Barrel-shaped rolls bring ultra-thin metal foils closer to perfect uniformity

Rolling a metal sheet down to foil thickness, below 200 micrometers, is one of the most demanding operations in cold rolling, and a new analytical study suggests that a surprisingly simple machine can do it with remarkable precision. Researchers have shown that a basic two-high rolling mill, the simplest roll configuration in metal forming, can be tuned through geometric optimization of its rolls to produce aluminum foil with a thickness variation of only about 1.3 micrometers across its width. The key was replacing perfectly cylindrical rolls with barrel-shaped, cambered rolls whose curvature was calculated iteratively so that the rolls bend and flatten under load into an almost perfectly flat contact surface.

The problem the study addresses is fundamental to foil manufacturing. When a thin metal strip passes between two counter-rotating rolls, the contact pressure in the deformation zone does two things to the rolls themselves: it makes them bend laterally, and it locally flattens their surfaces where they touch the foil. Both effects widen the effective roll gap, and because neither effect is uniform across the width of the foil, the emerging sheet ends up thicker in some places than others. In conventional practice, this non-uniformity is fought with backup rolls, multi-stage rolling, or sophisticated and costly thickness-control systems. The new work takes a different route: instead of adding hardware, it reshapes the work rolls themselves so that their elastic deformation under load is pre-compensated by design.

The analytical framework builds on a long line of rolling theory. Classic models by von Kármán, Orowan, and Bland and Ford, based on circular contact arcs and Hitchcock’s formula for the equivalent flattened roll radius, work well for thick sheets but break down for ultra-thin foils, where elastic roll deformation can no longer be neglected. Later contributions, including Fleck and Johnson’s foil-rolling theory that divides the contact area into five zones from elastic entry to elastic exit, and more recent roll-deformation models based on influence functions and elastic half-plane theory, progressively addressed these shortcomings. The present study extends a previously verified analytical model, which had shown less than 1 percent deviation from finite-element simulations for cylindrical-roll foil rolling, to the case of cambered-roll geometry and its influence on the transverse thickness distribution.

The mathematical machinery is built on a half-model of the roll-foil system, exploiting symmetry about the foil’s mid-plane. Roll deflection is treated as a beam problem: the roll, which has a larger radius in the rolling zone and a smaller shaft radius toward the bearings, is analyzed with the discontinuity-function method under symmetric loading, with boundary conditions of zero slope at the roll center, zero deflection at the bearings, and continuity of slope and deflection at the radius transition. The distributed contact pressure from the foil is then integrated to obtain the deflection field. Local flattening is computed from the theory of elasticity, and the flattened roll radius follows Hitchcock’s relation, which links it to the contact pressure and the local reduction in foil thickness. The contact pressure itself depends on the contact length and the yield stress of the foil, creating a mutually dependent set of quantities that must be solved iteratively until the pressure, contact length, and deformation fields become consistent with one another.

The case study rolls Al1050 aluminum foil, with a room-temperature yield stress of 110 megapascals, between tungsten carbide rolls with an elastic modulus of 500 gigapascals and a Poisson’s ratio of 0.25. The objective is a 36 percent thickness reduction, from 100 micrometers down to a target of 64 micrometers, on a foil 40 millimeters wide. The model was first verified against published results obtained under the same rolling conditions, for both cylindrical rolls and cambered rolls with a curvature radius of 300 meters, and showed good agreement in predicting the thickness distribution relative to the foil edge.

The baseline calculation with cylindrical rolls and a roll gap set exactly to the target thickness reveals why simple mills struggle. Starting from an assumed uniform contact pressure of 815 megapascals, the iterative scheme converged after seven cycles to a pressure profile ranging from 692 megapascals at the foil center to 740 megapascals at the edges. The foil-roll contact length, initially uniform at 0.795 millimeters, converged to between 0.675 millimeters at the center and 0.724 millimeters at the edge, because higher edge pressures produce more local flattening. Roll bending deflection peaked at 3.62 micrometers at the center, while flattening stayed near 2.84 micrometers across the central region before dropping to 1.82 micrometers at the edges. The combined result: the rolled foil came out at roughly 76.9 micrometers at the center and 72.0 micrometers at the edges, a variation of about 7 percent, with even the thinnest region sitting 8 micrometers above the 64-micrometer target.

