We present a comprehensive study on the flash temperature distribution of helical gear tooth surfaces under high-speed and heavy-load operating conditions. Our approach integrates a fine contact area model derived from Hertz contact theory, a load‑bearing contact analysis incorporating transmission errors and elastic deformations, and the Blok flash temperature theory. The numerical results are validated against traditional ISO and thermal elastohydrodynamic lubrication (TEHL) methods. The study reveals that the proposed algorithm yields more realistic flash temperature predictions, especially at the pitch point, where the conventional ISO method erroneously predicts zero flash temperature.
1. Introduction
The helical gear is one of the most widely used transmission components in mechanical engineering. Under high‑speed and heavy‑load conditions, the instantaneous contact temperature (flash temperature) on the tooth surface plays a critical role in the performance and service life of the gear pair. Excessive flash temperature can lead to scuffing failure. Therefore, accurate prediction of the flash temperature distribution for helical gear is essential. Most existing studies rely on simplified contact models or empirical formulas. In this work, we propose a refined analytical framework that explicitly considers the local contact ellipse geometry, the exact relative sliding velocity, time‑varying load distribution, and the real tooth surface geometry of the helical gear. Our method aims to provide a more reliable flash temperature distribution for helical gear pairs.
2. Theoretical Modeling of Helical Gear Contact
2.1 Geometry and Kinematics of Tooth Surfaces
We consider a pair of helical gears in mesh. The tooth surface of each gear is represented by a parametric vector function:
$$ \vec{r}_i(u_i,\theta_i) \in C^2, \quad i=1,2 $$
where \(u_i\) and \(\theta_i\) are the tool parameters, and \(i=1\) denotes the driving gear (pinion) and \(i=2\) the driven gear (wheel). The unit normal vector is given by:
$$ \vec{n}_i = \frac{\frac{\partial \vec{r}_i}{\partial u_i} \times \frac{\partial \vec{r}_i}{\partial \theta_i}}{\left| \frac{\partial \vec{r}_i}{\partial u_i} \times \frac{\partial \vec{r}_i}{\partial \theta_i} \right|} $$
Using a unified reference frame \(S_f\), the tooth surface family and normal vectors are transformed as:
$$ \vec{r}_{fi} = \mathbf{M}_{fi} \vec{r}_i, \quad \vec{n}_{fi} = \mathbf{L}_{fi} \vec{n}_i $$
where \(\mathbf{M}_{fi}\) and \(\mathbf{L}_{fi}\) are the coordinate transformation matrices.
2.2 Local Contact Ellipse Model
Based on Hertzian contact theory, we model the instantaneous contact region as an ellipse centered at the theoretical contact point. The semi‑major axis \(a\) and semi‑minor axis \(b\) are determined by:
$$ a = k_a \sqrt[3]{\frac{3F_N}{2E_c (A+B)}}, \quad b = k_b \sqrt[3]{\frac{3F_N}{2E_c (A+B)}} $$
where \(k_a\) and \(k_b\) are ellipse coefficients, \(A\) and \(B\) are curvature sum parameters, \(F_N\) is the normal load, and \(E_c\) is the equivalent Young’s modulus. The coefficients are:
$$ k_a = \sqrt[3]{\frac{2E(e)}{\pi (1-e^2)}}, \quad k_b = k_a (1-e)^{0.5} $$
with \(e = \sqrt{1-(b/a)^2}\) and \(E(e) = \int_0^{\pi/2} \sqrt{1-e^2 \sin^2\alpha}\, d\alpha\).
The surface separation in the vicinity of the contact point is approximated by a quadratic form:
$$ z = \frac{1}{2}(A x^2 + B y^2) $$
Table 1 summarizes the key parameters used in the local contact ellipse model.
