You are on page 1of 23

Measurements Comparing Hover Performance of Single,

Coaxial, Tandem, and Tilt-Rotor Configurations

Manikandan Ramasamy∗
University Affiliated Research Center
NASA Ames Research Center
Moffett Field, CA 94035

Abstract CT rotor thrust coefficient, = T /ρAΩ2 R2


CP rotor power coefficient, = P/ρAΩ3 R3
Hover performance characteristics of torque-balanced CP0 profile power coefficient, = σCd0 /8
coaxial, tandem, and tilt-rotor systems were compared d shaft-to-shaft distance
against single-rotor systems with equivalent solidity. To- D rotor diameter
wards this, a series of experiments were conducted on FM figure of merit
two identical, but independent, rotor test rigs that were L lift per unit span
configured to easily change from one system to another Nb number of blades
(i.e., coaxial to tandem etc.). The influence of various pa- P rotor power
rameters, such as the axial separation distance for coax- Q rotor torque
ial rotors or the shaft-to-shaft distance for tandem ro- R radius of the blade
tors, on the overall system performance was measured T rotor thrust
using untwisted (NACA 0012 airfoil) and highly-twisted vl induced velocity of the lower rotor
(xv-15 airfoil) blades. Additionally, performance char- vi hover induced velocity
acteristics of the two individual rotors (upper/lower or vu induced velocity of the upper rotor
front/aft) that comprised the coaxial and tandem rotor sys- z axial separation distance
tems were measured independently while operating in a α angle of attack
torque-balanced condition. The results, when compared θ collective pitch angle
against single-rotors, allowed several aerodynamic inter- κ induced power factor
ference loss factors to be determined. Thrust sharing ratio, κint interference loss factor
hovering efficiency of the rotor system, efficiency of the κov overlap factor
individual rotors, and influence of one rotor on the other κsep separation loss factor
were all measured at various rotor separation distances for κu , κl rotor-on-rotor influence factor
both coaxial and tandem rotor systems. Momentum the- σ rotor solidity
ory was used to compare and contrast the measurements φ inflow angle
against predicted values. Ω rotational speed of the rotor

subscripts:
Nomenclature a aft rotor
f front rotor
A rotor disk area i induced
c blade chord l lower rotor
Cd drag coefficient u upper rotor
Cd0 minimum (profile) drag coefficient
Cl lift coefficient
∗ Research Scientist.
Manikandan.Ramasamy@amrdec.army.mil Introduction
Presented at the AHS 69th Annual Forum, Phoenix, AZ,
May 21–23, 2013. This material is declared a work of the U. S. There is an ever increasing demand for helicopters to fly
Government and is not subject to copyright protection in the faster, quieter, carry more payload over a longer range,
United States. and at a cruise efficiency that is similar to fixed-wing
aircrafts. The next generation multi-role rotorcraft are alyzing more aspects of the coaxial rotor configuration,
expected to satisfy all these requirements without com- such as the influence of one rotor on the other (Ref. 11),
promising hover efficiency. Among the current genera- the effect of axial separation on the coaxial system per-
tion helicopters, both conventional (single main-rotor/tail- formance (Refs. 11, 12) etc., and to conduct feasibility
rotor) and multi-rotor configurations (such as tandem or studies for potential applications, such as micro-air vehi-
tilt-rotor aircraft) are unique in what they offer, and have cles (Ref. 13) and ducted coaxial rotor systems (Ref. 14).
secured a place both in military and civil applications. The Coleman (Ref. 15) reviewed and summarized the status
recent success of Sikorsky’s X2TM Technology Demon- of coaxial rotor research in1997, using various available
strator brought coaxial helicopters back in to considera- experiments and simulations. Since then, few experiments
tion for potential applications. The basic design for the have been conducted on coaxial rotors; baseline findings
next generation rotorcraft, therefore, can come from any from such experiments predominantly aligned with Cole-
of the existing configurations or from their combinations. man’s summary. Nevertheless, consistencies and con-
Any leap in terms of performance from the current genera- tradictions are expected from such a long list of experi-
tion helicopters to the next clearly depends on developing ments. While consistencies in the observation across var-
comprehensive understanding of rotor performance. ious measurements ascertain the fundamental knowledge
In the early years of helicopter research, developing sci- developed from the research, contradictions that are unex-
entific understanding of rotor performance was achieved plainable using simple physics are problematic. For exam-
primarily through experiments and analytical solutions. ple, both experiments and simulations consistently sug-
Knight & Hefner (Ref. 1) conducted a series of exper- gest that coaxial rotors perform better than a single-rotor
iments to understand the effect of solidity on rotor per- with equivalent solidity (Refs. 6–8, 11, 16–19). How-
formance nearly 75 years ago. Blade element momen- ever, the percentage gain in performance varied from 5%
tum theory (BEMT) was used by Gessow (Ref. 2) as to 11%, which is understandable considering the varia-
early as 1948 to conduct parametric studies on rotor per- tions in the blade planforms, Reynolds numbers, and axial
formance. Lately, with the evolution of computers, more separation distances at which the experiments/simulations
and more sophisticated computational tools are being de- were conducted.
veloped to understand and predict rotor performance. To The necessity for more experiments became apparent
this end, comprehensive codes (Refs. 3, 4) and high-end when defining the hover efficiency of the coaxial rotor sys-
computational simulations (Ref. 5) have reached matu- tem. Historically, figure of merit (FM) of coaxial rotor has
rity in predicting the hover performance of conventional been defined (Ref. 15) as
single-rotor systems. 3/2 √
Transforming a predictive tool into a design tool to- Tcoax / 2ρA
FM = (1)
wards developing next generation aircraft require substan- (Pu + Pl )
tial validation, especially using measurements from multi-
where Tcoax is the total thrust produced by the coaxial rotor
rotor systems. Tables 1 and 2 list the majority of the
system (i.e., Tu + Tl ), and A is the rotor disk area. Leish-
hover performance measurements conducted on coaxial
man (Ref. 20), towards accurately estimating the perfor-
and tandem rotor systems, respectively, along with the
mance of coaxial rotor systems, showed the deficiencies
complementing single-rotor measurements. Variations to
of using Eq. 1 that inherently assumes both rotors (upper
the coaxial (variable geometry and lift-offset) and tandem
and lower) produce an equal share of thrust. Using mo-
(tilt and side-by-side) configurations are included as well.
mentum theory, two more definitions were suggested to
Such a long list of experiments may appear comprehen-
appropriately reflect various possible experimental con-
sive, however, a careful analysis of their shortcomings
ditions. These definitions help in correctly establishing
substantiated the necessity for more experiments, not only
the momentum theory limits (ideal power), which should
to validate simulations but also to provide fundamental
translate into accurate estimation of induced power lost
understanding of the flow physics. A brief description of
through aerodynamic interference in coaxial rotor sys-
the coaxial rotor experiments (followed by the tandem ro-
tems. For a torque-balanced coaxial rotor system (realistic
tor experiments), in terms of their relevancy to the present
configuration, however, with unequal thrust sharing), the
study is discussed to establish goals for the experiments
ideal power and FM are given by
conducted in the current work.
Coaxial Rotors: Measurements on coaxial rotor con- 3/2
Tu +T
3/2
figurations, in the beginning, focussed on determining Pideal = √ L (2)
2ρA
the relative performance merits against single-rotor sys-
tems. The measurements were also used to develop an- and
3/2 3/2 √
alytical methodologies to predict coaxial rotor perfor- (Tu + Tl )/ 2ρA
mance (Refs. 6–8). Later experiments were aimed at an- FM = , (3)
Pu + Pl
Citation Configuration Radius Tip Speeds Nb σ z/D Conditions
ft ft/s
Harrington, 1951 (Ref. 6) single 12.5 540 2 0.027
coaxial 12.5 540, 450, 320 2+2 0.054 0.095 yaw trim
coaxial 12.5 540 2+2 0.054 0.095 no trim
single 12.5 390, 260 2 0.076
coaxial 12.5 390, 260 2+2 0.152 0.08 yaw trim
Dingeldein, 1954 (Ref. 7) single 12.5 500 2 0.027
coaxial 12.5 500 2+2 0.027 0.08 yaw trim
Cheney, 1969 (Ref. 8) single 2 n/a 3 0.082 n/a
coaxial 2 n/a 3+3 0.164 n/a lift offset
Rorke, 1976 (Ref. 9) single 26.5 450, 500 6 0.098
coaxial 26.5 450, 500 3+3 0.098 0.025 variable geometry
Arents, 1977 (Ref. 10) coaxial 18 650 3+3 0.076 0.07 yaw trim
Nagashima et al., 1978 (Ref. 11) single 1.25 400 2, 4 0.2
coaxial 1.25 400 2+2 0.4 0.105, 0.21 Math trim
coaxial 1.25 400 2+2 0.4 0.105 – 0.5 no trim
McAlister, 2006 (Ref. 12) single 2 165 3 .07
coaxial 2 165 3+3 0.14 0.1 – 0.73 yaw trim
Bohorquez, 2007 (Ref. 13) single 0.28 30 – 90 3 0.11
coaxial 0.28 30 – 90 3+3 0.22 0.09–0.31 yaw trim
Lee & Leishman, 2010 (Ref. 14) single 0.58 35–88 3 0.078
coaxial 0.58 35–88 3+3 0.157 0.07 – 0.15 yaw trim

