In the pursuit of advanced mechanical power transmission solutions for high-performance applications such as aviation engines, helicopter drive systems, and marine propulsion, the split-torque gear transmission system has emerged as a critical technology. Its defining characteristic is the division of input power into multiple parallel paths, which are subsequently recombined at the output. This architecture offers significant advantages, including high power density, reduced weight and volume, and improved efficiency, as the load is shared among several gear meshes. Among various gear types, herringbone gears are particularly favored in such demanding systems due to their inherent ability to cancel out axial thrust forces, leading to smoother operation and reduced bearing loads compared to single helical gears. The core of this study revolves around a specific configuration prevalent in aero-engine reducers: a split-torque system employing herringbone gears. A key, often underexplored, design parameter in such systems is the “shaft angle”—the angular arrangement formed by the geometric centers of the gears within the transmission layout. This paper investigates the profound impact of these shaft angles on the system’s dynamic behavior, including vibration response, dynamic load sharing between parallel paths, and dynamic transmission error.
The fundamental configuration under analysis is a two-stage, dual-path system. The first stage acts as the torque-splitting stage, where a single input herringbone gear simultaneously meshes with two intermediate herringbone gears. Each of these intermediate gears is co-axial with a second-stage herringbone gear on a compound shaft. In the second, torque-combining stage, these two second-stage gears simultaneously mesh with a single output herringbone gear. This arrangement creates a closed-loop kinematic chain, introducing a hyperstatic or over-constrained condition. This condition imposes strict geometric and phasing requirements on the system for proper assembly and load distribution, making the calculation of the correct shaft angles not merely a layout choice but a necessity for functionality.
Geometric Synthesis: Calculating the Shaft Angles
The spatial arrangement of gear centers, defined by shaft angles, must satisfy a set of geometric constraints to ensure feasible and optimal operation. Let the angles be defined as shown in the conceptual layout: α is the angle subtended at the input pinion center between the two first-stage compound gear centers, β is the angle subtended at the output gear center between the two second-stage compound gear centers, and δ represents the symmetrical angle on either side at the compound gear centers (assuming δ1 = δ2 = δ for symmetry). The gear teeth numbers are denoted as $z_1$ (input), $z_2$ (first-stage compound), $z_3$ (second-stage compound), and $z_4$ (output). The following conditions must be met:
1. Adjacency Condition: This ensures physical clearance between the non-meshing gears, specifically between the two first-stage compound gears. Their tooth tips must not collide, and in the limiting layout, they would be positioned 180 degrees apart.
$$ d_{a2} < L_{CD} \leq d_1 + d_2 $$
where $d_{a2}$ is the tip diameter of the first-stage compound gear, $d_1$ and $d_2$ are the pitch diameters of the input and first-stage compound gears, and $L_{CD}$ is the center distance between the two compound gears, given by $L_{CD} = (d_3 + d_4) \sin(\beta/2)$.
2. Concentricity Condition: This condition ensures that the centers of the first-stage and second-stage gears on the same compound shaft align with the shaft’s axis, forming consistent triangles from the input and output centers.
$$ \frac{d_1 + d_2}{2} \sin(\alpha/2) = \frac{d_3 + d_4}{2} \sin(\beta/2) $$
3. Geometric Closure Condition: The four angles in the quadrilateral formed by the gear centers must sum to 360 degrees (2π radians).
$$ \alpha + \beta + 2\delta = 2\pi $$
4. Assembly (or Phasing) Condition: This is the most critical condition for ensuring proper mesh phasing in the hyperstatic loop. It guarantees that starting from an ideally meshed tooth pair in one branch, the kinematic chain returns to the exact same phasing after traversing all meshes in the loop. For a system with herringbone gears, this condition accounts for the helix hand and ensures simultaneous engagement in both paths.
$$ \frac{z_1}{z_2}\alpha + \frac{z_4}{z_3}\beta – 2\delta = \frac{2\pi n}{z_2} $$
where $n$ is an integer. Typically, to simplify initial assembly, one branch is set to ideal mesh phasing at a nominal position, which often implies setting $\delta = 0$. This reduces the number of variables.
