Sizing of Overflow

[jcoots]

I have been charged with sizing the overflow nozzle for a large diameter storage tank. I've been given an input flowrate of 3,000gpm. Of course I need to make sure that my overflow nozzle is of adequate size. I think that as long as my outflow cross sectional area is equal to or greater than the inflow cross sectional area I should be ok. But, my limited work experience is telling me that this is probably not that simple. Can someone confirm my suspicion. What more do I need to look at?

Thanks,

 

[IFRs]

Depends on the type of flow and the head available. If the pipe is half or 3/4 full the flow is different than if it is flowing full with some head behind it. The fill velocity is probably higher than the overflow velocity because the fill is pumped with some pressure while the overflow is more like a wier, so equal areas will not work.

 

[JStephen]

Water tanks are typically designed for flow rates quite a bit higher than oil tanks. They will normally use a weir box inside. When the fluid level is just a few inches over the weir, the head at the overflow nozzle can be several feet. With this arrangement, it is usually possible to use the same size overflow as the inlet or outlet lines to the tank.

With just a nozzle, you may be able to size it the same as the inlet, but it will depend on how high the fluid can go above the nozzle without damaging the roof.

To check flow rates, apply Bernouli's equation between the fluid surface in the tank (zero velocity, zero pressure) and the outlet of the nozzle (unknown velocity, zero pressure). You can also figure in a local loss coefficient at the entrance into the nozzle.

If the overflow has a pipe running down the tank, check the nozzle itself as above, then check at other points down the piping, assuming zero pressure at different sections and finding the corresponding flow. With a long enough pipe, you'll have to start figuring in pipe friction as well.

 

[jcoots]

Thanks IFRs and JStephen

JStephen, let me see if I've got this right: If I take the Bernouli equation and set up the model you propose (fluid surface at p=0, v=o) and outlet at (v=?, p=0) that leaves me with

V2 (at outlet) = to [2*g*(z1-z2)]^0.5

If z2 elevation is 35ft, V2=47.5 ft/s

I then take V2 and work out a cross sectional area from the equation Q = V*A with Q=3000*gpm, A=20in^2

Should this be the cross sectional area of my outlet?

 

[JStephen]

First off, assume the overflow outlet nozzle is just a nozzle, with no pipe down the side of the tank. Then the z1-z2 term is the difference between the fluid surface while overflowing and the centerline of the nozzle- and should be just a foot or two, depending on the tank geometry. The z terms do not correspond to the tank height. Presumably you are not filling the tank 35' beyond the overflow nozzle.

If you have a riser pipe down the side of the tank and apply this same equation between the free fluid surface and the outlet at the bottom, it will give you much higher velocities (and I assume this is what you are doing above), but those velocities are not always achievable. If you take your calculated velocity in that case and calculate pressures up the drop pipe, you will find a partial vacuum or a negative absolute pressure. With the partial vacuum, you don't know if that vacuum is actually maintainable- if not, you won't get that flow rate. (Presumably, the fluid would be falling in globs down the pipe, as opposed to the pipe running full of liquid). The negative absolute pressure is physically impossible, or at least with the assumptions normally made in fluid problems. So typically, with a riser down the side of the tank, check velocity as if it were a nozzle only, and assume it may be limited to that amount. It might be further limited by pipe friction.

 

[jcoots]

Thanks again for the reply. In response I want to be sure I understand you correctly. I've tried to draw the FBD below. What I understand you saying is that the fluid surface will be 1-2ft above the nozzle. What dictates 1-2ft? Or is it more user defined being that if the fluid surface went any higher it would mean that the floating roof would hit the fixed roof, or come out of the tank?

|_ _ _ _z1_Fluid Surface _
|
V2,z2_____| 2ft
<----_____ ------
Nozzle |
|
|
|___________________Tank Bottom

If I use the assumption of 2ft as being the difference in elevations, would I then follow through with the method in my previous post. Using 2ft I come up with a req'd cross section of about 85 in^2 which sounds more reasonable than 20in^2

 

[IFRs]

The elevation of the overflow nozzle is determined by operations. Lower the nozzle = lower storage capacity and smaller nozzle to match the inlet rate. Raise the nozzle = more capacity and larger nozzle. API has guidelines on preventing overfill. You have to know your fill rate, tank size, floatign roof thickness, fixed roof thickness, lowest obstruction, time to shut off the valves, etc. It's definitely not arbitrary and using 24" could be way wrong!

 

[JStephen]

The height used would be based on actual tank geometry, whatever that happened to be. You can adjust the nozzle height or make the tank higher, but either decision impacts capacity and cost of the tank.

 

[jcoots]

Thanks for the education JStephen and IFRs. I always appreciate a good learning experience.

Just to confirm everything: The height above the overflow nozzle is determined by the how much room I have between the fluid surface and the nearest obstruction that will interfere with my roof. My calculation in my previous post should work and tell me what x-area I need for my overflow.

So, from your posts I can infere that if the roof is allowed to float way above the overflow nozzle without causing any problems then I have a lot of wasted space. I should want that elevation to be a safe but small number, right?

Thanks again.