DMalyuta

Hey all,

For my internship I have to carry out a batch of experiments on an electric motor - propeller combination for a drone in order to determine the most efficient motor - propeller choice. At my disposal I have a 6 DoF load-cell to which I mount the motor and which can then give me readings on thrust and torque. I will be using only the readings for thrust in the vertical axis and the torque about the vertical axis. I also have at my disposal a stroboscope to measure propeller RPM. I do not have a wind tunnel and thus all my experiments will have to be static-thrust (hence, advance velocity of propeller =0).

I have read a few papers on the subject (see here and here) and all of them use the advance/free-stream velocity as their independent variable because they all have a wind tunnel and so can plot graphs of (something) vs. advance ratio (=distance the propeller moves forward through the fluid during one revolution/diameter of the propeller). What's worse, they define the propeller efficiency proportional to advance ratio; since in a static test the advance ratio =0, then the formula for efficiency would give me propeller efficiency=0 which is obviously not true!

My question : how do I calculate propeller efficiency for a static test? Is there a way of changing free-stream velocity in order to "emulate" a wind-tunnel (some kind of idea like moving air around the wing rather than the wing through the air?)?

If you have done such experiments or know of experiments that have been done by others, please suggest to me what to measure and to compare keeping in mind that I am searching to find the optimal electric motor - propeller combo. So far, my idea is (excluding efficiency calculation which I hope you will be able to help me out with) to record:

• Motor model (pick 5 types)
• Propeller model (pick 5 types)
• Prop thrust --> thrust coefficient
• Prop torque --> torque coefficient
• Motor current draw
• Motor terminal voltage
• RPM of propeller

The experiment will go like this: for each motor type, test 5 prop types where for each prop type I apply different signal types to the motor (=> different duty cycles) in order to determine the best signal type for that motor-propeller combo. In total with 5 motor and prop types and, say, 5 different duty cycles it will make for a [5 X 5 X 5] thrust reading matrix of in total 125 elements. Therefore I will have to do 125 experiments.

I appreciate your help for the efficiency calculation and also for suggesting, if you can, what other data I should collect!

Best to you!

BMatthews

You're going to find that for static testing of props in the 8 to 12 inch diameter range once you get much over a 4 inch pitch that the blades will stall and pull a large turbulent bubble with them. You'll actually hear the difference between examples with a 4, 5 and 6 inch pitch as all the 4's will operate unstalled, some of the 5's will stall while others don't and the 6 inch pitch props are all stalled when tested in the static mode. When in that mode the thrust per watt of input power drops radically due to much of the power being used to fight the drag instead of moving air.

Moving up to the 14 to 18 inch diameter range I found from static testing that the critical angle altered a little to up around 8 inch pitch for running unstalled with a 10 inch pitch being stalled again.

Such props can be used for strong climbing flight or for models that hover. But they tend to be too low an advance ratio for any sort of faster flying. For that you need higher pitch ratios. And to allow testing them you need to test in a flowing airstream or figure out some way of simulating a moving plane.

Perhaps mount the motor on a rotating arm which allows you to rotate at a suitable airspeed? Make the arm quite long so the air has a few seconds to calm down before the arm comes around again. A brake on the rotor shaft could allow you to simulate the aerodynamic drag of different airframes or air speed reduction due to different climb angles.

bjr_93tz

I'm thinking for a meaningful efficiency result you'd need to look at a lift vs power relationship because at the end of the day you want the required lift for the least amount of electrical power input.

However a practical outcome probably won't keep an academic happy.

The first "paper" is quite happy to ignore how much of the electrical energy is being converted into kinetic energy and looks soley at the resultant force of that conversion and how fast that force is moving through a body of air. That assumption simplifies the calculations to how much meaningful work the propellor is doing vs the electrical power input and NOT how much actual work it's doing vs the electrical power input, and as you've stated that formula won't work for you when that force isn't going anywhere.

So for a static system you're really sort of stuck with a resultant force vs power input rather than power out vs power in.

