Pipe/pumping question

[AG1412]

Hi,

I have design situation where I have one main pipe that receives flows from several pipes at regular intervals in length of the main pipe. So ultimately the main pipe must handle flows from all individual small pipes and pump it to the next unit in operation. The flows from several small pipes are equal. I have to calculate head loss in the main pipe.

My approach-
Size the main pipe to handle the flows coming in from every small pipe. I sized it to maintain a minimum velocity in the main pipe. Because the flow increases over the length of the main pipe (due to flows from small pipes), the diameter increases over the length. Calculate the frictional head loss for every section of the main pipe based on pipe material, velocity in that length of the pipe and diameter. And add the frictional headlosses obtained so over the total length of the main pipe. Is this correct?
The pipe does not flow through gravity. How to view this pipe network -series/parallel?

 

[BigInch]

Yes.
Series.
Any two or more pipes can be a network if you want to call them that.

you must get smarter than the software you're using.

 

[AG1412]

@BigInch, Thank you for your reply. I would like to elaborate and confirm about one more point.
When I am sizing each section of the main pipe, the velocity is not constant (slightly decreasing and increasing) in every section due to the available pipe diameters. My question is as long as the velocity is above certain limit, the small differences in velocities between the sections of the main pipe should not matter? Or does the velocity in the main pipe must increase over the length?


What I mean by each section of a main pipe is that-length of the main pipe between the locations where small pipes discharge the flow into the main pipe.

 

[BigInch]

The velocities will remain constant over the length of one pipe, if volumetric flow rate remains constant and your fluid is not very compressible. Velocity changes will occur at changes in diameter, or more precisely, changes in cross sectional area. Those changes will be more or less localized and rapid. Some relatively small amount of pressure will be lost, or could be gained if the velocity drops significantly. The amount of pressure change at those points corresponds to the gain or loss in velocity head caused by the change in velocity at those points. Velocities will of course also change if volumetric flow rates change, where two pipes join and add their flows before continuing downstream.

you must get smarter than the software you're using.

 

[AG1412]

@BigInch, thank you. Yes I understand that the velocity changes with the volumetric flow. I attached a very crude form of the pipes diagram that I have been referring to. Arrows in the diagram refer to small pipes. My question: is it ok to have an increased and decreased velocities between multiple sections of the main pipe. This would be having V3>V4 (velocities) in the attached figure and again with V5>V4. V5 is the velocity towards the end of the main pipe later to which the liquid is pumped into another unit.
This irregularities in Velocities between the sections is arising from choosing the diameters that are commercially available to match the flow. And the velocities calculated so are used in calculating the frictional losses.
Because I lack practical experience, I am wondering if it is ok to have irregular velocities in a pipeline such as in this example? If irregularities in velocities is ok in the pipeline, then one can further play with the diameter of the main pipe in several sections to bring down the velocities and the associated frictional losses.
I do understand that the velocity change happens suddenly and continues to remain the same along the length of the main pipe as long as there are no changes in volumetric flow rate or diameter.

 

[bimr]

PAK11, within the normal range of flows that are used, changes in velocity do not matter.

If you revisit the Bernoulli principle, you should understand that the energy in the pipeline changes form from velocity head to pressure, static head etc., but that the total energy remains the same.

 

[BigInch]

It is OK to vary the velocity. In fact with various diameters it is difficult to avoid. It is more important to keep velocities and pressure drops within acceptable values. For liquid lines it is customary to keep velocities between 1 and 10 feet per second. 1 fps, to prohibit deposition of particulate matter. Over 10 and to 15 ft/s, risk of water hammer increases. Even higher, assuming water hammer does not become an issue, the pressure drop starts to become excessive if the pipes are long. Erosion of the pipe interior wall can also become worrisome if the liquid contains dirt or other hard particles, which is one more reason to keep towards the lower velocities when possible.

you must get smarter than the software you're using.

 

[katmar]

Depending on the diameter and length of your main collector pipe it may not be worth using multiple diameters. The 2 main reasons for using a smaller diameter pipe are price and keeping the velocity high enough to avoid sedimentation. If your fluid is clean then only price is relevant. The cost of purchasing additional pipe sizes and reducers and welding them all together may outweigh the saving made on the smaller diameter pipe. If your collector is 3 metres long and invloves 25, 40, 50 and 80 mm diameters there is a good chance that making it all 80 mm diameter would be cheaper. If the pipe is 3000 m long and varies from 300 to 800 diameter then you will get savings by introducing different sizes.

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[AG1412]

@BigInch and @katmar- Thank you. In the design problem the pipe is ~4000m. Now I have a question about pumps. I am assuming that the flow from small individual pipes be pumped into the main pipe. Then I have to count the frictional losses in the small pipes too? If the main pipe is a distribution kind of pipe, then it is ideal to have pump at the upstream. In my example, because the main pipe collects the flow over its length and it is horizontal, how to arrange a pump to push the flow from the main pipe to the next unit. I am thinking to have one pump at the downstream of the main pipe. Am I right? As long as the flow is continuous from small individual pipes into the main pipe, there should not be any problem in the main pipe for liquid flow?

 

[bimr]

Start at the end point (discharge point) with the desired outlet pressure and flow and work backwards to calculate the pressure drops across each segment of the system.

Each pump must have a discharge pressure greater than the pressure in the header pipe at the connection.

You have to add the headloss in the header pipe to the headloss in each pipe from each pump.

The last segment will be the farthest pump. The last segment will also have the largest friction head loss.

Would you explain the application?

Is there a requirement for constant flow from each pump?

If you have multiple centrifugal pumps, the flow from each pump will vary slightly with the discharge pressure in the main pipe.

