As a fundamental power transmission component, spur gears are ubiquitous in industrial machinery. The operational demands for higher loads and speeds in modern applications inevitably increase their susceptibility to failure. Among various failure modes, root cracks are particularly prevalent, accounting for a significant proportion of gear-related faults. The accurate assessment of gear health, therefore, hinges on precise modeling of fault signatures, with the time-varying mesh stiffness (TVMS) being a critical diagnostic parameter. This intrinsic characteristic not only governs the dynamic response but also sensitively reflects the integrity of the gear tooth structure. While the potential energy method remains a popular analytical approach due to its computational efficiency, its application to spur gears with large tooth numbers and root cracks requires careful refinement to align with geometric and physical realities. This article details an improved methodology for calculating the TVMS of such spur gears, addressing key modeling oversights in traditional energy methods and establishing comprehensive fault models for varying crack severities.
The core of the analytical calculation for spur gears lies in accurately modeling the tooth as a non-uniform cantilever beam. The total compliance during meshing stems from several energy storage mechanisms: Hertzian contact energy, bending energy, shear energy, axial compressive energy, and the foundation (or fillet) energy. The mesh stiffness $k$ is derived from the reciprocal of the sum of these compliances for both mating gears:
$$
\frac{1}{k} = \frac{1}{k_h} + \sum_{i=1}^{2}\left(\frac{1}{k_{b,i}} + \frac{1}{k_{s,i}} + \frac{1}{k_{a,i}} + \frac{1}{k_{f,i}}\right)
$$
where subscripts $b$, $s$, $a$, $f$, and $h$ denote bending, shear, axial, foundation, and Hertzian contact stiffness, respectively, and $i=1,2$ represents the driving and driven spur gears.
For spur gears manufactured by a rack-type cutter, the tooth profile consists of two segments: the involute working part and the trochoidal fillet (transition curve). A critical yet often overlooked geometric distinction arises for spur gears with a large number of teeth. When the tooth count exceeds approximately 41, the base circle $r_b$ becomes larger than the root circle $r_f$. In traditional energy models, the integration for calculating bending, shear, and axial energies is incorrectly performed from the root circle to the base circle for the involute part. The correct starting point for the involute profile is not its intersection with the base circle ($G_b$), but the point where it meets the transition curve ($G$). The radius of this point $r_G$ and its corresponding pressure angle $\alpha_G$ differ from $r_b$ and the standard pressure angle $\alpha_0$. Ignoring this distinction for large-tooth-count spur gears leads to an overestimation of the stored elastic energy and, consequently, an underestimation of the mesh stiffness. Therefore, a corrected tooth model is essential for accurate TVMS calculation of such spur gears.
The parameterization of the transition curve is crucial for this correction. For a standard rack-generated spur gear, the transition curve coordinates $(x, y)$, relative to the tooth centerline and root, can be expressed with parameter $\gamma$:
$$
\begin{aligned}
x(\gamma) &= r \cos \phi – (a / \sin \gamma + \rho) \sin(\gamma – \phi) \\
y(\gamma) &= r \sin \phi – (a / \sin \gamma + \rho) \cos(\gamma – \phi)
\end{aligned}
$$
where $\phi = (a \cot \gamma + b)/r$, $r$ is the pitch radius, $a = h_a^* m + c^* m – \rho$, $b = \pi m/4 + h_a^* m \tan \alpha_0 + \rho \cos \alpha_0$, $\rho$ is the cutter tip radius, $m$ is the module, $h_a^*$ and $c^*$ are addendum and dedendum coefficients, and $\gamma \in (\alpha_0, \pi/2)$. The corrected tooth thickness $h_x$ at a distance $d_y$ from the root for large spur gears ($N > 41$) is then:
$$
h_x(d_y) =
\begin{cases}
r \sin \varphi – (a/\sin \gamma + \rho) \cos(\gamma – \varphi), & 0 \leq d_y \leq d_1 \\
r_G[(\alpha_G – \alpha) \cos \alpha + \sin \alpha], & d_1 < d_y \leq d
\end{cases}
$$
Here, $d_1$ is the horizontal distance from the root to point G, $\alpha$ is the pressure angle at the current contact point, and $\alpha_G = \arccos(r_b / r_G) – (\pi/(2N) + \text{inv} \alpha_0 – \text{inv} \alpha_G)$.
