Velocity Variation through Pipeline 3

[Discreet544]

Hi

Imagine there is a closed-loop pipeline which we are pumping pure water. Reasonably, we expect the pressure-drop to increase towards the end, due to friction and minor losses. My question, however, is about the velocity.

When you half the length of the loop, at the same pump power, velocity increases, as there will be less total friction and maybe less pump slippage. However, at the same pump power, should we expect pressure drop/length to be different through pipeline?

In my experiment, I am experiencing different kPa/m at various positions. As the D and f is the same everywhere through pipeline, it can happen only if the velocity is different from point to point. But, when the pump power is kept fixed, the fixed amount of water is expected to be pumped and the velocity should be the same everywhere. Am I right ? If not, and the velocity reduces as well through pipeline, how do you justify the continuity of fluid ?

Thanks;

 

[btrueblood]

"In my experiment, I am experiencing different kPa/m at various positions. As the D and f is the same everywhere through pipeline, it can happen only if the velocity is different from point to point."

Right.

You have at least two sources of non-axial velocity, the pump and each elbow (of which you must have at least two). The swirl induced by the pump and the bends, if not treated by baffles, will persist for quite some distance downstream of each. You can look at ISO-5160 for estimates of the velocity effect (at least as measured by a square-edged orifice) for elbows in terms of the length of pipe to allow same to "settle".

 

[Discreet544]

Thanks btrueblood

Could you please send me a link of ISO-5160. I can't find it anywhere. Searching also results in cabinet related topics !

That can be really helpful.

Thanks;

 

[zdas04]

The density of the water is very close to constant throughout the length of the pipe. Mass flow rate has to be constant (continuity equation) as long as no mass is added or removed.

If you have constant density and constant mass flow rate and constant pipe size then the bulk velocity pretty much has to be constant over the loop for a given mass flow rate. There can certainly be local pertubations, but the sum of their impact pretty much has to be zero.

David

 

[ione]

I can’t completely grab what you mean with “In my experiment, I am experiencing different kPa/m at various positions”. The fluid always moves from higher to lower pressure (nothing new at the sun).
Dealing with water which can be assumed to be kind of incompressible, density is practically constant and, due to continuity, equation is velocity (as stated above by zds04). Changes in velocity are related to different loops which produce different pressure drop and lead to a different working points of the pump (intersection between pump performance curve and system curve).

 

[BigInch]

Imagine there is a closed-loop pipeline which we are pumping pure water. Reasonably, we expect the pressure-drop to increase towards the end, due to friction and minor losses.
Not really. This is true for gas, but for water (nearly incompressible), the difference is practically nothing, hence velocity remains essentially the same at all points in a pipeline of constant diameter.

My question, however, is about the velocity. When you half the length of the loop, at the same pump power, velocity increases, as there will be less total friction and maybe less pump slippage.
Halving the length of the loop at the same (or much less) power, increases velocity and reduces friction. You can't say anything about the pump, because you don't know if it moves to a more efficient or less eff point at the new flowrate.

However, at the same pump power, should we expect pressure drop/length to be different through pipeline?

Yes you should, because with a change of pipe length the system curve coefficient to calculate H vs Q changes. With half the length, the coefficient is less, but you don't know the new increased flowrate. It is not 2x original flow. Power requirements would not necessarily remain equal for 2x flow. As velocity increases and, remembering power is a function of the cube of velocity, the same power is going to give you a flowrate less than 2x. As velocity increases, friction increases, so the pressure drop per unit length will be steeper too. You cut the pipe length in half so roughly the same total head drop is not happening over half the length, so your pressure drop per foot must go UP.


Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso

 

[katmar]

Are you sure "D and f is the same everywhere through pipeline"? Do you have the same grade of pipe throughout the loop? Even if you do, there can be variations in diameter due to allowable tolerances.

In turbulent flow the pressure drop is proportional to approximately the 5th power of the diameter. This means that a 23 mm ID pipe has 50% more pressure drop per unit length than does a 25 mm ID pipe. Does all the pipe have the same history? If some of it has been corroded the friction factor could be much higher.

If I had to rate various explanations for what you are seeing I would say:-
Most likely - Error in reading instruments or out of calibration instruments.
Possibility - Variations in pipe diameter and surface condition.
Very unlikely - Density of water is varying.

