Академический Документы
Профессиональный Документы
Культура Документы
Composite materials have been increasingly considered for use in rotorcraft drive
system components due to their high strength to weight ratio, providing a means for significant
weight reductions while maintaining overall strength. This was originally considered for large
components like housings (Ref. 1) and drive shafts (Ref. 2). More recently, this has been
extended to gearing with the hybrid (steel-composite) gear, which replaces the steel in the gear
web with a lightweight carbon fiber composite. Original studies (Refs. 3,4) were conducted on
a 88.9 mm (3.50 in) pitch diameter spur gear that featured a design where a hexagonal void
was machined out of the gear web leaving the metallic tooth ring and hub.
The void was replaced by the carbon fiber composite. These studies focused on vibration
characterization and endurance testing of the hybrid gear. Later studies using this design
focused on multi-scale modeling of the composite material (Ref. 5) and loss-of-lubrication
performance of the gear (Ref. 6). The loss-of-lubrication tests showed failure of the composite
material at stress concentrations that exist at the corners of the hexagonal web. These initial
studies were conducted to determine the feasibility of the hybrid gear concept and a design
optimization was not performed.
The hybrid gear concept was extended to a 419 mm (16.5 in) aerospace bull gear (Ref. 7)
which has a greater potential for weight reduction since a greater amount of steel can be
replaced by composite material. This design features a sinusoidal interlock at the
compositesteel interfaces between the composite web and two steel adapters. This same design
was modified using a variable thickness web to reduce the amount of composite material used
in the gear (Ref. 8). A conceptual design analysis was conducted on the bull gear primarily to
maximize weight savings but also to design for redundancy and manufacturability (Ref. 9). The
study featured more than twelve different design concepts focusing on the most promising
designs. Many of the studies conducted on either the 88.9 mm or 419 mm gears have utilized
designs that feature inner and outer metallic components with a composite web in between.
Ideally, a composite web mated directly to an integral composite shaft would provide the
greatest weight savings. Darmstadt and Robuck (Ref. 10), conducted a systems level analysis
of the use of composites in advanced drive systems.
This study used various designs of composite shafting and hybrid gears to assess
transmission performance with the introduction of composite materials. The objective of the
current research is to develop an optimization technique for hybrid gear interlock design
seeking to minimize both the gear weight and the shearing forces that must be supported by
adhesive at the composite-metallic interface. For the sake of simplicity, this optimization
approach is applied to an 88.9 mm (3.5 in) pitch diameter spur gear here, but will be applied to
more representative rotorcraft gears in the future. To minimize stress concentrations at the
composite-metallic interface and ensure that most of the torque transmission from the teeth
through the gear to the shaft is supported by normal forces at the interface (i.e. not supported
by the adhesive), a sinusoidal interlock design is considered. Also, in order to design for
maximum weight savings, the design allows for an integral composite shaft rather than an inner
composite-metallic interface.
COMMUNICATION
PLANNING
MODELING
CONSTRUCTION
DEPLOY
Process planning
Raw material planning
Force analysis
Process scheduling
1.2.5 Deployment
Gears are mechanical components used for transmitting motion and torque from one
shaft to another. Ever since invention of rotating machines, gears existed. Early records states
that around 2600 BC Chinese used gears to measure the speeds of chariots. In 250 B.C
Archimedes used a screw to drive toothed wheels which were used in engines of war. In 4
century B.C., Aristotle used gears to simulate astronomical ratios. Greek and Roman literatures
mention the extensive use of gears in clocks of cathedrals and ecclesiastical buildings [1].
During early centuries gears were made of either wood or stone teeth set in wood. Later
during metal ages Iron, Bronze or tin were used instead of stone. There was no standard
procedure for gear manufacturing until 1835 when English inventor Whitworth patented the
first gear hobbing process [2].
The Pfauter of Germany patented the first gear hobbing machine capable of cutting both
spur and helical gearing in 1897, they introduced the first NC hobbing machine and in 1975
and they introduced the first all 6 axis gear hobbing machine in 1982.
Although gear manufacturing has achieved lots of advancement during its evolution, however
the failure of gear due to bending and contact stress still remained a challenge for designers
and manufacturers until 1892. In 1892 the Philadelphia Engineers club first recognized Wilfred
Lewis presentation of stresses on the gear tooth and it still serves as the basis to determine the
gear stress [3]. The Lewis bending equation has a lot of draw backs which include
3. The location of application of load is not true as the load is shared by the tooth.
The proposed design of the spur gear features a sinusoidal interlock and an integral shaft.
