The pursuit of high power density, compactness, and reliable power transmission has driven the development of advanced gear systems. Among these, the encased differential planetary transmission system, particularly when employing herringbone gears, represents a sophisticated solution for demanding applications in aerospace, marine propulsion, and heavy-duty industrial machinery. The unique double-helix structure of the herringbone gear effectively cancels out axial thrust forces, allowing for smoother operation and higher load capacity compared to single helical gears. The system’s architecture combines a differential planetary stage (Stage I) and a star gear stage (Stage II) in a closed-loop configuration, enabling significant speed reduction and torque multiplication through power splitting and reconvergence.
In this complex herringbone gear train, input power is first divided among multiple planet gears in the differential stage and subsequently recombined in the star stage. While this configuration offers substantial benefits, a critical challenge is ensuring equitable load distribution among the multiple power paths, i.e., among the planet gears in Stage I and the star gears in Stage II. Non-uniform load sharing can lead to premature failure of the most heavily loaded herringbone gear pairs, undermining the system’s inherent reliability and durability. The primary factors disrupting ideal load sharing are inevitable manufacturing inaccuracies, such as gear eccentricity and assembly (installation) errors, combined with system compliance. Therefore, a detailed static analysis to quantify the influence of these errors on load sharing is essential for robust design, precision tolerance specification, and informed assembly practices for the herringbone gear transmission.
System Configuration and Kinematic Fundamentals
The encased differential herringbone gear train, as analyzed herein, comprises two interconnected stages. Stage I is a differential planetary gearset consisting of a sun gear (Zs1), N planet gears (Zpi), a ring gear (Zr1), and the planet carrier (H). Stage II is a star gearset, which is a planetary system with a fixed carrier, containing a sun gear (Zs2), M star gears (Zmj), and a ring gear (Zr2). The two stages are connected via intermediate floating components—specifically, dual-gear couplings (Zg1, Zg2) and floating ring gears (Zf1, Zf2)—which are crucial for promoting load equalization. The input power is applied to the differential sun gear Zs1. The power is split, flowing through the planet carrier H and the ring gear Zr1 of Stage I. These two paths are then reconverged through the floating components onto the output shaft L. The kinematic relationships and torque flow are foundational for the static force analysis.
The fundamental speed ratio for the differential stage (Stage I) is given by:
$$ i_{I} = 1 + \frac{Z_{r1}}{Z_{s1}} $$
where $Z_{r1}$ and $Z_{s1}$ are the tooth numbers of the ring and sun gears in Stage I, respectively. The speed of the planet carrier $n_H$ relative to the sun gear speed $n_{s1}$ and ring gear speed $n_{r1}$ follows the planetary gear relation:
$$ n_{s1} – i_{I} \cdot n_{H} + (i_{I} – 1) \cdot n_{r1} = 0 $$
For the star stage (Stage II), with its carrier fixed, the speed ratio is:
$$ i_{II} = -\frac{Z_{r2}}{Z_{s2}} $$
The negative sign indicates direction reversal. The overall system ratio is determined by the constraint equations imposed by the connections between the floating ring gears (Zf1, Zf2), the intermediate components, and the output.
Static Load Sharing Model Formulation
A lumped-parameter static model is developed to analyze the load distribution. The model accounts for the translational and rotational degrees of freedom (DOF) of all major components, the meshing stiffness of the herringbone gear pairs, the support stiffness of bearings, and the stiffness of the intermediate floating connections. The total number of degrees of freedom for a system with N planet gears and M star gears is $18 + 3(N+M)$.
The generalized displacement vector $\mathbf{X}$ is defined as:
$$ \mathbf{X} = \{ x_{s1}, H_{s1}, V_{s1}, x_{pi}, H_{pi}, V_{pi}, x_{r1}, H_{r1}, V_{r1}, x_{g1}, x_{f1}, x_{s2}, H_{s2}, V_{s2}, x_{mj}, H_{mj}, V_{mj}, x_{r2}, H_{r2}, V_{r2}, x_{g2}, x_{f2}, x_{H}, x_{L} \}^T $$
where $x$ denotes rotational displacement (converted to linear displacement at the base circle or pitch circle radius), and $H$ and $V$ denote translational displacements in the horizontal and vertical directions (or in rotating coordinates for Stage I components).
