AIAA JOURNAL

Vol. 59, No. 3, March 2021

New Development of Classical Actuator Disk Model

for Propellers at Incidence

Rafael L. Rubin∗ and Dan Zhao†

University of Canterbury, Christchurch 8041, New Zealand
https://doi.org/10.2514/1.J059734
The actuator disk model thrust formula was mathematically expanded in series and divided into two parts to show
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

that propellers at incidence comprise an axial and a wing lift equivalent component. Both components share a
common induced speed w. This is done by considering an enhanced disk area for momentum balance, to match
Glauert’s hypothesis mass flowrate. To shed light on the theoretical developments, wind tunnel tests were conducted
on a two-bladed propeller at angles of incidence ranging from 0 to 90 deg. The wing component is shown to increase
with airflow velocity and angle of incidence. The axial component decreases with V, for all angles. The generally
observed thrust increase with angle of incidence is explained, by the theory, to be mostly due to the wing component
contribution. The theory also explains why at angles of incidence above 60 deg propellers inherently behave
differently than at lower angles. While thrust decreases with V at lower angles, it grows with airspeed at an angle
of incidence of approximately 60 deg or higher. This behavioral inversion happens as the wing component positive
sensitivity to V overcomes the negative sensitivity of the axial component. A simplified formula is presented for
predicting thrust at a given angle, based only on propellers data at an angle of incidence of 0, regardless of blade
geometry.

Nomenclature V disk = slip-stream velocity at rotor disk, m/s

AOA, αp = propeller’s angle of incidence, deg V disk = slip-stream velocity vector at rotor disk, m/s
V ult = slip-stream ultimate velocity, m/s
αslp = slip-stream angle of incidence at rotor disk, deg
V ult = slip-stream ultimate velocity vector, m/s
αult = slip-stream ultimate angle of incidence, deg WF = wing factor
CT = thrust coefficient, equal to T∕ρn2 D4 w = propeller’s average induced velocity at rotor disk,
D = propeller’s diameter, m m/s
e = theoretical entrainment factor w = propeller’s average induced velocity vector, m/s
eN = direction versor of propeller normal force due to wo = propeller’s average induced velocity at disk in
angle of incidence static condition, m/s
eT = thrust direction versor or propeller axial direction wult = ultimate induced velocity, m/s
Fx = wind tunnel net force measured on x axis, N ∂∕∂αp = derivative or sensitivity with respect to AOA, 1/deg
Fz = wind tunnel net force measured on z axis, N
∂∕∂V = derivative or sensitivity with respect to V, s/m
J = advance ratio, equal to V∕nD
ε, ϵ = angle between T and V disk , deg
Mp = propeller’s pitching moment due to angle of
ϵult , ϵult = angle between propeller axial direction and V ult ,
incidence, Nm
deg
My = wind tunnel net moment measured on y axis, Nm
ρ = atmospheric density, kg∕m3
m
_ = mass flowrate, kg/s σ = standard deviation
Np = propeller’s normal force due to angle of incidence,
N
n = propeller’s frequency, Hz Subscripts
p = static pressure, Pa ult = ultimate or far wake
patm = atmospheric pressure, Pa jV0 = static condition
q = freestream dynamic pressure, equal to 0.5ρV 2 , Pa jAOA0 deg = no incidence condition
RPM = propeller’s revolution per minute, rpm
Sdisk = propeller’s swept area or rotor disk area, m2
Seff = propeller’s momentum balance effective area, m2 I. Introduction
Swing = equivalent wing area for T wing calculation, m2
T
T axial
=
=
propeller’s thrust, N
axial component of thrust, N
L ARGE interest in vertical takeoff and landing (VTOL) tilt-mode
aircraft development appeared in the late 1940s, requiring
research on propellers performance for a wide range of angles of
T wing = wing lift equivalent component of thrust, N incidence. Recently, VTOL unmanned aerial vehicles (UAVs) projects
V = freestream or wind velocity, m/s are causing the resurgence of the subject for small scaled propellers. In
V = freestream velocity vector, m/s most cases, propellers are designed either for near-axial flight for
airplanes or to perform as helicopter rotors. The transition between
these flight modes in tilt-mode vehicles is mostly brief. Nonetheless,
Received 7 May 2020; revision received 2 September 2020; accepted for understanding and modeling propellers behavior at all angles is crucial
publication 2 September 2020; published online Open Access 11 December to determine those vehicles performance.
2020. Copyright © 2020 by Rafael Levy Rubin and Dan Zhao. Published by In the past century, several models with different accuracies and
the American Institute of Aeronautics and Astronautics, Inc., with permission. complexities were developed. The simplest model to assess propeller
All requests for copying and permission to reprint should be submitted to CCC thrust performance is the classical momentum theory or actuator disk
at www.copyright.com; employ the eISSN 1533-385X to initiate your request.
See also AIAA Rights and Permissions www.aiaa.org/randp. model. The model was first introduced for marine propellers by
*Ph.D. Candidate, Department of Mechanical Engineering; Private Bag Rankine [1] in 1865 and Froude [2] in 1889. Later, in the beginning
4800. [email protected]. of the 20th century, it was adopted for airscrews with the advent of the
† airplane. The theory applies the basics of fluid dynamics conserva-
Associate Professor, Department of Mechanical Engineering, Private Bag
4800; [email protected]. Associate Fellow AIAA. tion laws [mass (continuity), momentum, and energy] to provide a
1040
 RUBIN AND ZHAO 1041

general understanding of the propeller performance. Its assumptions at the disk edge must be included through a correction represented by a
disregard the propeller blades’ geometry and details of the flow about discrete vortex carrying an edge force. Goorjian [36] also states an
them. The propeller is considered as an infinitely thin actuator disk inconsistency in the general momentum theory, as mutual interferences
that impels a sudden increase in pressure on the fluid as it flows across between different annular elements are not considered. The existence
its surface. This causes an acceleration of the flow, so an induced of this inconsistency was also known to Glauert, but only in recent
velocity increment is modeled at the disk. The flow is assumed to be years, the errors associated with it have been quantified. This has been
incompressible and inviscid and rotation is neglected. Also, the veloc- done by Sorensen and Mikkelsen [37], Van Kuik and Lignarolo [38],
ity and pressure over the disk are considered to be uniform, and so and Bontempo and Manna [39] for the axial momentum theory, and in
thrust is distributed evenly over its surface. The major limitations of the the general case, to Sorensen [40], and Bontempo and Manna [41,42].
theory are that it does not take profile drag losses of the propeller blades Conway [43] developed an analytical closed solution for the linearized
nor blades’ tip vortices and rotation effects into account. Despite the actuator disk with arbitrary radial load distributions. As an extension of
simplifications assumed, the patterns of velocities and pressures in the linearized solution, a semi-analytical method was then developed
the actuator disk model have been verified experimentally [3], and for a nonuniform heavily loaded disk in [44]. Pitt and Peters [45]
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

