1. Introduction

Drs James and Jean Goodwin from the University of New Mexico School of Medicine, Albuquerque, USA, wrote an enlightening article in the May 1984 issue of the Journal of the American Medical Association (JAMA), titled ‘The Tomato Effect’ (Goodwin and Goodwin, 1984). They used this title to illustrate the rejection of the effectiveness of some medical treatments based on a ‘false belief’, in which received wisdom and practice declared that tomatoes should not be eaten because they are poisonous.

The present paper uses the term ‘tomato effect’ in its title to draw attention to a comparable phenomenon, where wisdom and practice have it that nitrogen and air are rejected for diving system leak detection based on the false belief that they cannot find leaks that helium can find. However, both are in fact perfectly capable of detecting relevant leaks and offer an advantage over helium, being significantly less expensive.

Goodwin and Goodwin (1984) explain that, for the old world inhabitants, the tomato was discovered in Peru and carried back to Spain. By 1560, the tomato was becoming a staple of the continental European diet. However, at the same time the tomato was shunned in North America; the reason tomatoes were not accepted until relatively recently in North America is very simple – they were poisonous. Everyone knew they were poisonous, at least everyone in North America. It was obvious. Tomatoes belong to the nightshade (Solanaceae) family. The word ‘nightshade’ is usually preceded by the word ‘deadly’, and for good reason … It simply did not make sense to eat poisonous food. Not until 1820, when Robert Gibbon Johnson ate a tomato on the steps of the courthouse in Salem, New Jersey, and survived, did the people of America, begrudgingly, we suspect, begin to consume tomatoes.

The International Association of Classification Societies (IACS) ‘class rules’ for diving system design, build and testing require gas pressure and leak testing (PLT) to be undertaken at various stages of the build or following particular maintenance tasks. Pressure testing examines the mechanical integrity of a system, while leak testing examines gas tight integrity. These class rules call for a gas mixture that is typically either the same as the intended in-service gas, or one with 10% helium content (Det Norske Veritas [DNV], 2010; Lloyd’s Register of Shipping, 1989). The rules vary between societies.

Everyone knows the reason for using a gas mixture with helium content when leak testing. It’s obvious – helium belongs to the noble gases family. It has a smaller atomic mass than nitrogen. Helium can find leaks other gases cannot. It simply does not make sense to look for leaks with air or nitrogen. Everyone knows this, at least everyone in the diving industry.

The use of a gas mixture containing helium is a class society requirement for pressure leak testing, and is the industry’s received wisdom, custom and practice. However, this so-called received wisdom that people generally believe is true is often false. If one considers the design and operational use of a diving system, the fact that a leak test could pass or fail as a result of the atomic mass of the test gas is surprising, especially as helium can be as small as 10% of the test gas mixture (DNV, 2010).

Moreover, when examining the pressures and volumes involved and the allowable pressure loss during testing, it should be quickly appreciated any relevant leaks would be revealed using nitrogen, air or indeed any gas. As a comparison, the atomic mass of helium is 4.007, nitrogen 14.007, and oxygen 15.99 (all within same order of magnitude; Clementi et al., 1963).

On a point of clarity, leak testing of saturation ‘diving systems’ should not be confused with those techniques used for leak testing vacuum systems, in which the test pressures and volumes are many orders of magnitude different. In vacuum testing, mass spectrometers and electronic helium detectors are used to locate extremely small but highly relevant leaks (of helium, a gas not appreciably present in the atmosphere) with up to one million times the sensitivity of pressure decay testing (i.e. gauge testing; Bialous et al., 1969). Vacuum systems are routinely tested for leak rates in the order of 1.013 × 10^-9 mbar·|·s^-1 (VTech Cool Technologies, 2009).

To give some perspective, a 1.013 × 10^-9 mbar·|·s^-1 leak is roughly equivalent to the release of 1cm³ (a sugar lump) of gas in 30 years, or one three-millimetre diameter bubble in nine months. In comparison, a typical diving chamber may be allowed to lose around six cubic metres of gas in the test period; this equates to a leak six million times greater than a typical vacuum system’s allowed leak rate.

Many factors influence the rate at which a gas will leak or effuse. Graham’s law of effusion states that the rate of effusion is inversely proportional to the square root of its particle mass (Laidler and Meiser, 1982). Thus, if the mass of one gas is four times that of another, it will diffuse through a leak path, small pinhole or similar in a vessel at half the rate of the lighter gas. Graham’s law provides a theoretical value and is most accurate for molecular effusion, which involves the movement of one gas at a time through a ‘hole’. However, for diving system testing this is never the case as there is always more than one gas present. In addition, the density, viscosity and geometry of any leak path will influence the flow and thus leak rate (Davis et al., 1986).