Katmar Software - Engineering & Risk Analysis Software

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[BigInch]

Sorry, "same total head drop is not now happening over half the length"


Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso

 

[jmw]

The point Btrueblood was making is that if you talk about velocity (without qualifying it), then velocity can vary because the flow path is not necessarily along the axis of the pipeline.
I assume that when you talk about velocity you refer to the mean velocity along the pipe axis.

If you just refer to velocity we ought to assume you mean point velocity. Point velocity can vary significantly.

Swirl is generated by elbows and other devices which means the flow follows a spiral path and hence the flow velocity in the direction of the flow is higher in order that the mean (axial) pipe velocity is constant.

This means the effective distance travelled is greater.

Another factor is that you cannot assume that all particles in the flow move at the same velocity and in the same direction.

Assume no swirl and straight pipe only:
If the flow is laminar then the fluid will be flowing parallel to the pipe axis but at different velocities across the pipe diameter with the slowest flow at the pipe walls and fastest in the pipe centre.

If the flow is turbulent then the flow will not be aligned with the pipe axis but disturbed. The flow velocity at any point and in the direction of flow may be higher than the mean axial flow. The flow direction will constantly change.

You will find that the flow at the pipe walls is less than the mean but away from the walls the mean flow velocity will be relatively uniform across the pipe.

But if you don't have just straight pipe, headloss is a function also of the various pipe components.
Elbows and fittings each add headloss.

If you have the same general layout number of elbows etc and all you do is reduce the straight pipe runs to reduce the distance travelled, then the reduction in headloss will not be a simple function of the distance travelled.

I assume from your expectation that because headloss changes that flow rate will change that you are using a centrifugal pump.
Now you have to establish where you are in the pump curve - the relationship is not linear.

A good reference to have is the Crane Fluid Flow Handbook.
Cheresources.com has a good section on working with this handbook in the student forums (Cheresources encourages students which Eng-Tips does not).


JMW

http://www.ViscoAnalyser.com

 

[jmw]

JMW

http://www.ViscoAnalyser.com

 

[btrueblood]

Sorry, I meant ISO 5167. There's an API standard, and a much older ASME standard that cover the same information. Or a good fluids book, or Crane TP 410.

 

[btrueblood]

Woops, jmw beat me to it.

One more thing, you stated that f is a constant, even though velocity is changing. In reality, f is a function of Reynold's number, and thus of velocity, though its change may be a small part of any variation in losses you may be seeing (at least, compared to the effect of bends).

 

[BigInch]

He simply means the average velocity across the cross-section, for nearly incompressible Newtonian flow at steady state, laminar, or turbulent, through a straight section or around a bend, v[sub]avg[/sub] is essentially constant for the entire length of a constant diameter pipeline made up of pipe with similar surface roughness. No need to cover all the bases when nobody is on base.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso

 

[jmw]

Well, the bases may not be covered yet but one question will always lead to another and it helps to be reminded that while a problem regarding pure water flow in a pipe seems simplistic, it may not be.
Is swirl going to increase, decrease or not affect headloss?
A simplistic view doesn't help one understand the contributions to headloss.
Here is an example of a paper that looks at the effects of swirl on headloss.

Incidentally, there is some detail on swirl effects in this on water transmission pipelines

JMW
[URL unfurl="true"]www.ViscoAnalyser.com

 

[BigInch]

of course it does and that is why we have tables and formulas for equivalent lengths of pipe to be included in the length of straight pipe x friction/foot to account for those extra losses due to swirl and momentum changing direction, etc. when fluid flows through ells, tees, reducers, ball and gate and globe valves and other non-straight-pipe elements.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso

 

[cvg]

nothing a better pressure gauge, an accurate flowmeter along with introductory courses in fluid mechanics and applied hydraulics can't solve

 

[btrueblood]

Big...

Those tables and charts are pretty generalized. Loss coefficients for fittings/valves are often based on measurements at the inlet and outlet flanges, and don't consider the losses due to dissipation of swirl, which can happen 10's of pipe diameters downstream. Then again, some tables and charts do consider such effects...

But, most of those charts treat the losses as mostly independent of pipe size, whereas the swirl and its dissipation can vary significantly as pipe size changes.

"No need to cover all the bases when nobody is on base. "

That's a bit dismissive of you and you know better than that, so does Dave. I think you two have been in the pipeline biz too long, and haven't dealt with enough elbows. Consider this an elbow...in yer ribs...

