Pump Control within efficiency curve

In that case, going back to the OP, I would suggest the PLC

  1. Set pump speed via a fixed frequency value e.g. 60Hz
  2. Control pump outlet pressure rise, measured between the pump and a valve downstream of and near the pump outlet, to the middle of "efficiency" range, using that same valve
  3. Measure the flow
Notes



  • Only one PID loop is required.
  • A measured flow greater than or equal to the flow for that speed and pressure means the pump meets the specs.
  • Several speeds could be used, as well as several pressures, but the "meets spec" metric is always the same (see next item below also)
  • It may be easier to control flow and measure pressure rise, in which case "meets spec" is a pressure rise less than the curve at the frequency and flow.
  • The measured pressure rise is between the pump outlet and the pump inlet, so if the pump outlet pressure is measured as PSIG wrt atmospheric, then any head in the sump above the level of the pump inlet must be subtracted from that measurement and used in the control.
  • Flow measurement is by nature one of the lowest-accuracy measurements, so it may be the source of debate with the customer. There are at least two ways to handle this:
    • Simplest: Make the meets/doesn't-meet flow number some percentage less than the curve.
    • Messy but eliminates any dispute over flow measurement: if the flow is small enough it may be possible to fill the sump and not recirculate the water, and convert a sump level change over time into a flow, or even use a time between two levels as the spec; this may change the outlet pressure, but that can be handled by integrating a sump level sensor into the measurement of the control variable.
 
You are making life tougher than it needs to be. No PID or control is required.

If you are looking to verify that impeller wear or other mechanical damage hasn't caused excessive drift from the manufacturer's performance curve just plot flow vs. pressure at 60 Hz (or whatever the manufacturer's original curve used for the base curve) and compare.

1) Pick 5 to 8 points flow and pressure points off the original curve and put them in Excell
2) Run the pump at the base speed and using a valve create backpressure. Measure pressure (suction) at the pump suction and discharge flanges (or as close to them as possible.) Make sure you have the same diameter pipe at your measurement point as the pump flange to avoid velocity head changes. You need both pressure measurements because the original curve is undoubtedly delta p across the pump. Measure the flow at that total pressure rise.
3) Repeat step 2) 4 to 7 times and put the points in the Excel spreasheet.
4) Use the Excel data to chart the original points with the curve and the new points without a curve. Use the trend function on the chart to plot the best fit curve for the test data - generally a polynomial. That gives you a smooth curve if the data has some scatter.
5) If the two don't line up within accepable limits send an invoice.
 
Last edited:
You are making life tougher than it needs to be. No PID or control is required....


Tom is exactly right, of course; my suggestion over-complicates the issue.


OT (but related, it's been that kind of Friday ...):



When testing large steam turbines (half a gigawatt and up, IIRC), my Dad (metioned twice on one day, and still another half a dozen threads to go, how about that?) said GE would run what they call a "valves wide open" test - isolate the process (turbine), open the inlets steam, and measure everything ASAP after commissioning to get a baseline (also to prove the as-built turbine meets the contract heat-rate, Btu/kWh, to ensure the final payment from the customer is made). Basically they treat the turbine as a flow element, analogous to an orifice meter or other pressure-drop flow measurement element.


Later, when the same valves-wide-open test is run and if the results are changed from the baseline, then something in the process changed (erosion, scale built up on the buckets, whatever).


With enough measurement data they could actually pinpoint where the changes occurred, at which point you can plan activities for the next maintenance an actually estimate an ROI. They could measure turbine performance to an absolute accuracy of around a quarter of a percent, so this metric was fairly useful.
 
This is making more sense now. At first there was talk of a VFD and changing speeds. The motor efficiency would change as a function of speed. Now the speed appears to be fixed at 60Hz. So is the VFD necessary to make sure the RPM is exactly 60Hz? If the pumps are not usually controlled by a VFD then there will be a little frequency droop. Shouldn't that be taken into account?





The problems I see is if there is a valve that is changing the system curve then how does one make sure the valve position and resulting system curve repeatable?


It seems there needs to be a known good pump that must be run from time to time to make a standard to which the pumps that are being tested can be compared.
 
It seems there needs to be a known good pump that must be run from time to time to make a standard to which the pumps that are being tested can be compared.


Not necessarily: Tom suggested half a dozen points or so; from that a curve can be plotted and compared to the design and/or baseline curves. If the valve is not at the exact same positions it won't matter: it is the pump's performance curve at "X" Hz that matters, and that is being compared; the system curves' values at specific valve positions are not needed.
 
Here is an example of a blower project I did several years ago comparing before an after tebuild test results. (The red and blue curves are at different air temperatures.)

Capture.JPG
 

Similar Topics

Hi everyone, I am a newbie and I have trouble in the control sequence. I have a system with 4 pumps. The required: Ex: Push a button to start in...
Replies
21
Views
2,703
Hi there, We have a system at a water treatment plant where large raw water tanks feed into the plant that's all on the same level. At high tank...
Replies
18
Views
3,637
I am at a loss... I have a pasteurizing system that i need to control product temperature with the speed of the product pump. I currently have a...
Replies
11
Views
2,559
I am controlling a very tight window in a 8 foot sump. My low level to keep from cavitating the pump is 44” and my high level is 55”. The problem...
Replies
9
Views
2,037
Hey everyone, I was on a customer's site on friday where they have a large bioreactor that they bought secondhand from a previous user. System is...
Replies
22
Views
4,385
Back
Top Bottom