AXILINEBearing Life, Vibration, and Load: What Balancing Actually Gets You
You're replacing the bearing on the induced-draft fan for the third time in two years, and the supplier says every time that the bearing is a good one. Vibration, meanwhile, sits at 6 mm/s, and everyone's gotten used to it. That's exactly how imbalance works: it doesn't break the assembly outright, it quietly adds a rotating force to the working load, one that travels around the bearing once every revolution. Below we go through the mechanics of that link, why calculated life reacts so sharply to load, and how to work out the benefit for your own machine without pulling numbers out of thin air.
What Imbalance Does Inside the Bearing
An unbalanced rotor pulls itself sideways. The centre of mass doesn't sit on the axis of rotation, so rotation produces a centrifugal force directed from the axis toward the heavy spot. It's a simple calculation: force equals the imbalance in kg·m times the square of the angular velocity in rad/s. Imbalance here means mass times radius — those same g·mm figures from the balancing report.
Take a typical example to get a feel for the scale. A fan wheel has picked up 50 g of buildup at a radius of 400 mm. That's 20,000 g·mm, or 0.02 kg·m. At 1500 rpm the angular velocity is 157 rad/s, and its square is about 24,700. The force comes out to roughly 490 N — the same as a 50 kg weight hanging off the wheel. The difference is that this "weight" travels continuously around the casing, 25 times a second. This example is illustrative; the numbers for your machine will be its own.
From there the force splits between the two bearing housings in proportion to their moment arms, exactly like a beam problem. On a rotor symmetric between its supports, that's roughly an even split. On an overhung wheel, the near support takes substantially more, and the far one often works in tension.
And here's the key part. The vector rotates together with the rotor. For half a revolution it points down and adds to the weight; for the other half it points up and subtracts from it. The constant load from the rotor's weight and belt tension stays put, while the rotating force rocks it back and forth: over one revolution the bearing passes through a maximum and a minimum load. It's this rocking, not the average value, that determines fatigue.
- In a bearing with a stationary outer ring, the load zone is limited: under the rotor's weight, only the lower sector of the raceway is loaded. The rotating force from imbalance drags this zone around the full circumference, and sections of the raceway that normally carry no load at all start carrying it.
- At the point of contact between a rolling element and the raceway, contact stresses act (stresses in a tiny contact patch, calculated using Hertzian formulas). The maximum shear stress lies not on the surface but beneath it, at a depth of a few tenths of a millimetre. That's where the damage starts: first a subsurface crack, then it breaks through to the surface, then spalling.
- Contact stresses grow nonlinearly with load, because the contact patch grows more slowly than the force does. The stress margin gets eaten up faster than the percentage increase in load would suggest.
- Clearance in the bearing works against you. The larger it is, the fewer rolling elements are carrying load at any one time, and the higher the load on each one.
Imbalance isn't the only rotating force in a machine. Misalignment, shaft bending, thermal bow, and uneven flow through the flow path also create variable loads. Balancing only removes the part that sits at the running-speed frequency, 1x, and is caused by the rotor's mass distribution.
One Revolution, One Cycle: the Arithmetic of Fatigue
Fatigue is counted in cycles, not hours. That's why speed matters more than the calendar in this story.
A machine at 1500 rpm makes 90,000 revolutions an hour, about 2.2 million a day, and roughly 650 million over a year of running with almost no stops. Every revolution is one loading cycle from imbalance. A point on the outer raceway takes a hit from every rolling element that passes it, and several pass per revolution.
Two practical consequences follow. First: the same imbalance is more dangerous at 3000 rpm than at 750, and doubly so. Force grows as the square of speed, while the number of cycles grows linearly. Second: the phrase "it'll keep running like this until the scheduled overhaul" doesn't mean a delay of a few months — it means hundreds of millions of cycles the assembly will burn through, irreversibly.
Spent fatigue life doesn't come back. Balancing the machine stops you from burning through it quickly, but it doesn't give back what's already been spent. That's why balancing right after a repair — not once the shaking starts — is economics, not pedantry.
- Record the actual working speed, not the nameplate figure. A machine on a variable-frequency drive lives at different speeds, so log the operating condition together with the measurement.
- Calculate how many revolutions the assembly makes between scheduled overhauls. That's the order of magnitude of the cycle count you're dealing with.
- If the machine runs at several operating conditions, break the running time down by condition: speed and load differ between them.
- Note starts and stops separately. During coast-down, a high-speed machine passes through its resonances, and the load on the supports is briefly higher than the working load there.