Although helium will pass through a hole more quickly than nitrogen or air, it is the ability to detect in the vessel’s surroundings if there is a leak rather than the lost volume and its effect on pressure that makes this an appropriate technique in vacuum testing. Accordingly, helium is used in vacuum testing not for its leak rate properties, but as a search gas. This is the ability to detect molecular quantities of helium gas escaping pressure decay by gauge or by covering the area of interest with liquid. By comparison, looking for bubble formation would simply fail to reveal this in an economic or practical test period (Leakdetection-Technology, 2013).

Table 1 lists the common forms of gas flow. Gas can also escape from a pressure vessel as a result of permeation, which is the passage of a fluid through a solid barrier that has no holes or geometric flaw. While permeation is a real phenomenon, it is not a leak in the sense of diving system leak testing (Norton, 1957).

Molecular flow is somewhat different to the other three listed in Table 1 and is seen in small leaks (<10^-7 mbar·|·s^-1) at low differential pressures (<0.01mbar), in which each gas particle travels and acts independently, unlike those in a fluid. Molecular flow is not relevant to diving system leak testing (Marchand et al., 2005). Turbulent flow and laminar flow are referred to as ‘viscous flow’, as the gas behaves in a similar way to a fluid. The leak rate is a value attributed to the amount of gas that flows (escapes) down a pressure gradient over a defined period.

Different gases have different masses and viscosities; therefore the leak rate through an orifice of given geometry will be different if the leaking gas is helium compared to nitrogen, for example. The SI unit for volume leak rate is mbar·|·s^-1 (Prosensys GmbH, 2014). One millibar litre/second is the amount of gas necessary to be removed from a one-litre container in one second to reduce the pressure by one millibar. Relevant gas leaks in diving systems would likely be characterised as ‘turbulent flow’. Turbulent gas flow is thought to be obvious to identify, locate and resolve, and tends not to have published leak rate values. However, in the present paper laminar flow leak rate is used to illustrate the difference between using air, helium or nitrogen for testing, as laminar flow leak rates have published reference values.

Regarding diving systems, the accepted industry standard for leak detection is to use pressure decay over time. Volume is not included in typical test criterion, but as a simple pass or fail subject to pressure loss over time, with a specified allowable pressure loss expressed as a percentage of the test pressure. Pressure decay has the required sensitivity to identify relevant leaks.

2. The test schedule

In diving system leak testing, class rules provide a test criterion. They define the test pressure (usually between maximum and 1.5 times maximum working pressure) and allow some pressure decay over the test period, with temperature fluctuations factored in. As an across class society interpretation, a pressure drop of 1% in 24 hours is the maximum allowed loss, with a minimum test period of six hours (DNV, 2010; Lloyd’s Register of Shipping, 1989).

Further to temperature stabilisation, most tests are over six hours for practical reasons. Importantly, small losses are allowed. As a comparison, the European Pressure Equipment Directive (PED 97/23/EC), under which many pressure system and parts are manufactured, has no special gas requirement when pressure or leak testing, even if a system is designed for gases such as helium or hydrogen. However, PED does have its own detailed test criterion (European Sealing Association, 1999).

It is interesting to note that the American Society of Mechanical Engineers’ (ASME) design code, which is often used outside of the EU, only calls for an oil-free gas that is non-flammable (American Society for Testing Materials [ASTM], 1973).

As class rules reveal, the so-called bubble tight – established by searching for leak with a soap solution – is not a requirement. However, it is common practice to search for leaks around penetrations, valves and other pressure boundary points using a water and soap solution, and most reputable manufacturers and system technicians will do this during a test. Bubbles are allowed in diving leak system testing, accordingly; this is a technique for locating rather than for characterising leaks.

3. Two examples of why the current requirement in class society rules and diving industry practice may benefit from a review

3.1. Example 1: Diving system leak test

This example compares a pressure test on two chambers of different volumes using the same gas and test pressure. With a chamber with a floodable volume of ten cubic metres and one with a floodable volume of one cubic metre, it would seem odd if both had identically sized leaks. The large volume chamber could lose just short of 1% of pressure over time and pass its test, whereas if the small chamber lost the same volume of gas, it would fail. Indeed, the larger chamber would need to lose a volume ten times greater than the smaller one before its pressure decayed to a failure threshold.