Solving this system of equations for a given set of gear parameters (like those in Table 1) yields specific shaft angle sets. The solution is not unique; different integer values for $n$ and slight adjustments can yield multiple valid configurations. For the purpose of dynamic comparison, two distinct sets of shaft angles, derived from the conditions above, are analyzed in this study. The basic gear parameters are summarized below:
| Parameter | Input Pinion (Stage 1) | Compound Gear (Stage 1) | Compound Pinion (Stage 2) | Output Gear (Stage 2) |
|---|---|---|---|---|
| Number of Teeth, $z$ | 29 | 108 | 29 | 116 |
| Normal Module, $m_n$ (mm) | 2.5 | 2.5 | 4.0 | 4.0 |
| Normal Pressure Angle, $\alpha_n$ (°) | 25 | 25 | 25 | 25 |
| Helix Angle, $\beta$ (°) | 30 | 30 | 30 | 30 |
Two example solutions for the shaft angles are:
| Shaft Angle Set | α (°) | β (°) |
|---|---|---|
| Set 1 | 2.9441 | 1.2661 |
| Set 2 | 3.0697 | 1.2723 |
Dynamic Modeling of the Herringbone Gear System
To analyze the dynamic effects of shaft angles, a comprehensive finite element node-based model is developed. This model incorporates the flexibility and inertia of all major components: shafts, herringbone gear rotors, bearings, and the gear meshes themselves. Each node in the model has six degrees of freedom (three translational and three rotational). For the system studied, a total of 67 nodes are used to discretize the input shaft, the two compound shafts, and the output shaft.
The governing equation of motion for the entire system is derived using rotor dynamics theory and can be expressed in matrix form as:
$$ \mathbf{M}\ddot{\mathbf{q}} + (\mathbf{C} + \Omega \mathbf{G})\dot{\mathbf{q}} + \mathbf{K}_f(\mathbf{q}) = \mathbf{F}(t) $$
where $\mathbf{M}$ is the global mass matrix, $\mathbf{C}$ is the damping matrix, $\mathbf{G}$ is the gyroscopic matrix, $\Omega$ is the rotational speed, $\mathbf{K}_f$ is the nonlinear stiffness matrix incorporating gear mesh stiffness, $\mathbf{q}$ is the vector of nodal displacements, and $\mathbf{F}(t)$ is the vector of external and internal excitation forces.
The modeling of the gear mesh is crucial and is where the shaft angles ($\alpha$, $\beta$) explicitly influence the dynamics. For a single helical gear pair, the relative displacement in the plane of action, $\delta_m$, is a function of the translational and rotational motions of the pinion (p) and gear (g). For a herringbone gear, each helix is modeled as a separate mesh with potential coupling. In a split-torque system with specific shaft angles, the effective pressure angle for each mesh is modified. The relative displacement for the i-th gear mesh in the system is given by:
$$ \delta_{ih} = (x_p – x_g)\sin\alpha_{m_i} + (y_p – y_g)\cos\alpha_{m_i} + (r_p\theta_{pz} + r_g\theta_{gz})\cos\beta_t + (r_p\theta_{py} + r_g\theta_{gy})\cos\alpha_{m_i} + (r_p\theta_{px} + r_g\theta_{gx})\sin\alpha_{m_i} + (z_g – z_p)\sin\beta_t – e_i(t) $$
Here, $\alpha_{m_i} = \alpha_n + \alpha_{p_i}$, where $\alpha_n$ is the nominal pressure angle and $\alpha_{p_i}$ is the angular offset due to the shaft angle for that specific mesh (e.g., $+\alpha/2$, $-\alpha/2$, $-\beta/2$, $+\beta/2$). $\beta_t$ is the helix angle, and $e_i(t)$ is the static transmission error excitation for that mesh.
The time-varying mesh stiffness, a primary source of parametric excitation, is modeled as:
$$ K_j(t) = K_{m_j} + \Delta k_j \sin(\omega_{h_i} t – z_i \gamma) $$
where $K_{m_j}$ is the mean mesh stiffness, $\Delta k_j$ is the stiffness variation amplitude, $\omega_{h_i}$ is the meshing frequency, $z_i$ is the number of teeth on the driving gear, and $\gamma$ is a phase angle directly related to the shaft angle ($\gamma = 0, \alpha, 0, \beta$ for the four meshes). This phase relationship is critical as it determines how the stiffness variations from different meshes interact, potentially amplifying or damping vibrations.
Similarly, the error excitation is modeled as:
$$ E_j(t) = E_{h_j}\sin(\omega_{h_i} t – z_i \gamma) + E_{s_j}\sin(\omega_{s_i} t – \gamma) $$
incorporating both manufacturing error harmonics and shaft order vibrations phase-shifted by the shaft angle.
Dynamic Analysis and the Impact of Shaft Angles
The dynamic analysis focuses on operational speeds relevant to aero-engine reducers: output speeds of 500, 800, 1000, and 1200 rpm. Key dynamic metrics examined include the dynamic transmission error (DTE) and the dynamic load factor (also called dynamic load coefficient) for each mesh. The DTE is a direct indicator of vibratory motion at the mesh, while the dynamic load factor indicates the magnification of load compared to the static condition, directly relating to load sharing between the two parallel paths.
