In the evolving landscape of electric vehicle powertrains, the demand for efficient, quiet, and reliable transmission systems has never been higher. Multi-stage helical gear reducers are a cornerstone of this technology, prized for their smooth operation and high load capacity compared to their spur gear counterparts. However, the very feature that grants them smoothness—the helical angle—also introduces significant complexity into their dynamic behavior. A central challenge in the accurate dynamic modeling of helical gear systems lies in the precise calculation of the time-varying meshing stiffness (TVMS) and the quantitative analysis of its fluctuation. This stiffness is not constant; it varies periodically as gear teeth engage and disengage, acting as a primary internal excitation source for vibration and noise. Therefore, accurately capturing this time-varying characteristic is paramount for predicting system dynamics, diagnosing faults, and optimizing design for noise, vibration, and harshness (NVH) performance.
Traditional methods for calculating TVMS include the finite element method, which is accurate but computationally expensive, and the Ishikawa formula, which offers simplicity at the cost of some accuracy. The potential energy method has emerged as a highly effective compromise, offering a good balance between computational efficiency and accuracy. This method decomposes the total elastic energy stored in a meshing tooth pair into several components: Hertzian contact energy, bending energy, shear energy, axial compressive energy, and the energy due to fillet foundation deflection. The total mesh stiffness is then derived from the sum of these complementary energy components. For spur gears, the application is relatively straightforward. However, for helical gears, the presence of the helix angle means the contact line is inclined and its length varies during meshing. To address this, the slicing method is employed, where the helical gear tooth is discretized along its face width into a series of thin spur gear slices. The stiffness of each slice is calculated using the potential energy method, and the total stiffness of the helical gear pair is obtained by summing the stiffness contributions from all slices in contact at any given time.
Consider a helical gear pair. The total mesh stiffness \( k(t) \) at time \( t \) for a pair of meshing helical gears can be expressed as the sum of the stiffnesses of \( n \) simultaneously contacting tooth pairs:
$$ k(t) = \sum_{i=1}^{n} \frac{1}{ \frac{1}{k_{h,i}} + \frac{1}{k_{b1,i}} + \frac{1}{k_{s1,i}} + \frac{1}{k_{a1,i}} + \frac{1}{k_{f1,i}} + \frac{1}{k_{b2,i}} + \frac{1}{k_{s2,i}} + \frac{1}{k_{a2,i}} + \frac{1}{k_{f2,i}} } $$
Where the subscripts \(b\), \(s\), \(a\), \(f\), and \(h\) denote bending, shear, axial, fillet foundation, and Hertzian contact stiffness, respectively, and subscripts \(1\) and \(2\) refer to the pinion and gear. The stiffness for each component for a single slice of thickness \(\Delta y\) is given by:
Bending Stiffness:
$$ k_b = \sum_{i=1}^{N} \Delta y \left/ \int_{-\alpha_1′}^{\alpha_2} \frac{ 3\{1 + \cos\alpha_1′[(\alpha_2 – \alpha)\sin\alpha – \cos\alpha]\}^2 (\alpha_2 – \alpha)\cos\alpha }{2E [\sin\alpha + (\alpha_2 – \alpha)\cos\alpha]^3} \, d\alpha \right. $$
Shear Stiffness:
$$ k_s = \sum_{i=1}^{N} \left[ \int_{-\alpha_1′}^{\alpha_2} \frac{1.2(1+\nu)(\alpha_2 – \alpha)\cos\alpha \cos^2\alpha_1′}{E [\sin\alpha + (\alpha_2 – \alpha)\cos\alpha]} \, d\alpha \right]^{-1} \Delta y $$
Axial Compressive Stiffness:
$$ k_a = \sum_{i=1}^{N} \left[ \int_{-\alpha_1′}^{\alpha_2} \frac{(\alpha_2 – \alpha)\cos\alpha \sin^2\alpha_1′}{2E [\sin\alpha + (\alpha_2 – \alpha)\cos\alpha]} \, d\alpha \right]^{-1} \Delta y $$
Fillet Foundation Stiffness:
$$ k_f = \sum_{i=1}^{N} \frac{ \Delta y \cos^2\alpha_1′}{E} \left/ \left\{ L^*\left(\frac{u_f}{S_f}\right)^2 + M^*\left(\frac{u_f}{S_f}\right) + P^* \left[1 + Q^* \tan^2(\alpha_1′)\right] \right\} \right. $$
The coefficients \(L^*\), \(M^*\), \(P^*\), and \(Q^*\) are determined by polynomial functions of gear geometry parameters.
