In this paper, I focus on the tooth modification technology of involute straight bevel gears used in automotive differentials. The aim is to reduce vibration and noise, and to improve stress distribution on tooth surfaces. I first propose a precise solid modeling method for straight bevel gears based on spherical involute theory, using Matlab and SolidWorks. Then, a systematic dynamic simulation approach based on ANSYS/LS-DYNA is developed to evaluate the meshing performance of both unmodified and modified straight bevel gears. Profile modification and longitudinal (tooth trace) modification are studied in detail. For longitudinal modification, I introduce a novel concept called isometric modification, which offers great flexibility in controlling the contact pattern. Dynamic simulation results demonstrate that both profile and isometric modifications can effectively reduce meshing impact and make the contact stress distribution more uniform. The proposed design and simulation procedures provide a practical and reliable basis for the engineering application of tooth modification in straight bevel gears.
1. Introduction
Straight bevel gears are among the earliest developed and most widely used gear forms in power transmission systems, especially in automotive differentials. The gear transmission system is the core component of a differential, and its working performance directly affects the overall vehicle characteristics such as fuel economy, riding comfort, and service life. For automotive straight bevel gears, the applied load varies over a wide range. Factors such as manufacturing and assembly errors, bending and torsional deformation of shafts, contact deformation of teeth, and bearing clearances often cause meshing impact at the entry and exit of tooth engagement, as well as load concentration along the tooth width. To alleviate these problems, gear tooth modification is recognized as one of the most effective and economical measures.
Tooth modification can be classified into two major categories: profile modification and longitudinal (trace) modification. Profile modification removes a small portion of material near the tooth tip or tooth root so that the actual base pitches of the mating gears become equal under load, thereby reducing vibration and noise. Longitudinal modification, on the other hand, adjusts the tooth surface along the face width direction to compensate for elastic deformation, misalignment, and manufacturing errors, resulting in a more uniform load distribution.
Although many research works have been devoted to cylindrical gear modification, straight bevel gears pose additional difficulties due to their complex spherical involute tooth surfaces. Traditional methods often approximate the spherical involute with plane involutes or circular arcs, leading to significant modeling errors. Moreover, the manufacturing of crowned teeth on bevel gears requires special machine adjustments and the amount and position of crowning are difficult to control. In this context, I aim to establish a complete design and evaluation methodology for tooth modification of straight bevel gears, combining precise 3D modeling, finite element dynamic simulation, and practical modification strategies.
2. Precise 3D Modeling of Straight Bevel Gears
2.1 Spherical Involute Theory
Unlike cylindrical gears whose tooth profiles are plane involutes, straight bevel gears have tooth profiles that are spherical involutes. The spherical involute is generated by a plane rolling without slipping on a base cone. Consider a base cone with a base cone angle \(\theta\). A generating plane is tangent to the base cone along a line. When the plane rolls over the cone, a straight line lying in the plane sweeps a spherical involute surface. The formation principle is shown in the following description.
Let the cone apex be the origin of a Cartesian coordinate system \((x,y,z)\), with the cone axis along the \(z\)-axis. A moving coordinate system \((x’,y’,z’)\) is attached to the rolling plane. During rolling, the instantaneous axis of rotation is the line of tangency. A generating line \(OA\) in the rolling plane makes an angle \(\varphi\) with the instantaneous axis. After a rolling angle \(\phi\), the position of \(OA\) in the fixed coordinate system can be obtained through coordinate transformation. The resulting spherical involute surface is described by:
$$ \begin{cases} x = l \left( \sin\varphi \cos\phi + \cos\varphi \sin\phi \cos\theta \right) \\ y = l \left( \sin\varphi \sin\phi – \cos\varphi \cos\phi \cos\theta \right) \\ z = l \cos\varphi \cos\theta \end{cases} $$
where \(l\) is the distance from the cone apex to the point on the spherical involute (the sphere radius), \(\varphi\) is the angle between the generating line and the instantaneous axis, and \(\phi\) is the rolling angle. The relationship between \(\varphi\) and \(\phi\) is \(\sin\varphi = \sin\phi \sin\theta\). For a given point on the tooth surface, the sphere radius \(l\) is constant, and the intersection of the spherical involute surface with that sphere is the spherical involute curve.
