Don sheds some light on prop braking!!!
Warning... Its a long one! ;-)

From : Don Stackhouse

OK, so far in this thread about the drag of a windmilling prop, I've heard enough partial truths, old wives' tales, gross oversimplifications and misconceptions to rival a downwind turn discussion! Let's see if we can distill a little order from all this chaos.

First we'll review a few basic concepts.

A propeller is for an airplane what the wheels and drivetrain are for a car. It connects the motor to the medium in/on which the vehicle operates. A propeller transmits power between the motor and the air.

Like most drivetrains, it can conduct power in either direction; i.e.: from the motor to the air, OR from the air to the motor. This second case is what windmilling drag is concerned with.

In addition, the propeller (like all drivetrains conceived by man) is not 100% efficient. For model propellers it's probably going to be in the range of 60% to 80% in typical cases, although really well designed props that are exceptionally well-matched to their application might in a few cases exceed 90%. Full-scale propellers tend to be more in the high 70's and low 80's in average applications, and also approaching or possibly exceeding 90% in some applications. The efficiency at low airspeeds and high power settings (takeoff, climb) will not be as good as at high speeds and lower power settings (cruise).

The efficiency in windmilling flight with the prop making negative thrust will not be as good as in powered flight with the prop making positive thrust of the same amount. It won't be as bad as what happens when you mount the prop backwards, because the leading edges of the blades in windmilling flight are still moving in the correct direction, but (like a prop mounted backwards), the camber of the blades will be facing in the wrong direction relative to the thrust being made.

Like a wing, a prop does its job by accelerating air. To make positive thrust, it accelerates the air passing through the prop disk so that the air behind the disk is moving faster than the air coming into the front of the disk. Approximately half this acceleration occurs in the region just ahead of the disk, and the other half behind it. Since the air has mass, the prop has to apply a force to the air to make it accelerate (F=Ma), and in accordance with Newton's Laws, the air applies an opposing reaction force to the prop that we call thrust.

To make negative thrust (i.e.: "drag"), the prop decelerates the air passing through it, instead of accelerating the air.

OK, I think that covers the key initial concepts we need to make some sense of this.

Now, to make either positive or negative thrust, power has to come from somewhere, AND it also has to go somewhere.

For windmilling drag, the first place it goes is to overcome the drag of the blades themselves. The blade have skin friction and inertial drag (the drag that results from overcoming the air's inertia as the blade shoves it out of the way as it passes by). This drag is a force, the blades are moving, and force times motion is energy. Force times motion in a given length of time is power. That power to move those blades through the air normally comes from the engine when the prop is making positive thrust. If the prop has an efficiency of, say, 70% (meaning that 70% of the power coming from the motor gets turned into useful thrust), then the energy from the motor that overcomes the blade's profile drag is a part of that remaining 30%.

If power is not coming from the motor for this, then the prop will extract this power from the air flowing through the disk, just like a windmill. Not a very efficient windmill, mind you, because the prop is designed to make positive thrust efficiently, and therefore tends to make negative thrust inefficiently. To the airplane this looks like drag, and is the first component of what we call windmilling drag.

Generally it's a relatively small amount of drag, usually less than the flat-plat drag of a stopped propeller, which is why rubber powered free flight models often have a windmilling device in their prop hub to let the prop spin as freely as possible after the rubber runs out. The other factor in that case is that rubber powered free flight props tend to have very high pitches (they need as low an rpm as possible to make the winds in the rubber last as long as possible), and the planes involved tend to have slow gliding speeds. If, due to all of this, the helical airspeed at the blades is low, then the drag force of the blades will be low. Since the rpm is also low, the product of the force times the motion is low, so the power absorbed (which the plane sees as windmilling drag) is therefore low. If the pitch was flatter (such as in many electric and gas engine props) and/or the plane's speed was higher, the drag due to these "profile losses" could be significantly greater.

For even more windmilling drag, we have to extract more energy from the airflow, and we have to find some place to dump it. Typically this is where the motor gets involved. The shaft bearings have friction, and the motor itself may be able to absorb energy.

