AI for Engineering Productivity
Bearing selection made practical: how to size for load, speed, and life, choose the right bearing type, and use AI to get cited, standards-backed answers.
·
⏱
9 min read
Michelle Ben-David
Michelle Ben-David is a mechanical engineer and Technion graduate. She served in an IDF elite technology and intelligence unit, where she developed multidisciplinary systems integrating mechanics, electronics, and advanced algorithms. Her engineering background spans robotics, medical devices, and automotive systems.
BOTTOM LINE
Bearing selection comes down to a disciplined sequence: define the radial and axial load, set a speed and life target, account for the environment, then match the dynamic and static load ratings to a bearing type that fits the space. The rating life formula is simple, but the judgment lives in the factors around it, the fits, the lubrication, and the misalignment the bearing will actually see. Most failures trace back to a missing input rather than bad math. Treat the load case as the real deliverable, verify every catalog value against its source, and keep a record of why each bearing was chosen. Do that consistently and bearings stop being a source of surprise field failures. Tools that surface cited data and past decisions make that discipline easier to hold to on every project.
Bearing selection is one of those tasks that looks routine until a bearing runs hot, seizes, or fails months after a product ships. A rolling element bearing sits where load, speed, temperature, and lubrication all meet, so a choice that looked reasonable on paper can quietly shorten service life or add cost that never needed to be there.
Most mechanical engineers learn the fundamentals in school, then build practical judgment over years through catalogs, standards, and the occasional field failure. This guide walks through how to select a bearing the right way, from defining the load case to reading a life calculation, and shows where an engineering AI assistant can cut the time you spend digging through catalogs and standards.
Why Bearing Selection Goes Wrong
Bearings rarely fail because an engineer lacked skill. They fail because one part of the load case was missed, or because a catalog value was read for the wrong operating condition. The most common problems repeat across industries.
Undersizing for the real dynamic load. A bearing sized for the steady running load can still be overwhelmed by start-up torque, shock loads, or a jammed mechanism.
Ignoring axial load. Many engineers size for radial load and forget that a helical gear, a fan thrust, or a preload spring pushes the bearing along its axis.
Missing misalignment. Shaft deflection and housing tolerances tilt the inner ring relative to the outer ring, and a rigid ball bearing does not tolerate that the way a self-aligning bearing does.
Wrong shaft and housing fits. A loose fit lets the ring creep and wear, while a fit that is too tight reduces internal clearance and can preload the bearing until it overheats.
Poor lubrication and sealing. More bearings die from contamination and the wrong grease than from fatigue, yet lubrication is often the last thing specified.
Getting the shaft and housing fit right is a selection decision in its own right. Our guide to engineering fits and tolerances covers how clearance, transition, and interference fits change the way a bearing seats and behaves.
IN PRACTICE
What Engineers Are Saying
"Leo uses a Large Mechanical Model trained on 1M+ technical sources. It also provides citations, so we don't have to guess whether a material property or tolerance is correct. We see 96% accuracy on technical queries."
- Dorian G., AI Engineer
The Parameters That Drive the Decision
Before you open a catalog, you need a clear picture of what the bearing has to survive. Five inputs do most of the work.
Radial and axial load. Separate the load into its radial component (Fr) perpendicular to the shaft and its axial component (Fa) along the shaft. The ratio between them decides which bearing types are even viable.
Speed. Rotational speed in revolutions per minute sets heat generation and the limiting speed of the bearing. High speed pushes you toward ball bearings and better lubrication.
Required life. Decide how many hours or revolutions the bearing must last at the design load. A conveyor gearbox and a surgical tool have very different targets.
Operating environment. Temperature, moisture, dust, and chemical exposure change the material, the seal, and the lubricant, not just the size.
Space and mounting. The available bore, outer diameter, and width often constrain the choice as much as the load does.
Two catalog numbers turn these inputs into a size. The basic dynamic load rating (C) is the load a bearing can carry for a basic rating life of one million revolutions. The basic static load rating (C0) is the load the bearing can hold without permanent deformation while stationary or barely turning. You compare C against the equivalent dynamic load and C0 against the equivalent static load. Material choice feeds into this too, and our overview of materials selection in mechanical design explains how property data drives these ratings.