although it is not very accurate for power estimates [4], it provides a presented a linear, unsteady actuator model considering the dynamic
very good approximation for thrust. The theory is, however, not inflow behavior of helicopter rotors. A model for axial and for skewed
expected to yield a good basis for rotor in plane analysis, and hence flows by Morillo and Peters [46] presents the solution of the complete
it is not suitable for propeller design by itself. The derivation of the velocity field above the disk by converting the potential flow equations
equations involved in the classical axial momentum theory for pro- into ordinary differential ones. Recently, Rosen and Gur [47] devel-
pellers can be found in several sources [5–8]. oped an axisymmetric axial actuator disk model including radial and
Subsequent models with higher complexity arose in the wake of tangential induced speed components. The model defines a pressure
the actuator disk theory. General or extended momentum theory, where ratio factor that depends on the blades geometry to conclude the
the incorporation of rotation is added to the model, was developed momentum balance, for which calculations are performed iteratively.
by Betz [9]. The blade element theory, which considers the geometry Later, Kominer and Rosen [48] adapted the model for asymmetric
of the blades, was first devised by Drzewiecki [10] in 1920. It ignored skewed inflows.
the effect of the induced flow inside the stream tube [11], as defined in Although the previously cited actuator disk models improve accu-
the momentum theory, taking into account only the freestream velocity racy, many of them are cumbersome to implement and end up missing
V and the propeller rotation for every blade element analysis. The more the advantage of the momentum theory simplicity. Another approach
precise blade element-momentum theory (BEMT) incorporated the for high accuracy is to use computational fluid dynamics (CFD) to
induced velocity. Other variations of BEMT calculate the induced solve the three-dimensional (3D) flow equations. These models may
velocity in different ways. Goldstein’s classical vortex theory [12] include viscosity, nonuniform propeller inflows, rotation, and more
related the induced velocity to the bound circulation around every blade complicating assumptions such as the addition of nacelles and wing
element. Vortex theory studies were later enhanced by Theodorsen [13]. interferences in the analysis [22,49,50]. Often, high-complexity
Extensive compilations of rotor aerodynamics analysis by Joukowsky in CFD models are coupled with experimental tests for validation. In
light of momentum theories and under vortex theories are presented in general, they are also computational expensive and time consuming
Refs. [14] and [15], respectively. to implement.
Previous relevant studies to incorporate nonaxial flow conditions In this Paper, we conduct theoretical studies by following the
were made by Ribner [16,17], who devised formulas, using BEMT, to classical momentum theory and Glauert’s hypothesis [7] for aerody-
calculate side forces and moments that appear when propellers namic propellers at incidence. By considering an enhanced mass flow
operate at incidence. The development assumed low angles of inci- rate through the disk, this new development allows for a mathematical
dence. De Young [18] later modified the Ribner formulas and decomposition of thrust into two parts: T axial , which is dependent on
expanded the use for high angles and also analyzed the effects of the axial component of the oncoming wind speed, and T wing which is
incidence on thrust. Experimental investigations have also been made sensitive to the wind component parallel to the rotor plane. This is done
in order to understand the behavior of full-scale propellers at a wide in Sec. II. The development quantifies the contribution of each com-
range of incidence in Refs. [19–21], in which it has been shown that ponent to total thrust and helps clarify, under the scope of momentum
thrust increases with increasing AOA and that thrust grows with the theory alone, why rotors in forward flight behave as wings and why
advance ratio J, at high AOA, as opposed to the effect at low angles. thrust increases with AOA. To validate the theoretical findings, exper-
Recently, experiments on several small scaled propellers for UAV imental wind tunnel tests are conducted on a two-blade propeller at
applications [22,23] also showed that for AOA < 60 deg, CT dimin- angles of incidence ranging from 0 to 90 deg. This is described in
ishes with the advance ratio J, and vice versa for higher angles of Sec. III. The effects of revolutions per minute, AOA, and oncoming
incidence. Previous studies on wind turbines under yaw conditions flow velocity on thrust measurements are experimentally evaluated,
are reported in Refs. [24–31]. and the propeller‘s behavior is explained, through the influence of
Glauert [7,32] conducted an analysis on helicopter rotors in for- the two components, in Sec. IV. Furthermore, as a consequence of the
ward flight using momentum theory. Although it is mentioned that no analysis, a simplified formula is derived to allow for the estimation of
solid proof has been given, he introduced a thrust formula for angles propellers thrust when at incidence, based on the performance data
of incidence that reverts to the classical static thrust equation in the measured at AOA  0. A discussion on slip-stream parameters in
case of zero forward speed, and at the other extreme at fast forward response to AOA is also provided. Key findings are summarized
flight, the equation takes the form of the elliptic wing lift formula, in Sec. V.
implying that the rotor behaves as a wing in that case. Bramwell [33]
investigated the validity of Glauert’s formula by solving the linear-
ized Euler’s equations with small disturbances in the velocity field, II. Momentum Theory for Propellers at Incidence
showing that, for lightly loaded propellers, the model is valid for the A. Theoretical Entrainment Factor e
axial case and for high speeds at 90 deg, regardless of blade geometry. The classical momentum theory models an inviscid flow, with a
Moreover, in the latter case, the linear theory appears to hold also for uniform pressure jump and an average uniform induced speed at the
nonuniform load distributions. Glauert’s hypothesis for propellers at actuator disk. Here, rotation is neglected. To apply the principles of
incidence has been widely used also for a wide angle of attack (AOA) momentum theory at incidence, it is first necessary to define the mass
range [3,4,8,11]. flow rate through the disk and the boundaries of the stream tube. As
The appeal of the actuator disk momentum theory simplicity the airflow velocity at the disk V disk is not normal to the disk plane
attracted further investigation to determine its limitations and possible (see Figs. 1a and 1b), initially one would assume the boundaries of the
improvements of its accuracy. Van Kuik [34,35] mentions that the stream tube touching the disk rim and the normal component of V disk ,
average induced velocity calculated with the assumptions of the in other words, V cos αp  w, to estimate the mass flux through the
classical theory is underestimated. For a uniform load, a singularity disk. However, it has been determined experimentally that the rotor
 1042 RUBIN AND ZHAO
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

Fig. 1 Scheme of a propeller at incidence: a) effective stream tube definition and b) illustration of velocities vectors.

entrains air from outside the rim stream tube [8]. The phenomenon area defined by the propeller diameter and normal to V disk , comes
can be explained to be caused by radial pressure gradients effects not from rotor analysis in forward flight and uses the analogy to the wing
modeled in the classical momentum theory. The radial pressure theory [3,5,7,52]. It was presented in Glauert’s hypothesis [7].
gradients cause an increase in mass flow that grows with AOA, by The freestream velocity projected on the propeller’s reference
means of tip vortices. A rigorous description of the phenomenon can frame can be shown as
be found in Refs. [34–37]. As the radial momentum balance is
disregarded, the uniform pressure load at the disk would cause an V  −V cos αp eT  V sin αp eN (3)
infinite radial pressure gradient at the edge and a velocity singularity,
which are inconsistent. This inconsistency in the model can only be By following the momentum theory, where the velocity at the disk
resolved through the addition of edge vortices, which act as natural is the sum of the axial induced velocity by the propeller and the
concentrators (see Refs. [34,51]), which increase the mass flow freestream speed, V disk  V  w, then it is quite obvious that the
through the disk. This effect grows with AOA and with the speed following two relationships hold:
component in the rotor plane (see Ref. [48]).
To incorporate an increase in mass flow rate to the model, an V disk  −V cos αp  weT  V sin αp eN (4a)
entrainment factor e is applied to the disk area, and a new effective q






























































⋅
area Seff  e Sdisk is defined. Then, m can be expressed as jV disk j  V cos αp  w2  V sin αp 2 (4b)
m
_  ρV cos αp  wSeff (1) As illustrated in Fig. 1b, it can be determined from vector geometric
Shapiro [3] also postulates the hypothesis of a much larger region relationship that
of induced flow affected by the rotor at incidence and assumes a wider V disk cos ε  V cos αp  w (5)
stream tube, where the mass of air that takes part in the exchange of
momentum is the mass flux through a projected area normal to V disk Substituting Eqs. (2) and (5) into Eq. (1) leads to the theoretical
and equivalent to Sdisk , regardless of the angle of incidence. In other entrainment factor e being determined as
words, this wider effective stream tube can be defined by flipping the
rotor disk area to a normal position relative to V disk (see Fig. 1). The s













































mass flow rate in the effective stream tube would then be given as V disk sin αp 2 1
e  1  (6)
V cos αp  w cos αp  w∕V2 cos ε
_  ρV disk Sdisk
m (2)
Hence, the effective area and the disk area are related to each other as
Now, stretching the disk area at the original position to reach the new
stream tube boundaries will define a new enhanced or effective disk Sdisk
area Seff that is adopted for the mass flux calculation in Eq. (1). The Seff  (7)
cos ε
following development is based on the assumption that Eqs. (1) and
(2) are equivalent for the definition of the mass flowrate m. _ This With e and the mass flow rate predicted, now we consider the
concept that determines the mass flowrate through a circle, with an momentum balance in the propeller axis direction eT , between the
 RUBIN AND ZHAO 1043

 2
ultimate wake section far downstream in the stream tube and section 1 T
far upstream of the propeller. As the classical momentum theory  V 2 w2  2Vw3 cos αp  w4 (18)
2ρSdisk
disregards the wake rotation effects and viscosity, then thrust can be
written as As thrust can be expressed as T  ρn2 D4 CT and the advance ratio
of the propeller is defined as J  V∕nD, then Eq. (18) can be
T  mV
_ ult cos εult − V cos αp  (8) rewritten as
 2
by applying geometry relationship from Fig. 1b it is possible to see 2n2 D2 CT
that V ult cos εult  V cos αp  wult , which leads to  V 2 w2  2Vw3 cos αp  w4 (19a)
π
T  mw
_ ult (9)  2  2  3  4
2CT w w w
 2 cos αp  (19b)
Substituting Eq. (1) in Eq. (9) allows thrust to be rewritten as πJ2 V V V
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