Table 1: Hover performance measurements on coaxial rotor systems

respectively. Equation 2 results in lower ideal power than are completely different at all axial separation distances.
the one used in Eq 1. Identifying the limit set by the laws Characterizing the performance of individual rotors is,
of physics is important when developing blade/rotor de- therefore, essential to successfully optimizing the coaxial
sign strategies aimed at reducing induced power of the rotor, as a system. The above examples clearly suggest the
coaxial rotors. To this end, Eq. 3 shows that the individ- need for more comprehensive coaxial rotor performance
ual thrust produced by the upper and lower rotor must be measurements.
known to accurately estimate the efficiency of the coaxial Another aspect of the coaxial rotor system that needs
rotor system. None of the existing measurements listed to be understood is the effect of axial separation dis-
in Table 1 provided the individual performance charac- tance, z, on the overall system performance. Again, Ta-
teristics of the upper and lower rotor when they oper- ble 1 shows limited measurements, where the effect of
ated as part of torque-balanced coaxial rotor system, ex- axial separation was studied at a balanced-torque condi-
cept for Refs. 11 and 12. Even among the two excep- tion. The challenges associated with spinning the rotors
tions, Nagashima et al. (Ref. 11) achieved torque bal- in close proximity limited the experiments from measur-
ance mathematically, i.e., thrust and power of the up- ing the performance less than an axial separation distance
per and lower rotors were originally measured on torque- of 0.1D, where D is the diameter of the rotor. With that
imbalanced coaxial rotors. Torque balance was achieved constraint, performance measurements made at axial sep-
through interpolation and/or curve fit. The procedure is aration distances greater than 0.1D showed no dependence
acceptable, however, the accuracy is dependent on the on z. (Refs. 11–14). Lee et al. (Ref. 18), and Wachs-
accuracy of the curve fit to the measured performance. press et al. (Ref. 21) predicted similar independence, as
McAlister’s (Ref. 12) experiments were aimed at analyz- well. However, using a vortex/Momentum/Blade element
ing the effect of axial separation distance (z) on the coaxial approach, Andrew (Ref. 22) suggested that the perfor-
system performance. Therefore, only one measurement mance of the coaxial rotor system improves with increas-
point was made at each z, which may not be sufficient to ing axial separation distance until 0.05D. Johnson us-
conduct rigorous analysis. ing CAMRAD II recently predicted that the performance
The importance of measuring the thrust from individ- gain can be achieved until 0.15D (Ref. 17). Unless high-
ual rotors is understandable considering the aerodynamic fidelity measurements are made at small axial separation
environment in which the upper and lower rotor operate distances (< 0.15D), conclusively ascertaining the thresh-
Citation Configuration Radius Tip speeds Nb σ d/D
ft ft/s
Dingeldein, 1954 (Ref. 7) single 7.5 500 2 0.054
tandem 12.5 500 2+2 0.054 1.03
Heyson, 1959 (Ref. 23) tandem 7.625 500 2+2 0.097 1.016
Sweet, 1960 (Ref. 24) single 7.5 500 2 0.0543
tandem 7.5 500 2+2 0.0543 1.02
single 7.625 500 2 0.097
tandem 7.625 500 2+2 0.097 0.51, 1.02
Stepniewski, 1984 (Ref. 25) single 2 330 – 425 3 0.06
tandem 2 330 – 425 3+3 0.06 0.625 – 1.03
Bender et al., 1984 (Ref. 26) tandem 30 675 3+3 0.085 0.65
Felker et al., 1987 (Ref. 27) single 15.24 590 – 720 3 0.1138
single with 15.24 590 – 720 3 0.1138 1.2
image plane

Table 2: Hover performance measurements on tandem/side-by-side rotor systems

old distance where performance gains diminish is diffi- Towards understanding the effect of shaft-to-shaft dis-
cult. tance between the rotors on the overall system perfor-
Axial separation distance between the rotors also influ- mance, Sweet (Ref. 24) conducted hover performance
ence the thrust sharing ratio between the two rotors. While measurements on a full-scale helicopter at two overlap
Nagashima et al. (Ref. 11) showed that the thrust sharing distances. A 14% increase in the total power was mea-
strongly depended (50% change in lower rotor thrust over sured when the overlap was approximately 50% compared
1D) on the axial separation distance, McAlister showed a to no-overlap case. Later, Stepniewski (Ref. 25) con-
weak dependence (10% change in lower rotor thrust over ducted a series of experiments on a side-by-side rotor sys-
1D). The difference in the observation between the two tem. The experiments were conducted by keeping the col-
results can be attributed to many differences between the lective pitch settings on both rotors unchanged while in-
two experiments, including the blade planform, Re num- creasing the overlap. The experiment was repeated at sev-
ber, and the trim condition. Despite the varying depen- eral collective pitch settings. Such an experiment requires
dence (which play an important role in the ideal power developing an analytical method or simulation to under-
estimation, shown in Eq. 2), the performance of the coax- stand the effect of shaft-to-shaft distance on the system
ial rotor system in both experiments did not vary much performance, as demonstrated by Harris (Ref. 28). Using
with the axial separation distance greater than 0.1D. More BEMT, Harris estimated the increase in induced power as
comprehensive measurements that can explain the varying a function of overlap distance. In general, as overlap in-
performance characteristics exhibited by the experiments creases, local disk loading increases that, in turn, reduces
are needed. the efficiency of the tandem rotor system. An engineer-
Tandem Rotors: The number of reported experiments ing approximation was also developed to relate the shaft-
on tandem rotor systems (and related variations) is smaller to-shaft distance between the rotors to the induced power
than those found for coaxial configurations. Similar to lost resulting from overlap. Key measurements missing
coaxial rotor configurations, several fundamental ques- in all the tandem rotor experiments, again, are the perfor-
tions prevail. For example, the effect of rotor overlap on mance characteristics of the front and aft rotor, when the
the system performance or the influence of the front rotor rotors operate as a part of torque-balanced tandem rotor
on the aft rotor and vice versa remain unclear. Consis- system. The importance of such measurements increases
tencies and contradictions among measurements are com- when there is an overlap between the rotors, combined
mon, as well. For example, Dingeldein (Ref. 7) measured with a finite axial separation distance. Characterizing the
a 18% increase in the efficiency of tandem rotor systems individual rotor performance is necessary to improve the
compared with single-rotors, when the shaft-to-shaft dis- overall efficiency of the tandem rotor, as a system.
tance was approximately equal to the diameter of the ro- To summarize all the past experiments on multi-rotor
tor (i.e., at no overlap). However, both Heyson (Ref. 23) systems, the complex nature of the set up did not al-
and Sweet (Ref. 24) did not measure any beneficial aero- low modifications to be made, leaving many questions
dynamic interference between rotors that resulted in net unanswered. The objective of the present experiment
improvement in the system efficiency. was two-fold. First, to develop a versatile, multi-rotor
(a) Tandem rotor (b) Coaxial rotor

Figure 1: Tandem and coaxial rotor stands

rig that allows simple modifications to be made quickly


and with relative ease, without compromising on the
quality of the measurements. Second, to conduct a se-
ries of high-fidelity hover performance measurements us- Thrust  sensor  
ing single, coaxial, and tandem rotor configurations with
configuration-specific goals that are listed below.
Torque  sensor   Transmission  (6:1)  
• Coaxial rotor: (1) Compare the coaxial rotor system
performance with single-rotor with equivalent solid- Motor  
Thrust  bearings  
ity, and with two isolated single-rotors whose com- Collec:ve  pitch    
servo  
bined solidity is the same as coaxial rotors, (2) quan-
tify the effect of axial separation on the system per-
formance at low and high system thrust conditions
with balanced torque, (3) measure the influence of Figure 2: Close up view of the set-up showing various
one rotor on the other at various axial separation dis- components of the rotor system
tances, and (4) compare and analyze the similarities
and differences between the untwisted and highly- Experimental Setup
twisted blade planforms for the four aforementioned
goals. Two identical, but independent, model-scale rotor rigs
were built at the U. S. Army test facility (26- by 24- by
• Tandem rotor: (1) Compare the performance of the 20-ft) located at the NASA Ames Research Center – see
tandem/tilt rotor configurations against two isolated Fig. 1. The rigs can be moved relative to each other, serv-
single-rotor systems whose combined solidity is the ing two purposes: (1) the set up can be converted from
same as the tandem rotor system, (2) measure the single to multi-rotor configurations with relative ease, and
induced power lost because of overlap by gradually (2) within a configuration, changes to the rotor system pa-
moving from tandem configuration to coaxial config- rameters can be made, allowing parametric studies (for
uration, (3) establish and understand the performance e.g., effect of axial separation distance on coaxial rotor
characteristics of the front and aft rotor relative to performance) to be conducted.
a single-rotor system during the transition from tan-
Figure 2 is a close up view of the test stand showing
dem to coaxial rotor configuration, and (4) compare
various components of the rotor system. Two brushed DC
and analyze the similarities and differences between
motors with 1.6-hp drove the rotor system. The rotational
the untwisted and highly-twisted blade planforms for
speed was controlled by adjusting the supplied current.
the three aforementioned goals.
A pair of Hall effect sensors attached to the rotor shafts
All the experiments were conducted at the same facility provided the 1/rev signal and was used to measure the ro-
using the same set of blades, instrumentation, and data tational speed. A pair of accelerometers attached to the
acquisition system. This allowed hover performance of stand measured the system vibration. The rotors were
various rotor configurations ignoring the facility effects. tracked and dynamically balanced using a digital signal
Parameter Untwisted Twisted
(a)   (b)  
Number of blades 2 to 6 2 to 4
Radius, ft 2.17 2.15
Chord, in 2.29 2.0
Root cut-out 19.1% 20.1%
Airfoil NACA 0012 see Fig. 4
Re1200 rpm 325,000 275,000
M1200 rpm 0.25 0.25

