Helical gears are widely employed in power transmission applications such as agricultural machinery, new energy vehicles, and wind power equipment due to their advantages of high load-carrying capacity, large contact ratio, and smooth operation. However, factors like assembly and manufacturing errors can induce meshing impact during tooth engagement, leading to increased vibration and noise. Diagonal modification has proven to be an effective method for mitigating these adverse effects by selectively removing material from specific regions of the tooth flank. Therefore, achieving a precise three-dimensional model of a diagonally modified helical gear is of paramount importance for subsequent analyses, including meshing impact force calculation, loaded tooth contact analysis (LTCA), and finite element simulation (FEA).
This study focuses on developing a high-fidelity reverse modeling methodology for diagonally modified helical gears. The process begins with establishing a comprehensive mathematical model of the gear tooth surface, encompassing both the standard involute working flank and the fillet transition curve. A diagonal modification scheme is then analytically defined and superimposed onto the standard flank. To address potential discontinuities at the boundary between the modified flank and the root fillet, a Hermite interpolation technique is employed to ensure a smooth connection. Finally, the digital tooth surface data is imported into CATIA V5 R20 to construct a precise 3D solid model, and its accuracy is rigorously verified.
The foundation of accurate modeling lies in the precise mathematical description of the helical gear tooth surface. The generation process is simulated using a imaginary rack cutter. A coordinate system transformation approach is employed to derive the equations of the generated gear flank. The right-hand flank of a right-hand helical gear is considered initially; the left flank can be obtained through symmetry or by defining the left cutting edge of the rack.
Let us define the coordinate systems. The rack cutter coordinate system \( S_t(x_t, y_t, z_t) \) is fixed to its left flank. The gear blank coordinate system \( S_1(x_1, y_1, z_1) \) rotates with the gear. Through a series of intermediate coordinate transformations involving translation and rotation related to the gear’s rotation angle \( \theta_1 \), helical motion, and tool geometry, the final transformation matrix \( \mathbf{M}_{1t} \) from \( S_t \) to \( S_1 \) is established. Its linear sub-matrix (removing the last row and column) is denoted as \( \mathbf{L}_{1t} \).
The right cutting edge of the rack (for generating the right gear flank) is a straight line in the normal section. Its position vector and unit normal vector in \( S_t \) can be expressed as:
$$ \mathbf{r}_t(u_t, l_t) = [u_t, 0, l_t, 1]^T $$
$$ \mathbf{n}_t(u_t, l_t) = [0, 1, 0]^T $$
where \( u_t \) and \( l_t \) are the surface parameters along the profile and length directions, respectively.
According to the theory of gearing, the position vector \( \mathbf{r}_1 \), unit normal vector \( \mathbf{n}_1 \) of the generated standard involute helical gear flank in \( S_1 \), and the equation of meshing \( f \) are given by:
$$ \mathbf{r}_1(u_t, l_t, \theta_1) = \mathbf{M}_{1t} \cdot \mathbf{r}_t(u_t, l_t) $$
$$ \mathbf{n}_1(u_t, l_t, \theta_1) = \mathbf{L}_{1t} \cdot \mathbf{n}_t(u_t, l_t) $$
$$ f(u_t, l_t, \theta_1) = \mathbf{n}_1(u_t, l_t, \theta_1) \cdot \frac{\partial \mathbf{r}_1(u_t, l_t, \theta_1)}{\partial \theta_1} = 0 $$
The fillet transition curve is generated by the tip rounding of the rack cutter. Assuming the rounding is a circular arc with radius \( r_w \), its position vector in \( S_t \) is parameterized by an angle \( \varphi_t \):
$$ \mathbf{r’}_t(\varphi_t, l_t) = [-r_w \sin \varphi_t – m_n \tan \alpha_n, \,\, r_w \cos \varphi_t – r_w, \,\, l_t, \,\, 1]^T $$
where \( m_n \) is the normal module, \( \alpha_n \) is the normal pressure angle, and \( \varphi_t \in [0, \pi/2 – \alpha_n] \). Its unit normal vector \( \mathbf{n’}_t \) is calculated as:
$$ \mathbf{n’}_t(\varphi_t, l_t) = \frac{ \partial \mathbf{r’}_t / \partial \varphi_t \times \partial \mathbf{r’}_t / \partial l_t }{ \lVert \partial \mathbf{r’}_t / \partial \varphi_t \times \partial \mathbf{r’}_t / \partial l_t \rVert } $$
The corresponding generated fillet surface on the helical gear and its meshing condition are:
$$ \mathbf{r’}_1(\varphi_t, l_t, \theta_1) = \mathbf{M}_{1t} \cdot \mathbf{r’}_t(\varphi_t, l_t) $$
$$ \mathbf{n’}_1(\varphi_t, l_t, \theta_1) = \mathbf{L}_{1t} \cdot \mathbf{n’}_t(\varphi_t, l_t) $$
$$ f(\varphi_t, l_t, \theta_1) = \mathbf{n’}_1(\varphi_t, l_t, \theta_1) \cdot \frac{\partial \mathbf{r’}_1(\varphi_t, l_t, \theta_1)}{\partial \theta_1} = 0 $$
Diagonal modification aims to remove material primarily from the regions near the tooth tips at the entry and exit ends of the meshing cycle, while the central portion of the tooth remains largely unmodified or lightly modified. This creates a localized relief pattern that is diagonal across the tooth face width when viewed from the side.