How to Calculate Bearing Life
Bearing life is a statistical quantity, not a guarantee. The basic rating life (L10) is the life that 90 percent of a group of identical bearings will reach or exceed under a given load. Ten percent are expected to fail earlier, which is why safety-critical designs use a higher reliability target.
The ISO 281 rating life equation is straightforward.
Find the equivalent dynamic load (P). Combine radial and axial load with the catalog factors X and Y so that P equals X times Fr plus Y times Fa. For pure radial load, P is simply Fr.
Apply the life formula. L10 equals the ratio (C divided by P) raised to the power p, in millions of revolutions. The exponent p is 3 for ball bearings and 10 divided by 3 for roller bearings, which is why roller bearings gain life faster as you oversize them.
Convert to hours. Bring in speed with L10h equal to (1,000,000 divided by 60 times n) times (C divided by P) to the power p, where n is speed in revolutions per minute. Hours are what most specifications actually call out.
Check the static safety factor. Divide C0 by the equivalent static load to get s0, and confirm it clears the value your application demands, which is higher for shock and vibration.
For a more realistic answer, ISO 281 adds a modified rating life that scales L10 by a reliability factor and a factor for lubrication and contamination. Clean oil, correct viscosity, and good sealing can multiply life several times over, while dirt and thin films cut it sharply. If you already run simulations, pairing the hand calculation with a model helps, as we describe in our look at FEA driven design optimization, and the arithmetic itself is the kind of task covered in AI for engineering calculations.
Choosing the Right Bearing Type
Once the load case is clear, the type usually picks itself. Each family trades capacity, speed, and tolerance for misalignment in a different way.
Deep-groove ball bearings. The general-purpose default. They carry moderate radial load, take some axial load in both directions, and run at high speed with low friction.
Angular-contact ball bearings. Built for combined radial and axial load in one direction. Mounted in pairs, they hold spindles and other precision shafts rigid.
Cylindrical roller bearings. High radial capacity thanks to line contact, and some designs let the shaft float axially to absorb thermal growth.
Tapered roller bearings. Heavy combined radial and axial load, which is why they show up in wheel hubs and gearboxes. They are mounted in opposing pairs and set with a specific preload.
Spherical roller bearings. High load capacity plus tolerance for misalignment, well suited to long shafts and heavy industrial equipment.
Thrust and needle bearings. Thrust bearings handle pure axial load, and needle bearings pack high radial capacity into a small radial space.
Availability matters as much as the ideal choice. A slightly larger standard bearing that a distributor stocks usually beats a perfect custom size with a long lead time.
Where AI Speeds Up Bearing Selection
The slow part of bearing selection is rarely the formula. It is gathering the right numbers: the correct X and Y factors for a given load ratio, the limiting speed for a seal type, the life adjustment for a contamination level, and the internal clearance class for a hot-running shaft. That information is scattered across standards, catalogs, and the memory of whoever last solved a similar problem.
Leo is an AI assistant built for mechanical engineers and trained on more than one million pages of standards, engineering books, and technical references. Instead of opening five catalogs and a copy of ISO 281, an engineer can ask for the equivalent load factors or the life adjustment and get an answer with a citation to the source, so the number can be verified rather than trusted blindly. That accuracy is the point, because a bearing sized on a misread table is a warranty claim waiting to happen.
Leo also connects to an organization's own knowledge base and offers integrations with leading PDM and PLM platforms, including SolidWorks PDM, Autodesk Vault, PTC Windchill, Siemens Teamcenter, and Arena PLM, among others. That means the bearing your team qualified for a similar gearbox two years ago, along with the reason it was chosen, is available while you design rather than lost when that engineer moves on. Pairing standards-backed answers with a searchable record of past decisions is how selection gets both faster and more consistent. The same idea applies to related design math, such as tolerance stack-up analysis, where a small error compounds across an assembly.
FAQ
Stop Guessing on Bearing Selection
Get cited, standards-backed answers while you design, not after.
Leo AI gives mechanical engineers standards-backed answers with citations, and connects to your PDM and PLM so past design decisions are there when you need them.
Schedule a Demo →
#1 New AI Software Globally - G2 2026
Enterprise-grade security
Trusted by world-class engineering teams
