Anda di halaman 1dari 12

First off, here's the nomenclature I will use:

Nomenclature used in this article:


F = thrust (N)
m_dot = mass flow rate (kg/s)
Ve = exit velocity of the airflow through the propeller (sometimes called the induced velocity, inflow
velocity, or velocity induced by the propeller) (m/s)
Vac = aircraft airspeed/velocity (not ground speed) (m/s)
Vpitch = propeller pitch speed (m/s)
= air density (kg/m3)
A = area through propeller, normal to airflow (ie: area that the propeller sweeps) (m 2)
P = pressure (Pa, N/m2, or kg/(m*s2))
SLUF = Steady, Level, Unaccelerated Flight
Here is some nomenclature *not* used explicitly in this article, but you should at least be
aware of it:
prop = propeller efficiency (-)
motor&prop = power system efficiency (of the motor, propeller, ESC, and wiring combined) (-)
L = lift (N, or kg*m/s2)
D = drag (N, or kg*m/s2)
W = weight (kg)
CL = lift coefficient (-)
CD = drag coefficient (-)
q = dynamic pressure (Pa, N/m2, or kg/(m*s2))
S = aircraft planform area (m2); could also be used as propeller planform area when looking at a
propeller as a wing

Background Equations
To get started, we must go all the way back to Newton. Many people say that Newton's 2nd Law
states that "force equals mass times acceleration:" F = ma. This is not quite what his law really
states. Rather, this is only a very special case of Newton's 2nd Law. Newton's 2nd Law really states
that "force is equal to the time rate of change of momentum," or F = d(mv)/dt, where momentum is
mass times velocity, or mv. In many cases, mass is constant, so it can be removed from the
derivative, and you get the following:

Equation 1: Newton's 2nd Law, where mass is held constant.

I should probably bring up Newton's 3rd Law here, which states that "for every action, there is an
equal and opposite reaction." A propeller produces thrust by accelerating air molecules. If we could
measure the acceleration of each air molecule by the propeller, and also know the mass of each
molecule being accelerated, then the thrust produced by each air molecule would be the mass of the
molecule times the acceleration of the molecule. The total thrust of the propeller would be the sum
of all of the individual thrusts produced by all of the air molecules being accelerated at any given

instant. Remember that to get F = ma above, we had to assume thatthe mass of each molecule is
constant. What happens, however, if we look at the problem a bit differently and assume that the
velocity of each molecule is constant instead? This brings us to our second approach.
A second approach is to consider a mass flow rate, or m_dot (written as m with a dot over it),
instead. Any time you see a dot over a variable, that means it is the derivative of the variable with
respect to (w.r.t.) time. So, m_dot = dm/dt. Its units are kg/sec. If we take Newton's 2nd Law and
hold the velocity of each molecule constant instead, we get the following:

Equation 2: Newton's 2nd Law, where velocity is held constant.

Now, we are considering the air molecules to have a mass flow rate through a propeller, at a
constant velocity. This is a very common starting point for jet propulsion engineers and rocket
scientists, as this equation is very common in both of those fields. For a static aircraft, we know that
all of the velocity of the air molecules passing through a propeller contributes to the thrust since the
air molecules started out stationary, and were accelerated to this velocity.
Therefore, propeller thrust, F, for a stationary (static) aircraft is:

Equation 3: Theoretical static thrust

For a moving aircraft, however, only the velocity of the air which is due to the air having
been accelerated by the propeller is what contributes to the thrust. In other words, only the change
of velocity is what matters:

Equation 4: Theoretical dynamic thrust

Note that based on this equation, as the aircraft velocity, Vac, increases, thrust decreases. This is
due to the fact that the propeller exit velocity (or induced velocity) is approximately constant, and
therefore the result of (Ve Vac) approaches zero as the aircraft top speed is reached.
Since m_dot is equal to the density of the air times the cross-sectional area through which the air is
flowing, times the velocity of the air, we get:

Equation 5: m_dot (left) & thrust w/m_dot plugged in (right)

where A is the cross-sectional area, or the rotor disc area covered by a spinning propeller. A is
therefore the area of a circle:

Equation 6: Area of a circle

where r is the propeller radius, and d is the propeller diameter, in units of meters.
Substituting in A above, we get the following for the theoretical propeller dynamic thrust equation:

