Slow leak testing

G'day all, I recently fixed a 15kw ducted system for a mate. It is 9 years old and it had a blocked orifice piston (looked like a piece of cellotape had been floating around in the system since manufacture). He had a pro come and look at it. The "pro" said it was leaking and charged them for 3 kilos of R22 (did not recover and weigh first). Told them the leak was so small it'd never be found and the 3 kilos would do them a couple of years. That did not fix the problem, so they called him back. They watched him recover 7kg of R22 (which is what the system holds). He did some pretty nasty things to the condenser pipework, brazed it all back up and put the gas back (supposedly). No dice. Fast forward 2 weeks and $1400 to the "pro" and I get a desperate call for help.

So the first thing I do is turn it on, find the blocked piston (hot one side, cold the other - simple) and pull the gas out. Hey at $220/kg, if it's leaking we don't want to lose any more. I removed the teflon tape he'd put on the service valves, and fixed the mating surfaces he'd buggered up with his wrench. I then put 150psi of Nitrogen in there and left it for 5 days.

In 5 days I still had the right pressure (when allowing for temperature - I ended up with 160, but it was hotter that day).

There is no way the system had lost 3 kg of gas. It went from working perfectly one day to dead the next, so I assume he charged for 3kg and put nothing in. The owner also saw him pull out 7kg, and I only found 5 in the system. The service valves were leaking, but not that much to lose 2 kilos in 14 days. The Big-Blu bubble fluid indicated a very slow leak, so we assume he nicked some gas as he went. I pulled nearly 2 rolls worth of teflon tape from the service valve threads. Why is it muppets think teflon tape is doing to help seal a flare ?

Anyway, my question is would 5 days with no pressure drop on nitrogen be enough to be sure the system was not actually leaking? My experience has been that even tiny leaks show relatively measurable pressure drops over 24 hours, and they also tend to prevent pulling and holding a good vacuum. I got no drop in 5 days, and it held below 500 microns for a couple of hours. Having said that, I've not worked on a system larger than about 9kw and 3kg of gas before.

Given infinite time and a client who wants it done *right* regardless of time/cost, how would you do a leak test for a possibly tiny leak?

[edit] Why are all my paragraph and line spaces being stripped out? I can't get this post to look anything like what I fed in. I have to add html break tags manually to get anything to work?

in fluid flow terms a tiny hole acts like a choked orifice so the rate at which mass flow out of the little hole is directly proportional to the upstream pressure and really does not care what the downstream pressure is...

So at 150 Psig (165 abs) you would lose mass at half the rate at which you would lose it at 315 Psig (330 Abs). The volume would also contain twice as much mass of gas to begin with....So with one hole in an otherwise tight system the rate of pressure decay is virtually the same....BUT if you have a type of leak that only develops with a high applied pressure (as in stem packings in some styles of valves and stressed flare connections) than you are going to miss it at a low pressure...

.

Also be aware that some leaks, leak more as the pipe expands or
contracts, when it heats up or cools down. Sometimes, some small
leaks only show when the system is running.

And as stated above by Sterl 150psi is only about 10 bar and R22 at
10 bar has a saturation temp of only about 26 degC, in the UK we work
to saturation temps of upto 55 DegC which gives working pressures of
21 bar ish (300 psi) and then we can go above that to prove the system.

If you can't find a leak at lower pressures you need to increase the pressure,
but only to the maximum your standards work to.

Rob

.

Is difficult see very small leak look at pressure change on 10-20 Bar pressurized system

do the math...

3 gram/year (maximal allowed leak refrigerant for one connection by law) is equal to 1.5 litre refrigerant gas in atmospheric pressure, 1500 cc / (365 * 24) give leak volume as 0.171 cc (ml) per hour or 0.0028 cc (ml) per minute.

If waiting 24 hour on free space inside 10 litre pressurized system with 10 Bar OFN and leak is 0.171 cc/h (maximal allowed rate for one connection) and system have 10 litre * 10 Bar = 100 litre of OFN or 100000 cc (ml).