Hertzian Contact Stiffness:
$$ k_h = \frac{\pi E L(t)}{4(1-\nu^2)} $$
Here, \(L(t)\) is the time-varying contact line length, which is the critical factor distinguishing helical gear stiffness calculation. Its value depends on the relationship between the transverse contact ratio \(\varepsilon_{\alpha}\) and the axial contact ratio \(\varepsilon_{\beta}\). For \(\varepsilon_{\alpha} < \varepsilon_{\beta}\):
$$
L(t) =
\begin{cases}
\frac{\varepsilon_{\alpha} P_{bt} \, t}{\sin\beta_b \varepsilon_{\alpha} t_z}, & t \in [0, \varepsilon_{\alpha} t_z] \\
\frac{\varepsilon_{\alpha} P_{bt}}{\sin\beta_b}, & t \in [\varepsilon_{\alpha} t_z, \varepsilon_{\beta} t_z] \\
\frac{\varepsilon_{\alpha} P_{bt} \, t}{\sin\beta_b \varepsilon_{\alpha}} \left(\varepsilon – \frac{t}{t_z}\right), & t \in [\varepsilon_{\beta} t_z, (\varepsilon_{\alpha}+\varepsilon_{\beta}) t_z]
\end{cases}
$$
And for \(\varepsilon_{\alpha} \ge \varepsilon_{\beta}\):
$$
L(t) =
\begin{cases}
\frac{b \, t}{\cos\beta_b \varepsilon_{\beta} t_z}, & t \in [0, \varepsilon_{\beta} t_z] \\
\frac{b}{\cos\beta_b}, & t \in [\varepsilon_{\beta} t_z, \varepsilon_{\alpha} t_z] \\
\frac{b}{\cos\beta_b \varepsilon_{\beta}} \left(\varepsilon – \frac{t}{t_z}\right), & t \in [\varepsilon_{\alpha} t_z, (\varepsilon_{\alpha}+\varepsilon_{\beta}) t_z]
\end{cases}
$$
where \(P_{bt}\) is the transverse base pitch, \(\beta_b\) is the base helix angle, \(b\) is the face width, \(\varepsilon = \varepsilon_{\alpha} + \varepsilon_{\beta}\) is the total contact ratio, and \(t_z\) is the time period for one base pitch.
The fluctuation of the TVMS, denoted as \(\Delta K\), is a key metric influencing dynamic response. A significant finding from applying this method is that \(\Delta K\) is not monotonically related to the helix angle \(\beta\). While increasing \(\beta\) generally increases the total contact ratio and smooths the engagement, the specific interplay between the transverse and axial contact ratios dictates the stiffness transition. To quantitatively predict the fluctuation without performing full TVMS calculations, a novel dimensionless parameter \(\tau\) is proposed. This parameter is defined as the ratio of the shortest distance of either contact ratio (\(\varepsilon_{\alpha}\) or \(\varepsilon_{\beta}\)) to the nearest integer, divided by the smaller of the two contact ratios:
$$ \tau = \frac{ \min\left\{ \min[ \text{ceil}(\varepsilon_{\alpha})-\varepsilon_{\alpha},\, \varepsilon_{\alpha}-\text{floor}(\varepsilon_{\alpha}) ],\, \min[ \text{ceil}(\varepsilon_{\beta})-\varepsilon_{\beta},\, \varepsilon_{\beta}-\text{floor}(\varepsilon_{\beta}) ] \right\} }{ \min(\varepsilon_{\alpha}, \varepsilon_{\beta}) } $$
Analysis shows a strong positive correlation: a smaller \(\tau\) value predicts a smaller TVMS fluctuation \(\Delta K\). This provides a powerful, simple tool for designers to pre-judge the potential dynamic excitability of a helical gear set based solely on its basic geometric parameters.