The presence of a root crack primarily degrades the bending and shear stiffness of the affected spur gear tooth. To model this, we assume a straight-line crack originating at the root-fillet junction point H, propagating at an angle $\upsilon$ relative to the tooth centerline, and spanning the full face width. The crack depth $q$ is often normalized as a Crack Level (CL):
$$
\text{CL} = \frac{q}{2 \cdot q_{1\text{max}}} \times 100\%
$$
where $q_{1\text{max}}$ is the maximum possible crack depth before crossing the tooth centerline. The crack’s effect is to reduce the effective cross-sectional area $A_x$ and the area moment of inertia $I_x$ along part of the tooth. A key modeling challenge is defining the “effective tooth thickness reduction limit line,” which demarcates the compromised (“dead”) zone of the tooth from the fully effective material.
| Parameter | Driving Gear | Driven Gear |
|---|---|---|
| Number of Teeth, $N$ | 75 | 55 |
| Module, $m$ (mm) | 2 | 2 |
| Pressure Angle, $\alpha_0$ (deg) | 20 | 20 |
| Face Width, $L$ (mm) | 20 | 20 |
| Elastic Modulus, $E$ (GPa) | 211 | 211 |
| Poisson’s Ratio, $\nu$ | 0.3 | 0.3 |
Historically, this limit line was often simplified as a straight line parallel to the tooth centerline, starting from the crack tip. For more severe cracks, this assumption loses accuracy. While a parabolic limit line better approximates the stress gradient, it complicates modeling. A practical and effective alternative proposed here for spur gears is an oblique (diagonal) line connecting the crack tip (point A or B) to the intersection of the involute profile and the line of action (point K). This line offers a better approximation than the straight, centerline-parallel model while being simpler than the parabolic model.
| Crack Scenario | Condition | Effective Area $A_x$ | Moment of Inertia $I_x$ |
|---|---|---|---|
| 1. Above centerline, tip before K | $h_{c1} \geq h_r$ and $\alpha_1 > \alpha_g$ | Piecewise | Piecewise |
| 2. Above centerline, tip beyond K | $h_{c1} < h_r$ or $h_{c1} \geq h_r$, $\alpha_1 \leq \alpha_g$ | $(h_{c1}+h_x)L$ | $(h_{c1}+h_x)^3L/12$ |
| 3. Below centerline, tip before K | $h_{c2} < h_r$ or $h_{c1} \geq h_r$, $\alpha_1 \leq \alpha_g$ | $(h_x – h_{c2})L$ | $(h_x – h_{c2})^3L/12$ |
| 4. Below centerline, tip beyond K | $h_{c2} \geq h_r$ and $\alpha_1 > \alpha_g$ | $(h_{c1} – h_{c2})L$ | $(h_{c1} – h_{c2})^3L/12$ |
| Oblique Limit Line Model | Crack tip at A (or B), limit line to K | $(g_x + h_x)L$ | $(g_x + h_x)^3L/12$ |
In the oblique line model, $g_x$ is the vertical coordinate of the limit line at horizontal distance $d_y$, derived from the linear equation connecting points A and K (or B and K). For spur gears, the coordinates of point K on the tooth tip are $y_K = \sqrt{r_a^2 – h_r^2}$, $x_K = h_r$, where $r_a$ is the tip radius and $h_r$ is the tooth thickness at the tip. The line equation is:
$$
\frac{x – x_A}{y – y_A} = \frac{x_K – x_A}{y_K – y_A}
$$
This allows for the calculation of the reduced effective thickness $(g_x + h_x)$ at any section, which is then used in the standard energy integrals for bending and shear deformation, thereby calculating the reduced stiffness of the cracked spur gear.
