Human Powered Helicopter
Rotor Test Stand
Advisor: Prof. William Crossley
Jeff Decker
Santosh Kuruvilla
Fall 2001
Abstract:
A rotor blade test stand has been designed and built to test a 1:8 scale human powered helicopter rotor blade, designed previously using a blade element method developed for low speed rotorcraft. This paper is a description of the design, analysis methods and manufacture of the test stand. Experimental procedures that have been developed to test the rotor blades are discussed.
Historical Background:
In the
past three decades, enthusiasts all over the world have enjoyed prolific
results within human powered flight. The Gossamer Albatross and Daedalus
both made great advances for fix-wing aircraft, and riding on their success,
attention soon turned to rotorcraft.The American Helicopter Society (AHS)
introduced the Sikorsky Prize in 1980: $20,000 to the first group that
can demonstrate controlled human powered vertical flight. In 1989,
California Polytechnic State University’s Da Vinci III demonstrated the
Sikorsky Award was achievable, albeit distantly. California Polytechnic
State University, Massachusetts Institute of Technology, and Nihon University
in Japan all put forth amazing contributions, though none capable of capturing
the Sikorsky Prize. Despite over twenty years of effort with no successful
attempt at the flight mandated by the Sikorsky Prize, work continues on
the idea with modest progress.
The Purdue university chapter of the AHS has been looking into the design and construction of a human powered helicopter (HPH) in response to the Igor I. Sikorsky HPH competition. The MATLABR program, BLADE, that implements Prouty’s blade element method has been developed by John Jameson in order to design the rotor blades.1,2
Purdue students, Luiz Valentini
and Dwayne Bevis, continued work on the project by building a 1:8 scale
test model of the rotor designed by Jameson.Valentini and Bevis also built
a test stand for the rotor; however, this hydraulically powered stand was
constrained by cost and time, and therefore did not have the accuracy or
usability required.
The helicopter test stand that is detailed
in this paper is to verify the validity of BLADE for low Reynolds numbers
expected in the HPH flight regime, using the rotor built by Valentini and
Bevis. The test stand may also help in identifying the interactions
that may be present between the different variables.Figure 1 shows the
influence of collective on hover performance; this was done using BLADE
and this is the type of results expected from the data taken using the
stand.
Figure 1. Influence of collective on hover using BLADE
Design Process:
To begin, the students brainstormed ideas and created various concepts. Figures 2 and 3 are the initial design sketches.
Each concept was
evaluated considering its practical issues: collecting accurate data, eliminating
unnecessary motion and friction, ease of manufacturing and portability.
The decision was made to take every effort to use high-quality materials
and instruments in order to collect precise and accurate data. However,
because the stand would be entirely student built, a simple structure the
better was deemed more feasible. The emergent design, represented
below in Figures 4 and 5, show its major characteristics. The motor,
which will spin the rotor, sits on a plate that rests on a load cell.
As the rotor is spun, thrust will be measured by the difference in apparent
weight of the motor acting on a load cell.
At this point in the design evolution, the students consulted several individuals to help trouble shoot the design. A few weeks time was spent gathering information on load cells, torque cells, and thin beam load cells, though their use was doubtful given the high acquisition cost. An alternative to costly load cells was needed, and students searched for a cost effective, relatively small, and efficient force measurement device. Fortunately, reallocation of departmental equipment provided the group with four Futek load cells.
To streamline the design and re-design iteration process, the group elected to learn new CAD software. A relatively small amount of time was devoted to learning IronCAD software. The use of a solid model assisted greatly in anticipating construction difficulties and their solutions. Throughout the design evolution, the solid modeling package allowed quick and easy adjustments to be made to the structure, thus actually saving time.
After modeling the test stand in
IronCAD, the group discussed the design’s weakness with Dr. William Crossley
a professor of design and Dr. John Sullivan a professor of experimental
aerodynamics. The outcome of these discussions was to eliminate the
“turntable” type interface and move to an arrangement with flexures.