For a right-hand helical gear left flank, the modification boundaries are defined. The tip relief termination is at the tip circle with radius \( r_{a1} \). The root relief termination is at a radius \( r_{k1} \), defined as the start of active profile (SAP) which ensures the minimum required length of contact for the involute profile. It can be calculated based on the gear pair geometry:
$$ r_{k1} = \sqrt{ r_{b1}^2 + \left[ (r_{p1}+r_{p2}) \sin \alpha_t – \sqrt{r_{a2}^2 – r_{b2}^2} \right]^2 } $$
Here, \( r_{b1}, r_{b2} \) are base circle radii, \( r_{p1}, r_{p2} \) are pitch circle radii, \( r_{a2} \) is the mating gear’s tip radius, and \( \alpha_t \) is the transverse pressure angle.
To define the modification, a “rotated projection plane” is utilized where the gear’s cylindrical coordinates are mapped: \( x_2 = \sqrt{x_1^2 + y_1^2} \) and \( z_2 = z_1 \). Key points are defined on this plane: point B at the tip-relief start (\(z_2^{(B)}, r_{a1} – AB\)), and point E at the root-relief start (\(z_2^{(E)}, r_{k1} + DE\)), where \( AB \) and \( DE \) are the specified modification heights, and \( B_1 \) is the face width.
The corresponding gear rotation angles \( \theta_1^{(B)} \) and \( \theta_1^{(E)} \) for these boundary points are solved using the surface equations and the mapping. By the principle that points on the same line of contact share the same \( \theta_1 \), the modification lengths \( AC \) and \( DF \) along the face width are determined. The start lines of modification (BC and EF) are approximated as straight lines on this projection plane for simplicity. Their effective helix angles \( \beta_a \) and \( \beta_d \) are:
$$ \beta_a = \arctan(AB / AC), \quad \beta_d = \arctan(DE / DF) $$
Within the triangular modification zones \( \triangle ABC \) (tip) and \( \triangle DEF \) (root), the modification amount \( \delta \) at any point \( G \) is defined based on its perpendicular distance \( l_G \) to the start line (BC or EF). The amount increases from zero at the start line to a maximum \( C_a \) or \( C_d \) at the boundary line (tip or root circle). The variation can be linear, parabolic, or of a higher order.
$$ \delta(z_2, x_2) =
\begin{cases}
\left( \dfrac{l_G}{l_a} \right)^{k_a} C_a, & \text{if } G \in \triangle ABC \quad \text{(Tip Relief)} \\[10pt]
0, & \text{elsewhere} \\[10pt]
\left( \dfrac{l_G}{l_d} \right)^{k_d} C_d, & \text{if } G \in \triangle DEF \quad \text{(Root Relief)}
\end{cases} $$
Here, \( l_a \) and \( l_d \) are the total distances from the boundary circle to the start line, and \( k_a, k_d \) define the modification curve exponent (e.g., 1 for linear, 2 for parabolic).
The modified flank surface \( \mathbf{r}_{1m} \) is modeled as the standard flank \( \mathbf{r}_1 \) plus a superposition of the modification amount \( \delta \) along the unit normal vector \( \mathbf{n}_1 \):
$$ \mathbf{r}_{1m}(u_t, l_t, \theta_1) = \mathbf{r}_{1}(u_t, l_t, \theta_1) + \delta(z_2, x_2) \, \mathbf{n}_{1}(u_t, l_t, \theta_1) $$
The unit normal vector of the modified surface \( \mathbf{n}_{1m} \) must be recalculated by taking the cross product of the partial derivatives of \( \mathbf{r}_{1m} \) with respect to the surface parameters.
A direct application of the above method can lead to a discontinuity or sharp edge at the boundary between the modified involute flank and the unmodified fillet curve. To ensure a smooth transition (\( G^1 \) continuity), a Hermite interpolation technique is applied in this junction region.