The elastic meshing force for an external herringbone gear pair (e.g., sun-planet in Stage I) along the line of action is calculated as:
$$ P_{spi} = K_{sp} \cdot \delta_{spi} $$
where $K_{sp}$ is the average mesh stiffness of the herringbone gear pair, and $\delta_{spi}$ is the relative deflection along the line of action, incorporating gear displacements and manufacturing errors $e_{spi}$:
$$ \delta_{spi} = \left( x_{s1} – x_{pi} \right) + \left( H_{s1}\cos\psi_{s1i} + V_{s1}\sin\psi_{s1i} \right) – \left( H_{pi}\cos\psi_{pi} + V_{pi}\sin\psi_{pi} \right) – e_{spi} $$
Here, $\psi$ terms represent the projection angles of translational displacements onto the line of action, determined by the pressure angle $\alpha$ and the planet’s position angle $\phi_i = 2\pi(i-1)/N$. A similar formulation, with appropriate sign conventions and projection angles, applies to internal meshes (planet-ring, star-ring) and meshes in Stage II.
The average mesh stiffness $K_m$ for a herringbone gear pair is computed as the parallel sum of the stiffnesses of its two constituent helical gear halves, following standards such as ISO 6336:
$$ K_m = \frac{1}{2} (K_{\varepsilon, helix1} + K_{\varepsilon, helix2}) $$
where $K_{\varepsilon, helix}$ is the total mesh stiffness per unit face width for a helical gear, accounting for base tooth stiffness and deflection factors.
The complete system of static equilibrium equations is derived from force and moment balance for each component. This includes equations for the sun, planet, and ring gears in both stages, the planet carrier, the intermediate floating components (Zg1, Zf1, Zg2, Zf2), and the output shaft. The equations consider input torque $T_D$, output load torque $T_L$, gravitational forces $G$, centrifugal forces $F_{pi}$ on planets in the rotating frame, and all elastic meshing and support forces. For example, the equilibrium for a planet gear in Stage I involves forces from its two meshes and its bearing support:
$$ \begin{aligned}
-P_{spi} + P_{rpi} &= 0 \quad \text{(Torque balance)} \\
K_p H_{pi} – P_{spi}\sin\alpha_2 + P_{rpi}\sin\alpha_1 &= F_{cent, H} \\
K_p (V_{pi} – x_H) – P_{spi}\cos\alpha_2 + P_{rpi}\cos\alpha_1 &= F_{cent, V} – G_p
\end{aligned} $$
The equations for the intermediate floating components are critical, as they enable self-alignment and load sharing. For instance, for the first intermediate component Zg1:
$$ -K_{g1q}(x_{r1} – x_{g1}) + K_{f1q}(x_{g1} – x_{f1}) = 0 $$
This equation ensures the torque transmitted through the floating connection is consistent.
Modeling of Manufacturing and Assembly Errors
The primary errors considered are gear eccentricity errors and assembly (installation) errors of the central members (sun and ring gears). These errors are modeled as equivalent displacements along the lines of action of the meshes.
Eccentricity Error: The eccentricity $E_j$ of a gear (e.g., the sun gear) introduces a harmonic displacement perturbation. Its projection onto the mesh line with a planet/star at position angle $\phi_i$ is:
$$ e_{ecc} = -E_j \sin(\omega_{rel} t + \varphi_j \mp \alpha) $$
where $\omega_{rel}$ is the relative rotational frequency of the gear, $\varphi_j$ is the initial phase angle of the eccentricity, $\alpha$ is the mesh pressure angle, and the sign depends on whether it’s an external or internal mesh. For the fixed carrier in Stage II, $\omega_{rel}$ is simply the gear’s absolute speed.
Assembly Error: The radial installation error $A_k$ of a central gear (e.g., sun gear bearing bore misalignment) acts as a static offset. Its projection is:
$$ e_{inst} = -A_k \sin(\mp \delta_k \pm \phi_i + \alpha) $$
where $\delta_k$ is the angular position of the maximum error vector. The phase relationship $\phi_i$ between different planet/star gears is key to understanding load distribution patterns.