Why Calculated Life Reacts So Sharply to Load
A bearing's basic rated life is calculated per ISO 281. The formula is short: life in millions of revolutions equals the ratio of the dynamic load rating C to the equivalent dynamic load P, raised to a power. That power is 3 for ball bearings and 10/3 for roller bearings.
From there, the arithmetic of the exponent takes over, and it's exactly what explains the whole interest in balancing. Reduce the equivalent load by 10%, and a ball bearing's calculated life rises by roughly 37%. Reduce it by 20%, and it very nearly doubles. Halve it, and life rises by a factor of 8 — roughly a factor of 10 for a roller bearing. Life doesn't change by a little; it changes by multiples.
Now for the honest part. These multipliers apply to the equivalent dynamic load P, not to vibration in mm/s, and not to the force from imbalance on its own. P already contains the rotor's weight, belt tension, and the axial and radial process loads. If the force from imbalance is a fifth of P, eliminating it completely will reduce P by roughly 20%, not by half. If it's the lion's share, the effect will be dramatic. The share is different for every machine, and the only way to know it is to calculate it for the specific assembly.
Three caveats, without which the calculation turns into self-deception:
- Basic rated life is a statistical quantity. It means that 90% of the bearings in a batch, under identical conditions, will run for at least that long. It says nothing about your specific bearing.
- ISO 281 also gives a modified life, where the basic figure gets adjusted with factors for lubricant viscosity and cleanliness, contamination, and the bearing's fatigue load limit. Poor lubrication or abrasive contamination in the assembly can easily wipe out the entire gain from reducing the load.
- The calculation is a tool for selecting a bearing at the design stage, not a forecast of the remaining life of an assembly in which a defect has already taken hold. Take the applicable edition of the standard and the methodology from the bearing manufacturer.
Calculating the benefit with this formula makes sense once you have the catalogue value of C, an estimate of the actual P, the speed, and the lubrication cleanliness class in hand. Without them, any claim that "balancing doubles bearing life" is a nice sales line, not a calculation. We don't quote figures like that.
Sources: ISO 281:2007
Why mm/s Can't Be Converted into Bearing Load
The temptation is understandable: vibration dropped from 7.2 to 1.8 mm/s, so the load on the bearing must have dropped fourfold. That's not how it works.
Vibration in mm/s is the response of the "rotor, supports, frame, foundation" system to a force. That response depends on stiffness, mass, and damping. The same imbalance on a rigid bedplate will give modest mm/s figures, because the structure barely moves and the force gets transmitted to the bearing almost in full. On a compliant frame, the same force will give high mm/s figures, but part of it goes into moving the rotor rather than into the support. Above the first critical speed (the speed at which the rotor passes through its own resonance), the rotor actually starts rotating around its own centre of mass, and the force transmitted to the supports drops — even though the imbalance hasn't gone anywhere.
Here's what follows from this in practice:
- For assessing machine condition and for trending, use overall vibration in mm/s RMS (root-mean-square) over the 10–1000 Hz band and the A, B, C, D zones per the applicable part of ISO 20816. That's what it's for.
- For assessing force, use imbalance, radius, and speed. Force is proportional to the imbalance in g·mm and to the square of the angular velocity, and it doesn't depend on the frame's stiffness.
- The most accessible source of a real number for your machine is the balancing report. The mass of the correction weights installed, multiplied by the installation radius, shows the imbalance you actually removed. Plug it into the centrifugal-force formula and you get the load you removed, in newtons.
- A reduction in the 1x amplitude by a given factor corresponds roughly to a reduction in imbalance by the same factor, at constant speed and stiffness. That's exactly why, for a conversation about load, 1x matters more than overall vibration.
There's an unpleasant conclusion here for rigid machines. A heavy pump on a massive foundation can read 2 mm/s while still driving a substantial rotating force through its bearings. Low mm/s doesn't prove the rotor is balanced. The proof is the residual imbalance in g·mm/kg relative to the chosen G accuracy class.
