Spiral bevel gears transmit motion and torque between intersecting axes through curved teeth that engage gradually along the face width. Their tooth flanks are generated by complex machine-tool motions, and the loaded contact pattern is sensitive to shaft deflection, bearing stiffness, mounting position, and assembly orientation. In aerospace, marine, and automotive drivetrains, design optimization therefore requires a continuous chain from system topology to component geometry, from geometry to structural stiffness, and from stiffness back to tooth contact.
General-purpose CAD and CAE software can build gear models and perform contact analysis, but the workflow usually involves repeated data exchange, manual assembly, and separate post-processing. Dedicated gear software offers interactive visualization, but the underlying data structures and rendering methods are often proprietary. A practical engineering requirement is to keep topology, geometry, stiffness, and contact results within one parameterized environment so that the model and the analysis output remain visible and traceable.
Topology representation and object data
A spiral bevel gear transmission system can be decomposed into shaft systems. Each shaft system contains a transmission shaft, one or more gears, bearings, and an external load. Gear pairs connect shaft systems and define the power flow. This hierarchy is represented by System, ShaftSystem, Shaft, Gear, Bearing, Load, and Pair objects. Material properties, geometric parameters, mounting positions, and load data are stored in the corresponding object, so the same parameter set can drive component modeling, assembly, and mechanical analysis.
Connectivity is encoded with graph adjacency matrices. For an input shaft system containing the input shaft, two bearings, the pinion, and the input torque, the adjacency matrix has the form
$$
\mathbf A_1=
\begin{bmatrix}
0&1&1&1&1\\
1&0&0&0&0\\
1&0&0&0&0\\
1&0&0&0&0\\
1&0&0&0&0
\end{bmatrix},
$$
where the rows and columns correspond to the input shaft, bearing 1, bearing 2, pinion, and input torque. A value of 1 indicates a connection; a value of 0 indicates no direct connection. A gear-pair matrix \(\mathbf B_{12}\) contains the coupling between the pinion on the input shaft system and the wheel on the output shaft system. For a multi-stage system, the shaft-system matrices are placed along the diagonal and the gear-pair matrices occupy the off-diagonal blocks:
$$
\mathbf A=
\begin{bmatrix}
\mathbf A_1&\mathbf B_{12}&\mathbf 0&\cdots&\mathbf 0\\
\mathbf B_{12}^{\mathrm T}&\mathbf A_2&\mathbf B_{23}&\cdots&\mathbf 0\\
\mathbf 0&\mathbf B_{23}^{\mathrm T}&\mathbf A_3&\cdots&\mathbf 0\\
\vdots&\vdots&\vdots&\ddots&\vdots\\
\mathbf 0&\mathbf 0&\mathbf 0&\cdots&\mathbf A_n
\end{bmatrix}.
$$
Traversing a row of the system matrix returns the components connected to the corresponding component. This makes it possible to query bearings, gears, loads, and adjacent shaft systems directly from the data model rather than reconstructing those relationships from geometric entities.
Geometric modeling of heterogeneous components
VTK, the Visualization Toolkit, provides an open-source, cross-platform, object-oriented environment for three-dimensional graphics. Its visualization pipeline separates data generation from rendering: sources create points and cells, filters transform the data, mappers convert data into graphics primitives, actors are placed in a renderer, and a render window displays the result. This separation is useful for gear transmission software because geometric parameters, analysis results, and display settings can be updated independently.
Shafts and bearings are rotational parts and are modeled by parametric cross-section extrusion. For a shaft, each segment is defined by its length \(L_i\), outer diameter \(D_i\), and inner diameter \(d_i\). A rectangular cross-section is constructed at the left end, and the section is rotated about the shaft axis with a rotational extrusion filter. Concatenating all segments produces the complete shaft. For an angular contact ball bearing, the bearing object stores the bore diameter \(d\), outer diameter \(D\), width \(B\), contact angle \(\alpha\), and rolling-element parameters. Inner- and outer-ring profiles are derived from these parameters, revolved into ring bodies, and combined with spherical rolling elements placed by diameter and number. The bearing components are constructed in a common coordinate system so that ring geometry constrains the rolling-element positions.
The gear model requires a different approach because the tooth flank is a complex generated surface. A Gear object stores blank dimensions, cutter parameters, machine settings, and tooth-surface parameters. Meshing equations map a regular grid on the tooth-surface topology into three-dimensional points. In the representative case, the face-width direction uses \(M=67\) points and the tooth-height direction uses \(N=101\) points, giving \(67\times101=6{,}767\) points per flank. The regular grid is important because it provides a structured point set for both surface construction and later contact-result mapping.
Each grid cell is split into two triangular cells. The concave flank, convex flank, toe end, heel end, tooth tip, and tooth root are assembled as six boundary surfaces, stored in a cell array, and integrated into a polygonal data set. A single tooth is then rotated and arrayed around the gear axis, and the gear blank is added to complete the model. This triangulation approach represents the generated tooth surface without requiring a closed-form solid model and preserves the point correspondence needed for stress visualization.
Assembly uses homogeneous coordinate transformations. The shaft coordinate system is the reference for each shaft system. A gear is placed at axial position \(S_j\) and rotated by angle \(\beta\) about the shaft axis:
$$
\mathbf V_g^{(s)}=\mathbf T(\beta,S_j)\,\mathbf V_g^{(g)},
$$
where \(\mathbf V_g^{(g)}\) is the gear point set in its local coordinate system, \(\mathbf V_g^{(s)}\) is the transformed point set in the shaft coordinate system, and \(\mathbf T\) is the corresponding \(4\times4\) transformation matrix. Bearings are oriented according to their mounting direction and placed at their axial positions. The shaft systems are then transformed into the gearbox coordinate system using a rotation matrix \(\mathbf M(\theta)\) and translation components \(t_x\), \(t_y\), and \(t_z\):
$$
\mathbf V_{ss}^{(\mathrm{box})}
=\mathbf T(\theta,t_x,t_y,t_z)\,\mathbf V_{ss}^{(s)}.
$$
This hierarchical assembly keeps each component parameterized in its local frame while allowing the complete transmission to be positioned by shaft-system-level mounting parameters.
Beam-based mechanical mapping and loaded contact analysis
The shaft system is the basic unit for mechanical analysis. A beam-element model is generated from the same geometric and assembly data used for visualization. Nodes are placed at shaft-section transitions \(L_i\), component mounting positions \(S_i\), load application points, boundary-condition locations, and internal-bore transitions \(l_i\). Each node has six degrees of freedom. Beam-element stiffness matrices are assembled at shared nodes, and bearing stiffness matrices are coupled at the bearing nodes to represent support flexibility. The resulting system equation is
$$
\mathbf K\boldsymbol{\delta}=\mathbf F,
$$
where \(\mathbf K\) is the assembled stiffness matrix, \(\boldsymbol{\delta}\) is the nodal displacement vector, and \(\mathbf F\) is the external load vector. The gear mesh force is calculated from the input torque and the gear-pair geometry and is applied at the corresponding node. Solving the equation gives the displacement and rotation at every node of the shaft system.
The displacements at the gear mounting positions are converted into relative mesh misalignment between the pinion and the wheel. Four quantities are used: \(\Delta P\), the relative displacement along the pinion axis; \(\Delta W\), the relative displacement along the wheel axis; \(\Delta E\), the relative displacement along the offset direction; and \(\Delta\Sigma\), the relative angular displacement about the shaft angle. These quantities shift the theoretical contact path away from the conjugate position and therefore change the loaded contact pattern.
Loaded tooth contact analysis combines the misalignment state with the tooth-surface geometry. A conjugate difference surface defines the initial contact path. Single-tooth stiffness is calculated with a slice method and a series-parallel stiffness model. Multi-tooth load sharing is then solved from moment equilibrium and deformation compatibility. The distributed normal load is introduced into a Hertzian contact model, which gives the contact ellipse and the contact stress. For a normal contact load \(F’\), ellipse semi-axes \(a\) and \(b\), and local coordinates \(x\) and \(y\),
$$
P_{\max}=\frac{3F’}{2\pi ab},
\qquad
P(x,y)=P_{\max}\sqrt{1-\frac{x^2}{a^2}-\frac{y^2}{b^2}}.
$$
The contact ellipses along the contact path form the contact imprint. Stress values at the tooth-surface points are mapped through a color lookup table, with low stress and high stress assigned to different colors, and the colored polygonal surface is rendered as a contact stress nephogram. Because the stress is attached to the same tooth-surface grid used for geometric modeling, the loaded contact state can be displayed without rebuilding the display mesh.
Case study: assembly, deformation, and contact pattern
A representative spiral bevel gear transmission was modeled with an input torque of \(400\ \mathrm{N\cdot m}\). The gear-pair parameters are listed in Table 1, the shaft sections in Table 2, and the bearing designations in Table 3.
| Parameter | Pinion | Wheel |
|---|---|---|
| Number of teeth | 28 | 41 |
| Module (mm) | 3.75 | 3.75 |
| Shaft angle (°) | 90 | 90 |
| Offset (mm) | 0 | 0 |
| Pressure angle (°) | 20 | 20 |
| Face width (mm) | 28 | 28 |
| Outer cone distance (mm) | 93.09 | 93.09 |
| Spiral angle (°) | 35 | 35 |
| Pitch cone angle (°) | 34.33 | 55.67 |
| Shaft | Section | Length (mm) | Outer diameter (mm) | Inner diameter (mm) |
|---|---|---|---|---|
| Input shaft | 1 | 35 | 60 | 25 |
| 2 | 90 | 55 | 25 | |
| 3 | 75 | 50 | 25 | |
| Output shaft | 1 | 30 | 90 | 40 |
| 2 | 100 | 85 | 40 | |
| 3 | 70 | 80 | 40 |
| Bearing position | Designation |
|---|---|
| 1 | 7011C |
| 2 | 7010C |
| 3 | 7017C |
| 4 | 7016C |
Using these parameters, the component models were generated and assembled into the complete transmission. The assembly reproduced the shaft, bearing, and gear geometry and showed no interference between components. This confirms that the object-level positioning data and the coordinate transformations are consistent from the component stage to the system stage.
The input-shaft deformation from the beam model was compared with a finite element solution at nine nodes. The results are listed in Table 4. The maximum relative difference was no more than 1.0%, and the Z-direction values agreed to within 0.1% at all listed nodes.
| Node | X direction (µm) | Y direction (µm) | Z direction (µm) | |||
|---|---|---|---|---|---|---|
| FEM | Beam model (error) | FEM | Beam model (error) | FEM | Beam model (error) | |
| 1 | -47.63 | -48.05 (0.9%) | 36.31 | 36.51 (0.6%) | 10.73 | 10.73 (0.0%) |
| 2 | -42.95 | -43.25 (0.7%) | 31.33 | 31.49 (0.5%) | 10.73 | 10.73 (0.0%) |
| 3 | -30.39 | -30.57 (0.6%) | 19.66 | 19.63 (0.2%) | 10.40 | 10.40 (0.0%) |
| 4 | -25.05 | -25.00 (0.2%) | 15.11 | 15.00 (0.7%) | 10.23 | 10.22 (0.1%) |
| 5 | -10.56 | -10.54 (0.2%) | 2.79 | 2.78 (0.4%) | 9.72 | 9.72 (0.0%) |
| 6 | 1.02 | 1.01 (1.0%) | -5.99 | -6.00 (0.2%) | 9.21 | 9.21 (0.0%) |
| 7 | 3.55 | 3.52 (1.0%) | -7.80 | -7.73 (0.9%) | 9.04 | 9.04 (0.0%) |
| 8 | 11.87 | 11.92 (0.4%) | -14.23 | -14.35 (0.8%) | 9.04 | 9.04 (0.0%) |
| 9 | 20.20 | 20.33 (0.6%) | -20.96 | -20.81 (0.7%) | 9.04 | 9.04 (0.0%) |
Using the relative displacements at the gear mounting positions, the mesh misalignment was calculated as listed in Table 5. The largest linear component was along the offset direction, \(-122.8\ \mu\mathrm m\), and the relative angular displacement about the shaft angle was \(0.262\ \mathrm{mrad}\).
| Quantity | Value | Meaning |
|---|---|---|
| \(\Delta P\) | 65.02 µm | Relative displacement along the pinion axis |
| \(\Delta W\) | -54.84 µm | Relative displacement along the wheel axis |
| \(\Delta E\) | -122.8 µm | Relative displacement along the offset direction |
| \(\Delta\Sigma\) | 0.262 mrad | Relative angular displacement about the shaft angle |
The initial design, under light load and without system deformation, placed the theoretical contact area as a diamond-shaped region in the middle of the tooth. When the load-induced shaft and bearing deflections were included, the contact area moved toward the outer end, or heel, and the tooth tip, producing a clear edge-contact condition. The tooth surface was then modified so that the unloaded contact area shifted toward the inner end, or toe, and the root, and the pattern was made narrower in the tooth-height direction. Under load, the contact area returned to the middle of the face, and no edge contact appeared at either the tooth tip or the root.
The loaded contact stress nephogram showed elliptical contact zones distributed along the contact path. The maximum stress occurred at the center of each ellipse and decreased toward the boundary. Neither the light-load nor the heavy-load contact pattern reached the tooth tip or root. This result illustrates why the deformation model and the contact visualization need to share the same tooth-surface grid: the shift from an edge-contact state to a centered contact state can be interpreted directly in terms of the calculated misalignment and the modified tooth geometry.
Engineering implications
The modeling chain has three connected elements. First, graph-based topology and object-oriented parameter storage provide a complete representation of the transmission configuration and its component relationships. Second, parametric section extrusion, swept construction, and triangulation in VTK allow shafts, bearings, and generated spiral bevel tooth surfaces to be built and assembled in one graphics environment. Third, the mapping from geometric nodes to beam elements and from tooth-surface points to contact-stress values allows system deformation and loaded tooth contact to be displayed without separating design geometry from analysis results.
For dedicated spiral bevel gear software, this organization reduces manual data transfer between CAD, structural analysis, and post-processing. The case study supports the beam model at the system-deformation level, with input-shaft displacements within 1.0% of the finite element reference. The contact results are conditioned on the beam stiffness, bearing stiffness, load-sharing assumptions, and tooth-surface modification used in the analysis. Within those conditions, the combination of topology data, VTK-based geometry, and loaded contact visualization provides a practical basis for interactive design and analysis of spiral bevel gear transmission systems.