Large straight spur gears, typically with diameters exceeding 0.8 m, are critical components in heavy machinery such as power generation equipment, mining machinery, shipbuilding, and rocket launch systems. The geometric accuracy of these gears directly affects the performance, reliability, and lifespan of the entire system. One of the primary challenges in the on-site measurement of large straight spur gears is the accurate determination of their geometrical center. Without a precise center reference, subsequent evaluations of tooth profile errors become unreliable. In this work, I present a novel method for calculating the geometrical center of large straight spur gears based on discrete data modeling of involute profiles. The method uses an iterative approximation approach, leveraging multiple tooth flanks to minimize the influence of measurement noise. I then compute the tooth profile total deviation using the estimated center and compare the results with those obtained using the theoretical center. The simulation results demonstrate that the proposed technique achieves sub‑micron accuracy in center determination and provides tooth profile error values within 10 μm of the theoretical values, even when the gear is subjected to random noise with amplitudes up to 318 μm. This method offers a practical and robust solution for the in‑situ metrology of large straight spur gears.
1. Introduction
The measurement of large straight spur gears has long been a difficult task due to their size, weight, and the need for on‑site evaluation. Traditional gear measuring centers, coordinate measuring machines (CMMs), and laser tracking systems have been employed, but they often require the gear to be precisely positioned relative to a known reference. For gears with diameters exceeding 6 m, even these advanced systems face limitations. A fundamental prerequisite for any accurate gear measurement is the determination of the gear’s geometrical center. Once the center is known, the actual tooth flank points can be compared with theoretical involute profiles to compute deviations such as the total tooth profile error. In this paper, I focus on straight spur gears as they are the most common type in large‑scale applications and serve as the foundation for studying other gear forms. I propose a computational method that uses the coordinates of multiple tooth flanks to iteratively refine the center estimate, followed by a rigorous derivation of the tooth profile total deviation. The method is validated through simulation with added noise, demonstrating its suitability for practical measurements.
2. Discrete Data Model of Involute Profile for Straight Spur Gears
To simulate the measurement process, I first establish a discrete coordinate model of an ideal involute tooth profile for a straight spur gear. The involute curve in Cartesian coordinates is given by:
$$
\begin{cases}
x_k = r_b (\cos \varphi_k + \varphi_k \sin \varphi_k) \\
y_k = r_b (\sin \varphi_k – \varphi_k \cos \varphi_k)
\end{cases}
\tag{1}
$$
where \(r_b\) is the base radius, and \(\varphi_k\) is the roll angle at the point on the involute. The range of \(\varphi_k\) is bounded by the starting and ending points of the active involute:
$$
\varphi_f = \tan \alpha_f = \tan\left(\arccos\frac{r_b}{r_f}\right), \quad
\varphi_a = \tan \alpha_a = \tan\left(\arccos\frac{r_b}{r_a}\right)
\tag{2}
$$
Here \(r_f\) and \(r_a\) are the root radius and addendum radius, respectively, and \(\alpha_f\) and \(\alpha_a\) are the corresponding pressure angles. For a tooth centered on the x‑axis, the left and right flanks are rotated by angles \(\theta_l\) and \(\theta_r\):
$$
\theta_r = -(\tan\alpha – \alpha + \pi/(2z)), \quad
\theta_l = \tan\alpha – \alpha + \pi/(2z)
\tag{3}
$$
For the i‑th tooth (i = 0, 1, …, z‑1), an additional rotation \(\omega_i = i \cdot 2\pi/z\) is applied. Thus, the coordinates of the k‑th point on the j‑th flank of the i‑th tooth become:
$$
\begin{bmatrix}
x_{ijk} \\ y_{ijk}
\end{bmatrix}
=
\begin{bmatrix}
\cos\omega_i & -\sin\omega_i \\
\sin\omega_i & \cos\omega_i
\end{bmatrix}
\begin{bmatrix}
\cos\theta_j & -\sin\theta_j \\
\sin\theta_j & \cos\theta_j
\end{bmatrix}
\begin{bmatrix}
r_b(\cos\varphi_k + \varphi_k\sin\varphi_k) \\
r_b(\sin\varphi_k – \varphi_k\cos\varphi_k)
\end{bmatrix}
\tag{4}
$$
Equation (4) generates a complete set of ideal tooth flank points for a straight spur gear. The table below lists the assumed parameters used in this study.