After 24 hour wait have 100000 cc - (24 * 0.171) cc = 99995.896 cc left of OFN and 99995.896/100000 = 0.9995896 time pressure left of starting pressure or pressure going from 10.00000 Bar to 9.9995896 Bar - I think your used gauge not have resolution to detect pressure change in this case lower than 0.5 mBar on 10 Bar measure point (i think practical resolution is around 0.1 Bar) and weather barometric pressure change in most case lot more after 24 Hour waiting time - and temperature shift in system below 1 degree C under waiting time does lot more pressure change than pressure change depend of tiny but to law limit leak-rate give in pressure change.

Even if system have 10 connection and all them leak 3 gram/year - I do have doubts if you can see any small pressure change as 4.5 mBar change on your gauge after 24 hour in measuring point of 10 Bar...

to find this type of leak - you need trace gas example OFN with 5% hydrogen and high sensitivity detector (start detect around 0.5 ppm hydrogen) and use soap to see bubble.... - is already over 10 time over allowed leak rate if you see first small tiny bubbles after some time waiting - remember allowed maximum leak rate is 0.0028 ml/minute for 3 gram/year refrigerant leak and this low leak rate, gas possible solve directly in soap-water without any shown of bubbles...

Quote:

---

Originally Posted by Rob White

.

Also be aware that some leaks, leak more as the pipe expands or
contracts, when it heats up or cools down. Sometimes, some small
leaks only show when the system is running.

And as stated above by Sterl 150psi is only about 10 bar and R22 at
10 bar has a saturation temp of only about 26 degC, in the UK we work
to saturation temps of upto 55 DegC which gives working pressures of
21 bar ish (300 psi) and then we can go above that to prove the system.

If you can't find a leak at lower pressures you need to increase the pressure,
but only to the maximum your standards work to.

Rob

.

---

Had a small tosh. R410a unit with a very slow leak, twelve months before noticable lack of performance that only held 0.7kgs could never find a leak when pressure tested at 500 psi ,eventually found leak when running in cooling mode with soapy suds on outdoor suction flare a bubble every now and then.

Standard procedure for ammonia plants is standing pressure test of 1.5 x the design pressure, standing for 24 hours, obviously safety reliefs blanked. It is a legal requirement here anyway for plant certification. isolate an open drive compressor so as to not blow the shaft seal.
apply same standard to any system and isolate compressors, particularly hermetic compressors. Then multi-vacuum system which is a negative one bar check but mainly a de-hydration check.
magoo

Thanks for the replies all. The system did not actually have a leak at the time as it turns out.... but.
The system holds 7kg of R22. The condenser holds about 4kg of liquid. The system has a 15KW scroll and as it turns out no high or low pressure cutouts. Something came loose in the system and blocked the metering device (looked like cellotape or cellophane), leaving it to try and force 7kg of gas into a condenser that holds 4kg. This physically ruptured a condenser tube and *now* it has a leak. I'm glad I wasn't around when it went off.

Who designs/builds a system that big and has no safety devices at all? Do scroll compressors have a built in pressure relief? Every other system I've looked at has at least a HPCO.

I have a test system at home I use to play with (a big ducted evaporator patched into an old 2HP wall banger) so I made a couple of brazed joints with teeny pin holes in them. Short of sticking some trace R22 in there and using the leak detector (which I didn't do but have in the past when I was desperate) the best result I got was with big-blu bubble fluid. A bit of patience saw even tiny leaks show up pretty clearly. You guys are spot on with the N2 pressure for detecting small leaks. Just too slow to be usable. I did find though that even a teeny tiny pin hole showed up very clearly in a limitation of ultimate vacuum. This test system is probably 30 years old and has no drier and appears to have been abused. Leaving the vac pump on for 24 hours I can get down to about 200 microns. I figure I just can't pull enough stuff out of the oil to get below that. With a miniscule leak I can't get below about 6-700 microns, so it becomes pretty obvious very quickly if there is even a tiny leak under vacuum.

As always, I very much appreciate the advice.