9 replies to this topic

[ulka_19]

Hi All,

 

I am specifying a centrifugal pump (60m3/hr) to carry Naphtha, Motor Spirit, Ethanol, Methanol fluids, Vapor Pressure- 0.6 bar, the storage tank is at atmospheric pressure and the minimum liquid level in tank is 1.2 m and the tank elevation is 1 m. Suction line is 175 m long and various head losses are around 3.7m due to fittings and valves. Line size is 4inch. Pump suction level is 0.8m. Taking all losses in consideration I am getting a negative NPSH value of 3.04m. However my line size is fixed so please suggest what i can do further to make it positive. Pump suction level doesnt help much. I cant go for a can pump.

 

Waiting for reply.

[breizh]

Hi,

 

You may post your calculation and a sketch , someone will comment.

my assumption ( density # 800 kg/m3 & viscosity # 2 cpo) leads to 1 bar head loss ...!

 

Reducing the flow rate by 25% should reduce the head loss by 2 >>>> is this concept acceptable?

 

Consider the search feature top right , you may find good support .

 

 

Breizh

Edited by breizh, 26 November 2013 - 09:24 PM.

[fseipel]

What is fluid specific gravity?  Assuming 0.7 SG, I come up with ~-4.4 m NPSHA close to your -3 m NPSHA:

 

NPSH_a = p_atm / γ - h_e - h_l - p_v / γ

where p_atm is atmospheric pressure

γ is specific weight of the fluid

h_e is elevation from surface to pump

h_l is head loss due to friction, from surface to impeller

p_vis fluid vapor pressure

 

NPSH_a = -4.4 m = 101.33kN/m^2/(9.8 kN/m^3*0.7 SG)+1.4 meters - (3.7 meters + 8 meters) - (101.33kN/m^2/1 bar*0.6 bar /(9.8 kN/m^3*0.7 SG))

 

where 1.4 meters = 1.2 m + 1 m - 0.8 m and

3.7 meters = friction losses in fittings and 8 meters = friction losses in 175 m segment

If you could indicate pipe schedule, include the requested diagram, and confirm 3.7 meters = losses from valves & fittings only that would be helpful.  Also include pipe type/material or outright roughness factor -- usually not so important but here you have an extremely long suction so it's very important.

 

If you half the flow rate you ought to be able to decrease the head loss in the straight section by about 6 meters at least getting you into positive NPSHA (please calc exact value for your mixture; that's a rough figure based on water and this mixture is not water, and also recalc the losses through valves/fittings at this reduced flow rate).  What is pump NPSHR?

 

Preliminary assessment indicates your calculations are correct.  Thus you may want to consider throttling pump discharge -- hardly ideal but it lowers NPSHR (due to shift leftwards on pump curve), and increases NPSHA -- increases NPSHA because if you throttle it enough to half the flow, the friction losses in the long suction line are greatly reduced.  If you can't half the flow rate then you need a booster pump closer to the tank or a larger suction line, higher tank liquid level, or need to cool liquid to lower vapor pressure, etc.  Also halfing flow is an approximation -- I don't know exact roughness factor of pipe or fluid properties, the halfing assessment is based upon water.  You should assess NPSHA at several flow rates, say, 25%, 50%, and 75% of desired flow, and compare to NPSHR at same flow to determine throttling required.  Or just run pump and observe where it cavitates and throttle discharge back slightly more to have a safety margin.  Another option is to use an eductor (nozzle) to boost the NPSH at the pump inlet.  This is a bit of an exotic technique, you will suffer an efficiency loss, because of the pumparound flow to the eductor, and eductors aren't cheap.  Jacoby Tarbox, Croll-Reynolds, Fox Eductor and others can supply nozzles.  I have used this technique to pump down tanks that are under vacuum, such as rotary vacuum filter separators, as an alternative to elevating them to provide adequate NPSHA.  If you decide to throttle, consider a locking stop on a gear actuator on a manual butterfly valve (so operator can't open it to where it cavitates), and/or a pressure transmitter on discharge and/or suction, and perhaps a control loop to automatically throttle.