The flexures eliminated friction between moving parts and zero-shift/ hysterisis
effects, while the turntable would have certainly shown frictional and
zero-shift errors. Also, it was decided that the test stand should
be able to vary the height of the rotor above the ground, in order to study
the effects of the ground on the thrust and torque. Additionally,
three load cells support the motor instead of one with guide shafts.
IronCAD was used to remodel the test stand once again, and the final iteration
is shown is Figures 6 and 7 below.
The
flexure size and material is of obvious importance. The group was
advised to use stainless steel, as it’s deflection and deformation properties
are well suited to the use. Hand calculations were performed to estimate
the initial dimension of the flexure. However, due to the complicated
combination of buckling and twisting loads exerted on the flexure, COSMOS
Design STAR, a Finite Element Analysis (FEA) tool, was used to verify the
students’ work. This particular FEA program worked very well in conjunction
with the IronCAD software and results were quickly obtained. A more
detailed explanation of the FEA is given in the following section.
Structural Analysis:
As mentioned above the
structural analyses of the flexures were done using the finite element
analysis program COSMOS Design Star. Different combinations of flexure
thickness and number of flexures were analyzed. This was done in order
to be able to determine the minimum thickness and number of flexures that
would allow a large enough deflection, which would allow the torque load
cell to deflect without affecting its output and still prevent the structure
from buckling. To check the validity of the program, a basic beam was analyzed
in buckling and the output values compared to standard expected values.
These values correlated with each other. The FEA program allows the input
of Parasolid files without any modification; the mesh generated automatically
by the program is a solid standard type mesh and Jacobian check was the
four-point rule. The resolution of the mesh had to be adjusted to allow
a proper mesh formation around edges and corners; the bolts that were modeled
in the CAD program were removed to allow COSMOS to run faster. The model
restraints were setup as fixed and the touching faces were set up as bonded.
Figure 8 shows the base plates and flexures with the restraints and loads.
Figure 8. Boundary conditions applied to the base plate
The maximum displacement
is about 6e-4in if 1/8-in flexures were used. This was not enough of a
displacement to be able to get a reading from the load cell, so the thickness
of the flexures was gradually reduced and a 20-gage stainless flexure was
deemed to be enough.
Figure 11. Plot of Static displacement, 50lb load.
Manufacturing Process:
This project was completed
entirely at Purdue University’s facilities, and all parts—except the rotor
hub—were student made. The fabrication processes were kept simple
and straight foreword in order to keep construction problems to a minimum.
All tools, machines, and equipment used were basic and relatively simple.
The majority of fabrication took place on an ordinary drill press and mill.
After the design was finalized, IronCAD was used to make mechanical drawings
for most major parts. These drawings are attached in appendix A. The test
stand was constructed in three primary sub-assemblies: the legs, the disks
and flexures, and the motor plate.
As one of the easier and
less intimidating parts to manufacture, the legs’ fabrication began first.
The method used to get the legs’ into the designed shape did not give the
precision required, but was easy to implement. This caused improper
positioning of the “feet” that contact floor. Each foot was canted
approximately 15? from the floor, such that the stand was very unstable
and susceptible to vibration. However, the feet were made from 1/8”
plate steel that was easily bent and hammered into a horizontal position.
The two disks and flexure
assembly presented the most challenge in terms of manufacturing difficulty.
Over 140 holes were drilled into the two plates. With the shear number
of holes being mated together, tolerances were quickly found to be extremely
tight, and due to the monotony and repetition, it was very easy to let
the precision drop and tolerances exceed limits. Alignment was sometimes
sacrificed in order to get these parts to fit. Flexure size was adjusted
after a test piece was constructed and predicted loads were applied.
Behavior was not as expected, mandating a modification. Arches were
cut in the flexures to reduce its’ moment of inertia, this immediately
improved the behavior of the flexure. Figures 12 and 13 show some of the
manufacturing difficulties encountered.
