Shaft Runout: How to Measure TIR with a Dial Indicator

Key takeaways

• TIR is the full sweep of the needle, so eccentricity is half the TIR: 0.08 mm TIR is 0.04 mm off the axis of rotation.
• Radial runout: a diameter not concentric with the axis of rotation. Axial or face runout: a face not square to it. Measure both.
• Measure at three or more axial planes. One plane cannot separate a bend, a taper and a cocked hub.
• Runout must be inside tolerance before alignment: a bent shaft carries its error into every alignment reading.
• Residual runout appears at 1x running speed, so a 1x problem that will not respond to balancing is usually runout or seating.

What runout actually is

Runout is how far a rotating surface moves against a fixed reference as the shaft turns one revolution. It does not measure size. It is a composite reading: the sweep combines the form error of the surface with its displacement from the axis of rotation, which is why the pattern of the sweep, not just its magnitude, tells you which fault you have.

• Radial runout: the stem sits on a diameter, perpendicular to the centreline, and moves because that diameter is not concentric with the axis. Hub seats, journals, seal surfaces.
• Axial or face runout: the stem sits on a face, parallel to the centreline, and moves because that face is not square to it. Hub faces, flanges, thrust collars.

The half-TIR rule, and why it gets missed

TIR, total indicator reading, or FIM, full indicator movement, is the difference between maximum and minimum needle position over one revolution.

If the maximum reading is +0.05 mm and the minimum is 0.03 mm below zero, the TIR is 0.08 mm. The centre of that surface is displaced from the axis of rotation by 0.04 mm, because as the shaft turns the high side swings toward the indicator and the low side swings away, so in one revolution the needle travels the offset once in each direction and the sweep is twice the offset.

This is the commonest arithmetic mistake in the trade: 0.08 mm TIR read against a 0.05 mm eccentricity limit condemns a serviceable shaft, and reversed it passes one twice as far out as allowed.

Most manuals state limits as TIR, but not all. Treat an unqualified "runout, max 0.05 mm" as TIR and note the assumption.

What runout separates, and how the plot reads

The check is diagnostic. Plot TIR at each plane against distance from a datum: the same sweep has different causes, and the pattern separates them.

• A bent shaft. An impact, an overhung load, a dropped rotor. TIR is near zero at the journals and rises smoothly to a maximum between them, high spot at one angular position throughout.
• A diameter machined off centre. Round and straight, but not concentric with the journals: constant TIR along that diameter, near zero on the journals.
• A cocked or badly seated component. A hub, sleeve or bearing pressed on square to nothing, or over rust, paint or swarf. Radial runout looks fine while hub face runout is large.
• A burr, nick or fretting on a seat. Localised: a single sharp step in the needle rather than a smooth sine sweep.
• Thermal bow. A rotor cooled unevenly after a hot shutdown reads bent, then straight hours later. Never condemn one after a hot stop.
• A measurement problem. TIR that changes with each rotation is no shaft fault. Check the base, the stem, the rotor's axial location.

How to measure properly

• Support the shaft the way it runs. Own bearings, V blocks under the journals, or between centres, never a support that reads a condition service never sees.
• Clean the surface. Stone down high spots, wipe off oil film, remove all burrs. A 0.02 mm burr is a 0.02 mm step in the sweep, because a local high spot adds its own height once per revolution instead of doubling the way an offset centre does, and it will still cost you an afternoon.
• Set the stem perpendicular to the surface, preloaded a quarter to a third of travel; at an angle it reads a cosine fraction.
• Mount rigidly on clean, flat, thick steel. If the bracket flexes, you are measuring the bracket.
• Zero at one angular position, mark the shaft, and rotate a full 360 degrees by hand in the normal direction, watching the needle throughout.
• Record maximum and minimum with their angular positions, then confirm the needle reads zero back at the mark; if not, suspect axial float or a loose base.
• Repeat at several axial planes: inboard bearing, midway, hub seat, seal area, free end.

Seating faults show as runout long before they show as noise, so this belongs in the acceptance step of any bearing installation.

Runout is not alignment, and it comes first

Alignment measures the relative position of two shafts; runout measures one shaft against its own axis. Every alignment method assumes the surface you measure from rotates concentrically, and a hub with 0.10 mm TIR adds or subtracts 0.05 mm on every reading, so you can chase shims for two hours and still have a machine that shakes. Record hub runout before the first coupling alignment reading.

Acceptance thinking

No universal number exists: tolerance comes from the manual and the application, tightening as speed rises and clearances close. Absent a manual figure, many plants work near 0.05 mm TIR on coupling hub radial and face runout at 1500 rpm, tightening toward 0.025 mm TIR at 3000 rpm and above, and the tighter figure for close clearance mechanical seals at any speed. Write it down, as with any acceptance check, including gearbox backlash measurement.

The vibration link

Residual runout produces a once per revolution forcing at 1x running speed, exactly where unbalance lives, so trim weights drop the amplitude a little and the 1x returns.

A 1x problem that does not respond to balancing is a runout, seating or alignment question. When you assess whether the residual level is acceptable, judge the broad-band RMS velocity against a severity framework such as ISO 10816-3 vibration severity, whose zones are defined for the overall broad-band level, 10 Hz to 1000 Hz for machines rated above 600 rpm and 2 Hz to 1000 Hz for rated speeds between 120 and 600 rpm, and not for a single order, and treat the 1x amplitude as the diagnostic component inside that overall figure rather than as the number the zones apply to.