| Parameter | Symbol | Value |
|---|---|---|
| Module | m | 40 mm |
| Number of teeth | z | 100 |
| Pressure angle | α | 20° |
| Helix angle | β | 0° |
| Base radius | rb | 1879.39 mm |
| Addendum radius | ra | 2000 mm |
| Root radius | rf | 1750 mm |
3. Geometrical Center Calculation Model for Large Straight Spur Gears
To determine the geometrical center of a large straight spur gear, I select two pairs of opposing teeth (or teeth and opposing tooth spaces when the tooth count is odd). Let the measured coordinates of the left and right flanks on tooth A be \((x_{Alk}, y_{Alk})\) and \((x_{Ark}, y_{Ark})\), with similar notation for teeth A′, B, and B′. An initial center estimate \(C_0(x_0, y_0)\) is obtained by averaging all flank points:
$$
\begin{aligned}
x_0 &= \frac{1}{8n}\sum_{k=1}^{n}(x_{Alk}+x_{Ark}+x_{A’lk}+x_{A’rk}+x_{Blk}+x_{Brk}+x_{B’lk}+x_{B’rk}) \\
y_0 &= \frac{1}{8n}\sum_{k=1}^{n}(y_{Alk}+y_{Ark}+y_{A’lk}+y_{A’rk}+y_{Blk}+y_{Brk}+y_{B’lk}+y_{B’rk})
\end{aligned}
\tag{5}
$$
Around \(C_0\) I generate four additional candidate centers at step size \(p\): \(C_1(x_0+p, y_0+p)\), \(C_2(x_0-p, y_0+p)\), \(C_3(x_0-p, y_0-p)\), \(C_4(x_0+p, y_0-p)\). For each candidate center \(C_l\) (l = 0,…,4), I compute the theoretical involute profiles of the four selected flanks. Then, for each flank, I intersect the measured points and the theoretical profile with \(N\) concentric circles centered at the origin O (the coordinate system of the measured points). For a given intersection point on the measured flank \(Q_i(x_i, y_i)\) and the corresponding theoretical point \(Q’_i(x’_i, y’_i)\), the squared distance is:
$$
w_i = (x_i – x’_i)^2 + (y_i – y’_i)^2
\tag{6}
$$
Summing over all \(N\) intersections on all eight flanks, the total squared error for candidate \(C_l\) is:
$$
\Psi_l = \sum_{k=1}^{8}\sum_{i=1}^{N}\left[(x_{k,i} – x’_{k,i})^2 + (y_{k,i} – y’_{k,i})^2\right]
\tag{7}
$$
I choose the candidate with the smallest \(\Psi_l\) as the new center. The step size \(p\) is then reduced to one‑tenth of its previous value, and the process is repeated iteratively until the center coordinates converge to within a desired precision (e.g., 0.1 μm). The final \(C_l\) is taken as the geometrical center of the straight spur gear.
4. Calculation of Tooth Profile Total Deviation
Once the geometrical center \((x_0, y_0)\) is known, the total tooth profile deviation for each flank can be computed following the standard defined in ISO 1328‑1:2013. The deviation is the distance between two design involutes that just enclose the actual profile within the evaluation range. For a measured point \(K(x_K, y_K)\) on the flank, I first determine its pressure angle \(\alpha_k\) and roll angle \(\varphi_k\) with respect to the found center:
$$
\alpha_k = \arctan\left(\frac{\sqrt{(x_K – x_0)^2 + (y_K – y_0)^2 – r_b^2}}{r_b}\right)
\tag{8}
$$
$$
\varphi_k = \frac{\sqrt{(x_K – x_0)^2 + (y_K – y_0)^2 – r_b^2}}{r_b}
\tag{9}
$$
The rotation angle \(\beta\) from the theoretical involute start point on the base circle to the point corresponding to \(K\) is:
$$
\beta = \varphi_k – \alpha_k
\tag{10}
$$
Using \(\beta\), the coordinates of the theoretical start point \(P\) on the base circle are obtained via a rotation of the point \(P’\) (the foot of the perpendicular from the center to the line through \(K\)). The set of all start points \(\{P_i\}\) for the measured flank points defines a cloud of positions on the base circle. Let \(\theta_i\) be the angle between the line from the center to \(P_i\) and a reference direction (e.g., the positive x‑axis). The total tooth profile deviation \(\Delta f\) is then:
$$
\Delta f = r_b \cdot (\theta_{\max} – \theta_{\min})
\tag{11}
$$
where \(\theta_{\max}\) and \(\theta_{\min}\) are the maximum and minimum angles among all \(\theta_i\). This calculation is performed for each flank separately.