Table 3: Rotor blade characteristics.


quired to spin the rotor blades, but also the transmission
Figure 3: Collective pitch measurement
losses such as frictional losses in the bearings, couplings,
analyzer. All systems had open loop control, meaning and gears etc. Such losses cannot be removed by sim-
any required adjustment to the rotor settings was achieved ply subtracting tares. The transmission efficiency changes
manually. with the load (thrust produced by the rotor) and the rota-
A thrust sensor (25-lb range) was mounted along the tional speed. In the present study, these losses were esti-
axis of each rotor shaft measuring the axial load. A re- mated by applying a known torque at a given rpm using
action torque sensor (50 in-lb) was mounted behind each a dynamometer-based brake at the hub end of the rotor
motor (similar to Harrington in Ref. 6) to measure the shaft and measuring the torque behind the motor using
torque exerted by the rotor systems. Thrust, torque, ro- the reaction torque sensor. This procedure was repeated
tational speed, voltage, current, and temperature of the to cover the entire range of experimental test conditions.
motor for both systems were acquired using in-house data The measured transmission efficiency was approximately
acquisition programs written in Labview. The measure- 88% and, as expected, changed with load and rpm. All
ments were acquired at 1 kHz for four seconds, which cor- the reported measurements in the present study were cor-
responded to 80 revolutions at 1200 rpm, and 50 measure- rected for transmission efficiency. Also, tares were esti-
ments within one revolution (7.2◦ resolution).The data set mated at each rotational speed and were subtracted from
was then used to estimate the mean and the standard de- the measurements.
viation. Both thrust and torque showed slightly less than Operating Conditions:
1% scatter in the data at all thrust conditions. The list of experiments conducted in the present study
The rotor systems were equipped with servo controls along with the associated trim conditions are listed in Ta-
to change the collective pitch on the fly. Collective pitch ble 4. All coaxial and tandem rotors measurements were
angles were measured by taking chordwise images of the made at yaw trim, meaning Q1 +Q2 = 0. Collective sweep
rotor blade using CCD cameras. To prevent introducing was conducted by keeping the separation distance (ax-
absolute errors in the process, rotors were spun at the re- ial, in the case of coaxial rotors and shaft-to-shaft in the
quired rotational speed and the pitch was adjusted until the case of tandem rotors) constant. When the separation dis-
thrust produced approached zero, at which point chord- tance was varied for parametric studies, two trim condi-
wise images were taken and were used as reference. Sim- tions were applied: (1) yaw trim, and (2) constant total
ilar images acquired at various thrust condition were com- thrust (Tu + Tl = T f + Ta =constant).
pared against the reference to estimate the collective pitch
of the blades. Representative images of the blade taken
during coaxial rotor measurement are shown in Fig. 3. Results
Rotor blades: The experiment was conducted using
two sets of blades: (1) untwisted, rectangular blade with The observations made from the experiments are dis-
NACA 0012 airfoil throughout the entire span, and (2) cussed in the following order.
highly twisted xv-15 blade with four different airfoils at
various sections of the blade span. Figure 4 shows the 1. Single-rotor measurements
twist distribution and the associated airfoil details for the
highly twisted blade. Table 3 shows essential geometric 2. Coaxial rotor measurements
details of both rotor blades.
Transmission Efficiency: Installing a reaction torque 3. Side-by-side/Tandem/Tilt-rotor measurements
sensor behind the motor to measure the aerodynamic
torque required by the rotor blades is not ideal. The Comparisons with momentum theory predictions of simi-
measured torque not only includes the actual torque re- lar rotor configurations are made wherever possible.
Figure 4: Details of twisted blade used in the study (from Ref. 12)

Rotor type Planform RPM Nb z/D d/D Trim condition(s)

Untwisted 800 &1200 2 to 6


single-rotor
Twisted 800 to 1400 2 to 4
Untwisted 1200 3+3 0.07 0
Qu + Ql = 0
Twisted 1200 3+3 0.07 0
Coaxial rotor
Untwisted1 800 3+3 0.05 to 1.5 0 Qu + Ql = 0 &
Twisted 800 3+3 0.05 to 0.75 0 Tu + Tl =const.
Side-by-side rotor Untwisted 1200 3+3 0 1.1 Q1 + Q2 = 0
Twisted 1200 3+3 0 1.1 Q1 + Q2 = 0
Untwisted 1200 3+3 0.07 0.9 to 1.02 Q f + Qa = 0 &
Tandem rotor
Twisted 1200 3+3 0.07 0 to 1.0 Tl + T f =const.

Table 4: List of experiments conducted in the present study.

Single-rotor periment was conducted and (2) established the baseline


characteristics of the single-rotor system to compare with
Performance measurements of a single-rotor system us- multi-rotor configurations. The latter is particularly im-
ing both twisted and untwisted blades are analyzed in this portant for model-scale experiments, considering the lim-
section. Many of the observations made here have been ited availability of airfoil tables at low Re numbers. Ac-
substantiated in the past using various experiments con- curately estimating the profile drag (and, therefore, pro-
ducted at different facilities, or through simulations and file power) is key to understanding the measured induced
analyses. However, these measurements were necessary power across various rotor configurations (Ref. 19).
and served two purposes: (1) evaluated the limitations of Figure 5 shows the thrust and power measured on a 3-
the rotor system, thereby, identifying the operating condi-
tions (such as the tip speed, thrust etc.) at which the ex- 1 Two cases, CT = 0.014, 0.007
Figure 6: Performance of a single-rotor system using
twisted blades at various rotational speeds.

conventional wisdom gained from various full-scale heli-


copter rotor analyses conducted in the past, where oper-
ating at low tip speeds has shown known to yield perfor-
mance gains among other operational benefits. To under-
stand the difference between full-scale and model-scale,
power must be analyzed in terms of its components: pro-
file power and induced power.
At full-scale, rotor blade tips operate at very high
Reynolds number (of the order of millions). Therefore,
a reduction in the Re number (by reducing the tip speed)
does not change the profile power substantially. How-
ever, the induced power is reduced because power load-
ing (thrust produced for a given power) varies inversely
with the tip speed. Consequently, a reduction in the
rpm increases the overall aerodynamic efficiency. On
Figure 5: Thrust and power measurements of a single- the other hand, model-scale experiments are conducted at
rotor system using twisted blades. low Reynolds numbers (of the order of 2×105 to 3×105 ),
where the gains made in the profile power by operating
bladed single-rotor system using twisted blades. Collec- at higher tip speeds (and higher Re numbers) significantly
tive pitch was kept constant during each run and the ro- outweigh the additional power lost in the form of induced
tational speed was gradually increased from 800 to 1400 power. This results in higher efficiency at higher tip speed,
rpm. The runs were repeated to cover a wide range of as seen in Fig. 6. Nevertheless, the baseline measure-
collective pitch angles (θtip ) from as low as -6◦ to 10◦ . ments showed the influence of tip speed on the rotor per-
The capacity of the motor primarily dictated the maxi- formance, and the necessity to conduct (and compare) sin-
mum achievable rpm/collective pitch combination during gle and multi-rotor measurements at the same Re number.
each run. The rotor system produced positive thrust even The baseline measurements also helped identify the ro-
when the tip of the blade was operating at relatively high tational speeds at which the rest of the study was con-
negative pitch angle (-6◦ ), a unique characteristic of the ducted. Both 800 and 1200 rpm were chosen, as 1200
highly-twisted blades. More details of this behavior will rpm represented the highest rotational speed and the high-
be discussed later. est measured FM before stall, and 800 rpm corresponded
The effect of Reynolds number on the performance of to the majority of the flowfield measurements made us-
the single-rotor system was readily apparent when the ing the same set of blades reported by Ramasamy et al.
measurements were plotted in terms of hover efficiency in Ref. 29. The velocity field measurements were found
– see Fig. 6. At all thrust conditions, figure of merit (FM) useful in explaining the observed performance character-
increased with increasing Re number. For example, a istics, especially when multi-rotor configurations were an-
50% increase in Re number (800 rpm to 1200 rpm), re- alyzed.
sulted in a 8.5% increase in FM. This is contrary to the Solidity Effect: Comparing the performance of multi-
(a) Untwisted blades

