In the realm of mechanical power transmission, helical gear pairs are extensively employed across various industrial sectors due to their smooth operation and high load-carrying capacity. However, the presence of manufacturing and assembly errors, such as geometric eccentricity, can significantly alter the meshing dynamics and internal excitations within the gear system. Among these internal excitations, friction excitation arising from tooth surface interaction is a critical factor. It not only accelerates wear, pitting, and crack initiation but also profoundly influences the dynamic characteristics and stability of the entire transmission system. The time-varying length of the contact line in helical gears is a primary source of this excitation, affecting time-varying mesh stiffness, damping, and consequently, the friction force and frictional torque. This article, from my perspective as a researcher in mechanical dynamics, delves into a comprehensive methodology for calculating the friction excitation of helical gear pairs when geometric eccentricity errors are present. I will establish a detailed geometrical model, derive key kinematic formulas, develop a piecewise function approach for contact line length, and subsequently calculate the friction force and torque. A numerical case study will be presented, comparing results with and without eccentricity errors, highlighting the increased complexity and spectral features introduced by such imperfections.
The analysis of helical gear dynamics necessitates an accurate geometrical representation, especially when errors are involved. I begin by constructing a two-dimensional geometric model for a single-stage helical gear pair in the transverse plane, considering the eccentricity of both the driving and driven gears. In the absolute coordinate system OXYZ with its origin at the driving gear’s rotational center O1, the geometric centers O’1 and O’2 of the driving and driven helical gears, respectively, are not fixed due to eccentricity. Their positions at any time t are given by:
$$ x_1 = e_1 \cos(\phi_1 + \omega_1 t), \quad y_1 = e_1 \sin(\phi_1 + \omega_1 t) $$
$$ x_2 = e_2 \cos(\phi_2 – \omega_2 t) + H, \quad y_2 = e_2 \sin(\phi_2 – \omega_2 t) $$
Here, \(e_1\) and \(e_2\) are the eccentricity magnitudes, \(\phi_1\) and \(\phi_2\) are the initial eccentricity angles, \(\omega_1\) and \(\omega_2\) are the rotational speeds, and H is the nominal center distance. The instantaneous geometric center distance \(l(t)\) becomes variable:
$$ l(t) = \sqrt{(x_1 – x_2)^2 + (y_1 – y_2)^2} $$
This variable center distance is fundamental as it modulates the mesh geometry. The base circle radii are \(r_{b1}\) and \(r_{b2}\). The line of action, tangent to both base circles, defines the dynamic pressure angle \(\alpha’_t\). Key geometric parameters are derived from the relative positions of the centers. The angle \(\psi\) between the line of centers and the line connecting the base circle centers relative to the line of action, and the orientation angle \(\gamma\) of the center line are:
$$ \psi = \arccos\left( \frac{a \cos \alpha_t}{l(t)} \right), \quad \gamma = \arctan\left( \frac{y_2 – y_1}{x_2 – x_1} \right) $$
where \(a\) is the theoretical center distance and \(\alpha_t\) is the transverse pressure angle. Consequently, the dynamic working pressure angle is:
$$ \alpha’_t = \psi – \gamma $$
The slope of the dynamic line of action is \(k = 1 / \tan \alpha’_t\). The coordinates of the tangency points A and B on the driving and driven gear base circles, respectively, can be found by solving the circle equations and the slope condition:
$$ x_{01} = x_1 + \frac{k}{\sqrt{k^2 + 1}} r_{b1}, \quad y_{01} = y_1 – \frac{r_{b1}}{\sqrt{k^2 + 1}} $$
$$ x_{02} = x_2 – \frac{k}{\sqrt{k^2 + 1}} r_{b2}, \quad y_{02} = y_2 + \frac{r_{b2}}{\sqrt{k^2 + 1}} $$
The intersection point P of the line of action with the line connecting rotational centers O1 and O2 (which is fixed along the X-axis at distance H) is crucial for defining the pitch point. Its x-coordinate is:
$$ x_P = x_{02} – y_{02} \frac{x_{02} – x_{01}}{y_{02} – y_{01}} $$
From this, the instantaneous pitch radii for the driving and driven helical gears are:
$$ r_1 = x_P, \quad r_2 = l(t) – r_1 $$
Furthermore, the instantaneous transmission ratio \(i_{12}(t)\) deviates from the nominal ratio due to eccentricity:
$$ i_{12}(t) = \frac{\omega_1}{\omega_2(t)} = \frac{H – x_P}{x_P} $$
Given a constant input speed \(\omega_1\), the output speed \(\omega_2(t)\) becomes time-varying and can be computed by solving the above equation iteratively with the geometric relations. The velocity along the line of action is:
$$ v(t) = \omega_1 x_P \cos \psi $$
This geometric model forms the foundation for all subsequent calculations regarding the contact and friction in the helical gear pair.
