Centrifugal pump loosing suction during tank changover

[omkar32]

Hi,

I need insights into the following problem. My problem is as follows:

A centrifugal pump is feeding to a crude unit @600 m3/hr. During tank change over, at first, the gate valve on the new tank is opened 25% and the gate valve on the feeding tank is closed by 25%. Then further 25% is opened on the new tank and to a similar extent, the feeding tank is closed. This process is repeated till the new tank is completely open and the feeding tank is completely closed. The reasoning given by the field operators to carry out feeding tank change over as above is if they open the gate valve on the new tank by 100%, the crude will gravitate from the new tank to the feeding tank and the feeding pump will lose suction. The field personal say they have experienced this phenomenon a number of times during normal operations.

I feel the above condition (feeding pump losing suction) should not arise considering there is a positive head in both the feeding tank and the new tank compared to pump suction. Is the reason given by field operators correct? Will the feeding pump lose suction if the gate valve on the new tank is opened by 100% without closing the feeding tank?

 

[razookm]

If the new tank is opened 100% without closing the feeding tank, tank level balancing will happen at a huge rate. That is there will be a flow to the feeding tank from the new tank because of the level difference. This may not be a preferred situation. The feeding pump losing suction is not likely but cannot be confirmed unless the complete arrangement is known.

 

[LittleInch]

This needs a diagram with lengths if possible but how come the outlet line from the tank to the pump doesn't have any check valve / non return valve?

Also size and speed of valve opening is important.

A gate valve even 25% open will effectively be about 75% open for flow restriction purposes.

But you're right - why would there be a big reduction in pressure isn't clear. It may be a result of the particular flow path, but seems a little odd.

Or if the line is two way, why isn't there check valves on the manifold upstream the pump??

This sort of thing is usually addressed as reverse flow in the HAZOP.



Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[omkar32]

Thank you razookm.

I agree that crude will gravitate from new tank to a feeding tank at a huge rate. Considering that the feeding tank valve will be closed, say 1 minute after the opening of the valve on the new tank the level increase in feeding tank may be acceptable.

I am attaching a simplified layout of the arrangement. See whether any comments can be drawn.

 

[omkar32]

Thank you LittleInch.

The line is common suction line from tanks to Pump. All the tanks suction is connected to this common suction line (24").

It is true that NRVs might have been provided to avoid reverse flow, these were not provided.

 

[razookm]

I presume that the height mentioned in the sketch (11m and 3.1 m) is the liquid level.
As you have mentioned that operators are confirming that they have faced this issue during operation, let me try to find a possible explanation for this. please check and decide for yourself whether this is a plausible reason.
The crude oil pressure at the bottom of the new tank will be around 1 barg considering the liquid head. The pressure will drop at the common suction if the valve is opened 100% and the crude oil gravitates to the feeding tank. Even for stabilized crude oil, there will be some dissolved gases present and these gases could be flash when the crude oil pressure reduces suddenly and eventually may cause cavitation in the pumps.

 

[LittleInch]

NRVs should have installed - not might have. That's a serious oversight.

However still can't see why this would result in loss of suction pressure / head to the pump.

But it's quite a long way from the tank to the pump so at low level it will be close to the minimum allowable inlet head / NPSH so might not take much to disturb it. If the line to the pump is actually a tee off then you might get some sort of venturi effect from a high velocity of oil going from one tank to the next.

Fitting NRVs should solve it though. Uncontrolled flow from one tank to the other isn't a great idea whatever the issue.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[razookm]

Agree with LittleInch that Uncontrolled flow from one tank to the other isn't a great idea whatever the issue.

 

[omkar32]

Thank you razookm and LittleInch for providing your inputs.

 

[crshears]

A possible technique would be to observe the pump suction pressure gauge [hopefully there is one] while closing the incumbent supply valve, stopping as soon as a slight drop is noticed.

Open oncoming supply valve until pressure returns to original value.

Iterate until changeover is complete.

It may take more than one person in constant walkie-talkie or other communication to complete this process, perhaps even one closing the first valve, one opening the second, and one observing the gauge and directing the other.

CR

"As iron sharpens iron, so one person sharpens another." [Proverbs 27:17, NIV]

 

[don1980]

This sounds like it may be a case in which this crude oil has some non-Newtonian characteristics. Such fluids resists flowing until they start to flow. For a non-Newtonian fluid the apparent viscosity decreases as the fluid starts to move ("shearing" starts to occur).