T  ρV cos αp  wSeff wult (10) According to Eq. (19a), an increase in the propeller frequency n, or in
revolutions per minute, is associated with an increase in w. The same
Considering the streamlines of the stream tube between section 1 and effect can be observed in terms of J from Eq. (19b), where a raise in J,
immediately before the disk (−) and from immediately after the disk or an increase in V relative to n or revolutions per minute, will cause a
() to the ultimate wake, the following two Bernoulli equations are decrease in w∕V and, vice versa, a rise in the ratio of the propeller
obtained: rotation to wind speed V will incur in an increase in w∕V. For the
particular case of static thrust, where V  0, one obtains the classical
1 1 induced velocity for hover condition wjV0 ,
p1  ρV 2  p−disk  ρV −2 (11a) s















2 2 disk
T
wjV0  (20)
1 1 2 2ρSdisk
pult  ρV 2ult  p
disk  ρV disk (11b)
2 2
For AOA  0 deg (i.e., no incidence) and if T > 0, the only physi-
Here, V −disk  V  cally valid solution of the four roots for Eq. (18) is wjAOA0 deg as (see
disk ≜ V disk for continuity through the disk, and
p1  pult  patm . This far wake pressure recovery can only be com- Ref. [4])
pleted in the inviscid actuator disk model, as the wake rotation is s






























 

neglected [7,37,39]. At the disk, there is a jump in pressure Δp 1 2T V
impelled by the propeller. Thus, p − wjAOA0 deg  V − 2
(21)
disk  pdisk  Δp. Manipulating 2 ρSdisk 2
Eqs. (11a) and (11b) and applying the pressure jump definition
leads to
Equation (21) can be rewritten as
1   s













































 
ρV 2ult − V 2   Δp (12) w 1 8CT jAOA0 deg 1
2  1− (22)
V AOA0 deg 2 πJ 2 2
where V ult as illustrated in Fig. 1b is given as
The derived equations from the classical momentum theory for
V 2ult  V  wult cos αp 2  wult sin αp 2 (13) propellers at a given incidence angle allows one to estimate the
theoretical slip-stream angle at the disk αslp and at the ultimate wake
By substituting Eq. (13) into Eq. (12) and considering that the force αslpult . As illustrated in Fig. 1b, it can be shown from Eqs. (13), and
associated with the jump in pressure is the thrust T, then the pressure (15b) that
jump can be rewritten as
0 1
T 1 B 2w∕V sin α p C
Δp   ρV  wult cos αp 2  wult sinαp 2 − V 2 
Seff 2
(14) αslpult  arcsin@q































































A (23)
1  4w∕V cos αp  4w∕V2
Substituting Eq. (10) into Eq. (14) and expanding terms in the rhs
leads to s


























































   2
w w
V ult V 14 cos αp  4 (24)
1 V V
ρV cos αp  wwult  ρw2ult  2wult V cos αp  (15a)
2
From Fig. 1b, (αslp ) at the disk is related to αp and ε as
wult  2w (15b)  
sin αp
αslp  αp − ε  αp − arctan (25)
The thrust T can be expressed by using Eqs. (15b) and (10) as cos αp  w∕V

T  2ρV cos αp  we Sdisk w (16) Further analysis shows

0 1
On the other hand, T can also be rewritten from Eqs. (2), (9), and
(15b) as B w∕V sin αp C
αslp  arcsin@q





























































A (26)
1  2w∕V cos αp  w∕V2
T  2ρV disk Sdisk w (17)

Equation (17) is the well-known equation of thrust from Glauert’s It has been shown that the angle ε influences T through the entrain-
hypothesis [7]. Substituting Eq. (4b) into Eq. (17) leads to a quartic ment factor e. For no incidence flight, ε equals αp at 0 deg. For
relationship in terms of w as αp  90 deg,
 1044 RUBIN AND ZHAO
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

Fig. 2 Variation of e determined from Eq. (6): a) e as function of ε and b) e as function of AOA and w∕V.

 
V V sin αp
ε  arctan (27) V cos αp  w  s



















































   (32)
w 1 1
−1 1
cos ε cos ε
Figure 2a illustrates the variation of the entrainment factor e
determined by using Eq. (6) with ε. The effects of AOA and w∕V After further simplification, we notice that the resulting expression
on e is shown in Fig. 2b. It can be seen that at low angles of incidence has the form of the wing lift formula from wing theory [3,5,7,52].
e tends to be unity. The effective area Seff is similar to the rotor disk Therefore, we define it as the wing equivalent component of thrust
area, as the propeller is operating at wind speeds near the axial T wing as
condition. Seff is increased with e and therefore with AOA. As
AOA approaches 90 deg and V cos αp tends to zero, e tends to r



















1 − cos ε
infinity, in theory at very low w∕V so as to increase the area Seff in T wing  2ρV sin αp wSdisk (33)
order to maintain the same finite value of mass flow rate m _ through 1  cos ε
the rotor, according to Eqs. (1), and (2). In these extreme cases, the
Because it is assumed in the disk actuator theory that the induced
angle ε is approaching 90 deg, as V disk is almost parallel to the rotor
speed w is constant across the rotor disk, T wing can be considered to
disk. The increase in e becomes more relevant at high angles of
be equivalent to the lift produced by an elliptic wing subjected to a
incidence and at very low w∕V values.
wind speed of magnitude V sin αp and having the wing surface area
equal to
B. Entrainment Factor Expansion and Introducing Axial and Wing
Equivalent Components of Thrust r



















1 − cos ε
The entrainment factor e can be expanded as a Maclaurin series as Swing  Sdisk (34)
1  cos ε
1 ε2 5ε4 61ε6 277ε8 50521ε10 The factor multiplying Sdisk and defining the equivalent wing area
e  1      ::: (28)
cosε 2 24 720 8064 3628800 will be regarded as the wing factor WF. It can be expressed also as
function of e as
By
P naming the higher-order terms of the rhs of Eq. (28) as r


















r












termsε, 1 − cos ε e−1
WF   (35)
1  cos ε e1
1 X
e 1 termsε (29) Substituting Eq. (6) on Eq. (35) and simplifying it leads to WF being
cos ε
given as
Applying Eq. (16) and using Eq. (29) to replace the entrainment sinαp
factor leads to the thrust T being determined as WF  q


























































(36)
1  2w∕Vcos αp  w∕V2  cosαp  w∕V
X
T  2ρV cosαp  wSdisk w  2ρV cosαp  wSdisk w termsε
Figure 3a shows the variation of WF with the angle ε. For low ε,
(30) WF implies a very small equivalent wing area, vanishing at ε 
0 deg (where also αp  0 deg and the wing component vanishes).
It can be seen that the first term on the rhs of Eq. (30) has the form of For ε → 90 deg, WF tends to unity. It yields then a full wing
an axial thrust, as the disk area is constant and independent of the equivalent area of Sdisk . It can be seen from Fig. 3b that WF can only
entrainment factor e for any angle of attack. There is only the reach unity for w∕V → 0, or at high speeds, and at AOA  90 deg,
component of the incoming wind velocity in the axial direction, with which corresponds also to ε → 90 deg. For AOA ≠ 0 deg, WF only
its absolute value changing with cos αp . Thus, that term is named the vanishes asymptotically as w∕V → ∞, when hovering.
T axial component of thrust, and it is defined as In the extreme case, at AOA → 90 deg and w∕V → 0, where
WF → 1, thrust can be shown as
T axial  2ρV cos αp  wSdisk w (31) T  T axial  T wing → 2ρSdisk w2  2ρVwSdisk (37)

The
P second term on the rhs of Eq. (30) can be rewritten by expressing Here, it can be seen that T axial takes the form of the static thrust
termsε  1∕ cos ε − 1 from Eq. (29) and by using the following formula and it contributes to total thrust even at forward speeds. T wing
relationship that can be obtained from the geometric graph in Fig. 1b: converges to the wing lift formula with full area Sdisk in the case of a
 RUBIN AND ZHAO 1045
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

Fig. 3 Variation of WF as determined by Eqs. (35) and (36): a) WF as function of ε and b) WF as function of AOA and w∕V.

