System curves

[dadfap]

Is a flat system curve typical of a discharge system that is characterised by friction losses rather than any vertical lift component?
In a flat system curve, a change in the speed of the pump by a VFD will not 100% follow the affinity laws?
Thanks
DADFAP

 

[TD2K]

The affinity laws apply to the different operating curves for a centrifugal pump at different impeller diameters or speeds.

You are talking about the system curve and how it intersects the adjusted pump curve (generated using the affinity laws).

 

[dadfap]

TD2K
I'm talking about the system curve. If it's flat, and I think that's where the greater % of the head requirement is friction, a change in speed via a VFD will move the pump operating point along the system curve. Therefor, if it's flat, rate will change but head remains almost constant
dadfap

 

[TD2K]

I'm a little confused about the comment then about the affinity laws but never mind.

Yes, if your system is such that the system curve is essentially flat then the flow will change and the head will remain constant.

I don't think such a system is where the greater % change is due to friction but perhaps I'm not following you. To me, a system where the system curve is flat would more where you are discharging to a constant pressure maintained system with minimal pressure drop through your piping. A piping system where a significant portion of the line losses are due to friction is not going to be a flat curve.

 

[BigInch]

Is a flat system curve typical of a discharge system that is characterised by friction losses rather than any vertical lift component?

A flat system curve is typical of a system that has relatively low friction losses in the piping system. If you use smaller pipe diameters, the system curve will become steep.

A "high lift component", or hi static head system, will shift the whole system curve up in relation to the pump curve; a low static head will shift the whole system curve down.

In a flat system curve, a change in the speed of the pump by a VFD will not 100% follow the affinity laws?

Pump affinity laws work for both flat curve systems and steep curve systems.

I think the answer you are really looking for is this,

VFD do not work as well with high static head systems as they can with low static head systems.

As you lower the speed using the VFD, the pump diff head will decrease to H2 = H1/(S1)^2 * (S2)^2
H1 = Differential Head at speed S1
H2 = Differential Head at speed S2

If the static head is high, the pump might NOT generate that high static head when you move to slower speeds. When diff head is not sufficient to overcome static head, the flow stops. Since diff head depends on the square of the speed, reducing speed a little decreases the head a lot.

If your system is 100% static head, forget the Variable speed drives (unless they're diesel engines).


Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[Artisi]

Your question is a little confused - but that's ok - lets see if we can clear it up for you.

The affinity law doesn't have any effect on a system curve, in a pumping system static head on the system remains the same as does the friction losses through the pipe.
If you were to change the pump characteristics by changing the speed or impeller diameter - what changes is the pump performance in relation to the system curve.

a typical system curve would start at the point of total head imposed on the pump at zero flow. I,E,. - no flow means no friction component in the system curve.

If you have a flat system curve it means that the friction component of the system is very low i,e,. pumping into a large diameter pipe in relation to the pump discharge rate.

A simplified example,

static head 100 ft
discharge pipe diam. 12" x 1000 ft long

point 1. of system curve

500 gpm = 0.6 ft friction + 100 ft static = 100.6 ft total head

point 2.

1000 gpm = 2.2 ft + 100 = 102.2 ft

point 3
1500 gpm = 5 + 100 = 105 ft

this is a flat system curve.

Note that the static head doesn't change - only the friction head changes.



Naresuan University
Phitsanulok
Thailand

 

[quark]

dadfap,

Constructing a flat system curve doesn't have any practical use. In the above case, Artisi proved a point where a high static head system (instead of frictional losses as you questioned)comprises of a flat system curve.

Generally, these system curves are plotted on a pump performance curve by starting the curve at 100 ft and then it takes the shape of a perfect parabola. The speed control band of a centrifugal pump should always be within its dynamic pressure range.

A centrifugal pump always follows the affinity laws at various speeds. Whether it is useful for you or not depends upon the system characteristic. In a case where static head is predominant (like Artisi's example), the speed control is a bad idea.

 

[BigInch]

I wasn't finished, just still thinking...

Pump speed control is not always a bad idea in all cases even if static heads are high. A good example would be when you would like to pump different products where their viscosities or specific gravities are quite different, or if the same product's viscosity varies significantly across your operating temperature range.

Pumping of Bingham fluids (catchup), where viscosity can change dramatically with flowrate is another excellent example that would have high head required to get flow going, but once moving, the head needed to keep it going decreases considerably.

Just goes to show you, like Abraham Lincoln said, "If I had 4 hours to do a job, I'd spend the first 3 thinking about how I was gonna' do it." ... or something like that.

Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[BigInch]

TIP: System curves are usually not perfect parabolas. Well, maybe you can get one in labratory conditions, if you have one product at uniform temperature. You could get a perfect parabola.

