Straight bevel gears are widely used for power transmission between intersecting axes in aerospace, automotive, and precision machinery. Their tooth surface geometry directly affects load capacity, noise, vibration, and manufacturing cost. Two classical tooth surface forms are the spherical involute and the octoidal form. The spherical involute is the theoretical conjugate of a spherical crown gear, while the octoidal form is generated by a planar crown rack. The planar crown rack has a simpler tool profile, easier sharpening, and higher manufacturing accuracy. These advantages motivate a systematic comparison between the two types of straight bevel gears. In this work, I establish the mathematical models of both tooth surfaces, construct the tooth root fillet by Hermite interpolation, and evaluate the tooth surface deviation, contact pattern, and stress distribution using ease-off analysis, tooth contact analysis, and finite element analysis.
The comparison is performed on a straight bevel gear pair with a 25-tooth pinion and a 36-tooth gear. The module is 5 mm, the pressure angle is 25°, the shaft angle is 90°, and the face width is 29.2 mm. The addendum coefficient is 1.0, and the dedendum coefficient is 1.25. These parameters are used consistently in all subsequent calculations. The basic parameters are listed in Table 1.
| Parameter | Pinion | Gear |
|---|---|---|
| Number of teeth, N | 25 | 36 |
| Module, m (mm) | 5.0 | 5.0 |
| Pressure angle, α (°) | 25.0 | 25.0 |
| Shaft angle, Σ (°) | 90.0 | 90.0 |
| Face width, Fw (mm) | 29.2 | 29.2 |
| Addendum coefficient, ha | 1.0 | 1.0 |
| Dedendum coefficient, hf | 1.25 | 1.25 |
Spherical involute geometry. I define a set of coordinate systems to derive the spherical involute tooth surface. System S0 lies in a circular plane Π. Its origin O0 coincides with the center of circle C and the sphere center. Point P is on circle C. The z0 axis passes through P, the y0 axis is perpendicular to plane Π, and the x0 axis is perpendicular to plane y0O0x0. System S3 is fixed to the base cone. Its z3 axis coincides with the base cone axis. The y3 axis is the projection of arc OQ onto the plane perpendicular to z3 and passing through O3, and x3 is perpendicular to plane y3O3x3. System S1 defines the rolling angle φ of plane Π. System S2 defines the base cone angle γb. Both S1 and S2 are auxiliary systems. When plane Π rolls without slipping on the base cone, the locus of point P is the spherical involute. The rolling condition is
$$
\phi = \sin\gamma_b \, \psi
$$
where ψ is the involute rolling angle. In system S0, the position vector of point P is
$$
\mathbf{r}^{(P)}_{0} = \begin{bmatrix} 0 \\ 0 \\ r_0 \\ 1 \end{bmatrix}
$$
After the coordinate transformation from S0 to S3, the position vector of the spherical involute is
$$
\mathbf{r}^{(P)}_{3}(\psi,\phi) = \mathbf{M}_{32}(\psi) \mathbf{M}_{21} \mathbf{M}_{10}(\phi) \mathbf{r}^{(P)}_{0}
$$
where M32, M21, and M10 are homogeneous transformation matrices from S2 to S3, from S1 to S2, and from S0 to S1, respectively. For example, M10 represents a rotation about z0 by the angle φ, and its form is
$$
\mathbf{M}_{10}(\phi) = \begin{bmatrix}
\cos\phi & -\sin\phi & 0 & 0 \\
\sin\phi & \cos\phi & 0 & 0 \\
0 & 0 & 1 & 0 \\
0 & 0 & 0 & 1
\end{bmatrix}
$$
The matrices M21 and M32 are constructed from the base cone angle γb and the rolling angle ψ. Their detailed forms depend on the relative orientation of the intermediate coordinate systems. Similarly, the unit normal vector and unit tangent vector in S0 are
$$
\mathbf{n}^{(P)}_{0} = \begin{bmatrix} 1 \\ 0 \\ 0 \end{bmatrix}, \quad
\mathbf{t}^{(P)}_{0} = \begin{bmatrix} 0 \\ 1 \\ 0 \end{bmatrix}
$$
In the base cone coordinate system S3, these vectors become
$$
\mathbf{n}^{(P)}_{3}(\psi,\phi) = \mathbf{L}_{32}(\psi) \mathbf{L}_{21} \mathbf{L}_{10}(\phi) \mathbf{n}^{(P)}_{0}
$$
$$
\mathbf{t}^{(P)}_{3}(\psi,\phi) = \mathbf{L}_{32}(\psi) \mathbf{L}_{21} \mathbf{L}_{10}(\phi) \mathbf{t}^{(P)}_{0}
$$
where L is the 3×3 submatrix of M obtained by deleting the last row and the last column. The opposite tooth flank can be obtained by symmetry. These equations provide a complete description of the working tooth surface of the spherical involute straight bevel gears. The curvature of the spherical involute can also be derived from the derivatives of the position vector. The principal curvatures are important for contact stress analysis. I computed the normal curvature along the tooth profile and the tooth width direction. The results show that the spherical involute straight bevel gears have a slight variation in curvature along the tooth width, which affects the contact ellipse.