The impact of model corrections, crack severity, and limit line type on the TVMS of spur gears is analyzed using the parameters in Table 1. Four crack levels are defined: A (0%, healthy), B (20%), C (40%), and D (60%). Different calculation methods are compared: Method 1 (uncorrected model, straight limit line), Method 2 (corrected model, straight limit line), and the proposed Method 3 (corrected model, oblique limit line). Results are validated against Finite Element Method (FEM) simulations.
| Crack Level (CL) | Method 1 | Reduction | Method 2 | Reduction | FEM | Reduction |
|---|---|---|---|---|---|---|
| A (0%) | 3.121 | 0% | 3.252 | 0% | 3.322 | 0% |
| B (20%) | 3.064 | -1.93% | 3.214 | -1.23% | 3.197 | -3.88% |
| C (40%) | 2.946 | -5.77% | 3.037 | -6.76% | 3.031 | -8.79% |
| D (60%) | 2.737 | -12.50% | 2.840 | -12.61% | 2.875 | -13.66% |
The results clearly demonstrate that the uncorrected model (Method 1) consistently underestimates the TVMS for large-tooth-count spur gears across all crack levels. This error stems from the erroneous integration over a non-existent involute segment between the root and base circles, which artificially increases the calculated compliance. The corrected model (Method 2) yields results closer to the FEM benchmark. Furthermore, the TVMS reduction intensifies with increasing crack depth. Notably, the stiffness loss is more pronounced in the second mesh interval (double-tooth contact period near the tooth root) compared to the first. This is because the cracked section’s compliance constitutes a larger portion of the total tooth compliance when the contact point is near the root, magnifying the crack’s effect in spur gears.
The choice of the effective tooth thickness reduction limit line significantly influences the results, especially for severe cracks. For low-level cracks (CL = 20%, 40%), all three analytical methods produce TVMS curves reasonably close to the FEM results. At these stages, the straight limit line assumption is acceptable, and the added complexity of the oblique line offers marginal accuracy gains.
| Method (Limit Line) | Point A’ Stiffness (×10⁸ N/m) | Error vs. FEM | Point B’ Stiffness (×10⁸ N/m) | Error vs. FEM |
|---|---|---|---|---|
| Method 2 (Straight) | 3.642 | +4.68% | 2.614 | -12.14% |
| Parabolic Model | 3.603 | +3.26% | 2.952 | -0.07% |
| Method 3 (Oblique – Proposed) | 3.590 | +3.19% | 3.144 | +2.09% |
| FEM (Reference) | 3.479 | – | 2.973 | – |
However, for a high crack level (CL = 60%), significant discrepancies arise. As shown in Table 4, at the end of the double-tooth contact zone (Point B’), the straight limit line model (Method 2) incurs an error exceeding 12% compared to FEM. In contrast, both the parabolic and the proposed oblique line models maintain high accuracy, with errors around 0-2%. The oblique line model strikes an excellent balance: it substantially outperforms the straight line model in accuracy for severe cracks in spur gears, while avoiding the parametric fitting complexities associated with a parabolic model. Its formulation requires only the coordinates of the crack tip and a reference point on the sound tooth profile, making it both robust and practical for implementation in analytical algorithms for spur gear health monitoring.
In conclusion, accurate calculation of time-varying mesh stiffness for spur gears with large tooth numbers and root cracks necessitates a geometrically precise model. The common oversight of confusing the involute start point with the base circle intersection must be corrected to avoid systematic underestimation of stiffness. The degradation of mesh stiffness in spur gears is progressive with crack depth and more severe in the root-adjacent mesh intervals. For modeling the compromised tooth section, the choice of the effective thickness reduction limit line is critical. For minor cracks, a straight line parallel to the centerline suffices. For significant crack propagation, however, this simplification leads to substantial error. An oblique line connecting the crack tip to a point on the sound profile offers a superior compromise, providing accuracy comparable to a parabolic model while ensuring simplicity and computational efficiency for the dynamic analysis of faulty spur gears. This refined approach enhances the fidelity of analytical models used in the condition monitoring and prognosis of spur gear transmission systems.