In recent years, the rapid development of the automobile industry has led to a constantly increasing demand for gears. Traditional cutting processes for gear manufacturing suffer from low material utilization, low production efficiency, and severe environmental impact. To address these challenges, precision cold forging has emerged as a promising net-shape forming technology. It offers higher material utilization, improved mechanical properties, better surface quality, and higher productivity. Straight bevel gears are widely used in automotive differentials, and their precision cold forging has become a key research focus. However, the extremely high forming loads in cold forging cause significant elastic deformation of the die cavity, and the forged gear experiences elastic springback after ejection. Both phenomena critically affect the dimensional accuracy of the final product.
I have conducted a comprehensive study on the precision cold forging of straight bevel gears using three-dimensional elastic-plastic finite element simulation. The purpose is to improve the accuracy of cold-forged straight bevel gears by compensating for die elastic deformation and workpiece springback. This article summarizes my research work, including gear modeling, process optimization, elastic deformation laws, springback behavior, and iterative die compensation. The goal is to provide theoretical guidance and practical reference for the precision forging of straight bevel gears.
1. Introduction and Research Background
Cold precision forging is a near-net-shape forming process in which metal billet is deformed at room temperature to obtain a finished or near-finished gear shape. Compared with traditional machining, cold forging of straight bevel gears saves material, reduces energy consumption, and improves production efficiency. Moreover, the continuous metal fiber lines in forged gears lead to higher strength and longer fatigue life. The automotive industry increasingly adopts this technology for differential gears, such as planetary gears and side gears.
Nevertheless, the precision control of cold-forged straight bevel gears remains a challenging problem. During cold forging, the forming load can reach several hundred kilonewtons, causing the die cavity to deform elastically. After the load is removed, the die recovers its original shape, while the forged gear undergoes elastic recovery. Both elastic phenomena alter the final tooth flank geometry. The deviation from the ideal gear profile may exceed the allowable tolerance for high-precision gears.
Therefore, my research focuses on quantifying the die cavity elastic deformation and the gear springback, and then applying a reverse compensation method to modify the die cavity geometry. By iteratively correcting the die shape, the accuracy of the forged straight bevel gears can be gradually improved. In this article, I present a systematic investigation using elastic-plastic FEM, combined with B-spline surface modeling, iterative die modification, and regression analysis for the correction coefficient.
2. Elastic-Plastic Finite Element Fundamentals
The finite element method (FEM) is a powerful numerical tool for analyzing metal forming processes. For cold forging of straight bevel gears, the workpiece experiences large plastic deformation, and the elastic recovery must be considered. The elastic-plastic finite element method is particularly suitable because it accounts for both elastic and plastic deformation, and it can accurately simulate the loading and unloading stages.
2.1 Basic Equations
For an elastic-plastic material, the total strain increment is decomposed into elastic and plastic parts:
$$
d\varepsilon_{ij} = d\varepsilon_{ij}^{e} + d\varepsilon_{ij}^{p}
$$
The stress increment is related to the elastic strain increment by Hooke’s law:
$$
d\sigma_{ij} = D_{ijkl}^{e} \left( d\varepsilon_{kl} – d\varepsilon_{kl}^{p} \right)
$$
After introducing the yield criterion and the plastic flow rule, the increment form of the elastic-plastic constitutive equation is:
$$
d\sigma_{ij} = D_{ijkl}^{ep} d\varepsilon_{kl}
$$
where \(D_{ijkl}^{ep}\) is the elastic-plastic tangent stiffness tensor, which depends on the stress state and the deformation history.
The equilibrium equation in the incremental form is solved using the virtual work principle:
$$
\int_{V} \delta \varepsilon^{T} D^{ep} \Delta \varepsilon dV = \int_{V} \delta u^{T} \Delta b dV + \int_{S} \delta u^{T} \Delta p dS – \int_{V} \delta \varepsilon^{T} \sigma dV
$$
Here, \(\Delta u\), \(\Delta \varepsilon\), and \(\Delta \sigma\) are the incremental displacement, strain, and stress fields, respectively.