(b) Twisted blades

Figure 7: Baseline performance of untwisted and twisted blade configurations at Vtip = 55m/s

rotor systems against single-rotor after matching solidity, example, fewer number of blades provided better effi-
as done primarily in the present study, would not necessi- ciency to produce a given thrust (attributed to the lower
tate a study on solidity effects. Previous experiments con- profile power associated with lower solidity).
ducted to compare single and multi-rotor performances The shortcomings of using fewer blades was obvious,
were made at unequal solidity, i.e., σsingle = 12 σcoax . In as shown in Fig. 7a. The two-bladed rotor system showed
order to enable comparison with previous experiments, signs of stall when the thrust coefficient was as low as
some of the analysis made in the present study were also 0.005, and was limited in its capability to produce high
conducted at unequal solidity. Therefore, establishing the thrust. Increasing the number of blades from two to three
effect of solidity on single-rotor performance before mak- pushed the stall boundary farther, resulting in higher val-
ing comparison with multi-rotors was essential. In the ues of CT . When four or more blades were used, the power
present study, solidity was changed only through chang- limitation of the motor prevented the rotor from reaching
ing the number of blades. the stall boundary.
In this section, baseline performance characteristics of Performance measurements using 2 to 4 twisted blades
single-rotor systems using both the untwisted and twisted are shown in Fig. 7b. Similar to untwisted blades, twisted
blades were analyzed by varying the number of blades. blades also exhibited expected fundamental performance
Both thrust and power were measured by conducting col- characteristics. These include, but not limited to, higher
lective sweeps at 800 and 1200 rpm, while changing the performance, and the thrust limitation resulting from stall
number of blades from 2 to 6 for untwisted blades, and 2 when fewer blades were used. However, two key differ-
to 4 for twisted blades. The results for the untwisted blade ences were observed. First, unlike the untwisted blades,
configuration at 800 rpm are shown in Fig. 7a. In general, the minimum power required by the twisted blades oc-
the measurements exhibited everything that is expected curred at a non-zero thrust condition. Second, the relation
from the past experiments and simulations/analyses. For between the power required and the thrust produced by the
(a) CT =0.0 (b) CT =0.007

Figure 8: Flow field of single-rotor system using untwisted blades.

(a) CT =0.0 (b) CT =0.007

Figure 9: Flow field of single-rotor system using twisted blades.

twisted blades were different from the untwisted blades, the present experiment.
especially at low thrust conditions. Representative flow field of the inflow distribution at all
The observed differences are explained using flowfield positive thrust conditions (until stall) is shown in Fig. 8b.
measurements and simple analysis tools such as momen- Such a clearly defined downwash allowed simple analysis
tum theory. Comprehensive velocity field measurements tools such as momentum theory to be used to establish the
made using particle image velocimetry (PIV) on the same efficiency of the rotor system by comparing the measured
set of untwisted and twisted blades at 800rpm were re- induced power to the minimum power required to produce
ported in Ref. 29. Representative time average measure- the same thrust. Power required by a rotor system, as de-
ments corresponding to the untwisted blades at two dif- fined by modified momentum theory, is given by
ferent thrust conditions are shown in Fig. 8. The color
3/2
contour represents the downwash velocity. At zero thrust, C
no flow in any direction was observed, and was expected C p = C p0 + κ √T , (4)
2
because all the sections of the blade (from root to tip) op-
erated at zero collective pitch. Therefore, the measured where C p0 is the profile power and κ is the induced power
power at zero thrust simply corresponded to the profile factor. Curve fitting Eq. 4 to the measurements shown in
power required to spin the rotor blades. The estimation of Fig. 7a yielded the values of C p0 and κ. A similar analysis
profile power allowed separating the induced power from for the twisted blades was very challenging.
the total power at higher thrust conditions, negating the For the twisted blades to produce zero thrust, the out-
need for airfoil tables (to estimate the sectional profile board sections of the blade must produce negative thrust
drag) that may not be available at low Re numbers, as in while the inboard sections produce an equal positive
(a) Untwisted blades

(b) Twisted blades

Figure 10: Baseline performance of untwisted and twisted blade configurations at Vtip = 85m/s

thrust. This behavior was reflected from the upwash found the power measurements cannot be expected to follow
near the tip and downwash at the inboard sections of the the functional dependency given in Eq. 4. As shown in
blade (see Fig. 9a). Geometrically, the collective pitch Fig. 7b, over a range of thrust, the power required varies
at the tip of the twisted blade must be at a high negative 3/2
linearly with thrust, as opposed to the CT relation shown
pitch angle at CT = 0. While the thrust from the inboard in Eq. 4.
and outboard sections cancelled each other, the associated
All the multi-rotor experiments using twisted blades
induced power added up. Consequently, the measured
in the present study were conducted at high thrust con-
power at zero thrust condition included both the profile
ditions, where there is no ambiguity in the flow direc-
power and the induced power resulting from the equal and
tion. Establishing the baseline characteristics of single-
opposite thrust. Hence, the measured total power is higher
rotor systems allows the net change (e.g., ∆κ) in perfor-
at the zero thrust condition than the minimum power re-
mance between the single and multi-rotor configurations
quired, as seen in Fig. 7b.
to be analyzed and quantified without needing the absolute
Reference 29 showed that when highly-twisted blades value of profile power. However, there are other alterna-
were used on single-rotor systems, the inflow distribu- tives, such as using CFD (Ref. 19) or MSES (Ref. 30) to
tion (from root of the blade to the tip) continuously var- accurately estimate the profile power.
ied with increasing thrust, especially at low thrust condi- Performance measurements of untwisted and twisted
tions. PIV measurements made at CT = 0.007 are shown in blades at 1200 rpm are shown in Fig. 10. Except for the
Fig. 9b. A clearly defined flow pattern emerged only when effects of Reynolds number discussed earlier, no differ-
CT > 0.004. This behavior suggests that until the collec- ences in performance are observed between the 800 and
tive pitch angles of the twisted blades are high enough 1200 rpm cases. The measurements shown in Figs. 7 to
that all sections of the blade produce non-negative thrust, 10 established the baseline characteristics of single-rotor
systems. Understanding the similarities and differences
between untwisted and twisted blades, in terms of their
performance and flow field distribution, provided insight
into multi-rotor performance analysis.

Coaxial Rotor
Measurements of a coaxial rotor system are analyzed with
the following specific goals: (1) compare the coaxial rotor
system performance against single-rotor configurations,
(2) quantify the effect of axial separation on the system
performance at low and high system thrust conditions, (3)
measure the influence of one rotor on the other at various
axial separation distances, and (4) compare and analyze (a) Single 6-bladed rotor vs. coaxial rotor (z/D = 0.07)
the similarities and differences between untwisted and
highly-twisted blades for the four aforementioned goals.
Results using untwisted blades are discussed first, fol-
lowed by twisted blades.
Performance Comparison: Two approaches were used
to evaluate coaxial rotor performance. The first approach
compares coaxial rotor performance with a single-rotor
at the same solidity. For the present study, a single
6-bladed rotor was compared against the coaxial rotor
(two 3-bladed rotors). Historically, this has not been the
method of choice because of the hub modifications needed
to accommodate twice as many blades as the coaxial rotor
hubs. Therefore, analytical methodologies (for example,
single-rotor theory (Ref. 6)) were developed to predict the
behavior of single-rotor with equivalent solidity. These (b) Two isolated 3-bladed rotors vs. coaxial rotor (z/D = 0.07)
methods were validated using single-rotor measurements
made on the upper (or lower) rotor that was operated in Figure 11: Performance analysis of coaxial rotors us-
an isolated condition, before comparing against the per- ing untwisted blades
formance of coaxial rotors. The difference between the
obtained after balancing the torque between the two ro-
predictions and the coaxial rotor measurements were at-
tors. Remember, there was no feedback control in any of
tributed to the anomalies found in the coaxial rotor flow
the systems used. So, the pitch angle and and the rpm of
field that were not present in the single-rotor system. The
the both rotors were adjusted manually until torque bal-
accuracy of this approach largely depends on the accuracy
ance was achieved. System thrust (Tcoax ) was obtained by
of the mathematical model, and the accuracy of the profile
adding the thrust from the upper rotor (Tu ) and the lower
drag estimated from the airfoil tables (Ref. 6).
rotor (Tl ). Similarly, system torque (Qcoax ) was obtained
The second approach compares the coaxial rotors by adding the absolute value of the individual torques, i.e.,
against two isolated single-rotors, whose combined so- |Qu | + |Ql |.
lidity is the same as the coaxial rotor. The goal of this Figure 11a shows the system thrust and the system
approach is usually to quantify the power lost/gained be- torque measured on a coaxial rotor system using untwisted
cause of the aerodynamic interference between the two ro- blades; measurements from a single-rotor system with
tors. Both approaches have advantages and provide useful equivalent solidity are also shown. The solid lines rep-
information that are unique to themselves. In the present resent a curve fit to the data using the following equation:
study, both approaches were followed with one key differ-
ence. In the first approach, measurements from six bladed T 3/2
rotors having the same solidity as the coaxial rotor system P = P0 + κ κsep √ (5)
2ρA
were used, instead of the analytical prediction technique
followed in the past. where κsep is given by
The first part of the experiment was conducted by keep-
ing the axial separation distance between the two rotors Pi(coax)
κsep = (6)
constant, i.e., z/D = 0.07. All the measurements were Pi(single)
at the same thrust. For single-rotor systems, κsep =1.
Coaxial rotors required ≈10% less power than an iso-
lated single-rotor with equivalent soldity to produce the
same thrust. The higher performance of the coaxial rotors
was not entirely unexpected, and can be attributed to two
sources: (1) swirl recovery, and (2) axial separation ef-
fects. Swirl recovery has been hinted in the literature as an
important source of power gain and, therefore, improved
performance whenever there is a need to emphasize the
benefits of coaxial rotor. The idea of reduced swirl losses
in coaxial rotor systems is logical considering the viscous
nature of the flow, however, no direct/indirect measure-
ment exists to validate the hypothesis.
The effect of axial separation on the performance of a
coaxial rotor system can be evaluated using momentum Figure 12: Performance of coaxial rotor (z/D = 0.07)
theory, albeit the procedure is not straightforward. For a using twisted blades
single-rotor system producing a thrust of 2T , the hover
induced velocity is given by in Fig. 11b are curve fits to the data using the following
s equations:
(2T ) T 3/2
vi = vh = , (7) Psingle = P0 + κ √ (13)
2ρA 2ρA