Equation 7: Theoretical propeller dynamic thrust equation with area substituted in

Simplifying, by factoring the area term out we get:

Equation 8: Theoretical propeller dynamic thrust calculation

Remember, Ve is the exit velocity of the air, or the induced velocity of the air by a propeller, through a
propeller, and Vac is the aircraft velocity (or, more specifically and accurately stated, the freestream
velocity). F is thrust, is air density, and d is the propeller diameter.
Ve is assumed to be approximately equal to the pitch speed of the propeller. The pitch of an RC
propeller is a unit, usually in inches, and represents the theoretical distance forward which a
propeller would move, based on its pitch angle, if it were rotated exactly one revolution in a magical
air which is more like Jell-O. Pitch speed is solely dependent upon the propeller RPMs and pitch,
and is as follows:

Equation 9: Propeller pitch speed

Now, you can plug the pitch speed (Eqn. 9) into Eqn. 8, in place of Ve, to get an initial estimate for
propeller thrust. If you do this, however, you'll notice that the equation comes up with absolutely
horrible estimates, and doesn't even show the appropriate trends for how thrust changes with
varying diameter and pitch propellers. Global trends are good when looking at changing
diameter or pitch, independently, but if you look at changing diameter and pitch together, trends are
not correct. So, here comes the hard part, but first, let's briefly talk about the air density term.
To get air density, , you could use the ideal gas law, P = RT. Using this equation you can solve
for density, rho, then plug in P, R, and T. You can get P and T, respectively, by measuring static air
pressure and temperature of the air where the propeller is being tested. If you don't have an
absolute pressure transducer, or a barometer, you can get static air pressure from a local weather
station, using this equation:

Equation 10: Converting the local weather station pressure to the actual static
pressure at that weather stations altitude

where w.s. = weather station.


You can, of course, get temperature from a thermometer or from the local weather station too. Make
sure temperature is in Kelvin, and pressure in Pascals. R is the specific gas constant of air, or
287 J/(kg*K).
This equation is essential since all weather stations provide only static pressures corrected for
altitude, as opposed to true, raw static pressures.
To keep things simple, however, and primarily since I have no idea what the static pressures were
during each of the thrust data points I collected from other sources, I just had to assume sea level
standard day, so I assumed that air density, = 1.225 kg/m3.
Now, setting Vac to zero, we have the following for the static thrust calculation, with all pieces
plugged in, and prop diameter, d, and prop pitch, pitch, being in units of inches, and thrust, F, coming
out in units of newtons (N). The 0.0254 is a conversion factor to convert inches to meters, since
there are 0.0254m/in.

Equation 11: Final theoretical static thrust calculation

At this point it is time to figure out empirical correction factors.

Theoretical to Empirical: Making the Equations Work


For Real Life
At this point we are ready to find empirical, or based on experimental data, correction factors to
make the thrust equation accurate. Now that it is based on some physics, we have something to
start with. However, it is important to note that the thrust equation derived above has some huge
assumptions. For instance, the assumption that induced velocity is approximately equal to pitch
speed (Ve Vpitch) is far from real life. Not only is it unrealistic, but we also need to realize that the
inflow velocity isn't even constant across the cross-sectional inflow area, or rotor disc area.
The helical twist on the propeller is an attempt to make the inflow velocity constant across the rotor
disc area, but due to many factors, it is not. Some of these factors include:

1) Non-constant inflow velocity profile across the propeller: I'd have to double-check to find the
precise value, but to make my point: the majority of the thrust is produced somewhere around the
70% span, measured as 0.7 x r from the propeller hub, towards the propeller tip, where r is the prop.
radius. This is for some of the following reasons: the propeller is like a wing moving through the air,
where the lift (thrust in this case) is proportional to the square of the linear velocity of the air striking
the wing. The linear velocity of the air striking the propeller is linearly proportional to the distance
from the hub. Therefore, maximum linear velocity occurs at the propeller tip, and at the center of the
hub, it is zero. To compensate for this velocity difference, the helical twist is such that the propeller
near the tip will have a relatively low angle of attack to the air, but the propeller near the hub will
have to have a much higher angle of attack to the air. However, the motor gets in the way at the
hub, no blade can exist at the hub, and even if blade could exist at the hub, the linear velocity at the
hub is zero. So, around the hub will be a low-flow zone. Additionally, the angle of attack of the
propeller blades near the hub may be high enough that the blade is stalled in that region. A stalled
blade can still produce thrust, but not nearly as much as an unstalled blade, and not nearly as
efficiently. Around the propeller tips, the inflow velocity will also suffer due to energy losses due to
propeller tip vortices. Therefore, the maximum inflow velocity, and consequently, thrust, occurs
somewhere around the 70%~80% radial span of the prop, measured from hub to tip.
2) Number of propeller blades affects the "grab" of the blades on the air, which affects how close the
air will get to approaching the propeller pitch speed. More blades = better "grab" on the air, which
means higher Ve, but it also means more blockage to inflow velocity. Each additional blade
decreases propeller efficiency, while increasing raw (total) thrust. Eventually, by adding too many
propeller blades, you will have blocked too much of the inflow area and caused so much additional
skin friction and parasitic drag, that total thrust will begin decreasing rather than increasing as you
add additional blades.
3) Propeller blade chord, planform area, and shape--I don't even take these things into account at all
in my equation, as I wanted to keep the equation simple.
These are just two factors I can think of, but the point is that (Ve Vpitch) is truly a poor
assumption. Since it's the best I've got, however, I'll work with it.
At this point I started collecting static thrust data, from as many sources as I could find, to develop
my empirical correction constants. I needed to know at a bare minimum the propeller pitch,
diameter, and RPMs, as well as the measured thrust for each data point. I ended up finding 149
static thrust data points to use, from places like OS engines, Grayson Hobby, and even HobbyKing,
pulling a few of the data points straight out of the user reviews.
Here are the 149 data points, for a whole variety of propeller sizes and RPMs. The thrust values
above 4kg were for large aircraft, some running gasoline engines and wooden props I think, but
most from OSEngines, running electric motors on 6~7S LiPos. The low line of values along the
bottom right, going up past 22,000 RPMs, were the values from Grayson Hobby, as RCPowers (their
affiliates), like to use very high-kV motors and small 5x5~6x4 props on their parkjets, which makes
for very high RPMs but relatively low static thrusts. The bulk of the data, in the "cloud" around
7,000~12,000 RPMs, is primarily from the OSEngines page previously linked above.

When I first plotted my estimates, compared to the actual thrust values, my estimates were *not* as
you see them on the plot now. Rather, they were far off of the real values. To correct my equation I
did the following:
First, I logically deduced (or rather hypothesized) that both the thrust, and also efficiency, of a prop
must be related to its diameter to pitch ratio, since the diameter of a prop directly affects the incident
(tangential) velocity (and hence also dynamic pressure and Reynolds number) of air striking the
blade at any given location of radial distance, r, from the hub. Hence, larger diameter props will
have not only increased thrust, but also increasedefficiency over a prop of the same pitch but
smaller diameter, since Re will be higher (and flow remain more attached) as you increase r.
However, pitch must also affect not only thrust but also efficiency. For any two props of the same
type and diameter, but different pitches, the prop with lower pitch will have lower incident angle of
attack than the prop of higher pitch. Lower incident angle of attack means better flow attachment,
less separation, less induced drag, and therefore less profile drag/pressure drag too. Therefore,
higher diameter and lower pitch props must be more efficient, but how are diameter and pitch
related? Well, I decided to do aconstant times the diameter to pitch ratio, then use goal seek in
Excel on this constant, to find out what value for that constant gave me the lowest deviation of
calculated static thrust from actual static thrust. At this point, the equation looked like this, where I
was goal-seeking to find the optimal value for the unknown empirical constant, k:

Equation 12: First attempt at finding an empirical correction factor, k*(d/pitch),


in this case, using goal-seek in Excel to optimize k

Pretty soon, however, I realized that the relationship of d/pitch to static thrust was nonlinear, so I
added in a second constant to raise the diameter-to-pitch ratio to some power, as shown here:

Equation 13: Second empirical attempt, this time with two unknown empirical constants, k1 and k2

where k1 is the "coefficient constant" I describe below, and k2 is the "power constant."
I then used trial and error to juggle around the two constants for several hours, incrementally
changing one, and goal-seeking the other. Repeated trials led me to choose 1.5 as the value for the
power constant, k2. This value is still enigmatic to me, and I do now know why it works best. The
value 1.5, or 3/2, just seems to be too perfect, and therefore perhaps has an analytical basis, but I
just dont know what. After choosing 1.5 as the power constant, performing the goal-seek on k1, the
coefficient constant for the diameter to pitch ratio, gave me a value of 1/3.29546 (or 0.30344777), as
the optimal coefficient constant necessary to minimize the average error between the predicted and
actual thrust values for the 149 data points I looked at.
At this point I was complete, and that brings me to my final static thrust equation at this time, shown
here:

Equation 14: Final static thrust equation, with empirical correction factor

Finally, I added Vac back in to the equation (refer back to Equation 8) in order to make it applicable
to dynamic thrust calculations, and I compared it against one single dynamic thrust test run
from Matthew McCrink of Ohio State University. Since my equation produced reasonable results,

even if only only for a single dynamic thrust data set, I consider it to be a success, though future
work will include applying an empirical correction constant to the dynamic thrust portion of the
equation as well, in order to improve the accuracy of the zero-thrust, or x-intercept point. My static
thrust empirical correction factor makes the y-intercept point accurate, so now if I can make the xintercept point more accurate, I can simply draw a line between the two in order to have the complete
dynamic thrust estimate for any given airspeed, Vac. After that, I can worry about the minor nonlinearity in the line segment that is noticeable between those two points, but it is so minor that
perhaps it can just be ignored. Additionally, I plan to look at some blade-element-theory type
techniques, breaking the propeller up into small segments and independently looking at each
segment, considering such things, perhaps, as apparent angle of attack to the inflow (to see which
segments of the prop may be stalled at any given time), and inflow velocity for that segment of the
propeller, since as discussed previously, I know the inflow velocity is not constant across the inflow
area. This will require using an annular (ring-like) cross-sectional inflow area (rather than circular),

for each prop segment I look at, since I will be looking at only segments of the prop. at a time. These
are just some ideas that I have, however, as I'll have to see what's truly feasible as I think through and
attempt the next steps.
One other idea, which I think is very feasible and will work very well, is to use multirotors/multicopters (ie: quadrotros, tricopters, hexacopters, etc) to further refine my static and
dynamic thrust estimates. For any vehicle in a steady, hovering state, the thrust is equal to the
vehicle weight. Knowing this, you can take a multirotor vehicle, hover it, use an Arduino
microcontroller to do real-time measuring and datalogging of each motor's RPMs (I've already
worked out an initial version of this code), and then apply my equation to refine the thrust estimate.
The thrust each propeller is producing is simply the total vehicle weight divided by the number of
propellers on the vehicle. Ex: for a standard quadrotor configuration, for a quadrotor weighing 1kg,
the thrust of each propeller, in any given steady, horizontally-stationary (horizontal velocity = 0),
unaccelerated hovering state, is 1kg/4.
Finally, here is my dynamic thrust equation, in its current final form, with the empirical correction
constants, as discussed above, applied. Note here, however, I previously chose to use V0 to represent the
aircraft velocity, instead of Vac. They are the same thing though, so ignore that minor difference. These are
the same equations that I show in Part 1 of this article.

Equations 15 (top) and 16 (bottom): Final, empirically-corrected, dynamic thrust equations for a propeller (top:
expanded form, bottom: simplified form)

Here is the single dynamic thrust test point from Matthew McCrink. I discuss this plot more

thoroughly in Part 1 of this article.

Here is a diagram I am adding for completeness, in case you'd like to study it for additional insight:

Application & Conjecturing:


One potential application of my dynamic thrust equation is to look at how power system choice
affects max speed of your aircraft. Realistically, we would need some more info. about your aircraft,
such as drag coefficient, so that in SLUF (Steady, Level, Unaccelerated Flight) we could assume
thrust = drag, but let's just make some generalized statements to get my point across for now. The
point is that your airplane can never fly faster than the airspeed indicated at the x-intercept of the
thrust vs. airspeed curve. If your power system produces zero thrust at a given airspeed, then your
plane will never fly quite that fast. If it was perfectly drag-free, it would fly exactly that fast.
Therefore, my equation at least gives us a decent idea of what the max theoretical, drag-free
airspeed might be for a given setup. Using my equation, let's see how prop pitch, diameter, RPMs,
and airspeed are related. Let's take a look.
Using my dynamic thrust equation, I created this plot below, trying to use reasonable values I
guestimated based on my personal RC experience, for one particular power system you might
have.