DMalyuta

Thank you BMatthews and bjr_93tz for your replies!

@BMatthews : thanks for the extra insight - it will help me in choosing the 5 motors and 5 propellers for the tests.

@Bjr_93tz : thank you for your suggestions. On my side, I have also concluded that I should do a (thrust out vs. electrical power in) scatter plot for the different engine - prop combos. This way, the best combo will be that which gives most thrust for least input power; it'll be interesting to also measure the weight of each combo to determine the thrust-to-weight ratio. I could then do an optimization and come up with the best combo which, hopefully, will give me an insight into the characteristics that I should be looking to optimize further for my application. This way I could perhaps do a second round of tests, now with combos in the ballpark of what I should really be looking for. I should hopefully be able to post back my results here in a few weeks!

In the meantime, if you have further suggestions, please do let me know!

Danylo.

airraptor

My Feedback: (66)

you can do all this testing and then when you put your best prop on the plane it doesn't perform like it should. A wind tunnel would be the best if can.

BMatthews

The power to weight of the motor ESC and battery means less than nothing. It's the power to weight of the overall model that matters. And it is that property that bjr was referring to.

You have not said what the plane will be tasked to do. Motor and prop results will be meaningless without knowing the duty to what they will be used for. Two models of the same weight can pull the same power from the battery. So the power to weight is the same for both. But if one is an electric powered sailplane and the other is a racing model the motor and prop are going to be at opposite ends of the selection spectrum. So I would start by assigning an anticipated goal to your study. And since you don't have a wind tunnel to allow for testing at the racing model end of the spectrum why not start off with the stated goal of selecting propellers for something like multi motor helicopter use? This neatly avoids the need for a wind tunnel and lets you focus on the results obtained from static testing. It also means that you can work up to the point where the blades stall and dismiss them as not being suitable for hovering and other helicopter flight mode study. Again you neatly side step the need for a flow of air onto the prop and put the focus on static testing.

It's also a highly valid study these days with so much work being done on multi-copter models for both fun and scientific work and even military and SAR use. In fact I can see the day when police and fire fighters use multicopter drones for quickly surveying a building or finding a fugitive before the big helicopter with the massive light can show up. So it's a growing field of study.

bjr_93tz

Whoops, I made the assumption he was talking about a hovering platform hence the talk about the static. If it's prop efficiency for a convential type plane then without a wind tunnel it's time to do real world tests.

If he can get input power vs airspeed logged or at the very least motor current vs airspeed and "guess" what the battery voltage was doing at the time, with the different motor/prop combinations then he's in with a fighting chance.

Trying to extrapolate static results isn't going to end well, although I'd almost bet my last dollar that you could hand those props to Red Bull F1, they'd scan them into their computer and tell you exactly how much thrust you'd get and how much torque it'd need to turn it at every airspeed/rpm combination imaginable...

BMatthews

Yep, I totally agree with the idea that extrapolating from static tests will be near useless. At least not without a lot of real world data to use for coming up with an algorithm.

I gave that example of a same weight and same power input sailplane vs racer above. That's an extreme of course. But I've been on the receiving end of a couple of real world examples of this myself. A four stroke glow engine model that was marginally powered which was fun to fly with a given size and brand of prop which became doggy and noticeably underpowered when flown with the same size prop by another maker. In that case even the shape of the prop came into play. I've also seen the same thing with the early electric models that used heavy battery packs and brushed motors where an old timer design's climb performance altered with minor changes in the prop.

Another point to ponder.... Years back a flying buddy bought out the old Y&O propeller line. He always found that these worked better than many options for his style of model flying. To try to find the cause he made up a pitch testing jig and spreadsheet to analyze the real pitch distribution from root to tip. He found that there isn't a prop on the market that carries the stated pitch value from hub to tip. And many didn't really match at more than a very small portion of the blade. So start by assuming that no prop from any maker is actually the pitch it says it is. You'll want to mimic this measuring yourself.

