Gears

Intuition

A gear trades speed for torque at approximately fixed power. The kinematics and gear ratio follow from pitch radii. The difficulty lies in tooth contact. Two curved surfaces roll and slide under Hertzian pressure near a gigapascal, millions of times, in oil, for years. Everything else is bookkeeping.[fn:: This is why gear design looks deceptively simple in a machine design textbook. One chapter, a few Lewis-formula examples. Gear engineering is in fact a multi-billion-dollar industry. It has its own standards body (AGMA), its own metallurgy (carburized 8620, nitrided 4140), its own lubrication regime, and its own dedicated software (KISSsoft, Romax).]

The power transmits through a line contact of infinitesimal width. See the Tribology note on Hertz contact. The load per unit face width is limited by what that contact can survive over 10 to the 8th to 10 to the 9th cycles. The gear designer is, fundamentally, a contact fatigue engineer who also worries about bending and scuffing. The bending fatigue from Lewis-style tooth breakage is the catastrophic failure. Designers usually check it first. The contact fatigue from pitting is the life-limiting failure. It actually retires the gear.

Gear Types

The taxonomy is driven by the geometry of the shafts and the noise and speed requirements. Each type has a characteristic failure signature.

Spur

Teeth are parallel to the axis. This is the default type. Contact occurs along a line instantaneously appearing across the full face width and then disappearing. Spur gears are noisy at speed. They are limited to about 20 meters per second pitch-line velocity in commercial practice.[fn:: The instantaneous line across the face is an idealization. Real teeth have tip and root relief. Manufacturing errors smear the load transition over a few degrees of roll angle. The noise is the acoustic signature of the idealized load transfer, modulated by the imperfections. Gear noise control is essentially the art of deliberately making the tooth profile less ideal in a way that softens the load transition.]

Helical

Teeth are inclined at helix angle psi. Contact begins at one end of the tooth and sweeps across the face. Multiple teeth share load at all times. Helical gears are smoother, quieter, and carry higher load for the same module. They are the workhorse of anything above about 5 meters per second. The price is an axial thrust component (F axial equals F tan times tan psi) that must be reacted by a thrust bearing or canceled by a herringbone (double-helical) arrangement. The thrust is not a side-effect. It is often the governing constraint on helix angle selection.

Bevel

Bevel gears connect intersecting shafts with conical pitch surfaces. Straight, spiral, and Zerol variants exist. Bevel gear design is mostly empirical and CAM-driven. Lewis-style design is only a rough screen. Spiral bevels are quieter, stronger, and costlier to make than straight bevels.

Worm

A worm is a screw meshing with a worm wheel. It gives 10:1 to 100:1 ratios in a single stage. Contact is sliding-dominated. Efficiency ranges from 50 to 90%, lower for high ratios. High-ratio worms can be self-locking. The wheel cannot back-drive the worm. This helps hoists but traps designs that assumed reversibility.[fn:: Self-locking is not binary. It depends on lead angle, friction coefficient, and whether the system is vibrating. A self-locking worm that unlocks under vibration is a known and occasionally lethal failure mode. Design for self-locking with margin, or use a brake.]

Planetary (epicyclic)

A planetary gear has a sun, planets, ring, and carrier. Three coaxial members can be input, output, or fixed. This yields compact, high-ratio power transmission. Load shares among N planet meshes. Torque density is high. Planet load sharing is the real difficulty. Perfect sets share one-third each. Real sets share unevenly due to carrier errors and ring flexibility. A floating design lets one member find its radial position. It trades geometric precision for elastic averaging.

Ratios, Torque, and Speed

The gear ratio i = ω_in/ω_out = N_out/N_in = r_out/r_in. Ideal torque T_out = i × T_in. Real gears lose 1–3% per mesh to friction, plus windage and churning losses. Single-stage ratios above 8:1 (spur/helical) or 6:1 (worm) grow uneconomical. Ratio choice is a system problem dictated by motor and load speeds. The gear designer minimizes cost, weight, and noise within the envelope.

Contact Ratio

Contact ratio is the average number of tooth pairs in mesh. It must exceed 1.0 for continuous motion. Spur gears target 1.2–1.6. Helical gears target 2.0–3.0 with face contact ratio added. Higher ratios mean smoother motion, lower peak loads, and less noise. The cost is tighter manufacturing tolerances. A ratio of 1.0 barely transmits motion and does so loudly.