where T = W /2, and W is the gross weight of the heli- and


copter.The associated induced power is κ (2T )3/2
Pcoax = 2P0 + κint √ √ , (14)
s 2 2ρA
(2T ) where κint is the interference loss factor, which represents
Pi(single) = (2T ) (8)
2ρA the increase in the induced power resulting from oper-
ating the two rotors in coaxial configuration rather than
Leishman (case 2, Ref. 20), using simple momentum the- an isolated condition. Coaxial rotors showed nearly a
ory, related the induced power lost by the single-rotor to 22% increase in the induced power compared to two iso-
two isolated rotors producing the same total thrust as lated single-rotors producing the same thrust. The loss
√ was expected, again, based on simple momentum theory
Pi(single) = 2 [2 Pi(isolated) ] (9) that predicts 26% increase in the induced power (case
4b, Ref. 20), and is understandable considering one of
Next, for a torque-balanced coaxial rotor with the lower
the two rotors always operates in the wake of the other.
rotor operating in the fully contracted wake of the upper
Figure 12 shows similar performance comparison when
rotor, momentum theory based induced power is given by
twisted blades were used.
Pi(coax) = Pu + Pl = Tu vu + Tl (vu + vl ) (10) While momentum theory results confirmed the validity
of the present measurements, at least in terms of perfor-
Similar to the single-rotor case, Pi(coax) was related to two mance gains/losses, the assumptions made in momentum
isolated single-rotors producing same total thrust (case 4a, theory are substantially different from the present exper-
Ref. 20), as imental conditions. For example, the present measure-
ments were conducted at an axial separation distance of
Pi(coax) = 1.281 [2 Pi(isolated) ] (11) 0.07D, that is, the lower rotor operated much closer to
the upper rotor than where the fully contracted wake from
Therefore the upper rotor was expected (0.2D), based on the single-
Pi(coax) √ rotor flow field measurements (reported in Ref. 29 using
κsep = = 1.281/ 2 = 0.90, (12) the same set of blades). Consequently, the lower rotor
Pi(single)
may begin to influence the performance of the upper rotor,
which is close to the measured value. which is not accounted for in momentum theory. Further-
Results using the second approach comparing the per- more, momentum theory assumes uniform inflow, which
formance of the coaxial rotors against two isolated single- is not realistic.
rotors, whose combined solidity is equal to the solidity Rotor-on-Rotor Influence: The next step in the coaxial
of the coaxial rotor is shown in Fig. 11b. The solid lines rotor analysis is to understand and measure the influence
(a) Untwisted blade (b) Twisted blade

Figure 13: Coaxial rotor measurements quantifying rotor-rotor influence.

of the lower rotor on the upper, and vice versa. The thrust which is discussed next.
and power measurements from the individual rotors (op- Axial Separation Distance: A series of experiments
erating in a torque balanced condition), when compared were conducted to understand the effect of the axial sepa-
with the isolated rotor measurements at the same solidity, ration distance (z/D) on the coaxial rotor performance, as
provided the necessary information to understand the in- well as to measure the resulting power loss factors, such as
fluence of one rotor on the other. Figure 13a shows one κint and κu|l . Thrust and torque of the upper and lower ro-
such measurement when untwisted blades were used. The tors were measured by changing the axial separation dis-
solid lines are curve fits to the equation tance from 0.05D to 1.5D. The system thrust (Tu + Tl ) was
3/2
kept constant and the torque was balanced (Qu + Ql = 0)
C throughout the experiment. Collective pitch angles were
CQ = CQ0 + κ κu|l √T (15)
2 also measured to understand the exhibited rotor perfor-
where κu and κl are the rotor-on-rotor (ROR) influence mance characteristics.
factors that represent the induced power lost by the up- Two sets of measurements were made in the case of
per rotor because of the influence of the lower rotor and untwisted blades that corresponded to system coefficient
vice versa, respectively. The induced power factor, κ, es- of thrust, CTcoax , of 0.014 and 0.007. In the case of
timated from isolated rotor measurements was 1.25. Both twisted blades, the measurements were limited to one case
the upper and lower rotor performed poorly when com- CTcoax = 0.014. The coaxial rotor figure of merit was esti-
pared with the isolated rotor. The upper rotor downwash mated using
3/2 3/2
CTu CT
is known to degrade the performance of the lower rotor, √ + √l
2 2
and was evident from the estimated κl . The lower rotor FMcoax = (16)
needed about 35% more induced power than the isolated CQu +CQl
rotor to produce the same thrust. Also, the results sug- As discussed earlier, Ref. 20 explains the importance of
gested a measurable influence of the lower rotor on the using Eq. 16, which accounts for the thrust sharing be-
upper rotor, leading to a 9% increase in the induced power tween the rotors, as opposed to using system thrust and
of the upper rotor. system torque, to evaluate coaxial rotor performance.
Similar measurements using twisted blades are shown Figures 14 and 15 show the hovering efficiencies of
in Fig. 13b. The induced power factor, κ, estimated from the coaxial rotor system at various axial separation dis-
isolated rotor measurements was 1.1. Although κ was tances using both untwisted and twisted blades. The as-
much lower compared to the isolated rotor system using sociated interference, and rotor-on-rotor influence factors
untwisted blades, the overall ROR influence was identical. are shown in Fig. 16. Independent of the blade planform
A flowfield analysis of the wake geometry may explain or the thrust conditions, performance of the upper rotor
the similarity between the two blade planforms, however, was better than the lower rotor at all measured z/D. The
is beyond the scope of the current work. The inflow dis- downwash from the upper rotor deteriorated the perfor-
tribution, which is the primary difference between the un- mance of the lower rotor even when the rotors were far
twisted and highly-twisted blades, did not seem to affect apart (z/D > 0.75). Measurements of coaxial rotors made
the ROR influence factors. However, the ROR influence using untwisted blades were analyzed in detail, followed
factors must change with the axial separation distance, by the twisted blades.
(a) FM vs. z/D (b) FM vs. CT

Figure 14: Effect of axial separation distance on FM - untwisted blades at CT coax = 0.007 and 0.014.

(a) FM vs. z/D (b) FM vs. CT

Figure 15: Effect of axial separation distance on FM - twisted blades at CT coax = 0.014.

Figure 14a compares the hovering efficiency of the ually decreased when z/D decreased. The lower rotor, on
coaxial rotor system using untwisted blades at two dif- the other hand, showed gradual increase in efficiency. The
ferent thrust conditions, measured at various axial sepa- overall coaxial rotor performance, therefore, showed no
ration distances. Performance measurements of the indi- measurable change. Region 3 corresponds to z/D < 0.2,
vidual rotors (when they operated as part of the coaxial where the upper rotor showed accelerated reduction in
rotor system) are also shown. Because the measurements performance as z/D decreased. This was accompanied by
were made at a balanced-torque condition, the difference an accelerated increase in lower rotor performance. How-
in FM between the upper and the lower rotor also reflects ever, unlike region 2, the overall FM of the coaxial rotor
the difference in the thrust produced by the two rotors for decreased.
the same power. Observations made at a higher thrust con- To understand the observed difference between region 2
dition (CT coax = 0.014) were analyzed first, before making and 3 more clearly, the effect of varying the axial separa-
comparison with the lower thrust condition case. tion distance on various rotor parameters was examined.
Three distinct regions can be identified. Region 1, when The exhibited reduction in thrust by the upper rotor and
z/D > 0.75, performance of the coaxial rotor was invari- the associated increase in thrust by the lower rotor when
ant to the changes in the axial separation distance. The in- the rotors came closer (in region 2 and 3) can be explained
dividual rotor characteristics showed indifference, as well. using blade element momentum theory. The forces nor-
Region 2, 0.2 < z/D < 0.75, FM of the upper rotor grad- mal and parallel to the tip path plane of rotor blade per
(a) Interference power loss factor (b) Rotor-on-Rotor influence factors