For example, an electric motor usually converts electric power to mechanical power. However, many motors (including most of the ones in our models) can also function as a generator. If you drive their shafts with mechanical energy extracted from the airflow by the propeller, they can convert that mechanical energy into electricity. Of course that electricity has to have some place to go, so the motor has to be connected to a closed electrical circuit for this to work. The braking circuit in an ESC does this. Typically it connects a resistor across the motor terminals, and the electric current generated by the motor (driven by the windmilling prop) gets turned into heat, which is then dissipated into the air. Without such a closed circuit, the motor absorbs very little power, and therefore adds very little additional windmilling drag. It makes voltage as it windmills, but without a closed circuit to use that voltage there is no current. Power is volts times current, and if the current is zero, then there isn't any power absorbed by the motor, other than tiny amounts through things like eddy currents within the iron in the armature, etc. Thus, the windmilling drag of a prop + electric motor combination when there is no closed circuit across the motor terminals tends to be relatively small, limited mainly to just the power absorbed by the friction in the bearings and the profile drag of the propeller blades.

You can see this for yourself using an ordinary can motor and a direct drive prop, such as a Speed 400 motor with one of those plastic Gunther props used on so many of the arfs. If you spin the prop with nothing connected to the motor terminals, it's fairly easy to spin, and will continue spinning a relatively long time after a quick flip with a finger. Now, connect the motor's two terminals together with a piece of wire. That's right, just short them together. The wire closes the circuit, and unless you happened to use some superconducting wire (properly chilled, of course), it also has resistance (as do all the wires inside the motor itself). Current now flows, the resistance of the wire converts that current into heat, and the motor now absorbs a lot more power from the propshaft. You will feel a noticeably higher torque when you flip the prop, and the prop will stop spinning much sooner.

In the case of a model with a folding prop, you're trying to get the prop to fold in flight to minimize drag during power-off soaring, after an initial powered climb. A freely windmilling prop by itself has less drag than a stopped propeller, but more drag than a folded propeller. However, an unrestrained windmilling prop will probably windmill fast enough to make enough centrifugal force to hold the blades open, preventing them from folding properly. Applying a brake in the ESC (connecting a resistance across the motor terminals and completing the circuit) dramatically increases the windmilling drag, but it also restrains the prop enough to force its rpm to drop low enough to cause the blades to fold. Once the blades are folded, the drag is much lower.

A dead piston engine driven by a windmilling prop can absorb massive amounts of power. A piston engine is essentially an air pump, and it takes power, lots of it, to pump large amounts of air. In addition, since the prop is now making large amounts of thrust (negative thrust, in this case), it now has large induced losses (the losses that are the by-product of the thrust-making process, just as induced drag of a wing is the by-product of making lift). These large induced losses add even more windmilling drag.

The case of a turbine engine gets more complicated. There are essentially two main classes of turbines, "fixed shaft" and "free shaft".

A fixed shaft engine such as the Rolls Royce "Dart" in the Vickers Viscount or the Allison T-56 in a C-130 Hercules has the compressor, the compressor turbine (i.e.: the turbine wheels that extract power from the hot gasses to drive the compressor), the power turbine (i.e.: the turbine wheels that extract power to drive the prop), and the output to the propeller gearbox all connected to the same shaft. The key here is that the prop is mechanically connected to the compressor, and the compressor of a turbine engine is an air pump of truly colossal capability! The horsepower that the compressor can absorb can be far greater than the rated max horsepower of the entire engine. If you start dumping power from the windmilling prop into the compressor, the drag can be phenomenal.

In addition, turbines typically make large amounts of power, and therefore need large props to efficiently transmit that power. The area of the wing and tail surfaces that is immersed in the propeller slipstream can be a very significant portion of their total area, and it tends to be in the middle of them, right where most of the lift is produced. In the case of a windmilling prop making large amounts of windmilling drag, it is decelerating the airflow over those portions of the wing and tail, which dramatically reduces their ability to make lift, perhaps even enough to cause a stall.

Turbines typically have an overspeed governor as backup to the primary governor, to kick the blade angle up if the rpm gets too high. In addition, fixed-shaft turbines usually have something called an "NTS" (Negative Thrust Sensor) System that will increase the blade angle if it detects that the prop is making any more than a very small amount of negative thrust.

Free shaft turbines such as the Pratt & Whitney PT6A and PW100 series use a separate shaft for the power turbine and gearbox output, so the prop is therefore not directly connected to the compressor. On these engines it's actually possible to hold the prop stationary and run the engine (do not try this at home!). In fact on some versions of the PW100 series they include a prop brake for exactly this purpose, so that the engine can be run on the ground to provide electric power and "bleed air" from the compressor to run the plane's air conditioning, eliminating the expense and weight of an auxiliary power unit ("APU").