 

[AG1412]

@BIMR:It is an aquaculture application. Idea is to withdraw water at a rate from tanks, collect in main pipe, treat and return.I am assuming a constant flow. I came across this in one of the text books, but could not completely view it. So I thought why not calculate to understand the losses estimations in pipes. Yes, as I explained, I calculated head losses in the main pipe and in laterals at every point and added these while moving forward to the next lateral.

 

[itdepends]

I don't suppose you could use gravity to feed your main pipe could you? Having overflow systems on each tank (assuming you can run them full) which then gravity flow through the individual overflow pipes to the main header. The header would discharge to a break tank or directly to the treatment system if you have enough height to get it there. Then you can control the volume that comes out of each tank by controlling how much flow you return from the treatment unit to each individual tank.

As a chem eng/metallurgist the first part of any answer I give starts with "It Depends"

 

[bimr]

I am in agreement with itdepends.

If you have multiple tanks, you need a device to control the level in each tank. Otherwise, you may have a condition where flow from one tank moves to another tank.

There is no way that you would have equal flow from each tank unless you have equal flow moving into the tanks. This is accomplished with a splitter box with overflow weirs.

Maintaining liquid level in tanks is accomplished using weirs or pipe overflows.

 

[AG1412]

@itdepends- I did not assume gravity could feed the main pipe. Flows in all pipes is pressurized flow. I attached a figure showing approximate arrangement of 1) flows from individual pipes and header pipe to the treatment system and 2) return flows from the treatment system back to the tanks.

In figure 1, after the flow is collected in header pipe, it must be pumped to at least 4m height.I know the frictional head losses in the individual and header pipes. Based on this total head (static head+frictional head (converting into discharge pressure) I have to select pumps. I am thinking to choose a pump at every individual pipe with a discharge pressure greater than the discharge pressure at the connection with header pipe. Is that correct?, If so, in order to pump 4 m height from header to the treatment system, where should the pump be located. Do I need to have a pump/lift station to do that? How do I develop a system curve for the pump that lifts flow from header to the treatment unit.

In figure 2, when flows from two treatment units join together and to be pumped back to the tanks through a main distribution line, do I need to have a pump station? I know the frictional losses in the entire pipe network. How does pump arrangement in this case works?

If there is any textbook, design examples one knows to understand about this, please let me know.

 

[itdepends]

In Figure 1- why are the flows from the tanks to the header pipe pressurised? Is the pressure coming from a pump on each inflowing stream? Or is the tank pressurised?

If you have a tank pumping flow out of each tank into the header then you don't need a pump on the main header.

Sizing is done as follows- determine the frictional loss and head requirement to get flow through the header pipe to the treatment system. Set that pressure as the discharge pressure for the inlet feeders. Then size a pump for each inlet feeder based on its respective frictional and head loss with the extra pressure required to push flow through the header. You may need to specify individual discharge pressures if the header is long enough, or varies enough in height to impact the pressure that each feeder needs to push against.

Each individual pump needs its own flow control to (I assume) control the upstream tank level. That's why a gravity system is so much simpler. You put a sump that's low enough for all the feeders to gravity flow to and then a single pump to return the combined flow to the treatment system.

In Figure 2- A bit more complicated. I'd have a control valve on the outlet line to each tank- that would control the flow to each tank. Each treatment plant would have a discharge pump feeding the header. The pumps will need to have their own control valve or variable speed drive so that you can push more flow from treatment plant A or B depending on your needs. It's not an ideal setup but I hope you've got a decent control system/engineer. How critical is the flow split between the tanks on the discharge side?

I've designed a number of systems where I've had multiple dischage lines with a single pump controlling the header pressure but not two separate pumps from different systems where you need to control the relative inflow rates and then split it between multiple tanks- that will be fun.



As a chem eng/metallurgist the first part of any answer I give starts with "It Depends"

 

[bimr]

AG1412, you seem to be lost in the details.

Size the pump intake lines at 3 ft/sec and minimize the length of the pump intake pipes. Size the pump discharge lines at 5 ft/sec.

At these velocities, the frictional headloss will be small and insignificant as compared to the static head loss that you will be pumping against.

Calculate the total frictional headloss to the last fitting in each run of pipe. (For each pipe run.)

If you want to divide flows, use weirs or valves. It is too difficult to try to balance flows using the frictional headloss.

http://www.fao.org/docrep/field/007/af011e/AF011E09.HTM

 

[bimr]

Another link to assist you.

http://www.aces.edu/users/davisda/classes/facilities/presentations/introduction_to_pumps.pdf

 

[LittleInch]

Your system will (probably) achieve some sort of steady state flow. however what that flow is will not be anywhere near what your calcualtions are unless you start to think about how important it is to achieve certain exact flowrates or introduce some sort of control.

I would think seriously about PD pumps where flow is pretty fixed and forget about trying to balance a system with pumps and pipe. Far too many variables and big changes for small changes in temperature, head etc

My motto: Learn something new every day

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

 

[AG1412]

@itdepends- Thank you. Pressure is from the pump. Tank is not pressurized. I do not know any details about the ground levels. So assumed everything is pumped. These assumptions may be revised/revisited later once I know the details. The work I am doing is preliminary to assess how much operational energy is required to transport contents between the units in the site. My job here is to find about how much head is needed to be able to calculate the energy and make sure that it makes sense and seems realistic.
I did basic calculations pretty much similar to that you explained. My only question is how to know the extra pressure in the header pipe.

I am right out of school. All my school experience was in classroom completely different to what I am doing now. So it is a bit challenging to imagine and feel whatever I am doing is correct.

 

[AG1412]

@bimr-thank you very much for the links you provided. I will go through them and get back to you.
@Littleinch-Thank you. I considered your suggestion of PD pumps.