Any pump geeks able to lend me a hand?

[gitchegumee]

Looking for a bit of help using VFD to hit sweet spot on my pump curve. Anybody able to help confirm my use of affinity laws? Thanks in advance!

[WaterEng]

Does the pump manufacturer give you the pump curves at various frequencies?

Don't you actually mean "hitting the sweet spot" on your system curve? Pump efficiency and proper VFD settings can't be estimated without the system curve. If you have a pressure reading somewhere in your system, you'll at least have the head condition. If you have a flow reading, you'll have a point on the system curve.

Of course, you can estimate the system curve based on the pipe diameter(s), material(s), and fittings. However when restrictive fittings like spray balls are in the system, then tend to dominate the headlosses and in a way make the system curve estimation easier.

[gitchegumee]

Hitting the sweet spot

Let me give a bit of background that may help: Electricity is extremely expensive for this installation. I'm trying to run my glycol circulation pump at the exact speed required to keep a required minimum flow rate through the freon exchangers, provide enough flow rate to service whatever fermenters/BBTs might be active, and add no more energy. Any more is not only a waste of energy into the system, but also takes a tremendous amount of refrigeration to remove it, and robs me of cooling for my tanks. Classic double-edged sword scenario. Most pump manufacturers publish performance curves for 2-pole and 4-pole motors running at 3,500rpm and 1750rpm with 60Hz supplies. A few go so far as to publish curves for 50Hz/2900rpm too. Normally to meet a sweet spot on a pump curve (ie: the best efficiency point for flow vs pressure) we would find our desired pressure & flow and then trim the impeller to meet it on the performance curve. Not so easy with changing flow rates, or with a non-ideal pump choice to begin with. As an alternative--or arguably a better solution--to trimming impellers, VFDs can change those curves by changing rpm to better match the pump performance with the changing flow requirements of the system. Since manufacturers don't publish different curves for every frequency, we have to use pump affinity laws to estimate those new points on a new curve. These affinity laws allow us to take one point on the published curve and forecast the new point for the new rpm. Basically, these laws are: flow is proportional to rpm; head is proportional to the square of rpm; and power is proportional to the cube of rpm. To assist with developing the new curves, the proportionality constants are similar to those in the published pump curves. That is to say that the new curves have lines of similar modes to the published curves. Okay, that's where I'm starting to need help. I'll grind my way through the literature, but sure would be nice to have some help!

[Starcat]

Thoughts

Let me give a bit of background that may help: Electricity is extremely expensive for this installation. I'm trying to run my glycol circulation pump at the exact speed required to keep a required minimum flow rate through the freon exchangers, provide enough flow rate to service whatever fermenters/BBTs might be active, and add no more energy. Any more is not only a waste of energy into the system, but also takes a tremendous amount of refrigeration to remove it, and robs me of cooling for my tanks. Classic double-edged sword scenario. Most pump manufacturers publish performance curves for 2-pole and 4-pole motors running at 3,500rpm and 1750rpm with 60Hz supplies. A few go so far as to publish curves for 50Hz/2900rpm too. Normally to meet a sweet spot on a pump curve (ie: the best efficiency point for flow vs pressure) we would find our desired pressure & flow and then trim the impeller to meet it on the performance curve. Not so easy with changing flow rates, or with a non-ideal pump choice to begin with. As an alternative--or arguably a better solution--to trimming impellers, VFDs can change those curves by changing rpm to better match the pump performance with the changing flow requirements of the system. Since manufacturers don't publish different curves for every frequency, we have to use pump affinity laws to estimate those new points on a new curve. These affinity laws allow us to take one point on the published curve and forecast the new point for the new rpm. Basically, these laws are: flow is proportional to rpm; head is proportional to the square of rpm; and power is proportional to the cube of rpm. To assist with developing the new curves, the proportionality constants are similar to those in the published pump curves. That is to say that the new curves have lines of similar modes to the published curves. Okay, that's where I'm starting to need help. I'll grind my way through the literature, but sure would be nice to have some help!

