Spiral bevel gears are widely used for intersecting-shaft power transmission in aero-engines, vehicle axles, marine drives, and industrial machinery. Their curved teeth provide smooth conjugate motion and high torque density, but the loaded contact pattern remains a sensitive design variable. A footprint that looks acceptable under light load can migrate toward the heel or toe, or toward the tooth top or root, as torque rises. Once the footprint reaches an edge, the effective contact area collapses, contact pressure rises sharply, and scoring or tooth-root fatigue becomes more likely. The engineering problem is therefore not only to place a contact ellipse at one operating instant, but to control how much room the pattern has to expand before it reaches any tooth-surface boundary.
Why a Static Contact Pattern Is Not Enough
Conventional tooth contact analysis (TCA) predicts the contact path, transmission error, and the no-load or lightly loaded footprint. Loaded tooth contact analysis (LTCA) extends this to tooth deformation and pressure. These methods are essential, but a single footprint does not describe the time-varying expansion of the contact region. Under heavy load, the pattern grows and shifts; bending of the teeth and housing, mounting misalignment, and manufacturing scatter all influence the direction of that shift. In practice, a static pattern is often placed slightly toward the toe because load-induced deformation tends to move the footprint toward the heel. That rule of thumb is useful, but it is not a quantitative control target for the full expansion process.
Contact pattern expansion (CPE) treats the pattern as a region that must retain sufficient margin on four sides: the tooth-top boundary and the root boundary on the pinion and gear, and the heel and toe boundaries along the face width. The objective is to equalize the available margin so that the pattern can grow uniformly, rather than reaching one edge early while unused area remains on the opposite side. This is particularly important in lightweight, high-power-density designs, where the tooth-surface margins are small and the penalty for edge contact is severe.
Boundary-Based CPE Metrics
The CPE calculation starts from the same geometric information used in TCA. At each meshing instant, the contact ellipse is obtained as the intersection of two offset surfaces representing the two tooth flanks under load. Superimposing the ellipses over the mesh cycle produces the loaded contact pattern. The boundary points of each ellipse are then projected onto the axial section of the gear blank, where distances to the tooth-top, root, heel, and toe boundaries can be measured.
Four metrics are used: the tooth-top margins \(\delta_{ag}\) and \(\delta_{ap}\) for the two members, and the heel and toe margins \(s_h\) and \(s_t\). Because the tooth surface is curved and distorted, the midpoint of the blank does not coincide with the midpoint of the actual flank. The heel and toe margins are therefore measured along the flank line by arc length, not by a simple straight-line distance from the blank center. A centered expansion reserve is defined by two equalities:
\[
\delta_{ag}=\delta_{ap},\qquad s_h=s_t .
\]
The first equality gives the pinion and gear footprints the same remaining margin to the tooth-top boundary. The second gives the gear footprint equal clearance to the heel and toe. These conditions do not force the pattern to the geometric center of the blank; they equalize the distance available for growth, which is the quantity that governs whether edge contact occurs. The gear-side footprint is used as the control reference because the two members do not have identical contact footprints under load. The corresponding pinion margin is still evaluated, but the control target is expressed on the gear side to avoid overconstraining the two non-identical contact regions.
Regulating the Flank Before Solving for Machine Settings
Directly adjusting every machine setting to move a contact pattern is inefficient because many settings influence the pattern in coupled ways. The CPE model instead introduces a small normal deflection field on the pinion flank and uses that field as an intermediate design variable. Two scalar controls define the field. A tooth-height control \(\delta_h\) is applied at the top heel and top toe and set to zero at the flank midpoint. A face-width control \(\delta_l\) is applied at the heel top and heel root and set to zero at the midpoint. Three constraints determine a plane in the flank coordinates, so the deflection field is smooth and can be superimposed on the existing tooth surface without changing the basic gear geometry.
Adding this field changes the normal gap between the mating flanks, which moves the contact point and changes the four CPE metrics. The inverse problem is then to find machine settings that reproduce the desired deflected flank after grinding. The tooth surface is discretized into a grid of points, and the normal deviation between the target flank and the flank generated by the machine settings is minimized with the Levenberg–Marquardt method. In the aerospace case described below, the radial cutter position and the cutter pressure angle were selected as the two sensitive settings for this inverse solution. This reduces the computational cost while preserving the dominant effect on the CPE metrics.
Case Study: An Aerospace Face-Milled Pair
The procedure was demonstrated on a face-milled spiral bevel gear pair intended for aero-engine service. The pinion has 12 teeth and the gear has 49 teeth; the blank geometry is summarized in Table 1. The pair is right-hand driven on the pinion and left-hand on the gear, with a relatively narrow face width and a large outer cone distance, which is typical of aircraft bevel-gear proportions.
