Pump Control within efficiency curve

Hello Forum,

I am developing a Pump Control System and one of the most important modes of operation, is to keep the Pump operating inside it's efficiency curve range. As seen in the picture below, the chart represents the Pump's efficiency curves at different operating frequencies and the one I have highlighted is the 60Hz curve which is the most ideal. The circle on that curve is the Best Efficiency Point and the Control System needs to target that. X-Axis is Flow and Y-Axis is Pressure.

I created a limiting block that defines upper and lower limits for both Flow and Pressure and at the moment the only way to stop the Pump from operating at Max Flow limit, I created a "Power Limit" block which goes closer to 0 when the Pump's Flow is reaching the higher Limit and thus slows down the Pump's speed.
So, the Pump is operating at 100% of allowed speed until it reaches the Max Limit on the flow in that frequency range but I need to keep it operating between Min and Max limit while trying to achieve Best Efficiency Point. I can limit it from hitting Upper Limit with my "Power Limit" block but I just wanted to ask your suggestions as to what would be the best way to take this logic forward and have the Pump
(a) Keep operating between the Lower and Higher limits for both Flow and Pressure
(b) Target Best Efficiency Point when running at 60Hz frequency.

Keep in mind there is Pump's Speed to Control and also Control Valve Position to obtain maximum safe performance.

Thanks.
https://photos.app.goo.gl/HAUMnXxHfQnqJbkY6

Operating in the most efficient part of the curve doesn’t always mean you’re going to mean that you’re using less power. You can force the operating point to the efficiency hump with flow restriction and speed control, but you may be actually wasting energy to stay there.

RET said:

Operating in the most efficient part of the curve doesn’t always mean you’re going to mean that you’re using less power. You can force the operating point to the efficiency hump with flow restriction and speed control, but you may be actually wasting energy to stay there.

Click to expand...

^This.


Is pump efficiency the goal, or is total system efficiency the goal? The latter saves money, the former may or may not.


Reminds of the old carbureted vehicle example: do you get better mileage from 0-60 as a leadfoot (pedal to the floor, come off when you reach 60) or by holding accelerator at, or even gradually moving to, position for steady-state 60 from the start?


Turns out the former is better mileage, because the throttle is open and it takes less work for the engine to pull in the air.


Obviously accelerator pump could affect the result.

RET said:

Operating in the most efficient part of the curve doesn’t always mean you’re going to mean that you’re using less power. You can force the operating point to the efficiency hump with flow restriction and speed control, but you may be actually wasting energy to stay there.

Click to expand...

drbitboy said:

^This.


Is pump efficiency the goal, or is total system efficiency the goal? The latter saves money, the former may or may not.


Reminds of the old carbureted vehicle example: do you get better mileage from 0-60 as a leadfoot (pedal to the floor, come off when you reach 60) or by holding accelerator at, or even gradually moving to, position for steady-state 60 from the start?


Turns out the former is better mileage, because the throttle is open and it takes less work for the engine to pull in the air.


Obviously accelerator pump could affect the result.

Click to expand...

Thanks for the replies. These curves are actually provided by the customer who manufactures these pumps and they want a Control System that keeps the pump operating with those curves. Yes, I do agree on achieving Total System Efficiency to be the goal.

Software wise, I came up with the following algorithm:
2 separate PID loops; one controlling flow, with the flow feedback and the output controlling the pump speed; and the other controlling pressure, with the pressure feedback and the output modulating the control valve.

But, if the Pressure and Flow are correct at a certain point, should I go for a lower Pump Speed if the Pump can still maintain that Pressure and Flow?

rzuhair said:

]Yes, I do agree on achieving Total System Efficiency to be the goal.

Software wise, I came up with the following algorithm:
...modulating the control valve.

Click to expand...

What is missing from that plot is the system characteristic.


Pump curves (aka pump characteristics) are used for choosing a pump after the system has been designed and the system characteristic is set (or at least has an estimated model); they could also be used for designing the system, to get a favorable system characteristic.


