Temper mill roll wear gets blamed for shape defects it does not cause. A single-stand temper mill has one job on shape: hold the loaded gap where the strip needs it. Strip waviness, quarter buckle and centre buckle are not separate problems with separate causes. They are the gap profile showing up in the strip.
This study measured both roll types across a full service cycle, then separated what wear actually does from what the ground crown does. The conclusion runs against common practice: temper mill roll wear is not the main shape variable, and shortening roll change intervals to chase it wastes grinding capacity.
What the gap says about temper mill roll wear
Under symmetrical conditions, the loaded gap decomposes into a fourth-order polynomial with constant, quadratic and quartic components. Two of those components map directly onto the waviness patterns that show up on the strip.
The quadratic component produces second-order waviness: double edge wave and centre wave. Most shape adjustment in production works on this quadratic crown, because it is the term the bending system controls most directly.
The quartic component produces fourth-order waviness: quarter wave and combined edge-centre wave. On thin strip, this is the term that produces the high-frequency defects a temper mill struggles with.
Both crowns convert quantitatively from the polynomial coefficients of the gap. Therefore the gap decomposition gives the shop floor a way to quantify how a roll profile change will translate into strip shape, and to predict a defect before it appears.
What temper mill roll wear actually changes
Across a complete service cycle, work roll wear was light and the distribution was even. Only the drive side wore slightly more than the operating side. The quadratic and quartic crown values before and after service moved within a narrow band.
That matters because it inverts the usual assumption. If wear were the dominant shape variable, crown would drift steadily through the cycle. Shape would then degrade from first coil to last. Plutôt, the gap profile barely shifted. Temper mill roll wear, in other words, is not the main cause of shape fluctuation on this mill.
Cependant, a word of caution belongs here. “Even and light” describes a properly run temper mill. A mill running excessive rolling force, poor lubrication or contaminated coolant will wear rolls differently. This conclusion does not transfer to it.
The real cause: ground crown against strip width
The problem sits upstream of the mill, in the roll shop. The base crown produced by grinding does not match the target crown for the strip widths being run, and therefore the mismatch widens as strip gets narrower.
Measured against target, narrow product came out with less than half the crown it needed. That single number explains most of the quarter buckle seen on narrow strip.
The quartic data makes the mechanism explicit. On an unused roll, the quartic crown reached more than 50% of the quadratic crown. A gap with that much fourth-order content will generate quarter wave regardless of how the bending system is set. Bending acts mainly on the quadratic term, while the quartic term rides straight through.
Par conséquent, the shop floor sees quarter buckle concentrated on narrow strip, and adjusts bending without effect. The bending force was never the problem.
Backup roll wear: 120 µm that barely moves the gap
Backup rolls show the opposite behaviour. Service life runs only 10 à 14 jours, yet local uneven wear is pronounced. The maximum diameter wear difference between upper and lower rolls reached 90 µm and 120 µm respectively, so loss across the body varies sharply.
That sounds severe, and the intuitive response is to grind mid-cycle. The measurement says otherwise.
A finite element model of the roll stack was built and run with measured new and worn backup roll profiles. The resulting gap was then calculated. The quadratic crown difference between new and worn backup rolls came out at only 1 à 6 µm, and the quartic crown showed almost no change at all.
Autrement dit, uneven backup wear changes the gap only indirectly, by altering how the work roll bends. Its grip on the overall gap crown is weak. Across a full service cycle, the original backup roll profile held the shape control function steady, and local wear did not produce sustained or large-scale shape fluctuation.
What to change in the roll shop
Three actions follow directly from the data.
Grind crown by strip width. Set a dedicated target crown for each finished strip width. For narrow product, raise the base quadratic crown and reduce the quartic share at the same time. This attacks quarter buckle at its source, which is the grinding template rather than the mill.
Stop shortening roll change intervals for wear. Because even wear does not shift the gap, there is no case for changing work rolls early on wear grounds alone. Rely on periodic profile measurement, and regrind only when the original ground contour drifts out of tolerance.
Inspect profile before the roll goes on. Use a displacement-sensor profile gauge to check diameter variation across the full body before mounting. This catches a bad grind before it produces a batch of defective strip, which is far cheaper than finding it in the finished coil.
Backup rolls take a lighter touch. Local wear does not justify extra mid-cycle grinding, and the existing 10 à 14 day cycle can stand. Bending force and rolling force baselines do not need to be re-tuned as backup wear progresses. The single control point that matters is the initial grind: get the new roll profile right, and service wear can be left out of the daily shape adjustment variables.
Operationally, this simplifies defect hunting. Read the quadratic and quartic crown data from the gap first. A quartic shift points to work roll profile, while a quadratic shift points to process parameters or bending. Two numbers separate the two families of cause, and the technician stops guessing.
The underlying rule holds across single-stand temper mills. Put the effort into the grinding template, and treat wear as a slow and even background. Then stop using bending force to compensate for a crown that was wrong from the start.