Figure 12. Multiple holes being drilled into the base plate.
Figure 13. Flexure deflecting under load
The
motor plate was one of the simpler assemblies, and therefore had fewer
tolerance issues than the rest of the assemblies. However, the configuration
still presented a few problems with mating parts together. This assembly
was simple enough that the problem areas could be corrected by greatly
reducing tolerances with no overall adverse effects.
In general, the test stand’s
physical dimensions are exactly as planned. It looks and acts precisely
like what was wanted. However, the fabrication process broke down
a few times, making the structure a little different than the design.
Most of the imperfections are due solely to the lain experience of the
students who where building the test stand. Problems areas were dealt
with easily so that the test stand is not significantly different from
what was designed.
Experiment Setup
and Test Procedure:
Steps for the experimental setup and calibration are in appendix B.
Taguchi’s design of experiments
is used to develop an experimental procedure for the rotor blade tests.
This allows a psudo-optimal solution and tests interactions between the
three factors varied by the stand. Collective, chord to height ratio and
rotation rate are the factors varied by the stand and thrust/torque is
the result that is tested. An L27 array is chosen, since it allows the
analysis of three factors and interactions between the three factors. The
L27 array also uses three levels, this allows the interactions to be better
noted and since the array has thirteen positions for factors, if the blades
are changed, they could be used as another factor.
The levels of the factors
were chosen based on historical values used for testing collective, and
z/d.1 The rotation rate was based on max h.p., which is constrained by
what a human can produce. Table 1 is the list of factors and levels. Analysis
and performance calculations’ can be done using the method in reference
3.
|
Factor
|
L1
|
L2
|
L3
|
|
Collective (Deg)
|
0
|
6
|
12
|
|
z/D
|
0.15
|
.25
|
0.375
|
|
? (rad/sec)
|
0.314 (3 RPM)
|
0.5234 (5 RPM)
|
1.256 (12 RPM)
|
Summary
and Conclusion:
The objective of this semester’s AAE490HPH class was to design and construct a test stand for a 1:8 scale rotor. The stand was to measure both thrust and torque, while also varying the rotational speed of the rotor. Furthermore, the test stand should have the ability to vary the height of the rotor above the ground. Beginning with previous AHS’s members’ work, a design was chosen based on constraints and intentions. The stand underwent several iterations and a finite element analysis tool was used to analyze the stand’s properties. Students built the stand with ordinary shop tools at Purdue University’s facilities. Though the manufacturing process presented a few difficulties, all the flaws were worked out. A test procedure is suggested, for the analysis of the rotor. A LabView data acquisition VI written for the stand not been tested yet, but will be shortly.
Future work includes the actual testing of the rotor to verify BLADE and produce a trade study similar to Figure 13. Monday, December 17, 2001, will see the 1:8 scale rotor spun, and the human powered helicopter will be one small step closer.
Figure 13. Factors affecting Hovering performance
Experimental Helicopter Rotor Procedure / Checklist
- Assemble Legs
- Mount Motor to Upper Load plate
- Attach Hub to Shaft
- Attach Rotor to Hub
- Adjust AOA to 0 deg setting
- Mount legs to base plate
- Position stand, secure legs to ground
- Connect load cells to D/A board
- Visually check load cells
- Check Controller is in off Position RPM = 0
- Check For obstructions and people (Min 5 ft from blade tip)
- Plug motor into electric socket
- Turn on motor increase to 5 RPM for 1 min
- Turn off motor
- Disconnect from electric socket
- Check bolts to hub and AOA control rod
- Set AOA Tighten Hub/AOA
- Check Controller is in off Position RPM = 0
- Check For obstructions and people (Min 5 ft from blade tip)
- Zero Balance
- Plug motor into electric socket
- Turn on motor to specified RPM, then wait 1 min for rpm to stabilize, Take Data for 30 Sec MIN (at least one every 5 Sec)
- Follow Taguchi combinations specified in Report.
|
|
|
|
|