To investigate the system-level dynamic consequences, a lumped-parameter model of a two-stage helical gear system for an EV reducer is developed. The model incorporates 12 degrees of freedom (DOF), accounting for the transverse vibrations (\(x, z\) directions) and torsional vibration (\(\theta\)) of each of the four gears. Crucially, the intermediate shaft, which carries two gears, is modeled with both axial (compressional) stiffness \(k_s\) and torsional stiffness \(k_t\), along with their corresponding damping elements \(c_s\) and \(c_t\). This detail is essential as it captures the coupling and interaction between the two gear stages through the shared shaft. The bearings supporting the shafts are modeled as linear spring-damper elements in both transverse directions. The time-varying mesh stiffnesses \(k_{1m}(t)\) and \(k_{2m}(t)\), calculated via the potential energy method, are introduced as the key internal excitations at the two meshing interfaces.
The equations of motion are derived using D’Alembert’s principle. For illustration, the equations for the pinion (Gear 1) and the intermediate shaft gear (Gear 2) of the first stage are shown. The relative displacements along the line of action (x-direction) and axial direction (z-direction) between Gear 1 and Gear 2 are:
$$ \delta_{1x} = (x_1 + r_{b1}\theta_1) – (x_2 + r_{b2}\theta_2) $$
$$ \delta_{1z} = (z_1 – \bar{x}_1 \tan\beta_b) – (z_2 – \bar{x}_2 \tan\beta_b) $$
The resulting mesh forces are:
$$ F_{1x} = k_{1mx}(t) \delta_{1x} + c_{1mx} \dot{\delta}_{1x} $$
$$ F_{1z} = k_{1mz}(t) \delta_{1z} + c_{1mz} \dot{\delta}_{1z} $$
where \(k_{1mx} = k_{1m}(t)\cos\beta_b\), \(k_{1mz} = k_{1m}(t)\sin\beta_b\), and the damping coefficients \(c\) are expressed as \(c = 2\xi \sqrt{k_{avg} I_{eq}}\), with \(\xi\) as the damping ratio and \(I_{eq}\) as the equivalent inertia.
The final system dynamics equation in matrix form is:
$$ \mathbf{M}\ddot{\mathbf{q}} + \mathbf{C}\dot{\mathbf{q}} + \mathbf{K}(t)\mathbf{q} = \mathbf{F} $$
where \(\mathbf{q} = \{x_1, z_1, \theta_1, x_2, z_2, \theta_2, x_3, z_3, \theta_3, x_4, z_4, \theta_4\}^T\) is the displacement vector, \(\mathbf{M}\) is the mass matrix, \(\mathbf{C}\) is the damping matrix, \(\mathbf{K}(t)\) is the time-varying stiffness matrix containing the mesh stiffnesses \(k_{1m}(t)\) and \(k_{2m}(t)\), and \(\mathbf{F}\) is the force vector containing input torque \(T_p\) and load torque \(T_g\).
| Parameter | Stage 1 (Gear 1 / Gear 2) | Stage 2 (Gear 3 / Gear 4) | Unit |
|---|---|---|---|
| Number of Teeth, \(z\) | 17 / 35 | 18 / 66 | – |
| Normal Module, \(m_n\) | 2.75 | mm | |
| Pressure Angle, \(\alpha_n\) | 20 | ° | |
| Helix Angle, \(\beta\) (Variable) | 9, 12, 15, 18, 21, 24 | ° | |
| Face Width, \(b\) | 35 | 30 | mm |
| Young’s Modulus, \(E\) | 2.06e11 | Pa | |
| Poisson’s Ratio, \(\nu\) | 0.3 | – | |
The model is solved numerically using the Runge-Kutta method under a constant input speed of 1500 rpm and an input torque of 100 N·m. The dynamic response is analyzed for different helix angles to validate the influence of the parameter \(\tau\) and the TVMS fluctuation \(\Delta K\).
