Best Of

The IRC/IFGC and NFPA 54 all use 3 psi x 10 minutes minimum for small residential. High volume pipe systems can require longer test durations.

The use of high pressures for leak detection is faulty. Consider the test gauge: the code allows for a maximum resolution of 5 times the test pressure. A 3 psi gauge can be 15 psi max. With the spacing between the graduations, you have a chance of detecting needle movement. Now, consider a 150 psi gauge for a common 30 psi test. If there is a low-level leak, it misleads you into thinking the needle is not moving. High pressure testing has been known to actually cause leaks. See the key valves below

Another big difference is gauge size. It is much easier to read needle deflection on a 4.5 or 6" lollly pop gauge than a little 2" typical one. You also have considerations such as lighting, parallax error, or a defective gauge. Our inspectors bleed our gauges just to prove the needle can move.

Everything leaks: spaceships, submarines, governmental agencies, bladders- they all leak. It, therefore, becomes a question of allowable leakage RATE. Gas appliance valves allow for 235cc/hr through the seals of a valve and 250 cc/hr through the main operator at 3/4 psi per ANSI. Yet, the gas codes, in their arrogance, simply stamp their feet and demand no detectable leakage. A gas piping system or component can be leaking, and the standard test miss it simply from the eyesight and patience of the inspector.

The test prescribed in the gas codes is considered a "pressure decay test". You pressurize then observe any detectable leakage from the equipment specified. An industry variation would be a differential decay test where the subject is tested against a reference piece known not to leak. The difference is the leakage rate. Again, gas is intolerant of any detectable decay. So, how to localize the problem? Well, you first need to consider what can cause needle drift on a 3 psi pressure decay test. You can have legitimate leaks- poorly formed threads, poorly sealed joints, faulty mating surfaces, improper types of valves and connections, a drop in temperature (even in one part of the grid), etc. The very test apparatus should periodically be bench tested for leaks. This may entail a submersion test for bubbles. This has limitations and is subjective. You may think you see micro bubbles leaking when, in fact, they were formed upon submersion and attached to the exterior of the device at a joint. It can be difficult to separate tiny bubbles continuing to stream out of a leak versus a growing thin film soap bubble. Soap bubble solutions notoriously have their own bubbles premixed from jostling around or hurried application.

The most common leak points are joints and devices. Sounds simple, right? Not so fast. It is too common to find a nail or screw piercing a CSST line. Water or oils in a sediment trap can absorb gas under pressure over time. This is why the old copper air chambers failed over time to provide water hammer buffering. They would fill with water as the air dissolved into solution. Poly gas lines used outside for in-ground trenching at pressures above 5 psi can expand under Hoop Stress causing vasodilation resulting in a drop in pressure from the increase in volume. Isolate plastic gas piping from hard pipes.

Pressure testing at greater than 1/2 psi/ 14 wci requires ALL appliances to be disconnected and the PIPING capped. Piping is within 6 LF of the appliance valve. Note that ground unions are notorious leakers. The mating surface is ground but not lapped in most cases. The brass ring inset was deformed and mated to that female coupling. Sometimes guys take them apart and mix mating pieces of used fittings. You want all new unions.

If anyone used unapproved fittings or made them up improperly, they can leak slowly. It is common to find compression joints on copper tubing as well as pipe dope or PTFE tape on flare joints or ground unions. These are metal to metal joints. Old ground key valves can leak. The 'Dante' floor/ wall key valves for gas fireplaces and log lighters are notorious leakers. These valves were typically rated at 3 psi only. High pressure testing ruined them causing leaks. We had to replace many. The newer ball valves are better. In fact, many current ANSI Z21.15 gas cocks and ball valves are tested only to 3 or 5 psi. Globe and gate valves are inappropriate. Swing joints are susceptible to leaks.

Some things may seem tight at test temperature but leak at operating temperature or vice versa. Static pressure decay testing is not representative of conditions of use. In a large building, I've seen a section exposed directly to the cold cause a pressure drop mimicking a leak even over a short test period. We separated the cold section in the unconditioned space from the heated space and both passed. Overnight testing is useless because of the temperature drop causing a false positive 'leak' when it's just Charles Law in effect.

You could try a variety of other leak detection methods including vacuum testing, static liquid pressure, electronic ion detection, ultrasonic detection, or tracer gas. Filling gas pipes with water may be the more effective but least practical, especially if the leak is in a wall. Oops. Tracing helium requires an expensive sniffer but comes with its own problems. Helium molecules are so tiny they can leak out of just about anything. You could find yourself chasing a ghost. Detection of refrigerant gases has been found to be rather reliable for very small leaks but not recommended for large leaks. You don't want the EPA mad at you.

For the OP's case, I'd strongly consider doing the refrigerant under low pressure with small holes poked into wall cavities and the sniffer inserted. Consider whether the test gas is lighter or heavier than air so you'll know where to test.

i would be concerned that trying to cram insulation inside the baseboard covers would impede the airflow through the element.

unless there is something very wrong with the system with baseboard on a cast iron boiler the return water temp should be well above flue gas condensing temps.

I found a way to make some really cool graphics for my articles, videos etc. Go to your favorite AI. Mine is Grok Upload a picture of a component and ask it to make a vector image. Works really well. Below are some examples.