Centrifugal pump test

[petertha]

Good to know about the PC cooler. Another difference is a car radiator operates at some pressure, 15 psi or so? The radiator fluid keeps the pump perpetually primed but its negligible pressure contribution due to short hydrostatic head, under 1 psi. Operating pressure also elevates the boiling point temperature, 50/50 coolant mix boils at 129C vs 100C for fresh water.

I've read that model engines run much cooler but not sure I've come across typical temperatures. One calculation suggests using cylinder/head heat dissipation at 30% * BHP. Assuming guestimate 4 HP (2984 watts) yields these coolant temps at range of flow rates. Of course this is full throttle, reduced RPM would be ~proportionately less heat. For reference, Charles pump curve at 3000 rpm was 6-7 ml/sec = 0.36 - 0.42 litre/min

<edit> removed incorrect chart attachment, see subsequent discussion

[Charles Lamont]

Beware that those curves were not for a centifugal pump. They were for a 'turbine' or 'regenerative' pump.

[petertha]

Beware that those curves were not for a centifugal pump. They were for a 'turbine' or 'regenerative' pump.

Thanks for catching this Charles. I believe the correct chart is this one? I'll edit my comments.

[Charles Lamont]

Yes, that graph show curves of the characteristic shape for a centrifugal pump. I need to update that thread, but for now, here is the graph with more recent lines added. You may find of interest the olive green line rising from the origin, which shows the flow resistance of my cooling circuit. Also the pale blue curves indicate a pretty good showing from a double-shrouded impeller.

[dieselpilot]

You learn something every day. I knew automotive fuel pump were "turbine pumps". This didn't really make sense to me (my mind went straight to axial flow turbines) and I didn't take the one I replaced apart to look inside. Now, I understand they are in fact turbine regenerative designs.

The theory about model engines running cool has to do with the idea that they have more surface area for displacement at this scale. Potentially it's easier to cool them, but they don't actually run cooler. You can estimate heat load by fuel consumption. For a display engine I think radiators are sized to match the scale of the engine with little thought into how much cooling is actually required. They simply don't have any requirements to run continuously. Most PC coolers aren't rated and obviously capacity is determined by both media's flow rates. Making 4HP continuously will require different cooling than starting starting the engine and cracking the throttle a few times on the display table.

[Vixen]

The theory about model engines running cool has to do with the idea that they have more surface area for displacement at this scale. Potentially it's easier to cool them, but they don't actually run cooler. You can estimate heat load by fuel consumption. For a display engine I think radiators are sized to match the scale of the engine with little thought into how much cooling is actually required. They simply don't have any requirements to run continuously. Most PC coolers aren't rated and obviously capacity is determined by both media's flow rates. Making 4HP continuously will require different cooling than starting starting the engine and cracking the throttle a few times on the display table.

And the PC cooler can have an electric fan added to one or both sides of the radiator to aid the heat dissipation.

Mike

[dieselpilot]

Yes of course. Many are low power single digit watt ratings., but powerful 20-30W units are available, though they tend not to be quiet.

[gbritnell]

Peter,
I use a centrifugal type pump on all my engines for water circulation. I never bench tested them  but rather just used traditional centrifugal pump design in making them. I can honestly say that they all perform to the point that after the engine is running the radiators get hot so I know that they do work. On my Holt for instance, it doesn't get spun very fast in normal operation and the water flows as it should. I'm attaching the drawing of my pump which is very similar to yours.

[petertha]

Thanks so much George. Yes there is some similarity. That is encouraging. I will draw it up & take a close look dimensions & clearances. I have seen your 4-cyl OHV 'watchmaker' pump but I was trying to make my bench tester a bit bigger. Well I am making progress. I have my stationary motor aligned & driving the stationary pump. My water supply bottle & discharge tank is still a bit ungainly, but it is now moving water at moderate RPM. I figured 3D printing the parts would allow me lots of changes & no machining involved. But it brings up new challenges like fitting hardware components & tolerances in general so more tweaking involved there. Note to self: if dry spinning the pump to evaluate bring son the same smell as 3D printing, you know something is rubbing HaHa. I've been applying a thin grease film between certain surfaces so the don't gall & have to pay attention to tightening the peripheral bolts. Carry on!

[petertha]

I modified the thread title deleting the word 'fail'. Today after some futzing & head slap moments, I have 'success'.