Phillip, I seem to recall you are running a single pump type unit.
I would want to be able to read minimum GPM through the HX and decide what the baseline Frequency will be for that operating condition. This would assume that there is not a wild alteration in characteristics with diversity of loads in your system as regards the piping. I would want to read the actual watts at that point and it would seem to be logical that you would need somewhat more output for the loop after you determine HX MINIMUM. I would look at WATTS vs. PERFORMANCE. Ideally the drive is not hunting and its possible to widen the setpoint differential on the chiller as possible. This is all a low Tech approach. Just a perspective from that angle without too much math.

[CPESystems]

We can generate curves at any speed for you with PumpFlo software

Hi Phillip

You can generate pump curves at what ever speed you need using PumpFlo software. I have a copy and can do it for you as long as the pump manufacturer has signed up and most have.

What pump do you have?

This is an example of a 3000 rpm curve for a Inoxpa SEN-15 pump.

[gitchegumee]

Awesome!

I have a Goulds NPE series 2ST1G2C4 pump that I'm trying to evaluate as glycol pump. Is this software something you can share with me? I don't have a set speed for a set curve. That's my whole dilemma. I need to evaluate the pump at lowest speed (as in fermenters off and maintenance for glycol reservoir temperature) through highest speed (as in many fermenter jackets open). If I can find a pump that performs well in my range of operation, then I use pressure transducer, controller, and VFD to control the speed to just enough. Not more.

[Starcat]

Pumps on 10 HP System

Phillip, currently running:
Goulds 2ST1E1E3 3450 RPM / 1 HP on about 600 BBL Diversified Load. Performance is unreal. It mystified one guy from a Chiller MFG. [Loop Pump]

Goulds 2ST1F1D4 3450 RPM / 1.5 HP [Chiller HX pump only]

I am taking note of similar design 2 stage impeller pumps as used on Della Toffola units I have worked on in the past and wondering about how speed control would behave on those and if there might be any advantage at lower RPM........

[gitchegumee]

Really not a good match for my glycol system

Pressures are far too high and flows far too much for my 4T + 4T system. Even with the trimmed impeller. I'm sinking tons of power into my glycol for no reason. Got my head around the pump affinity laws now and should have some good numbers soon. Doing them by hand; would love to get my hands on software that does it for me. At least as a check on my work.

[Starcat]

Readings

Pressures are far too high and flows far too much for my 4T + 4T system. Even with the trimmed impeller. I'm sinking tons of power into my glycol for no reason. Got my head around the pump affinity laws now and should have some good numbers soon. Doing them by hand; would love to get my hands on software that does it for me. At least as a check on my work.

Phillip, your process has me intrigued and I would like to follow you on it.
Also I would be curious to know what your running Voltage and AMPs are currently at full load on this pump.
Also chiller details would be helpful.

[CPESystems]

I think you can do it online yourself

Hi Phillip,

If you click on this link it will take you to the Pump-Flo website and you should be able to set up an account and do it yourself. The online version is not quite as good as the paid version I have but might give you enough info.

At least it might get you close and then I can refine the curve for you with my paid version of the software.

https://www.storefronts.pump-flo.com/storefronts

Let me know how it works out for you.

[gitchegumee]

This is too much fun!

Thank you so much! I've been playing with this non-stop last few hours to visualize those numbers I've been cooking. I was amazed at how well my affinity laws predicted results. I was able to use published data for 3500rpm, 2900rpm, and 1750rpm to confirm my calculations at target operating point within a very few percent. And this curve generator further confirmed those results. But more importantly, they helped me see efficiency lines and select the right impeller and the right motor speed to give me target flow and target head at best efficiency point. Working on the rest of my report now to show just how much we can save using a VFD and smart selection. Thank you again for another tool to put in my digital toolkit! Starcat, I'll get results out in a few days when I'm finished. Thanks again!

[Starcat]

I'd like to see if I can walk through the process for the problem child I have been speaking to you about.
Thanks

[CPESystems]

Hi Warren,

I will give you a call and we can run your application through the program.

[Starcat]

CPE this would be helpful when time permits.
I am in the shop generally Sun-Thurs. 6-2.
You guys have helped me greatly in the past.