However, even though the power absorbed by a free-wheeling power turbine is far less than that of a fixed shaft engine's compressor (which is why free-shaft turbines don't normally need NTS systems), the profile losses of that huge prop can still be a major factor in the performance and controllability of the plane, especially if a failure in the propeller control system allows the prop to get to an abnormally flat pitch in flight. Normally there is a "flight-idle stop" in a turbine propeller's control system that prevents the prop from going below that pitch in normal flight operations. Below that pitch is what's called the "beta range". "Beta" is prop engineer lingo for the pitch angle of the blades. Above the flight idle stop, the blades go to whatever angle the prop governor thinks it needs to hold the desired RPM. However, in the ground and in reverse, the aerodynamics of the blades at very low blade angles (especially in reverse) would confuse a governor, so instead the blade angle is controlled by a cam, linked to the power lever in the cockpit. If a failure occurs, either mechanical or human-induced, that allows the blade angle to drop into the beta range in flight, the results can be catastrophic.

The EMB 120 crash that killed Senator John Tower and about 20 others some years ago was just such an event. There was abnormal wear in the components that connected each prop to its PCU ("Propeller Control Unit"). The spline in one of the PCU's wore enough to allow the prop pitch feedback to be lost and the blade angle to drop down to around flat pitch, well below the flight idle stop position, while the plane was on final approach at about 600 feet. The resulting asymmetric drag and the lift loss from twelve feet of prop blanketing a large portion of one wing made the airplane uncontrollable, and it essentially fell out of the sky. There have been other crashes of turbine aircraft due to human error (typically pilots deliberately trying to go below flight idle in flight in order to use the extra drag to help decelerate the plane from a too-fast or too-high approach) that had similar consequences, for much the same reasons. Mis-rigging the flight idle blade angle by just a few degrees can also result in the pilot having a very bad day. Propellers have a far more profound effect on both handling and performance than most folks (including some aero engineers who should know better) realize.

On very small and light airplanes with large props, we can see significant effects from windmilling drag as well, even if the windmilling drag is just that which is due to the prop's profile losses, without any significant additional drag due to motor braking effects. For example, on many of our "Roadkill Series" models, the prop diameter is a significant portion of the wing span. On one of the single-motor WW II warbirds it's typically about the middle 5" of a 20" to 21" span, or in other words about the middle fourth of the total wing span. In a small twin like the P-38 it is approximately the middle 37% of the span. Even with just the profile losses alone, the effect of this can essentially "shut off" the airflow to those areas of the wing and tail, resulting in significantly higher stall speeds, and very mushy control effectiveness. On these models it's best to leave the throttle open a little on final approach, providing just enough power to the motor to match the profile losses, resulting in what's referred to as a "zero thrust" condition.

On our RK Series Northrop XB-35 prototype this is essential. Remember how half the acceleration of the air through the prop disk occurs in front and the other half behind the disk? Even though the props are behind the elevators, they can still block the flow over the elevators. Without a little power left on there is nearly zero elevator authority available to flair for touchdown. That airplane definitely needs to be "flown-on" with about 20-25% throttle stick position.

So what about those rules of thumb about the windmilling drag being equivalent to a flat plate the size of the prop disk, or some other "one-size-fits-all-cases" formula? Well unfortunate thing about "rules of thumb" is that they are usually only good at measuring thumbs, and not much else. In the case of windmilling drag, as you can see from the above there can be a wide range of different results, determined by a whole shopping list of factors. If it's really an issue, then it needs to be calculated/measured for the individual case in question. In addition, since we're dealing with the performance of the blades' airfoils at highly negative angles of attack (which are often not as well documented as the behavior at positive angles), and may well involve negative stall conditions over parts of the blade (even less well understood), and that stalling on a propeller blade doesn't behave the same as stall on a wing (the flow tends to move outboard and then reattach, instead of staying separated), it can be fiendishly difficult to calculate a reasonably accurate value for windmilling drag under the best of circumstances. Ultimately some flight testing is usually required to verify the answers.

Stephen Smith asks: ...Many years ago I was told by one of the "old pros" that there was definitely a "best" position to stop a feathered 3 or 4 blade prop with a wing mounted engine. The 3 bladed solution was easy, that being the inverted "Y" position. The rationale was that less disturbed airflow over the top of the wing would result in more lift. The 4 bladed solution was a little more hard to accept, that was stopping 2 of the blades in line with the leading edge to produce less drag. Would you have a comment?...