| Helix Angle \(\beta\) (°) | Stage | Trans. Ratio \(\varepsilon_{\alpha}\) | Axial Ratio \(\varepsilon_{\beta}\) | Parameter \(\tau\) | TVMS Fluctuation \(\Delta K\) (N/m) |
|---|---|---|---|---|---|
| 9 | 1 | 1.423 | 0.648 | 0.226 | ~4.8e8 |
| 2 | 1.412 | 0.556 | 0.206 | ~5.5e8 | |
| 12 | 1 | 1.452 | 0.867 | 0.314 | ~7.2e8 |
| 2 | 1.440 | 0.743 | 0.345 | ~7.9e8 | |
| 15 | 1 | 1.498 | 1.087 | 0.016 | ~1.5e8 |
| 2 | 1.486 | 0.931 | 0.074 | ~2.1e8 | |
| 18 | 1 | 1.562 | 1.308 | 0.126 | ~3.0e8 |
| 2 | 1.550 | 1.121 | 0.284 | ~4.8e8 | |
| 21 | 1 | 1.647 | 1.534 | 0.241 | ~4.1e8 |
| 2 | 1.634 | 1.315 | 0.224 | ~3.9e8 | |
| 24 | 1 | 1.756 | 1.770 | 0.138 | ~2.7e8 |
| 2 | 1.742 | 1.517 | 0.148 | ~2.9e8 |
The results clearly demonstrate the predictive power of \(\tau\). For both gear stages, the minimum TVMS fluctuation \(\Delta K\) occurs at a helix angle of 15°, which corresponds to the smallest values of \(\tau\) (0.016 and 0.074). This confirms that \(\tau\) is an effective indicator for stiffness smoothness.
The dynamic system response correlates strongly with this finding. The vibration displacement, dynamic mesh force, and dynamic transmission error (DTE) all show optimal performance at \(\beta = 15°\).
Vibration Analysis: The vibration acceleration of gears on the intermediate shaft reveals strong coupling between stages. The tangential (\(x\)-direction) vibration of the first-stage gear is dominated by the third harmonic of its own mesh frequency (e.g., \(3f_1\)), while the second-stage gear’s vibration is more broadly distributed across the harmonics of its mesh frequency (\(f_2\)), but also shows influence from the first-stage excitation. Axial (\(z\)-direction) vibrations are generally of higher magnitude than tangential vibrations due to lower axial support stiffness.
Dynamic Mesh Force: The dynamic component of the mesh force, which superimposes on the static load, is a critical factor for fatigue life and noise generation. The root-mean-square (RMS) values of the dynamic mesh force for the first stage across different helix angles are summarized below:
| Helix Angle \(\beta\) (°) | 9 | 12 | 15 | 18 | 21 | 24 |
|---|---|---|---|---|---|---|
| Tangential Force \(F_x\) (N) | 425 | 580 | 210 | 310 | 390 | 255 |
| Axial Force \(F_z\) (N) | 68 | 115 | 55 | 80 | 105 | 95 |
The dynamic mesh forces are minimized at \(\beta = 15°\), corresponding to the smallest \(\Delta K\). The case of \(\beta = 12°\) shows the largest forces, which aligns with its large \(\tau\) and \(\Delta K\) values.
Dynamic Transmission Error (DTE): DTE, defined as the relative displacement of mating gears along the path of contact (\(\delta_{1x}\)), is a primary source of vibration excitation. The frequency-domain amplitude of DTE at the fundamental mesh frequency for both stages is highly sensitive to the helix angle.
| Helix Angle \(\beta\) (°) | DTE Amplitude – Stage 1 (µm) | DTE Amplitude – Stage 2 (µm) |
|---|---|---|
| 9 | 4.8 | 5.2 |
| 12 | 6.5 | 7.1 |
| 15 | 1.2 | 1.8 |
| 18 | 2.7 | 3.9 |
| 21 | 3.5 | 4.3 |
| 24 | 2.1 | 3.0 |
Again, the minimum DTE, indicating the smoothest transmission and lowest vibration excitation potential, is achieved at \(\beta = 15°\).
In conclusion, this analysis underscores the critical role of accurately modeling the time-varying meshing stiffness in the dynamic study of helical gear systems. The potential energy method, combined with the slicing technique, provides an efficient and accurate framework for this calculation. The introduction of the geometric parameter \(\tau\) offers a significant advancement, allowing designers to quantitatively estimate the stiffness fluctuation \(\Delta K\) and, consequently, predict the dynamic excitation level of a helical gear pair without performing complex dynamic simulations. For the two-stage EV reducer studied, a helix angle of 15° for all helical gears resulted in the smallest \(\tau\) and \(\Delta K\), which directly translated to superior dynamic performance: minimized vibration displacement, reduced dynamic mesh force, and lower dynamic transmission error. This work verifies that optimizing gear geometry to minimize \(\tau\) is a viable and accurate strategy for pre-judging and enhancing the overall dynamic performance of multi-stage helical gear systems, providing a valuable theoretical and practical tool for the design of quiet and reliable electric vehicle drivetrains.