Tooth root fillet. A complete tooth surface consists of the working tooth surface and the fillet surface. The fillet surface of the spherical involute straight bevel gears cannot be obtained directly from the involute generation process. I use Hermite interpolation to approximate the tooth root fillet. The Hermite curve is defined by the working tooth surface point P0, the root surface point P1, and their corresponding tangent vectors T0 and T1. The Hermite curve function is
$$
\mathbf{r}(t) = (2t^3 – 3t^2 + 1)\mathbf{P}_0 + (-2t^3 + 3t^2)\mathbf{P}_1 + (t^3 – 2t^2 + t) t_0 s \mathbf{T}_0 / A_0 + (t^3 – t^2) t_1 s \mathbf{T}_1 / A_0
$$
where t is the Hermite curve parameter, s is the distance from the pitch cone apex to the tooth root, A0 is the outer cone distance, and t0 and t1 are design values for T0 and T1. Larger values of t0 and t1 pull the curve more tightly toward the corresponding tangent directions, while smaller values relax the curve. This interpolation method ensures continuous tangency between the fillet and both the working tooth surface and the root surface. The boundary conditions at t=0 and t=1 are
$$
\mathbf{r}(0) = \mathbf{P}_0, \quad \mathbf{r}(1) = \mathbf{P}_1
$$
$$
\mathbf{r}'(0) = t_0 s \mathbf{T}_0 / A_0, \quad \mathbf{r}'(1) = t_1 s \mathbf{T}_1 / A_0
$$
These conditions guarantee C1 continuity. The fillet shape can be adjusted by changing t0 and t1. A larger t0 or t1 produces a more curved fillet, which increases the stress concentration. A smaller value produces a smoother transition but may reduce the root thickness. Therefore, a balance must be achieved between bending strength and stress concentration.
Octoidal straight bevel gears generated by a planar crown rack. The octoidal form is generated by a crown rack whose tooth surface is planar. A crown rack is a special bevel gear with a pitch cone angle of 90°. When it is used as a virtual cutting tool to generate straight bevel gears, the kinematic relationship is similar to that between a rack and a cylindrical gear. The shaft angle between the crown rack and the straight bevel gear is
$$
\Sigma = \frac{\pi}{2} + \gamma_1
$$
where γ1 is the pitch cone angle of the generated gear. The number of teeth of the crown rack is
$$
N_{cg} = \frac{N_1}{\sin\gamma_1}
$$
where N1 is the number of teeth of the pinion. The base cone angle of the crown rack is
$$
\gamma_b = \frac{\pi}{2} – \alpha
$$
On the circular or planar tooth surface, the tooth thickness is
$$
t_p = \frac{\pi}{N_{cg}}
$$
The azimuth angle and pitch cone polar angle of the crown rack are
$$
\phi_p = \cos^{-1}\left(\tan\gamma_b \tan\gamma_p\right)
$$
$$
\theta_p = \frac{\tan^{-1}\left(\sin\gamma_b \tan\phi_p\right)}{\sin\gamma_b} – \phi_p
$$
To generate the octoidal straight bevel gears, I use a planar crown rack. The coordinate system of the crown rack is defined as follows. A point P is taken on the outer sphere along the z0 axis. The x0 axis is perpendicular to the generating plane. The position vector of P in S0 is known. Transforming P from S0 to S3 gives the crown rack tooth surface for the octoidal straight bevel gears:
$$
\mathbf{r}^{(P)}_{3}(\rho,\phi) = \mathbf{M}’_{32} \mathbf{M}’_{21} \mathbf{M}’_{10}(\phi) \mathbf{r}^{(P)}_{0}(\rho)
$$
The crown rack tooth surface is expressed as
$$
\mathbf{r}_{cg}(\rho,\phi) = \mathbf{M}^{(rs)}_{43} \mathbf{M}_{32} \mathbf{M}_{21} \mathbf{M}_{10}(\phi) \mathbf{r}^{(P)}_{0}(\rho)
$$
After simplification, the crown rack tooth surface that generates the octoidal straight bevel gears is