2.2 FE Simulation Software
I used the commercial FEM software DEFORM-3D, which is specifically designed for metal forming simulations. The software consists of a pre-processor, a solver, and a post-processor. The pre-processor allows geometry import, mesh generation, material assignment, contact definition, and simulation parameter setup. The solver performs the incremental elastic-plastic analysis. The post-processor provides results such as stress distribution, strain distribution, displacement fields, and load-stroke curves.
In my simulations, the workpiece was meshed with approximately 50,000 tetrahedral elements, with a minimum element size of 0.3 mm. The step length was set to 0.1 mm to ensure accuracy and convergence. The material of the workpiece was 20CrMnTi, which is commonly used for automotive gears. Since this material was not available in the DEFORM material library, I used 20MnCr5, which has very similar mechanical properties.
3. Three-Dimensional Modeling of Straight Bevel Gears
To build a realistic FE model, I first created a precise three-dimensional geometric model of the straight bevel gears. The gear geometry was constructed using the B-spline surface method, which can accurately represent the complex tooth flank shape.
3.1 Design of the Forging Preform
Based on the gear specifications, I designed the forged gear model with appropriate machining allowances, draft angles, and corner radii. The main parameters of the planetary gear and side gear are summarized in the following tables.
| Parameter | Value |
|---|---|
| Module | 5 mm |
| Number of teeth | 10 |
| Pressure angle | 22°30′ |
| Shaft angle | 90° |
| Full tooth height | 8.94 mm |
| Pitch diameter | 65 mm |
| Pitch cone angle | 37°34′ |
| Root cone angle | 31°44′ |
| Effective tooth height | 8 mm |
| Parameter | Value |
|---|---|
| Module | 5 mm |
| Number of teeth | 10 |
| Pressure angle | 22°30′ |
| Shaft angle | 90° |
| Full tooth height | 8.94 mm |
| Pitch diameter | 65 mm |
| Pitch cone angle | 52°26′ |
| Root cone angle | 44°32′ |
| Effective tooth height | 8 mm |
For the planetary gear, the forging was accomplished in a single step without preforming. For the side gear, a preform was designed to reduce die wear and forming load. The preform geometry was a cylinder with a conical top, as shown in the model section.
3.2 B-Spline Tooth Surface Construction
I generated a set of \(m \times n\) control points on the tooth flank. For each tooth, a B-spline surface was fitted through these points to obtain a smooth and accurate tooth surface. The gear model was then created in the UG software. This virtual machining approach simulates the actual gear generation process and minimizes geometric errors.
The coordinate system used for the tooth flank is defined as follows:
- X-axis: radial direction from gear axis to tooth tip
- Z-axis: axial direction from small end to large end
- Y-axis: tangential direction perpendicular to X and Z
4. Die Design for Cold Forging of Straight Bevel Gears
The die structure directly affects the forming quality and service life. For the high forming loads encountered in cold forging of straight bevel gears, I designed a two-layered combined die with an optimal interference fit. The overall die assembly consists of the upper punch, lower punch, tooth cavity die, and back-cone cavity die.
4.1 Combined Die Optimization
The total diameter ratio of the die was selected in the range \(a = 4 \sim 6\), which is the reasonable range for cold extrusion dies. For a two-layer combined die, the allowable specific pressure is 1100–1400 MPa. Through optimization, I obtained an interference of 0.45 mm and an axial pressing amount of 10 mm. The prestress ring was made of SKD11, which has good toughness and can effectively absorb forming stresses.
| Parameter | Value |
|---|---|
| Die type | Two-layer combined die |
| Interference | 0.45 mm |
| Axial pressing amount | 10 mm |
| Prestress ring material | SKD11 |
| Cavity material | High-strength alloy steel |
4.2 Overall Die Assembly
The die assembly is designed for a double-action press. During forming, the upper and lower dies close first to form a closed cavity. Then the punches move to deform the billet. After forming, the upper die opens, and the lower ejector pushes the forged gear out of the die. The detailed assembly includes a backup plate to provide sufficient rigidity.
5. Optimization of the Cold Forging Process
Cold forging of straight bevel gears can be performed in either open-die or closed-die mode. Open-die forging reduces the forming load but results in lower accuracy and material utilization. Closed-die forging provides better accuracy and material utilization but requires higher loads and reduces die life. I proposed a hybrid process combining the advantages of both methods.