The total equivalent error $e_{total}$ on a specific mesh is the sum of the contributions from the eccentricities and installation errors of the two gears forming that pair. The error vectors for all planets/stars form a set of forcing functions in the static equilibrium equations. The phase differences between these error excitations are governed by the mesh phasing relationships. For a system with equally spaced planets/stars, the phase shift $\Phi$ between the error excitation for planet $i$ and planet $1$ in a sun-planet mesh is:
$$ \Phi_{sp}(i) = (\omega_s – \omega_c) \cdot \frac{Z_s}{N} \cdot (i-1)\frac{2\pi}{N} $$
where $\omega_s$ and $\omega_c$ are the absolute speeds of the sun and carrier, and $Z_s$ is the sun gear tooth number. This phasing determines whether error effects accumulate or cancel across the gear mesh.
Load Sharing Coefficient and Solution Methodology
The uniformity of load distribution is quantified using the Load Sharing Factor (LSF). For the planet/star gears, it is defined based on the external mesh force. The LSF for the i-th planet gear in Stage I is:
$$ \Omega_{pi} = \frac{F_{spi} \cdot N}{T_D / r_{bs1}} $$
where $F_{spi}$ is the calculated sun-planet mesh force, $N$ is the number of planets, $T_D$ is the input torque, and $r_{bs1}$ is the base radius of the sun gear Zs1. Similarly, for the j-th star gear in Stage II:
$$ \Omega_{mj} = \frac{F_{smj} \cdot M}{T_2 / r_{bs2}} $$
where $T_2$ is the torque acting on the sun gear Zs2. A perfectly uniform distribution yields $\Omega = 1.0$ for all paths. The overall stage LSF is taken as the maximum deviation:
$$ \Omega_{stage} = \max(|\Omega_i – 1|) + 1 $$
The static equilibrium equations, incorporating the stiffness matrix and the error displacement vectors, form a large linear system of the form $\mathbf{K} \cdot \mathbf{X} = \mathbf{F} + \mathbf{F}_{error}$. The vector $\mathbf{X}$ of all displacements is solved numerically. Subsequently, the mesh forces $F_{spi}$ and $F_{smj}$ are computed from the displacement differences, leading to the calculation of the LSFs.
Analysis of Results and Parametric Studies
The following analysis is based on a representative high-power encased differential herringbone gear transmission system with the parameters summarized in Table 1. Stage I has three planet gears (N=3), while Stage II has five star gears (M=5). The base values for eccentricity and installation errors are set at 32 μm.
| Parameter | Stage I (Differential) | Stage II (Star) |
|---|---|---|
| Module (mm) | 7 | 7 |
| Number of Planets/Stars | N = 3 | M = 5 |
| Sun Gear Teeth | Zs1 = 37 | Zs2 = 71 |
| Planet/Star Gear Teeth | Zp = 56 | Zm = 39 |
| Ring Gear Teeth | Zr1 = 149 | Zr2 = 149 |
| Helix Angle | β₁ = 22° | β₂ = 22° |
| Pressure Angle | αn₁ = 20° | αn₂ = 20° |
Influence of Eccentricity Errors
The impact of eccentricity errors in the central gears (sun and ring) on the LSF is systematic. As shown in Figure 1 (conceptual trend), increasing the eccentricity error of either the sun or ring gear in either stage leads to a nearly linear increase in the Load Sharing Factor for that stage. This is because the error directly creates unequal paths for load transmission. A notable finding concerns the eccentricity errors of the planet and star gears themselves. When the initial phase angles $\varphi_{pi}$ or $\varphi_{mj}$ of these errors are random or different from one another, they contribute to load imbalance. However, if the phase angles are intentionally set to be identical for all planets (or all stars), a significant cancellation effect occurs. In this synchronized condition, the equivalent mesh displacement caused by the planet eccentricities becomes identical for all planets, and the floating central members (sun/ring) can compensate uniformly, rendering the magnitude of the planet/star eccentricity irrelevant to the final LSF. This is a crucial insight for high-precision assembly of critical herringbone gear systems.
