Pump Speed Change Spreadsheet

[Guyinfield]

I made up a spreadsheet to figure the centrifugal pump curve at reduced speeds. Just punch in the full speed data points on the 60hz line and voila. Just wanted to put it out there if anyone's interested.

I was skeptical at first but GPM actually seems to trend +/-10% against the DP on my cooler barrels.

Long story short we have 2 new centrifugals installed and the scope of the job resulted in a kind of hacked together VFP system. Trying to show that the magnetic type flow-meters are unreliable/poorly located. Would attach pics but wouldn't want to defame the installer. Haven't gotten any direction from clueless balancer on system DP setpoints so we're playing around ourselves. At night I have very stable flows so I tweak here and there.Pump Speed Change Template x1Hz.xlsx

[WAYNE3298]

You should add the design and actual system curve.

[Guyinfield]

I've been using the part-load data from york for the barrel flows vs DP and interpolating between points.

As far as the system curve where would I start? Building is three "zones" each with dedicated AHUs. Residential is 24hr operating all equipment. Retail and office on normal operating hours. Would I have three different curves to look at? Condo only, office condo, and retail office condo?

[WAYNE3298]

You should do the system curve as near full load as possible. One reason is to be sure you don't have an adverse at the pump inlet. I like to plot the system curve on the pump curve. If you get the pump DP at full load you can calculate the entire curve. I also like to compare the pump flow per the curve with the chiller DP. The chiller DP is the best and that gives you a good idea how close the pump is operating per design. They are usually pretty close but I only saw two in twenty five years that were dead nuts on.
I had to do that as part of the balance report but assume you are doing it for the practice. I applaud you for that.

[Guyinfield]

I've been logging things so I might have what I need to calculate in that case.

I don't get to see the plant under high load at night. Covid has brought occupancy way down so even on the warmest days the load isn't the same.

As mentioned pump and barrel match up pretty well. We have 4 new magnetic flowmeters, one for each machine and one total. None are even close. Also 3 ultrasonics we used for billing btus. We calc the total through BAS and they're also off.

Design flow is 1367gpm. Most nights I've been running around 850. Near design flowmeters read 1050-1200. Around 850 flowmeters read 500-600. Ive been pushing that the barrel DP is correct so we need a true dp sensor. There just isn't a straight run to use the meters we have.

I'll add that the condo zone has 4 AHUs with 2 way valves, about 350 fan coils with binary 3 way valves and a secondary loop for 2 AHUs also with 3 way valves. Delta t is an issue at times.

[WAYNE3298]

I agree that the chiller barrels are a lot more reliable for determining flow than anything else. I had a flow meter guy swearing that his meter was reading accurately when all the other data said it wasn't. What made it a problem was the design engineer wanted to use the flow meter for control. We had a conference call with the chiller factory engineer and he told them he knew exactly what the flow was from the chiller pressure drop because they had been testing them for 50 plus years. He said as far as we are concerned flow meters have never been invented. That ended that argument.
I liked the mag meters if there was a decent place to install them because they have the best turn down ratio and require less undisturbed flow. They are so expensive that they are usually used when nothing else will work but unfortunately they can't either. The biggest problem with the ultrasonic meter was they wouldn't track flow if the flow varied a lot.

You are relying on the right information and have a good grasp of what is going on. You have enough information to run the numbers you want. The mixing of two and three-way valves is not good and will screw with your Delta-T. It also screws with control and can be hard to deal with.

Keep us posted.

[Guyinfield]

Will do. Thanks Wayne.