Spiral bevel gears transmit motion between intersecting shafts in aerospace, vehicle, mining, and industrial drives. Their curved tooth trace provides gradual engagement, a controllable contact pattern, and high load capacity, but the pinion root remains a fatigue-critical region. Tooth breakage is governed by root bending stress, and a small difference in calculated stress can change the apparent safety margin of a design. ISO 10300-3 offers two calculation routes for tooth root strength: method B1 and method B2. Both use load factors, geometry factors, life factors, and a safety-factor check, yet they refer the tooth load to different sections of the bevel gear and use different correction coefficients. The engineering question is therefore not simply which formula is longer, but how much the two routes differ, which one is more severe, and how either compares with a three-dimensional contact solution.
Two ISO 10300 Routes to Root Stress
Method B1 is built on a virtual cylindrical gear at the mid-face of the bevel tooth. The nominal tangential force is converted to the virtual gear, and the root stress is assembled from a tooth-form factor, a stress-correction factor, a contact-ratio factor, a spiral-angle factor, and a load-sharing factor:
$$
\sigma_{F,\mathrm{B1}}=
\frac{F_{vmt}}{b_v m_{mn}}
Y_{F\alpha}Y_{S\alpha}Y_\varepsilon Y_{BS}Y_{LS}
K_A K_V K_{F\beta}K_{F\alpha},
$$
$$
F_{vmt}=\frac{F_{mt}}{\cos\beta_v\cos\beta_{m1}},
\qquad
F_{mt}=\frac{2000T}{d_m}.
$$
Here \(T\) is the nominal torque in N·m, \(d_m\) is the mid-face pitch diameter in mm, \(b_v\) is the virtual gear width, and \(m_{mn}\) is the mid-face normal module. The factor \(Y_{LS}\) represents load sharing among adjacent teeth. Because the reference section is at mid-face, B1 uses the geometry and load intensity of the central part of the tooth.
Method B2 refers the calculation to the outer end of the bevel gear and applies a Lewis-type cantilever beam with a 30° tangent to locate the critical root section. Its working stress is written as
$$
\sigma_{F,\mathrm{B2}}=
\frac{F_{mt}}{b m_{mn}}
\frac{Y_A}{Y_J}
\frac{m_{mt}m_{mn}}{m_{et}^{2}}
K_A K_V K_{F\beta}K_{F\alpha},
$$
where \(b\) is the face width, \(m_{et}\) is the outer transverse module, \(m_{mt}\) is the mid-face transverse module, \(Y_J\) is the geometry factor in the denominator, and \(Y_A\) is a root-stress adjustment factor. The geometry factor \(Y_J\) combines tooth form, load position, effective face width after lengthwise crowning, and some adjacent-tooth load sharing. It is a single divisor rather than the set of separate linear factors used in B1.
| Item | Method B1 | Method B2 |
|---|---|---|
| Reference section | Mid-face virtual cylindrical gear | Outer-end virtual gear, Lewis beam |
| Critical root section | 30° tangent to root fillet | 30° tangent, cantilever model |
| Load factors | \(K_A,\ K_V,\ K_{F\beta},\ K_{F\alpha}\) | Same factor set |
| Geometry factors | \(Y_{F\alpha},\ Y_{S\alpha},\ Y_\varepsilon,\ Y_{BS},\ Y_{LS}\) | \(Y_J\) as divisor, \(Y_A\) as adjustment |
| Life factor | \(Y_{NT}\) | \(Y_{NT}\) |
| Surface and notch factors | \(Y_{\delta,\mathrm{relt}},\ Y_{R,\mathrm{relt}},\ Y_{ST},\ Y_X\) | \(Y_{\delta,\mathrm{relt}},\ Y_{R,\mathrm{relt}},\ Y_{ST},\ Y_X\) |
| Rating rule | \(S_F=\sigma_{FP}/\sigma_F>S_{F,\min}\) | Same rule |
Both methods use the same load-factor framework: an application factor \(K_A\), a dynamic factor \(K_V\), a face-load distribution factor \(K_{F\beta}\), and a transverse-load distribution factor \(K_{F\alpha}\). In the representative comparison discussed below, both routes used \(K_A=1.0\), \(K_V=1.0975\), \(K_{F\beta}=1.50\), and \(K_{F\alpha}=1.10\). The difference between the methods therefore comes mainly from the reference section, the geometry factors, and the material-related factors in the allowable stress.