Figure 16: Power loss factors

(a) Collective sweep θ (b) Thrust ratio, Tu /Tl

Figure 17: Rotor system parameters and thrust ratio at various z/D

unit span are given by where φ is the ratio of the normal velocity (inflow veloc-
ity) to tangential velocity (Ωr) in hover, when small angle
dFz = dL cos φ − dD sin φ (17) assumptions are made. More details of the blade element
momentum theory are found in Ref. 31.
and When z/D decreases (i.e., when the lower rotor comes
dFx = dL sin φ + dD cos φ (18) closer to the upper rotor) in region 2, the influence of the
respectively. Lift (L) and drag (D) per unit span are given lower rotor on the upper rotor increases. This is under-
by standable from inflow perspective or through Biot-Savart
1 law. As a result, the inflow angle of the upper rotor φu
dL = ρV 2 c Cl dy (19) increases, thereby, causing the angle of attack (αu ) to de-
2
crease for a constant collective pitch. This, in turn, re-
and duces the magnitude of thrust (Fz ) produced by the upper
1
dD = ρV 2 c Cd dy (20) rotor. Also, increasing the inflow angle increases Fx and,
2 therefore, torque. In summary, by moving the lower ro-
where Cl and Cd are lift and drag coefficients, and V is the tor closer to the upper rotor, the thrust produced by the
resultant velocity. The relation between the angle of at- upper rotor is decreased and the torque is increased, pro-
tack, blade collective pitch, and the inflow angle are given vided that the collective pitch was unchanged. Figure 14b
by confirmed the sequence of events, showing a reduction
α = θ−φ (21) in thrust and an associated reduction in performance (be-
cause of increased torque/power). The collective pitch of rington. The argument was based on the occurrence of
the upper rotor remained unchanged in region 2, as shown lower rotor stall at high thrust conditions, preventing the
in Fig. 17a. applicability of single-rotor theory to explain coaxial ro-
A reduction in the thrust produced by the upper rotor tor characteristics. However, no difference in the mea-
directly translates in to lower downwash velocity on the surements were found when Dingeldein repeated the ex-
lower rotor. Following the same argument, the reduction periment using Harrington rotor (showing no stall effects),
in downwash velocity reduces the inflow angle (φl ) and forcing a conclusion that the single-rotor theory was suffi-
increases the angle of attack (αl ), when θl is unchanged. cient to predict coaxial rotor performance. Unfortunately,
Consequently, the lower rotor produces more thrust. And, both his own measurements and Harrington’s measure-
the torque decreases. However, the experimental trim con- ments were made at one axial separation distance and,
ditions enforced more changes. Because the two rotors more importantly, at thrust conditions too low to stall the
must operate at the same power (torque balanced condi- rotor blades. Unless the thrust conditions are high and
tion), either Ql must be increased to match Qu or Qu must the rotors are placed very close to each other (which may
be reduced to match Ql . The former can be achieved by vary between model-scale and full-scale), stall may not
increasing the collective of the lower rotor (Tl increases) occur. This was evident from the present measurements
and the latter by reducing the collective pitch of the upper when the rotors were operated at low thrust conditions
rotor (Tu decreases). The choice depends on the second (CTcoax = 0.007, similar to Harrington’s measurements).
experimental trim condition, which forces the total thrust Coaxial rotors showed no signs of stall and did not show
to remain constant. It is apparent from the measurements a reduction in performance even for z/D < 0.15. Never-
shown in Fig. 17a that the second trim condition forced theless, the lower rotor blades are expected to stall (invali-
the lower rotor to operate at higher collective to match the dating the use of single-rotor theories, as intended by Din-
torque of the upper rotor. This resulted in more lift pro- geldein), provided that the coaxial rotor system operate at
duced by the lower rotor that has the same effect as bring high thrust conditions, and at very small axial separation
the lower rotor closer to the upper rotor. The cycle con- distance between the upper and lower rotor.
tinues until an equilibrium is reached, at which point, the Figure 16a shows the effect of axial separation distance
lower rotor produced more thrust than it was when the ax- on the performance of the coaxial rotor system in terms of
ial separation distance was higher. The upper rotor, on the κint loss factor. As mentioned earlier, kappaint represents
other hand, produced less thrust. The resulting thrust ra- the additional induced power required by the coaxial ro-
tio, as the axial separation distance changed is shown in tor system when compared against two isolated 3-bladed
Fig. 17b. rotor systems. The value of interference loss factor κint ,
Region 3 is different from region 2 in that the collec- for regions 1 and 2 in the case of coaxial rotors using
tive pitch of both the upper and lower rotor increased with untwisted blades, remained constant. As z/D decreased
decreasing z/D. As the lower rotor moved closer to the (region 3), κint increased rapidly towards higher values,
upper rotor, the collective pitch of the lower rotor gradu- approaching 1.41 set by momentum theory.
ally increased. At z/D = 0.15, the collective pitch of the The effect of axial separation distance on the rotor-on-
lower rotor operated at/near the stall angle θtip = 15◦ - rotor influence factors is shown in Fig. 16b. The estima-
see Fig. 17a. While the inner portions of the blade expe- tion of κu and κl requires the induced power values of the
rience downwash from the upper rotor (thereby reducing individual rotors when operating as a part of a coaxial ro-
the effective angle of attack), the outboard sections of the tor, as well as in an isolated condition. To estimate the in-
blade must operate in a stalled condition or in the non- duced power required by the isolated rotor to produce the
linear range of the cl − α curve producing lesser increase same thrust as the upper or lower rotor, momentum the-
in thrust for ∆α than in the linear range. Also the lower ory based curve fits to the measured data on single-rotor
rotor is very close to the upper rotor in region 3, the inflow systems were used. The three regions that were used to ex-
velocity from the lower rotor plays a much bigger role in plain the changes in FM with the axial separation distance
increasing the inflow angle of the upper rotor. The loss of are consistent with the observation made earlier. Power
lift by the lower rotor (because of stall), when combined loss incurred to the lower rotor predicted using momen-
with the experimental trim requiring constant total thrust, tum theory is also shown for comparison. Momentum
forces the collective of the upper rotor to be increased. theory assumes the upper rotor to be independent of the
When the collective for both the upper and the lower ro- lower rotor, consequently, κu = 1.
tors increases to produce the same total thrust, induced Figure 15 shows performance characteristics of coax-
power of the system increases. Consequently, the overall ial rotors using twisted blades for CT = 0.014. Only two
performance of the coaxial rotor decreased quickly. regions were needed to explain the changes in FM. For
Dingeldein expected similar results while questioning region 1, defined as z/D > 0.2, the upper, the lower and
the validity of the single-rotor theory proposed by Har- the coaxial rotor as a system did not show any change in
approaching 41% loss when the rotors are placed in the
same plane). As mentioned earlier, the twisted blades
more closely reflect the flow field assumptions made in the
momentum theory, such as uniform inflow. For z/D < 0.2,
the collective pitch angles remained unchanged. Yet, the
total performance decreased as shown in Fig. 15, which
is confirmed through the increased interference losses in
Fig. 16a. A comprehensive flow field analysis of coax-
ial rotors using highly twisted blades at various separa-
tion distances is needed to understand the observed per-
formance characteristics.

Tandem and Tilt-rotor


(a) Untwisted blades
As mentioned earlier, performance measurements from a
tandem/tilt-rotor were analyzed with the following spe-
cific goals: (1) Compare the performance of the tan-
dem/tilt rotor configurations against two isolated single-
rotor systems whose combined solidity is the same as the
tandem rotor system, (2) measure the induced power lost
because of overlap by gradually transitioning from tandem
configuration to coaxial configuration, (3) establish and
understand the performance characteristics of the front
and aft rotor relative to a single-rotor system during the
transition, and (4) compare and analyze the similarities
and differences between the untwisted and highly-twisted
blades for the aforementioned goals.
Nomenclature needs to be established clearly to distin-
guish the various configurations that were analyzed.
(b) Twisted blades - tilt-rotor
• Side-by-side rotor: Rotors operating in the same
Figure 18: Rotor performances at d/D = 1.1 and plane (z/D = 0) using untwisted and highly twisted
z/D = 0. xv-15 blades. When xv-15 blades are used, the con-
figuration resembles tilt-rotors