There are four separate components that make up the design: a web composite, two outer
composite components and a metallic outer tooth ring. This design can be seen in Figure 1.
Notice that the integral shaft is not shown in the figure. Instead during the optimization this
diameter remained fixed and zero displacement boundary conditions were enforced at this
surface.
The effects of loading on the gear teeth are not of interest here so the geometry for
optimization was replaced by a gear blank with an outer diameter equal to the pitch diameter
of the gear. This may change the magnitude of the stresses at the interface and may affect the
magnitude of the mass, but the simplified gear blank provides a test article for developing the
optimization technique. Future work will investigate the effect of gear teeth on the optimal
solutions. Sinusoidal Interlock The sinusoidal interlock that defines the compositemetallic
interface is defined analytically by Equation 1, where r is the radial coordinate, Rc is the radius
of the centerline, A is the amplitude of the curve, Nc is the number of cycles in the sinusoid, θ
is the angle and φ is a phase shift which will rotate the curve in-plane relative to the location
of load application.
Several of the design parameters of the gear remained fixed. These include the shaft
diameter (Ds), the pitch diameter (Dp) and the height of the components (H). The specific
values for these parameters can be seen. The design parameters for the interlock that were
varied during optimization are the radius of the centerline of the sinusoidal curve (Rc), the
amplitude of the curve (A) and the number of cycles in the sinusoid (Nc).
The number of cycles must be an integer value. However, Borg needs to define non-
integer values in the given range. To get around this, the values output by Borg are rounded to
the nearest integer for use in the objective functions. One final design parameter that was varied
was the outer diameter of the outer composite (Do). The range of possible values for these
variables, as well as the constraints on some parameters can be seen. The range for Rc and Do
were chosen based on values that are physically possible with the given constant values of Ds
and Dp. The range of possible values for A were chosen based on the maximum possible value
if Rc was half the difference between Dp and Ds . The range of values for Nc were chosen
arbitrarily though it is expected that fewer than 5 cycles is not practical. An example design
configuration using 10 cycles of a sine wave can be seen. The dedendum used in the two
constraints was 2.870 mm (.113 in) and was calculated by subtracting the addendum from the
whole depth for this particular gear, provided in (Ref. 3). to an all steel gear. The shear tractions
at the interface are of interest since adhesives like those that could be used at the composite-
steel interface would likely fail in shear.
This is a problem for rotating components that can have high in-plane shear tractions at the
interface. By minimizing this traction it will be possible to prevent adhesive failure at the
interface. A summary of the objectives is provided in Table 3. The values of ε are important
parameters and were chosen based on the expected range of masses and shear traction results.
The importance of the ε values is discussed in the section on optimization.
Material Properties
The material properties of the metal and composite are both important parameters and
remained fixed throughout the optimization. The metal used was 9310 steel and the assumed
properties are given in Table 4. A tri-axially braided composite laminate, modeled as
orthotropic, was used for the optimization.
This is the same material used in the original design of the gear (Ref. 3). The orthotropic
properties of the individual composite lamina were determined using a multi-scale modeling
approach (Ref. 5). These properties are shown in Table 5. The in-plane lamina properties are
heavily influenced by the carbon fiber’s high longitudinal properties. The transverse properties
of the fiber are much lower, therefore the resin matrix has a greater effect on the out of plane
lamina properties. This accounts for the large difference between the lamina elastic and shear
moduli. An arbitrary layup of [60/0/- 60]2s was used for this study resulting in a quasi-isotropic
laminate. Effective in-plane laminate properties were determined for the optimization using
classical laminated plate theory. The effective out of plane properties were assumed to be the
same as the lamina properties. The effective laminate properties are also shown
Optimization
As mentioned before, Borg MOEA was used to perform the design optimization. Borg is
an auto-adaptive multiobjective evolutionary algorithm, developed at The Pennsylvania State
University (Ref. 11). Borg iterates through combinations of values for the decision variables,
which for this study are the design parameters, and uses these values in the objective functions
to calculate the values of the objectives. As already stated the objectives for this study are to
minimize the mass of the gear and the interfacial shear traction. Borg compares the values of
the objectives from each combination of design parameters to determine a Pareto-optimal set
of designs for the spur gear. Borg also has the option to check constraints that may need to be
enforced on the possible decision variables.