rotor at fast forward flight (w ≪ V), when T axial is negligible. As

T wing becomes dominant, the propeller behaves as a wing. These two
expressions are mentioned by Glauert [7] in the development of his
thrust hypothesis in the analysis of helicopters in forward flight.
Ariza [22] presents results from 3D numerical simulations showing
induced vortices produced by the rotor when at high angles of attack
in the same manner as finite wing vortices. It is suggested that a
possible explanation for the phenomenon in those conditions is that
the propeller could behave as an elliptic wing. Johnson [11] also
mentions that the helicopter rotor behaves as a wing at forward speeds
and that these two expressions are the limits of thrust when V ≫ w
and when in hover, where V  0. Although he claims that there is no
theoretical justification for the approach at intermediate forward
speeds, good agreement has been found with measured rotor perfor-
mance and with vortex theory results, suggesting therefore that it
should be accepted for the entire range of speeds. McCormick [5] also
mentions the analogy of the propeller at forward flight to an elliptic Fig. 4 Variation of Twing ∕Taxial with w∕V and AOA.
wing when w∕V → 0 and when V  0 to the hover case. The
derivations of Eqs. (31) and (33) in this Paper allow us to assume
that indeed the two components T axial and T wing are always present
and comprise the total thrust for the whole operational envelope, at III. Experimental Wind Tunnel Tests
any angle of incidence and velocity.
A. Experimental Setup and Tests Procedures
From Eqs. (31) and (33), it is possible to calculate the ratio
T wing ∕T axial to show the contribution of the wing component to the The experimental tests were conducted at the University of Canter-
axial component on thrust, according to bury’s closed circuit subsonic wind tunnel that has a rectangular cross
section of 0.9 m in height by 1.20 m in width, providing maximum
r

















test speeds of 60 m∕s. The force balance is a six-axis JR3 45E15A4
1 − cos ε r


















2ρ V sin αp w Sdisk sensor capable of measuring the forces and moments in the three-axis
T wing 1  cos ε 1 − cos ε xyz. It can stand loads of up to 400 N with nominal accuracy of
  tan ε (38)
T axial 2ρV cos αp  wSdisk w 1  cos ε 0.25%. A fourth-order Butterworth analog filter with eight differ-
ent cutoff frequencies ranging from 6.3 to 926 Hz can be selected via
Figure 4 shows the variation of T wing ∕T axial with AOA and w∕V, jumper plugs. The default frequency at the wind tunnel is set to 6.3 Hz
determined by Eqs. (6) and (38). The region where T wing ≥ T axial is for maximum noise attenuation. An aluminum rig with variable AOA
highlighted. It can be seen that the contribution of T wing to the total settings was manufactured to hold a 6 mm carbon fiber square rod and
thrust T wing T axial increases with AOA and with V. As w∕V is the 3D-printed motor/propeller assembly, shown in Fig. 5.
reduced to less than 0.6 and AOA grows higher than 60 deg, even- The propeller used is a two-blade, 6.0 in. diameter, 4.5 in. pitch
tually T wing overcomes T axial and becomes dominant, in other propeller, manufactured by HQProp, powered by an Ethix Team-
Blacksheep Silk V2 2345 V electric motor. An 80A T-motor elec-
words, T wing ∕T axial > 1. tronic speed controller (ESC) is used to control the motor. A power
At high AOA and at high speeds (w∕V → 0), then T wing ∕ supply continuously provides 15.7 V and electric current set accord-
T axial → ∞, so T is composed mainly by T wing. In those cases, T → ing to power requirement to maintain a desired revolutions per
2 ρ V w Sdisk . At V  0 in hover, it becomes T  2 ρ Sdisk w2 . This minute, limited at around 20 A for motor constraints. The input
agrees well with the findings reported by Glauert [7]. By manipulat- electric current for the motor is measured through a current meter
ing Eqs. (35), (36), and (38), it is possible to rewrite T  T axial 1  that communicates with data acquisition hardware (DAQs) from
T wing ∕T axial  as National Instruments to a desktop PC prepared with a LabVIEW

8 9
>
> >
>
>
> sinαp  2 >
>
<1  2s



















































3=
T  T axial       2   (39)
>
> w 4 w w w 5>
>
>
> cosαp  12 cosαp   cosαp  >
>
: V V V V ;
 1046 RUBIN AND ZHAO
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

Fig. 6 Scheme of forces for the propeller wind tunnel tests.

Equations (40a) and (40b) enable the calculation of T and N p from

forces Fx and Fz measured at the wind tunnel tests, according to the
scheme depicted in Fig. 6. The analysis of experimental test data is
performed by using the thrust values acquired for every test condition
and subsequently inputting those into Eq. (18) to calculate the
associated induced speeds w. Once w is obtained, then all the
remaining calculations can be performed.

Fx  cos αp T − sin αp N p (40a)

Fig. 5 Wind tunnel settings: a) the propeller test rig and b) experiment
hardware. Fz  sin αp T  cos αp N p (40b)

program that records electric power, current, voltage, and motor The experimental data were acquired for around 20 s per experimental
revolutions per minute. A Monarch remote optical light-emitting test. Batches of three tests were performed for every tested condition, in
diode (LED) sensor able to measure up to 250,000 rpm from up to other words, RPM, AOA, and V, comprising around 80 points per
0.9 m distance, at a maximum of 45 deg angle, is installed at the wind condition. The experimental measurements were averaged, and the
tunnel test section facing the motor that is prepared with a reflective precision error evaluated as one standard deviation and presented in
tape on top of half of its circumference. An exponential smoothing percentage terms relative to the average for V and RPM. The errors for
filter for revolutions per minute readings was applied in the Lab- the forces Fx , and Fz are presented in N, rather then in percentage
VIEW program. Another LabVIEW program is used for the force values, as in many experimentally tested cases the very small average
values obtained would result in meaningless very high percentage error
measurement and includes a low-pass filter with adjustable cutoff
values. A total of 182 tested conditions were measured. The figure
frequencies. Oscillatory nature of thrust, moments, and lateral forces
shown in Appendix C, presents the measurements errors obtained for
from propellers at incidence are mitigated by the use of the filter. A
all tested conditions. The maximum error for V was around 5%,
1 Hz cutoff frequency has been chosen throughout the tests in order to
although the vast majority of the tests showed errors below 2.0%.
achieve more stable average readings. Experimental test data are
The maximum errors observed for RPM, Fx , and Fz were 0.8%, 0.14 N
compiled and analyzed with MATLAB®.
and 0.65 N, respectively. The errors in T were calculated based on the
Propeller thrust and normal forces reading were acquired at wind
F , and Fz errors, according to error propagation rules, σ T 
qx

























































tunnel speeds ranging from zero to 25 m∕s and motor speed rotations
from 9000 to 18,000 rpm. Angles of attack ranged from 0 to 90 deg. cos αp σ Fx 2  sin αp σ Fz 2 . The cumulative distribution function
Electric input motor power is also measured and analyzed for each for all σ T points is presented in the subgraph f in Appendix C, where it
test condition. For a wind speed of 25 m∕s and AOA higher than is shown that for 85% of the cases σ T < 0.1 N and that the measure-
60 deg, it was not possible to reach revolutions per minute higher than ment errors tend to a log-normal distribution behavior as verified by a
15,000 as power requirement exceeded motor limitations. corresponding data fit.

B. Experimental Data Postprocessing

Before the experimental tests, the holding set without the propeller IV. Results and Discussions
was exposed to the same wind speeds and AOA values as of the A. Thrust Measurements Analysis
propeller tests in order to evaluate the resistance of the rig. The forces Figure 7a shows thrust performance of the propeller as a function
readings acquired were fitted to spline surface functions. These are of RPM at different wind speeds, at no incidence (AOA  0 deg). As
netted from the final propeller tests readings. The extra effect of the expected from the momentum theory, the freestream velocity in-
propeller slip stream on the holding sting was not evaluated, and the crease at constant RPM (or a raise in J) will cause a reduction in
total net thrust could be expected to be somewhat higher. However, the available thrust [5]. It is also noticeable from propeller tests results
the slip stream is expected to impinge only on the sting behind the presented in previous works [19–23,53,54], and the thrust exhibits a
propeller and not on the entire holding set. parabolic variation with RPM as in Refs. [19–23,53–56]. Figure 7b
 RUBIN AND ZHAO 1047
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

Fig. 7 Variation of thrust T with RPM: a) as V is set to five different values and AOA
0 deg and b) as V
20 m∕s and AOA is set to four different
values.

