E1002 specifies a value of approximately 1.5Á10 −2 Pa m 3 /s for this leakage rate. 35 In a contribution about the current developments of the acoustic ultrasonic leak detection, the theoretical sensitivity is given as 10 −3 Pa m 3 /s (flow of leaking air at ambient pressure). In practical application, it is more likely to be 10 −1 Pa m 3 /s. 36

| Ultrasonic gas bubble detection
In the case of industrially used bubble test, the manual underwater visual inspection often fails because it is not sufficiently reproducible and only rarely economical. 37 As mentioned above, the attention of the inspection personnel is not guaranteed over the entire test period and the visibility may be poor. 29 When integrating the bubble test into series production, automated ultrasonic gas bubble detection offers an alternative 37 ( Figure 5).
Ultrasonic waves are scattered when they hit gas bubbles on their way through a liquid. 37 From an ultrasonic sensor, which can be operated both as a transmitter and as a receiver, a wave packet is emitted in the water basin. If there is a leaking part and thus air bubbles in the basin, the wave packet is scattered prematurely on them. The evaluation of the scatter signal is carried out via digital signal processors.
Because of the propagation time of the sound, the system calculates the position of the bubbles and evaluates the component as leakproof or leaking on the basis of the specified limit leakage rate. 29 The advantages of the ultrasonic bubble detection in the bubble test are that even with turbid test liquid, the sensitivity of the system is maintained and that even under conditions of series production, a quantification is possible. 37 With ultrasound, air bubbles with a diameter of only 0.1 mm can be detected-in contrast to optical monitoring. The theoretical detection limit of this method is therefore 10 −9 Pa m 3 Table 4 lists the individual aspects of the two methods.
The airborne ultrasound detection does not have the necessary sensitivity q min,std,ref of 5Á10 −5 Pa m 3 /s (see Table 2). In addition, it is questionable whether at the relatively low overpressure level, as prescribed in the bubble test, a flow can be caused in a magnitude which lies in the turbulent-viscous flow region.
In terms of sensitivity, ultrasonic bubble leak detection would theoretically be appropriate. The suitability with regard to the cycle time of the respective production line would have to be checked.

| Gas leak detection
In the field of test gas method, the following two cases can basically be distinguished: gas flow into the test object and gas flow out of the test object. 14,40 Because the test gas methods, which are based on a gas flow from the outside into the test object, contradict the requirement for the preferred direction of flow in the case of dangerous goods packagings mentioned in Section 2.4, this will not be discussed in the following.
In the field of test methods with a gas flow from the inside of the test object to the outside, there are generally three methods allowing an examination of the total area of the packaging.

| Pressure technique by accumulation
The test object is pressurised with tracer gas and is the placed into a test chamber. The tracer gas, usually helium or a halogen, flows out through leaks into the surrounding volume, causing a concentration increase in the chamber, whose free inner volume is homogenised with fans. After a certain accumulation period, this concentration increase is measured with a suitable leak detector. This method (technique B.3 in previous publications 14,40 ) can be used for objects that can be filled with a tracer gas at a pressure greater than atmospheric pressure. Its sensitivity is 10 −7 Pa m 3 /s (SLR) under industrial conditions, depending on accumulation period 14 ( Figure 6).
Because in this method the chamber does not have to be evacuated but is under atmospheric pressure, no mass spectrometer is required as leak detector. In the automated helium leak test, a system that uses a quartz membrane sensor without a vacuum chamber has instead been used for some time. 41,42 In the test chamber, the test piece is charged with helium via its test gas connection. The quartz membrane sensor is capable of safely resolving increases in helium The just mentioned sensitivities, expressed as helium leakage rates, can be equated with the air leakage rates assuming a laminar viscous flow regime, because the dynamic viscosities of these two gases are similar. 44,45 It is also possible to combine the immersion bubble test and the pressure technique by accumulation. 46 The test sample is filled with helium and is immersed in a water bath, which is placed in an accumulation chamber connected to a helium mass spectrometer in sniffing operation.

