Spiral bevel gears transmit motion between intersecting shafts through curved teeth that engage gradually and carry high contact loads. Their use in aerospace, automotive, and energy transmissions places severe demands on tooth-surface integrity. Grinding is often the final shaping operation, and the residual stress field it leaves behind influences crack initiation, tooth-root fatigue, load-carrying capacity, and service life. A compressive layer near the surface can restrain crack opening, whereas a tensile zone can accelerate surface-initiated failure. The central process question is therefore not simply whether grinding produces compressive or tensile stress, but how the stress profile through the tooth depth responds to grit geometry, depth of cut, and wheel speed.
Grit-scale removal and the origin of residual stress
A grinding wheel behaves as a multi-point cutting tool. Each active abrasive grit removes a small volume of material through three sequential regimes: rubbing, plowing, and chip formation. In the rubbing stage, the grit contacts the tooth surface elastically and generates friction heat. In the plowing stage, the grit penetrates and pushes material sideways, producing plastic flow, side ridges, and strain hardening. In the chip-formation stage, the penetration depth exceeds a critical value, shear localizes ahead of the grit, and a chip flows over the rake face.
The residual stress that remains after cooling is a path-dependent consequence of this history. Mechanical loading imposes local compression and plastic strain; the temperature rise causes thermal expansion against the surrounding colder material; and elastic recovery after the grit passes redistributes the incompatible strains. The final profile is the balance among these contributions. In the single-grain model discussed here, the balance favors compression near the tooth surface, with a maximum compressive value below the surface and a tensile balancing zone deeper in the material. This outcome is specific to the modeled grinding conditions: a stronger thermal cycle or a different cooling path can shift the near-surface stress toward tension, so the sign and magnitude of the surface stress should not be generalized without checking the process window.
Finite-element representation of single-grain grinding
The grit was represented as a rigid conical indenter 0.07 mm high, with a 90° apex angle and a 0.01 mm edge radius. The workpiece was a small AISI 4340 steel block measuring 0.4 mm × 0.3 mm × 0.15 mm. The bottom of the block was fixed, the grit was treated as rigid, and the tooth material was treated as an elastic-plastic body. A penalty friction model with a friction coefficient of 0.3 represented the grit-workpiece interface. The grinding stage was solved with Abaqus/Explicit, and the subsequent cooling stage was solved with Abaqus/Standard after the tool and contact definition were removed.
AISI 4340 steel was described with the Johnson-Cook flow-stress law and the Johnson-Cook fracture criterion. The flow stress is
$$
\sigma=(A+B\varepsilon^{n})\left(1+C\ln\frac{\dot{\varepsilon}}{\dot{\varepsilon}_{0}}\right)\left[1-\left(\frac{T-T_{0}}{T_{m}-T_{0}}\right)^{m}\right]
$$
where A is the yield stress, B is the hardening modulus, n is the strain-hardening exponent, C is the strain-rate sensitivity, m is the thermal-softening exponent, T0 is room temperature, and Tm is the melting temperature. The parameters used for the workpiece are listed in the table below.
| Parameter group | Values |
|---|---|
| Flow stress | A = 792 MPa; B = 510 MPa; n = 0.26; C = 0.014; m = 1.03 |
| Fracture | D1 = 0.05; D2 = 3.44; D3 = -2.12; D4 = 0.002; D5 = 0.61 |
Residual stress was evaluated after cooling from the normal stress component S11, which acts along the grinding direction. At each depth layer, 170 equally spaced nodal values were averaged, with strongly disturbed regions excluded. This spatial averaging reduces the influence of local element-scale fluctuations and gives a representative stress-depth profile.
Residual stress profile through the tooth depth
After cooling, the computed stress profile follows a characteristic sequence: compressive stress at the surface, a maximum compressive value in the subsurface layer, a transition to tensile stress, and a gradual decay toward zero in the unaffected bulk. Compressive stress is often reported as a magnitude, while finite-element output stores compression with a negative sign. At the optimized condition, the maximum compressive magnitude reached 638.6 MPa and the maximum tensile stress was 24.9 MPa.