Its the same basic pump design shown thus far, just a better, more controlled setup. I'm using a windshield washer fluid just to show the fluid stream more readily. I'm having difficulty getting a good RPM read with my laser dot RPM gun. Maybe affected by the coupler color/finish contrast to white tape or spline grooves? It seems to be reading my drill as expected. That said, I'm estimating the low stream at ~800 RPM & high stream at ~1800 RPM. The setup is ~1" of head from fluid level to pump inlet & 4" head from pump discharge to container inlet nozzle shown. Its a healthy squirt. Now a column of fresh water is only 0.433 psi/foot so we are not talking a lot of pressure here. I wont try & correlate performance into a nice format like Charles assimilated. I'd need a taller column & other stuff. I will probably try some other rotor configurations for fun. I need to improve the wishful thinking seal between pump faces anyway. I suspect a groove with 1mm O-ring cord or something. I also want to get on with a mockup 3DP cylinder / water jacket & try & flow through that restricted space which is the end goal anyway.

I'm wondering out loud if there is a rudimentary way to get a handle on power used when the time comes. For example if I measure voltage & current at the motor just circulating fluid through negligible pressure (dumping into the tank) & call that baseline power. Then I attach discharge to a more constricted cylinder jacket + radiator mockup. Adjust for TLAR flow & repeat measurement, is the delta watts telling me power consumed & I could relate that to what the engine is likely producing at that RPM?

[petertha]

I mentioned previously that the outside of pump body was rough looking cosmetically. That's because I orientated the print to favor a smoother inside flow surface - where it counts. Therefore the exterior surface shows stringy artifacts of slicer tree support. The filament is PLA using 0.4mm nozzle on Bambu A1.

[MJM460]

Congratulations on getting your pump working. Always a satisfying achievement.

I tend to agree with you that the shiny coupling is possibly confusing the tacho by providing extra reflections that make a definitive count more difficult.  I would suggest some black tape right around the coupling first, then put your reflector spot on top of the black tape.

It’s not easy to infer the power drawn by a pump in a circulating system from a single operating point test in an open system.

First, the power estimated by measurement of the input electrical power includes the efficiency of the motor as well as the efficiency of the pump.  It is difficult to separate these two.

Second, the pump drive power required is proportional to the head times volume divided by efficiency.  The efficiency ranges from zero at zero flow to a maximum then back to zero at the maximum flow with an open discharge.  Of course, there is power absorbed by friction in seals and bearings, and some hydraulic issues.  This is additional to the hydraulic power requirement.

  The maximum efficiency is usually at a flow a bit over 50% of the maximum.  Your little pump efficiency is probably less than 50% but even so, that provides a large range for error when you know neither head nor flow in your engine cooling circuit.

With a few more tests to determine the range of input power at a suitable range of speed, you could at least estimate a likely maximum.  I don’t really know the motor efficiency, but  someone else may be able to suggest an appropriate number to use.

MJM460

[Charles Lamont]

In my pump tests I never tried to measure the power requirement. The maximum power output I calculated was about 23mW, at 25cm head and 9.5ml/sec flow rate. That was with a 1/2" diameter impeller at 4000rpm.

The impeller diameter is one of the prime dimensions of a centrifugal pump, and with a cursory check back through the thread I don't think we know how big yours is, if that is not an impertinent question.

Scaling up linear dimensions increases the power output as the 4th power (unless I am having an off day). Power output is also proportional to the cube of the speed.

Using a really rather inadequate motor, I was very conscious that the mechanical friction was very variable, and absorbed much more power than actual pumping did. 

[petertha]

Charles, here are some dimensions of my pump. It wasn't sized to any particular task, more or less TLAR to get started & something that would reasonably print. My CAD file is somewhat parametric, so if I start with a different rotor diameter & volute rise (=throat height, is that the right word? maximum dimension above rotor at throat) then much of the body self-updates. Not quite that simple but that was the general idea.

Rotor diameter = 27mm. Inlet pipe ID = 6mm. Inlet eye diameter = 7.8mm (what the inlet pipe trumpets outward to and the open circle at center of vanes). Throat height = 4mm. Throat width = 6mm = width of vanes which are constant width in this version. Throat rectangular sectional area = 24mm2. Vane to opposing pump face clearance ~ 0.2mm. Radial gap at cutwater ~1mm. Outlet pipe ID = 6mm.

Sheesh I should just make you a dimensioned drawing but hopefully this helps. I'm sure I've misinterpreted some design parameters along the way. I think the only meaningful way I can attempt to correlate to your pump curves is to measure flow