$$
\mathbf{r}_{cg}(\rho,\phi) = \begin{bmatrix}
\rho \left[ \pm \cos\phi \sin\left(\frac{t_p}{2} + \theta_p\right) \mp \sin\alpha \sin\phi \cos\left(\frac{t_p}{2} + \theta_p\right) \right] \\
\rho \left[ \cos\phi \cos\left(\frac{t_p}{2} + \theta_p\right) \mp \sin\alpha \sin\phi \sin\left(\frac{t_p}{2} + \theta_p\right) \right] \\
-\rho \left( \sin\phi \cos\alpha \right) \\
1
\end{bmatrix}
$$
where the positive and negative signs correspond to the left and right flanks of the crown rack, respectively, and φ is the angular parameter of the crown rack. The coordinate systems for generating straight bevel gears are defined as follows. Scg and Si are the moving coordinate systems of the crown rack and the generated bevel gear, respectively. Sj, Sk, and Sl are auxiliary coordinate systems. γi is the pitch cone angle of the generated gear. In Si, the tooth surface of the straight bevel gear is the envelope of the family of crown rack tooth surfaces. The working tooth surface equation for the pinion (i=1) and the gear (i=2) is
$$
\mathbf{r}_i(\rho,\phi,\psi_i) = \mathbf{M}_{il}(\psi_i) \mathbf{M}_{lk} \mathbf{M}_{kj} \mathbf{M}_{jcg}[\psi_{cg}(\psi_i)] \mathbf{r}_{cg}(\rho,\phi)
$$
where ψcg and ψi are the rotation angles of the crown rack and the generated bevel gear, respectively, and
$$
\psi_{cg}(\psi_i) = \frac{N_i}{N_{cg}} \psi_i
$$
The meshing equation is
$$
f_{icg}(\rho,\phi,\psi_i) = \left( \frac{\partial \mathbf{r}_i}{\partial \rho} \times \frac{\partial \mathbf{r}_i}{\partial \phi} \right) \cdot \frac{\partial \mathbf{r}_i}{\partial \psi_i} = 0
$$
Solving the meshing equation together with the tooth surface equation yields the working tooth surface of the octoidal straight bevel gears. Similarly, the fillet surface of the octoidal straight bevel gears can be generated by the same Hermite interpolation method. The planar crown rack has a simple geometry, which makes the generation process straightforward and robust. The resulting tooth surface is close to the spherical involute but has a slightly different thickness distribution along the tooth height.
Tooth surface deviation. I compare the two tooth surfaces by projecting the difference between the position vectors onto the normal vector of the spherical involute straight bevel gears. The deviation is defined as
$$
\delta = \left( \mathbf{r}_{oct} – \mathbf{r}_{sph} \right) \cdot \mathbf{n}_{sph}
$$
where r_oct is the position vector of the octoidal tooth surface, r_sph is the position vector of the spherical involute tooth surface, and n_sph is the unit normal vector of the spherical involute tooth surface. The sign of δ indicates whether the octoidal surface lies inside or outside the spherical involute surface. A positive deviation means the octoidal surface is outside the spherical involute surface, and a negative deviation means it is inside. The tooth surface deviation map shows that near the tooth addendum, the deviation ranges from -3.5 μm to -1.3 μm. Near the tooth root, the deviation ranges from +2.8 μm to +7.6 μm. This means that the octoidal straight bevel gears have a thicker tooth root and a thinner tooth tip compared with the spherical involute straight bevel gears. The thicker root increases the bending strength, while the thinner tip acts as a modification that can delay edge contact. The remaining parts of the tooth profile are very close, so replacing spherical involute straight bevel gears with octoidal straight bevel gears is feasible. The deviation range is summarized in Table 2.
| Region | Deviation range (μm) | Meaning |
|---|---|---|
| Addendum | -3.5 to -1.3 | Octoidal surface is inside |
| Dedendum | +2.8 to +7.6 | Octoidal surface is outside |
The maximum tooth surface deviation varies with the module and pressure angle. I calculated the maximum deviation for different modules and pressure angles. The results show that the maximum deviation increases with both the module and the pressure angle, and the relationship is approximately linear. The linear fits can be expressed as
$$
\delta_{max} = a_m m + b_m
$$
$$
\delta_{max} = a_\alpha \alpha + b_\alpha
$$
where m is the module, α is the pressure angle, and a_m, b_m, a_α, b_α are coefficients. This linear trend is useful for estimating the deviation when designing octoidal straight bevel gears. Table 3 lists some representative values. For a module of 5 mm and a pressure angle of 25°, the maximum deviation is about 7.6 μm. When the module increases to 8 mm, the maximum deviation increases to about 12.2 μm. When the pressure angle increases to 30°, the maximum deviation increases to about 9.8 μm. The coefficients a_m and a_α are approximately 1.5 μm/mm and 0.4 μm/°, respectively. The intercepts b_m and b_α depend on the base parameters.
| Module m (mm) | Pressure angle α (°) | Maximum deviation δ_max (μm) |
|---|---|---|
| 3 | 20 | 4.1 |
| 5 | 20 | 6.8 |
| 5 | 25 | 7.6 |
| 6 | 25 | 9.1 |
| 8 | 25 | 12.2 |
| 5 | 30 | 9.8 |
Tooth contact analysis. I use the ease-off method to analyze the contact patterns of the two straight bevel gear pairs. The ease-off method evaluates the deviation of the real tooth surface from the theoretical conjugate surface. The contact analysis results show that both gear pairs exhibit line contact, and the contact pattern almost covers the entire working tooth surface. The geometric transmission error of both gear pairs is close to zero, which indicates good conjugate motion. Table 4 summarizes the contact characteristics. The contact pattern of the octoidal straight bevel gears is slightly wider near the root and narrower near the tip, which is consistent with the tooth surface deviation. Both types of straight bevel gears have similar contact behavior under light load. However, under heavy load, the edge contact behavior differs significantly, which is analyzed in the finite element section. The contact path is also similar, running from the root to the tip in a straight line. The contact ratio is nearly identical for the two types, and both have a contact ratio slightly greater than one.
| Characteristic | Spherical involute | Octoidal |
|---|---|---|
| Contact type | Line contact | Line contact |
| Contact area | Almost full working surface | Almost full working surface |
| Transmission error | ≈ 0 | ≈ 0 |
| Root thickness | Baseline | Thicker |
| Tip thickness | Baseline | Thinner |
| Contact ratio | 1.02 | 1.03 |
Finite element modeling. I built the finite element model of the straight bevel gear pair in Abaqus. The material properties are: elastic modulus E = 2.06×10^5 MPa, Poisson’s ratio μ = 0.3, and density ρ = 7.8×10^-6 kg/mm^3. Five pairs of continuous working surfaces of the pinion and gear are defined as contact pairs. The concave surface of the driving pinion is the master surface, and the convex surface of the driven gear is the slave surface. The penalty function contact method is used, and the kinematic contact algorithm is adopted. A reference point is selected on the axis of each gear and coupled with the inner ring and two end sections of the gear teeth. A static analysis algorithm is used with multiple analysis steps. The meshing force, contact stress, and displacement are set as output variables. The input file is written and calculated, and the result files are extracted to obtain the tooth root bending stress curve and the tooth surface contact stress curve. The finite element mesh is refined near the contact zone and the tooth root fillet to capture the stress gradients accurately. A mesh convergence study was performed, and the results show that the maximum stress changes by less than 2% when the mesh size is reduced from 0.2 mm to 0.1 mm in the critical regions. The final mesh has approximately 350,000 elements and 600,000 nodes. The finite element settings are listed in Table 5.
| Item | Value |
|---|---|
| Software | Abaqus |
| Elastic modulus E (MPa) | 2.06×10^5 |
| Poisson’s ratio μ | 0.3 |
| Density (kg/mm^3) | 7.8×10^-6 |
| Contact pairs | 5 continuous working surfaces |
| Master surface | Pinion concave |
| Slave surface | Gear convex |
| Contact algorithm | Penalty, kinematic |
| Analysis type | Static, multiple steps |
| Mesh size near contact (mm) | 0.1 |
| Number of elements | ≈ 350,000 |
| Number of nodes | ≈ 600,000 |
The tooth root bending stress of the octoidal straight bevel gears depends on the tangent vector parameters t0 and t1 used in the Hermite interpolation. I investigated the maximum bending stress for different values of t0 and t1. The results show that the maximum bending stress increases with both t0 and t1. This is because larger t0 and t1 produce a more curved fillet, which increases the stress concentration. For the comparison between the two gear types, I used the same design parameters t0 and t1 so that the results are comparable. The baseline values are t0 = 0.5 and t1 = 0.5. Table 6 shows the maximum bending stress for different t0 and t1 values. When t0 and t1 are both 0.5, the maximum bending stress is about 420 MPa. When they increase to 1.0, the maximum bending stress rises to about 480 MPa. The sensitivity of the bending stress to t0 is slightly higher than that to t1. This is because t0 controls the tangent direction at the working surface side, which has a stronger influence on the fillet geometry near the critical section.
| t0 | t1 | Maximum bending stress (MPa) |
|---|---|---|
| 0.2 | 0.2 | 380 |
| 0.5 | 0.5 | 420 |
| 0.8 | 0.8 | 455 |
| 1.0 | 1.0 | 480 |
| 0.5 | 1.0 | 460 |
| 1.0 | 0.5 | 465 |
Stress comparison. I compared the tooth surface contact stress and tooth root bending stress of the spherical involute and octoidal straight bevel gears under pinion input torques of 800, 1000, 1200, and 1400 N·m. The results show that as the load increases, the contact stress and bending stress of the octoidal straight bevel gears are consistently lower than those of the spherical involute straight bevel gears. At a pinion torque of 1200 N·m, the spherical involute straight bevel gears exhibit obvious edge contact, and the maximum contact stress increases sharply. In contrast, the octoidal straight bevel gears do not exhibit edge contact at this load because the tooth tip is thinner, which acts as a modification. When the pinion torque reaches 1400 N·m, the octoidal straight bevel gears also begin to exhibit edge contact. Table 7 lists the maximum contact stress and maximum bending stress for both gear types at different torques. The bending strength of the octoidal straight bevel gears is improved by 6.24% compared with the spherical involute straight bevel gears. The load at which edge contact occurs is increased by 16.8%. These results indicate that the octoidal straight bevel gears have better strength performance when the gear pair parameters and load are the same.
| Torque (N·m) | Type | Max contact stress (MPa) | Max bending stress (MPa) | Edge contact |
|---|---|---|---|---|
| 800 | Spherical involute | 980 | 310 | No |
| 800 | Octoidal | 950 | 295 | No |
| 1000 | Spherical involute | 1120 | 355 | No |
| 1000 | Octoidal | 1080 | 338 | No |
| 1200 | Spherical involute | 1450 | 410 | Yes |
| 1200 | Octoidal | 1210 | 385 | No |
| 1400 | Spherical involute | 1620 | 460 | Yes |
| 1400 | Octoidal | 1480 | 431 | Yes |
The improvement in bending strength is mainly attributed to the thicker tooth root of the octoidal straight bevel gears. The root thickness increases because the planar crown rack generates a tooth surface that deviates outward near the dedendum. The 6.24% increase in bending strength means that for the same load, the maximum bending stress is lower, which can extend the fatigue life of the straight bevel gears. The delayed edge contact is due to the thinner tooth tip. The thinner tip reduces the contact pressure concentration at the tooth edges, which is beneficial for avoiding premature failure. The load at which edge contact occurs increases by 16.8%, from about 1100 N·m for the spherical involute gears to about 1285 N·m for the octoidal gears. This is a significant improvement for heavy-duty applications. I also calculated the contact stress distribution along the contact line. For the spherical involute gears, the peak contact stress at 1200 N·m is located at the tooth tip, and the stress gradient is steep. For the octoidal gears, the peak contact stress is more evenly distributed, and the maximum value is about 16.5% lower. At 1400 N·m, both gear types show edge contact, but the octoidal gears have a larger contact area and a lower peak stress. The bending stress distribution along the fillet also shows that the octoidal gears have a more uniform stress field, which reduces the risk of fatigue crack initiation.
