Euskara
Zulu
Latinus
Cymraeg
தமிழ்
Slovak
Galician
AfrikaansRoller Runout on Film Lines: Causes, Measurement, Correction and Total Cost
Nov 18,2026
Roller Runout on Film Lines: Causes, Measurement, Correction and Total Cost
Quick Answer
Excessive runout is the geometric error that survives every control loop on a film line, because it repeats once per revolution and cannot be averaged out. It comes from journals, shell geometry, bearing fits, thermal distortion and surface build-up, and it is identified fastest from the repeat length it prints on the web. Correction follows a fixed sequence — journals, face, surface, crown, balance — and the sequence matters more than the individual operations, because each one references the one before it.
Key Takeaways
- TIR combines roundness, eccentricity, bearing movement and elastic deflection into one number.
- Repeat length equals πD; πD/2 and πD/3 indicate lobing rather than eccentricity.
- Film-line rolls are ground to 0.005–0.020 mm TIR against 0.05–0.10 mm as-turned.
- A 400 mm shell can read 0.012 mm at 20 °C and 0.04 mm at 200 °C; hot acceptance is not optional.
- Balancing must follow the final grinding pass, or the certificate describes a rotor that no longer exists.
- Regrinding removes 0.05–0.15 mm per cycle and costs 25–40% of a new roll.
- Plating and polishing reproduce geometry; grinding generates it.
- Runout is also a cost line: every unwarranted regrind and every resin trial it triggers is avoidable spend.
Summary Table: Runout Specification by Roll Duty
Roll duty | TIR target (in housing, hot) | Surface | Balance | Dominant defect |
Casting roll, capacitor or optical | ≤0.005–0.010 mm | Ra 0.005–0.02 µm | G2.5 / G1.0 | Quench-side gauge band |
Chill roll, BOPP and BOPET | 0.010–0.015 mm | Ra 0.02–0.05 µm | G2.5 | Contact and heat-transfer variation |
MDO preheat roll | 0.010–0.020 mm | Ra 0.4–0.8 µm | G2.5 | Thermal and draw-gap variation |
MDO stretching roll | 0.010–0.020 mm | Ra 0.1–0.3 µm | G2.5 | Draw-ratio stripe |
Coating or metering roll | 0.005–0.010 mm | Ra 0.02–0.1 µm | G2.5 | Wet film thickness variation |
Idler or guide roll | 0.020–0.050 mm | Ra 0.4–1.6 µm | G6.3 | Tracking disturbance |
Winder drum | 0.010–0.020 mm | Ra 0.2–0.4 µm | G2.5 | Wound-in bands, starring |
Knowledge Graph: Previous, Current, Next
Previous: melting, filtration and die delivery. The die sets the incoming gauge profile, and no roll can correct a profile it never received — which is why gauge and geometry faults must be separated before a roll is condemned.
Current: roll geometric accuracy in the Film Industry — runout, roundness, cylindricity, crown and texture, all referenced to the journals and all measured at temperature.
Next: quenching on the Casting Roll, drawing between MDO Stretching Roll pairs, cooling on the PRS Cooling Roll and winding. The same 0.02 mm becomes a gauge band upstream and a winding defect downstream.
Market Context
Three developments in the past week sharpen the case for tighter geometric control.
Ultra-thin capacitor film (Xiamen, November 2026). A second BOPP electronic-film line has started producing 1.9 µm, 2.2 µm and 2.4 µm capacitor grades on a simultaneous biaxial line, with the producer describing tens of nanometres of thickness deviation as significant at these gauges. For equipment engineers, that statement is a runout specification: a roll running out by 20 µm cannot deliver nanometre gauge regardless of what the die does.
700 m/min packaging lines (Hubei, November 2026). A 30 billion RMB programme covering four BOPP lines over two years, including matte, pre-coated and high-end packaging grades, with the fastest machines at 700 m/min. Excitation from residual error rises with the square of rotational speed, so a tolerance written at 400 m/min will not hold at 700.
Wide simultaneous lines with online coating (Cangnan, November 2026). A 120,000 t/yr BOPP and electronic-film project is in main construction, pairing a simultaneous stretching line with online coating and multi-layer co-extrusion. Multi-layer structures concentrate risk in the casting section, where a single out-of-round chill roll disturbs the skin layer before the core has solidified.
Small cast-film increments (Guangzhou, November 2026). An approval adds 1,200 t/yr each of TPU and EVA on cast lines. On small lines, runout is routinely misattributed to resin because no roll register exists to check a repeat against.