UStik

There is a nice tool by M. Hepperle: PropellerScanner (here at his Software page). All you have to do is photograph or scan a front and side view of a prop. The geometry including the pitch (varying over the radius) is taken from the pictures fairly accurately. It spares you the tedious pitch measurements.

Next step would be entering the geometry data into M. Hepperle's JavaProp applet and calculate a coefficient table. The tool neglects some aerodynamic effects and may overestimate thrust, for instance. Yet the results are quite useful and far better than values extrapolated from static measurements. Of course, the calculated coefficients are useless for speeds (advance ratios) where there is some blade stall, but the tool indicates blade stall quite reliably. Besides, if a drive with maximum efficiency is required there has to be no blade stall, anyway.

To find the optimum drive for a given task one could use my spreadsheets which I mentioned in another thread. They use the coefficients calculated in JavaProp and motor data from measurements or specified by the manufacturer.

If there is no way to measure in a wind tunnel this "workflow" is far better and economic than static testing. It doesn't even require having the motors and ESCs, just the props.

da Rock

There have been pitch gauges for sale since the 60s. I've had two different ones since then. They both were capable of very good accuracy for measuring the angle of the back of the prop. No gauges that I've seen made any attempt to measure anything other than that. It was up to the user to match the tool to the profile he hoped to measure. Luckily most props are relatively flat bottom airfoils. Unfortunately, very few flat bottom airfoils AOA match the angle of the bottom. And of course, to expect a prop-profile's AOA to advance at the "pitch rate" fails to understand what's actually going on with that means of propulsion.

But it's about all we got to work with. So measuring the pitch however you do it at least gives you some means of describing something about your different props.

One thing you discover with a pitch gauge is just how far from perfect our model props are. And about the time you recognize that, you discover how little that matters with our models.

Back in the past, the C/L speed guys and FF guys were carving their own props and had an excellent situation for judging their handiwork. They were some of the first to produce prop gauges. The C/L aerobatics guys also had a few who were that serious and some Rat Race boys. If you know any of them, they'd be worth talking to in order to dig one up. I don't think I've ever seen one advertised by a model retailer. I think one of mine from the old days was made recently by Edmunds Engineering. Yup... found a pix... Can't find them however.

You should notice how limited the prop size is...

da Rock

In fact many of the most popular props don't measure the same for both blades of the same prop. (even if you were to "true" the hub to the blades)

BMatthews is sorta playing with you guys..... Props aren't that simple anyway and he knows it.


What a pitch gauge is going to measure isn't pitch at all. It's going to measure what it can: the AOI of the bottom of the prop airfoil. And no gauge I've ever seen is going to measure the AOA of the airfoil since they can't deal with the LE radius most airfoils have to begin with nor can they predict the AOA at all.

The pitch of a prop is the result of a lot of things and measuring just one of them isn't going to provide a useful prediction of the pitch it'll produce in the air.

I'm afraid DMalyuta, that you will need to look into your project requirements and how deeply you need to classify your 5 props. There is a very good chance that simply relying on their advertised pitch as gospel just might jeopardize the credibility of your findings. Accurately describing the 5 props would be sensible. Accurately measuring them would go hand in hand with that. Area, twist etc..... Those props have more chance of distorting your testing than you do. Hopefully they are manufactured to more exacting standards than our model props are. If so, you can ignore all of my advice....

UStik

From Hepperle's PropellerScanner:

UStik

Example: old Graupner 7x5.
Front and side view from a flatbed scanner.
The same turned into silhouettes in a grafics program.
PropellerScanner window with parameters and calculated geometry.
Blade chord, twist (blade angle beta), and pitch (!!!) distribution over radius.
JavaProp parameters.
Geometry data from PropellerScanner with front, side, and top view.
Calculated power and thrust coefficients as well as efficiency.
Flow field meaning prop wash speed relative to flight speed (idealized).

As I said, not perfect but better than wind tunnel measurements you don't have.(Notice the table above the coefficient diagrams.