Time and Frequency Domain Response: At an output speed of 500 rpm, the dynamic response for the two shaft angle sets shows notable differences. While the time-domain responses for the left and right helices of each herringbone gear pair are similar, they are not identical, with more significant differences observed in the first-stage herringbone gears. The most striking difference lies in the frequency domain. The system with Shaft Angle Set 1 exhibits a frequency spectrum dominated by a few distinct harmonics of the mesh frequency. In contrast, the system with Shaft Angle Set 2 excites a much richer set of frequency components, including more sidebands and higher-order harmonics. This suggests that Shaft Angle Set 2 creates a phasing condition that more effectively excites the system’s natural modes and/or promotes nonlinear interactions, even at the same nominal operating condition.
Dynamic Load Sharing and System Performance: To quantitatively compare the shaft angles across different speeds, the dynamic load factor and the root-mean-square (RMS) value of the DTE are calculated for all four operating points. The results are summarized conceptually below. Analysis focuses on two aspects: 1) The difference in response between the two parallel branches (indicative of load-sharing uniformity), and 2) The absolute magnitude of the response (indicative of overall vibration level).
Dynamic Load Factor Analysis:
* At 500 rpm, Set 1 results in a higher dynamic load factor with a slightly larger branch difference.
* At 800 and 1000 rpm, Set 1 yields a lower absolute dynamic load factor, but the imbalance between branches is more pronounced.
* At 1200 rpm, Set 1 again shows a lower dynamic load factor with better balance between branches.
* The dynamic load factors for the second-stage meshes are relatively similar for both angle sets across speeds.
Dynamic Transmission Error Analysis:
* The RMS value of the DTE is consistently and significantly lower for Shaft Angle Set 1 across all four operating speeds.
* Furthermore, the difference in DTE between the two branches is smaller for Set 1, indicating superior dynamic load-sharing characteristics from a vibration perspective.
This presents a nuanced picture. Judging by dynamic load factor alone does not give a clear winner; the preferred set depends on the operating speed. However, when considering dynamic transmission error—a more direct measure of vibration and noise potential—Shaft Angle Set 1 demonstrates unequivocally better performance, offering both lower vibration levels and more balanced operation between the power paths across the entire speed range examined.
Discussion and Engineering Implications
The findings underscore that shaft angles are not merely passive geometric parameters in a split-torque herringbone gear system; they are active design variables that fundamentally shape the system’s dynamic signature. The mechanism of influence is primarily through the phasing of excitations—both parametric (time-varying stiffness) and kinematic (transmission error)—across the four primary gear meshes. Different angle sets change the phase relationship $\gamma$ in the excitation terms $K_j(t)$ and $E_j(t)$. This phasing controls whether excitations from different meshes arrive at the shared components (input and output shafts) in phase, causing constructive interference and high vibration, or out of phase, causing partial cancellation and smoother operation.
The results highlight a critical task for designers: the multi-objective optimization of shaft angles. The goal is to find the angle set that simultaneously minimizes dynamic load (for durability), minimizes dynamic transmission error (for noise and vibration), and ensures equitable branch load sharing under all operational conditions. This often involves trade-offs, as seen in the dynamic load factor results. The superior DTE performance of Set 1 suggests it may be the preferred choice for applications where smooth, quiet operation is paramount, even if the dynamic load factor is marginally higher at certain speeds.
This analysis also reinforces the unique advantages of using herringbone gears in such systems. The double-helix design naturally mitigates axial vibration and provides a degree of redundancy and load distribution within each gear body, which interacts complexly with the system-level load sharing governed by the shaft angles. Future work could integrate more detailed models of bearing nonlinearities, housing flexibility, and the effects of minor manufacturing deviations (like lead and profile errors) from the nominal shaft angle design to create an even more robust design methodology for these high-performance herringbone gear split-torque transmissions.
Conclusion
This investigation into the dynamics of a split-torque transmission system employing herringbone gears has conclusively demonstrated the significant influence of shaft angles on system behavior. Through geometric constraint analysis, valid shaft angle sets were derived. A detailed finite element node-based dynamic model was developed, explicitly incorporating the shaft angles into the gear mesh excitation phasing. The dynamic analysis revealed that:
- Shaft angles drastically alter the frequency content of the system’s dynamic response. Different angle sets can excite a broader or narrower spectrum of frequencies, with significant implications for resonance avoidance and noise generation.
- The impact of shaft angles on key performance metrics like dynamic load factor and dynamic transmission error is speed-dependent. No single angle set is optimal for all metrics across all speeds, necessitating a system-level, multi-speed design evaluation.
- For the specific system and speed range studied, one shaft angle set (Set 1: α≈2.94°, β≈1.27°) provided consistently superior performance in minimizing dynamic transmission error and improving branch balance, making it a strong candidate for applications prioritizing vibrational performance.
Therefore, the careful calculation and selection of shaft angles must be considered an integral part of the design process for high-performance split-torque systems using herringbone gears, as they hold the key to unlocking the full potential of this compact and powerful transmission architecture.