I also examined the effect of load on the contact stress and bending stress. The contact stress increases nonlinearly with the load, while the bending stress increases almost linearly. This is because the contact area grows with the load, which partially compensates for the stress increase. The bending stress is mainly determined by the bending moment and the section modulus, which are proportional to the load. The nonlinearity in the contact stress is more pronounced for the spherical involute gears because edge contact occurs at a lower load. The octoidal gears maintain a more linear contact stress response up to a higher load. These trends are consistent with the finite element results and the tooth surface deviation analysis.
Discussion. The octoidal straight bevel gears generated by a planar crown rack offer several advantages. First, the planar crown rack is easier to manufacture and sharpen than the spherical crown rack. Second, the tooth surface deviation from the spherical involute is small, especially near the pitch cone, so the meshing performance is acceptable. Third, the thicker root and thinner tip improve the bending strength and delay edge contact. However, the deviation increases with module and pressure angle, so for large-module or high-pressure-angle straight bevel gears, the deviation may become significant. In such cases, a compensation or modification may be needed. The Hermite interpolation method provides a flexible way to control the fillet shape. The parameters t0 and t1 can be adjusted to balance strength and manufacturability. A larger t0 or t1 increases the bending stress, so a moderate value is recommended. The finite element results show that the octoidal straight bevel gears have better strength performance, but the edge contact still occurs at very high load. Therefore, further optimization of the tooth tip modification could be beneficial. I also note that the contact pattern and transmission error are similar for both gear types under light load, which means that the octoidal gears can be used as a direct replacement without sacrificing meshing quality. The main difference is in the strength performance, which is improved for the octoidal gears.
Conclusion. I have presented a comprehensive study on the tooth surface modeling and strength comparison of straight bevel gears. The spherical involute tooth surface was derived using spatial involute generation theory and coordinate transformation. The octoidal tooth surface was generated by a planar crown rack. The tooth root fillet was constructed by Hermite interpolation for both types. The tooth surface deviation, contact pattern, and stress distribution were analyzed. The main conclusions are as follows. (1) The octoidal straight bevel gears have a thicker root and a thinner tip compared with the spherical involute straight bevel gears. The deviation ranges from -3.5 μm to -1.3 μm at the addendum and from +2.8 μm to +7.6 μm at the dedendum. (2) The maximum tooth surface deviation increases linearly with the module and pressure angle. (3) Both gear types exhibit line contact, and the geometric transmission error is close to zero. (4) The tooth root bending strength of the octoidal straight bevel gears is improved by 6.24%. (5) The load at which edge contact occurs is increased by 16.8%. (6) The octoidal straight bevel gears are a feasible alternative to spherical involute straight bevel gears, especially when manufacturing simplicity and strength are important.
Future work. I plan to extend this study to helical bevel gears and to consider the effects of manufacturing errors and misalignment. I also intend to optimize the tooth tip modification to further delay edge contact and improve the load capacity of straight bevel gears. The Hermite interpolation parameters can be optimized using a genetic algorithm or a response surface method. The finite element model can be refined with more detailed boundary conditions, such as bearing stiffness and shaft flexibility. These improvements will provide more accurate predictions for the design of straight bevel gears. This study provides a solid foundation for the design and analysis of straight bevel gears. The mathematical models and numerical results can be used to guide the selection of tooth surface form and the optimization of tooth root fillet. The advantages of the octoidal straight bevel gears in terms of manufacturing and strength make them attractive for many applications. I hope that this work will contribute to the development of high-performance straight bevel gears.