Spiral bevel gears are widely used where a right-angle or offset drive must transmit high torque smoothly. Their overlapping tooth action provides a high contact ratio, a stable instantaneous ratio, and low noise, which explains their presence in automotive axles, aerospace accessory drives, and mining machinery. In the Gleason system, the gear member is commonly cut by the formate method and the pinion by a generating or cutter-tilt method. The final tooth surface is a conjugate surface modified by machine settings, and the contact pattern is highly sensitive to small changes in tooth-surface curvature. Traditional development therefore relies on trial cuts: a blank is cut, checked on a rolling tester, and the machine settings are adjusted. That loop consumes material and retains a stochastic element.
A virtual manufacturing chain can replace part of that trial-and-error loop. The chain combines solid-model cutting simulation, finite element contact analysis, curvature-driven grinding-wheel design, virtual grinding, and a final rolling inspection. The case considered here is a Gleason-type hypoid spiral bevel gear pair with 43 gear teeth and 11 pinion teeth. Its main blank parameters are listed in Table 1.
Table 1. Main blank parameters of the gear pair.
| Parameter | Gear | Pinion |
|---|---|---|
| Number of teeth | 43 | 11 |
| Module | 7.00 mm | 7.00 mm |
| Hand | Right-hand | Left-hand |
| Mean pressure angle | 22.5° | 22.5° |
| Midpoint helix angle | 31.84° | 50° |
| Face width | 26.92 mm | 34.31 mm |
Cutting simulation and model accuracy
The gear member is cut by the formate method with a dual-face cutter. During cutting, the gear blank and the cutter have no generating relative motion; each boolean operation removes the material of one complete tooth slot. The blank is indexed by 8.372° after each cut, and 43 cuts produce the full gear. The pinion is cut by the cutter-tilt method with a single-face cutter, so the convex and concave flanks are produced separately. For the convex flank, the inner cutter rotates through 1° while the pinion blank rotates through 4.039°; 49 cutting steps generate the tool-mark traces. For the concave flank, the outer cutter rotates through 1° while the blank rotates through 3.4266°; 39 steps complete the trace pattern. The traces are then lofted into a surface, converted into a single tooth, and arrayed around the axis.
The accuracy of the resulting CAD model is checked against the theoretical tooth surface. Forty-five theoretical points are used: nine along the tooth-height direction and five along the face-width direction. Their distances to the simulated tooth surface are measured in Pro/E. The maximum deviation is 0.003 mm on the gear convex flank and 0.004 mm on the gear concave flank. This level of agreement supports the use of the cutting simulation as a geometric input for contact analysis.
Finite element contact of the unmodified tooth surface
For the contact analysis, the gear and pinion are assembled and simplified. The shafts are removed, and the gear is truncated to 12 teeth so that the 11-tooth pinion can complete one revolution while three tooth pairs remain available for contact. The simplification reduces the mesh size without removing the tooth-root and tooth-flank features required for contact.
The analysis uses a static general step in Abaqus. The pinion flank is the master surface and the gear flank is the slave surface. The contact friction coefficient is 0.1, taken from the gear design handbook. Reference points are placed on the gear and pinion axes and coupled to the inner rings. A resisting torque of 500 N·m is applied at the gear coupling point to represent the working load. The material and mesh settings are summarized in Table 2.
Table 2. Finite element settings for the contact analysis.
| Item | Setting |
|---|---|
| Material | 20CrMnTi |
| Density | 7,860 kg/m3 |
| Young’s modulus | 207,000 MPa |
| Poisson’s ratio | 0.3 |
| Element type | C3D8R |
| Contact friction coefficient | 0.1 |
| Applied torque | 500 N·m |
Contact stress plots are extracted at gear rotation angles of 2°, 4°, 6°, 8°, 10°, 12°, and 14° after the initial 0° position. On the unmodified gear flank, the contact pattern is not the desired central ellipse. Severe root contact and outer-end contact appear, and the contact line is long, covering much of the tooth surface. In curvature terms, the tooth-height curvature is too small, and the face-width curvature is not appropriate for a controlled elliptical contact. The contact pattern therefore needs correction before the gear can be accepted as a representative design.