The 9th column indicates that there is no blade stall at advance ratios (v/nD) bigger than 0.25. Best effiency is 70% at 0.65 advance ratio, not bad for such a small prop.

The prop's efficiency depends mainly on a proper choice of diameter, pitch, and rpm. Compared to that, the actual geometry is of minor importance. Of course, you'd notice a big difference (between various geometries) in contest flying. Meseems in the first place the OP's task is rather finding the best motor/prop combo for the job than optimizing it.

UStik

The whole drive (prop, motor, ESC, battery, plugs and cables). I like such drive layout with quite big prop pitch. It's not that efficient in climb (40%, 12.5 A current draw) but it is in cruise (47%, 3.75 A).

Overall efficiency is the light blue line in the lower diagrams. Note the small rpm change ("unloading") over flight speed (yellow line).

el touristo

My Feedback: (2)

much of this is above my head or ambitions....but on the subject of prop efficiency I've concluded there is only one method
that suits me: telemetry data. airspeed vs watts is about all that's needed, I think (without getting into marginal concerns).
Rate of climb (altitude sensor) might help you tell when you were level (and for other purposes) but would it really be needed? RPM for fun and sharing but not required. Why bother with a wind tunnel when you can get real data easily? GPS and/or pilot tube. I think tubes are more fun but I've no real experience. Adding this sort of telemetry is my next project. When flying level the data tells you about your airframe's 'best' and also 'max' cruise speed (unless prop was way under-pitched I think). That's what we're really after here, isn't it? This would make is easy to choose a prop after testing a short list of best-estimated candidates. If you logged RPM you could also do a real decent job of calculating the airframe's actual drag by seeing how much thrust was required for level flight at a given speed.This would also be really cool info and would be as good (or better from a practical view) than wind tunnel. What I would really like to see in the RC airplane hobby is people publishing logged telemetry data (speed vs watts) for airframe/motor/prop combos, AUW if beyond norms, (excessively larger batt, FPV equipment, etc)..more induced drag might mean one size bigger. If more people would just get one system that they could test in all their airframes we could have a bunch of data. BTW, anyone know of any compilations of such data?

alex design

thankssssssssssssssssssssssssssss

HighPlains

My Feedback: (1)

I would look up the work of Dr. Eugene Larrabee on props. I believe he was at MIT.

jybp

Hi UStik hope its ok to bump this, I'm trying to replicate the way you copied the propeller geometry.
I got the same spreadsheet results in PropellerScanner, but geometry is off when I copy to JavaProp.
I've messed around with the airfoil and modify tabs and spinner value, but the results are way different.
Could you please share how to do this in the latest JavaProp version, or is there a better method with 3D scanning?

UStik

Nothing wrong with bumping an 8 years old thread, just stressing my memory. Going to the hard-disk memory of my PC may help though.

Actually I'm at a loss what your problem might be. What struck me is that you wrote "geometry is off when I copy to JavaProp". A straight copy won't work. From PropellerScanner, I always export into a text file. In a text editor, the three columns before "r" are deleted as well as the header row (see attached examples for the 7x5 Graupner prop). In the text editor, the whole content is copied to the clipboard (Ctrl-A, then Ctrl-C). In JavaProp, it's inserted with the Import button. (It uses the first three columns r, c, and beta and ignores the rest.)

Then you have to enter the correct Design Parameters (except Velocity and Power) before calculating a Multi Analysis. Rotational speed has to be assumed, for instance something like expected full-power static rpm (no precision needed).

At least that's what I remember...

jybp

Thank you!
The problem was I copied the scanner results directly to Javaprop without editing it, with your reference file and spinner diameter set to 0 it worked perfectly.
Another possible solution I found was to 3D scan it and use OpenVSP 'fit model' command.

gow589

My Feedback: (1)

About 14 years ago I was making a large quad. I wanted to test static thrust of propellers so I made a rig with a scale. The recommended propeller for flight gave about 6lbs of thrust but after extensive testing I zeroed in on the size and pitch which gave me over 10lbs static: