Spiral bevel gears transmit torque between intersecting shafts through a curved tooth trace. Their contact imprint—its size, position, and shape—controls load sharing, transmission error, noise, and fatigue life. In high-speed and high-load applications, the design problem is not only to generate a conjugate flank but to predict how that flank behaves when the gear pair deflects under load. A geometrically exact model is therefore a prerequisite for reliable finite element contact analysis.
The modeling route follows the Gleason face-milling principle. A virtual generating gear meshes with the workpiece without clearance, and the cutter blade sweeps a conical surface that maps onto the tooth flank. Machine settings such as radial cutter position, angular cutter position, cradle angle, and workpiece mounting angle are carried through the coordinate transformation, so the resulting surface remains tied to the manufacturing process rather than to a purely geometric approximation.
Gleason Generation and Tooth-Surface Geometry
The cutting tool is represented in its own coordinate system Sq(Xq, Yq, Zq). For the straight cutting edge, the cutter cone Σq is parameterized by the edge coordinate Sq and the angular coordinate θq:
$$
\mathbf r_q(S_q,\theta_q)=
\begin{bmatrix}
(R_q-S_q\sin\alpha)\cos\theta_q\\
(R_q-S_q\sin\alpha)\sin\theta_q\\
-S_q\cos\alpha\\
1
\end{bmatrix}
$$
where Rq is the cutter tip radius and α is the tooth profile angle. The unit normal follows from the cross product of the two surface tangents:
$$
\mathbf n_q(\theta_q)=
\frac{\partial \mathbf r_q/\partial S_q \times \partial \mathbf r_q/\partial \theta_q}
{\left\|\partial \mathbf r_q/\partial S_q \times \partial \mathbf r_q/\partial \theta_q\right\|}
=
\begin{bmatrix}
\cos\alpha\cos\theta_q\\
\cos\alpha\sin\theta_q\\
-\sin\alpha
\end{bmatrix}
$$
The fillet region is generated by a circular arc of radius ρq with center coordinates Cx and Cy. Its normal is obtained from the same cross-product construction. This distinction matters because the working flank and the transition fillet are not produced by the same edge geometry, and the transition region influences stress concentration near the tooth root.
The cutter surface is then transformed into the workpiece coordinate system. If [M2] is the position transformation matrix and [L2] is its rotational part, the flank point and normal are
$$
\mathbf r_2=[M_2]\mathbf r_q,\qquad
\mathbf n_2=[L_2]\mathbf n_q
$$
During generation, the cutter and workpiece maintain a meshing relationship. In the machine coordinate system Sm2, the meshing equation is
$$
f_2=\mathbf n_{m2}\cdot \mathbf v_{m2}^{(q2)}=0
$$
where nm2 is the unit normal of the cutter-generated cone and vm2(q2) is the relative velocity between the cutter surface and the workpiece. Solving this equation together with the surface parameterization produces discrete flank points. Interpolation, trimming, and lofting then convert the point set into continuous tooth surfaces. The same construction applies to the gear concave flank and to the pinion, with appropriate changes in blade geometry and machine settings.
Parametric Solid Model and Assembly
The mathematical flank is converted into a solid model in SolidWorks. The process begins with a two-dimensional sketch of the gear blank, followed by generation of the three-dimensional blank. Tooth-slot point-cloud data are imported along the spiral trace, and a loft cut removes material to form a single slot. A circular pattern then reproduces the slot around the blank. Changing the tooth count, module, spiral angle, face width, and cone angles regenerates the corresponding gear and pinion models without rebuilding the entire feature tree.
The gear pair used in the contact analysis has 29 gear teeth and 12 pinion teeth, a 9 mm transverse module, a 35° spiral angle, a 90° shaft angle, and a 46 mm face width. The principal design parameters are summarized below.
| Parameter | Gear, 29 teeth | Pinion, 12 teeth |
|---|---|---|
| Number of teeth | 29 | 12 |
| Transverse module | 9 mm | 9 mm |
| Spiral angle | 35° | 35° |
| Shaft angle | 90° | 90° |
| Face width | 46 mm | 46 mm |
| Hand | Right | Left |
| Addendum | 4.74 mm | 10.56 mm |
| Dedendum | 12.25 mm | 6.43 mm |
| Pitch cone angle | 67.31° | 22.28° |
| Root cone angle | 62.33° | 19.52° |
| Face cone angle | 70.70° | 27.26° |
The gear and pinion are assembled at the design shaft angle and center distance. A dynamic interference check in the SolidWorks motion module is used to remove unintended contact caused by assembly error before the model is transferred to finite element analysis.
Finite Element Contact Setup
The contact model uses a seven-tooth sector to reduce computational cost while retaining the meshing region of interest. The material properties are an elastic modulus of 202 GPa, a Poisson ratio of 0.273, and a density of 7.91 g/cm³. Adaptive tetrahedral meshing is applied: the contact zone uses a refined element size of 1 mm, while non-contact regions use a coarser global size. The resulting sector contains 228,332 nodes and 133,453 elements.
The gear pair is modeled with face-to-face contact, a friction coefficient of 0.2, and the Augmented Lagrange algorithm. The pinion rotates at 154 rad/s. The rated torque on the gear is 8000 N·mm. Three torque levels—400, 8000, and 15,000 N·mm—are applied to represent a load spectrum.
