Why is the curvature of pump rotor blades "backwards"?

[NiHaoMike]

I'm designing a rotor for a small centrifugal oil pump that is to be 3D printed, to circulate mineral oil for cooling electronics. (Before anyone says to just buy one, I haven't found one of the right specs that's anywhere near a reasonable price. It and the rest of the project is also a learning experience.)

When it comes to designing the rotor blades, I noticed that most centrifugal pump rotors seem to have the blades curving "opposite" to what I would expect, in that the convex side is driven into the liquid. I would have expected the concave side to be "pushing" so that it would act as a wedge with a changing angle, initially a shallow angle giving more force to get it moving and then steeper angles to continue accelerating the fluid that is already set in motion by the shallow part. Indeed, most fans are built that way. Or looking at it a different way, most airplane wings are concave on the bottom to accelerate the air downwards.

Thinking it might be some nonintuitive effect of incompressible liquids as opposed to compressible gases, taking a look at boat propellers seems to disprove that since the concave side is pushing, same as for fan blades.

I'm sure there's a good reason why centrifugal pumps use the "backwards" curve, what is it?

[bdunham7]

Does the oil accelerate as it reaches the outside of the pump or does it slow down? 

[Berni]

If you look wide enough you can find both directions, some bend forward some bend backward, or some are just straight. It's an optimization for the job the centrifugal blades are designed for.

The logic behind the backward bent blades is that the pumped medium has to slow back down from the rotation inside the rotor. This is most easily facilitated by the blades bending backwards since it provides a nice gradual transition from the high speed center to the slowly moving outside collection chamber. Eventually the medium has to go into the output flange and that is moving at the output flow rate, no faster, no slower. So such a rotor doesn't waste mechanical energy spinning the medium where it doesn't want to spin anyway.

So then why are not all of them backwards if its so nice and efficient? Well it turns out that if what you want is high flow rate at low pressure (like a lot of air moving fans) the forward bent blades are better. They act as scoops to grab as much air as possible, get it up to speed quickly as possible and fling it out like a shovel. This lets the same size rotor create more flow rate.

As for straight blades, it might be that they are designed to spin in both directions. Can also help if the medium has a lot of solid particles since they have a harder time hitting straight blades. Perhaps they actually needed some compromise in between spot between forward or backward bent designs. Or they didn't bother to optimize it, the flat design was easiest to manufacture and worked well enough for the job.

[beanflying]

What type of Oil are you planning to pump? Apart from maybe mineral types of weights seriously look at a vane type of impeller over a centrifugal one in particular as it is likely small. You will likely find the vanes print better than the complex curves.

Commercially look at the small weed or ag type spray diaphragm pumps or the Jabsco and Johnson ranges of flexible impeller pumps.

Or if you have a known flow, pressure and oil type (or viscosity will do) throw it here and I will see what suits it what I did when I had a real job.

[armandine2]

if taken to the extreme the opposite curvature would not release the liquid - the centrifugal pump isn't a positive displacement device, it accelerates the liquid. Look at the relative linear speeds of points along the curvature.

[MarkF]

Cavitation

[IanB]

Here is a technical explanation with some relevant equations:

https://youtu.be/cj1edFvGqRk

[thm_w]

An existing watercooling pump could work for the task, assuming the seals are compatible or changed out to something compatible.

https://forum.level1techs.com/t/mineral-oil-in-a-water-cooling-loop/70339/14

[Whales]

In the world of air impellers my vague understanding is:

 - backwards curved impeller blades are quieter and more energy efficient
 - forwards curved impeller blades are more space efficient (ie more pumping power out of a given size impeller and fixed RPM)
 - straight impeller blades are a compromise between the two

This may all be completely different for less-compressible or incompressible fluids, I don't know.  Cavitation is not a problem for air pumping AFAIK (I wonder if you can pump so hard you cavitate into plasma?  Or separate the gases into low-high density?)

[bdunham7]

This may all be completely different for less-compressible or incompressible fluids, I don't know.  Cavitation is not a problem for air pumping AFAIK (I wonder if you can pump so hard you cavitate into plasma?  Or separate the gases into low-high density?)