Table 1. Blank geometry of the aerospace spiral bevel gear pair.
| Parameter | Pinion | Gear |
|---|---|---|
| Number of teeth | 12 | 49 |
| Hand | Right | Left |
| Face width (mm) | 41.05 | 37 |
| Outer cone distance (mm) | 109.12 | 114.20 |
| Pitch cone angle (°) | 17.5833 | 71.8167 |
| Face cone angle (°) | 21.3667 | 72.6167 |
| Root cone angle (°) | 16.8167 | 67.9333 |
| Addendum (mm) | 6.98 | 1.44 |
| Dedendum (mm) | 2.72 | 8.13 |
The initial TCA gave \(\delta_{ag}=0.5\) mm, \(\delta_{ap}=1.6\) mm, \(s_h=11.5\) mm, and \(s_t=9.3\) mm, with a peak transmission error of \(66\ \mu\text{rad}\). The pinion footprint sat near the tooth top, while the gear footprint sat near the root. The heel margin was larger than the toe margin, so the expansion reserve was not balanced. Applying a tooth-height deflection of \(\delta_h=-0.0015\) mm and a face-width deflection of \(\delta_l=0.0025\) mm, followed by the machine-setting inverse solution, produced \(\delta_{ag}=\delta_{ap}=1.1\) mm and \(s_h=s_t=10.7\) mm. The peak transmission error remained \(66\ \mu\text{rad}\), so the correction improved the pattern position without materially changing the conjugate motion.
Table 2. CPE metrics before and after regulation.
| Metric | Before | After |
|---|---|---|
| Tooth-top margin \(\delta_{ag}\) (mm) | 0.5 | 1.1 |
| Tooth-top margin \(\delta_{ap}\) (mm) | 1.6 | 1.1 |
| Heel margin \(s_h\) (mm) | 11.5 | 10.7 |
| Toe margin \(s_t\) (mm) | 9.3 | 10.7 |
| Peak transmission error (\(\mu\text{rad}\)) | 66 | 66 |
The equalized margins place the loaded footprint near the middle of the working region rather than at an edge. The change is small in absolute terms—tenths of a millimeter to a few millimeters—but the CPE metrics are precisely the quantities that determine how much load-induced growth can occur before boundary contact. This is the main reason the method is useful for high-performance gears: it converts an experienced designer’s visual judgment into a measurable boundary condition.
Loaded Contact Pressure After Regulation
The corrected flank was evaluated with a finite-element loaded contact model in Abaqus under the analyzed LTCA load case. Before regulation, the pinion peak pressure was 1978.3 MPa near the tooth root and the gear peak was 1773.9 MPa near the tooth top. These concentrations are consistent with a footprint that is not centered in the expansion sense: one side of the tooth carries the load peak while the opposite side retains unused margin. After regulation, the pinion peak fell to 1351.1 MPa and the gear peak to 1500.5 MPa, both in the central region of the flank. The pressure at the root and tooth-top boundaries fell to 0 MPa.
Table 3. Peak contact pressure before and after CPE regulation.
| Quantity | Before | After |
|---|---|---|
| Peak contact pressure, pinion (MPa) | 1978.3 | 1351.1 |
| Peak contact pressure, gear (MPa) | 1773.9 | 1500.5 |
| Pinion peak reduction | — | 31.7% |
| Gear peak reduction | — | 15.4% |
The pressure distribution along the face width also changed from a non-monotonic profile with a local rise near the heel to a smoother increase and decrease from toe to heel. Along the tooth height, the concentrations near the pinion root and gear top were largely removed. These results show that the CPE correction does more than move a visual footprint: it redistributes the loaded contact and reduces the peak pressure that drives pitting and bending fatigue.
Experimental Contact Areas Under Light and Heavy Load
The adjusted machine settings were transferred to the CNC system of a Phoenix 600G grinder, and the pair was ground for contact-area inspection. The contact areas were examined at 10 N·m and 1500 N·m to represent light and heavy operating conditions. At 10 N·m, the unregulated pattern was centered along the face width but biased toward the tooth top. After regulation, the pattern shifted slightly toward the toe but became more uniform in the tooth-height direction. This is consistent with a static pattern that has been positioned to leave room for load-induced growth toward the heel.
At 1500 N·m, the difference was more pronounced. Before regulation, the contact area shifted toward the heel and crossed the tooth boundary; roughly one quarter of the flank did not participate in contact. After regulation, the pattern sat near the middle of the flank, the distances to the heel and toe were approximately equal, and the contact area expanded to nearly the entire tooth surface. The experimental observations therefore support the predicted CPE behavior: equalizing the boundary margins allows the pattern to use more of the available flank under heavy load instead of losing area at an edge.
What the Method Establishes, and What It Does Not
The case demonstrates a practical modeling route for CPE control in spiral bevel gears. A small, parameterized flank deflection is used to reshape the contact pattern; a nonlinear inverse solution then converts that deflection into machine settings that can be used on a grinding machine. The result is a pattern that is centered in terms of expansion reserve, with lower peak pressure and a larger effective contact area at high load. In principle, a larger contact area provides room to reduce gear size while maintaining fatigue life, which is attractive for lightweight transmission systems. That weight-saving outcome is a design implication, not a demonstrated hardware result.
The scope of the evidence should also be kept in view. The validation covers one aerospace face-milled gear set, one grinding process, and the selected LTCA load case. The CPE target assumes that expansion is approximately uniform; actual service conditions can introduce misalignment, thermal growth, and manufacturing scatter that alter the pattern. The control target is expressed on the gear-side footprint, and the inverse problem was reduced to two sensitive machine settings. These choices make the method efficient, but they also mean that the result depends on the selected parameters and on the quality of the underlying TCA/LTCA model. Within those limits, the boundary-based CPE approach gives engineers a quantitative way to keep a spiral bevel gear pattern away from its edges as load increases, rather than relying only on a single static footprint.