Notice that dynamic thrust is linear. Based on my equation, this makes sense, because although
thrust has a squared relationship to induced velocity through the propeller, it has only a linear
relationship to aircraft speed, or airspeed. Looking at Figure 4 on this website
(http://www.allstar.fiu.edu/aero/BA-Form&gra.htm), you can see that thrust of a prop-driven airplane
does indeed taper off almost linearly. You can also see that in the wind tunnel data I posted above.
This is in contrast to jet airplane. From a few plots I saw in my core aero book (ISBN-13: 9781563477010, found here), for a jet airplane, thrust vs. airspeed is basically just a straight horizontal
line. I believe this is because for a jet airplane, V0 (ie: Vac) is insignificant compared to Ve (ie: Ve >>
Vac), so thrust stays nearly constant across various aircraft airspeeds.
To make a point, looking at the various curves on the thrust vs. airspeed plot above, you can
see that despite having the lowest static thrust, the smallest diameter, highest-pitch prop in
this case (7x6 at 13,500 RPM) has the highest potential to get the plane going fast. In other
words, it has the highest high-speed dynamic thrust, since its curve intersects the x-axis the
farthest to the right. Assuming that your propeller/battery comparisons maintain constant
power for a given motor (since a power system: motor, ESC, battery, etc., is in fact power
limited, due to heat generation), that is always the tradeoff: high static thrust versus high
dynamic thrust (& hence also high pitch speed, & high top speed). In this made up case, the
7x6 prop. has the lowest static thrust, but the highest dynamic thrust. Therefore, it might be
a poor choice on this power setup for a 3D plane, which needs very high static thrust to be
able to hover and accelerate vertically, but it might be ideal on a sport plane you want to fly
around fast, since high-speed flight requires a high pitch speed and higher dynamic thrust.

The fast plane sacrifices high static thrust, as indicated by a lower thrust curve near the left
of the plot, in exchange for higher dynamic thrust and top speed, as indicated by extending
the curve to the right on the plot. This trade-off in choosing your propeller is like shifting
gears in a car. Either you get a really good top speed with lousy initial acceleration (like
5th gear), or you get a really low top speed with really good initial acceleration (like 1st gear).
Assuming constant power draw through your given motor/battery/ESC setup, a 10x4.7 prop
is more like 1st gear, and a 7x6 prop is more like 5th gear.
One more note: the area under each curve is equal to the Power of the propulsion system. If
put in standard metric units, the y-axis, thrust, would be in Newtons, and the x-axis, airspeed,
would be in m/s. The area under the curve would be in units of thrust x airspeed = Nm/s = J/s
= W = Watts. This is the unit of power. There are 745.7 watts per horsepower (W/hp).
In order to make the hypothetical propeller comparisons above more realistic, I need to redo
that plot, this time ensuring that the areas under the curves are equal, thereby ensuring
constant power for each propeller being used.
[Todo: redo the above plot, ensuring equal areas under the curves, & thereby equal power
draw for each setup. Also, calculate the power for a given system, & see if this correlates
accurately to the real-life measured power draw through a system, or if I am
misunderstanding something ~GS.]
One more thing to keep in mind: These plots above assume constant RPM. However, as
Vac increases, the RPM data from the wind tunnel experiment above shows that RPM also
increases. This seems explainable to me by the fact that effective Angle of Attack (AoA, or alpha) on
the prop blades decreases with increasing airspeed (see my figure above, showing the side profile of
the propeller). Since effective AoA decreases, pressure drag and induced drag will decrease on the
blades. This, perhaps, is what allows the propeller RPM to increase. I hypothesize that decreasing
effective AoA on the prop blades reduces resistive torque, which allows the motor to speed up a
little. I will have to look into this phenomenon more in the future. If you look at the Cessna 172 data
in Figure 4 of this source, you'll see that the thrust curve is very slightly parabolic, bowing up in the
center. The wind tunnel data also slightly shows this phenomenon. I would like to know what
causes this, so that is something else I will look into in the future, as I continue learning and growing.

Anda mungkin juga menyukai