| Pressurisation-evacuation (bombing test)
This method (technique B.5 in previous publications 14,40 ) is applicable to test objects that are sealed prior to leak testing and that are not equipped with a connection for filling with test gas. 31 The object is first placed in a bombing chamber and is pressurised with tracer gas (Figure 7). The tracer gas flows through leaks from the outside into the test object. After this 'bombing period', the object is placed in a vacuum chamber connected to a mass spectrometric leak detector.
These objects are generally of small dimensions (for example, semiconductor devices and hermetically enclosed relays). The magnitude of its sensitivity is between 10 −9 and 10 −6 Pa m 3 /s (SLR) under industrial conditions. Test gas adsorbed on the surface represents a disturbing influence during the measurement. Therefore, test objects shall be flashed with tracer-gas free air or nitrogen before testing in the vacuum chamber. 14,40 ISO 12807 indicates a sensitivity between 10 −9 and 10 −4 Pa m 3 /s (SLR) for this procedure. 6 The bombing test is used, for example, in the leak testing of cardiac pacemakers, components for mobile phones or encapsulated electronic modules for applications in the automotive or aerospace sector. 47 For economic reasons, this method can only be used for components with a volume of approx. 10 cm 3 , otherwise the times for pressure storage, flushing and test time will take too long. 31 An application of this method in dangerous goods packagings is therefore out of the question. It can also be assumed that most types of dangerous goods packagings are not suitable for withstanding an external high vacuum, as is required in the operation of a mass spectrometer.

| Vacuum chamber technique
The test object, filled with tracer gas, is placed into a test chamber, which is evacuated to a pressure lower than the internal pressure of the test object ( Figure 8). Tracer gas flowing through leaks into the chamber is measured using a leak detector (technique B.6 in previous publications 14,40 ).
The maximum sensitivity of this method is indicated by 10 −9 Pa m 3 /s (SLR). 6,14 A typical industrial application of this procedure is the testing of 200-L steel drums. In this application, the test gas may be either pure air or an air-helium mixture. The drums are filled with the test gas under ambient pressure conditions, closed and placed in a large test chamber, which is evacuated to high vacuum to allow testing with a mass spectrometer. The production cycle is 360 parts per hour. The entire measuring process including evacuation and aeration is completed within a period of approximately 5 s. Because the test pressure difference is 100 kPa, the drums must be stabilised in the bottom area and lid area so as not to be irreversibly deformed. When using helium as the test gas, its content in the helium-air mixture is 1%. This means a consumption of about 2 L of helium per test. 26,48,49 The sensitivity of this application under production conditions, expressed as 100% helium leakage rate, is <10 −9 Pa m 3 /s. 26 Because of the use of a mass spectrometer and the necessary presence of a high vacuum, the pressure level does not match that of the bubble test. When testing other types than steel drums, it is to be expected that they will be irreversibly deformed by the test and thus destroyed. Table 5 lists the individual aspects of the three methods.
Of these three gas detection methods, the pressure technique by accumulation formally has the best conditions for industrial leak testing of dangerous goods packagings. The vacuum chamber technique is also suitable, provided that it is possible to prevent the packaging from irreversible deformation. Depending on the surface tension of the soap solution used, it is even possible to detect leaks whose diameter lies below the minimum detectable leak diameter d min of the conventional bubble test. 50,51 Theoretically, therefore, the sensitivity, expressed as a quantitative leakage rate, would be higher than in the actual bubble test. Only suitable for low production quantities.

| Summary of suitable methods
Because of the individual consideration of the different methods in  Other methods such as the pressure rise method (vacuum chamber), the ultrasonic bubble leak detection or the soap bubble test are basically able to fulfil the requirements. In the last two, however, the suitability for production lines with high cycle times must be checked.
In the field of gas detection methods, the pressure technique by accumulation and the vacuum chamber technique are most suitable.
Currently, leak testing methods are often used in industrial series production, which do not correspond to the requirements of the bubble test. These test methods are not equally effective procedures, as