Moving the valve changes the system characteristic e.g. the locus of [deltaPressure = C FlowSquared], so that would be a parabola starting at the origin of your plot, then moving right and accelerating upward, and the pump speed chooses one of the existing pump curves. Where the two curves intersect is where the process will run.


The valve position changes C: closing the valve X% increases C and the system characteristic shifts upward, so the intersection with the pump curve moves to the left and up, but assuming flow is controlled to be constant, the flow control would increase the pump speed, which moves to a higher pump curve, so the net intersection would move up i.e. constant flow.


By the same token, opening the valve X% decreases C and the system characteristic shifts downward, so the intersection with the pump curve moves to the right and up, but the flow control slows the pump speed and the net move is down.


The primary operating cost is likely power to the pump, which is proportional to the product of deltaPressure and Flow, divided by efficiency; so lines of constant cost (power) are, to first order, hyperbolas (K = c dP F) on that plot, with lower-cost (smaller K) closer to the origin and the the abscissa and ordinate.


Flow is probably constant for any steady-state operation, and efficiency will likely be near-constant, so cost is proportional to deltaPressure, so yes, you want to keep the valve as open as possible, which flattens (shifts downward) the system characteristic, so a lower-speed pump curve can meet the Flow requirement, and the system runs at the hyperbola with the smallest K (cost) i.e. closest to the abscissa = lowest pressure.


What I am saying here is nothing more than the justification for VFDs that has be a fact of life for so many decades that few even thinks about it any more.



If there is a requirement for a minimum pressure at the outlet of the pump* (why there would be is beyond me, but I don't know your process), then the valve may have to be less than full open, to shift the system characteristic back up to meet that pressure. So yes, controlling pressure with the valve (system characteristic) for a given flow should work. If the pump can meet a target flow at some positive pressure rise, it can almost certainly meet that same flow at a lesser pressure rise.


* I have been assuming the pump is upstream of the system, but the "pressure" in the model above and in the power equation is actually pressure rise across the pum; there may be both upstream and downstream characteristics to consider, but the basic approach here does not change.

rzuhair said:

But, if the Pressure and Flow are correct at a certain point, should I go for a lower Pump Speed if the Pump can still maintain that Pressure and Flow?

Click to expand...

Won't the pump speed and flow be proportional? The difference will be the torque required to achieve the desired pressure.


Valves cause an energy loss. If you can I would control the flow with the pump speed.
Doubling the pump speed will quadruple the pressure and multiply the power usage by a factor of 8 given the system piping stays the same.


The operating point is determined by the intersection of the pump curve and the system curve. The pump manufacturer has no control over the system curve because that is determined by the customer's system. The best the pump manufacturer can do is use efficient motors.


I don't see why there is a need for two PIDs. You can't control pressure and flow at the same time.

Peter, I disagree,
Quote: I don't see why there is a need for two PIDs. You can't control pressure and flow at the same time.
Rossi Catelli produced a steam injector system that took 10 bar down to 2 bar with a controller and controlled the temperature with a second controller
I programmed one by replacing the two controllers by integrating them into a Siemens S5 controller, the spec was a product temperature of 101.0 Deg. C +- 0.5, this worked a treat. very often such systems use a pressure reducing valve anyway to reduce the pressure form the standard 10 bar to 2-5 bar depending on the spec of the injector core or HE plates.

parky said:

Peter, I disagree,
Quote: I don't see why there is a need for two PIDs. You can't control pressure and flow at the same time.

Click to expand...

your example is not the same.




The pump manufacturer only has control of the pump curves. The pump manufacturer can increase the area under the pump curve by changing the speed. That is one PID.


A second PID could be added to control the pressure drop or system curve but where is the pressure being measured? Between the pump and valve or after the valve which will be at the operating point.

You mentioned working for the pump manufacturer. Are you building a system to test these pumps or to try to verify that they actually do meet the published specifications?

That would make some sense of your request.

ndzied1 said:

You mentioned working for the pump manufacturer. Are you building a system to test these pumps or to try to verify that they actually do meet the published specifications?

That would make some sense of your request.

Click to expand...

The pump manufacturer is my customer. I am building a PLC Controlled VFD package for their Pump with sensors and Tank Level automation. The customer is adamant on keeping the pump within the curves.