The optimization then reshapes the rolls. A two-parameter function, combining an adjusted edge roll gap and a circular camber profile, is fitted to the discrepancy between the target thickness and the elastically deformed roll surface. The fitting yielded initial values of a curvature radius of 98 meters and an edge roll gap of 55 micrometers, and the analysis was repeated with the updated geometry. Because changing the gap and curvature alters the pressure distribution, which in turn changes the deformation, the procedure loops until convergence, defined as changes below 1 micrometer in the gap and below 1 meter in the curvature radius between successive iterations. After four complete rolling analyses, the optimal parameters converged to a curvature radius of 59 meters and an edge roll gap of 57 micrometers, with a fifth iteration confirming that the convergence criterion had been met.

The mechanism behind the improvement is intuitive once the numbers are laid out. Decreasing the curvature radius increases the convexity of the rolls, which compensates for their natural tendency to bend more at the center under load, where the foil would otherwise emerge thicker. The optimized configuration produces the most uniform contact pressure of all geometries examined, ranging from 816 megapascals at the center to 827 megapascals at the edges, a difference of only 11 megapascals. The contact length becomes nearly uniform as well, varying from 0.794 millimeters at the center to 0.805 millimeters at the edge, compared with a 0.05-millimeter spread for cylindrical rolls. Interestingly, the roll deflection itself grows larger under the optimized geometry, from 5.00 micrometers at the center to 2.70 micrometers at the edge, and total elastic deformation reaches 8.80 micrometers at the center, but this is precisely the point: the added convexity is sized to offset the enlarged deformation, so the effective roll gap under load becomes nearly flat.

The final thickness profile demonstrates the payoff. With the optimized rolls, the foil thickness varies from a maximum of 64.30 micrometers to a minimum of 63.04 micrometers, a non-uniformity of 1.26 micrometers, down from 4.92 micrometers with cylindrical rolls. At the foil center, the thickness exceeds the target by only 0.02 micrometers, and across most of the width the deviation from the target stays within about 0.30 micrometers. Excluding the outermost 5 percent of the foil width, the deviation drops below 0.55 micrometers, and the maximum deviation from the target is limited to 1.0 micrometer, roughly 3 percent of the total thickness reduction applied. The residual edge non-uniformity is attributed to the inability of a circular roll profile to perfectly compensate the deformation distribution, and to boundary effects at the free foil edges, where the material can deform more freely and undergo additional local thinning.

Perhaps the most striking comparison comes against a more complex machine. The optimized two-high results were benchmarked against published analytical results for a four-high rolling mill, equipped with two cambered work rolls of 150-meter curvature radius and two backup rolls, rolling the same aluminum foil from 100 micrometers. That four-high configuration produced an estimated thickness non-uniformity of about 3 micrometers and a maximum deviation from target of 2.7 micrometers, whereas the optimized two-high mill achieves 1.26 micrometers and 1.0 micrometer respectively, an improvement factor of roughly 2.3 in uniformity. The edge-thickness gradient is also markedly milder in the optimized case. As a final practical measure, the study notes that trimming 1 millimeter from each edge of the 40-millimeter-wide foil leaves a 38-millimeter strip averaging 64 micrometers in thickness with a widthwise variation below 0.3 micrometers. The authors present the approach as a practical and computationally efficient framework for foil-rolling design, one that achieves, in a simple two-high configuration without backup rolls, thickness accuracy comparable to or better than that predicted for more complex four-high systems under the conditions considered.

Subject of Research: Geometric optimization of cambered work rolls in a two-high mill to improve thickness uniformity in foil rolling

Article Title: Optimization of a two-high cambered mill for the foil rolling

Article References: Mirabadi, Y. A., & Alizadeh Kaklar, J. (2026). Optimization of a two-high cambered mill for the foil rolling. Heliyon, 12(16), Article e45623. https://doi.org/10.1016/j.heliyon.2026.e45623

Image Credits: AI Generated

DOI: 10.1016/j.heliyon.2026.e45623

Keywords: foil rolling, two-high rolling mill, cambered rolls, roll deflection, roll flattening, thickness uniformity, cold rolling, aluminum foil, contact pressure, Hitchcock roll-flattening, iterative optimization, metal forming