General aviation multi-engined recips generally head straight for the nearest airport after an engine failure (at least if the pilot has any common sense!), so they aren't really in a position in most cases to benefit significantly from little details like that. Besides, the pilot of a light recip twin with one dead engine is too busy flying the plane to have any spare time to fiddle with trying to position the dead prop exactly right. Better to put up with what is at best a very minor bit of performance loss than to lose control of the airplane altogether while dealing with unnecessary distractions and details. If they were going to be spending a long time in the air (such as a four-engined recip on a transoceanic flight) where fuel consumption might be an issue and where there are other crew members available to deal with it, then it might be worth the trouble.

The friction in a turbine engine is so low that trying to get the prop to hold still at some specific position is an exercise in futility to begin with, and it isn't desirable anyway. When the prop is on such a low-friction mount, the minimum drag of the prop does not occur with the prop perfectly stopped, but rather at a very small but positive RPM. I believe it's because of the twist in the blades, the average angles of attack along the blades in feather (and therefore the feathered drag of the prop) will actually be reduced by letting the prop windmill very slowly. The feather blade angle is determined by this rpm, measured at the speed they will fly at in an engine-out situation. Note, this would not be desirable in a recip, because of the power absorbed by the engine while it's in motion.

Steven Bixby asks:

Having read all of this with interest, I'm still curious about some factors.

Given an aircraft with a fixed-pitch prop of medium size, like you might find on your average EFlyer:

On one end of the spectrum, there is a stopped prop - it will have a drag component that represents the parasitic drag the blades have, ignoring any airflow effects on the airfoil. From my understanding of the recent articles, this is the lowest possible amount of drag on the aircraft

Steven Bixby asks:

Then at the other end, lets say the prop is completely freewheeling - no shaft drag at all, or at least, just shaft drag from ball bearings, so it's relatively small. At this point, much of the drag to the airframe is from that of the 'wrong way' inefficiency of the prop being spun up by the airflow of the plane in unpowered glide.

However, if it's a fairly fast airplane and/or the prop has relatively flat pitch, the rpm will be high, which means the profile drag of the blades will be high, which also means that the blades have to make a lot of lift to overcome the profile drag, which means the induced losses become very high as well.

The extreme case of this is accidents like the crash of the EMB 120 airliner I described, where the prop (due to a catastrophic system failure) went to an extremely flat pitch in flight, resulting in very high windmilling drag plus the slowing down of the airflow over 12 feet of the wing, resulting in a total loss of control. Fortunately such failures are indeed extremely rare. BTW, that was a competitor's prop in that accident, but I am very familiar with its operation and with the circumstances of that accident. That was a free-shaft turbine engine, so the power absorbed by the engine was not significant, it was all the result of the abnormally low blade angle on a very fast airplane and the extremely high rpm it caused, which resulted in very high profile and induced drags from the propeller blades themselves. If the members of this group would really like to hear the grisly details of exactly how and why this failure occurred, I can relate that whole unfortunate tale another time.

In the aircraft industry (including propellers), any reasonable failures and any reasonable combinations of failures that can result in a catastrophic condition (i.e.: loss of the aircraft) must have a probability of occurrence of less than once per billion flight hours. The expected accumulated flight hours of the entire fleet of that type of aircraft should be comfortably less than that, so that according to probability that failure or combinations of failures should never occur.

Of course the whole problem with statistics in the first place is that even though the probability might be once in a billion flight hours, that says nothing about during WHICH of those billion flight hours that failure will occur. It could be in the last hour, or after 500 hours, or even in the very first flight hour; the probabilities for each of those different individual hours is exactly the same. In addition, the analysis of what exactly could happen in the case of something that hasn't happened yet is not easy to do or certain, either. Sorry, life isn't perfect, and engineering never has, is not, and never will be an "exact" science. No matter how thoroughly you engineer something, there are still ways for it to fail and non-zero probabilities of that actually happening. Also, another thing folks often overlook is that it's almost inevitable (as in right up there in the same league with death and taxes) that if you add safety in one area, you will almost certainly have to give up some safety in some other area.

Sitting at home in your living room is dangerous too. With all due respect to any product liability lawyers out there, we need to get over it and go on with our lives. Sorry about the rant, but after spending the last several decades watching designs get picked apart with twisted, self-serving logic by Monday morning quarterbacks who just happened to have membership cards for the Bar Association and ATLA, it's a bit of a sore point with me. My apologies for digressing, I'll get off of my soap box now and answer the rest of Steven's question.

Steven Bixby asks:

Finally, somewhere in the middle of all of this there is some combination of shaft drag from turning an electric motor or piston engine or turbine.