What to record

• Asset, component, date, technician, support condition.
• Indicator resolution, base location, each plane's distance from a datum.
• Maximum, minimum, angular positions, TIR, eccentricity, the tolerance and its source.

In Fabrico, a runout check lives as a recurring PM task with a checklist, so readings and verdicts sit against the asset, visible when the next set is taken. A failed manual check can trigger a follow up task, and a machine stop from PLC or OEE data can become a work order, so a repeat 1x complaint becomes a per asset record an auditor can read: book a short demo.

A worked example with real numbers

A single stage centrifugal water pump runs at 2950 rpm. Its mechanical seal has failed twice in five months, and the vibration route shows 4.8 mm/s RMS at 1x on the inboard bearing, unchanged by a field balance.

The coupling is off, the shaft stays in its own bearings, and it is indicated at five planes measured along the shaft from the inboard journal, high spot 70 degrees from the keyway.

• Plane A, 10 mm, journal shoulder: max +0.005, min 0.005 below zero. TIR 0.010 mm, eccentricity 0.005 mm.
• Plane B, 55 mm, seal running surface: max +0.021, min 0.019 below zero. TIR 0.040 mm, eccentricity 0.020 mm.
• Plane C, 100 mm, mid span: max +0.036, min 0.034 below zero. TIR 0.070 mm, eccentricity 0.035 mm.
• Plane D, 145 mm, drive end seal running surface: max +0.024, min 0.022 below zero. TIR 0.046 mm, eccentricity 0.023 mm.
• Plane E, 190 mm, outboard journal: max +0.006, min 0.006 below zero. TIR 0.012 mm, eccentricity 0.006 mm.

The house tolerance for this machine class, a close clearance mechanical seal pump running just under 3000 rpm, is 0.025 mm TIR on both seal running surfaces, the plant applying its close clearance seal figure rather than its general purpose one. Plane B reads 0.040 mm TIR, which is 0.015 mm over the limit and 1.6 times the allowed value. Plane D reads 0.046 mm TIR, which is 0.021 mm over and 1.84 times the limit. Both fail.

Now read the pattern: near zero at both journals, peaking at 0.070 mm at mid span, falling again toward the outboard journal, a smooth curve rather than a step, one clock position throughout: a bent shaft.

The mid span bow is 0.035 mm of true eccentricity, and it explains the repeat 1x and the seal failures, because a seal running surface eccentric by 0.020 mm forces the seal's dynamic element to follow a 0.020 mm radius orbit, a 0.040 mm total excursion, once per revolution at 49 revolutions per second, flexing the secondary seal and lifting the faces apart on every turn. The shaft is replaced, not straightened.

Before release: indicate the new shaft at the same five planes, confirm 0.025 mm TIR or better at both seal running surfaces, with the journal and mid span planes inside the general purpose 0.05 mm TIR figure, check the refitted hub for radial and face runout, align only then, and recheck vibration at 24 hours against the 4.8 mm/s baseline, a sequence for the pump's preventive maintenance plan.

Common mistakes

• Treating TIR as eccentricity. Halving is not optional.
• Measuring at one plane. One number cannot separate a bend from a taper from an eccentric diameter, and the fault you assume decides the part.
• An angled indicator stem. It under reads by an amount nobody can quantify afterwards.
• Skipping the return to zero. That hides a loose base, a sticking stem or axial float.
• Indicating a hot rotor. A thermally bowed shaft reads bent, then straight once it equalises.
• Aligning first, checking runout later. The numbers look convincing, mean nothing, and the shims come out again.
• Recording only the verdict. Pass or fail with no readings, planes or tolerance cannot be compared with next year's check.

Frequently asked questions

Is TIR the same as eccentricity?

No. TIR is the full sweep of the needle over one revolution. Eccentricity is the displacement of the surface centre from the axis of rotation, and equals half the TIR: 0.08 mm TIR is 0.04 mm. Most manuals state limits as TIR, so confirm the convention.

Can I measure runout with the shaft in the machine?

Yes, and usually you should. The shaft in its own bearings gives the most representative reading, because that is the condition it runs in. Rotate by hand in the normal direction, drive isolated and locked off. V blocks suit shafts already out of the machine.

How many measurement planes are enough?

Three as a minimum, four or five if the shaft has several functional diameters: one near each bearing journal, one at mid span, and one at every seat that matters. Record each distance from a datum, because the shape across planes is the diagnosis.

My machine shows high 1x vibration. Is that runout or unbalance?

Both peak at 1x running speed, which is why they are confused. The discriminator is response to balancing: if trim weights bring the level down and it stays down, it was unbalance; if the gain is partial, indicate the shaft and the hub instead.

What runout tolerance should I use if I have no manual?

Set a documented house standard. Many plants work near 0.05 mm TIR on coupling hub radial and face runout around 1500 rpm, tightening toward 0.025 mm TIR at 3000 rpm and above, and tighter for close clearance seals. An unwritten tolerance drifts with whoever holds it.