Fig. 8 Thrust T varied with AOA and RPM: a) V
10 m∕s and b) V
20 m∕s.

illustrates the influence of AOA on thrust for a constant wind speed

V  20 m∕s. The thrust T is increased, as the angle of incidence
rises. This finding is consistent with the previous concluding remarks
in Refs. [19–23].
Figure 8 presents the thrust measured from the propeller as func-
tion of AOA and RPM, for a) V  10 m∕s and b) V  20 m∕s.
Again, from both graphs at AOA  0 deg, it can be seen that the
thrust is reduced with the increase of V. For instance, at 18,000 rpm,
in Fig. 8a, T  6 N, whereas, in Fig. 8b, T  4 N, approximately.
As AOA rises, T increases in both cases. However, the slope of thrust
increase with AOA ∂T∕∂αp is higher in the case of the higher speed.
Another interesting observation from Fig. 8 is that at low RPM and
low AOA, no thrust is available, which is also seen in Fig. 7. In that
region, the propeller will perform in a windmill state until a sufficient
value of RPM is achieved. For example, in Fig. 8a in which
V  10 m∕s, no thrust is generated before the propeller reaches
around 5000 rpm at low angles of attack. For the case of
V  20 m∕s in Fig. 8b, the value required is 9500 rpm at no inci-
dence, and the region of no thrust achievable (T ≤ 0) is wider,
reaching up to AOA ≈ 60 deg. However, as AOA is increased, less
RPM is required to achieve some thrust in both cases. Fig. 9 Variation of T with AOA at different RPM, as V is set to three
Figure 9 shows how thrust T is varied with AOA, for different values
V  10; V  15, and V  20 m∕s and rotation speeds of 12,000,
15,000, and 18,000 rpm. In all tested cases, the thrust T is found to and the inversion of behavior at a higher AOA are found in the
always increase with AOA, more noticeably for AOA > 20 deg. A experimental tests in Refs. [21–23].
higher slope ∂T∕∂αp is associated with a higher V, for a given RPM. Figure 10a shows the thrust T contour surface at 15,000 rpm. The
Therefore, the analysis here means to be extended also to the advance analysis here, at constant RPM, again means that an increase or
ratio J, as an increase in velocity, at a given constant RPM, represents decrease in V is also associated with a corresponding increase or
also an increase in J, and vice versa for a decrease. This difference in decrease in J. Again, at a low AOA value, T is decreased as V is
slopes will cause a change in thrust sensitivity to V at high angles of growing. However, at a high AOA value, beyond 60 deg, the thrust
incidence; while at low angles T decreases with V (∂T∕∂V < 0), for variation behavior inverts to increase with V. Also, T is always
AOA around 60 deg and over, T eventually starts to increase with increasing with AOA at constant speeds. T axial is presented in
increased V (∂T∕∂V>0). Similar results for CT growing with AOA Fig. 10b, and T wing is depicted in Fig. 10c, calculated according to
 1048 RUBIN AND ZHAO
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

Fig. 10 Variation of the thrust T, Taxial , and Twing with AOA and V: a) T measured at 15,000 rpm, b) calculated Taxial , and c) calculated Twing .

Eqs. (17), (31), and (33). T axial behaves in the same way for all angles (static thrust tests performed at different AOA) is probably associated
of incidence (AOA), decreasing as V is increased. T wing is rising with with ground effects from the wind tunnel. The expected behavior
V and AOA, peaking at AOA  90 deg and at higher speeds. The would be the same static thrust at all incidence angles.
slope of T wing increase with AOA, (∂T wing ∕∂αp ) also grows with V. Figure 11 shows the measured T and its components T axial and
Note in Figs. 10a and 10b the slight increase in T with AOA for V  0 T wing calculated for the cases of fixed velocities V  10 m∕s,

Fig. 11 T measured, Taxial , and Twing calculated according to Eqs. (31) and (33).
 RUBIN AND ZHAO 1049

V  15 m∕s, V  20 m∕s and RPM values of 9000 and 15,000. The

increase of thrust with AOA (∂T∕∂αp > 0) follows to a great extent
the behavior of T wing (∂T wing ∕∂αp > 0) from which it can be inferred
the wing component to be mostly the reason for the phenomenon.
Also, it can be observed from Figs. 11b, 11d, 11f and 11a, 11c, 11e
that as V (and J) is increased the sensitivity slopes of T wing to AOA
are also increased. T wing contribution below AOA ≈ 30 deg is
negligible even at high advance ratios J as the wing factor WF and
sin αp are low at small angles. At AOA  0 deg, T wing vanishes, and
so T  T axial , which formula reverts to the classical momentum
theory thrust equation at no incidence. T wing relevance starts at
AOA > 30 deg, and a high J value causes the contribution of
T wing relative to T axial to become very important in the region around
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

AOA ≈ 60 deg, where T wing eventually surpasses T axial as illustrated

in Figs. 11a–11f. This can only be observed for w∕V lower than
around 0.6 (see Fig. 4), which is associated with a higher J [see
Eq. (19b)]. At very high angles and as J is increased, T wing composes
an ever-larger part of T. Note also that as J is increased (decreased
w∕V) the crossing of the two components is possible at a lower angle
than 90 deg, toward 60 deg (Figs. 11a, 11c, 11e and 11b, 11d, 11f). Fig. 12 The w∕V obtained from T data at AOA compared to w∕V
This is in accordance with the theoretical prediction behaviors as estimated from TjAOA
0 deg and extrapolated to other AOA values
shown in Fig. 4. (solid lines).
As T axial is predominant at angles AOA < 60 deg, T is decreased
with increasing speed (and J), and accordingly (∂T∕∂V < 0). For
higher angles, T is growing with V (∂T∕∂V > 0) (see Fig. 11a, 11c, AOA  0 deg. This assumption comes from the fact that T axial is
11e at 9000 rpm and Figs. 11b, 11d, 11f at 15,000 rpm). The inversion relatively insensitive to AOA as seen in Fig. 11 unless at high speeds.
from negative to positive sensitivity (see the figure in Appendix B) Figure 12 illustrates w∕V calculated from measured T data for all
could be explained as the influence of T wing on T (∂T wing ∕∂V > 0.) angles against the approximation model of w∕V in solid lines. It is
being predominant over T axial (∂T axial ∕∂V < 0) in that region. In seen that there is a good match of the simplified model with the
Ref. [21], several experimental tests were performed for CT vs J and measurement data down to w∕V  0.2 or J up to 0.53. For
AOA, ranging from 0 to 85 deg on three different full-scale propellers J > 0.53, there is a detachment past AOA  30 deg that grows with
at many different blade pitch angles, where the results obtained showed J. At J  0.87 and AOA  0 deg, the propeller is windmilling and
inversions of slope ∂CT ∕∂J ocurring at angles of incidence higher TjAOA0 deg  0 as in Fig. 11e, so the model is not suitable as it
than 60 deg, for all blade angle configurations. It is interesting to note predicts w  0 for all angles.
that this inversion is in the region of high AOA, where the theory Figure 13 presents the results obtained from actual thrust mea-
predicts that T wing and its effects begin to overcome T axial if under a surements against predicted T through the simplified model accord-
large enough velocity V (high J) (see Fig. 4). A possible explanation ing to Eq. (39), using the values of T at AOA  0 deg. Also, w∕V
for this would be that different blade configurations would have differ- projected to be used in Eq. (39) is obtained using T at AOA  0 deg
ent slope sensitivities of T to V∂T∕∂V as different propellers should in either Eq. (18) or Eq. (19) as illustrated in Fig. 12. As the simplified
have different T surfaces vs w∕V and AOA, and so ∂T wing ∕∂V > model relies on a projection of thrust at AOA  0 deg and as T axial is
∂T axial ∕∂V would happen at different angles for different propellers. relatively constant up to AOA around 60 deg, at low advance ratios
However, we assume that for any propeller the relation between T wing (see Fig. 11 for J < 0.53), good agreement is observed between the
and T axial must follow Eq. (38), as illustrated in Fig. 4, to fulfil the model and the actual thrust (see Figs. 13a, 13b, and 13d). In those
momentum theory assumptions. tested cases, w∕V projected from the simplified model is very close to
T axial is relatively insensitive to AOA at low advance ratios (up to w∕V obtained from experimental data (see Fig. 12). At around
J  0.53 for the propeller tested) and up to angles around 60 deg (see AOA > 70 deg, the model overestimates the practical test data. This
Figs. 11a, 11b, 11d, and 11f). A decrease in T axial at higher AOA is should be expected as the model assumes a T axial constant for all
noticed in all cases, being more accentuated as J is growing (w∕V incidence angles, but a decrease in that component at a higher AOA
diminishing), where also T wing growth is more intense. At higher can be observed from Figs. 11a, 11b, and 11d.
advance ratios, T axial is increased with AOA to peak at around 60 deg A small detachment of w∕V projected by the model, from mea-
for the propeller studied (see Figs. 11c and 11f). As T wing tends to be sured w∕V is seen at J  0.53 in Fig. 12, which will cause the start of
the main component of T at high J and AOA, the thrust formula is the detachment of T estimated by the model from the real T measured,
indeed reduced to the second term of Eq. (37) or T → 2ρVwSdisk as as in Fig. 13f. As J grows, the detachment of T projected by the model
predicted by Glauert [7] and described in Ref. [11] for helicopterro- increases as in Fig. 13c until the model eventually loses validity as in
tors at high translational speeds (high J), when the propeller behaves Fig. 13e. This happens when T jAOA0 deg measured at no incidence is
as a wing. vanished (windmill/brake state), alongside w∕V, which is used by the
model to project T at incidence. The growing detachments of the
B. Simplified Formula for Estimating T at Incidence from Data model w∕V, from w∕V calculated based on the real experimental
at AOA
0 Deg data, occur in consonance with T axial no longer being relatively
The thrust measurements obtained previously from the wind tunnel constant and similar to T jAOA0 deg at angles up to around 60 deg,
experimental tests at any AOA allow us to calculate w∕V from Eq. (18) for increasing J values. In these cases, the T axial growth with AOA is
and estimate T axial and T wing at any point. However, it is not possible to not captured by the simplified model that underestimates T up to
measure these two components separately except in the extreme cases those angles.
of hover and at high V (high J, low w∕V), at AOA → 90 deg, when
T axial is negligible and T wing becomes the main component of T, i.e. C. Slip-Stream Parameters
T → T wing as per Eq. (37), which is in accordance with Glauert [7]. To Table 1 summarizes the calculation of the relevant angles and
prove the validity of the theory, we assume the values of thrust measured velocities, as depicted in Fig. 1, from the experimental data. At given
at AOA  0 deg and calculate the value of w∕V from either Eq. (18) or AOA, RPM, and V conditions and from the thrust T measured,
Eq. (19), extrapolating it to any AOA, but using T obtained at calculations are performed according to Eqs. (4b), (6), (15b), (18),
 1050 RUBIN AND ZHAO
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