A Hermite curve between two points \( P_0 \) and \( P_1 \) with specified tangent vectors \( T_0 \) and \( T_1 \) is defined by the cubic polynomial:
$$ \mathbf{r}(t) = (2t^3 – 3t^2 + 1)P_0 + (-2t^3 + 3t^2)P_1 + (t^3 – 2t^2 + t)T_0 + (t^3 – t^2)T_1 $$
where \( t \in [0, 1] \) is the curve parameter. The blending functions are:
$$ b_1(t)=2t^3-3t^2+1, \quad b_2(t)=-2t^3+3t^2, \quad b_3(t)=t^3-2t^2+t, \quad b_4(t)=t^3-t^2 $$
For the gear fillet problem, \( P_0 \) is the last point on the modified involute curve (near the root), and \( P_1 \) is a point on the original fillet curve. The tangent vectors \( T_0 \) and \( T_1 \) are set equal to the tangents of the respective curves at these points. To control the curvature of the interpolated segment, these tangent vectors are scaled by the product of the normal module \( m_n \) and a shape control parameter \( t_h \) (typically between 0.5 and 1.5). The coordinates of any point \( P \) on the Hermite curve are:
$$ \begin{aligned}
x_P(t) &= b_1 x_{P_0} + b_2 x_{P_1} + b_3 x_{T_0} + b_4 x_{T_1} \\
y_P(t) &= b_1 y_{P_0} + b_2 y_{P_1} + b_3 y_{T_0} + b_4 y_{T_1} \\
z_P(t) &= b_1 z_{P_0} + b_2 z_{P_1} + b_3 z_{T_0} + b_4 z_{T_1}
\end{aligned} $$
This creates a smooth bridge between the two surfaces, eliminating the discontinuity.
The proposed methodology is implemented for a specific helical gear pair with parabolic diagonal modification. The basic gear parameters and the modification parameters are listed in the tables below.
| Parameter | Pinion | Gear |
|---|---|---|
| Handedness | Right | Left |
| Number of Teeth | 30 | 72 |
| Normal Module (mm) | 5.000 | |
| Normal Pressure Angle (°) | 20.000 | |
| Helix Angle (°) | 33.273 | |
| Tip Radius (mm) | 94.706 | 220.294 |
| Root Radius (mm) | 83.456 | 209.044 |
| Face Width (mm) | 40.000 | |
| Parameter | Tip Relief | Root Relief |
|---|---|---|
| Modification Exponent \( k \) | 2 | 2 |
| Modification Height (mm) | 5.000 | 5.000 |
| Modification Length (mm) | 18.519 | 23.597 |
| Maximum Modification \( C \) (µm) | 30.000 | 30.000 |
| Start Line Helix Angle (°) | 15.109 | 11.964 |
Using MATLAB, the complete tooth surface coordinates for the diagonally modified helical gear are calculated, densely discretized along both the profile and lead directions. The Hermite interpolation parameter \( t_h \) is tuned to 0.9 to achieve a well-balanced, smooth transition curve that closely follows the original fillet geometry.
The computed point cloud data is imported into CATIA V5 R20 using the Digitized Shape Editor (DSE) module. The ‘Power Fit’ command in the Quick Surface Reconstruction (QSR) module is used to generate a high-quality, single tooth gap surface from the point cloud. This surface is then extended, trimmed, and closed to form a solid tooth. The solid tooth is patterned around the gear axis to complete the full three-dimensional model of the modified helical gear.
To validate the modeling accuracy, a deviation analysis is performed within CATIA. The ‘Deviation Analysis’ command compares the reconstructed modified flank surface against the theoretically calculated discrete points of the nominal standard flank. Deviations are measured along the surface normal direction. The analysis is conducted separately for profile sections (at different face-width locations \( z_1 \)) and lead sections (at different radii \( r \)).
For profile sections at \( z_1 = -20 \) mm and \( z_1 = 20 \) mm, the maximum deviation after subtracting the intentional 30 µm modification is 0.189 µm and 0.380 µm, respectively. For the central profile section at \( z_1 = 0 \) mm, which contains both modified and unmodified regions, the maximum fitting deviation in the unmodified region is found to be 1.103 µm. This occurs on the left flank and represents the overall maximum profile fitting error.
For lead sections at the modified boundaries (\( r = 85.666 \) mm and \( r = 94.706 \) mm), the maximum fitting deviations after subtracting the 30 µm modification are 0.380 µm and 0.189 µm, respectively. For the lead section at the pitch radius (\( r = 89.706 \) mm), the maximum fitting deviation in the unmodified region is 0.691 µm. Therefore, the maximum overall fitting deviation for the entire modified helical gear flank is approximately 1.1 µm. This extremely low error confirms the high precision of the reverse modeling methodology.
This study presents a comprehensive and precise workflow for the reverse modeling of diagonally modified helical gears. The key steps involve the rigorous derivation of the complete tooth surface geometry, the analytical definition and application of diagonal modification, the implementation of Hermite interpolation for smooth root transitions, and the final high-fidelity model reconstruction in CATIA based on digitally computed surface coordinates. The accuracy of the generated three-dimensional model, with a maximum fitting deviation of about 1.1 micrometers, meets the stringent requirements for advanced engineering analyses such as contact stress simulation, dynamic behavior prediction, and manufacturing process validation. This methodology provides a reliable digital foundation for the design and analysis of high-performance, low-noise helical gear transmissions.