An Attempt at Complete Assembly Contact Analysis of A High Precision Reduction Gear
An Attempt at Complete Assembly Contact Analysis of A High Precision Reduction Gear
An Attempt at Complete Assembly Contact Analysis of A High Precision Reduction Gear
1. Introduction
A high precision reduction gear is a complex mechanism that achieves a large reduction ratio by
means of precession motion driven by plural gears accompanied by rolling and sliding on many
pins inside a hub. In recent years, the optimum design for strength and durability has been
accompanied by demands for higher loads and further space savings. To achieve this optimum
design requires stress analysis that accounts for the deformation of all parts in the assembly. The
problems with this kind of analysis are the large number of elements required for a complete
model and the treatment of contact between the many bearings, pins and gear teeth. Two kinds of
approaches were employed: a simplified model using spring elements at contact surfaces, and a
more rigorous full scale contact model.
3. Analysis
3.1 Models
3.1.1 The spring model.
This model is composed of six kinds of parts and is shown in Figure 3(a).
Spring elements were used in the bearings parts and in the pins between the gears and hub.
All six degrees of freedom were restrained in the flange part and rotation was restrained in
the cover side face of the Crankshaft. Output torque was set in the output side of crankshaft
(Setting the output side torque confirms the existence of input side torque)
Spring rigidity was calculated from analysis results on an individual part (Figure 4.) With
the exception of the springs between the gears and the hub, each spring was set to be
effective only in the direction radiating from the central axis of rotation of the part.
Both hexahedron and triangular pyramid elements were used. (Figure 5.)
3.2 Software
ABAQUS version 6.6 was used as solver and ABAQUS/CAE version 6.6 was used in both the pre
and the post-processing.
Hub
Spring element
(Gear/hub/pin)
Rotational degree of
freedom restrained
Torque
Shaft
Cover
Flange
(a) Spring model. (Spring elements were used for all contact surfaces.)
All degrees of freedom restrained
Contact
Rotational degree of
freedom restrained
Torque
Needle
Pin
(b) Contact model. (Contact elements were used in parts of the spring model.)
Figure 3. Simulation models. (Boundary conditions.)
Load
Rigid surface
Figure 4. Typical result of the FEA to determine the rigidity (elasticity) of a pin.
Contact model
Number of elements
240,000
650,000
Number of nodes
610,000
1,250,000
4. Results
4.1 Comparison between theoretical calculation and spring model result
Pin load values from a theoretical calculation of an individual pin were compared with those from
the FEM spring model analysis. The FEM analysis, which took into account the elasticity of all
parts, gave a greater pin load variation around the hub than did the theoretical calculation, which
was based on a rigid model. However, the total reaction force (given in Table 2.) shows that,
although the distribution of pin load differs markedly between analysis and calculation, the
difference in the total pin load is only about 5%.
Theoretical calculation
of the rigid model
0.95
4.2 Comparison between the spring model and the contact model
Overall appearance
The calculation results of stress analysis on the reduction gear assembly are shown in
Figure 6, and clearly reveal the load transmitted to the external face of the outer cover. In
addition, the results with the cover and the flange removed are shown in Figure 7. In the
spring model, the springs are displayed as white in the red circles. In the contact model,
springs are replaced by contact elements at the needles and pins and these are shown on the
right of Figure 7. Stress can be observed as a result of load transmitted from the crankshafts
to the needles and the gears. In the spring model, it occurs through springs. In the contact
model, it occurs through contact between each part. These results are as expected and show
that the calculations finished successfully. Figure 8 shows an enlarged view around the
gear/pin/hub mating part. As a result of the variation from pin to pin mentioned in 4.1, the
load transmission at different mating points has a wide distribution.
Spring model
Figure 6.
Contact model
Spring
Spring model
Figure 7.
Contact
Contact model
Stress contour plot for each model (with cover and flange removed).
Spring model
Figure 8.
Contact model
contact model results show that contact stress varies widely and continuously, but the peak
stress occurs at the edge. This follows from the crankshaft expressing the effect of bending
as a result of the reaction force. In the needle itself, the highest stress occurs at the edge of
the contact part.