Curvature calculation and grinding-wheel profile correction
The cut tooth surface is treated as a free-form surface. Let the surface be described by a two-parameter vector function
$$ \mathbf{r}=\mathbf{r}(i,j)=\bigl(x(i,j),\,y(i,j),\,z(i,j)\bigr). $$
The first fundamental form follows from the differential of the surface:
$$ ds^{2}=E\,di^{2}+2F\,di\,dj+G\,dj^{2}, $$
where \(E\), \(F\), and \(G\) are the first fundamental quantities. With the second fundamental quantities \(L\), \(M\), and \(N\), the normal curvature along a direction \(\lambda=dj/di\) is
$$ k_{n}(\lambda)=\frac{L+2M\lambda+N\lambda^{2}}{E+2F\lambda+G\lambda^{2}}. $$
The principal curvatures \(k_{1}\) and \(k_{2}\) are the extrema of \(k_{n}(\lambda)\). They can be written as
$$ k_{1,2}=H_{m}\pm\sqrt{H_{m}^{2}-K}, $$
with the mean curvature and Gaussian curvature
$$ H_{m}=\frac{EN-2FM+GL}{2(EG-F^{2})},\qquad K=\frac{LN-M^{2}}{EG-F^{2}}. $$
Changing the tooth-surface curvature changes the length and orientation of the contact line. The correction strategy is therefore to remove a thin layer of material from the tooth surface, starting from a fixed reference point \(M\) near the center and increasing the offset toward the tooth boundaries. The reference point keeps its spatial position; the other nodes move along their local surface normals. Three correction levels are defined, with different maximum offsets in the tooth-height and face-width directions.
Table 3. Three curvature-correction levels for the gear convex flank.
| Correction level | Maximum offset along tooth height | Maximum offset along face width | Relative material removal |
|---|---|---|---|
| Curvature 1 | 0.020 mm | 0.081 mm | Least |
| Curvature 2 | 0.0215 mm | 0.121 mm | Intermediate |
| Curvature 3 | 0.025 mm | 0.161 mm | Most |
For the largest correction, the offset decreases from the tooth top toward the root in a sinusoidal manner until it reaches 0.025 mm, and from the large end toward the small end until it reaches 0.161 mm. The node coordinates and normal vectors are extracted from the gear convex flank, and a program calculates the offset of each node. These offsets are then applied to the cutter-section coordinates to obtain the corresponding grinding-wheel section. This step is the practical link between the desired tooth-surface curvature and the grinding process: the wheel profile is corrected rather than the cutting model being rebuilt.
Virtual grinding and post-grinding contact
The corrected wheel is used in a five-axis virtual grinding environment built in VERICUT. The machine model follows a Gleason PHOENIX series spiral bevel gear grinder and uses a Fanuc control configuration. It includes the X, Y, and Z translation axes and two rotary axes for the wheel and workpiece, matching the motion capability of the actual machine. A bowl-shaped grinding wheel is created from the corrected section, and the G-code drives the wheel in the five-axis forming-grinding mode.
After virtual grinding, the gear flank has lost the designed layer of material, and the tooth slot is slightly wider. The flank is smoother than the cut surface, and the node positions have shifted according to the calculated offsets. The modified gear is then exported, assembled with the pinion, simplified in the same way, and analyzed in Abaqus with the same contact and load settings.
The post-grinding contact pattern is substantially different from the unmodified pattern. The contact spots form an ellipse and lie near the middle of the tooth in both the tooth-height and face-width directions. The severe root and outer-end concentrations of the unmodified gear are no longer present. For a spiral bevel gear pair, this central, limited elliptical contact is the desired form because it reduces edge loading and leaves room for load-dependent contact migration.
A rolling inspection is used to check whether the simulated pattern corresponds to the physical process. Red lead oxide is painted on the gear flank, and a motor drives the gear pair. As the teeth contact, the powder is wiped away and the contact region becomes visible. Before correction, the contact region on the gear flank is shifted toward the small end. After correction, it lies near the middle of the tooth and has an elliptical or rectangular shape. The simulated and observed patterns are consistent, which supports the validity of the machine settings used in the virtual model.
Implications for spiral bevel gear development
The case shows that tooth-surface curvature can be adjusted at the grinding stage without reconstructing the cutting model. The cutting simulation provides a high-precision geometric starting point; the finite element analysis identifies the contact defect; the curvature calculation converts that defect into node offsets; and the corrected wheel profile carries the offsets into virtual grinding. This sequence follows the actual process chain of cutting followed by grinding, so the virtual result is tied to a manufacturable wheel profile rather than to an abstract tooth-surface modification.
The practical value is a shorter development loop. Instead of repeated trial cuts with new blanks, the contact pattern can be screened by changing the wheel section and the grinding motion. The same workflow can be used to search for wheel profiles and machine parameters that place the contact ellipse correctly and reduce transmission error. The present case carries the largest correction level through grinding, contact analysis, and rolling inspection at a single torque of 500 N·m. A production design would still need a map over the intended torque and speed range, but the curvature-adjusted virtual chain provides a coherent way to generate and test candidate spiral bevel gear designs before physical trials.