Relief Header Sensor

[brian_chapin]

It was recently recommended in our PHA revalidation that we install NH3 detectors in our relief headers. I've done some research and talked to some technicians who say that these 1-2% relief line sensors are notoriously unreliable.

A consultant heard that some people are placing a small stub of pipe on the relief header. At the far end of the pipe is a pressure sensor with a 10# trip. At the end of the pipe that is welded to the relief header they are placing a 10# burst disc. Has anyone heard of such an application? I can't see a reason that this wouldn't work - of course it would not detect a weeping pressure relief valve, but we are primarily concerned with larger releases.

It would certainly be less $$$ than a $2000 manning sensor and likely would not need to be frequenly excercised and calibrated as NH3 sensors must.

Then again, since we have pressure burst discs on all the relief valves going into those headers, do we really need a relief NH3 sensor at all?

[TXiceman]

The Manning sensors are very gas specific and in my opinion, the best on the market. You can buy less expensive units, but they do not perform very well. I feel you would be money ahead to get a sensor that you know will work.

As to needing the sensors in the relief headers, you need to look at the building code for your location. Most are using the international code and adding as required.

Ken

[nh3wizard]

[quote=brian_chapin;71303]It was recently recommended in our PHA revalidation that we install NH3 detectors in our relief headers. I've done some research and talked to some technicians who say that these 1-2% relief line sensors are notoriously unreliable.

A consultant heard that some people are placing a small stub of pipe on the relief header. At the far end of the pipe is a pressure sensor with a 10# trip. At the end of the pipe that is welded to the relief header they are placing a 10# burst disc. Has anyone heard of such an application? I can't see a reason that this wouldn't work - of course it would not detect a weeping pressure relief valve, but we are primarily concerned with larger releases.

It would certainly be less $$$ than a $2000 manning sensor and likely would not need to be frequenly excercised and calibrated as NH3 sensors must.

Then again, since we have pressure burst discs on all the relief valves going into those headers, do we really need a relief NH3 sensor at all?

A few of our warehouses have the sensors installed on the end of the relief header, they are prone to false alarms, I have seen headers installed with a diaphragm type disk installed with setting of approx. 5lbs which would let you know if you have a relief valve leaking by or if a relief has tripped.

Hope this helps

[brian_chapin]

I'm starting to lean towards simply adding pressure sensors on the burst discs - the gauges that are there now are constantly vibrating out of calibration from the compressors anyway and they use the same port.

Furthermore, I already have the inputs on our refrigeration control computer.

[US Iceman]

At the end of the pipe that is welded to the relief header they are placing a 10# burst disc.

This would affect the relief valve capacity, since this increases the initial relief setting of the valve. If the relief valve is set for 250 psi, adding a 10 psi burst disc increases the relief pressure to 260 psig. I doubt that is exactly what ASME had in mind (or ASHRAE for that matter).

If a rupture disc is used upstream of a relief valve you have to derate the actual valve capacity, so an additional rupture disc downstream would present a similar constraint.

I'm not sure I buy this methodology. The person who should advise if this is acceptable is the local inspector.

I like rupture discs, but would use a design pressure of 300 psi (for any vessel in a normal situation!) and then use a rupture disc calibrated for 300 psi with a very small tolerance of say 1% or less. That is much tighter than a normal relief valve, and rupture discs do not weep. They are either sealed or blown.

To get this kind of ability you have to use industrial process related rupture discs (Fike or BS&B are two that come to mind). They are more expensive, but less problems than a relief valve.

And, if a rupture disc blows you won't have to worry about sensors or gauges. It will make a nice noise to provide notice of a relief event.

[NH3LVR]

I like rupture discs, but would use a design pressure of 300 psi (for any vessel in a normal situation!) and then use a rupture disc calibrated for 300 psi with a very small tolerance of say 1% or less. That is much tighter than a normal relief valve, and rupture discs do not weep. They are either sealed or blown.

And, if a rupture disc blows you won't have to worry about sensors or gauges. It will make a nice noise to provide notice of a relief event.

Iceman:
Are you advocating Rupture Discs without relief Valves?
We had problems like this in one plant recently. We installed Rupture Discs under ALL the relief valves (about 20 as I recall). Problem solved.
Weepage from relief valves is a constant problem these days. Actually I think they allways have weeped, it is just since we started installing sensors that we know about it

[US Iceman]

What I would rather advocate is to use higher design pressures for new vessels. A 250 psi design working pressure does not leave very much margin of error in a working situation.

