9 replies to this topic

[Quados]

Hi all, i'm really stuck at this problem since the beginning of pump commissioning

First of all, I have various problems with my vertical turbine pump that pumps propylene from the spherical tank. There are two majors problems that we've faced which are;

1. Mechanical seal leaks and trips the pump many times.
2. Start a pump for a minute and then the pump is tripped by low current protection. In this case, the operating discharged pressure and flow cannot be built. **Noted : we used the by pass line (minimum flow line) to test the pump.

At that time, we assumed two plausible causes ; the first one is NPSHa is insufficient and the second one is the discharge vent line is not properly vented back to tank due to there are many low spots. Normally, refer to Gould's pump manual, vent piping must continuously rise back to the source with no low spots. So, we changed the vent design to comply with standard but those problems still occur!!.

Now, i've tried to calculate NPSHa by canceling out the the vapor pressure and tank pressure in the NPSHa equation. Then only hydrostatic head of liquid minus by friction loss in a suction line is left in the equation. According to my calculation, the pipe friction loss is 1 meter and the hydrostatic head is 7 meters, so, the NPSHa result is 6 meters. This pump requires only 4 meters and it seems OK with my calculation.

As i know, Propylene or liquefied gas is very sensitive with temperature change. My point here is our spherical tank is insulated but the suction line and discharge line are not. Therefore, by canceling out the vapor pressure and tank pressure is not applicable for this case right? Regarding to my investigation, the tank temperature is 28.6 degree C with pressure at top tank 11.74 kg/cm3. And the pump temperature (measured by temp transmitter at pump discharged nozzle during pump stops) is around 40 degree C with 12 kg/cm3 that measured by the suction pressure gauge.

My second question is "Are there any possibility that the suction fluid flow piping design is independent to the friction loss in the NPSHa equation and can significantly affect the pump? because now we have suspected that the suction pipe size is too small. Normally, the suction pipe size shall be larger than the suction pump nozzle 1 or 2 size with eccentric reducer at the suction flange but our suction line is 8" along the way over 40 meters to the pump nozzle which is also 8". (Actually the suction line is branching from the 16" header under the tank) Anyhow, the suction loss calculation is low.


Any help would be appreciated.

Regard.

[Quados]

Anyone can help me this?

[GS81Process]

This appears to be a serious safety problem. Propylene is very flammable and leaks pose a serious process safety hazard.

Unfortunately I don't have my copy of GPSA with me. It has a good section on physical properties of propylene.


A few thoughts for you:

1. Are you located in a hot climate? Do you leave the suction line filled with liquid propylene for very long before starting the pump? If so, the propylene in the suction line may be vapourized by heat input from solar radiation.

2. If the vapour pressure of propylene is greater than the operating pressure in any part of the suction line then you will generate propylene vapour. You need to look at a detailed pressure drop profile through the entire suction line and identify if there are any locations where this may occur.

"Therefore, by canceling out the vapor pressure and tank pressure is not applicable for this case right?" If the temperature is higher at the pump suction than it is at the tank (caused by say external heat input from solar radiation), then the propylene vapour pressure is higher than it is in the tank! In this case you cannot cancel these two terms from your NPSH equation.

3. Do you have reduced port ball valves (soft seated?) in the pump suction line which you have not properly accounted for in your NPSHa calc? Any other possible sources of unaccounted pressure drop?

The possible solutions depend on your answers to the questions above.

Edited by GS81Process, 27 January 2012 - 11:15 PM.

[Quados]

Thanks a lot for your answer.

Some clarification for your questions

1. Yes, Thailand has a very hot climate. The maximum temperature could lead to 42 degree C. and Yes we leave the suction line filled with liquid for a long time due to this pump is not used much.

2. One thing that conflicts my theory about temperature change while I ran a pump 2 days ago is that the temperature at the pump was reduced dramatically from 40 degree C to 28-29 degree C when operators opened the controlled valve under the tank before running the pump. it causes the pump temperature and tank temperature are the same. So, that means I still can cancel out the vapor pressure and tank pressure in the NPSH equation right?