Most relevant is that the gas used for testing in this first theoretical example makes no difference; it’s the volume to leak size that matters, not the leak rate and gas used (see Table 2). It is also interesting to note that for a larger volume chamber, such as a modern saturation diving systems sleeping compartment with a floodable volume of around 20 cubic metres, the allowable loss of pressure could equate to a volume of six cubic metres of gas. Such a leak would be very evident regardless of the gas used. Allowable leaks in a diving system are many orders of magnitude greater than vacuums systems allow.

Mixtures can be even more confusing. If one takes test mixtures as advised by some class societies, for example (a typical in service gas) of 98% helium, 2% oxygen or an alternative advised test gas mixture of 10% helium in nitrogen, at test pressure the partial pressure of helium in each mix is quite different (see Table 3). It can be an order of magnitude difference with one society asking for one mixture and one asking for the other. In a mixture, each gas will contribute to how that gas behaves (flows) as a fluid.

Flow is the volume of fluid that passes through a given surface per unit time. The SI unit is m³/_s (cubic metres per second). The viscosity of a fluid is a measure of its resistance to gradual deformation by shear stress or tensile stress. Viscosity can be thought of as measure of a fluid’s resistance to flow (Davis et al., 1986). The SI unit of dynamic viscosity is the Pascal-second (Pa·s), known as the Poiseuille. The property of a turbulent gas that relates closely to how it behaves or flows is its viscosity (see Table 4).

While all four gases have different viscosities, all are within the same order of magnitude. As a further comparison, if one compares the published leak rate of helium under laminar flow conditions to air or nitrogen, there is only a 7.7% difference between air and helium, and a 12% difference between nitrogen and helium (Leakdetection-Technology, 2013).

3.2. Example 2: Diving system leak test

This example is meant to give perspective to the leak rates in section 3.1. If one has a pressure vessel under test at 33 bar and a leak that allowed the test gas air to escape at such a rate, the vessel would fail its test at exactly the 24 hour point, i.e. it lost 1% of its test pressure or 0.33 bar.

A revealing question would be: at what time would the vessel fail if 100% helium had been used for the test? As helium has a leak rate 7.7% greater than air, the test fails 7.7% of 24 hours sooner (110.88 minutes or one hour and 40 minutes) – i.e., the leak would lead to a failed test at 22 hours and 20 minutes.

For practical reasons, tests are normally carried out over six hours. This means that a chamber tested using air that fails right on the six-hour mark would fail 27.7 minutes sooner at five hours and 33 minutes if pure helium were used. This would be the same as testing with air or nitrogen and simply increases the test period to allow for the difference in theoretical leak rate – i.e., by adding 27 minutes to the test period if using nitrogen.

Of practical relevance, it should be noted that 100% helium is not commonly available within the diving industry, and as the leak rate is proportional to viscosity, any other gas mixed with the helium would influence the theoretical disclosure time in this example (International Marine Contractors Association [IMCA], 1986). Critically, in this example, air or nitrogen will all find any leak that is present. Although it is true that different gases have different leak rates as a result of their mass and viscosity (in the examples given all within the same order of magnitude), under practical tests, the differences are all but academic other than, importantly, the cost of the gas.

As can be seen in Table 5, the cost of testing is increased elevenfold as a result of using heliox rather than nitrogen. For a 20-cubic-metre chamber, this equates to GB£3960 extra for completing the leak test in six hours rather than six hours and 27 minutes. Therefore, compressed clean air is equally acceptable and significantly more cost effective.

It is not the case that received wisdom and practice has led to the requirement of helium for diving system leak testing to find those ‘small’ leaks that air or nitrogen simply cannot find, when in actual fact when closely examined, the reasoning offered for using helium is far from convincing at a practical level for the pressures and volumes commonly encountered.

Custom and practice based on belief is potent. Foot binding in China lasted 1000 years, however with education and understanding it was eradicated in the 20th century in one generation (Elliott, 2001).

The offered explanation of the benefit of using a helium mixture for leak test diving systems is seductive. However, experience in carrying out numerous gas leak testing, and having witnessed chambers and their associated pipe-systems pressure gauge stable and bubble tight to snoop® on doors, ports and penetrations, has led the author to the question: why is helium relevant?

It is one of the most difficult things to let go of something one has held true and preached. However, it may be time to eat a tomato and begrudgingly review this requirement, particularly as helium is a very valuable earth resource that is becoming increasingly rare and expensive. Its use for leak detection in diving systems should therefore not be mandated, nor should it be used in preference to nitrogen or air.

This paper first appeared in the Society for Underwater Technology journal Underwater Technology,Volume 32, Number 2, July 2014.