5.1 Three Process Schemes
To compare different forming strategies, I defined three schemes:
- Scheme 1: Fully closed-die forging.
- Scheme 2: Open-die forging with a flash gutter.
- Scheme 3: Composite process: open-die preforming first, then closed-die finishing.
All schemes were simulated with the same material, mesh, friction coefficient, and velocity. The friction coefficient was set to 0.12, the punch velocity to 30 mm/s, and the initial temperature to 20°C.
5.2 Simulation Results
The forming load curves for the three schemes are shown in the following table, which summarizes the maximum loads and filling characteristics.
| Scheme | Maximum load (N) | Filling quality | Springback behavior |
|---|---|---|---|
| 1 (closed die) | 5.94 × 10⁵ | Excellent filling | Large springback |
| 2 (open die) | 4.46 × 10⁵ | Incomplete filling at large end | Moderate springback |
| 3 (composite) | 5.46 × 10⁵ | Nearly identical to closed die | Close to scheme 2 |
Scheme 3 reduced the maximum load by about 10% compared with the fully closed-die scheme, while maintaining the same filling quality. The load reduction helps increase die life and also reduces the elastic deformation of the die. Therefore, I selected scheme 3 as the optimal process for the subsequent accuracy study.
For the side gear, a preforming step was introduced. Although the final forming load remained the same with and without preforming, the preforming reduced die wear and improved die service life. The springback remained the same because it was proportional to the final forming load.
6. Elastic Deformation of the Die Cavity
During cold forging, the die cavity is subjected to high internal pressure from the workpiece. Although the die is made of high-strength steel, it experiences elastic deformation. This deformation is directly transferred to the forged gear, causing dimensional errors. To study this effect, I treated the die as an elastic body in the FE model, while the workpiece was defined as elastic-plastic. The combined simulation allows accurate prediction of the final gear shape, including both die elastic deflection and workpiece springback.
6.1 Analysis Model
To quantify the deformation of the die cavity, I selected 9 tooth section curves from the small end to the large end of the tooth cavity. On each curve, 5 points were chosen from the tooth tip to the root. The points were tracked during the simulation to obtain their displacement vectors.
6.2 Deformation Patterns
The results show that the die cavity deformation is not uniform. On a given tooth section curve, the maximum deformation occurs at the tooth tip. The deformation decreases gradually toward the tooth root, reaching a minimum at the tooth middle (the widest part of the tooth), and then increases again near the root.
From the small end to the large end of the die cavity, the deformation decreases continuously. The largest deformation appears at the small end, and the smallest at the large end. This pattern is explained by the pressure distribution during forging: the small end is formed first and receives higher pressure, while the large end is formed last and is closer to the free surface, which reduces pressure.
The elastic deformation values for the die cavity can be expressed as a function of the position along the tooth flank. Let \(\delta_d(s)\) represent the deformation at a normalized position \(s\) along the tooth curve, \(0 \le s \le 1\) from tip to root. The deformation follows approximately:
$$
\delta_d(s) = \delta_{tip} – (\delta_{tip} – \delta_{mid}) \cdot \left( \frac{s – s_{mid}}{1 – s_{mid}} \right)^2
$$
for \(s \ge s_{mid}\), where \(\delta_{tip}\) is the deformation at the tip, \(\delta_{mid}\) is the minimum at the middle, and \(s_{mid}\) is the position of the minimum.
7. Elastic Springback of Forged Straight Bevel Gears
After the die opens and the forging is ejected, the internal residual stresses in the workpiece are released, causing elastic springback. The springback magnitude is related to the residual stress distribution, which depends on the forming load and metal flow.
7.1 Springback Measurement
To measure springback accurately, I exported the simulated gear model and compared it with the ideal gear geometry. The deviation values at each node were calculated using a relative coordinate system. The same 9×5 point grid used for the die analysis was applied to the forged gear. I used HyperMesh to extract the coordinate data and MATLAB to compute deviations.
7.2 Springback Laws
The springback behavior of the forged straight bevel gears exhibits the following characteristics:
- On a given tooth curve, the tooth tip has the largest springback. The springback decreases toward the tooth root, reaches a minimum at the middle of the tooth flank (the widest point), and then increases slightly near the root.
- From the small end to the large end, the springback decreases gradually. The small end has the maximum springback, while the large end has the minimum.