Fig. 13 T measured vs simplified model projected from TjAOA
0 deg on Eqs. (18) and (39).

Table 1 T measured, Twing , and Taxial calculated from the theory and slip-stream parameters
AOA, deg RPM V, m∕s J w∕V T, N T axial; N T wing , N e ε, deg αslp , deg αslp ult , deg V disk , m∕s V ult , m∕s
30 17,864 10.4 0.23 0.832 7.001 6.715 0.285 1.0 16.4 13.6 18.8 18.4 26.8
30 14,989 15.5 0.41 0.301 4.050 3.723 0.327 1.1 23.2 6.8 11.2 19.7 24.0
30 9024 20.1 0.87 0.006 0.103 0.089 0.014 1.2 29.8 0.2 0.3 20.2 20.3
60 17,802 10.5 0.23 0.938 7.668 6.569 1.100 1.2 31.1 28.9 40.0 17.6 26.6
60 15,002 15.7 0.41 0.386 5.204 3.721 1.483 1.4 44.4 15.6 25.7 19.5 24.2
60 9074 19.9 0.86 0.081 1.480 0.824 0.655 1.8 56.1 3.9 7.4 20.8 21.7
90 17,865 10.3 0.23 1.155 8.289 6.267 2.022 1.3 40.9 49.1 66.6 15.8 26.0
90 14,861 15.6 0.41 0.518 6.296 2.896 3.400 2.2 62.6 27.4 46.0 17.6 22.5
90 8979 19.9 0.87 0.160 2.828 0.446 2.382 6.3 80.9 9.1 17.7 20.2 20.9

(23), (24), (26), (31), and (33). T wing and T axial are also presented at αslp-ult is always greater than αslp at the disk as wult is higher than w
different flow conditions. [see Fig. 1 and Eq. (15b), wult  2w]. At high AOA, the contribution
For AOA  30 deg, the theoretical entrainment factor e is close to of T wing becomes more relevant, and for w∕V values lower than 0.6
unity, as expected from the momentum theory. For increasing V values, and angles higher than 60 deg, T wing eventually surpasses T axial as in
αslp gets smaller as it is harder for the propeller to turn the flow, the last two cases where AOA  90 deg and w∕V  0.518 and 0.16.
especially at low RPM (high J, low w∕V), which is in agreement with Note the entrainment factor e is also high (e  2.2 and e  6.3)
Ref. [22], and if AOA is high, so is e whenever αslp is small. The angle accompanied by a high angle ε in these instances.

Fig. 14 Angles variation with J and AOA: a) slip-stream angle at the disk αslp and b) angle ε, between V disk and thrust T.
 RUBIN AND ZHAO 1051

Figure 14a depicts the behavior of αslp as a function of J and AOA behavior is interpreted as being the consequence of T wing impact
for the propeller being tested in this Paper. It is seen that for static tests overcoming the contribution of T axial effects. Different propellers and
J  0 αslp  AOA. At no incidence AOA  0 deg, αslp also blade configurations should present different surfaces of T vs
vanishes as the flow is axial. As J is increased, αslp tends to diminish, w∕V and AOA. Therefore, different slope sensitivities ∂T wing ∕∂V,
and for a constant J, αslp grows with increased AOA. Figure 14b ∂T axial ∕∂V and [∂∕∂V∂T wing ∕∂αp  should be expected, for a given
shows that the angle ε is varied with AOA and with J. For static test RPM. It is believed that this could explain the inversion of ∂T∕∂V at
J  0; ε  0 deg as w is aligned with T. At AOA  90 deg, ε different angles of incidence for different propellers seen in other
tends asymptotically to 90 deg with the growth of J (which implies studies. However, the ratio T wing ∕T axial must follow the theory, and
growth in V∕w). The analysis of ε is of interest as it defines the the inversion of thrust behavior should happen around the region where
entrainment factor e and the wing factor WF and therefore is directly T wing becomes relevant and ∂T wing ∕∂V overcomes ∂T axial ∕∂V,
related to T wing , as indicated in Eqs. (33) and (35) and shown in Fig. 3. which happens at high angles and at high speeds (w∕V < 0.6). This
The maximum ε is associated with the highest T wing , which occurs at verification could be done in further work.
high J (low w∕V values) and high angles of incidence (see Fig. 10c). At AOA → 90 deg, T axial converges to the static thrust formula,
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

being always present in the thrust composition even in forward flight.

At hovering condition, when T wing vanishes, then T axial becomes the
V. Conclusions sole contributor of T, in which case the theory is proven to be
consistent with the traditional formula T  2ρSdisk w2 . The proposed
In this Paper, a series of experimental tests was conducted to theory is also shown to agree completely with the classic momentum
investigate the aerodynamic performance of a two-blade propeller theory at AOA  0 deg. To a great extent, the increase of T with
operating at angles of incidence (AOA) ranging from 0 to 90 deg. AOA is due to the wing component. T wing peaks to compose most of
This is done in in a closed-loop wind tunnel at University of Canter-
the thrust at very high speed and high AOA. In this case, the proposed
bury. An alternative approach on analyzing aerodynamic thrust per-
theory is shown to be consistent with Glauert’s hypothesis, reverting
formance is proposed. It is based on the classical momentum theory
to the wing lift formula used for helicopter rotors in fast forward
by defining a theoretical entrainment factor e accounting for the mass
flight, T → 2ρ Vw Sdisk .
flow rate through an enhanced area of the rotor disk to equalize
The development presents an alternative simplified formula for
Glauert’s hypothesis mass flow definition. The factor e is found to
estimating T at any AOA up to 90 deg for any propeller, based
depend on the angle ε, defined as the angle between V disk and the
on data acquired at no incidence. (T or CT and V, RPM, or J).
thrust vector T. It is shown mathematically that the thrust T consists
The formula showed good agreement with the experimental results
of two components: one is the axial component T axial , and the other is
for the propeller tested up to intermediate advance ratios and
the wing lift equivalent component T wing .
AOA ≈ 80 deg.
T axial behaves similarly to a propeller operating at no incidence A concluding remark is that one could assume the thrust of a
under an axial stream of magnitude V cos αp . Therefore, for a given propeller at incidence to be interpreted as the thrust of a propeller
RPM, it is decreased with increasing airspeed (∂T axial ∕∂V < 0), for all in axial flow condition under incoming speed of V cos αp with a
angles of incidence. T axial is shown to be not so sensitive to AOA at thrust addition equivalent to the lift produced by an elliptic wing of
low airspeeds (low J), especially at AOA < 60 deg, while at high area Sdisk WF, under incoming speed V sin αp , where both compo-
airspeeds (high J), it grows with the angle of incidence up to nents share a common induced speed w.
AOA ≈ 60 deg. At higher angles, T axial is decreased with increasing
AOA. This decrease is more intense at higher speeds.
T wing provides the equivalent lift of an elliptic wing under velocity Appendix A: General Thrust Formula for Propellers
V sin αp of magnitude with a variable area equal to Sdisk multiplied at Incidence Based on Thrust at AOA
0 Deg
by a factor WF. This factor ranges from zero at AOA  0 deg to A final equation for thrust prediction based on propellers data at
WF → 1 as AOA approaches 90 deg at high airspeed V. T wing was AOA  0 deg is presented based on the assumption of T axial rela-
found to rise with AOA and V. Also, the slope ∂T wing ∕∂αp  is tively constant with AOA up to around 60 deg for intermediate
increased with V and ∂T wing ∕∂V is increased with AOA, for a J values. The simplified equation is an approximation based on
given RPM. Eq. (39),