The OP said exactly what you and Dave keep harping on, which is that the bulk AXIAL velocity is constant. Duh. The OP went on to state that he does not see a steady rate of pressure loss along his straight sections - even though we all agree that the bulk average axial velocity is and must be constant. So what's different? How can he have a non-linear decrease in head when the tables and charts and equations all say that he should?

What is not constant is the true velocity, or total velocity, which can vary due to non-axial (i.e. radial and tangential) components, commonly referred to as swirl.

I'm right, and Jmw is right, and all you have to do is admit it. Come on, say it.

 

[Discreet544]

btrueblood: You can look at ISO-5160 for estimates of the velocity effect (at least as measured by a square-edged orifice) for elbows in terms of the length of pipe to allow same to "settle".
I have more than 35D entrance length, so the flow is already fully developed.

jmw: If you just refer to velocity we ought to assume you mean point velocity. Point velocity can vary significantly.
I am using magnetic flow meter, so the velocity I am referring to is "bulk" velocity. My test sections are two straight sections of lengths of 2.5m and 8.5m.

BigInch: Not really. This is true for gas, but for water (nearly incompressible), the difference is practically nothing, hence velocity remains essentially the same at all points in a pipeline of constant diameter.
Referring to standards for pure water (e.g.

http://www.engineeringtoolbox.com/pressure-loss-steel-pipes-d_307.html)

dp/dx for given D and V is gonna be fixed, independent of the position you measure it.

katmar: Are you sure "D and f is the same everywhere through pipeline"? Do you have the same grade of pipe throughout the loop? Even if you do, there can be variations in diameter due to allowable tolerances.
I am running a 2" diameter pipeline. Not really sure how much the diameter tolerances would be effective in this length !


Thanks Everybody for All the Helpful Comments. But let me explain my situation a bit more in detail:
I am working on hydro-transporting coarse materials (slurry of liquid-solid). For two years, I've been able to calibrate my loop based on water first, as I was able to get exactly the same pressure drops as in the link above. However, since end of September, within one week, everything changed and, now, I am seriously struggling with this issue (see the image below). By now, we have changed all the blade, plates, and sealings of the pump to remove the possibility of any sort of cavitations. Also, I have recalibrated digital pressure transducers, while they are confirming the results of parallel mechanical ones. I've checked the pipeline as well, step by step, to make sure there is no clogging or so on. Now, the problem is still there: I can't reproduce old results anymore.

I used to have only one test section of 8.5m. However, last week I made new positions for new gauges on the other side of the loop (2.5m test section) to compare with the old section. Now the results over there are also absolutely meaningless. The pressure drops per unit length are far different from the other section, also from the standard.

My toolbox of ideas is now totally empty, and I am in a real trouble ! If you kind guys can make any guess, please do so. That might be my answer!

 

[zdas04]

The "current" graph is really smooth. That wasn't true for your experimental data. In fact the current condition looks like the result of an algorithm instead of measured data (most of your experimental data that matched the standard looks like measured data except maybe Jason might have cooked his a bit).

Your current condition is about 0.4 kPa/m lower than the previous data. If I assume that your instruments are calibrated to +/-0.5% of full range, and that full range is 0-500 kPa then the uncertainty is +/-2.5 kPa and the difference is within gauge uncertainty. If the instruments are calibrated 0-50 kPa (which is a dang narrow range) then the uncertainty is 0.25 kPa for each instrument or +/-0.5 kPa, again less than the uncertainty.

This isn't feeling very real to me right now. The numbers are simply too tight for real gauge accuracy. This feels like a school project to me.

By the way, if you use engineering.com to upload attachments we won't have to deal with 20-30 pop-up adds and might end up being nicer.

David

 

[BigInch]

Referring to standards for pure water (e.g.

http://www.engineeringtoolbox.com/pressure-loss-steel-pipes-d_307.html)

dp/dx for given D and V is gonna be fixed, independent of the position you measure it.

Only true for incompressible, steady state, Newtonian flow ...no matter what the toolbox says. I didn't say anything else in the context you placed the problem. I said dP/dx would change AS YOU changed the velocity by virtue of changing the system curve when you halved the length of the pipe. Obviously the same head loss over half the pipe length gives twice the dH/dx as well as change the flowrate and the velocity!



Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso