Comparing technical tightness requirements with leak rates considered safe for hydrocarbon systems.

A year or so ago, I wrote a couple of columns here about tightness testing. The discussions about suitable tightness levels and appropriate procedures to verify them rumble on, and they led me to a quite startling realization that I want to share with you.It is easy for us to forget what an odd segment of engineering the refrigeration world is. Without a second thought, we do all sorts of things that seem counterintuitive or even downright crazy to engineers working in other fields. One example is the level of tightness that is written into our codes and standards. For example, ISO 5149 on Refrigeration Safety, which is under review at the time of writing and may be published later this year, requires that a tightness test is conducted with equipment “with a capability of 5 g (0.18 oz) per year of refrigerant or better.” The system passes the tightness test if no leaks are detected when this specification of test equipment is used.There are several good reasons for setting such a tight tolerance on the tightness test before the system is charged with refrigerant. The charge of refrigerant is sealed, and often the system is designed to make the charge as low as possible. However, several studies have shown that even a small loss of refrigerant over time can seriously impair the plant’s performance in several ways. The first thing to go is the system efficiency. The room (or water, or product) will still be cold enough and nobody complains, but the energy bill will be a bit higher than it was last month. If the leak persists, then the next thing to go after efficiency is effectiveness. Now the room (or water, or product) is no longer in specification and attention is drawn to the plant performance, but the slight loss of charge might still not be obvious and at least the plant is still running. The third stage is loss of reliability. The plant breaks down and cooling is no longer happening at all. To put these into perspective, loss of efficiency might happen with 10% refrigerant loss, loss of effectiveness with 30% refrigerant loss and loss of reliability when more than 50% of the charge is gone.In contrast, the natural gas network in the UK, which is typical of all such services, loses 3% of its inventory per year through leaks. For a chiller with 10 kg of propane that would be 300 g (10.58 oz) per year—sixty times higher than our threshold.A lot of the debate in the standards working groups has been around the level of leakage that can be tolerated from a hydrocarbon system without creating an unsafe condition. I came up with a number of 750 g (27 oz) in three days: I reckoned that this level of leakage would not create an explosive atmosphere under almost all circumstances. A quick calculation showed that this is 10 g (0.36 oz) per hour, which is 17,500 times higher than the detection threshold in ISO 5149. I struggled to come up with a meaningful illustration for this and eventually visualized a seesaw with a five ton elephant on one end and a hamster on the other.A colleague from the standards working group then pointed out that the IEC standard 60335-2-40 considers a maximum safe leak rate for hydrocarbons of 9 g (0.32 oz) per minute—the equivalent of having 55 elephants on that end of the seesaw. The very counterintuitive conclusion of this thought experiment is that the tightness level required to qualify a hydrocarbon system as “technically tight” (see “Getting Tight,” ASHRAE Journal, December 2024) has got nothing whatsoever to do with protection against ignition hazards. Tell me it ain’t so!