Two solution strategies are compared. The steady contact analysis applies a step load, releases inertia, and reaches static equilibrium quickly; it is efficient for evaluating the basic contact state but omits dynamic impact. The transient analysis applies a ramp load to reduce shock. For the 8000 N·mm case, the first load step runs from 0 to 1 s and is divided into 1000 time steps, with Δt = 1 ms and a load increment of 8 N·mm per step. The second step runs from 1 to 3 s at constant load with automatic time stepping.
Contact Imprint and Stress Response
At 8000 N·mm, the steady analysis gives a peak contact stress of 9.90 × 109 Pa. The transient analysis gives 1.47 × 1010 Pa, about 48% higher. The difference shows that inertia and dynamic impact amplify the local stress when the load is applied. The steady step loading also produces a higher initial stress than the transient ramp, which indicates that the loading path itself affects the predicted contact state.
The transient response shows a pronounced stress fluctuation near 0.6 s. This timing corresponds to the transition in which the next tooth pair enters mesh while the previous pair leaves. The change in contact conditions produces a complex dynamic response and a local stress peak. Both steady and transient analyses produce an elliptical, tapered imprint, but the transient case also captures how the contact region expands as the load rises. That expansion is not visible in a quasi-static snapshot.
As the applied torque increases, the contact imprint grows nonlinearly and its center shifts toward the large end of the tooth. The contact zone is concentrated in the mid-flank region and initially biased toward the small end. Elastic deformation under load enlarges the contact area and moves the center toward the heel. The overall contact pattern remains stable, which supports the consistency of the tooth-surface topology. At the light-load condition, the pinion still shows a relatively large contact area; this is attributed to the lower contact stiffness of the soft-flank design rather than to a loss of geometric conformity.
Fatigue Life and Load Dependence
Fatigue assessment uses the strain–life (ε–N) framework in the ANSYS fatigue tool. The minimum computed life values are 333.69 at 400 N·mm, 77.82 at 8000 N·mm, and 18.78 at 15,000 N·mm. The reduction is strongly nonlinear with load. At 400 N·mm, surface fretting wear dominates the life estimate. At 8000 N·mm, the contact stress field becomes more severe. At 15,000 N·mm, non-proportional cyclic hardening appears through the stress–strain hysteresis loop, accelerating fatigue damage and causing a sharp drop in the minimum life value.
| Pinion speed | Gear torque | Observed contact behavior | Minimum life value |
|---|---|---|---|
| 154 rad/s | 400 N·mm | Large pinion contact area under light load; imprint in the mid-flank region | 333.69 |
| 154 rad/s | 8000 N·mm | Imprint expands; transient peak exceeds steady peak | 77.82 |
| 154 rad/s | 15,000 N·mm | Imprint grows further and shifts toward the heel; non-proportional cyclic hardening | 18.78 |
Rolling Contact Verification
The finite element results are checked on a universal rolling contact tester. The gear pair is mounted with adjustable offset and axial positions. The offset error ΔE, gear axial displacement ΔG, pinion axial displacement ΔP, and shaft angle Σ are set with handwheels and monitored with dial indicators and an electronic angle meter. The controlled installation errors are ΔE ≤ ±0.03 mm, ΔG and ΔP ≤ ±0.05 mm, and Σ ≤ ±40″.
| Parameter | Adjustment step | Resolution | Controlled accuracy | Monitoring |
|---|---|---|---|---|
| ΔE | ±0.01 mm | 0.01 mm | ≤ ±0.03 mm | Dial indicator |
| ΔG, ΔP | ±0.02 mm | 0.01 mm | ≤ ±0.05 mm | Dial indicator |
| Σ | ±10″ | 10″ | ≤ ±40″ | Electronic angle meter |
Before loading, the gear teeth are coated with marking pigment. The test starts at 20 rad/s for 1 min to seat the assembly, then accelerates at 0.08 rad/s² to 154 rad/s. A PID controller raises the torque at 200 N·mm/s to the target values of 400, 8000, and 15,000 N·mm. Each condition is held for 5 min, after which the machine is stopped and the tooth-surface imprint is inspected.
At all three torque levels, the measured imprint position and orientation agree with the finite element prediction. The experimental contact zones fall within the design requirement for the contact region, and the stress distribution pattern is consistent with the simulation. This agreement supports the use of the parametric model and the transient contact procedure for design evaluation within the tested operating range.
Implications for Design and Manufacturing
The main value of the workflow is the direct link between machine settings and the solid model. Once the cutter geometry and machine parameters are defined, the flank is generated as a process-consistent surface rather than a manually approximated shape. This reduces the iteration needed when tooth count, spiral angle, face width, or cone angles change, and it provides a reliable geometry for contact simulation.
The transient results also show why a steady contact analysis alone is insufficient for dynamic load cases. Inertia and the transition between tooth pairs raise the local stress peak, while the contact imprint expands and shifts with torque. These effects matter for load sharing and for fatigue assessment. The strain–life results further show that life reduction is not proportional to torque: the high-load condition introduces cyclic hardening and a much shorter minimum life than a simple load-scaling estimate would suggest.
Within the tested range of 154 rad/s and 400–15,000 N·mm, the combination of Gleason-based parametric modeling, transient finite element contact analysis, and rolling contact inspection provides a coherent basis for spiral bevel gear design optimization, process planning, and life prediction.