• Tandem rotor: Rotors operating at an axial separation


FM due to the variations in z/D. Region 3 was reduced
distance (z/D) of 0.07. The front rotor operated at a
to z/D < 0.2, where the performance of the coaxial ro-
lower plane when compared with the aft rotor. When
tor system decreased with decreasing axial separation dis-
d/D = 0 (i.e., coaxial rotors), the front rotor becomes
tance, a behavior that is similar to when untwisted blades
the lower rotor, and the aft rotor becomes the upper
were used.
rotor.
While the performance characteristics of coaxial rotors
using untwisted blades were relatively simple to under- Rotor Performance: Figure 18 shows the performance
stand from the fundamental principles, twisted blade be- of the side-by-side rotor and the tilt rotor configurations
havior was completely different. The upper and lower at a shaft-to-shaft distance (d/D) of 1.1. All the mea-
rotor collective pitch angles changed very little over the surements were made after balancing the torque between
entire z/D. However, thrust sharing clearly reflected the the two rotors. Isolated 3-bladed single-rotor measure-
absence of region 2 (exhibited by the untwisted blades). ments were also shown for comparison. Both the rotors
When z/D > 0.2, the ratio of thrust produced by the lower (front and aft), when operating as part of the side-by-
rotor to the upper rotor changed vert little and was inde- side configuration or as a tilt-rotor, did not deviate from
pendent of the axial separation distance. The values of the isolated single-rotor measurements. At this separa-
κint and κu reflected similar independence over the same tion distance, there was no measurable aerodynamic in-
region. The magnitude of κint measured by twisted blades terference between the two rotor systems that resulted in
remained closer to the values predicted by momentum the- any additional loss/gain in the induced power. The ro-
ory (26% loss when rotors were placed farther apart and tors essentially operated as isolated single-rotor systems.
Consequently, this means that the side-by-side and tilt ro-
tor configurations perform better than single-rotor systems
(Nb = 6) with equivalent solidity. These observations were
extrapolated to assume that both the front and the aft ro-
tors would continue to operate as single-rotor systems at
higher d/D. However, when d/D is reduced (overlapping
rotors), aerodynamic interaction between the two rotors
increases, garnering more interest from research perspec-
tive.
Effect of Overlap: The rotor systems used in the present
study were unsynchronized, therefore, overlap between
rotors (d/D < 1) operating in the same plane (z/D = 0)
was not possible. To study of the effect of overlap on the
system performance, one of the two rotors was displaced
(a) Performance
axially downward by 0.07D (the same distance at which
coaxial rotor performance measurements were made). All
measurements were made after satisfying two conditions:
(1) balanced torque between the two rotors, and (2) the
total system thrust (CT f +CTa ) was kept constant at 0.014
(same as coaxial rotor measurements). Again, perfor-
mance measurements of the untwisted and the twisted
blades were analyzed. Irrespective of the overlap, the disk
area of the rotor system was assumed as twice the area
of the front (or the aft) rotor. Following the procedure
used throughout this paper, untwisted blades were ana-
lyzed first, followed by twisted blades.
Torque balance between the rotors made the front and
the aft rotors to produce unequal thrust. Therefore, the fig-
ure of merit must be calculated using the same procedure (b) Power lost/gain from overlap
followed in the coaxial rotor analysis, which is given by
3/2 3/2 √ Figure 19: Effect of shaft-to-shaft distance on rotor
(CTa +CT f )/ 2 performance using untwisted blades.
FMTandem = (22)
CQa +CQ f
lost through aerodynamic interference between the two ro-
Results using untwisted blades are shown in Fig. 19a. The tors. Even though the system performance was similar be-
overlap was varied from 0.9 to about 1.025. The coaxial tween the tandem and the side-by-side configurations, the
rotor (when d/D = 0) and isolated 3-bladed single-rotor individual rotors (front and aft) of the two configurations
measurements are also shown to establish the trend and behaved completely differently. In the case of side-by-
to compare the performance of the individual rotors (front side rotor systems, the individual rotors mimicked the per-
and aft), respectively. The induced power loss factor re- formance characteristics of the isolated 3-bladed single-
sulting from overlap κov is given by rotor. However, in the case of tandem rotor systems, the
Pia + Pi f performance of the front (lower) rotor was better than the
κov = (23) aft (upper) rotor, such that the resulting performance of
2 × Pi(single)
the tandem rotor (as a system) equaled the performance of
where Pi(single) is the power required by the single-rotor two isolated single-rotors. More importantly, the perfor-
to produce half the total thrust as the tandem rotor. mance of the front (lower) rotor was better than the iso-
Figure 19b shows the estimated κov at various overlap lated 3-bladed single-rotor. One possible reason could be
distances, along with predictions using momentum the- that the tip vortices from the aft (upper) rotor counteract-
ory. Several interesting observations can be made when ing against the self-induced velocity of the lower rotor tip
Fig. 19a and 19b are analyzed together. vortices, resulting in a lower induced power of the front
The performance of the tandem rotor using untwisted rotor.
blades was at its best when d/D = 1, which matched the Nevertheless, the performance of the tandem rotor sys-
performance of the side-by-side configuration at d/D = tem decreased on either side of d/D = 1. As the shaft-
1.1. This resulted in κov = 1, i.e., no additional power was to-shaft distance increases (for d/D > 1.1), the individual
where the upper rotor appear to perform better than lower
rotor for any further increase in overlap. The crossover
point also correlates well with the location of the slip
stream boundary of the single-rotor (see Ref. 29) for the
z/D used in the present measurements.
Figure 20 shows measurements made using twisted
blades to quantify the effect of overlap on the tandem rotor
performance. Unlike when untwisted blades were used,
the performance of tandem rotor system using twisted
blades at d/D = 1.0 was lower than that of two isolated
single-rotors. Also, none of the individual rotors (front or
aft) performed better than the single-rotor system at any
d/D. However, similar to untwisted blades, the tandem
rotor system using twisted blades showed continuous re-
(a) Performance of tandem rotor duction in performance as overlap increased.
The front (lower) rotor performed better than the aft
(upper) rotor over half the overlap distance, i.e., until
d/D = 0.5, which corresponds to the condition where the
shaft of each rotor aligns with the blade tip of the other
rotor. The crossover in the individual rotor performance
occurred at d/D = 0.5, i.e., the aft (upper) rotor began
performing better than the front (lower) rotor. Although
unexpected, the observation is consistent with flow field
measurements shown in Ref. 29, which suggests that the
front (lower) rotor may operate in a relatively clean flow
and not in the downwash of the aft (upper) rotor. The
lower rotor stands to gain from the upwash produced by
the upper rotor tip vortices, counteracting the self-induced
downwash produced by lower rotor tip vortices. While
(b) Overlap power factor, κov this explains the higher performance of the front rotor over
the aft rotor when 0.8 < d/D < 1.0, no explanation can
Figure 20: Effect of shaft-to-shaft distance on rotor be given for the region 0.5 < d/D < 0.8, where down-
performance using twisted blades. wash from the upper rotor must adversely affect the lower
rotor performance. Only flow field measurements on sim-
rotors should begin performing like isolated rotors with ei- ilar rotor configurations can reveal the physics behind the
ther rotor having no influence on each other. This can be observed performance.
expected based on the κov measurements made at z/D = 0, Even though the performance crossover occurred at
shown in Fig. 19b. In the present study, experiments were d/D = 0.5, the performance of the aft (upper) rotor ap-
limited to d/D < 1.1 to prevent wall effects. peared uninfluenced by the front (lower) rotor from as
Increasing the overlap, i.e., reducing d/D, presents early as d/D = 0.75 until d/D = 0. The reduction in
even more interesting performance characteristics. The performance exhibited by the front (lower) rotor reduced
front (lower) rotor continued to outperform the aft (up- the overall tandem rotor performance, as explained by
per) rotor when 0.9 < d/D < 1.02. And, the tandem rotor the thrust and power ratios of the upper and lower rotors
system showed a overall reduction in performance. This shown in Fig. 21.
is because a reduction in d/D increases the disk loading The overlap power factor, κov , shown in Fig. 20b re-
(therefore, lower FM) for the same total thrust. The poor flects the observed performance characteristics. Two mo-
performance was also reflected by the increase in κov as mentum theory solutions were plotted, as well, where the
shown in Fig. 19b. The trend should continue until d/D solid line corresponds to the rotors operating in-plane and
reduces to zero (i.e., coaxial rotor). While the induced the dashed line corresponds to the front rotor operating at
power losses associated with overlap were expected based a large z/D relative to the aft rotor, i.e., at the axial lo-
on momentum theory, the measurements showed higher cation where the wake from the aft rotor contracts fully
losses than the theoretical prediction. to form vena contracta. Performance predicted by Har-
At d/D = 0.9, the measured FM of the upper and lower ris (Ref. 28) is plotted as well. All the theoretical so-
rotor were equal. This represents an interesting condition lutions underpredicted the measured losses, except when
to provide the baseline rotor characteristics. Multi-rotor
measurements were compared against single-rotor sys-
tems to quantify various kinds of interference losses, as
well as to measure the rotor-on-rotor influence. The fol-
lowing are the configuration-specific conclusions derived
from the present study.