Two film-line rolls supported on inspection stands with journals exposed for a TIR check
ALT: Two film-line rolls supported on inspection stands with journals exposed for a TIR check
The Five Sources, and How to Separate Them
Source | Indicator signature | Diagnostic step | Correction |
Journal out of round or tapered | TIR differs between stations | Micrometer at two planes | Straighten, re-grind or build up |
Shell eccentric to journals | Constant TIR, one high point | Face and journal in one setup | Grind face to journals |
Shell lobed | Two or three peaks per turn | Slow rotation, marked datum | Grind with corrected support |
Bearing seat or housing fit | TIR changes with load | Clearance and bore measurement | Re-metalise seat, correct bore |
Thermal distortion | TIR grows after warm-up | Cold and hot at same angle | Fix flow symmetry |
Build-up is the sixth source and the cheapest to fix: plate-out, dried coolant solids, polymer film and corrosion add local thickness. A 0.015 mm deposit reads as 0.015 mm of TIR and looks exactly like a machining fault until the surface is cleaned and re-measured.
The separation logic is simple but routinely skipped. If TIR is constant along the face with one high point, the shell is eccentric and the journals are the suspect — because the face was machined on a centre that the journals do not share. If TIR changes monotonically end to end, the roll is tapered or the axis is misaligned. If two or three peaks appear per revolution, the shell is lobed and the previous grinding setup is the suspect. If TIR only appears after warm-up, the channels are the suspect.
Section by Section: Where It Bites
Casting. The sheet is pinned and quenched while still molten. A runout error here changes contact and air-gap geometry once per revolution, which modulates the quench rate and produces a gauge band of the order of a micron on a 25 µm sheet — enough to fail a ±2% gauge tolerance. This is why casting rolls are ground at the tight end of the band and why surface finish is specified separately from geometry.
MDO. The heating side of the same section is covered in the notes on MDO Preheating Rolls for BOPP and BOPET Lines. Runout on a preheat or stretching roll changes the gap between paired rolls, and therefore the local draw ratio. The result is a stripe rather than a uniform band, and it is often mistaken for a temperature-zone problem. Verifying the draw-gap with feeler or impression material at four positions across the face separates geometry from thermal causes in minutes.
Cooling and PRS. The quench duty itself is described in the guide to PRS Cooling Rolls for BOPP, BOPET and BOPA Lines. Runout here does two things: it varies contact, which varies heat transfer, and it varies tension, which propagates to the winder. A cooling-section error frequently surfaces as a winding defect metres downstream, which is why investigations that start at the winder rarely find the cause.
Converting. Among the Film Converting Rolls the metering roll is the most sensitive. In coating and lamination, a metering roll error converts directly into wet-film thickness variation, because the gap is the film thickness. The relevant tolerance is the one the coating weight can accept, not the one the roll drawing inherited.

Roll end face with journal and bearing seat visible, the datum for shell concentricity
ALT: Roll end face with journal and bearing seat visible, the datum for shell concentricity
Signature Analysis
Because the web advances one circumference per revolution, a fixed angular feature contacts the web every πD millimetres.
Diameter | Repeat | Diameter | Repeat |
250 mm | 785 mm | 500 mm | 1,571 mm |
300 mm | 942 mm | 600 mm | 1,885 mm |
400 mm | 1,257 mm | 800 mm | 2,513 mm |
450 mm | 1,414 mm | 1,000 mm | 3,142 mm |
Two corrections are essential. Slip: a roll running 2% fast shows a true 1,414 mm repeat at 1,386 mm or 1,443 mm, so a single measurement may be wrongly dismissed — count twenty repeats. Harmonics: an oval shell prints at πD/2 and a three-lobe shell at πD/3. A repeat at a clean fraction of the circumference is a geometry diagnosis, and the repair changes from polishing to grinding.
The practical tool is a roll register sorted by circumference. It converts an open-ended search into a lookup and is the single most useful document a film plant can maintain.
Measurement Protocol
Four stations. Drive end, quarter point, centre, far end. Monotonic trend means taper; single peak means local defect; multiple peaks mean lobing.
Stated datum. Between centres isolates geometry. In the housing includes clearance, bore error and frame deflection. The in-machine reading is what the web sees, and a large difference between the two is itself a diagnosis.