I would say most applications are interested in keeping the pump under the curves.

To keep the pump ON the curve you may have to induce an artificial load (pressure relief valve) which will waste energy and your time.

I hope this is just a misunderstanding in what they are asking. Or, perhaps, this is someone overstepping their knowledge base and not realizing what they are asking for.

If you are doing tank level and want to wow the customer, search for tank control posts from Peter Nachtwey. Awesome stuff there.

Peter Nachtwey said:

Won't the pump speed and flow be proportional? The difference will be the torque required to achieve the desired pressure.


Valves cause an energy loss. If you can I would control the flow with the pump speed.
Doubling the pump speed will quadruple the pressure and multiply the power usage by a factor of 8 given the system piping stays the same.


The operating point is determined by the intersection of the pump curve and the system curve. The pump manufacturer has no control over the system curve because that is determined by the customer's system. The best the pump manufacturer can do is use efficient motors.


I don't see why there is a need for two PIDs. You can't control pressure and flow at the same time.

Click to expand...

I came up with 2 PID loops if in case it maintains the desired Flow due to pump speed but not able to maintain desired Pressure, then i would regulate the valve accordingly. But, Yes. I can go with only Speed Control too.

Peter Nachtwey said:

your example is not the same.





The pump manufacturer only has control of the pump curves. The pump manufacturer can increase the area under the pump curve by changing the speed. That is one PID.


A second PID could be added to control the pressure drop or system curve but where is the pressure being measured? Between the pump and valve or after the valve which will be at the operating point.

Click to expand...

Discharge is measured between the Pump and the Control Valve.

Peter Nachtwey said:

Won't the pump speed and flow be proportional? The difference will be the torque required to achieve the desired pressure.

Click to expand...

I don't remember seeing if this is a positive displacement pump or an impeller pump? That could make a difference.

This is obviously a centrifugal pump.

You need to keep the fundamental principles in mind.

1) The objective should be minimizing power consumption, not maximizing efficiency. The end user pays for power used, not efficiency lost.

2) Throttling controls flow by creating head. The pump must create the higher head, which is dissipated as heat and is wasted energy. Throttling a valve basically creates a steeper system curve.

3) Varying pump speed creates flow by reducing the flow itself. It eliminates the creation of excess head and thereby reduces power demand. This is the effect of the affinity laws Peter mentions.

I have a hard time envisioning a condition where you can save more power by throttling than you can by reducing speed. Throttling is by definition wasting power.

There are circumstances at the exreme left and right of the pump performance curve where inefficiency is so extreme it creates vibrations and levels of heat that damage seals and bearings. The only time you should be throttling is if the pressure and flow at reduced speed move the pump into that range.

Norm, Tom is right about the pump. One can tell by the pump curve.

I have a hard time envisioning a condition where you can save more power by throttling than you can by reducing speed. Throttling is by definition wasting power.

Click to expand...

YES! I point out below how one can determine the power. Throttling waste energy. I have said that above. So one only needs to control the speed. if that is necessary. Some pump motors may have two sets of windings that will allow 1800 RPM or 900 RPM minus slippage speeds.



I don't make a big deal out of this but I was a RCO ( reactor controls officer ) on a nuclear submarine and one of the most important parts are the main coolant pumps that pump water through the core. We had to learn everything about everything and that includes the main coolant pumps. That includes system and pump curves and where the operating points would
be at different pump speeds and how much power was used.


The power used is proportional to the flow times the pressure. In imperial units the HP = P x Q / 1714 where P is pressure in psi and Q if flow in gallons per minute. One must divide the HP by an efficiency factor which is typically about .95 so the actual HP required is a little more than the theoretical. Now you can find the intersection of the pump and system curves to find the operating pressure and flow and the resulting power which is basically the area under the rectangle defined by the origin and the operating point.
BTW, I was interviewed by Adm Rickover twice and I trained on the prototype of the USS Nautilus. The movie is pretty realistic especially the part where candidates were being interviewed. I know because I went through it. You can find the movie on YouTube.