Sources: ISO 20816-1:2016
What Benefits Alongside the Bearings
The bearing is the most load-sensitive component, but it's not the only one. The rotating force doesn't stop there: it travels into the casing, the feet, the frame, the foundation, and comes back around. So balancing shows up in the maintenance record more broadly than through a single line item, "bearing replacement."
| Component | What the Rotating Load Does to It | What You Get After Balancing |
|---|---|---|
| Impeller and its welds | Gussets, blades, and discs work under alternating bending, cracks grow from stress concentrators at the welds | Cracks stop growing quickly, weld inspections turn up fewer findings |
| Shaft and impeller fit | Cyclic shaft bending, fretting (wear from microscopic relative movement between parts) on the fit surfaces and in the keyway, a worn fit | The fit lasts longer, the impeller doesn't work loose on the shaft between overhauls |
| Fasteners and feet | Bolts back off, holes wear oversized, soft foot develops on its own | Tightness holds, retightening is needed less often |
| Frame, anchor bolts, grout | The concrete under the base wears away, the grout washes out, the anchor bolts lose preload | The base stops "breathing," and the results of any work done on it hold |
| Shaft alignment | Worn feet and a settled frame drag the alignment out of true, misalignment comes back after every alignment job | Alignment holds between overhauls, not just until the end of the week |
| Seals and coupling | Increased runout accelerates wear of lip seals and elastic elements, the mechanical seal starts leaking | Fewer leaks and less rubber dust under the coupling |
| Belts and pulleys | Tension pulsation, slippage, localized groove wear | Belts last longer, tension is more stable |
| Machining accuracy | Spindle vibration transfers to the workpiece: lobing (out-of-roundness), waviness, accelerated tool wear | Consistent surface finish and fewer rejects for geometry |
| Noise and working conditions | The running-speed component creates a hum that travels through piping and building structures | Quieter at the workstation, less secondary vibration on neighbouring equipment and instruments |
These benefits are harder to put a number on than bearing life, but they show up in the maintenance log. Look at this asset's two-year history. If it regularly features bolt retightening, repeat shaft alignment, belt replacements, and lip-seal replacements, a rotating load is almost certainly among the causes.
How to Honestly Calculate the Benefit for Your Own Machine
There's no universal multiplier, but you can get a measurable answer for your specific asset. It takes discipline in record-keeping, not complicated mathematics.
- Change one factor at a time if you want to know what actually worked. Balanced the rotor, switched lubricant, aligned the shafts, and installed a different bearing brand, all at once? You'll get a benefit, but you won't know why.
- Keep your measurements in one consistent format: point, direction, sensor mounting, speed, operating condition, Fmax, and number of lines. Change a setting without logging it, and your annual trend turns into a set of numbers you can't compare.
- Log running time in hours and in revolutions, not just dates. For a machine with variable speed, dates are deceptive.
- Photograph the bearings you pull. The nature of the damage is a diagnosis of the cause, and it's worth more than any estimate of remaining life.
- 01
Take a Baseline Before the Work
Overall vibration in mm/s RMS plus 1x amplitude and phase at both bearing housings, at the same points, in the same direction, with the same sensor mounting. Plus the speed, operating condition, and load of the unit. Without this record, you'll have nothing to compare against later.
- 02
Record How Much Imbalance You Removed
Write down the mass of the correction weights installed and the installation radius from the report. Their product in g·mm is exactly the imbalance you removed. Multiply it by the square of the angular velocity and you get the rotating force removed, in newtons. That's your one honest figure for the load.
- 03
Pull Up the Bearing Replacement History
For this specific asset: replacement dates, running hours between them, and what was found on the bearings that were pulled. Raceway spalling, cage wear, overheating marks, and abrasive marks each point to different causes, and balancing doesn't cure all of them.
- 04
Cost Out an Unplanned Stoppage in Your Own Numbers
Hours of downtime, the cost of the bearing and the labour, lost output, overtime. Our figures won't fit here — you already have your own in your records. This is the baseline for comparing against the cost of a site visit or an instrument.
- 05
Repeat the Measurement at the Same Operating Condition
A month later, three months later, six months later. One measurement shows a condition; decisions get made from a trend. A twofold rise from the baseline is reason enough to investigate, even within an acceptable mm/s zone.
- 06
Compare Running Time Between Replacements a Year Later
Before and after. This is the only answer that actually applies to your specific machine. A single asset isn't a statistic, so if you have a fleet of similar units, compare groups: balanced ones against the rest.
If vibration dropped after balancing but bearings keep failing at the same rate, that's a valuable result, not a failure. It means imbalance wasn't the main load on this assembly, and you need to look at lubrication, installation, alignment, or bearing selection instead. That's how you'll actually save the next bearing.
Sources: ISO 20816-1:2016
What Balancing Doesn't Cure
Balancing reduces exactly one force: the rotating centrifugal force from unbalanced mass, the one that sits at running-speed frequency. It doesn't touch anything else in the bearing assembly. Below are the reasons bearings die prematurely on machines that are already balanced.