The two constraints used for this study are shown in Table 2. These constraints are in
place to ensure that combinations of the design variables do not produce geometry that is not
possible and that cause the finite element analysis to fail. If the constraints are not satisfied the
combination of design parameters are infeasible and are not considered for the Pareto-optimal
set. Borg has the ability to be run in parallel testing several cases simultaneously on different
processors and returning the results to a master processor. This capability was utilized for the
optimization and the analysis was run on one compute node with 48 processors and a total
memory of 256 GB.
Objective Functions
This section describes the two objective functions in detail. Both objective functions were
coded in C++ for easy integration with Borg and the in-house finite element solver used. The
mass objective function uses the geometry of each component to calculate the volume. The
geometry is created using the design constants in Table 1 and Borg’s values for the design
variables in Table 2. The volume of the two outer composites is calculated as a hollow cylinder
using the shaft (Ds) and outer composite (Do) diameters and the height of the components (H).
The volume of the web composite is approximated as a hollow cylinder using the shaft
diameter, the radius of the centerline (Rc) and the height. The volume of the tooth ring is
approximated as a hollow cylinder using the radius of the centerline, the pitch diameter (Dp)
and the height. The volumes of each component are then multiplied by the density of the
respective material and the results summed to give the total mass of the gear.
CHAPTER 4
FINITE ELEMENT ANALYSIS
The first step in any FEM is to divide the given continuum in to smaller region called
element. The type of elements has to be taken depending on type of analysis carried out like
one dimensional, two dimensional, and three dimensional.
This step involves determining of global stiffness matrix. This is done by using the
compatibility conditions at the nodes. The displacement of a particular node must be the same
for every element connected to it. The externally applied loads must also be balanced by the
forces on the elements at these nodes.
1) Elimination method
2) Penalty method
3) Multi constraint method
These boundary conditions are incorporated into the system of linear algebraic
equations, which can then be solved to obtain a unique solution for the displacements at each
node.
After solving the global equations, displacements at all the nodal points are determined.
From the displacement values, the element strains can be obtained from the stress-strains
relations. In FE formulation only the displacements are the independent variables, that is,
forces, strains and stresses are obtained from the displacements
Convergence is a process of refining mesh, as the mesh is refined, the finite element
solution approach the analytical solution of the mathematical model. This attribute is obviously
necessary to increase the confidence in FEM results from the standpoint of mathematics.
The fundamental premise of FEM is that as number of elements (mesh density) is
increased, the solution gets closer and closure, however solution time and compute resources
required also increases dramatically as we increases the number of elements to the true solution.
The objective of analysis decides how to mesh the given geometry, if we are interested in
getting accurate stress; a fine mesh is needed, omitting geometric details at the location we
needed. If we are interested in deflection results, relatively course mesh is sufficient.
There are two convergence studies, h-convergence study and p-convergence study
h- Convergence study is done by increasing number of elements which can be done by making
mesh size finer, and it is important to maintain continuity in meshing and element check should
be done for aspect ratio, warping angle, skew ratio and others The elements must have enough
approximation power to capture the analytical solution in the limit of a mesh refinement
process. p- Convergence study is done by increasing number of nodes.
Meshing of a given model will be done depending on geometry of the model, it is better
to have more degrees of freedom hence more number of elements so that results obtained will
be closure to analytical results. In two bay panel analyses, crack region is meshed with more
number of elements when compared with other parts of fuselage, for obtaining a converged
solution which in turn a better solution.
Structural analysis is probably the most common application of the finite element
method. The term structural implies not only civil engineering structures such as bridges and
buildings, but also naval, aeronautical, and mechanical structures such as ship hulls, aircraft
bodies, and machine housings, as well as mechanical components such as pistons, machine
parts, and tools.
The FE analysis used for the major part of this work is static analysis which involves both
linear and nonlinear structural analysis. Hence more prominence is imparted on Linear and
nonlinear analysis in further sections.
In many engineering problems, the behavior of the structure may depend on the load history
or may result in large deformations beyond the elastic limit. The assumptions/ features in
nonlinear analysis are:
Where [K] is the linear elastic stiffness. When the above assumptions are not valid, one
performs nonlinear analysis.
Geometric nonlinearity
Geometric nonlinearity occurs when the changes in the geometry of a structure due to
its displacement under load are taken into account in analyzing its behavior. In geometric
nonlinearity, the equilibrium equations take into account the deformed shape. As a
consequence of this, the strain-displacement relations may have to be redefined to take into
account the current (updated) deformed shape. That is, the stiffness [K] is a function of the
displacements {u}.
