Introducing the idea of “tightness” as a measurable design and maintenance target.

I am a believer in the law of unintended consequences. For example, if you install a fancy burglar alarm and expensive locks on your house, you are more likely to attract the attention of the local criminals because they reason that you must have something worth protecting and therefore worth stealing. The introduction in 2006, for example, of the restrictions on fluorinated gases (f-gas) in European law carried several unintended consequences.

One of these was that we had to reexamine our terminology with regard to the integrity testing of refrigerating systems before putting them into service. Everybody knew that a “pressure test” was required, but the term was used rather loosely and the process was not clearly defined. After 2006 we had to differentiate between a “strength test” and a “tightness test.”

A strength test is typically conducted using oxygen-free nitrogen (OFN) at a pressure somewhat higher than the stated maximum allowable pressure, which became known, rather confusingly, as the “service pressure” (“pression service” in French, and hence the abbreviation “PS”). During the strength test, the system must not be approached as it is seeing these high pressures for the first time and the test is, by nature, a straight pass/fail. You don’t want to be near if it fails.

The tightness test used to be referred to as a “leak test”—not unreasonable, as the objective is to find leaks. However, the subtle distinction introduced by the f-gas regulations was that the term “leak test” is now reserved for the procedure used when the system has been charged with refrigerant and is in service, hence a new name was needed for the integrity test of the installation once the strength test had been completed.

A system is referred to as “tight” if it is not leaking. However, everything, even steel pipe, allows a certain passage of gas molecules, so the system is called “technically tight” if the leakage is less than a specified threshold. The leakage rate is measured in the odd-looking units of Pa.m³.s⁻¹ or Pascal meter cubed per second. This is the gas leakage rate required to raise an external volume of one cubic meter by one Pascal every second. The American equivalent for leak rate is the standard cubic foot per minute, or scf/min. 1 Pa.m³.s⁻¹ = 0.021 scf/min.

Testing a system joint with a soapy liquid to detect bubbles of leaking gas, if it is done well, can pinpoint a leak of 1×10⁻⁶ Pa.m³.s⁻¹ or 2 x 10⁻⁶ scf/min. This size of leak would release 140 g (5 oz) of R-134a in a year but for refrigerants with a GWP greater than 150, the f-gas regulation requires testing with a calibrated instrument with a sensitivity of less than 5 g (0.2 oz) per year for the system to be considered “technically tight.” Adding a tracer gas to the OFN and using an electronic detector improves the sensitivity to 0.14 g (0.005 oz) per year.

Some folks think that holding the strength test gas at the service pressure for a few hours before releasing it is a suitable tightness test. I calculated that you would need to observe that test for several decades to spot a 5 g per year leak, even in a relatively small system.

Getting tight is difficult.