Selecting a Pump Based on Flow Rate and Line Pressure

[MechEng176]

Hi Everyone,

Been following eng-tips for awhile now and finally made an account to ask a question. Junior mechanical engineer here working on my first project. I'm working on a project where I have to pump 500m3 of oil(or any other hydrocarbon) from tanks into a pipeline operating at 500psi in 1hr (so 8m3/min). For the setup I have tanks, centrifugal pumps(to feed the PD pumps), positive displacement pumps, 60m of 4'' pipe, a metering skid (rated for 10MPA/1450psi), and then the pipeline (which has the same fluid flowing through it at 500psi).

I used a head loss calculator from

http://www.flowoffluids.com(see

attached picture)and have determined that at the rate the fluid will be flowing, there will be a pressure drop of 140psi and head loss of 480ft through the 60m section of pipe.

When speccing the pump, I was told to look for a pump that will put out 10MPa at 8m3/min.

So my questions are:

1. If I add the 140psi drop to the lines 500psi operating pressure I get 640psi. Does that mean I need a pump that will put out atleast 641psi?

2. Will the 641psi pump be able to push my fluid at 8m3/min into the pipeline which is at 500psi?

3. Where does hp come into pump selection?

4. Where does the head loss of 480 feet come into play?

5. Looking to order the Flow of Fluids Through Valves, Fittings, and Pipe TP410 by Crane. Yay or Nay? Any other book recommendations?

http://www.amazon.com/gp/product/B0...pf_rd_t=101&pf_rd_p=1389517282&pf_rd_i=507846

 

[petrochemical87]

I must say I do not fully comprehend the problem. You say "I have tanks, centrifugal pumps(to feed the PD pumps), positive displacement pumps".
Does this mean you intend to use centrifugal and PD pumps in series to achieve a flow rate of 500 m3/hr @480 feet. I believe this can be done with centrifugals alone . Please clarify
1. If I add the 140psi drop to the lines 500psi operating pressure I get 640psi. Does that mean I need a pump that will put out atleast 641psi?
Yes your pumping system needs to deliver 500 m3/hr against the sum of static head of 500 psi and frictional head of 60 m of piping(which you have calculated as 480 ft)

2. Will the 641psi pump be able to push my fluid at 8m3/min into the pipeline which is at 500psi?
make a system curve and match it with your pump curve

3. Where does hp come into pump selection?
You have to choose the most economical pump for the duty!!

4. Where does the head loss of 480 feet come into play?
This is frictional head for your pump design

5. Looking to order the Flow of Fluids Through Valves, Fittings, and Pipe TP410 by Crane. Yay or Nay? Any other book recommendations?
this is a mandatory read for all pump engineers

 

[MechEng176]

The centrifugal and PD pumps will be in series. According to the operations folk I talked to, our PD pumps can't suck so they need a centrifugal pump to feed them the fluid.

Also, since our PD pumps can only go 2m3/min (since it's oil and we don't want to burn up the packing), we will have 4 PD pumps in parallel all fed by the same centrifugal pump.

2, 3, 4, Understood.

Thanks for your input and I will definitely be picking up a copy of that book!

 

[petrochemical87]

I think it is possible to get a centrifugal pump for your duty(500m3/hr,480ft). I dont think you require the 4 PD's in series.

 

[Latexman]

Is there an elevation difference between the low operating level in the tanks to the pipeline tie-in point? If so, you have to account for that.

Good luck,
Latexman

Need help writing a question or understanding a reply? forum1529

 

[Latexman]

Also, and this is probably minor, but the pump also has to increase the velocity of the oil in the tank (v = 0?) to the velocity in the pipeline.

The two things I mention can be calculated using the Bernouilli equation.