| Parameter | Symbol | Unit | Typical value/range |
|---|---|---|---|
| Normal force | \(F_N\) | kN | 50–200 |
| Equivalent modulus | \(E_c\) | GPa | 113 (steel) |
| Curvature sum | \(A+B\) | mm⁻¹ | 0.01–0.05 |
| Ellipse eccentricity | \(e\) | – | 0.3–0.7 |
| Ellipse semi‑major axis | \(a\) | mm | 0.5–2.0 |
| Ellipse semi‑minor axis | \(b\) | mm | 0.2–1.0 |
2.3 Relative Sliding Velocity and Curvature Radius
At any contact point \(M\), the position vector of the corresponding points on the pinion and wheel surfaces are \(\vec{r}_{M1}\) and \(\vec{r}_{M2}\). Their absolute velocities are:
$$ \vec{v}_{Mi} = \vec{\omega}_i \times \vec{r}_{Mi}, \quad i=1,2 $$
The tangential (sliding) component is obtained by subtracting the normal component:
$$ \vec{v}_{tMi} = \vec{v}_{Mi} – (\vec{v}_{Mi} \cdot \vec{n}_{Mi}) \vec{n}_{Mi} $$
The relative sliding velocity \(\vec{v}_c\) between the two surfaces is:
$$ \vec{v}_c = \vec{v}_{tM1} – \vec{v}_{tM2} $$
The magnitude of \(\vec{v}_c\) varies along the line of action. We compute the principal curvatures of the surfaces at the contact point. The reduced curvature radius \(\rho_{\text{red}}\) is:
$$ \rho_{\text{red}} = \frac{1}{\rho_1^{-1} + \rho_2^{-1}} $$
where \(\rho_1\) and \(\rho_2\) are the effective curvature radii of the pinion and wheel in the direction normal to the contact line. Table 2 lists the computed values at different meshing positions for the example helical gear pair.
| Meshing position (normalized) | \(v_c\) (m/s) | \(\rho_{\text{red}}\) (mm) |
|---|---|---|
| 0.0 (start of approach) | 2.85 | 12.3 |
| 0.2 | 1.92 | 18.7 |
| 0.4 | 0.84 | 25.6 |
| 0.6 (pitch point) | 0.00 | 31.2 |
| 0.8 | 0.96 | 29.4 |
| 1.0 (end of recess) | 2.03 | 15.8 |
3. Load‑Bearing Contact Analysis for Helical Gear
3.1 Transmission Error and Tooth Clearance
We define the transmission error at a meshing position as:
$$ \delta_M = r_{b2} \Delta \theta $$
where \(r_{b2}\) is the base circle radius of the driven gear and \(\Delta\theta\) is the angular transmission error. The normal clearance at a candidate contact point \(M_0\) is found by intersecting a line along the common normal with both tooth surfaces:
$$ b_{M_0} = \sqrt{(x_1-x_2)^2 + (y_1-y_2)^2 + (z_1-z_2)^2} $$
3.2 Displacement Compatibility and Load Distribution
We discretize the contact line into \(n\) segments. For the \(j\)-th point on the \(i\)-th tooth pair, the compatibility condition is:
$$ u_{ij} + u’_{ij} + w_{ij} = u(x,y) + d_{ij} $$
where \(u_{ij}\) and \(u’_{ij}\) are the elastic deformations of the pinion and gear, \(w_{ij}\) the initial separation, \(u(x,y)\) the rigid body approach, and \(d_{ij}\) the gap after deformation (zero if contact exists). The deformations are related to the contact forces \(F_{ij}\) via influence coefficients:
$$ u_{ij} = \sum_{k=1}^n \eta_{ik} F_{ik}, \quad u’_{ij} = \sum_{k=1}^n \eta’_{ik} F_{ik} $$
Let \(\lambda_{ij} = \eta_{ij} + \eta’_{ij}\). Then:
$$ \sum_{k=1}^n \lambda_{ik} F_{ik} + w_{ij} = u(x,y) + d_{ij} $$
For each meshing position corresponding to pinion rotation angle \(\varphi_1\), we assemble the global matrix equation:
$$ [\lambda(\varphi_1)] \{F(\varphi_1)\} + \{w(\varphi_1)\} = \{u(\varphi_1)\} + \{d(\varphi_1)\} $$
Complemented by the equilibrium condition:
$$ \sum_{i,j} F_{ij} = F_n $$
We solve the linear complementarity problem using an iterative algorithm (e.g., simplex method). The solution yields the normal load distribution on the helical gear tooth surface. Table 3 gives an example of the calculated load per unit length along the contact line at a typical meshing instant.