Allowable Stress and Material Sensitivity
The allowable root stress has the same general structure in both methods:
$$
\sigma_{FP}=\sigma_{F,\lim}Y_{ST}Y_{NT}
Y_{\delta,\mathrm{relt}}Y_{R,\mathrm{relt}}Y_X,
\qquad
S_F=\frac{\sigma_{FP}}{\sigma_F}>S_{F,\min}.
$$
The differences lie in the relative notch-sensitivity factor \(Y_{\delta,\mathrm{relt}}\) and the relative surface-condition factor \(Y_{R,\mathrm{relt}}\). Method B1 makes \(Y_{R,\mathrm{relt}}\) depend on the gear material and the root roughness. For carburized and quenched-and-tempered steel with root roughness \(R_Z<1\ \mu\mathrm{m}\), for example, the value is 1.12; for structural steel it is 1.07; for gray cast iron, nitrided steel, and carbonitrided steel it is 1.025. For \(1\ \mu\mathrm{m}\le R_Z\le 40\ \mu\mathrm{m}\), B1 uses material-specific polynomial expressions. Method B2 is simpler: for \(R_Z\le 16\ \mu\mathrm{m}\), it sets \(Y_{R,\mathrm{relt}}=1.025\).
The notch-sensitivity treatment also differs. B1 uses a slip-layer expression involving the material-dependent thickness \(\rho’\), the local tooth geometry, and the size factor. B2 uses a step rule based on the root fillet coefficient \(q_s\): \(Y_{\delta,\mathrm{relt}}=1.0\) when \(q_s\ge 1.5\), and \(Y_{\delta,\mathrm{relt}}=1.5\) when \(q_s<1.5\). B1 therefore gives a more differentiated description of material and surface condition, while B2 gives a more compact, conservative-style geometry correction. For spiral bevel and hypoid gears, ISO 10300 uses \(S_{F,\min}\ge 1.3\); for straight bevel gears or spiral bevel gears with mean spiral angle \(\beta_m\le 5^\circ\), the minimum is \(S_{F,\min}\ge 1.5\).
| Factor | Method B1 | Method B2 |
|---|---|---|
| \(Y_{R,\mathrm{relt}}\) | Material- and roughness-dependent; e.g. 1.12 for carburized steel with \(R_Z<1\ \mu\mathrm{m}\) | 1.025 for \(R_Z\le 16\ \mu\mathrm{m}\) |
| \(Y_{\delta,\mathrm{relt}}\) | Slip-layer expression with \(\rho’\), local geometry, and size factor | 1.0 for \(q_s\ge 1.5\); 1.5 for \(q_s<1.5\) |
Stress and Safety Factors for a Representative Pair
A 16/27-tooth Gleason curved-tooth pair provides a concrete comparison. The pinion and wheel share the same outer transverse module, mean spiral angle, pressure angle, face width, material, and accuracy grade, but have opposite modification coefficients.
| Parameter | Pinion | Wheel |
|---|---|---|
| Number of teeth | 16 | 27 |
| Outer transverse module \(m_{et}\) (mm) | 4.25 | 4.25 |
| Mean spiral angle \(\beta_m\) (°) | 35 | 35 |
| Normal pressure angle \(\alpha_n\) (°) | 20 | 20 |
| Tangential modification coefficient | 0.005 | −0.005 |
| Radial modification coefficient | 0.252 | −0.252 |
| Face width \(b\) (mm) | 17 | 17 |
| Material | 20CrNiMo | 20CrNiMo |
| Accuracy grade | 7 | 7 |
At nominal torques of 200, 300, and 400 N·m, the two ISO routes return the following root stresses. The values are rounded to two decimal places.
| Torque (N·m) | B1 wheel | B1 pinion | B2 wheel | B2 pinion |
|---|---|---|---|---|
| 200 | 280.46 | 289.13 | 292.61 | 302.61 |
| 300 | 420.69 | 433.70 | 438.92 | 453.92 |
| 400 | 560.92 | 578.26 | 585.22 | 605.23 |
| Torque (N·m) | B1 wheel | B1 pinion | B2 wheel | B2 pinion |
|---|---|---|---|---|
| 200 | 4.47 | 4.34 | 4.01 | 3.88 |
| 300 | 2.98 | 2.89 | 2.67 | 2.58 |
| 400 | 2.24 | 2.17 | 2.00 | 1.93 |
Three trends are clear. First, B1 returns stresses about 4.4% lower than B2; averaged over the six gear-and-load combinations, the B1 result is 95.56% of the B2 result. The corresponding safety factors are higher for B1. Second, the pinion stress exceeds the wheel stress at every torque, by roughly 3–4%, and the pinion safety factor is correspondingly lower. This follows from the asymmetric working flanks of a spiral bevel pair: the pinion works on the concave flank, where the root fillet is tighter and stress concentration is stronger, and the smaller tooth number reduces the critical root-section thickness. Third, both standards scale stress linearly with torque. Raising the torque from 200 to 400 N·m exactly doubles the calculated stress and halves the safety factor. That algebraic linearity is convenient for rated-load screening, but it is a weak representation of part-load behavior.