2.2 Geometric Parameters of the Studied Straight Bevel Gears
The gear pair analyzed in this paper is the differential gear set from a passenger car. The planetary gear (small gear) meshes with the side gear (large gear) at a shaft angle of 90 degrees. The basic geometric parameters are summarized in Table 1.
| Parameter | Planetary gear | Side gear |
|---|---|---|
| Shaft angle | 90° | |
| Number of teeth | 10 | 15 |
| Module | 4.438 | |
| Pressure angle | 22.5° | |
| Pitch diameter (mm) | 44.38 | 66.57 |
| Cone distance (mm) | 40 ± 0.025 | |
| Pitch cone angle | 33.69° | 56.31° |
| Face cone angle | 46° | 66° |
| Root cone angle | 24° | 44° |
| Addendum (mm) | 4.98 | 3.28 |
| Whole depth (mm) | 9.17 | 9.169 |
2.3 Solid Modeling Procedure
To avoid the approximation errors introduced by using plane involutes or circular arcs, I generate the exact spherical involute curves in Matlab and then import them into SolidWorks to build the 3D solid model. The procedure is as follows:
First, I write a Matlab script to compute the coordinates of points on the spherical involute according to the equations above. For a given gear with base cone angle \(\theta\), the script outputs a set of \((x,y,z)\) coordinates for a range of \(\phi\) values. By changing the sphere radius \(l\), I obtain the spherical involute curves at the large end and the small end of the tooth. These point sets are then imported into SolidWorks using the “Curve through XYZ points” feature.
Second, I create the tooth flank surface by lofting or filling between the two spherical involute curves while also incorporating the cone apex. The symmetrical tooth flank is generated by mirroring the surface about the central plane. The addendum and root cone surfaces are created by revolving the corresponding lines. A ruled surface tangent to the spherical involute surface and intersecting the root cone is added to close the tooth space volume. After sewing all the surfaces, I obtain a solid body representing a single tooth space.
Third, the tooth space body is circularly patterned according to the number of teeth. The main blank body of the gear is then created by revolving the outer contour. By combining the patterned tooth spaces with the blank and performing a Boolean subtraction, I obtain the complete gear solid model. Finally, the central hole is added and fillets are applied at the tooth root. This parametric approach allows easy modification of the model by simply changing the input parameters such as pressure angle, number of teeth, module, and sphere radius.
The 3D models of the planetary gear and the side gear, as well as their assembly, are then used for the subsequent finite element analysis. The assembly is created by aligning the two gear axes at 90 degrees so that the pitch cone apices coincide. The correct meshing position is determined by rotating the gears until the tooth flanks are in tangency without interference.
3. Finite Element Modeling of Straight Bevel Gears Based on ANSYS/LS-DYNA
3.1 General Analysis Procedure
For the dynamic analysis of gear meshing, I employ the explicit finite element code LS-DYNA. The combination of SolidWorks, ANSYS, and LS-DYNA provides a powerful workflow. I first build the solid model in SolidWorks. Then the model is imported into ANSYS in Parasolid format. In ANSYS, I define the element types, material properties, mesh, contacts, boundary conditions, and loads. ANSYS generates the LS-DYNA keyword file (K-file), which can be further edited to refine the control parameters. Finally, LS-DYNA performs the computation, and the results are processed by LS-PREPOST or ANSYS postprocessors.
3.2 Assumptions and Material Properties
To make the simulation tractable without losing the essential physics, I make the following assumptions:
- The gear material is continuous, linear elastic, homogeneous, and isotropic.
- Deformation at the contact line follows the normal direction of the tooth surface, and the contact surface is smooth.
- The solid model is perfect (no manufacturing or assembly errors), and equivalent nodal parameters are used for contact conditions.
- Thermal deformation is neglected, and the friction obeys Coulomb’s law.
The gear material is 20CrMnTi steel with density \(\rho = 7.8 \times 10^3 \, \mathrm{kg/m^3}\), Young’s modulus \(E = 2.07 \times 10^5 \, \mathrm{MPa}\), and Poisson’s ratio \(\nu = 0.3\).
3.3 Element Types and Mesh
For the three-dimensional dynamic contact analysis, I use the SOLID164 element for the gear body. SOLID164 is an 8-node hexahedral element for explicit dynamic analysis, supporting large deformations and nonlinearities. It has no rotational degrees of freedom. To apply the rotational motion to the gears, I model the inner ring surface as a rigid body and mesh it with SHELL163 elements, which do have rotational degrees of freedom. The nodes of the shell elements are shared with the adjacent solid elements so that the rotation of the rigid inner ring drives the gear.