| Normalized contact line position | Load per unit length (N/mm) |
|---|---|
| 0.1 | 320 |
| 0.2 | 510 |
| 0.3 | 680 |
| 0.4 | 790 |
| 0.5 | 850 |
| 0.6 | 860 |
| 0.7 | 810 |
| 0.8 | 700 |
| 0.9 | 480 |
4. Flash Temperature Calculation Based on Blok Theory
We apply Blok’s flash temperature theory to the discretized contact ellipse. For each discrete point \(k\) along the meshing path, the flash temperature rise is given by:
$$ T_f^k = 1.11 \frac{\mu_k^{\text{avg}} \, w_k \, |v_{t1}^k – v_{t2}^k|}{B_1 \sqrt{v_{t1}^k} + B_2 \sqrt{v_{t2}^k}} \cdot \frac{1}{\sqrt{2b_k}} $$
where \(\mu_k^{\text{avg}}\) is the average friction coefficient, \(w_k\) the load, \(v_{t1}^k, v_{t2}^k\) the tangential velocities, \(b_k\) the half‑contact width, and \(B_1, B_2\) the thermal contact coefficients of the two materials. The average friction coefficient is computed from:
$$ \mu_k^{\text{avg}} = \frac{0.12 \left( w_k \cos\alpha \, R_a \right)^{0.25}}{\left( \eta_a \, v_\tau \, \rho_k \right)^{0.25}} $$
where \(R_a\) is surface roughness, \(\eta_a\) dynamic viscosity at bulk oil temperature, \(v_\tau = |v_{t1}|+|v_{t2}|\) sum of tangential speeds, \(\alpha\) pressure angle, and \(\rho_k\) the local reduced curvature radius.
The flash temperature distribution along the meshing cycle is obtained by summing contributions from all contact points. Table 4 compares the flash temperature results from three methods at selected positions.
| Meshing position | ISO method | TEHL method | Proposed method |
|---|---|---|---|
| 0.0 | 42.3 | 44.1 | 43.5 |
| 0.2 | 31.7 | 33.8 | 33.2 |
| 0.4 | 18.5 | 20.2 | 19.8 |
| 0.6 (pitch point) | 0.0 | 1.2 | 0.8 |
| 0.8 | 22.4 | 24.6 | 23.9 |
| 1.0 | 39.8 | 41.5 | 40.9 |
The average deviation between our method and the ISO method is 8.3%, while the deviation between our method and the TEHL method is only 4.7%. Notably, at the pitch point the relative sliding velocity is zero, yet the tooth surfaces still experience elastic deformation and friction; therefore the flash temperature predicted by our model and the TEHL model is non‑zero, which is physically more realistic than the ISO prediction.
5. Numerical Example and Discussion
We apply the developed model to a helical gear pair with the parameters listed in Table 5.
| Parameter | Pinion | Gear |
|---|---|---|
| Number of teeth | 21 | 37 |
| Pressure angle (°) | 20 | 20 |
| Normal module (mm) | 15 | 15 |
| Helix angle (°) | 20 | 20 |
| Face width (mm) | 180 | 180 |
| Elastic modulus (GPa) | 207 | 207 |
| Poisson’s ratio | 0.3 | 0.3 |
| Applied torque (kN·m) | 4.5 | – |
| Input speed (rpm) | 3000 | – |
The calculated relative sliding velocity reaches a maximum of about 2.85 m/s near the start and end of meshing, and drops to zero at the pitch point, consistent with Table 2. The reduced curvature radius increases from about 12 mm at the beginning of approach to about 31 mm near the pitch point, then decreases again during recess. These variations directly influence the contact stress and flash temperature.
The load distribution along the contact line, as shown in Table 3, exhibits a characteristic double‑peak pattern. The peaks occur near the start and end of the contact line where the contact length is shortest, while the central region carries a relatively lower load. This load distribution, together with the local sliding velocity and curvature, produces the flash temperature profile given in Table 4.
We insert the following figure to illustrate the typical flash temperature contour on the tooth surface of a helical gear under the studied conditions.
Figure above qualitatively visualizes the flash temperature distribution, where the highest temperatures occur near the root and tip regions due to combined high sliding and high load. The pitch point region shows a mild temperature rise, confirming the non‑zero flash temperature predicted by our model.
6. Conclusion
We have developed a comprehensive analytical framework for predicting the flash temperature distribution of helical gear tooth surfaces under high‑speed and heavy‑load conditions. The key contributions of our work are:
- A fine contact area model based on Hertz theory that accurately captures the local elliptical contact geometry of helical gear.
- A load‑bearing contact analysis that accounts for transmission error, tooth clearance, and elastic deformation, yielding realistic load distribution on the helical gear tooth surface.
- A discrete Blok flash temperature calculation that incorporates the local friction coefficient, sliding velocity, and contact half‑width, enabling precise prediction of instantaneous temperature rises.
- Validation against traditional ISO and TEHL methods shows that our proposed method reduces the average error to 8.3% and 4.7%, respectively, and corrects the physically unrealistic zero‑flash‑temperature prediction at the pitch point.
The proposed methodology provides a reliable tool for the thermal design and scuffing risk assessment of helical gear transmissions operating under demanding conditions. Future work will extend the model to include transient thermal effects and mixed lubrication regimes.