8 9
>
> >
>
>
> sin αp  2 >
>
T <1  2s
























































3 =
       2   (A1)
T jA0 deg >> w 4 w w w 5>>
>
> cos αp  12 cos αp   cos αp  >
>
: V V V V ;

The theory shows that T wing surpasses T axial at AOA ≈ 60 deg or where the value of w∕V is obtained from solving Eq. (18) or Eq. (19),
above and at high speeds (w∕V < 0.6). Also, as the theoretical ratio while using thrust data acquired at AOA  0 deg.
T wing ∕T axial stems from the momentum theory development, which
disregards blade geometry and propeller design, the authors assume
that it would be valid for any propeller. The influence of T wing on T
overcomes that of T axial at high angles and speed as in those con- Appendix B: Measured Different Thrust Behaviors
ditions the positive sensitivity ∂T wing ∕∂V is larger than the negative for Tested AOA Range
sensitivity value of ∂T axial ∕∂V. Constant AOA surfaces are depicted in Fig. B1, which shows
Thrust is found to be decreased with V (and J) (∂T∕∂V < 0) at low T as function of V and RPM. It can be seen that below
angles, as T axial is dominant, whereas at around AOA ≈ 60 deg or AOA ≈ 60 deg, T decreases with V, or (∂T∕∂V < 0), for any
higher and at high airspeeds (and J), T changes its behavior to given RPM, while for AOA > 60 deg, T grows with V, or
increase with increasing V, in other words, (∂T∕∂V > 0). This (∂T∕∂V > 0). This can be interpreted as the contribution of
 1052 RUBIN AND ZHAO

Acknowledgments
The authors would like to express appreciation to the technical
staff who helped make the experimental tests achievable, Julian
Murphy, Julian Philips, Kenneth Brown, David Fanner, Natalia
Kabaliuk, and Tony Doyle, and to Emeric Jago, an undergraduate
engineering student. The first author would like to thank Dan Zhao
for supervising his Ph.D. research and encouraging this work. The
authors would also like to thank the anonymous reviewers for their
insightful and detailed comments.

References
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

[1] Rankine, W. M. J., “On the Mechanical Principles of the Action of

Propellers,” Transactions of the Institute of Naval Architects, Vol. 6,
April 1865, pp. 13–39.
[2] Froude, R. E., “On the Part Played in Propulsion by Differences of Fluid
Pressure,” Transactions of the Institute of Naval Architects, Vol. 30,
1889, pp. 390–405.
Fig. B1 Variation of the thrust T with V and RPM, as AOA is set to three [3] Shapiro, J., Principles of Helicopter Engineering, McGraw–Hill, New
different values. York, 1956, pp. 1–23.
[4] McCormick, B., Aerodynamics, Aeronautics & Flight Mechanics,
Wiley, New York, 1979.
[5] McCormick, B. W., Jr., Aerodynamics of V/STOL Flight, Academic
T wing influence overcoming that of T axial , in other words, Press, New York, London, 1967.
∂T wing ∕∂V > ∂T axial ∕∂V at high angles, and vice versa at low [6] Glauert, H., The Elements of Airfoil & Airscrew Theory, Cambridge
angles. Univ. Press, Cambridge, England, U.K., 1948, pp. 199–207.

Appendix C: Experimental Measurement Error Analysis

Fig. C1 Measurement precision errors: a) in V, b) in RPM, c) in Fx , d) in Fz , e) inferred for T from Fx and Fz , and f) CDF, cumulative distribution
function of the errors in T.
 RUBIN AND ZHAO 1053

[7] Glauert, H., “Airplane Propellers,” Aerodynamic Theory—Vol. IV, 4574.

edited by W. F. Durand, Springer-Verlag, Berlin, 1935, pp. 169–359. https://doi.org/10.3390/en12234574
[8] Dommasch, D. O., Elements of Propeller and Helicopter Aerodynam- [32] Glauert, H., On the Horizontal Flight of a Helicopter, Her Majesty’s
ics, Pitman, London, 1953, pp. 21–126. Stationery Office, 1928.
[9] Betz, A., “Eine Erweiterung Der Schraubenstrahl Theorie,” Zeitschrift [33] Bramwell, A. R. S., “Some Remarks on the Induced Velocity Field of a
für Flugtechnik und Motorluftschiffahrt, Vol. 11, 1920, pp. 105–110. Lifting Rotor and on Glauert’s Formula,” Aeronautical Research Coun-
[10] Drzewiecki, S., Théorie Générale de l’hélice: Hélices Aériennes et cil Current Papers—C.P. No. 1301, Her Majesty’s Stationery Office,
Hélices Marines, Gauthier-Villars, Paris, 1920. 1974.
[11] Johnson, W., Helicopter Theory, Princeton Univ. Press, Princeton, NJ, [34] Van Kuik, G. A. M., “On the Revision of the Actuator Disc Momentum
1980, pp. 28–129. Theory,” Wind Engineering, Vol. 5, Jan. 1991, pp. 276–289.
[12] Goldstein, S., John, S., and Wilhelm, K., “On the Vortex Theory of [35] Van Kuik, G. A. M., “The Edge Singularity of an Actuator Disc with a
Screw Propellers,” Mathematical and Physical Character, Vol. 123, Constant Normal Load,” Proceedings of ASME Wind Energy Sympo-
No. 792, 1929, pp. 440–465. sium, ASME Paper WIND2003-356, New York, Jan. 2003, pp. 62–72.
[13] Theodorsen, T., Theory of Propellers, McGraw–Hill, New York, 1948. https://doi.org/10.1115/wind2003-356
[14] Van Kuik, G. A. M., Sørensen, J. N., and Okulov, V. L., “Rotor Theories [36] Goorjian, P. M., “An Invalid Equation in the General Momentum
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734