Mesh density plays a critical role in the accuracy of contact stress calculation. I refine the mesh in the tooth flank and tooth root regions, where large stress gradients occur. The mesh in these critical areas is about half the size of that in the rest of the gear body. The final mesh of the gear pair is shown in the previous figure. The total numbers of elements are approximately 86,000 for the planetary gear body, 175,000 for the side gear body, and about 3,500 for the rigid rings combined.
3.4 Contact Definition
Since the exact contact location is unknown in advance, I adopt the general surface-to-surface contact algorithm (STS) available in LS-DYNA. The planetary gear flank is defined as the contact surface, and the side gear flank is the target surface. The static friction coefficient is set to 0.5 and the dynamic friction coefficient to 0.3. The contact stiffness is automatically calculated by the penalty method.
3.5 Loads and Boundary Conditions
The boundary conditions are imposed on the rigid inner rings. The ring of the planetary gear is allowed to rotate about its own axis, while all other degrees of freedom are constrained. Similarly, the ring of the side gear rotates about its axis. The rotation is driven by prescribing an angular velocity on the planetary gear and a resisting torque on the side gear. Because this is a transient explicit analysis, the loads are defined as functions of time. I use constant angular velocity and constant torque to focus on the internal excitation of the gear pair itself. The applied values are 157 rad/s for the planetary gear and 10 N·m resisting torque on the side gear.
In ANSYS, these loads are applied using the EDLOAD command. The arrays for time, angular velocity, and torque are defined as shown in Table 2.
| Time (s) | Angular velocity (rad/s) | Torque (N·m) |
|---|---|---|
| 0 | 157 | 10 |
| 1 | 157 | 10 |
The total solution time is set to 0.004 s, which is sufficient to observe the engagement of one or two teeth. To reduce CPU cost without losing accuracy, mass scaling is used only on elements smaller than the specified minimum time step.
3.6 Model Verification
Before using the model for modification studies, I verify it by checking the contact stress against the Hertzian theory for a simplified contact between two cylinders. The standard finite element results are in good agreement with the analytical solution, confirming that the mesh density and contact algorithm are adequate. The proposed modeling procedure is thus reliable for simulating the complex meshing of straight bevel gears.
4. Profile Modification of Straight Bevel Gears
4.1 Principle of Profile Modification
In an ideal gear pair, the base pitches of the mating gears are exactly equal, so the transmission ratio is constant. However, in reality, due to manufacturing tolerances, assembly errors, and elastic deformation under load, the base pitches become unequal. Consequently, the actual point of contact is displaced from the theoretical pitch point, causing a varying transmission ratio and inducing dynamic loads. These dynamic loads produce impact, vibration, and noise.
Profile modification is intended to remove the interfering portion of the tooth profile, usually near the tip or root, so that the transition between single and double pair contact is smooth. Figure 4.1 in the original thesis illustrates the modification concept. The modification curve is defined by the amount of material removed and the height over which it is removed.
4.2 Determination of Profile Modification Parameters
The key parameters are the amount of modification, the height of modification, and the shape of the modification curve. The amount of modification should compensate for the elastic deformation of the tooth and a portion of the manufacturing errors. The elastic deformation can be estimated from the unit load and mesh stiffness:
$$ \delta_a = W_t \, c_r $$
where \(\delta_a\) is the elastic deformation in micrometers, \(W_t = F_t / b\) is the tangential load per unit face width, \(F_t\) is the tangential force, \(b\) is the face width, and \(c_r\) is the mesh stiffness, approximately 20 N/mm·µm for standard external gears. In this study, I also obtain the deformation directly from the finite element model by comparing the circumferential displacements of a point in the meshing zone and a point far away from the contact. This value is taken as the modification amount at the large end. The modification amount at the small end is then scaled according to the cone geometry:
$$ S_1 = S_2 \frac{R – b}{R} $$
where \(R\) is the cone distance, \(b\) is the face width, \(S_1\) is the small-end modification, and \(S_2\) is the large-end modification. The modification amounts for the two gears are listed in Table 3.
| Gear | \(S_2\) (large end) µm | \(S_1\) (small end) µm |
|---|---|---|
| Side gear | 29.03 | 17.33 |
| Planetary gear | 27.69 | 18.06 |
For the modification height, I adopt the “short modification” scheme, which starts at the entry of the meshing zone and extends to about one-third of the active profile. This scheme retains a part of the double-pair contact region and is suitable for straight bevel gears. The modification heights at the small and large ends are given in Table 4.