Steven Bixby asks:

So my question is, what does the drag profile look like on a chart, where the y-axis is the drag on the airframe and the x axis is the prop's rotating RPM - for a given average glide airspeed? I assume the profile of this chart would change for airspeed change, so lets stick to one airspeed. I'm also trying to ignore the energy dissipation. So this question is purely dependent on the prop's RPM vs the drag is produces.

What I'm envisioning is that it would be a curve, with the lowest point on the left (stopped prop)

Steven Bixby asks:

a highest point somewhere in the middle where the drag is 'optimized' by restricting the prop speed, and lower again where the prop is windmilling freely.

Is this a correct guess? Or - does anyone know of some charts of this nature?

The profile drag of the blades increases with the square of their airspeed (and therefore approximately with the square of the rpm, assuming the plane's airspeed is constant), and the power that represents increases with the cube of the rpm, so induced losses needed to overcome the profile drag will increase approximately with the cube of the rpm as well. Therefore the graph will have a high drag at zero rpm, decreasing very quickly to a minimum at some low but positive rpm, then increasing very rapidly from there.

Adding in the braking torque from a motor will increase the induced losses in proportion to that torque, which will tend to add a linear increase on top of the nearly exponential one I described above. The result will reduce the optimum rpm and make the drag at higher rpms increase even more rapidly. If the braking torque is very great, such as with many piston engines, it could shift the optimum rpm down to zero. This is why if your full-scale Piper or Cessna suffers a catastrophic engine failure in flight (i.e.: no hope of restarting it), it could be worthwhile to slow the airplane down to around stall speed long enough to stop the prop, before accelerating again to best L/D speed.

The bottom line is that if you plan to fly any significant distance with no power from the engine, keep your airspeed down, let the prop spin as freely as possible, or better yet get a feathering prop, or better still a folding prop. If you can;'t let it spin freely because your engine's minimum windmilling drag is high compared to the flat plate drag of the prop itself, try to stop the prop.

Steven Bixby asks:

Lets say, for the sake of this scenario, that there's no chance of supplying power to the shaft again, *but* you can supply drag to the propshaft. IE, a largish amount of drag to stop the prop, all the way down to no drag at all, where the prop windmills.

From what you describe, then, the curve on my imaginary chart would be something more of a 'V' shape, with the zero-RPM portion being moderately high in drag, then quickly tapering off to some minimized amount of drag when the prop is able to rotate at some RPM - then after that, the drag keeps increasing as the prop windmills.

Steven Bixby asks:

So, theoretically, if trying to stretch a glide back to the airport, you would want to allow the prop to windmill some small extent, rather than completely stopped or spinning at it's self-sustaining RPM when there's no shaft drag?

Assuming the prop can't be feathered, the lowest windmilling drag will occur with the airspeed of the plane as low as possible AND the drag on the shaft as low as possible. Any time you put a drag on the shaft, you start extracting power from the prop (which itself is extracting that power from the airflow though it, just like a windmill). This adds to the drag that the prop is already making by itself through its own profile losses.

Steven Bixby asks:

I'm guessing also that the ideal situation of being able to control the RPM of the windmilling prop doesn't really happen in real life, so this is probably just conjecture out of curiosity.

Or am I totally misunderstanding this discussion?

If you can't feather the prop, then the best option is to let it spin as freely as possible, and also keep your airspeed as low as possible (typically the best L/D speed of the airplane or a little below, to get the best compromise between aircraft performance and minimum prop drag). Any braking effect you apply to the shaft will just add to the prop's own drag.

The exception is when you have a motor that inevitably puts a lot of drag on the shaft with very little capability to change it, such as a piston engine. In this case the induced losses from letting the prop drive the motor are greater than the profile losses of the prop itself. The minimum then occurs when you eliminate the motor drag. Power is proportional to torque and rpm; if you make the rpm equal to zero, then it doesn't matter how much torque the motor applies to the shaft. Zero times any amount of torque is still zero, so the motor braking effects will not consume any power. If the savings from that are greater than the flat plate drag of the blades, then it's probably worth stopping the prop.

ADDENDUM:

I should know better than to write these things when I'm tired. There is one case with a free-wheeling prop where you would be better off to stop it. If the prop has very flat pitch, and the airplane's gliding speed is relatively fast in comparison to that pitch, it is possible that the natural unrestrained windmilling rpm of the prop is fast enough that the profile losses are greater than the flat plate drag of the stopped prop. It is probably not a good idea to just restrain it to a lower rpm through braking, since that introduces induced losses that will probably exceed the savings in profile losses. Of course all of this begs the question of why there is such a fine pitch prop on such a fast airplane in the first place!

Don Stackhouse
DJ Aerotech