Output torque comparison
As mentioned in 3.1.1, this analysis calculated input torque. The output torque to input
torque ratio was then evaluated and used to convert actual output torque from the driving
motor to output torque of the reduction gear assembly. This enabled a comparison between
output torque derived from a theoretical calculation, and that from both the spring and
contact models. The results of this comparison are shown in Table 3 as the ratio of output
torque to theoretical result. Compared to the theoretical value, output torque is 10% lower
in the spring model and 10% higher in the contact model. (The measured output torque,
including friction, is about 10% less than the theoretical value)
Spring model
Contact model
0.90
1.10
Calculation time
(hours)
Spring model
Contact model
3.9
21.8
5. Discussion
Pin load
The difference between the pin load value obtained from the spring model analysis and that
from the theoretical calculation is only about 5%. Despite load transmission occurring
through many complex parts a good degree of accuracy is achieved. Furthermore, although
the load distribution obtained from analysis does not closely fit that from the theoretical
calculation, this is to be expected as the deformation of the hub that supports the pin is not
considered in the calculation but is accounted for in the analysis models. In addition, the
results of gear and crankshaft deformation reflect the direction of the individual reaction
forces in the hub, the pins and the gear
Output torque
As the measured output torque including friction is about 90% of the theoretical value, it
follows that the influence of the deformation of parts on torque appears extremely small.
This result shows that the torque can easily be evaluated within an error margin of about
10%. The reason for the output torque value from the spring model being less than the
theoretical value is the spring elements inability to completely match the actual direction
of the load. In the contact model, the course nature of the mesh around the needles and pins
might be responsible for an output torque value greater than the theoretical value.
As mentioned above, the deformations and reaction forces between contacting parts can be
calculated to a relatively high degree of accuracy with both the spring and contact models. This
follows from the fact that the deformation of the crankshafts and the hub that cause the reaction
forces can be calculated even by the spring model. Therefore, the spring model can be used for the
calculation of stress and deformation in parts such as the pins, crankshafts and gears.
However, the contact condition in the needles and the pins cannot be evaluated in the spring model.
The contact model overcomes this limitation, as using this model calculates the precise edge
contact condition caused by the bending of a crankshaft and hence enables evaluation of the
maximum local pressure on each of the contact parts. This now provides a means for determining
the rolling motion fatigue strength in the sliding parts such as the needles and pins. This important
value cannot be evaluated by the rigid body model used in standard design calculations. The mesh
around the pins and needles is still a little coarse in the contact model, and so to calculate an even
more accurate load distribution and maximum local pressure the mesh in the contact areas should
be further refined. However, an increase in the number of elements is difficult without excessively
increasing analysis time and cost and the problem of finding an optimum distribution of the
element density between contact and non contact parts needs to be addressed.
6. Conclusions
i. As a result of this complete assembly analysis of a precision reduction gear, the output
torque can be evaluated within an error margin of about 10% and the reaction force in the
sliding surfaces through precession movement can be determined to within 5%.
ii. Evaluation of the deformation of complete parts such as the gears, the crankshafts, the
sliding pins and the needles is now a possibility. These results confirm the considerable
effect that the operating loads and stresses have on such parts.
iii. The evaluation of reaction force, deformation and stress is possible even with the spring
model. This means that the fatigue strength of components such as the crankshafts can be
evaluated with the more simple spring model.
iv. In the contact model, the local contact conditions that cause deformation of parts such as
the pins, the gears and the needles can be calculated. Consequently, the evaluation of
rolling motion fatigue in the sliding parts has become possible.
Through this analysis, the multipoint contact analysis (190 places) converged comparatively easily.
This result shows that ABAQUS has a high functionality in solving contact problems.
7. Acknowledgements
This paper has been organized based on the content of the paper presented at the ABAQUS Japan
Users Conference held on October 30-31, 2006. The authors would like to take this opportunity to
thank the support members of ABAQUS Japan for their assistance.