Good luck,
Latexman

Need help writing a question or understanding a reply? forum1529

 

[ccfowler]

Here is some advice, statements of prejudices, etc. from an old geezer:

First, do your own calculations. All the handy little Internet gizmos, canned programs, etc. can be useful and sometimes a bit overly handy, but you never know what idealizations, assumptions, limitations, ... may be imbedded that can give you a misleading or outright wrong answer.

Second, pumps in series are usually not the best configuration. I would not use such an arrangement without some compelling reasons.

Well chosen centrifugal pumps are usually the most efficient and reliable option, but all types of pumps exist because there are very good reasons for all of them.

You suggest that your pumping system may be handling a wide range of fluid characteristics, so some form of single stage positive displacement pump may be a better choice. Centrifugals are not my first choice where fluid viscosity gets to be a problem--that is where PD pumps usually serve better.

Centrifugal pumps do not produce pressure, they generate head with pressure being related to fluid density or specific gravity.

Also, pumps do not suck--period! All pumps rely on sufficient pressure being available to cause the fluid to flow into the pump inlet. Always make sure that you fully understand the implications of NPSHa and NPSHr for the specific situation.

Be sure that you fully understand the actual characteristics and range of characteristics for the pumps that you are considering. Published data will be accurate, but it can be very misleading if you do not fully appriciate the potential implications of industry standards regarding such data. In my experience, pump manufacturers and their representatives can be of enormous help, but that help will only be fully realized when you hold up your end by being well informed and completely honest about all aspects of your complete system's performance characteristics. So called pump problems are almost always due to the connected piping and equipment rather than the pump.

Without being silly about it, don't skimp on motor size where varying conditions are likely. Some modest sacrifice in motor efficiency is far better than a burned out under-sized motor.

The Crane publication is excellent, but don't rely on it exclusively or use it without careful thought about your specific application.

Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.

 

[LittleInch]

This site is called Eng-tips, so let me give you a few tips which I hope you can take forward or revise your post a bit.

1) Please try to use consistent units as it makes it confusing and you are then prone to make mistakes. So far you have psi, MPa, ft, m, m3/min, head, pressure. I personally use mainly metric units as you also need to be aware there are two gals, US and imperial. US engineers tend to forget that both exist and just quote gpm meaing US gpm, which is a bit smaller than imperial gpm. Typical units used are gpm, m3/hr, bbls/day, l/sec. Mixing these up can cause a lot of confusion.

2) Unfortunaltey there are many discrepancies and "holes" in your data. So far you have provided about 20% of the required information for anyone to give a reasoned view on your issues. You need to think about the system curve (no pipeline will accept 500 m3/hr at a fixed pressure without this additional flow affecting the pressure input). You need to provide a system diagram or sketch so that we can understand what is where in your system.

3) You appear to pumping petrol (gasoline) which has notorious NPSH problems, especially on low tank head, high temperatures (>20C) and high elevations (>750 masl). However you then mention other hydrocarbons. Beware for all pumps that head remains the same for centrifugals, but pressure and power increase with density, therefore if this is going to be used by multi products, you need to specifiy the worst conditions on your pump data sheet.

4) I'm surprised no one has commented so far on your totally ridulous pressure drop and velocity down 60m of 4" line. This is just enormous - I calculated 16m/sec and this will just burn power and would have truly horrenodus surge pressures. For 500m3/hr you need at least a 10" or 12" pipe to get your velocities down to around 2-3 m/sec and hence your pressure drop and friction losses down.

5) Centrifugal to PD pumps is very odd. Be aware that PD pumps at fixed motor speed pump fixed volume and pressure can be very high if the downstream resistance rises. However you then say you were told to look for a pump doing 10 MPa (100bar or 1440 psig). This is riduclaous for a pressure demand of 500 psi, but would seem to eliminate the need for the PD pumps. what exactly are you being asked to design??

6) A tip - Beware of "operations folks" telling you things. I love operators and they do a great job, but many do not understand the concepts and basics behind fluid flow. They just know how their system works, even if it is not the way that it was designed or intended to work... If a PD pump can't "suck" and I agree with ccfowler use the correct terminology, then there is something serously wrong with your system which needs investigation.