Despite the obvious difference between air and liquids, the same ideas do apply because even though liquids are mostly incompressible,  with air it is still much more efficient to avoid excessive compression.  "Cavitation" in air would make a lot of noise, but you really have to work at getting your pump rotating fast enough (transonic speeds).

[Ground_Loop]

The comparison with a fan is not very good as a fan is axial flow whereas a centrifugal pump is radial flow. Also, if you look at the impeller at flow entry whether fan or cent pump the fluid 'sees' forward swept vanes. At the exit the fluid 'sees' reverse swept vanes. I have seen only one example where the reverse was true. I had a dishwasher whose pump had extreme swept impeller blades that upon reversing would actually pump water from the pump casing back through the impeller eye to drain the basin. Forward rotation was trailing blades at impeller exit and pumped water through the wash bars.

Generally, straight blade are for high flow and swept blades are for high pressure.

[CatalinaWOW]

There may be additional factors besides the 'pump theory' ones mentioned before.  If the pump is designed with little or no clearance to the blade tips, friction between the blade tip and wall will tend to drag a forward facing blade into the wall, with the reverse for a backward blade.  In one case increasing pump power requirements and reducing leakage, while the other case minimizes power at the potential cost of leakage.  Chatter could result from the forward configuration.  Several other possibilities.

[beanflying]

Some of what has been posted above isn't really correct.

Centrifugal Pumps come in a range of types from Axial, Radial and Mixed Flow impellers depending on application

Also they can be run in reverse (tips forward for a crude description) but they run very inefficiently with lower flow and reduced pressure (Typically in the 25-30% range).

[Ground_Loop]

Centrifugal pumps are by definition radial flow. Fluid may flow through the casing axially, but within the impeller it is only radial.

[beanflying]

Tell that to the industry 

Axial Flow impellers do not have the fluid flow radially in the casing at all (middle of the picture above). Typical applications are high flow low head applications like flood irrigation or ground dewatering and in some cases longer shaft drives are used. The impeller is a propeller and just because it spins radially doesn't mean the flow past the impeller is.

Some more reading here https://www.batescrew.com/axial

[Ground_Loop]

That is not a centrifugal pump as they say in the first sentence. An axial flow impeller is a screw or propeller impeller as they also describe. Centrifugal force is orthogonal to the axis of rotation and is, again, radial.

[beanflying]

In the industry they fall under one on the THREE types of typical Centrifugal Pump impellers in use. If you want the industry to change because you demand it then good luck with that 

There is also in addition to those the types a bunch of application specific ones I have left out as they are for process, chemical or waste/sewage applications.

The bottom line for the OP still is a Centrifugal Pump for oil is not really the best solution before we get to far away from that.

[amyk]

The bottom line for the OP still is a Centrifugal Pump for oil is not really the best solution before we get to far away from that.

This. There's a reason nearly all oil pumps (except possibly the huge ones in industrial plants) are vane, gear, or gerotor types. If you're 3d-printing one, a gear pump would probably be easiest.

[NiHaoMike]

The bottom line for the OP still is a Centrifugal Pump for oil is not really the best solution before we get to far away from that.

My application needs a good amount of flow but not much pressure, quite a different case from most oil pumps with relatively low flow but high pressure. The oil is farm grade mineral oil (often used to cool amateur radio dummy loads), not specified for viscosity but much thinner than car engine oil.

[beanflying]

The bottom line for the OP still is a Centrifugal Pump for oil is not really the best solution before we get to far away from that.

This. There's a reason nearly all oil pumps (except possibly the huge ones in industrial plants) are vane, gear, or gerotor types. If you're 3d-printing one, a gear pump would probably be easiest.

The reason not to go for a gear pump here is tolerance and wear on the sides and faces of the gear with 3D printed bits. And blip or bump on a surface is a grab point. Also extra complexity with needing a second shaft. Some of this can be overcome with installing very thin shims of maybe Teflon or Brass on the sidewalls for the 3D printed Gears to run against but you still have gear to gear mesh wear.

Vane as I offered initially makes a lot of sense and will be fine with mineral types of oil. Use the same thoughts as the gear with a thin shim on both sides of the case. It is also 'normal' on smaller vane pumps to leave the impeller floating on the shaft to align with the cases so depending on the motor a keyway and a slightly loose fit will work.

