Spiral bevel gears are widely used to transmit torque between intersecting shafts, commonly at a 90° shaft angle. Their curved, gradually engaging teeth provide high load capacity, smooth motion, and efficient power transfer. The same geometric features, however, make the contact pattern sensitive to small deviations in tooth-surface shape. A mismatch that concentrates load near the tooth tip or the tooth end produces edge contact: the nominal contact area becomes very small, and the local contact stress rises far above the level expected from the applied torque. Tightening manufacturing and installation tolerances can reduce this risk, but at increasing cost. Controlled tooth-surface modification is therefore an important alternative for improving load distribution and reliability.
The practical question is not simply whether modification lowers stress, but which components of the tooth-surface deviation control the peak stress, and how those components should be selected together. For spiral bevel gears, the contact path, curvature, and ease-off are more complex than for spur or helical gears, so modification parameters cannot be transferred directly from simpler gear types. This article examines a five-parameter modification space applied to the pinion of a 14-tooth/25-tooth spiral bevel gear pair, quantifies the sensitivity of maximum contact stress to each parameter, and describes an optimization procedure that uses a transient finite-element model together with a surrogate model and a non-dominated sorting genetic algorithm.
Gear Geometry and Material
The analyzed pair consists of a 14-tooth pinion and a 25-tooth gear with a 90° shaft angle. The main geometry is listed in Table 1. Both members have a face width of 16 mm, a large-end transverse module of 3.5 mm, a normal pressure angle of 20°, and a spiral angle of 35°, with right-hand teeth on the pinion and left-hand teeth on the gear. The pinion and gear use +0.025 and −0.025 tangential modification coefficients, respectively, and +0.27 and −0.27 radial modification coefficients. The material is 20CrMnTi, with a density of 7,800 kg/m³, an elastic modulus of 207 GPa, and a Poisson’s ratio of 0.25.
| Parameter | Pinion | Gear |
|---|---|---|
| Number of teeth | 14 | 25 |
| Large-end transverse module / mm | 3.5 | 3.5 |
| Spiral angle / (°), hand | 35, right-hand | 35, left-hand |
| Face width / mm | 16 | 16 |
| Normal pressure angle / (°) | 20 | 20 |
| Shaft angle / (°) | 90 | |
| Pitch cone angle / (°) | 29.249 | 60.751 |
| Root cone angle / (°) | 26.171 | 55.544 |
| Face cone angle / (°) | 34.456 | 63.829 |
| Cone distance / mm | 50.143 | 50.143 |
| Whole tooth depth / mm | 6.608 | 6.608 |
| Tangential modification coefficient | 0.025 | −0.025 |
| Radial modification coefficient | 0.27 | −0.27 |
Transient Finite-Element Contact Model
The mesh uses hexahedral elements, with 70 divisions along the tooth profile and 50 divisions along the face width. The resulting model contains 1,308,220 elements. A mesh-independence check was performed before the final density was selected. Contact between the pinion and gear teeth is defined as a pinion-tooth contact surface against a gear-tooth target surface. Because ISO 10300-1:2023 neglects friction when calculating the load capacity of spiral bevel gears, the contact is modeled as frictionless.
The inner rings of the two gears are coupled to rotational joints that release only the axial rotational degree of freedom. This arrangement allows a rotational speed to be applied to the pinion and a resisting torque to be applied to the gear. Both speed and torque are ramped over 0.002 s. The subsequent analysis interval is 0.015 s, corresponding to one revolution of the gear. The total simulated time is therefore 0.017 s. To improve contact convergence, the gear position at each meshing location is adjusted so that contact is just established at the start of the step.
Baseline Contact Stress and Edge Contact
Before modification, the contact-stress distribution remains at a high level throughout the mesh cycle. The maximum contact stress reaches 5,595 MPa during meshing-out, and the peak is located at the tooth tip. This behavior is characteristic of edge contact: the load is carried by a small region near the tooth boundary rather than by an elliptical area in the middle of the tooth surface. Because the contact zone is narrow, the local stress is high even when the applied load is unchanged.
The baseline result also shows that the stress problem is not limited to a single instant. The stress fluctuates periodically with the mesh cycle, and the high-stress condition persists over a substantial portion of that cycle. Reducing the peak therefore requires changing the contact pattern, not merely adjusting one local point on the tooth surface.
Five-Parameter Tooth-Surface Modification
Modification is applied to the pinion because it has fewer teeth and because pinion machining offers more adjustable parameters. A second-order surface is used to approximate the tooth-surface deviation, and the deviation is decomposed into five independent directions: spiral-angle correction, pressure-angle correction, tooth-length crowning, tooth-profile crowning, and tooth-surface torsion. The design vector is
$$ \mathbf{A}=(a_1,a_2,a_3,a_4,a_5), $$
where \(a_1\) is the spiral-angle correction, \(a_2\) the pressure-angle correction, \(a_3\) the tooth-length crowning, \(a_4\) the tooth-profile crowning, and \(a_5\) the tooth-surface torsion. All five coefficients are expressed in micrometers. The optimization problem is written as
$$
\begin{aligned}
\min_{\mathbf A}\quad & f_1(\mathbf A)=\sigma_{c,\max}(\mathbf A)\\
\text{s.t.}\quad
& -20\le a_1\le 20,\\
& 0\le a_2\le 55,\\
& 0\le a_3\le 20,\\
& 0\le a_4\le 55,\\
& -55\le a_5\le 5,
\end{aligned}
$$
where \(\sigma_{c,\max}\) is the maximum tooth-surface contact stress. The asymmetric range for \(a_5\) allows the torsion correction to act in either direction while keeping the search space bounded.
A design of experiments (DOE) was used to sample the modification space. An optimal Latin hypercube design was selected, with the number of design points estimated from
$$ N=\frac{(n+1)(n+2)}{2}, $$
where \(n\) is the number of input variables. For five variables, this gives 21 points. Thirty design points were used instead to provide a denser sample for surrogate construction. Each design point defines a modified gear model, which is then evaluated with the transient finite-element procedure described above.
Contact-Stress Sensitivity to Modification Parameters
The maximum contact stress from each modification case was extracted and used for sensitivity analysis. The resulting influence ranking is given in Table 2. Tooth-profile crowning is the dominant parameter, accounting for 42.57% of the sensitivity magnitude. Tooth-surface torsion is second at 22.07%, followed by pressure-angle correction at 15.29%, spiral-angle correction at 14.48%, and tooth-length crowning at 5.59%. All five parameters exceed 5%, so none can be treated as negligible within the studied design space; the practical issue is how to combine them.
| Coefficient | Modification direction | Range / μm | Sensitivity magnitude / % |
|---|---|---|---|
| \(a_1\) | Spiral angle | −20 to 20 | 14.48 |
| \(a_2\) | Pressure angle | 0 to 55 | 15.29 |
| \(a_3\) | Tooth-length crowning | 0 to 20 | 5.59 |
| \(a_4\) | Tooth-profile crowning | 0 to 55 | 42.57 |
| \(a_5\) | Tooth-surface torsion | −55 to 5 | 22.07 |
The dominance of tooth-profile crowning indicates that the profile curvature is a primary control on whether contact concentrates near the tooth tip or root. Tooth-surface torsion is the next most important because it changes the asymmetric distribution of the gap along the tooth surface, which is relevant when one end of the contact path carries more load than the other. Pressure-angle and spiral-angle corrections shift the contact position and orientation and therefore also affect the peak, but they are secondary to profile crowning and torsion in this design. Tooth-length crowning has the smallest sensitivity, yet its 5.59% contribution is still large enough that it should be optimized rather than fixed arbitrarily.
Surrogate Model and NSGA-II Optimization
A Kriging surrogate model with a Gaussian correlation function was built from the 30 finite-element cases to represent the relationship between the five modification coefficients and the maximum contact stress. The fitted model has a coefficient of determination \(R^2=0.933\) and a mean relative error of 0.051, indicating that the surrogate reproduces the finite-element results closely enough for optimization.
The modification parameters were optimized with the non-dominated sorting genetic algorithm with elitism, NSGA-II. The population size was 40, the number of generations was 400, the crossover probability was 0.9, the crossover distribution index was 10, and the mutation distribution index was 20. The objective was to minimize the maximum tooth-surface contact stress. The optimized coefficient set was then returned to the transient finite-element model for verification, so the reported stress reduction is based on a direct contact calculation rather than on the surrogate alone.
Optimized Contact Pattern and Stress Reduction
With the optimized modification, the maximum tooth-surface contact stress falls to 1,120 MPa, a reduction of about 80% from the 5,595 MPa baseline. More importantly, the contact pattern changes qualitatively. Instead of loading the tooth tip, the contact region lies in the middle of the tooth surface and forms an elliptical area. The maximum stress occurs at the center of this ellipse rather than at an edge. This shift explains the lower peak: the same load is distributed over a larger and more favorable contact area.
The stress histories before and after modification both show periodic fluctuation at the gear mesh frequency. After modification, the maximum contact stress is lower throughout the cycle, with the difference becoming especially clear in the middle and late stages of meshing. This is consistent with the removal of the tooth-tip edge contact that dominates the unmodified case. The optimized geometry does not eliminate the periodic nature of gear contact; it changes the contact pattern so that each mesh cycle operates at a lower stress level.
Design Implications for Spiral Bevel Gears
The results support a clear design priority for this class of spiral bevel gear. Tooth-profile crowning and tooth-surface torsion should be treated as the primary modification variables when the goal is to reduce peak contact stress. Pressure-angle and spiral-angle corrections provide useful secondary adjustments, while tooth-length crowning alone is unlikely to correct a severe edge-contact condition. A five-parameter optimization is more appropriate than a single-variable trial-and-error approach because the contact pattern is controlled by the combined gap distribution, not by one curvature component in isolation.
The surrogate-based workflow also has practical value. Once the DOE cases have been calculated, the Kriging model allows many candidate combinations to be screened without a full transient finite-element run for each one. The final candidate still requires finite-element verification, as was done here, because the surrogate is an approximation and because contact stress is sensitive to the actual loaded geometry.
The absolute stress values apply to the modeled geometry, material, mesh, and loading condition; they are not general allowable stresses for other spiral bevel gear designs. The broader finding is the sensitivity ranking and the contact-mechanics result: in this design space, edge contact at the tooth tip produced a peak stress of 5,595 MPa, and an optimized five-parameter modification moved the contact ellipse to the tooth center and reduced the peak to 1,120 MPa. For right-angle power transmissions in which compactness and load capacity are both required, that kind of contact-pattern control is a direct route to lower contact stress and improved reliability.