The near-surface compression is consistent with the local mechanical action of the grit: the tooth material is compressed and plastically deformed directly beneath the contact. The subsurface maximum reflects the combined influence of constrained plastic flow and the thermal cycle, which peaks below the immediately contacted layer. Deeper in the material, the mechanical influence decreases and elastic recovery of the surrounding material produces a tensile balancing zone. This explanation is mechanistic rather than a unique decomposition: the model does not separately isolate the mechanical and thermal contributions, and the relative weight of the two can change with grinding speed, cooling intensity, and material condition.
Parameter sensitivity
Single-factor simulations were used to rank the influence of grinding speed, grinding depth, and abrasive rake angle. The factor levels and observed trends are summarized below.
| Factor | Levels examined | Observed effect on residual stress |
|---|---|---|
| Grinding speed, vs | 10, 15, 20, 25 m/s | Increases with speed, but the effect is the weakest of the three factors |
| Grinding depth, ap | 0.01, 0.02, 0.03, 0.04 mm | Increases with depth; second in influence |
| Abrasive rake angle, γ | -20, -30, -40, -50° | Increases as the angle becomes more negative; strongest influence |
At a fixed speed of 25 m/s and a depth of 0.01 mm, changing the rake angle from -20° to -50° produced the largest change in residual stress. As the rake angle becomes more negative, the grit removes material less by sharp cutting and more by pushing and plowing. The plastic zone extends over a larger volume and to a greater depth, so the incompatible strain left after the grit passes increases. The result is a stronger residual stress field, including a larger compressive magnitude and a larger balancing tensile stress.
At a fixed speed of 25 m/s and a rake angle of -20°, increasing the grinding depth from 0.01 mm to 0.04 mm increased the residual stress. A deeper cut enlarges the contact area between the grit and the tooth, raises the grinding force, and drives plastic deformation further into the workpiece. The material removal rate also increases, so the surface experiences more intense cutting and compression per unit time. Among the three factors, depth ranks second because it changes both the force amplitude and the volume of deformed material.
At a fixed depth of 0.04 mm and a rake angle of -20°, increasing the speed from 10 m/s to 25 m/s increased the residual stress, but the change was modest. Higher speed brings more grits into contact per unit time and tends to reduce the undeformed chip thickness per grit, which can lower the force on an individual grit. At the same time, the increased sliding velocity raises heat generation and thermal stress. These opposing effects explain why speed is the weakest of the three factors in this single-grain model, even though it remains technologically important through its influence on wheel wear and thermal damage.
Multi-grit interaction and model scope
A three-grit model with randomly distributed abrasives was also evaluated at 25 m/s, 0.01 mm depth, and a -50° rake angle, using the same material properties, analysis steps, mesh size, and contact constraints as the single-grain case. The multi-grit model produced a lower maximum compressive stress than the single-grain model, while the stress-depth trend remained similar. The reduction occurs because adjacent grits generate overlapping stress fields: a compressive region produced by one grit can coincide with a tensile region produced by a neighboring grit, so part of the compression is cancelled.
This comparison supports the use of single-grain results as a fast screening map for parameter trends. It does not make the single-grain model a quantitative substitute for a full wheel. Actual wheel topography, grit spacing, grit wear, and repeated passes all affect the superposition of stress fields, and those effects become more important as the contact area and the number of active grits increase.