The time-varying total length of the contact lines is a critical parameter influencing load distribution and friction excitation in helical gears. For an ideal helical gear pair without errors, the contact line length variation within a mesh cycle is well-documented and depends on the transverse contact ratio \(\varepsilon_{\alpha}\) and the axial contact ratio \(\varepsilon_{\beta}\). However, with eccentricity, the instantaneous geometric center distance \(l(t)\) modulates the effective transverse contact ratio, making the contact line length calculation more complex. I adopt a piecewise function approach to describe the length of a single tooth pair’s contact line as it moves through the plane of contact.
In the contact plane projection, the total possible contact region is bounded by the gear addendum circles. Let \(b\) be the face width, \(\beta_b\) the base helix angle, \(P_t\) the transverse base pitch, and \(L_{CD}(t)\) the instantaneous length of the path of contact in the transverse plane. The axial contact ratio \(\varepsilon_{\beta} = b \tan \beta_b / P_t\) remains constant, while the transverse contact ratio \(\varepsilon_{\alpha}(t) = L_{CD}(t) / P_t\) varies with \(l(t)\). The lengths \(L_{AC_0}\) (initial distance from start of contact to a reference point) and \(L_{AC}(t)\) are:
$$ L_{AC_0} = \sqrt{l(0)^2 – (r_{b1}+r_{b2})^2} – \sqrt{r_{a1}^2 – r_{b1}^2} $$
$$ L_{AC}(t) = \sqrt{l(t)^2 – (r_{b1}+r_{b2})^2} – \sqrt{r_{a1}^2 – r_{b1}^2} $$
$$ L_{CD}(t) = L_{AC}(t) – L_{AD}, \quad \text{where } L_{AD} = \sqrt{r_{a1}^2 – r_{b1}^2} $$
At any time \(t\), the distance from the contact start point C to the lower end E of the contact line for the i-th mating tooth pair is:
$$ D_i(t) = \int_0^t v(\tau) d\tau + (i – 1)P_t + L_{AC_0} – L_{AC}(t) – m P_t $$
where \(m\) is the integer part of \(\int_0^t v(\tau) d\tau / P_t\). Due to eccentricity, the index \(i\) can start from 0, indicating early entry into mesh when the center distance decreases. The length of the contact line for a single tooth pair \(L_i(t)\) is expressed as a piecewise function based on the position \(D_i\):
$$
L_i(t) =
\begin{cases}
D_i / \sin \beta_b, & 0 \leq D_i < \min(b \tan \beta_b, L_{CD}) \\
\min(b \tan \beta_b, L_{CD}) / \sin \beta_b, & \min(b \tan \beta_b, L_{CD}) \leq D_i < \max(b \tan \beta_b, L_{CD}) \\
\left[ L_{CD} – (D_i – b \tan \beta_b) \right] / \sin \beta_b, & \max(b \tan \beta_b, L_{CD}) \leq D_i < L_{CD} + b \tan \beta_b \\
0, & D_i \geq L_{CD} + b \tan \beta_b
\end{cases}
$$
The total contact line length \(L_{total}(t)\) is the sum over all actively mating tooth pairs:
$$ L_{total}(t) = \sum_{i} L_i(t) $$
This formulation captures the effect of eccentricity-induced center distance variation on the engagement pattern of the helical gear pair.
To compute the friction excitation, I first determine the friction force and frictional torque arising from the sliding action on the tooth flanks. The direction of sliding reverses at the pitch line (instantaneous pitch point P in the contact plane). Therefore, each contact line is divided into two segments at the pitch line: the approach segment (from the start of contact to P) and the recess segment (from P to the end of contact). The friction force on each segment acts opposite to the relative sliding direction. I define the friction force direction as positive on the recess side (right of pitch line) and negative on the approach side (left of pitch line). Similarly, the frictional torque is considered positive if it acts in the direction of rotation and negative otherwise.