Finite-Element Contact Model
To test the standard predictions, a three-dimensional model was generated from the tooth-surface data of the same pair, assembled, and checked by a simulated roll test for contact-pattern regularity. The analysis used Ansys Workbench with 20CrNiMo properties. The pinion tooth surfaces were defined as contact surfaces and the wheel tooth surfaces as targets, with a friction coefficient of 0.03 and rotating joints on both members. A hexahedral-dominant mesh was used at an overall size of 1 mm, with local refinement to 0.1 mm at the root. The pinion was rotated through 180° while torques of 200, 300, and 400 N·m were applied to the wheel. Weak springs and large deflection were enabled, with 1000 initial and 2000 maximum substeps over an analysis time of 60 s.
Root stress histories were extracted for teeth 1, 3, and 4. Each tooth shows a double-peak history. On the wheel, the first peak corresponds to the tensile action as the tooth enters mesh, and the second peak arises from the compressive contribution as the following tooth takes up load. On the pinion, the order is reversed and the second peak is used. Because the equivalent von Mises stress is unsigned, both events appear as positive peaks; the extraction must therefore follow the loaded flank rather than the absolute maximum alone.
| Torque (N·m) | Wheel B1 | Wheel B2 | Wheel FEA, teeth 1/3/4 | Pinion B1 | Pinion B2 | Pinion FEA, teeth 1/3/4 |
|---|---|---|---|---|---|---|
| 200 | 280.46 | 292.61 | 293.55 / 296.43 / 298.43 | 289.13 | 302.61 | 296.52 / 310.08 / 301.32 |
| 300 | 420.69 | 438.92 | 392.57 / 396.45 / 396.75 | 433.70 | 453.92 | 409.00 / 413.85 / 408.59 |
| 400 | 560.92 | 585.22 | 496.14 / 495.25 / 499.43 | 578.26 | 605.23 | 512.16 / 505.12 / 509.31 |
Comparison and Divergence
The finite-element peaks are not identical from tooth to tooth. Contact ratio, tooth compliance, and the instantaneous contact pattern distribute load unevenly, so teeth 1, 3, and 4 differ by a few megapascals even under the same torque. At 200 N·m, the wheel FEA peaks of 293–298 MPa sit close to the B2 prediction of 292.61 MPa and above the B1 value of 280.46 MPa. The pinion FEA peaks of 297–310 MPa also lie near the B2 value of 302.61 MPa. At 300 N·m, the FEA wheel values of 393–397 MPa and pinion values of 409–414 MPa fall below both standards, but they are closer to B1. At 400 N·m, the FEA wheel peaks of 495–499 MPa and pinion peaks of 505–512 MPa are about 11–13% below B1 and about 15–18% below B2.
The comparison therefore supports the general finding that B1 is closer to the finite-element result than B2 at the higher torques, with differences around 6% in the moderate-load range. It also shows why that statement must be scoped: the 400 N·m points expose a larger gap, and B2 happens to match the 200 N·m peaks more closely. The reason is not a simple offset in a coefficient. The ISO routes treat load sharing as a factor evaluated at the rated condition, whereas the contact solution recomputes the sharing as the teeth deflect and the contact pattern shifts. The standard’s stress remains proportional to torque; the FEA root stress grows sublinearly from 300 to 400 N·m because additional teeth and a changing contact distribution absorb part of the increase.
Method B2 gives the higher working stress and the lower safety factor of the two ISO routes, so it is the more severe stress rating for this pair. Method B1 gives a more differentiated treatment of material and surface condition, but its lower stress should not be read as extra margin unless the rated load and the actual contact pattern support that conclusion. The finite-element result also depends on mesh refinement, friction definition, and peak extraction; it is a verification tool, not an exact replacement for a fatigue test.
Design and Standardization Implications
For rated-load screening, both B1 and B2 are useful, but they answer slightly different questions. B1 is convenient when material, roughness, and mid-face geometry are well defined and a more differentiated root rating is needed. B2 is convenient when a compact geometry factor is available and a more severe stress estimate is acceptable for sizing. Using only the lower B1 stress without a cross-check can understate the root load, particularly when the contact pattern is not fully developed or the tooth is lightly loaded.
The linear scaling of ISO 10300 is appropriate for a single rated operating point, not for a duty cycle with frequent part-load operation. For mining, vehicle, or aerospace transmissions with variable torque, the root damage should be accumulated over a load spectrum, and a contact finite-element analysis should be used where the contact pattern or adjacent-tooth sharing changes appreciably with load. The pinion should be checked separately: in the present pair it carries about 3–4% higher root stress and a correspondingly lower safety factor than the wheel, a difference that matters in strength-balanced design.
Standardization also benefits from this kind of cross-check. GB/T 10062 is derived from the ISO 10300 series, and both B1 and B2 remain part of the international framework. Updating a national standard requires more than adopting the latest formula set: the reference section, the material factors, the load-sharing assumption, and the treatment of part-load behavior must be validated against contact analysis and, ultimately, against strain or fatigue tests. For spiral bevel gears, the practical conclusion is to use the ISO routes as rated-load screening tools, to compare B1 and B2 rather than treating them as interchangeable, and to reserve finite-element contact analysis for the cases where the contact pattern, load spectrum, or safety margin makes the linear standard insufficient.