If a relief valve begins to weep, this seems to start about 10-15% of the valve set pressure. That puts the maximum discharge pressure somwhere in the range of 212 to 225 psig. An older valve may be lower due to aging of the valve spring itself (my opinion only).

I have worked on jobs where only rupture discs were used, with the agreement of the owner. The downside is; once they blow... that's it! You loose all of the refrigerant.

Somwhere in one of the safety codes there is a stement similar to this: To determine the design pressure of the vessel/system you have to use the maximum ambient temperature + 25°F (I think..., this should be confirmed) as the maximum design saturation temperature.

So... If the summer design is 100°F + 25° = 125°F, which is 293 psig.

Sure, the normal condensing temperature is only 95°F, but if the receiver is setting in the sun during a shutdown condition you could pop the relief valves.

I really think we need to re-evaluate the 250 psi design basis more carefully.

I should also clarify my earlier post on the use of 300 psi rupture discs. You would only do this IF the vessel design were rated at 300 psi!

The relief device set pressure must never be higher than the vessel design working pressure.

[TXiceman]

I to prefer to see ammonia systems with a 300 PSIG DWP and PSV's.

One thing I have seen done in the chemical industry, where they are very concerned about relief capacity and leakage is to use PSV's with RD following. They then mount a inexpensive pressure transducer between theRd and PSVto monitor any pressure build up.

Doing this they specificy PSV's with bubble tight seats to minimize the potential for leakage.

If you go the route of a end of header RD with a minimal set pressure, you will need a high quality RD such as a Fike or like. Also, you will need to recheck every vessel PSV for proper capacity and set pressure with a back pressure on the header or valve outlet. i many cases you will need to go to a pilot operated PSV in order to get the right relief pressure.

I short, I think you may be better off selecting a single quality ammonia monitor in the common relief header.
Ken

[NH3LVR]

I would not be comfortable with using rupture discs alone in NH3 system except in the case of a very small one.
A popped relief valve is a rare thing. It does happen, but usually as the result of a chain of errors, or ignorance.
In one situation I know of a milk silo was isolated. Unfortunately the clean up crew was not aware of this nor of the consequences of sanitizing the interior with steam.
Even if you have a water tank to absorb relief valve discharge it is usually designed to absorb the discharge of the largest relief valve for a period of one hour. Without calculations I believe this could well be insufficient.
In some systems the discharge is mixed with city water and then put down the drain. This can cause a lot of problem at the sewer plant.
There is probably a 90% chance a relief valve will close when the pressure is relived.
My vote is for rupture discs plus relief valves.
As to increasing the vessel working pressure to 300lbs you will get no disagreement from me. One of the most troublesome plants I ever worked on in regard to relief valve seepage was a spiral freezer package where the condenser was sized just big enough. Hot humid weather had us running at 190lbs discharge, with the screw backing of on high amperage. Of course had the condenser been sized properly, we would have been fine.
A "pet peeve" of mine is when contractors talk about "design days" and say that the condenser will be adequate except for "two or three days a year". In a production setting, any slowdown is not tolerable. The choice to save a few thousand on the condenser will be paid back many times over during the life of the system.

[US Iceman]

A "pet peeve" of mine is when contractors talk about "design days" and say that the Condenser will be adequate except for "two or three days a year".

You are not alone in your concern for this. For me, this is a bad way to design systems and constitutes a way to save money on condensers to get a job.

As you might guess, this really gets me wound up!

The owners don't know or understand what happens, until it is to late. That's expensive, not the few extra dollars for a larger condenser.

I want to come back to some other comments later, when I have more time.

[nh3wizard]

I would rather have the condenser over sized than under especially here in Florida, summer time temps average 95 plus degrees

[US Iceman]

Here is some further reading...

http://www.fike.com/pub/fmpdocs/B9098.pdf (3.2 MB)

http://www.fike.com/pub/fmpdocs/tb8100.pdf (0.35 MB)

http://www.fike.com/pub/fmpdocs/tb8102.pdf (0.2 MB)

http://www.fike.com/pub/fmpdocs/tb8103.pdf (0.2 MB)

http://www.fike.com/holders_products.html

http://www.fike.com/rdaccessories_products.html

http://www.bsbsystems.com/products/c...-Rupture-35882

http://www.bsbsystems.com/products/c...Standard-11481

After reading all of this you will gain a new appreciation for what goes into the design and application of these devices.