These observations are consistent with the residual stress distribution. The small end of the gear is pressed harder and retains higher residual stresses after unloading. The tooth middle is supported by the surrounding material and thus has lower residual stress.
The relationship between springback \( \delta_s \) and the residual stress \(\sigma_{res}\) is approximately linear:
$$
\delta_s = C \cdot \sigma_{res}
$$
where \(C\) is a constant related to the elastic modulus and the geometry of the gear.
8. Accuracy Inspection and Tooth Surface Contact Analysis
For straight bevel gears, not only the dimensional accuracy but also the tooth surface contact pattern is crucial. A gear with accurate size but poor contact may generate excessive noise and fail prematurely. I therefore performed both geometric accuracy inspection and tooth contact analysis (TCA) after each die correction step.
8.1 Measurement Method
The simulated gear surface was digitized by extracting the coordinates of the 5×9 node grid. The deviations were computed with respect to the theoretical B-spline surface. These deviations were used to evaluate whether the gear met the ISO 7-grade accuracy requirement.
8.2 Contact Analysis
To ensure the corrected gear still meets the meshing requirements, I performed a finite element contact analysis using the gear pair model. The conditions for conjugate contact are:
- The normal vectors of the contacting surfaces are equal.
- The radial vectors are equal.
- The relative sliding velocity normal to the contact surface is zero.
These conditions are expressed mathematically as:
$$
\mathbf{n}_1 \cdot \mathbf{V}_{12} = 0
$$
where \(\mathbf{n}_1\) is the normal vector of the driving gear tooth surface and \(\mathbf{V}_{12}\) is the relative velocity at the contact point.
After the second iteration of die correction, the contact pattern remained within the acceptable region, and the transmission error was below the allowable limit.
9. Die Cavity Correction by Iterative Reverse Compensation
To compensate for the combined effects of die elastic deformation and workpiece springback, I adopted a reverse compensation method. The principle is to modify the die cavity in the opposite direction of the measured deviation. Each node on the die cavity is displaced by a certain compensation amount.
9.1 Iterative Procedure
The iterative procedure consists of the following steps:
- Perform the initial finite element simulation with a nominal die cavity.
- Extract the deviation values \(\Delta_i\) at each node of the forged gear tooth flank.
- Compensate the die cavity by \(\Delta_i\) in the reverse direction: \(x_{die}^{new} = x_{die}^{old} + \Delta_i\), where \(x\) is the coordinate.
- Fit a smooth B-spline surface through the compensated points.
- Build a new die model and repeat the simulation.
- Evaluate the new deviations. If the maximum absolute deviation is within the required tolerance, stop. Otherwise, repeat from step 2.
I used single-rate compensation (i.e., the first correction amount was equal to the deviation value). This ensures convergence. After the first iteration, the deviations became negative, meaning the gear was undersized. The second iteration then reduced the deviations to within ±0.025 mm, satisfying the ISO 7-grade requirement.
9.2 Deviation Results
Table 5 lists the deviation values at selected nodes before correction, after the first correction, and after the second correction. The deviations are in mm.
| Node | X (mm) | Z (mm) | Δ1 (initial) | Δ2 (after 1st) | Δ3 (after 2nd) |
|---|---|---|---|---|---|
| 1 | 17.947 | -18.392 | 0.1209 | -0.0606 | -0.0267 |
| 2 | 19.343 | -19.764 | 0.1344 | -0.0741 | -0.0211 |
| 3 | 20.738 | -21.136 | 0.1436 | -0.0851 | -0.0169 |
| 4 | 22.132 | -22.509 | 0.1486 | -0.0937 | -0.0140 |
| 5 | 23.525 | -23.883 | 0.1493 | -0.0998 | -0.0124 |
| 6 | 24.917 | -25.257 | 0.1457 | -0.1034 | -0.0121 |
| 7 | 26.308 | -26.631 | 0.1378 | -0.1046 | -0.0132 |
| 8 | 27.699 | -28.006 | 0.1256 | -0.1033 | -0.0156 |
| 9 | 29.089 | -29.382 | 0.1092 | -0.0996 | -0.0193 |
After the second iteration, the maximum absolute deviation was about 0.027 mm, which is below the tolerance range. The gear was considered acceptable.