Backlash

Backlash is deliberate circumferential clearance between mating teeth. It accounts for thermal expansion, lubricant film, and manufacturing tolerance. Any reversing load makes teeth cross the gap before picking up load. This causes impact, noise, and lost motion in servo systems. Anti-backlash strategies include split gears, modified forms, and skew-cut pairs. All trade cost, friction, or load capacity.

Materials and Heat Treatment

Gears face a surface-vs-core tension. The tooth surface must be hard to resist pitting and wear. The core must be tough to resist bending impact and tooth fracture. The canonical solution is case hardening. Carburizing or nitriding a low-carbon alloy steel produces a hard surface of 58 to 62 HRC over a tough core of 25 to 35 HRC.

Through-hardening (4140 QandT to about 30 HRC) is the cheap option for low-to-medium duty. It gives no surface gradient and is limited by the bulk hardness. Carburizing depth is typically specified as 0.5 to 1.5 mm. Pitting that penetrates the case progresses rapidly once it reaches the softer core.[fn:: The case depth is a free parameter framing understates how much of gear cost and distortion is determined by the heat treatment. Carburizing at 925 C for 8 hours followed by oil quench and temper distorts the blank measurably (runout, lead, profile). High-precision gears are cut after carburizing and ground into the hardened case. A second operation that roughly doubles the gear cost. Nitriding, being lower-temperature, can be done after finish cutting. It skips the grind, at the cost of lower case hardness and thinner case.]

Failure Modes

Bending fatigue (tooth breakage)

The tooth is a cantilever beam loaded at the tip. The critical section is at the root fillet. The stress concentration factor ranges from about 1.5 to 2.0. The stress concentration and the bending moment combine. The Lewis equation sigma equals Ft over (b m Y) is the first-order model. M is module. Y is the Lewis form factor. This is catastrophic failure. A broken tooth ends the gear. It is usually the checked mode rather than the experienced mode in well-designed gears. See the Fatigue and Fracture note for the S-N and delta-K framework.

Pitting (contact fatigue)

Subsurface Hertzian shear stress initiates a crack below the surface. The crack propagates to the surface. A flake of material spalls off. This is the life-limiting mode for hardened steel gears. The AGMA contact stress formula is essentially a Hertzian contact stress (sigma H is proportional to the square root of F over b d1) modified for geometry, material, and dynamic factors. The formula checks this stress against an allowable stress derived from material S-N data.

Scoring / scuffing

When the elastohydrodynamic lubricant (EHL) film breaks down, asperities weld and tear. This breakdown occurs at high speed, high load, or inadequate lubrication. Unlike pitting, scoring is adhesive wear, not fatigue. It can destroy a gear in minutes rather than years. It is temperature-driven. The flash temperature at the asperity is the trigger. The Blok flash-temperature criterion and the AGMA scuffing index are thermal calculations, not stress calculations.[fn:: Scoring separates "gear" from "gear in a gearbox". A gear set run open and dry at moderate load may pit slowly for years. The same set in a sealed gearbox with inadequate lubricant cooling can score in an afternoon. The lubricant is not a consumable added to the gear. It is a structural member of the gear system.]

The AGMA Approach

The American Gear Manufacturers Association (AGMA 2101 / 2001) is the de-facto standard for enclosed gear design. Most industrial gears are sized against this framework. The structure is empirical-analytical. You compute the bending stress at the tooth root and the contact stress at the pitch line from closed-form expressions. These expressions are modified by a stack of empirical factors. The factors include overload (KO), dynamic (KV), size (KS), load distribution (Km), and others. You compare the modified stresses against allowable stresses further adjusted by life, reliability, and temperature factors.

sigma bending = Ft KO KV KS Km / (b m J) is less than or equal to sigma allow, bending sigma contact = ZE sqrt(Ft KO KV KS Km (ZR/ZI) (1/(b d1))) is less than or equal to sigma allow, contact

The factors are separable. Each captures one physical effect and can be refined independently. The weakness is that the factors are correlated in reality. A misaligned gear is also dynamically loaded differently. Multiplying the factors can over- or under-predict depending on the direction of correlation. The approach works because it is calibrated to a century of test data. The designer who follows it is standing on a pile of broken gears that calibrated the safety factors.[fn:: The ISO 6336 method is structurally similar but differs in factor definitions and in some of the underlying S-N curves. The two standards do not give identical answers for the same gear. This is a source of quiet transatlantic disagreement in gear design reviews. The discrepancy is typically 5 to 15 percent, enough to matter at a margin but not enough to matter in a comfortable design.]