[US Iceman]

...but we are primarily concerned with larger releases.

What constitutes a large release Brian?

What we should be looking for are the prevailing indicators that lead to incidents and releases. A relief valve cutting loose is a secondary effect, not the root cause.

Relief valves weep from one of two causes I am aware of:

1. inlet pressure to the valve getting with the tolerance of the valve set pressure, or
2. the valve spring is aging and allowing the tolerance for the weep pressure to reduce

I think part of the logic behind the 5 year replacement plan here in the US is based on the latter (#2 above) and also includes some concern over prolonged exposure to ambient conditions affecting the relieve pressure.

That brings us back to number 1. What causes the system to hit the lower pressure (that is the valve weeping pressure)?

Part of this may be mitigated by forcing the compressors to unload when the discharge pressure reaches a certain pressure. Yes, this does affect the total cooling capacity, but what's better? A near release event, or a minor controlled (forced) reduction in system capacity until the problem is past?

Is the purger operating correctly to eliminate the non-condensable gases?

Since we seem to have some fixation on 250 psi relief valves, the valve tolerance on a 250 psi valve places the upper limit for operating discharge pressure around 210-225 psi (with a new valve in good condition).

With our normal operating summer discharge pressures reaching 180 psig, any slight amount of air or further reduction in the weep pressure of the relieve valve can result in preventable problems.

Personally, I think the bigger problem is a rush to use band-aids on problems rather than fixing the root cause of the problem.

In some respects, some attempts at fixing problems are similar to putting on tourniquets, instead of hiding the knives.

[US Iceman]

Here is a good link for some information.

http://www.henrytech.com/Tips/HT-TT1.pdf

and some basic info:
http://www.henrytech.com/Tips/HT-TT11.pdf

[NH3LVR]

Relief valves weep from one of two causes I am aware of:

1. inlet pressure to the valve getting with the tolerance of the valve set pressure, or
2. the valve spring is aging and allowing the tolerance for the weep pressure to reduce

Another cause I would like to add-poor quality on the manufacturers part.
In the plant I referred to earlier with rupture discs we changed reliefs on vessels where I have no reason to suspect pressures had never been even close to set pressure.
The manufacturer admitted they had been having problems on R-22 systems. I see no reasons that the valves would not weep just as bad on NH3

[brian_chapin]

This would affect the relief valve capacity, since this increases the initial relief setting of the valve. If the relief valve is set for 250 psi, adding a 10 psi burst disc increases the relief pressure to 260 psig. I doubt that is exactly what ASME had in mind (or ASHRAE for that matter).

If a rupture disc is used upstream of a relief valve you have to derate the actual valve capacity, so an additional rupture disc downstream would present a similar constraint.

If the rupture disc covered the header outlet I could see your point - but in this case I was talking of a 1" pipe welded to the side of the 4" header. It was at the end of the 1" that the burst disc was to be at.

It's all a moot point now - pressure switches are going to replace the gauges currently on the rupture discs that are placed before the relief valves.

[TXiceman]

I recently had an ammonia rink chiller blow the reliefs on the condenser due to a comedy of non-maintenance. The company that had been doing the FULL maintenance on the job, had not bothered to check or test the high pressure safeties and the PSVs were 12 years old. The customer lost the spay header pump on the evap condenser and the valves lifted. This site has two rinks and both were evacuated.

When we arrived on site, we managed to get the water pump running and the units running. On investigation, the PSVs lifted at 235 psig and the HP shutdown was set for 255 psig.

EPA was on site asking question very soon after the incident. So, it still goes back to proper maintenance on the system. We also found that the unit did not have a single ammonia sensor installed in the equipment room.

What gets me wound up is a service company that is being paid to maintain a system and not doing it. Also got to do top ends on two of the three compressors and a bearing and ring job on the other one. The records showed it was over 3 years of operation and the compressor heads had not been off. Discharge temps were high and oil carry over was a big problem.

Ken