10. Best Correction Coefficient K and Its Regression Equations
In the die correction process, the optimal correction amount for each node is not equal to the initial deviation value. I defined the correction coefficient \(K\) as:
$$
K = \frac{\Delta_{correction}}{\Delta_{deviation}}
$$
where \(\Delta_{correction}\) is the best total compensation amount applied to the die, and \(\Delta_{deviation}\) is the initial deviation of the forged gear from the target geometry.
For a single tooth curve, the values of \(K\) show a linear trend along the X and Z axes. I therefore fitted linear equations for each of the eight tooth curves. The equations are presented in Table 6.
| Curve | K as a function of X | K as a function of Z |
|---|---|---|
| 1 | \(K_1 = 1.03724 – 0.08663X\) | \(K_1 = 2.45376 + 0.10527Z\) |
| 2 | \(K_2 = 0.69888 – 0.05941X\) | \(K_2 = 1.87875 + 0.07270Z\) |
| 3 | \(K_3 = 0.56200 – 0.04642X\) | \(K_3 = 1.59557 + 0.05652Z\) |
| 4 | \(K_4 = 0.64690 – 0.04580X\) | \(K_4 = 1.62188 + 0.05575Z\) |
| 5 | \(K_5 = 1.00237 – 0.05745X\) | \(K_5 = 2.01642 + 0.06980Z\) |
| 6 | \(K_6 = 0.93017 – 0.04976X\) | \(K_6 = 2.04297 + 0.06831Z\) |
| 7 | \(K_7 = 0.88070 – 0.04371X\) | \(K_7 = 1.78732 + 0.05685Z\) |
| 8 | \(K_8 = 0.78284 – 0.03624X\) | \(K_8 = 1.45635 + 0.04410Z\) |
Note that the Z coordinate values are negative in the model. The equations allow the calculation of the correction coefficient for any point along a given tooth curve. The ninth curve was not included because its deviations were very small and the linear fit had low confidence.
The general trend is that \(K\) decreases from the small end to the large end, and on each curve \(K\) decreases from the tooth tip to the tooth root. This reflects the fact that the combined elastic deformation and springback are larger at the small end and at the tooth tip.
With these regression equations, the die correction process can be accelerated: for a new simulation, one can directly compute the required compensation amount for any point using its coordinates, without performing multiple iterative corrections.
11. Conclusion and Outlook
In this research, I have systematically studied the precision cold forging of straight bevel gears using elastic-plastic finite element simulations. The main conclusions are as follows:
- The combined approach, where the die is treated as an elastic body and the workpiece as an elastic-plastic body, realistically captures the coupled effects of die deformation and gear springback. This is crucial for accurate prediction of the final gear geometry.
- The die cavity elastic deformation is non-uniform. It is largest at the tooth tip and at the small end of the gear, and smallest at the tooth middle and at the large end. The workpiece springback follows a similar pattern, with the largest springback at the tooth tip and at the small end.
- The composite process (open-die preforming followed by closed-die finishing) reduces the maximum forming load by about 10% compared with fully closed-die forging, while maintaining the same filling quality and precision. This process is therefore recommended for the cold forging of straight bevel gears.
- The iterative reverse compensation method effectively improves the accuracy of cold-forged straight bevel gears. After two iterations, the tooth profile deviations were reduced from about 0.15 mm to less than 0.03 mm, meeting the ISO 7-grade requirement.
- The correction coefficient \(K\) varies linearly with the coordinates on each tooth curve. The fitted regression equations provide a convenient way to determine the optimal die correction amount for any point, thus greatly reducing the die try-out time.
This study provides a theoretical basis and practical guidance for the production of high-precision cold-forged straight bevel gears. Future work may include the influence of temperature, the use of advanced meshless methods to avoid volume loss, and experimental validation of the simulation results. The combination of numerical simulation and physical experiments will further improve the reliability and applicability of the proposed method.
In summary, precision cold forging of straight bevel gears is a promising manufacturing technology. Through detailed numerical analysis of elastic deformation and springback, combined with rational die compensation, it is possible to produce straight bevel gears with high accuracy and consistency. The methods and findings presented in this article will contribute to the wider application of cold forging in the automotive gear industry.