Assuming a constant average coefficient of friction \(\mu\) and a uniform distribution of the normal load along the contact line for simplification, the total normal force \(F\) is shared among all contact lines in proportion to their lengths. For the j-th segment (j=1 for approach, j=2 for recess) of the i-th tooth pair’s contact line, with instantaneous length \(L_{ij}(t)\), the normal force on that segment is:
$$ F_{ij} = \frac{F}{L_{total}(t)} L_{ij}(t) $$
The friction force on that segment is \( \mu F_{ij} \), with the sign determined by its position relative to the pitch line. The lengths \(L_{i1}\) and \(L_{i2}\) for the two segments are determined based on the distance \(D_i\) and the distance \(L_{CP}(t)\) from the contact start C to the pitch point P, and \(L_{PD}(t)\) from P to the contact end D:
$$ L_{CP}(t) = \sqrt{(x_{01} – x_P)^2 + y_{01}^2} – L_{AC}(t), \quad L_{PD}(t) = L_{CD}(t) – L_{CP}(t) $$
The segment lengths are:
$$
\begin{aligned}
& \text{If } 0 \leq D_i < L_{CP}: & L_{i1} = 0, & L_{i2} = L_i \\
& \text{If } L_{CP} \leq D_i < L_{CP} + b \tan \beta_b: & L_{i1} = \frac{D_i – L_{CP}}{\sin \beta_b}, & L_{i2} = L_i – L_{i1} \\
& \text{If } L_{CP} + b \tan \beta_b \leq D_i < L_{CD} + b \tan \beta_b: & L_{i1} = L_i, & L_{i2} = 0 \\
& \text{Otherwise:} & L_{i1} = 0, & L_{i2} = 0
\end{aligned}
$$
The net friction force from the i-th tooth pair is:
$$ F_{f,i}(t) = \mu \frac{F}{L_{total}(t)} (L_{i2} – L_{i1}) $$
The total friction force on the gear pair is:
$$ F_f(t) = \sum_i F_{f,i}(t) $$
To calculate the frictional torque, the moment arm for each friction force segment is needed. The lever arm \(S\) is the perpendicular distance from the driving gear’s rotational center O1 to the line of action, which is \(S = r_1 \sin \psi\). The effective moment arm for each segment is approximated by taking the midpoint of that segment. For the approach segment (length \(L_{i1}\)), the distance from its midpoint to the pitch point along the line of action is \(L_{i1} \sin \beta_b / 2\). Similarly for the recess segment. Thus, the moment arms \(S_{ij}\) are:
$$
\begin{aligned}
& \text{If } 0 \leq D_i < L_{CP}: & S_{i1} = 0, & S_{i2} = S – L_{CP} + \frac{L_{i2} \sin \beta_b}{2} \\
& \text{If } L_{CP} \leq D_i < L_{CP} + b \tan \beta_b: & S_{i1} = S + \frac{L_{i1} \sin \beta_b}{2}, & S_{i2} = S – \frac{L_{i2} \sin \beta_b}{2} \\
& \text{If } L_{CP} + b \tan \beta_b \leq D_i < L_{CD} + b \tan \beta_b: & S_{i1} = S + L_{PD} – \frac{L_{i1} \sin \beta_b}{2}, & S_{i2} = 0 \\
& \text{Otherwise:} & S_{i1} = 0, & S_{i2} = 0
\end{aligned}
$$
The frictional torque contribution from the i-th tooth pair is:
$$ T_{f,i}(t) = \mu \frac{F}{L_{total}(t)} (L_{i2} S_{i2} – L_{i1} S_{i1}) $$
The total frictional torque on the driving gear is:
$$ T_f(t) = \sum_i T_{f,i}(t) $$
These equations provide a complete framework for computing the time-varying friction excitation in a helical gear pair with geometric eccentricity.
To illustrate the methodology and analyze the effects of eccentricity, I perform a numerical simulation for a specific helical gear pair. The parameters are summarized in the table below.
| Parameter | Driving Helical Gear | Driven Helical Gear |
|---|---|---|
| Number of teeth, \(z\) | 18 | 69 |
| Normal module, \(m_n\) (mm) | 2.5 | 2.5 |
| Normal pressure angle, \(\alpha_n\) (°) | 20 | 20 |
| Helix angle, \(\beta\) (°) | 12 | 12 |
| Face width, \(b\) (mm) | 50 | 50 |
| Rotational speed, \(\omega\) (rad/s) | 157 | Variable |
| Input power, \(P\) (kW) | 45 | 45 |
The derived transverse pressure angle \(\alpha_t\) and base helix angle \(\beta_b\) are:
$$ \alpha_t = \arctan\left( \frac{\tan \alpha_n}{\cos \beta} \right), \quad \beta_b = \arcsin(\sin \beta \cos \alpha_t) $$
For the simulation, I set eccentricities \(e_1 = e_2 = 0.2 \text{ mm}\), initial angles \(\phi_1 = 0\), \(\phi_2 = \pi/2\), and a constant input torque corresponding to the 45 kW power at 157 rad/s. The average friction coefficient \(\mu\) is taken as 0.08. The equations are solved numerically over several mesh cycles.
