PEX-Pump Tables

[ewdudley]

Attached are some tables for matching a pump to a PEX circuit.

[Edit: Updated for more pipe lengths.]

Computations are done for equivalent pipe length circuits of 100 to 500 ft in 50 ft increments. Equivalent pipe length is the technique of converting elbows, tees, valves, and so forth to an equivalent length of pipe, so 125 feet of pipe plus a bunch of fittings might come out to something like 150 feet equivalent pipe length. Search online for more information on equivalent pipe length.

Circuit equivalent pipe lengths are round-trip, not one-way.

For each equivalent pipe length, PEX pipe of nominal diameters 0.75", 1.00", 1.25", and 1.50" are selected.

Then for each equivalent pipe lengh/PEX diameter combination, various pumps are selected.

For variable speed pumps two speeds are selected: max speed and minimum or close to minimum speed. Speeds are designated by suffix comprising a hyphen and Roman numerals. So if maximum speed is too much gpm and minimum speed is too little gpm then the pump will likely have a speed in between that is right for your application.

Constant speed pumps have no suffix.

Table columns are

fps=feet per second;

psi=psi pressure drop;

head=feet H20 pressure drop;

gpm=gpm;

watt=power estimate;

gpwh=gallons-per-watt-hour, which is a measure of relative efficiency. A bigger number is better, more flow with less electricity;

%gpm=percentage of maximum flow for that pump at that particular speed. Typically %gpm will be between %33 and %67 when the pump is well matched to the circuit.

pump=pump model

 

[varadhammo]

Where'd you get the pump curve formulae? Or did you do a polynomial approximation by entering some data points from the published curves, or just do a bunch of manual lookup?

 

[ewdudley]

did you do a polynomial approximation by entering some data points from the published curves

Exactly as you suppose. Just eyeballed about eight points per pump speed from the manufacturer's sheet and turned Steffen Meuller's (Eric W. Weisstein) Algorithm-CurveFit loose, constrained to a plain old ax^2 + bx + c quadratic. When I spit computed points back out and compared them to the input graph and estimates, the results seemed pretty darn nice, with all due apologies to Euler for my cavalier lack of rigor.

Wattage is a bit of a WAG, but clearly the ECM pumps really shine in terms of efficiency.

The Hazen-Williams coefficient used is C=150. H = (2.3066587 * 4.526 * Q**1.85) / (C**1.85 * d**4.865).

I worked through a couple operation points using pfcalc's Darcy-Weisbach with 170 degF water and flows were higher.

 

[jebatty]

Saved this in my "boiler" folder. Thanks.

 

[heaterman]

Nice work EW!
It's always enlightening to see in black and white how flow velocity impacts head. Once you get above the magic mark of 4.5-5 fps the head elevates exponentially.
All to often this is the case with the OWB dealers who seem to think they can ram any amount of btu's needed down 1" pex by just installing a larger pump.
But then........I forget they as a breed are immune to the laws of physics.

Somewhere in my archives I have a home made chart showing circuit length, tube size and pump model required to deliver xxx number of btu's .
I should hunt that down and see how closely it matches what you did there. I'd be willing to bet they are pretty tight.

 

[DaveBP]

A pity more (any?) OWB dealers aren't familiar with tables like this when they sell yet another customer a 600' roll of 1" PEX along with their largest boiler.

 

[jebatty]

A quick rule of thumb is that if the flow is doubled, the pump head is squared. I believe the actual amount of increase in pump head is to the power of 1.75, not 2, but squaring gets pretty close.

 

[ewdudley]

Once you get above the magic mark of 4.5-5 fps the head elevates exponentially.

I like the looks of the two biggest small ECM pumps -- the Taco_VT2218 and the B&G_Vario -- when it comes to hitting a sweet spot of less than 4 fps with minimum power. For less than $200 they look like a real good buy.

The other three small ECM pumps also perform very well where a little less flow is required.

 

[velvetfoot]

They apparently really liked the Varios on this project as well:
(I know I've posted this before, but I love it):

(broken image removed)

 

[ewdudley]

They apparently really liked the Varios on this project as well:
(I know I've posted this before, but I love it):

Good grief. Each one of those pumps could easily move 80,000 btu per hour!

"I love my cigar too, but ..."

 

[heaterman]

They apparently really liked the Varios on this project as well:
(I know I've posted this before, but I love it):

(broken image removed)

That's a supply house salesman's wet dream right there.

Plumbers use pumps. Boiler techs will use zone valves on a system like that as a rule.

.....I shouldn't critique another guys work without knowing all the circumstances really. Maybe they had a good reason for selecting that method for the job.

 

[varadhammo]

Just eyeballed about eight points per pump speed from the manufacturer's sheet and turned Steffen Meuller's (Eric W. Weisstein) Algorithm-CurveFit loose, constrained to a plain old ax^2 + bx + c quadratic.

Just on a mathematical side note, three points define a unique quadratic, so I guess that algorithm is just throwing out the extra five. With your 8 data points you could get a 7th degree approximation... like you said though, I doubt it would give any significant differences, but it might make Euler proud...

Edit: Or actually I suppose it could generate quadratics based on subsets of 3 and choose the one which minimizes the residual of the other points....?