7) PD pumps at 2m3/sec (125 m3/hr) doing this flow as "we don't want to burn up the packing". Can you explain what this means as it doesn't look good to me.

So, draw out your system, with pipe OD, lenght, vertical differences, liquid level, location of pumps, type of pumps existing now that you want to re-use, pressures, flow etc and then come back to us because at the moment it simply doesn't make sense I'm afraid. You need to work out your system curve as the first thing i.e. the static head plus frictional resistance of your pipe system downstream of the pumps plotted as m head vs flow. Oh and make your 4" pipe a 10" pipe and always look at velocity.

I hope that you are coming here to find a few things out before you talk to your supervison / lead engineer / manager as they should be guiding you better than we ever can on somehting as basic as this.

Good luck and hope this helps

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[MechEng176]

Thanks for all the comments and input. And yes, I'm coming here to get a handle on things before meeting with my supervisor to hammer out the details. I realize now I left out a lot of info, in my head it all made sense but that's only because I know what we're doing.

The reason centrifugal pumps are out of the question is because the client insisted on PD pumps. I'm sure there was a good reason but it's escaping me right now.

4. For our l6m/s fluid speed in 4'' piping. The reason we went with 4'' for now is because that's the absolute minimum NPS we are willing to consider because it is readily available to us. Will try to go bigger to reduce pressure drop and friction losses.

5. We're designing a system to reinject oil from 400bbl tanks manifolded together back into a pipeline (just 500m3 of product, that's it, one time deal). The tanks outlet will be connected to a blender unit(blenders are used in well fraccing to mix fracturing fluids before feeding them to frac pumps) which will feed the 4 PD pumps. Reason I was told to look for upto 10MPa is because that is the maximum pressure that a piece of equipment between our temporary piping and the main pipeline can handle.

7. The ''burn up the packing'' was said to me by an operations personnel when I said I'm looking for something that can put out 10MPa at 8m3. He said that their frac pumps can normally do 2m3 but since it's oil he wouldn't go over 1.5m3 to avoid burning up the packing. I figured he meant oil just gets hot when it's being pumped that fast and hard.

I'll try to hammer out something much more detailed (length, type of pumps, pipe ID, layout) in the meanwhile so you're not forced to assume things.

My takeaways so far have been a lot of good info and that I still have a very long way to go. Going to be lurking these forums for awhile.

Thanks again,

 

[1gibson]

So the word "viscosity" only appears once in the thread (twice now) and was mentioned in passing, a comment on selecting centrifugal vs. PD. That is a problem when you are pumping "oil or any other hydrocarbon."

Your first step is to fully define the complete range of fluid properties (temp, specific gravity, viscosity, vapor pressure) but you've skipped that completely and started off listing various flowrates, pipe sizes, pressures, etc.

Step 1a: Don't skip step 1
Step 1b: Fully define fluid properties
Step 2: Start to define system properties (pipe size and length -> head loss, suction conditions, discharge conditions) so you can create a system curve
Step 3: Now you can start trying to pick some pumps

P.S. I'm going to make one quick correction to your last post, you will thank me for it later, and you will be a better engineer if you remember this:

I'm sure there was a good reason but it's escaping me right now.

 

[LittleInch]

3a, I'm taking a guess here that the client understands that pd pumps are essentially constant flow pumps which have a max pressure, hence the 10 MPa quote. This means he doesn't have to worry about the system curve.

Centrifugal pumps on the other hand are essentially constant pressure pumps, + / - 10% so if you get the system pressure wrong it either doesn't work or you go off the end of the curve and flow 2 to 3 times as much flow.

4) Good, you won't regret it

5) See above

6) ?

7) The only way you can make pd pumps pump less flow is either a VSD drive or a lot of flow lost via a re circ valve. I'm still non the wiser.

LI

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way