First, the output speed \(\omega_2(t)\) of the driven helical gear is no longer constant. Its time-domain waveform shows periodic fluctuations, and its frequency spectrum reveals dominant peaks at the input shaft rotational frequency \(f_{in} = \omega_1/(2\pi) \approx 25 \text{ Hz}\) and the output shaft rotational frequency \(f_{out} \approx 6.5 \text{ Hz}\), as expected from the eccentricity modulation.
The total contact line length \(L_{total}(t)\) is computed. The following table compares key characteristics of the contact line length with and without eccentricity errors.
| Aspect | Helical Gear without Eccentricity | Helical Gear with Eccentricity |
|---|---|---|
| Time-domain pattern | Regular, periodic with mesh cycle | More complex, amplitude and period modulated |
| Average length | Constant over time | Varies slightly with center distance changes |
| Frequency spectrum peaks | At mesh frequency \(f_m\) and harmonics | At \(f_m\), harmonics, \(f_{in}\), \(f_{out}\), and sidebands around \(f_m\) |
| Sideband spacing | None | Equal to \(f_{in}\) and \(f_{out}\) |
The mesh frequency \(f_m = z_1 \omega_1 / (2\pi) \approx 450 \text{ Hz}\). The presence of eccentricity introduces sidebands around the mesh frequency and its harmonics, spaced by the rotational frequencies. This modulation effect is a classic signature of gear eccentricity in vibration spectra. When the center distance increases due to eccentricity, the transverse contact ratio decreases momentarily, causing \(L_{total}(t)\) to drop below its ideal value. Conversely, when the center distance decreases, \(L_{total}(t)\) exceeds the ideal value. This modulation directly impacts the friction excitation.
The friction force \(F_f(t)\) and frictional torque \(T_f(t)\) are calculated using the described method. Their time-domain behaviors mirror the complexity of the contact line length. For the ideal helical gear pair, \(F_f(t)\) and \(T_f(t)\) exhibit a periodic pattern primarily at the mesh frequency. With eccentricity, their waveforms become richer, showing beats and amplitude variations. The frequency spectra confirm this: strong components appear at the mesh frequency, its harmonics, and the rotational frequencies. Furthermore, distinct sideband families emerge around the mesh frequency components, with spacing equal to \(f_{in}\) and \(f_{out}\). These sidebands indicate that the friction excitation process is modulated by the slow rotation of the eccentric gears, potentially exacerbating vibration and noise in the transmission system.
To generalize, the impact of eccentricity in a helical gear pair can be quantified by the following key formulas summarizing the modulation effects:
$$ \text{Modulated Center Distance: } l(t) = a + \Delta l(t), \quad \Delta l(t) \propto e_1 \cos(\omega_1 t + \phi_1) + e_2 \cos(\omega_2 t + \phi_2) $$
$$ \text{Modulated Mesh Stiffness (related): } k_m(t) \approx k_{m0} \left[ 1 + \epsilon \cos(\omega_{mod} t) \right] $$
$$ \text{Friction Force Modulation: } F_f(t) = F_{f0}(t) \left[ 1 + \sum_m \xi_m \cos(m \omega_{mod} t + \theta_m) \right] $$
where \(\omega_{mod}\) represents the modulation frequencies (e.g., \(\omega_1\), \(\omega_2\)), and \(\epsilon\), \(\xi_m\) are modulation indices dependent on eccentricity magnitudes.
In conclusion, the methodology developed here provides a comprehensive approach for calculating friction excitation in helical gear pairs subject to geometric eccentricity errors. By establishing a detailed geometrical model that accounts for the time-varying center distance, dynamic pressure angle, and instantaneous transmission ratio, I derived formulas essential for kinematic analysis. The piecewise function formulation for the time-varying contact line length effectively captures the engagement variations induced by eccentricity. Subsequently, the friction force and frictional torque calculations incorporate the segmentation of contact lines at the pitch line and consider the variable moment arms. The numerical analysis for a specific helical gear pair clearly demonstrates that eccentricity errors introduce significant complexity into the system’s dynamic response. The contact line length, friction force, and frictional torque all exhibit modulated time-domain signals with enriched frequency spectra featuring mesh frequency harmonics, rotational frequency components, and prominent sidebands. These spectral features, characteristic of modulation due to eccentricity, can degrade the stability and smoothness of the helical gear transmission. Therefore, accurately modeling these effects is crucial for high-precision gear design, noise-vibration-harshness (NVH) prediction, and condition monitoring strategies aimed at detecting mounting or manufacturing faults in helical gear systems. Future refinements could incorporate more realistic factors such as a spatially and temporally varying friction coefficient, non-uniform load distribution along the contact line, and the effects of other gear errors like profile deviations to further enhance the predictive capability of this helical gear friction excitation model.