| Gear | Small end \(L_1\) (mm) | Large end \(L_2\) (mm) |
|---|---|---|
| Planetary gear | 1.79 | 3.0 |
| Side gear | 2.4 | 3.0 |
4.3 Modification Curve
Many types of curves have been proposed for profile modification, such as Walker’s parabola, the quadratic curve, and the straight line. They can be expressed generally as:
$$ \Delta(x) = \Delta_{\max} \left( \frac{x}{l} \right)^n $$
where \(\Delta(x)\) is the modification amount at a distance \(x\) from the starting point, \(\Delta_{\max}\) is the maximum amount at the tip, \(l\) is the modification height, and \(n\) is the exponent. In this work, I select a circular arc as the modification curve because it is easy to model and manufacture, and it ensures a smooth transition with the original involute profile. The circular arc is constructed such that it passes through the starting point and the tip point, with its center located on the normal to the involute at the starting point.
4.4 Dynamic Simulation Results of Profile-Modified Gears
Using the finite element model described in Section 3, I simulate the meshing of straight bevel gears before and after profile modification. The results are compared in terms of contact stress and acceleration.
4.4.1 Contact Stress
The contact stress distribution on the tooth flank is shown in the finite element contour plots. For the side gear, the maximum contact stress is reduced from about \(3.5 \times 10^9\) Pa to \(2.9 \times 10^9\) Pa, a reduction of 17.1%. For the planetary gear, the maximum contact stress drops from \(2.4 \times 10^9\) Pa to \(1.7 \times 10^9\) Pa, a reduction of 29.1%. These results indicate that profile modification effectively reduces the magnitude of the impact stress and improves the overall stress level.
4.4.2 Angular and Axial Acceleration
Vibration and noise in gear transmission are directly related to fluctuations in angular acceleration and axial acceleration. I extract the accelerations of a node on the tooth tip for both the modified and unmodified cases. The angular acceleration curves show that after modification, the amplitude of angular acceleration fluctuations is significantly reduced for the planetary gear. The axial acceleration also decreases, indicating reduced axial vibration. For the side gear, the angular acceleration amplitude is also reduced, but the axial acceleration change is less noticeable because the axial acceleration of the side gear is inherently small.
These results confirm that profile modification of straight bevel gears can effectively reduce meshing impact and improve the smoothness of transmission, thereby reducing vibration and noise.
5. Longitudinal (Tooth Trace) Modification of Straight Bevel Gears
5.1 Necessity of Longitudinal Modification
Under real operating conditions, the load distribution along the face width of a straight bevel gear is rarely uniform due to elastic deflections of shafts and gears, manufacturing and assembly errors, and thermal effects. The contact often concentrates at one end of the tooth, leading to severe edge loading, pitting, or tooth fracture. Longitudinal modification, also known as trace modification, is a common solution to improve the load distribution by making the tooth flank slightly convex near the middle or at a desired position.
5.2 Crowning Modification
The most common longitudinal modification is crowning. A crowned tooth (Figure 5.1) has a barrel-shaped flank. The design parameters include the crown radius and the amount of crowning. For an arc-shaped crowned tooth, the relationship between the radius of curvature \(R_c\), face width \(b\), and crown amount \(\Delta\) is:
$$ 2 R_c = \frac{b^2}{8\Delta} $$
The amount of crowning can be determined from standards or empirical formulas. The ISO standard recommends values between 10 and 40 µm for general precision gears, with an additional allowance for manufacturing errors. A more specific formula is:
$$ \Delta = 0.5 f_{sh} + f_{H\beta} + 5 \sim 10 \, \mu\mathrm{m} $$
where \(f_{sh}\) is the meshing tooth trace error under unit load and \(f_{H\beta}\) is the helix slope deviation. Alternatively, a simpler formula often used is:
$$ \Delta = \left( 0.25 \times 10^{-3} b + 0.5 \right) \, \mu\mathrm{m} $$
Although crowning works well for cylindrical gears, applying it to straight bevel gears is difficult because the special machinery required to generate a true crown on a bevel gear is expensive and the position of the crown center is hard to adjust. Thus, I propose a new approach called “isometric modification” for straight bevel gears.
5.3 Isometric Modification Concept
Isometric modification is defined as removing a uniform layer of material from a selected region of the standard tooth flank in the direction normal to the flank. The new surface is still a spherical involute surface, parallel to the original one. By choosing the location and the thickness of the removed layer, the contact area of the gear pair can be controlled. In this method, only the driving gear (planetary gear) is modified; the driven gear (side gear) remains unmodified, because the deformation of the small gear is more significant.
