SYSTEM RESISTANCE CURVE NOT INTERSECTING PUMP CURVE

[davincigee]

Dear guys,
A recent project was undertaken and I was asked to review the design of the pumping system and determine if it was appropriate or not. I determined the various parameters and came up with a graphical representation of the system resistance curve superimposed on the curve of the pump that was selected for that project. I have attached the file here. It is obvious that the pump curve does not intersect the system resistance and thus the design was not done appropriately. Nonetheless practically, the pump works! Can I get a thorough analysis from any of you pump experts? What is actually happening? Why is the pump working in reality?

 

[1gibson]

I would imagine the answer is "because there is a valve after the pump discharge."

 

[77JQX]

You're running at the extreme end of the pump curve, where the manufacturer decided to stop drawing it because they don't recommend running there. Just because the manufacturer stopped drawing the curve doesn't mean the pump won't operate out there if you're willing to push it there. Running out at the end like that could be expected to shorten the life of your pump. Off the top of my head, things to look out for are suction cavitation (check NPSHr and NPSHA), excess bearing wear, poor efficiency, and an overloaded motor.

 

[Pumpsonly]

In addition to 1gibson and 77JOX comments, did you verify with site measurement of the actual differential head and flow of the pump is operating. The system curve is not far from the end of the curve. Your system curve could be under estimated.

 

[Artisi]

Why is the pump working, because there is nothing stopping the pump from running where it is operating, as already pointed out - the pump curve will extend well beyond where it has been stopped (who stopped it there you or Sulzer?)

You need to show us the REAL pump curve not some graphic representation, showing the full H-Q curve, power curve, NPSHr together with the site conditions including NPSHa etc.
The two line plot is next to useless for anything meaningful.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[LittleInch]

I agree with all of the posters above. Your pump is operating at the end of its curve as drawn by the vendor, so on that basis will be doing about 65 (m3/hr?). You need to go and get some more info to prove it, but if, for instance your system curve started a few metres higher then the flow will gradually come into a more normal point on the pump curve. At a guess this pump is pumping into a tank - are you looking at the low level case and in fact the pump normally operates into a tank with a higher level than that? What is the height range of the liquid in the tank? Draw a max tank height case and a min tank level case and then re-post.

One easy way to see what is happening to your pump is to go check the amps it is consuming. Nearly all switch boards show this or record it as a standard feature so you don't need to go and fix any new instruments.

My motto: Learn something new every day

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

 

[chicopee]

Assuming all elements that were used to develop the two graph curves, you may want to explore the affinity laws governing this pump. At first I would derive a new curve based on reducing the pump RPM. The new curve will assume nearly a similar shape as the original one but will drop below the original one. Pump efficiency should remain the same at any point.

 

[TenPenny]

I have attached the file here. It is obvious that the pump curve does not intersect the system resistance and thus the design was not done appropriately. Nonetheless practically, the pump works!

Technically, if the pump works, then the pump curve MUST intersect the system curve, because the point where you are operating is the intersection of those two curves.

In order to give any meaningful analysis, you'd need to supply at the very least the actual pump curve, and also verify operating speed as compared to the speed shown on the pump curve.

 

[BigInch]

A pump curve and system curve do not have to intersect, as in the case where the system curve has a high static head, say 100 ft, but the pump has a shutoff head of only 50 ft. Oh wait, it intersects when the pump is in reverse flow!

I hate Windows 8!!!!

 

[TenPenny]

A pump curve and system curve do not have to intersect

No, you're right, they only have to intersect in order for the pump to do anything.
As long as you're happy for the pump to not actually do anything at all, they don't need to intersect.

 

[1gibson]

Definitely agree with checking power, pump curve shape looks like low enough specific speed that power should be continuously rising with flow.

 

[davincigee]

Thanks a lot guys for the inputs. Also, LittleInch, the pump is not pumping into a tank. It is delivering to a loading gantry that has 18 loading arms. And infact guys, the static head is the difference between the height of gasoline in the tank (21m) and the highest elevation of pipelines at the loading gantry (3m). I'd wanted to attach the excel sheet I generated for obtaining the pump curve and the system resistance curve but I might be confusing. Nonetheless, I am pretty sure that I performed the right calculations. In fact, I assumed the highest case scenario where the tank was filled to the brim i.e. 21m. It thus means that if I reduce the static head to actual situations of like 16m where the tank is filled to a height of 19m (safe fill level), you might not get the curve getting even close to the pump curve. What do u think guys? I have attached the pump curve that we received from the manufacturer.

 

[TenPenny]

I'm definitely confused by your description.

Your highest tank level is 21m with a full tank and the loading arm is 3m, and you say that you reduce the head with a lower tank level. That makes no sense to me, if the tank level is lower, the static head should be higher.

Also, for clarity, your switching of units between the sulzer curve and your spreadsheet definitely makes it more confusing.

But I'm baffled by your static head / tank level comments, I must be missing something.

 

[LittleInch]

As usual things keep appearing that weren't mentioned before. A pump verve versus system curve graph needs to have the same datum point, usually the pump. Therefore anything that is going on on the inlet side, e.g. static head, but also friction losses on the inlet side are added to the pump curve and then any static head effects and increasing friction losses in the system calculated for the system downstream of the pump.

Therefore your curves seem incorrect as the head at 0 is shown as nearly 20m. This doesn't match your data. However with 18 loading Bays, how many have you assumed on. I assume your pumps all work in parallel. I'm taking a guess here that your loading Bays actually have a valve or other device which restricts Max flow to a certain level. On a head chart this is repeated by a vertical line.

The static head at zero flow is the head you need to reach to get flow, e.g. you have an empty tank 15 m higher than your pump, then until you get a head of 15m you won't get any flow. This doesn't look like your system, but only you know and we've got about 10% of the information you have, but you seem to expect us to guess it all....

Your system sounds much more complex than a simple chart as you've drawn it, though I'm puzzled by your x axis. You shows about 65m3/hr as the end curve point, but the pump curve is nearly 240. What's going on?

I think you need to post your system diagram and your excel sheet if we're going to be able to see where you seem to be going wrong.

My motto: Learn something new every day

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

 

[1gibson]

New curve he posted has duty point at a weird RPM, and says "VFD" on it...

 

[LittleInch]

Well spotted. That throws everything into the mix again.

Davincigee, please give us the full picture about what you're trying to do and what you're system is and how it works.

My motto: Learn something new every day

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

 

[Artisi]

A bit of background history might help, why have the pump/s been selected for 145m3/h at 62.2m and why is the speed 2226rpm.

Seems a major cock-up somewhere and the pumps are way oversized.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)