3. Actually, we have 2 full port ball valves (normally open) in the suction line which are already accounted in my calculation. Moreover, we have a big T strainer in the suction line but I ignored it in my calculation due to the pressure drop between inlet and outlet is 0 kg/cm2. Is that the right thing to do? If i need to include pressure drop in strainer, what is the K factor for T strainer? or using 0.1 kg/cm2 in my calculation?


Thanks in advance

Regard,

Quados

[GS81Process]

So if the liquid propylene is allowed to heat up in the uninsulated suction line, are you causing a vapour lock before you start the pump?

[Quados]

So if the liquid propylene is allowed to heat up in the uninsulated suction line, are you causing a vapour lock before you start the pump?

No, we didn't have that problem before.

[GS81Process]

But you are commissioning the pump now and have not yet had successful operation, correct?

How are you able to rule out vapour formation in the suction line? DId you check the pressure drop profile to see if the vapour pressure would exceed the suction pressure at any point in the line? Is the suction piping sloped appropriately all the way to the pump suction or are there any pockets where vapour could be trapped?

"One thing that conflicts my theory about temperature change while I ran a pump 2 days ago is that the temperature at the pump was reduced dramatically from 40 degree C to 28-29 degree C when operators opened the controlled valve under the tank before running the pump. it causes the pump temperature and tank temperature are the same."

It sounds to me as if the propylene is heating up to 40C when left as a liquid in the uninsulated suction line, and then the temperature drops as soon as cooler propylene at 29C flows down by gravity from the tank.

Yes, you should account for the temporary strainer in your pressure drop calc. Is it a witch's hat? K factors for this are readily available in publications like Crane Technical Paper 410.

Edited by GS81Process, 28 January 2012 - 10:51 AM.

[Quados]

"But you are commissioning the pump now and have not yet had successful operation, correct?"

We used to run the pump successfully when the liquid level in the tank was high.

"How are you able to rule out vapour formation in the suction line?

We have a vent line at the suction can of pump which is normally opened and jumped to the discharge vent line which have already been modified to continuously rise back to the tank.

"DId you check the pressure drop profile to see if the vapour pressure would exceed the suction pressure at any point in the line?"

Honestly, I don't have a pressure drop profile to see it clearly where the liquid is starting to boil

Is the suction piping sloped appropriately all the way to the pump suction or are there any pockets where vapour could be trapped?

Yes, the suction pipe is sloped up to the tank with 1 cm/100 cm and i don't see any part of the suction pipe where vapor could be trapped

"It sounds to me as if the propylene is heating up to 40C when left as a liquid in the uninsulated suction line, and then the temperature drops as soon as cooler propylene at 29C flows down by gravity from the tank. "

i think so, and it seems to me that i can ignore vapor pressure and tank pressure in my calc. Anyway, i really don't know whether the existing vapor in the suction line can vent back to the tank completely when the liquid flows down by gravity from the tank.

Yes, you should account for the temporary strainer in your pressure drop calc. Is it a witch's hat?

No, it's not a witch's hat. It's like a basket type strainer

K factors for this are readily available in publications like Crane Technical Paper 410.

I didn't see the K factor for T strainer in that paper but I normally used 0.1 kg/cm2 for strainer. However, if i add pressure drop of strainer in my calc, the NPSHa will be lower than NPSHr. which means i need to remove strainer from the suction line right? because the pressure drop from fittings, valve and pipe is only 0.03 kg/cm2

regard

quados

[GS81Process]

Does the strainer vendor have a curve of pressure drop versus flowrate available? If so then use that as a reference.

If the NPSHa you calculate with the strainer is less than NPSHr, then you have likely found your problem. What is the strainer's mesh liner size? You might just need to determine what mesh size will give you an appropriate pressure drop, and switch to a coarser screen size.

Edited by GS81Process, 29 January 2012 - 10:54 AM.

[pavanayi]

Quados,
I have been following this thread from start. I do believe the problem with the pump is vapour generation. You mentioned in the first post that NPSHa is 6m while NPSHr is 4m. That itself is a close range for comfortable operation, with the temperature of the suction line varying.
You have also mentioned in the post 8 that the pump used to work successfully when the liquid level in the tank was high. This also points to the same direction.