by Professor Joukowsky: Momentum Theories,” Progress in Aerospace Theory of Actuator Disk,” AIAA Journal, Vol. 10, No. 4, 1972,
Sciences, Vol. 73, Feb. 2015, pp. 1–18. pp. 543–544.
https://doi.org/10.1016/j.paerosci.2014.10.001 https://doi.org/10.2514/3.50146
[15] Okulov, V. L., Sørensen, J. N., and Wood, D. H., “The Rotor Theories by [37] Sorensen, J. N., and Mikkelsen, R., “On the Validity of the Blade
Professor Joukowsky: Vortex Theories,” Progress in Aerospace Scien- Element Momentum Method,” EWEC-CONFERENCE, edited by P.
ces, Vol. 73, Feb. 2015, pp. 19–46. Helm, and A. Zervos, WIP—Renewable Energies, Munich, 2001,
https://doi.org/10.1016/j.paerosci.2014.10.002 pp. 362–366.
[16] Ribner, H. S., “Propellers in Yaw,” NACA Rept. 820, 1943. [38] Van Kuik, G. A. M., and Lignarolo, L. E. M., “Potential Flow Solutions
[17] Ribner, H. S., “Formulas for Propellers in Yaw and Charts of the Side- for Energy Extracting Actuator Disc Flows,” Wind Energy, Vol. 19,
Force Derivative,” NACA Rept. 819, 1943. No. 8, 2016, pp. 1391–1406.
[18] De Young, J., “Propeller at High Incidence,” Journal of Aircraft, Vol. 2, https://doi.org/10.1002/we.1902
No. 3, 1965, pp. 241–250. [39] Bontempo, R., and Manna, M., “Verification of the Axial Momentum
https://doi.org/10.2514/3.43646 Theory for Propellers with a Uniform Load Distribution,” International
[19] McLemore, H. C., and Cannon, M. D., “Aerodynamic Investigation of a Journal of Turbomachinery, Propulsion and Power, Vol. 4, No. 2, 2019,
Four-Blade Propeller Operating Through an Angle-of-Attack Range Paper 8.
From 0° to 180°,” NACA TN-3228, 1954. https://doi.org/10.3390/ijtpp4020008
[20] Kuhn, R. E., and Draper, J. W., “Investigation of the Aerodynamic [40] Sorensen, J. N., General Momentum Theory for Horizontal Axis Wind
Characteristics of a Model Wing-Propeller Combination and of the Turbines, Vol. 4, Springer International Publ., Switzerland, 2016.
Wing and Propeller Separately at Angles of Attack up to 90°,” NACA https://doi.org/10.1007/978-3-319-22114-4
Rept. 1263, 1956. [41] Bontempo, R., and Manna, M., “Analysis and Evaluation of the Momen-
[21] Yaggy, P. F., and Rogallo, V. L., “A Wind-Tunnel Investigation of Three tum Theory Errors as Applied to Propellers,” AIAA Journal, Vol. 54,
Propellers Through an Angle-of-Attack Range from 0° to 85°,” NACA No. 12, 2016, pp. 3840–3848.
TN D 318, 1960. https://doi.org/10.2514/1.J055131
[22] Ariza, D. G., “Study of the Sensitivity to the Lateral Wind of a Mini [42] Bontempo, R., and Manna, M., “Effects of the Approximations Embod-
Unmanned Aerial Vehicle with VTOL Flight Capabilities,” Ph.D. Dis- ied in the Momentum Theory as Applied to the NREL PHASE VI Wind
sertation, Inst. Supérieur de l’Aéronautique et de l’Espace, Toulouse, Turbine,” International Journal of Turbomachinery, Propulsion and
France, 2013. Power, Vol. 2, No. 2, 2017, Paper 9.
[23] Serrano, D., Ren, M., Qureshi, A. J., and Ghaemi, S., “Effect of https://doi.org/10.3390/ijtpp2020009
Disk Angle-of-Attack on Aerodynamic Performance of Small Propel- [43] Conway, J. T., “Analytical Solutions for the Actuator Disk with Variable
lers,” Aerospace Science and Technology, Vol. 92, Sept. 2019, pp. 901– Radial Distribution of Load,” Journal of Fluid Mechanics, Vol. 297,
914. Aug. 1995, pp. 327–355.
https://doi.org/10.1017/S0022112095003120
https://doi.org/10.1016/j.ast.2019.07.022
[44] Conway, J. T., “Exact Actuator Disk Solutions for Non-Uniform Heavy
[24] Øye, S., “Induced Velocities for Rotors in Yaw,” Proceedings of the
Loading and Slipstream Contraction,” Journal of Fluid Mechanics,
Sixth IEA Symposium, Energy Research Centre of the Netherlands,
Vol. 365, June 1998, pp. 235–267.
Petten, 1992, pp. 1–5.
https://doi.org/10.1017/S0022112098001372
[25] Haans, W., “Wind Turbine Aerodynamics in Yaw: Unravelling the
[45] Pitt, D. M., and Peters, D. A., “Theoretical Prediction of Dynamic-
Measured Rotor Wake,” Ph.D. Dissertation, Delft Univ. of Technology,
Inflow Derivatives,” Vertica, Vol. 5, No. 1, March 1981, pp. 21–34.
Delft, The Netherlands, 2011.
[46] Morillo, J. A., and Peters, D. A., “Velocity Field Above a Rotor Disk by a
[26] Micallef, D., Sant, T., Aissaoui, A. G., and Tahour, A., “A Review of
New Dynamic Inflow Model,” Journal of Aircraft, Vol. 39, No. 5, 2002,
Wind Turbine Yaw Aerodynamics,” Wind Turbines-Design, Control
pp. 731–738.
and Applications, IntechOpen, London, July 2016, Chap. 2.
https://doi.org/10.2514/2.3019
https://doi.org/10.5772/63445 [47] Rosen, A., and Gur, O., “Novel Approach to Axisymmetric Actuator
[27] Rahimi, H., Hartvelt, M., Peinke, J., and Schepers, J. G., “Investigation Disk Modeling,” AIAA Journal, Vol. 46, No. 11, 2008, pp. 2914–
of the Current Yaw Engineering Models for Simulation of Wind Tur- 2925.
bines in BEM and Comparison with CFD and Experiment,” Journal of https://doi.org/10.2514/1.37383
Physics: Conference Series, Vol. 753, Sept. 2016, Paper 022016. [48] Kominer, S., and Rosen, A., “New Actuator-Disk Model for a Skewed
https://doi.org/10.1088/1742-6596/753/2/022016 Flow,” AIAA Journal, Vol. 51, No. 6, 2013, pp. 1382–1393.
[28] Schepers, J., “An Engineering Model for Yawed Conditions, Developed https://doi.org/10.2514/1.J052057
on Basis of Wind Tunnel Measurements,” 37th Aerospace Sciences [49] Patterson, M. D., “Conceptual Design of High-Lift Propeller Systems
Meeting and Exhibit, AIAA Paper 1998-0039, 1998. for Small Electric Aircraft,” Ph.D. Dissertation, Georgia Inst. of Tech-
https://doi.org/10.2514/6.1999-39 nology, Atlanta, GA, 2016.
[29] Schepers, J. G., “Engineering Models in Wind Energy Aerodynamics: [50] Veldhuis, L. L. M., “Propeller Wing Aerodynamic Interference,”
Development, Implementation and Analysis Using Dedicated Aerody- Ph.D. Dissertation, Delft Univ. of Technology, Delft, The Netherlands,
namic Measurements,” Ph.D. Dissertation, Delft Univ. of Technology, 2005.
Delft, The Netherlands, 2012. [51] Van Holten, T., “Concentrator Systems for Wind Energy, with Emphasis
[30] Rahimi, H., Garcia, A. M., Stoevesandt, B., Peinke, J., and Schepers, G., on Tipvanes,” Wind Engineering, Vol. 5, No. 1, 1981, pp. 29–45.
“An Engineering Model for Wind Turbines Under Yawed Conditions [52] Dommasch, D. O., Sherby, S. S., and Connoly, T. F., Airplane Aerody-
Derived from High Fidelity Models,” Wind Energy, Vol. 21, No. 8, 2018, namics, Pitman Aeronautical, New York, 1951, pp. 112–145.
pp. 618–633. [53] Deters, R. W., Ananda, G. K., and Selig, M. S., “Reynolds Number
https://doi.org/10.1002/we.2182 Effects on the Performance of Small-Scale Propellers,” 32nd AIAA
[31] Lin, M., and Porté-Agel, F., “Large-Eddy Simulation of Yawed Applied Aerodynamics Conference, AIAA Paper 2014-2151, June 2014,
Wind-Turbine Wakes: Comparisons with Wind Tunnel Measurements pp. 1–43.
and Analytical Wake Models,” Energies, Vol. 12, No. 23, 2019, Paper https://doi.org/10.2514/6.2014-2151
 1054 RUBIN AND ZHAO

[54] Ananda, G. K., Selig, M. S., and Deters, R. W., “Experiments of [56] Brandt, J. B., Deters, R. W., Ananda, G. K., and Selig, M. S., UIUC
Propeller-Induced Flow Effects on a Low-Reynolds-Number Wing,” Propeller Database, Univ. of Illinois at Urbana–Champaign, http://m-
AIAA Journal, Vol. 56, No. 8, 2018, pp. 3279–3294. selig.ae.illinois.edu/props/propDB.html [retrieved 22 April 2020].
https://doi.org/10.2514/1.J056667
[55] Hrishikeshavan, V., and Chopra, I., “Performance, Flight Testing of a F. N. Coton
Shrouded Rotor Micro Air Vehicle in Edgewise Gusts,” Journal of Associate Editor
Aircraft, Vol. 49, No. 1, 2012, pp. 193–205.
https://doi.org/10.2514/1.C031477
Downloaded by KOREA ADVANCED INST OF SCIENCE & TECHNOLOGY (KAIST) on May 2, 2021 | http://arc.aiaa.org | DOI: 10.2514/1.J059734