The isometric modification region is determined by the desired contact pattern. According to the Chinese standard GB/T11365-1989, the ideal contact area for a 6-7 grade bevel gear should be 50–70% along the tooth length and 55–75% along the tooth height. Based on practical experience, the contact area should be located near the small end, approximately one-third of the face width from the small end. For the planetary gear, I choose the modification window with dimensions: distance from small end \(a = 2.3\) mm, length \(b = 3.2\) mm, height \(c = 2.3\) mm, and \(d = 2.4\) mm as indicated in the original illustration.
5.4 Determination of Isometric Modification Amount
The isometric modification amount \(h\) is the thickness of the removed layer. It can be estimated by the same methods as for crowning. However, one must also verify that the modification does not significantly affect the kinematic accuracy of the gear pair. The modification changes the position of the contact point and the angular position of the driven gear. I derive a simple relationship between the modification amount and the angular error using differential geometry.
Let \(\Delta \Phi_2\) be the change in the angular position of the driven gear caused by the modification amount \(h\). For a straight bevel gear, the relationship is approximately:
$$ \Delta \Phi_2 = \frac{h}{R_b} $$
where \(R_b\) is the base circle radius of the equivalent spur gear. The calculated \(\Delta \Phi_2\) must be smaller than the tangential composite tolerance of the gear pair specified in the standard. Using the finite element method, I compute the relative circumferential displacement between a point in the meshing zone and a point far away on the same tooth. This displacement is taken as the isometric modification amount \(h = 30 \, \mu\mathrm{m}\). Substituting into the formula above gives an angular error well below the tolerance, validating the selected amount.
5.5 Simulation Results and Discussion
5.5.1 Stress Distribution along Face Width
I extract the maximum equivalent stress of a series of contact elements distributed along the face width on the tooth flank. The stress distribution before and after isometric modification is shown in the line chart. Before modification, the stress is highly non-uniform with a clear concentration at the large end. After modification, the stress distribution becomes more symmetrical and is confined within the modified region near the small end. The large-end stress concentration is significantly reduced. Although the maximum stress on the planetary gear slightly increases, the stress on the side gear does not increase significantly. The overall improvement is beneficial for avoiding edge wear and pitting.
5.5.2 Tooth Root and Contact Area Stresses
For the planetary gear, the tooth root stress decreases from \(0.96 \times 10^3\) MPa to \(0.65 \times 10^3\) MPa after isometric modification. Meanwhile, the contact stress in the modified zone increases from \(1.0 \times 10^3\) MPa to \(2.1 \times 10^3\) MPa, because the load is now concentrated in a smaller area. For the side gear, which is not modified, both the contact and root stresses are reduced slightly due to the more uniform load distribution. The stress-time curves for the contact and root elements illustrate that the dynamic load fluctuation is smaller after modification.
The isometric modification method offers several advantages over traditional crowning. It can be easily implemented in the forging die cavity or by machining because the modified surface is simply a parallel offset of the original spherical involute. The position of the contact area can be adjusted by moving the modification window, and the amount can be controlled by varying the offset distance. This makes it possible to design gears for different operating conditions, accuracy levels, and load spectra.
6. Conclusion
In this paper, I have systematically investigated the tooth modification design of automotive straight bevel gears, including profile modification and longitudinal (isometric) modification. The main conclusions are as follows:
(1) The exact spherical involute of straight bevel gears can be generated parametrically in Matlab and imported into SolidWorks to create high-fidelity solid models. This method eliminates the approximation errors associated with plane involute or circular arc substitutions and facilitates rapid remodeling for different gear parameters.
(2) The SolidWorks-ANSYS-LS-DYNA combined simulation procedure is well suited for the dynamic analysis of straight bevel gear meshing. Proper selection of element types, mesh density, contact algorithm, and boundary conditions ensures accurate prediction of contact stress, acceleration, and other transient responses.
(3) Profile modification of straight bevel gears reduces the meshing impact and angular acceleration fluctuations. The contact stress on the modified gears is significantly lower than that on the unmodified gears, confirming the effectiveness of profile modification in improving the transmission smoothness of straight bevel gears.
(4) The proposed isometric modification concept provides a practical alternative to crowning for straight bevel gears. It allows precise control of the contact region and the amount of modification, leading to a more uniform stress distribution along the tooth width and a lower tooth root stress. The method is easy to implement in manufacture and offers great flexibility for various engineering requirements.
Future work should investigate the effect of different modification curve shapes on the dynamic performance of straight bevel gears and include the influence of variable loads, thermal deformation, and installation errors to better represent real operating conditions.