Recorded temperature. A 400 mm steel shell with asymmetric channels reads 0.012 mm at 20 °C and 0.04 mm at 200 °C. Shells above 300 mm need at least 45 minutes at setpoint. An unlabelled reading cannot be compared with anything, including a previous measurement of the same roll.
Instrument and setup. A lever-type indicator with 1 µm graduations is sufficient; the limiting factor is access and rigidity, not resolution. Mount independently of the roll frame so frame vibration is not measured as geometry. Take two readings per station with the roll returned to the same angle; a spread above 2 µm means the setup cannot judge a 10 µm tolerance.
How Tolerances Are Actually Set
The tolerance on a roll drawing should be derived from the defect the customer rejects, not inherited from a previous drawing. Work backwards: a 2 µm gauge band is rejectable, the roll that produces it at the casting position runs out by 0.008 mm, therefore the casting roll is specified at 0.005–0.008 mm and everything upstream — journal concentricity, bearing fit, housing bore — is tightened to leave margin.
This is why copying tolerances between applications fails. A BOPP packaging line can accept 0.03 mm on an idler and never see it; a capacitor film line cannot accept 0.01 mm on a casting roll. The number is meaningless without the duty, the speed and the product.
Balance: A Separate Certificate
ISO 1940 grade G2.5 permits about 1.5 g·mm of residual unbalance per kilogram of rotor mass. For a 2,000 kg roll that is 3,000 g·mm, and the force it generates rises with the square of speed — sixteen times greater at 1,200 rpm than at 300 rpm. A balance certificate without a quoted speed is therefore not a specification.
Balancing must also come last. Grinding removes mass, so a roll balanced before finish grinding is certified in a condition that no longer exists. In practice this is the most common procedural error we see in incoming inspection packs: a G2.5 report dated before the final grind.

Large mirror-finish roll being ground to correct shell geometry
ALT: Large mirror-finish roll being ground to correct shell geometry
The Correction Sequence
Journals. Straighten if bent; strip and re-grind if worn; build up and re-machine if undersize. Every later operation references the journals, so error here is multiplied.
Face. Grind concentric to corrected journals, holding roundness and cylindricity. If lobing is present, fix work support and dress the wheel before finish passes — lobing is a setup fault far more often than a wheel fault.
Surface. Strip and re-plate where remaining chrome in a loaded zone is below about 0.03 mm, typically 0.05–0.15 mm at 800–1000 HV, then finish-grind the chrome. Final tolerance is generated on the chrome.
Crown. Verify with impression film or lead foil under production nip load: narrow contact at the edges means insufficient crown or excessive deflection; narrow at the centre means over-crown. A 0.03 mm crown error across a 2,500 mm face is enough to drop local nip pressure below the bonding threshold in many processes.
Balance. Two-plane, at maximum operating speed, after the final grind.

Technician re-cutting a large roll journal in a lathe before grinding the face
ALT: Technician re-cutting a large roll journal in a lathe before grinding the face
Cost of Getting It Wrong
Cost item | Typical magnitude | Comment |
Unwarranted regrind | 25–40% of a new roll | Removes 0.05–0.15 mm of life |
Resin and die trials chasing a band | Days of production | The band is geometry, not formulation |
Scrap from out-of-tolerance gauge | Lot-dependent | Highest on capacitor and optical film |
Bearing replacement from vibration | Set per event | Imbalance and runout both accelerate wear |
Line stop for roll change | Hours per event | Two to three regrind cycles per life |
Premature replacement | One roll | Caused by grinding a good roll to chase a fit |
The largest single avoidable cost is not the roll. It is the production time spent changing everything except the roll, because no repeat-length check was made in the first two hours.
Failure Analysis
Symptom | Measurement | Cause | Action |
Band at πD | One high point, constant | Shell eccentric | Grind to journals |
Band at πD/2 or πD/3 | Two or three peaks | Lobed shell | Grind, correct support |
Worse with speed | Vibration, TIR acceptable | Imbalance or clearance | Balance at speed |
Only when hot | TIR grows after soak | Thermal distortion | Fix flow symmetry |
Local mark | Step at one angle | Deposit or ding | Clean, inspect |
Slow drift | Rising clearance | Seat wear | Re-metalise seats |
Edge wrinkles | Good TIR, poor alignment | Non-parallel rolls | Realign frame |
Maintenance Programme
Maintain a register per roll: diameter, face length, chrome thickness history, balance grade with speed, and every TIR reading with its temperature and datum. Measure critical rolls annually, after every bearing change, and after any event that could bend a shaft. Protect journals with sleeves in transit — a single ding becomes an eccentric face at the next regrind. Treat chrome as a consumable with a minimum thickness and plan re-plating one cycle ahead.