The Drivetrain Hub notebook collection is the best freely available modern reference for working through AGMA-style gear calculations with current factor values.

Worked Sketch — Spur Gear Pair

A through-hardened steel spur pair. Pinion N1 is 20 teeth. Gear N2 is 60 teeth (ratio i equals 3:1). Module m is 3 mm. Face width b is 30 mm. Pinion speed n1 is 1800 rpm. Material is 4140 QandT with sigma allow,bend approximately 200 MPa and sigma allow,contact approximately 1200 MPa.

Pitch radii: r1 equals m N1 over 2, which equals 30 mm. R2 equals 90 mm. Pitch-line velocity v equals pi d1 n1 over 60, which equals pi times 60 times 1800 over 60, or 5.65 meters per second. This value is within the spur regime but approaches the upper limit.

Transmitted load (assuming 10 kW at the pinion): Ft equals P over v equals 10 000 over 5.65 equals 1770 N.

Bending (Lewis first order, neglecting K factors for the sketch): sigma equals Ft over (b m Y), with Y approximately 0.32 for a 20-tooth pinion. sigma equals 1770 over (0.030 times 0.003 times 0.32) equals 61.5 MPa, which is comfortable against 200 MPa.

Contact (Hertzian, first order): sigma H equals ZE times the square root of (Ft over b d1 times (i+1)/i), with ZE approximately 191 square root MPa for steel-steel. sigma H equals 191 times the square root of (1770 over (0.030 times 0.060) times 4/3) equals 191 times the square root of 1.31 times 10 to the 6th power, which is approximately 691 MPa. This is comfortable against 1200 MPa.

The point of the sketch is not the numbers. The numbers change materially once KO, KV (approximately 1.2 to 1.4 at this speed), and Km (approximately 1.2 to 1.6 depending on alignment) are included. The point is the order of magnitude. The contact stress is the binding constraint (691 divided by 1200 is 0.58 utilization versus 61.5 divided by 200 is 0.31 for bending). This is the generic result for hardened steel gears. The AGMA contact check is the one that usually governs the size.[fn:: A common student reflex is to size on bending and then check contact. It is the opposite for case-hardened steel gears. The contact stress almost always governs. The bending check is satisfied with margin. The exception is low-speed, high-shock applications such as hoist drives and mill drives. Those use through-hardened or normalized gears because contact is not the limit.]

Pitfalls

1. Specifying only the ratio. "I need a 10:1 gearbox" is not a gear specification. The torque, speed, duty cycle, input power, mounting, and lubrication all change which gear type and material are appropriate. A 10:1 worm at 0.1 kW and a 10:1 helical at 100 kW share only the integer 10.

2. Ignoring the thrust. A helical or spiral bevel gear generates axial load that the shaft and bearings must react. A gearbox designed without the thrust bearing is a gearbox that will fail its bearings.

3. Treating the lubricant as an afterthought. The EHL film thickness (typically 0.1 to 1.0 micrometers) is the structural element separating the tooth surfaces. Without it, the gear operates in boundary lubrication, and scoring is a matter of time. The lubricant viscosity and the additive package are gear-design parameters, not procurement decisions.

4. Underestimating the dynamic factor. The AGMA KV accounts for tooth meshing dynamics. It increases sharply at high pitch-line velocity. A gear sized at low speed and run at high speed will be under-strength by a factor that the static calculation never showed.

5. Assuming "case hardened" is uniform. The case depth, case hardness gradient, and core hardness are all specified separately. A gear with adequate case hardness but inadequate depth will pit through the case and fail rapidly. The specification is three numbers, not one.

Related

• Power Transmission — the system the gear lives in. Ratio, duty, and mounting context.
• Bearings — gear shafts need them. The thrust bearing is often the gear governing constraint.
• Journal Bearings — the sliding-contact alternative for high-speed, lightly-loaded shafts.
• Fasteners — mounting gears to shafts. The spline or key is part of the gear system.
• Tribology — the EHL film and the Hertzian contact that govern gear life.
• Lubrication — the lubricant as structural member. Viscosity, additives, film thickness.
• Fatigue & Fracture — the S-N framework behind both bending and contact fatigue limits.
• Material Selection — the case-vs-core tension and the carburize/nitride decision tree.