 

Breizh: Yes, that sounds right -- Reading off a water table such as http://www.engineeri...ml#.UpIdhT-zmlo, at the 60 m^3/hr flow rate, pressure loss is 40,000 Pa/100m x 175m/100m = 70,000 Pa = ~7.1 meters + 3.7 meters friction loss in fittings = ~11 meters = ~1.1 bar since 10.33 m w.c. = 1 atm.  Correcting for density takes this to ~1 bar -- I suspect viscosity is close to that of water, but if OP could provide composition also that would help assessment.

 

One other comment: If you want to get fancy, and have a highly-instrumented plant, you can also throttle the pump [discharge] automatically or decrease its speed when tank level decreases, if it's on a frequency drive. 

Edited by fseipel, 24 November 2013 - 09:09 PM.

[S.AHMAD]

1. Since the line size is fixed then the only choice (short term solution) is to revise the minimum liquid level to 1.2 + 3 = 4.2m (or slightly higher to include safety factor) . This limit the operating volume and increase the "dead" volume of the storage tank.

2. As a long term and optimum solution you need to revise both minimum level and the pipe size. Other options are expected to be more costly.

Edited by S.AHMAD, 24 November 2013 - 06:50 PM.

[ulka_19]

Thanks for your replies. Fluid specific gravity is 0.87 and viscosity is 5cp. Tank is receiving these products from jetty so for tank unloading process theses pumps are specified. Reducing pump flow upto 25% gives me (-1.49) and required npsh is +2.1m. Piping Material is API5L Gr.B sch.40.

3.7m head loss includes loss due to strainer and 1 valve, 12 tees, two 90 deg elbow and two 45deg elbow.

Edited by kankshu19, 24 November 2013 - 11:20 PM.

[fseipel]

Okay, you're probably looking at very close to 250 meters of equivalent length for straight pipe+fittings -- I don't know exact valve type/strainer so can't pin down their L/D.  With your additional details, my initial estimate of halfing flow seems overly optimistic; to have 2.1m+0.5m safety margin, flow would be more like 15 m^3/hr or 25% of required flow.  However I wasn't clear how much the NPSHR dropped between 75% of desired flow and 25%.  I'd repeat the formal calculation at 25% and 50% of desired flow and interpolate between the two to determine the operating point. What is preventing you from setting pump closer than 175m away from tank?  You state you're 'specifying' the pump, but the remainder of line (size) is fixed, to me that indicates a new pump?

[breizh]

Based on your data ( 870 kg/m3; 5 Cpo,; TV =0.6 bar , Leq # 180 m ;source elevation :1.4 m) , considering 0.1 mm absolute rugosity :

Qv (m3/h)  :                         60    ;  30

Head Loss (mbars) :           819   ; 290

NPSH a      (mcl)     :           -3.3   ;  2.8

NPSH r       (mcl)     :           2.1    ;  2.1

 

Short term , reduce the flow rate !

 

Breizh

Edited by breizh, 26 November 2013 - 09:59 PM.

[ulka_19]

Hi fseipel,

 

Actually there are total eight tanks which have to be unloaded through 4 pumps and these pumps are located in a pump house. Suction line of the last tank is the longest line and equal to 175 m. So basically this NPSH problem is arising for last two tanks but we cant make seperate arrangement for those tanks. One specification has to match for all tanks. SInce line sizing was done earlier and pipes have been ordered so line size cant be changed now otherwise with 6" line there would not have been any problem at all. Halving the flow rate will increase the time for unloading. One tank is of capacity 4800m3 and there are total eight such tanks (4800*8=38400m3). I suggested for VFD pumps and lowering the flow rate as the level gets down as u mentioned but I think that option is going to be costly.

[curious_cat]

What was the original designer's basis for using small diameter lines? And does the pump house have to be so far from the tanks? You could have a smaller capacity booster / low NPSHr pump to be used only for the two problem tanks when level drops below a certain point. 

[Bayo Alabi]

Hi ulka_19

 

The following site may be of help:

 

http://www.process-d...centrifugal.jsp

 

It can also be used in what-if scenarios.