Case Studies
All three examples below follow the same diagnostic order, and comparable projects are collected in the Case Studies library.
A Southeast Asian BOPP producer chased a 1,414 mm band for three weeks across resin lots, die seals and cooling setpoints. Measured in the housing at temperature, the Ø450 chill roll read 0.07 mm with a single high point and no axial trend. Journals were re-ground, the face ground to them, the roll balanced after grinding; final TIR was 0.012 mm hot and the band disappeared with no process change.
A South American wide-line producer found a three-lobe pattern on a 600 mm roll at a third of the circumference. Regrinding alone would have hidden it; the root cause was inadequate work support during the previous regrind, so the setup was corrected as part of the repair. That project's wider roll programme is documented in our 7.4 m BOPP case, where specifying only the tolerances the product needed cut procurement cost by more than 60%.
A North American converter running an 8.7 m line replaced a chill roll set after a repeating band traced to one roll whose TIR had grown to 0.09 mm. The replacement set was specified at 0.012 mm measured hot, with the balance report quoted at line speed rather than at shop speed. The band disappeared and no further resin trials were run.

Mirror-polished roll surface after finish grinding
ALT: Mirror-polished roll surface after finish grinding
Procurement and Acceptance
Specify TIR as a value with a method: datum named, temperature stated, four stations recorded, angular position of the high point marked. For heated rolls, require cold and hot readings. Require the balance report to quote grade and speed, chrome thickness with a minimum, and roundness and cylindricity reported separately from TIR.
At HUATAO Industrial Rolls, runout is treated as one of several independently certified characteristics rather than a single pass/fail number. Bi-Oriented Film Rolls and Casting Film Rolls are ground on journal datums, finish-ground after plating and balanced after the last grinding pass, with hot readings recorded for heated shells. The BOPP Chill Roll range and MDO Preheating Roll range follow the same sequence, and worn rolls are re-machined to the original datum rather than to the worn face.
For related diagnostics on adjacent equipment, the Industry Blog covers tension, tracking and heat-transfer faults that mimic runout.
FAQ
What is a normal TIR for a film-line chill roll?
0.010–0.015 mm in the housing at operating temperature for BOPP and BOPET; tighter for capacitor and optical grades.
Why does TIR read higher in the machine than on the bench?
Bearing clearance, housing bore error and frame deflection are included in the in-machine reading.
Can a roll be reground without removal?
Only surface deposits and light damage can be addressed in place. Geometry requires a machine tool with journals as datum.
How much life does a regrind consume?
0.05–0.15 mm of diameter per cycle, with two to three cycles typical over a 10–15 year life.
Is polishing ever a valid fix for a band?
Only when the cause is surface texture or a deposit. It cannot change geometry.
What balance grade should be specified?
G2.5 as standard, G1.0 for large and fast shells, always with the speed stated.
Why do bands appear only at high line speed?
Excitation rises with the square of speed and the web has less time to conform to a short-wavelength gap change.
Does a new roll guarantee no band?
Only if it was specified and accepted the way the web measures it — in the housing, hot, at four stations.
How long should a shell soak before a hot reading?
At least 45 minutes at setpoint for shells above roughly 300 mm.
When is replacement better than repair?
When journals are at minimum diameter and chrome or wall thickness cannot absorb another cycle.
Conclusion
Runout is diagnosable in an afternoon and correctable in a sequence. The repeat length identifies the roll, the hot in-housing reading characterises it, and the datum order — journals, face, surface, crown, balance — determines whether the repair holds. Everything else, including the resin trials and the extra grinds, is the cost of skipping those two steps.
Hashtags
#RollerRunout #TIR #FilmProduction #BOPP #BOPET #ChillRoll #PrecisionGrinding #RollerBalancing #WebHandling #Manufacturing
Contact Us
Jenny Liu
Email: sale05@huataogroup.com
Website: https://www.huataomulti-indusrolls.com/
Phone / WhatsApp: +86 177 4378 9775
Contact Us
E-mail:
team3@huataogroup.com
Phone/WhatsApp:
+86-13331385676
Address:
NO.298 Zhonghua North Street,Shijiazhuang city, China.