1. Coaxial rotors were found to perform better than a


single-rotor with equivalent solidity. Nearly 9% im-
provement in FM was achieved using coaxial ro-
tors. However, when compared against two isolated
single-rotors whose combined solidity is equal to the
solidity of the coaxial rotors, aerodynamic interfer-
ence between the rotors resulted in approximately a
20% increase in the induced power for coaxial rotors.
Figure 21: Thrust and power ratio of the front and aft The latter result is independent of the blade planform
rotor for a torque-balanced tandem rotor system used. Momentum theory results suggests 10% and
26% for the two similar comparisons, respectively.
2. The upper rotor influenced the performance of the
lower rotor even when the axial separation distance
was large. The lower rotor required nearly 60% more
induced power than an isolated single-rotor of equal
solidity, when the rotors were 1.5D apart, to produce
the same thrust. The increase in the induced power
required by the lower rotor reduced with decreas-
ing axial separation distance. At 0.05D, lower rotor
needed only 37% more induced power than an iso-
lated single rotor of equivalent solidity.. The result
was similar for both blade planforms.
3. The adverse influence of the lower rotor on the upper
was measurable when the axial separation distance
Figure 22: Measurements to determine rotor-on-rotor
was less than 0.75D. The effect increased with the re-
influence factor in tandem rotor configuration
duction in the axial separation distance, as expected.
d/D is closer to 0, i.e., coaxial rotor condition. Figure 22 At 0.05D, the lower rotor caused the upper rotor to
compares the hovering efficiencies of the individual rotors consume 23% more induced power to produce the
of the tandem rotor system against isolated single-rotor. same thrust.
Unlike the effect of axial separation distance on the rotor-
4. Rotor axial separation distance had a different effect
on-rotor influence found in coaxial rotors, the effect of
on different performance parameters. The overall
overlap distance on the rotor-on-rotor influence factor in
performance of the coaxial rotor was independent of
tandem rotor configuration cannot be ascertained conclu-
the axial separation distance when z/D > 0.15. For
sively.
decreasing z/D, the performance of the coaxial rotor
started decreasing at higher thrust conditions because
of stall experienced by the lower rotor blades. At low
Conclusions thrust conditions, no stall was observed and so, no
performance penalty was encountered even when the
A series of experiments were conducted on an easily con- rotors were as close as 0.05D. This behavior was also
figurable rotor stand. Performance measurements were reflected in the interference power factor, which was
acquired for various rotor configurations, such as sin- constant until 0.15D. When z/D was reduced further,
gle, coaxial, tandem, and tilt-rotors using untwisted and the induced power started approaching higher values
highly-twisted blades. Within a configuration, rotor sys- set by momentum theory.
tem parameters were varied to conduct parametric stud-
ies. single-rotor measurements were used to establish the 5. Thrust sharing between the two rotors changed with
physical operating limits of the rotor stands, as well as axial separation distance, and with the blade plan-
form. The lower rotor produced only 70% of the References
upper rotor thrust when z/D was higher. However,
1 Knight, M., and Hefner, R. A., “Static Thrust Analysis
as z/D decreased, the lower rotor produced more
and more thrust, approaching 95% at 0.05D. While of the Lifting Airscrew,” NACA TN 626, December 1937.
twisted and untwisted blades behaved similarly at 2 Gessow, A., “Effect of Rotor-Blade Twist And Plan-
small and large z/D, the difference occurred for
0.3 < z/D < 0.75, where the changes to the thrust Form Taper on Helicopter Performance,” NACA TN
sharing was more gradual in the case of untwisted 1542, February 1948.
blades compared with the twisted blades. 3 Johnson, W., “Technology Drivers in the Development
of CAMRAD II, American Helicopter Society Aerome-
6. Side-by-side rotors and tilt-rotors showed no addi- chanics Specialists Conference, San Francisco, CA, Jan-
tional loss/gain from aerodynamic interference be- uary 19–21, 1994.
tween the rotors when the shaft-to-shaft distance
(d/D) was more than 1.1. The front and aft rotors 4 Datta, A., and Chopra, I., “Validation and Understand-
performed very similar to isolated rotors. ing of UH-60 Vibratory Loads in Steady Level Flight,”
Journal of the American Helicopter Society, Vol. 49,
7. Independent of the blade planform, the perfor- No. 3, July 2004, pp. 271–287.
mance of tandem rotors decreased with increasing 5 Wissink, A.M., Sankaran, V., Jayaraman, B., Datta,
overlap, resulting from increasing local disk load-
A., Sitaraman, J., Potsdam, M., Kamkar, S., Mavriplis,
ing. At d/D = 1, tandem rotors using untwisted
D., Yang, Z., Jain, R., Lim, J., and Strawn, R. C., “Ca-
blades showed no reduction in hover efficiency,
pability Enhancements in Version 3 of the Helios High-
however, twisted blades showed a 3% reduction.
Fidelity Rotorcraft Simulation Code,” AIAA-2012-0713,
In general, when the measured losses were com-
50th AIAA Aerospace Sciences Meeting and Exhibit,
pared against momentum theory predictions, mea-
Nashville, TN, January 2012.
surements showed more induced power than the pre-
dicted values. 6 Harrington, R. D., “Full-Scale Tunnel Investigation of

the Static-Thrust Performance of a Coaxial Helicopter Ro-


8. Overlap distance had a different effect on the front tor,” NACA 2318, March 1951.
and aft rotor. Surprisingly, the front rotor (which
7 Dingeldein,R. C., “Wind Tunnel Studies of the Per-
is lower relative to the aft rotor in terms of axial
position), operated at higher efficiency than the aft formance of Multirotor Configurations,” NACA TN 3236,
rotor when overlap increased. However, this be- August 1954.
havior changed beyond a threshold distance, which 8 Cheney Jr.,
was different for the two different planforms used M. C., “The ABC Helicopter,” AIAA/AHS
in the present study. Untwisted blades showed the VTOL Research, Design and Operations Meeting, At-
crossover in performance occurring at 0.9D; the lanta, GA, February 17–19 1969.
crossover for twisted blades was at 0.5D. The 9 Rorke, J. B., “Hover Performance Tests of Full Scale
crossover corresponds to the location where the up- Variable Geometry Rotors,” United Technologies Corpor-
per rotor begins performing better than the lower ro- tation 2713, Washington, D. C., August 1976.
tor for further increase in overlap. Only a flowfield
analysis can explain the observed performance char- 10 Arents, D. N.,
“An Assesment of the Hover Performace
acteristics. of the XH-59A Advancing Blade Concept Demonstration
Helicopter,” USAAMRDL TN-25, May 1977.
11 Nagashima, T., O. H., and Sasaki, F., “Optimum Per-
Acknowledgements formance and Load Sharing of Coaxial Rotor in Hover,”
Journal of Japan Society for Aeronautics & Space Sci-
The author gratefully acknowledges the discussions with ences, Vol. 26, No. 293, June 1978, pp. 325–333.
Drs. Wayne Johnson, Gloria Yamauchi, Hyeonsoo Yeo,
Mahendra Bhagwat, Joon Lim, and Mark Potsdam that 12 McAlister, K. W., Tung, C., Rand, O., Khromov, V.,
were helpful throughout the work. The support of Mr. and Wilson, J. S., “Experimental and Numerical Study of
Perry Kavros in setting up the experiment and keeping the a Model Coaxial Rotor,” Proceedings of the American He-
rotors operational through several months of rigorous test- licopter Society 62nd Annual National Forum, Phoenix,
ing was greatly appreciated. AZ, May 9–11 2006.
13 Bohorquez, F., “Rotor Hover Performance And System 26 Bender, G. L., Sullivan, P. J., Yamakawa, G. M., Rob-

Design Of An Efficient Coaxial Rotary Wing Micro Air bins, R. D., Herbst, M. K., and Williams, R. A., “Airwor-
Vehicle,” PhD thesis, University of Maryland, College thiness And Flight Characteristics Test of the CH-47D He-
Park, MD, 2007. licopter,” UAAAEFA AD-A151 280, Edwards Airforce
Base, CA, February 1984.
14 Lee, T. E., and Leishman, J. G., “Design and Perfor-
27 Felker, F. F., Signor, D. B., Young, L. A., and Betz-
mance of a Ducted Coaxial Rotor in Hover and Forward
Flight,” M. S. Thesis, Univeristy of Maryland, College ina, M. D., “Performance and Loads Data From a Hover
Park, MD, October 2010. Test of a 0.658-Scale V-22 Rotor and Wing,” NASA TM
89419, April 1987.
15 Coleman, C. P., “A Survey of Theoretical and Exper-
28 Harris,F. D., “Twin Rotor Hover Performance,” Jour-
imental Coaxial Rotor Aerodynamic Research,” NACA
TN 3675, Moffett Field, CA, March 1997. nal of the American Helicopter Society, Vol. 1, No. 1, Jan-
uary 1999, pp. 34–37.
16 Bourtsev, B. N., Kvokov, V. N., Vainstein, I. M., and 29 Ramasamy, M., Gold, N. P., and Bhagwat, M. J.,
Petrosian, E. A. “Phenomenaon of a Coaxial Helicopter
High Figure of Merit at Hover,” Twenty Third European “Flowfield Measurements to Understand Effects of Wake
Rotorcraft Forum, Dresden, Germany, September 1997. Behavior on Rotor Performance,” AIAA 2010-4237, 28th
AIAA Applied Aerodynamics Conference, Chicago, IL,
17 Johnson, W., “Influence of Lift Offset on Rotor Per- June 28–July 1 2010, AIAA.
formance,” Proceedings of the AHS Technical Specialists 30 Lim, J. W., McAlister. K. W., and Johnson, W., “Hover
Meeting on Aeromechanics, San Francisco, CA, January
Performance Correlation for Full-Scale and Model-Scale
23–25 2008.
Coaxial Rotors,” Journal of the American Helicopter So-
18 Lee, J., Chae, S., Oh, S., and Yee, K., “Paramet- ciety, Vol. 54, No. 3, July 2009, pp. 032005–1–032005–
ric Study for Hovering Performance of a Coaxial Rotor 14.
Unmanned Aerial Vehicle,” Journal of Aircraft, Vol. 47, 31 Leishman,J. G., Principles of Helicopter Aerodynam-
No. 5, September 2010, pp. 1517–1530.
ics, Cambridge University Press, New York, 2000.
19 Kim, H. W., and Brown, R., “A Comparison of Coax-
ial and Conventional Rotor Performance,” Journal of
the American Helicopter Society, Vol. 55, No. 1, January
2010, pp. 012004–1–012004–20.
20 Leishman, J. G., and Syal, M., “Figure of Merit Defi-
nition for Coaxial Rotors,” Journal of the American Heli-
copter Society, Vol. 53, No. 3, July 2008, pp. 290–300.
21 Wachspress, D. A., and Quackenbush, T. R., “Impact
of Rotor Design on Coaxial Rotor Performance,” Proceed-
ings of the American Helicopter Society 62nd Annual Na-
tional Forum, Phoenix, AZ, May 9–11, 2006.
22 Andrew, M. J., “Coaxial Rotor Aerodynamics in
Hover,” Vertica, Vol. 5, No. 2, 1981, pp. 163–172.
23 Heyson, H. H., “An Evaluation of Linearlized Vortex

Theory As Applied To Single And Multiple Rotors Hov-


ering In And Out Of Ground Effect,” NASA TN D-43,
September 1959.
24 Sweet,G. E., “Hovering Measurements For Twin-
Rotor Configurations with and without overlap,” NASA
TN D-534, Langley Field, VA, November 1960.
25 Stepniewski,W. Z., and Keys, C. N., Rotary-Wing
Aerodynamics, Dover, New York, NY, 1984.

You might also like