Influence of Installation Errors
Installation errors of the central gears also degrade load sharing. Similar to eccentricity, larger installation errors $A_k$ result in higher LSFs, as depicted in Figure 2 (conceptual trend). An important distinction arises between the two stages due to kinematics. In Stage I (differential), the planet gears revolve around the sun. Therefore, a static installation error of the sun or ring gear manifests as a time-varying excitation from the perspective of each planet gear as it orbits. Consequently, the load on each planet gear in Stage I fluctuates with the carrier’s rotation. In contrast, in Stage II (star), the carrier is fixed. The installation error presents a constant offset relative to each stationary star gear. Thus, the load imbalance among the star gears is static and does not vary with time, although it still results in a non-uniform LSF.
Combined Error Effects and Stage Sensitivity
The effects of eccentricity and installation errors are cumulative. When both types of errors are present, the resulting LSF is greater than that caused by either error type acting alone. This additive nature underscores the importance of controlling all sources of geometric inaccuracy in the manufacture and assembly of the herringbone gear train.
A critical observation from the analysis is the differential sensitivity of the two stages to errors. Consistently, the Load Sharing Factor for Stage I (differential) is found to be significantly higher than that for Stage II (star) under identical error conditions. This can be attributed to the fundamental load levels in each stage. Stage II, being the final reduction stage before the output, carries a much higher torque load than Stage I. It is a well-established principle in gear system design that higher transmitted loads tend to “swamp” the effects of fixed errors due to increased elastic deflections, promoting more uniform load sharing. Consequently, Stage II exhibits a more robust and less error-sensitive load distribution compared to Stage I. This makes the precision control of the differential stage particularly vital for the overall system’s reliability.
| Error Condition | Stage I (Differential) LSF | Stage II (Star) LSF | Key Observation |
|---|---|---|---|
| Eccentricity Error Only (Base) | 1.0522 | 1.0252 | Stage I is more sensitive. |
| Installation Error Only (Base) | 1.0255 | 1.0121 | Stage I LSF varies with time; Stage II LSF is constant. |
| Combined Errors (Base) | 1.0771 | 1.0370 | Effects are cumulative. Stage I LSF > Stage II LSF. |
| Planet/Star Eccentricity (Synchronized Phase) | ~1.0698 (constant) | ~1.0452 (constant) | LSF becomes independent of planet/star eccentricity magnitude. |
Conclusions
The static analysis of the encased differential herringbone gear transmission system reveals intricate relationships between manufacturing/assembly errors and load sharing performance. The model, incorporating the essential compliance of gears, supports, and the pivotal intermediate floating components, provides a reliable tool for such investigations. The key conclusions are:
- Error Impact: Both gear eccentricity and installation errors degrade load sharing, and their effects are additive. Larger errors lead to higher Load Sharing Factors (LSFs).
- Phase Synchronization of Planet/Star Errors: A strategically important finding is that if the initial phase angles of eccentricity errors are identical for all planet gears (or all star gears), the magnitude of these errors has no influence on the system’s static LSF. This presents a potential precision assembly technique—phase matching—for achieving superior load sharing in high-performance herringbone gear systems, albeit at potentially higher cost.
- Kinematic Effect on Error Manifestation: Installation errors affect the two stages differently. In the rotating differential stage, they cause time-varying planet loads. In the fixed-carrier star stage, they create a constant, time-invariant load imbalance among the star gears.
- Stage-Wise Sensitivity: The differential stage (Stage I) is significantly more sensitive to errors and exhibits a higher LSF than the star stage (Stage II). This is primarily due to the lower torque load in Stage I, making its load distribution more susceptible to disruption by fixed geometric errors. Therefore, tighter tolerance controls should be prioritized for the components of the differential stage in the design and manufacturing of the herringbone gear train.
This work establishes a foundational understanding for the static load sharing behavior in complex encased differential systems utilizing herringbone gears. The insights gained are directly applicable to setting error tolerance limits, guiding assembly processes, and informing the design of floating elements to ensure reliable and efficient operation of these advanced transmission systems.