5. Simulation and Results
To validate the method, I generated ideal tooth flank points using the gear parameters from Table 1. I then added random noise with uniform distribution in the range \([-\delta/2, \delta/2]\) to each point, with \(\delta\) taking values 0, 10 μm, 50 μm, 100 μm, 150 μm, 200 μm, 250 μm, 300 μm, and 318 μm. For each noise level, I applied the proposed center‑finding algorithm and computed the tooth profile total deviation \(F’_\alpha\) using the estimated center. For comparison, I also computed the deviation \(F_\alpha\) using the true theoretical center \((0,0)\). The results are summarized in Table 2 and Table 3.
| Noise amplitude δ (μm) | Estimated xl (μm) | Estimated yl (μm) |
|---|---|---|
| 0 | –3.832 × 10–2 | +3.205 × 10–2 |
| 10 | –4.564 × 10–2 | +5.861 × 10–2 |
| 50 | –3.017 × 10–2 | +7.407 × 10–2 |
| 100 | –2.894 × 10–2 | +7.531 × 10–2 |
| 150 | –3.248 × 10–2 | +7.177 × 10–2 |
| 200 | –5.645 × 10–2 | +4.779 × 10–2 |
| 250 | –5.783 × 10–2 | +4.642 × 10–2 |
| 300 | –6.488 × 10–2 | +3.936 × 10–2 |
| 318 | –5.199 × 10–2 | +5.225 × 10–2 |
| Noise δ (μm) | Fα (true center) | F’α (estimated center) | Difference |Fα – F’α| |
|---|---|---|---|
| 0 | 0 | 0 | 0 |
| 10 | 10 | 10 | 0 |
| 50 | 50 | 59.9 | 9.9 |
| 100 | 99.8 | 102.2 | 2.4 |
| 150 | 149.5 | 159.1 | 9.6 |
| 200 | 199.4 | 208.6 | 9.2 |
| 250 | 249.6 | 258.6 | 9.0 |
| 300 | 299.5 | 306.8 | 7.3 |
| 318 | 317.2 | 318.5 | 1.3 |
The results show that the estimated geometrical center consistently deviates by less than 0.1 μm from the true center, even when noise as high as 318 μm is added. The tooth profile total deviation computed using the estimated center differs from the true deviation by no more than 10 μm. In many cases, the difference is below 5 μm. This demonstrates that the proposed iterative center‑finding method is highly robust to measurement noise and provides a reliable foundation for tooth profile error evaluation in large straight spur gears.
6. Discussion
The method relies on the fact that the involute profiles of a straight spur gear are uniquely determined by the base radius, which is known from the gear design parameters. By minimizing the distance between measured points and theoretical profiles over a set of candidate centers, the algorithm effectively filters out random noise. The choice of two opposing pairs of teeth ensures that the error is balanced and reduces the influence of localized form deviations. The step‑size reduction strategy guarantees convergence to the true center within sub‑micron precision. The maximum error of 10 μm in the tooth profile deviation is acceptable for many industrial applications, especially for large gears where the tolerance grades are often wider. The method can be extended to helical gears or other tooth forms by modifying the theoretical profile equations accordingly.
7. Conclusion
In this work, I have presented a systematic approach for determining the geometrical center of large straight spur gears from discrete measured data and for calculating the tooth profile total deviation based on that center. The method combines an averaging initial guess with iterative refinement using least‑squares fitting to theoretical involutes. Simulation results confirm that the center can be determined to within 1 μm and the tooth profile error to within 10 μm for noise levels up to 318 μm. This technique offers a practical, noise‑resistant solution for the on‑site metrology of large straight spur gears, enabling accurate quality assessment without the need for expensive and bulky reference fixtures. Future work will focus on experimental validation using a real large gear and integration with portable coordinate measurement systems.