For seals Nitrile lip seals should be ok but maybe Viton if you can source them for not a lot more is a good idea. You can also go a full mechanical seal but it adds to the size and complexity.

More width = more flow and diameter is proportional to pressure both will increase with speed is a 'sort of' linear fashion minus some slip.

For info

[IanB]

My application needs a good amount of flow but not much pressure, quite a different case from most oil pumps with relatively low flow but high pressure. The oil is farm grade mineral oil (often used to cool amateur radio dummy loads), not specified for viscosity but much thinner than car engine oil.

For a general answer to your question, spinning blade (impeller) pumps fall under two broad categories: axial flow pumps and centrifugal pumps. Axial flow pumps are similar to a fan: they move liquid through from one side to the other like a propeller. Centrifugal pumps work a different way: they take liquid in at the center, spin it round with the blades inside a closed casing and push it out at the edge using centrifugal force (hence the name).

Axial flow pumps generally favor higher flows with lower pressure rise, whereas centrifugal pumps favor lower flows at higher pressures. So for your application, you might be better with an axial flow design, though this is not guaranteed. For example, garden fountain pumps are usually centrifugal, but they circulate water with a reasonable flow and not all that much pressure. You will need some pressure to move the oil through the various heat exchangers in your cooling circuit (this may not be trivial). Selecting a pump requires knowledge of how much flow and how much pressure rise (head) is needed.

[Berni]

Impellers and centrifugal pumps are great pumps for when you need lots of flow rate at low pressures. They don't mind there output being blocked and will just keep making a steady pressure. They are easy to make because no tight tolerances are needed, nothing actually rubs and the only seal is around the drive shaft. For 3D printing this is the best design since anything in the right shape will work, having really loose tolerances just means that you will have somewhat reduced pumping performance.

The reason for positive displacement pumps like gear or vane pumps is that they can produce much higher pressures. Once there output is blocked off they will create as much pressure as you can put force into its drive shaft. For pumping lubricating oil this is usually what you want since you need to squeeze the thick oil trough tight passages into small gaps, at the same time the oil itself lubricates all tight tolerance rubbing surfaces in the pump. This makes most of these pumps not very 3D printer friendly.

Pumping mineral oil in cooling should be fine for a centrifugal pump. The stuff is reasonably thin and you typically need mostly flowrate not pressure for cooling. Tho if you had the idea to push mineral oil trough standard PC cooling waterblocks then you might need a bit more pressure (the fins in those are really closely spaced), so for that case you might have to spin your pump at rather high RPM or go for a multistage design that puts 3 or 4 of these pumps in series to build up extra pressure.

[beanflying]

You should never dead head (no flow) any pump regardless of type.

If you don't circulate so fluid into and out of the pump body you will see an increase in temperature of that fluid. It is fairly common for example for a water pump left running to turn the casing water into steam vapor and kill impellers and seals.

Re Vane pumps without getting into petrochem grade tolerances are generally going to top out pressure wise at 30-60m maximum head (60-90 psi) and I doubt you will get the tolerances on a small 3DP impeller much over 10m without a lot of work post printing. Even with that said nearly all industrially plumbed options will either have an internal and or an external pressure relief valve. The little domestic and fire service jacking pumps generally don't.

If you make a low tolerance centrifugal impeller then you will be back into the realms of the junk pond pumps and a few M maximum which may or may not be an issue. Still not a fan of unspecified mineral oil and a centrifugal regardless.

[jmelson]

My understanding is "forward" facing vanes, meaning concave toward the direction of rotation,  are used to impart more energy to the fluid, giving higher pressure differential at lower flow rates.  "Backward" facing vanes are for higher flow rates but lower pressure.

Jon

[RJSV]

I'm mainly curious (gets me in trouble, fast, often):
   ?: I've thought, as to wing shape, it was speed, that essentially 'lowers' air pressure, although now I'm not convinced, that the reduction is (?) at rt. angles, to air flow direction?
  So, having a wing 'hump' upwards shape, THATs going to produce a volume with lower air pressure,
this regular pressure (underneath) pushes UP, more than the upper surface is getting.
Part of that, is an 'attitude'; that 'vacume' doesn't exist,
(without real 'pressures' elsewhere).
  Anyway, no thoughts, about a 'downward directed' air stream, is that real ?