Response-surface model and multi-objective optimization
A 15-run response-surface design was used to relate the two residual-stress responses to the three grinding parameters over the ranges 10–25 m/s, 0.01–0.04 mm, and -50° to -20°. Quadratic regression models were fitted to the maximum compressive stress S1 and the maximum tensile stress S2. Terms with P > 0.05 were removed, and the reduced models were checked by analysis of variance and normal residual plots.
| Response | Model F | R² | Adjusted R² | Factor F: speed / depth / rake |
|---|---|---|---|---|
| Maximum compressive stress, S1 | 47.96 | 97.29% | 95.27% | 15.73 / 40.30 / 91.89 |
| Maximum tensile stress, S2 | 57.67 | 97.74% | 96.05% | 5.66 / 9.48 / 18.71 |
Both models were significant at P < 0.0001, and all retained factor terms had P < 0.05. The F values confirm the same ranking for both responses: abrasive rake angle > grinding depth > grinding speed. The high R² and adjusted R² values indicate that the quadratic surfaces reproduce the simulation data closely, while the residual plots show no strong departure from normal behavior.
The optimization objective was to maximize the compressive magnitude S1 while minimizing the tensile stress S2, subject to the parameter bounds:
$$
\max S_{1},\quad \min S_{2},\quad 10\le v_{s}\le25\ \mathrm{m/s},\;0.01\le a_{p}\le0.04\ \mathrm{mm},\;-50^{\circ}\le\gamma\le-20^{\circ}
$$
The best compromise was obtained at a grinding speed of 25 m/s, a grinding depth of 0.01 mm, and a rake angle of -48°. This condition gave a maximum compressive stress of 638.6 MPa and a minimum tensile stress of 24.9 MPa. The result combines a shallow cut, which limits the deformed volume and the tensile balancing field, with a strongly negative rake angle, which promotes compression, and a relatively high speed, which has only a weak effect on the stress magnitude.
| Case | Speed (m/s) | Depth (mm) | Rake angle (°) | Compressive S1 (MPa) | Tensile S2 (MPa) |
|---|---|---|---|---|---|
| Optimized | 25 | 0.01 | -48 | 638.6 | 24.9 |
| 1 | 30 | 0.01 | -48 | 676.1 | 30.7 |
| 2 | 25 | 0.02 | -48 | 657.3 | 29.5 |
| 3 | 25 | 0.01 | -45 | 610.9 | 19.8 |
| 4 | 25 | 0.01 | -50 | 662.4 | 30.2 |
| 5 | 20 | 0.01 | -48 | 624.4 | 22.1 |
The control cases show that the optimum is a compromise rather than a single-factor maximum. A 30 m/s case raised the compressive magnitude to 676.1 MPa, but it also raised the tensile stress to 30.7 MPa. A -45° rake reduced the tensile stress to 19.8 MPa, but it lowered the compressive magnitude to 610.9 MPa. The selected -48° condition sits between these extremes, maintaining a high compressive magnitude while keeping the tensile stress relatively low. This trade-off is important in gear design: a larger compressive layer is not automatically better if it is accompanied by a stronger tensile zone that promotes crack initiation or subsurface damage.
Engineering implications and limits
For spiral bevel gear grinding, the results point to a clear hierarchy. Abrasive rake geometry and grinding depth should be treated as primary controls because they change the plastic zone size and the force transmitted into the tooth. Grinding speed is better used as a fine-tuning variable, but it cannot be ignored because it affects heat generation, wheel wear, and thermal damage. A shallow depth combined with a strongly negative rake angle favors compression, while excessive depth or a less negative rake angle tends to intensify both the compressive and tensile parts of the profile.
The single-grain model is computationally efficient and useful for ranking parameters, constructing response surfaces, and screening process windows before full wheel trials. Its absolute stress values, however, depend on the assumed conical grit, the small workpiece volume, the simplified cooling boundary, and the absence of grit wear, wheel topography, repeated passes, and microstructural phase transformations. The predicted profile should therefore be calibrated against measured residual stress data before being used as a production acceptance criterion. Within those limits, the analysis provides a consistent mechanistic basis for selecting grinding parameters that increase beneficial compression while limiting the tensile stress that drives fatigue damage.