1) Large strain assumes that the strains are no longer infinitesimal (they are finite). Shape
changes (e.g. area, thickness, etc.) are also accounted for. Deflections and rotations may
be arbitrarily large.
2) Large rotation assumes that the rotations are large but the mechanical strains (those
that cause stresses) are evaluated using linearized expressions. The structure is assumed
not to change shape except for rigid body motions.
4) Spin softening: The vibration of a spinning body will cause relative circumferential
motions, which will change the direction of the centrifugal load which, in turn, will tend
to destabilize the structure. As a small deflection analysis cannot directly account for
changes in geometry, the effect can be accounted for by an adjustment of the stiffness
matrix, called spin softening.
SOLID45 is used for the 3-D modeling of solid structures. The element is defined by
eight nodes having three degrees of freedom at each node: translations in the nodal x, y, and z
directions.
The element has plasticity, creep, swelling, stress stiffening, large deflection, and large
strain capabilities. A reduced integration option with hourglass control is available. Fig( 4)
SOLID45 Geometry
The objectives are to minimize the gear mass and the interfacial shear tractions. Borg
MOEA was used to perform the optimization and determine the Pareto-optimal set of solutions.
The optimization was run for 500,000 function evaluations and resulted in a set of six optimal
designs. The optimization results show that changes can be made to the ranges of some of the
design variables for future analyses.
REFERENCES
1. Cecil, T. M., Ehinger, R. T., and Kilmain, C., “Application and Configuration Issues of
Resin Transfer Molded Composite Transmission Housings - A Program Overview,”
American Helicopter Society 63rd Annual Forum, Virginia Beach, VA, May 2007.
2. Lin, S. and Poster, S., “Development of a Braided Composite Drive Shaft with Captured
End Fittings,” American Helicopter Society 60th Annual Forum, Baltimore, MD, June
2004.
3. Handschuh, R. F., Roberts Gary, D., Sinnamon, R., Stringer, D. B., Dykas, B. D., and
Kohlman, L. W., “Hybrid Gear Preliminary Results-Application of Composites to
Dynamic Mechanical Components,” American Helicopter Society 68th Annual Forum,
Fort Worth, TX, May 2012.
4. Handschuh, R. F., LaBerge, K. E., DeLuca, S., and Pelagalli, R., “Vibration and
Operational Characteristics of a Composite-Steel (Hybrid) Gear,” NASA TM 216646,
2014.
5. Catera, P. G., Gagliardi, F., Mundo, D., De Napoli, L., Matveeva, A., and Farkas, L.,
“Multi-Scale Modeling of Triaxial Braided Composites for FE-Based Modal Analysis
of Hybrid Metal-Composite Gears,” Composite Structures, Vol. 182, September 2017,
pp. 116–123.
6. Laberge, K. E., Berkebile, S. P., Handschuh, R. F., and Roberts, G. D., “Hybrid Gear
Performance Under Lossof-Lubrication Conditions,” American Helicopter Society
73rd Annual Forum, Fort Worth, TX, May 2017.
7. Laberge, K. E., Handschuh, R. F., Roberts, G., and Thorp, S., “Performance
Investigation of a Full-Scale Hybrid Composite Bull Gear,” American Helicopter
Society 72nd Annual Forum, West Palm Beach, FL, May 2016.
8. Laberge, K. E., Johnston, J. P., Handschuh, R. F., and Roberts, G. D., “Evaluation of a
Variable Thickness Hybrid Composite Bull Gear,” American Helicopter Society 74rd
Annual Forum, Phoenix, AZ, May 2018.
9. Morales, G., Anderson, C., and Chavez, A., “Conceptual Design and Analysis of
Hybrid Composite Power Gearing in a Fielded Drive System Configuration,” American
Helicopter Society 74th Annual Forum, Phoenix, AZ, May 2018.
10. Darmstadt, P. R. and Robuck, M., “Composites for Advanced Drive Systems, a Systems
AnalysisRevolutionary Vertical Lift Technology (RVLT),” American Helicopter
Society 74th Annual Forum, Phoenix, AZ, May 2018.
11. Hadka, D. and Reed, P., “Borg: An Auto-Adaptive Many-Objective Evolutionary
Computing Framework,” Evolutionary Computation, Vol. 21, (2), 2013, pp. 231– 259.
12. Campbell, R. L., Fluid Structure Interaction and Design Simulations for Flexible
Turbomachinery, Ph.D. thesis, The Pennsylvania State University, University Park, PA,
USA, August 2010.