Lubrication
The wrong grade, the wrong interval, over-greasing, foaming, water washout, loss of viscosity from overheating. The lubrication regime feeds directly into the modified life calculation: with a poor oil film, even a perfectly balanced rotor won't save the raceway.
Contamination
Abrasive particles, water, wear debris, swarf. A hard particle rolled over by a rolling element leaves a dent, and that dent becomes a stress concentrator and the starting point for spalling. Seals and cleanliness during installation matter more here than weights.
Installation
A fit outside tolerance, press-fitting force transmitted through the rolling elements, torch heating, hammer blows on the ring, a raceway dented during installation. A significant share of premature failures are born on the workbench, ten minutes before installation.
Ring Misalignment
The outer ring is tilted in the housing, the fit surfaces are out of line with each other, the shaft is bent, the cap is drawn down unevenly. The rolling elements travel the raceway at the wrong angle, and the load zone slides toward the edge of the raceway. Weights don't change this geometry.
Incorrect Selection
A bearing selected for the wrong load, speed, or temperature. Two rigidly fixed bearings with no floating support end up pinched by the shaft's thermal expansion and take on an axial load that was never in the calculation.
Misalignment and Drive Tension
An over-tensioned belt and misaligned coupling halves produce a constant component of load. It goes into the equivalent dynamic load in full, and the only way to remove it is through shaft alignment and correct tensioning, not a correction mass.
Bearing Currents
A variable-frequency drive with no insulated bearing or grounding ring burns the raceway with electrical discharges. It looks like a rolling-contact defect on the surface, but the fix is electrical.
Resonance
At resonance, vibration is inflated by the structure's response, not by the force itself. Balancing there gives an unstable result, while the bearing keeps taking load from the structure being shaken. We have a separate article on the signs of resonance and how to detune it.
- There's an honest scenario in which balancing will do almost nothing for bearing life: a slow-running assembly under a heavy process load — a crusher, a slow-turning drum, a gearbox shaft. There, the weight and the working load are an order of magnitude larger than the force from imbalance, and that's not the force you need to reduce.
- The reverse scenario is just as real. A light impeller at high speed: the force from imbalance can easily become the assembly's dominant variable load, and this is where balancing pays off the most.
- Balancing doesn't undo a raceway defect that's already developed. A spalled bearing gets replaced, and balancing removes the cause so the new one doesn't go the same way. We have a separate piece on the signs and stages of a defect, on diagnosing bearings from vibration.
Check whether the 1x phase repeats from run to run. A worn bearing with excessive clearance behaves like looseness: the phase drifts, the influence coefficient comes out unreliable, and balancing won't converge. A drifting phase is a signal to sort out the mechanics first, not to increase the trial weight.
Where to Start and How AXILINE Can Help
The work sequence is simple and doesn't call for heroics. First, a measurement you can trust: sensors on the bearing housings as close to the bearing as possible, rigid mounting on a stud or a magnet on a clean spot, the same radial direction from run to run, a laser tachometer on the reflective tape mark. Then, an assessment of whether it's actually imbalance: the 1x share of overall vibration, the spectrum, repeatability across three runs. Only after that, weights.
Balanset-1A is built for exactly this task. Two accelerometers, a laser phase sensor, a two-channel USB module with preamplifiers, integrators, and an ADC, Windows software. It measures overall vibration and 1x with phase at both bearing housings simultaneously, shows speed, builds the FFT spectrum and the time waveform, balances in one and two planes using the influence coefficient method, calculates the tolerance by G class, splits the mass across fixed positions, calculates drilled-hole corrections, works from saved influence coefficients, and files the results into an archive for the report. There's a version without the case for building into machine tools and test stands.
The report here isn't a formality, it's a record-keeping tool. It preserves the baseline before the work, the mass and radius installed, the residual 1x, and the speed. A year later, exactly these records are what let you compare bearing running time before and after, instead of guessing.
If you balance regularly, it's cheaper to keep the instrument on hand. If it's a one-off or a disputed case, AXILINE's engineers will come out and do it on site, with no disassembly and no removing the rotor: they'll measure, assess the cause, balance, and leave a report. Balanset instruments are designed and manufactured by the same engineers who go out and balance with them themselves, so you get consulting support on choosing planes, placing sensors, and working through a run that didn't go as planned.
- Machine type, working speed, drive power, rotor mass if known.
- Bearing replacement history for this asset: dates, running time, what was found on the ones removed.
- Measurements, if you have them: overall vibration, 1x amplitude and phase at both bearing housings, spectrum, operating condition.
- Photos of the rotor from both sides and of the spots where sensors could be mounted.
Frequently asked questions
By how much will balancing extend bearing life on my machine?
We won't say without a calculation, and that's the honest answer. It takes the specific bearing's dynamic load rating C from the catalogue, an estimate of the actual equivalent dynamic load P, the speed, and the lubrication conditions. ISO 281 sets the order of magnitude of the effect: life depends on P raised to the power of 3 for ball bearings and 10/3 for roller bearings, so a 20% reduction in load nearly doubles the calculated life. But the share of the total load that comes from imbalance is different for every assembly, and a ready-made figure like 'doubles it' without your data is marketing, not a calculation.
Can mm/s be converted into force on the bearing?
No. A figure in mm/s is the structure's response to a force, and it depends on the stiffness, mass, and damping of the supports and the frame. The same force gives different mm/s readings on a rigid bedplate and a compliant one. Force is estimated through imbalance: the correction weight's mass, multiplied by the installation radius, gives the removed imbalance in g·mm, and multiplying that by the square of the angular velocity gives the force in newtons.
Vibration is in zone A. Is there any point in balancing more precisely for the sake of the bearings?
Zone A per the applicable part of ISO 20816 answers the question of whether the machine's condition is acceptable, not whether the rotor is balanced. On a rigid foundation, low mm/s readings coexist comfortably with a substantial rotating force. If the rotor is light and fast-running, there's a point to it, and you should assess the residual imbalance in g·mm/kg against the chosen G accuracy class. If the assembly is loaded mainly by process load, prioritize lubrication, installation, and shaft alignment instead.
Does balancing replace bearing replacement?
No. It doesn't smooth over a spall on the raceway or restore clearance. The logic runs the other way: the bearing gets replaced, and balancing removes the cause that made it fail prematurely, so the new one doesn't share its fate. If you replace the bearing without removing the load, the next one will last about as long.
How often should balancing be repeated to preserve bearing life?
By trend, not by calendar. Take a baseline right after balancing and repeat the measurement at the same operating condition and the same points. A twofold rise from the baseline is reason enough to come out with an instrument. Fans with buildup and blade wear can drift out of normal within weeks, while a motor after a good balancing job can hold steady for years.
Does the instrument calculate bearing life?
No, and no vibration analyzer does. The instrument measures vibration, extracts 1x with phase, and calculates correction masses and the tolerance by G class. Calculated life is worked out per ISO 281 from the bearing manufacturer's data and the actual loads. Vibration answers the question of what's happening and how urgent it is, not how many hours are left.
Related content
Vibration Measurement Units: mm/s, g, µm, and What RMS Means
Vibration is described by three quantities: displacement in µm (usually peak-to-peak), velocity in mm/s (usually RMS — root-mean-square value), and acceleration in m/s² or g. Displacement works at low frequencies and on shafts, velocity gives a universal condition assessment for housings over the 10–1000 Hz band, and acceleration shows up high frequencies, bearings and impacts. Converting between quantities is only possible for a single component at a known frequency, and the 1.41 factor between RMS and peak only holds for a sine wave. That's why a figure with no stated quantity, amplitude type, frequency band and measurement point simply has nothing to be compared against.
Savings from balancing without removal: how to calculate the full cost and avoid overpaying in downtime
On-site balancing wins not on the price of the work itself, but on what it doesn't include: disassembly, rigging, transport, reassembly, shaft alignment after reassembly, and calendar days of downtime. Calculate both options against the same boundary: from the moment the machine stops to the moment it's back within tolerance and running again. In the overwhelming majority of cases, the outcome comes down to two line items: your hourly downtime rate, and whether you have a standby unit. Removal stays cheaper wherever a weight physically can't be fitted on site, where the rotor is flexible, where the geometry is damaged, or where acceptance requires a report against a balance quality grade G.
On-site balancing of any rotor: the five conditions that make it work
Yes, we balance non-standard rotors and rotating assemblies on site, in their own bearings, under five conditions: there's somewhere to mount vibration sensors on the bearing housings; speed can be measured with an optical tachometer off a reflective mark; the machine can be safely started and stopped two or three times; there's somewhere on the rotor to add or remove mass; and rotation speed stays stable during the measurements. The type of machine doesn't matter. If even one condition isn't met, we say so before the visit and suggest another path.
Describe your equipment and the problem
We'll answer your questions, clarify the details, and let you know what's needed for an estimate and a visit.