First one I can think of is when transitioning from laminar to turbulent flow. That produces discontinuities.

There's lots of non-Newtonian fluids, which give system curves that are radically non-parabolic. Catchup, Sulfur, slurries...

Classic examples would be for,

1.) Hot heavy Orinoco crude pipeline, which exhibits some degree of non-Newtonian characteristics and it also cools as it is pumped from the oilfield down to a marine termainal on the coast, so viscosity changes along the length of the pipeline. It can also transition from turbulent to laminar somewhere along that length as well, or
2.) For a liquid sulfur pipeline. Liquid sulfur is highly non-Newtonian and worse at some temperatures than others.

3.) Both of the examples are complicated by cooling as they flow down a long pipeline.

4.) A product pipeline with more than one product being transported at any given time of course would have a big discontinuity at the batch interfaces. Many product pipelines can have several batches contained at any one time. That would be a system curve that continuously varies with the density and the position of the batches.

5.) Two-phase flow varies with %gas to % liquid ratios

6.) A pipeline that must switch between packed flow and slack flow at some flowrate,

Let's see, how much more thinking time do I have left???


Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[BigInch]

Did I really say, "labratory"? Can you tell I went to HS with GWB?

Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[katmar]

The system curve has no effect on how closely the pump curves will follow the affinity laws. An analogy would be that in your car, the steepness of the incline you are climbing has no impact on the rpm vs power realtionship of your car's engine - but together the angle of the road and the power relationship of your engine will determine the speed you achieve relative to how hard you stand on the accelerator.

Also, the shape of the system curve is not affected by the static head - as BigInch clearly stated above all the static head does is move the system curve up or down on the H-Q plot, but the shape is fixed by the friction losses.

The missing bit of information in the earlier posts is that the affinity laws as stated above give the relationship between head and speed at the same flowrate. Using the affinity laws allows you to generate new pump curves for different speeds by calculating new heads for a range of flowrates. If the speed of the pump is decreased the new pump curve will sit to the left and below the base curve (on a standard H-Q plot).

So if you decrease the speed you have to look for the new intersection of the pump and system curves, and if the system curve is relatively flat it will occur at almost the same head but with a reduced flowrate. Which is exactly why you use a VFD to vary the flowrate.

regards
Harvey

Katmar Software
Engineering & Risk Analysis Software

http://katmarsoftware.com

 

[25362]

A short section from Sulzer's Centrifugal Pump Handbook:

When varying speed the following equations hold good in accordance with the law of similarity:

Q₁/Q₂ = n₁/n₂
H₁/H₂ = (n₁/n₂)²
NPSH₁/NPSH₂ = (n₁/n₂)²
............................................................................................

With small speed changes (up to 10%) efficiency remains virtually unchanged. With bigger speed changes the velocity in the channels alters and with it the Reynolds number. The efficiency factor must be downgraded at lower speeds and upgraded at higher ones.

 

[BigInch]

Katmar, I appreciate your confirmation of the above, and like your car-hill analogy. It helps to discuss things in terms related to day to day experiences.

And you are right, I did leave out the Q2 = Q1/S1 * S2 relationship, which is an important point that really should not have gone missing. That probably points to some neuron damage on my part.

But fortunately, I still have one sparking and its telling me that the relationship, H2 = H1/S1^2 * S2^2, between head and speed is not at the same flowrate. When you change speed from S1 to S2, the flow simultaneously changes to Q2 = Q1/S1 * S2. So when constructing the new system curve for speed S2, the old point, (Q1,H1) moves to the new point at (Q2,H2) concurrently with the speed change from S1 to to S2. As you note in your next paragraph, a new slower speed system curve at speed S2 would move lower and to the left, which is true, so if H2 was indeed at the original flowrate Q1, the new curve would only move downward and not to the left, so I think you already agree. Right?

Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[katmar]

BigInch, yes you are correct - the curve moves down and to the left and I was wrong to say "at the same flowrate". I realised this the second I hit the "post" button and immediately red flagged my own post but it takes time for these things to run their course.

What I should have said was - for each point on the base curve a new point at the new speed has to be calculated by calculating the corresponding head and flowrate from the affinity laws. The curve at the new speed is made by joining up these points and if the new speed is lower than the base speed the curve will be lower and to the left of the original curve.

Thanks for the correction.

Harvey

Katmar Software
Engineering & Risk Analysis Software

http://katmarsoftware.com

 

[BigInch]

Looks good.

My cursor on this laptop touchpad sometimes jumps off the screen. It's hit the post button a couple of times without my approval.

Cheers.

Going the Big Inch!

http://virtualpipeline.spaces.msn.com