Pumping Aq NH3 against a closed valve 2

[bif]

We have an operation which transfers Aq NH3 (34% w/w) via centrifugal pump (~4 barg) from one vessel to another. Both are at atmospheric pressure.

I have looked at potential hazards if a valve in the suction line is left shut, but now want to look at the situation if the suction valve is open and the discharge valve shut.

Aq NH3 has a very low boiling point (~20°C). If left pumping against a 'dead head', could the increase in pressure damage the pipework/pump or will it back feed through the pump and storage vessel and relieve through the vent scrubber system on the vessel.

 

[Montemayor]

bif:

I believe you are wrong in your statement that the aqueous Ammonia has a low boiling point. The NH3 is in solution and the gas is what will evolve out of solution as the temperature increases - the water will remain and not boil until it reaches approx. 212 oF.

You are correct in presuming that the evolved NH3 gas will seek the lower pressure and, if allowed, will migrate out the suction line if the discharge remains dead-headed.

The best, overall design is where you absolutely avoid the dead-headed scenario and do not allow the NH3 to come out of solution. I recommend you design for this type of operation. With today's instrumentation technology you should have no problem doing this.

Dead-heading a centrifugal under this service is not recommended. You will not damage the piping; however, you are putting maximum loads on bearings, impellar, seals, and other components. The best answer is: don't do this. Why put such undue stress on a good pump and also generate fugitive emmisssions? It doesn't make good sense and can't be regarded as good design.

Art Montemayor
Spring, TX

 

[bif]

Thanks for the quick response.

This is not a normal operation and usually the valve will be open - the answer of 'don't do it' is not really adequate as it was being considered as part of 'no flow' deviation in a HazOp study - the plant already exists and although we have good procedures, trained operators etc the situation could still occur after maintenance activity

 

[BenThayer]

I don't disagree with Art's overall response of "don't do this" but I do disagree slightly with what 'aqua ammonia' is.

Aqua ammonia is commonly sold in the 29.8%wt ammonia in water is more accurately NH4OH or ammonium hydroxide in water. The reason for the ~30% is due to the vapor pressure of this solution is essentially 14.4 psia or effectively one atm. This allows the use of low pressure tanks.

The heat of the dead-headed pump will cause the ammonia to be released and when the seal is damaged, it will stink really, really bad. Like Art says, don't do it.

 

[Montemayor]

bif:

Under the basic data you gave, I still maintain that you shouldn't do it. Now, you're giving the "rest of the story": this is about a HazOp. When you don't give the complete basic data up front, you're not going to obtain good, applicable responses to your problem.

But this time my response still stands. The HazOp will want to prevent the hazardous emission of NH3 vapors (which will seriously attack mucous membranes like your eyes, lips, nasal passages, etc.) and you have to have a response for the credible scenario. As I said, instrumentation can mitigate the event. Put a limit switch on the discharge block valve wired in series with your pump motor and the pump cannot be started with the discharge block valve in the closed position. Besides this, I would also always CC (chain closed) the block valve during operation and especially maintenance procedures. Now, you can go before your HazOp fully armed with a logical answer to the credible scenario.

By the way, if you believe that the emitted NH3 vapors are only going to "stink" or be annoying, try standing in the area for about 5 minutes while you are generating the NH3 vapors. I have seen the results and it isn't pretty. This is a serious, hazardous scenario and I wouldn't trust administrative control (simply training operators and relying of manual checks) to avoid it. If your pump seals fail totally (which is another credible scenario during dead-head) you have a potential massive spill in the area and you will have to evacuate - it's that serious.

Art Montemayor
Spring, TX

 

[smckennz]

Hi Bif
there are several ways you can look at this.
As you point out the pump can back vent so your piping is not in immediate danger.
As a start you should look at the dead head pump power and the heat dissipation rate from the pump casing; a heat balance. From this you can figure out the rate of gas production under dead head conditions. It may be zero, but it needs to be worked out.
Then you need to look at the pump itself; if it is canned then there is a very slim risk of a dry bearing and a seized pump. If it is conventional mechanical seal, then the seal can be destroyed and gas will leak out. Highly unlikely but difficult to disprove.

The question is how long do you want to run with the discharge valve closed?
If is were my job, and the heat balance indicated gas evolution, I would throw in a flow switch and a timer or equivalent method of flow proving, such as rate of level drop or rise in a tank. The timer would be set to trip out the pump if the pump had been started and flow/level change had not been detected within the time it would take for the (dead headed)pump to heat up to cause gassing. The flow rate trigger point would be the flow rate sufficient to remove dead head heat input without gassing. If the pump could dissipate heat faster than the motor heat input, then I would do nothing other than do a heat balance for boiling water conditions (or 80 deg C to be safe).
My feeling is that under dead head conditions, the ammonia would be driven off as the temperature increased until you had a pump casing full of near water.
The thing I would not do would be to put a padlock on the discharge valve.
I would be interested to know your reasoning behind closed suction valve being a non-issue. It would worry me more than a closed discharge.

Cheers

Steve

 

[bif]

Thanks smckennz,
I now have a basis to work from.

On the point of the suction side valve, I didn't say it was a non-issue, just that I am aware of the possible hazards that may arise and the safety measures we have to prevent it happening.

I was just the discharge side I was a bit unsure of.

Thanks again.

 

[25362]

Perry VI brings Tables 3-21+ with vapour pressures for water, ammonia, and for both together, in aqueous solutions of various concentrations and at various temperatures. These also show that, although in minor proportions, also water evaporates.

As a corollary to BenThayer's post on pressure vessels to store aqueous ammonia at ambient temperatures: a 30% molar solution of ammonia (=28.81 mass%) at temperatures above 80^oF has vapor pressures above atmospheric. For example, at 100 deg F, 21.3 psia; at 120 deg F, 31.7 psia! Table 3-24 on Perry VI gives more details.