Mudanças na Lacuna do Rolo do Calendário: O que o acoplamento de tração deve acomodar?

A calender is unusual among driven machines in that its working geometry changes during normal operation. The roll gap is adjustable, rolls move relative to each other on purpose, and some of them can be opened wide for threading and closed down for production. Every one of those positions is a position the drive train has to work in.

Most drive train specifications are written for one alignment condition, usually the one the machine sits in when it is assembled and checked. A calender has no single alignment condition. It has a range, and the range is bounded by the positions no one takes a measurement in, because they are transient, or awkward, or simply not where the machine was set up.

The question worth asking before a calender drive is specified or changed is not what the coupling can compensate. It is what the coupling is being asked to compensate at the worst roll position, and whether the answer is inside the coupling's rated capability.

What a Coupling Can and Cannot Absorb

Start with the point that ends arguments quickly. A flexible coupling's ability to accommodate parallel offset, angular misalignment, and axial displacement is a property of that coupling's design, and within a design, of its tolerances. There is no general figure for "how much misalignment a flexible coupling takes," because the answer depends entirely on which coupling.

This is stated directly in published technical guidance from laser alignment equipment manufacturers, which goes on to describe what happens when the limit is approached. Once a coupling's permissible operating range is used up, secondary vibration and increased restoring forces appear even while the misalignment is technically still inside the limit. Those forces act on the bearings and seals of both machines, and in extreme cases on the shafts themselves.

The practical consequence is that staying inside the rating is a statement about whether the coupling can operate, not about how long it will. A misaligned coupling generates restoring forces continuously for as long as it turns, and those forces grow with the displacement rather than staying at a fixed level. They have to go somewhere, and where they go is into the bearings and seals on both sides of it. How much life that costs is a question for the coupling manufacturer's own data for that coupling; what the geometry can say is that the forces are present and that they scale with the displacement.

The forms of misalignment are worth naming precisely, because calender drives produce all of them. Parallel offset means the two shaft axes are displaced laterally but remain parallel. Angular misalignment means the axes meet at an angle. Axial displacement means the shaft ends move toward or away from each other. Real installations combine all three, in proportions that change as the machine moves.

Misalignment never announces itself as misalignment. It arrives as bearing and seal distress, coupling wear, leaking seals, loosened foundation bolts and a vibration reading that keeps creeping up. So it gets repaired component by component for years, without the geometry ever being measured. On a calender that progression runs faster, because the displacement at the coupling is not a fixed error but a variable that moves with the roll gap. A coupling that is comfortable at the production setting can sit at the edge of its range at the setting used to thread the machine.

Why a Calender Produces Misalignment on Purpose

On most machines, misalignment is an error. On a calender, some of it is designed in.

The rolls are not fixed in position. A calender's whole function is to pass material through a controlled gap between rotating rolls, and that gap is set by moving rolls relative to each other. Depending on the machine type, this may mean one roll adjusts against a fixed roll, several rolls adjust, or the entire roll stack can be opened. The roll arrangements themselves vary widely: two-roll, three-roll, four-roll, and five-roll machines exist, in L, inverted-L, F, Z, and S configurations, and each puts the drive connections in different places with different geometry.

How the rolls are driven is a machine-specific matter. Some calenders drive every roll from its own motor or its own gearbox output, some drive several rolls from a common line shaft with fixed gearing between them, and some combine the two. What the speeds are, and whether each roll can be set independently, is a question for the machine's drawings rather than a general property of calenders.

Where speeds do differ, they usually follow a profile across the stack rather than a single value: the down-stack rolls commonly run progressively faster to maintain a slight tension or over-speed in the sheet as it passes from one nip to the next, with the exact ratios set by the material and the product. The profile is not a universal set of numbers. It is written for a specific compound and a specific product, and it is one of the first things a supplier of a given machine will specify and a general article cannot. A drive train has to be able to hold whatever profile that machine requires, including across the last roll, the take-off, the cooling, and the winding.

So a calender drive has to deliver several different output speeds at once, to rolls that move relative to one another, through couplings whose misalignment changes as they move, while transmitting the torque needed to shear the material in the nip.

That combination is why the drive train on a calender is more sensitive to geometry than the drive train on a machine with fixed roll centers.

The Intermediate Shaft Is Not an Ideal Force Transmitter

Where a cardan shaft or a double-jointed intermediate shaft connects a gearbox to a roll, there is a piece of mechanical behavior worth understanding before specifying.

A single universal joint produces non-uniform output velocity: as the joint angle increases, the output shaft speed oscillates twice per input revolution, with the amplitude growing with the joint angle. A double universal joint arranged correctly, with the two joint angles equal and the yokes in the correct phase, cancels that oscillation and produces a uniform output.

Real installations rarely achieve perfect equality, because the geometry changes as the rolls move. An intermediate shaft between two universal joints therefore carries bending moments, arising from joint friction and from the velocity fluctuations the joint angles produce, and those effects break the ideal linear relationship between input and output.

The bending moments are the part that matters for the gearbox. They do not stay in the shaft. They are reacted at the shaft ends, which means they load the gearbox output shaft and its bearings, and the roll end bearing, with forces that were not part of the torque calculation. A gearbox sized for torque and installed with a cardan shaft at a large joint angle is being asked to carry a radial and moment load nobody specified.

The mechanism is worth taking seriously because the load path is easy to underestimate. A joint angle exists to be accommodated, and everything that accommodates it produces a reaction somewhere, which means the gearbox output bearing and the roll end bearing carry forces that no torque figure describes. The larger the joint angle through the adjustment range, the larger those reactions, and the less of the bearing capacity is left for the duty it was sized against.

If a calender drive uses cardan shafts, the joint angles at the extreme roll positions belong in the specification. If the machine uses a flexible coupling, the equivalent parameters are the offset and angular displacement at those same positions.

Checking the Worst Position, Not the Setup Position

The productive exercise is a position-by-position check across the full roll gap range, and it takes less time than it sounds.

Então, for each coupling, compare the required compensation against the coupling's rated capacity for parallel offset, angular misalignment, and axial displacement separately, and against the combined rating, which is usually more restrictive than any single axis.

The gearbox side of that comparison is where a heavy-duty gearbox for calendering machines enters the specification, and the reason it enters here rather than at the torque calculation is that the output bearing arrangement and the shaft end geometry are what the coupling acts on.

The table lists seven positions, and the practical question is where to start measuring. The worst position is not automatically the widest gap. It depends on where the roll axes actually sit through the adjustment range, on the length of the intermediate shaft, on how thermal movement and the adjustment vector combine, and on which deviations add in the same direction at the same setting. On some machines that lands at the threading gap, on others at a single-sided open position used for crown work, and on others somewhere in between. The way to find it is to plot offset and angle against gap position from the machine's own drawings, with the shaft length and the adjustment directions on them, and take the setting where the combined displacement peaks. The production gap is the reference case, not the governing one, and it is the position where a problem is least likely to be found.

Two entries on that list are the ones that produce surprises.

A reground roll changes the geometry, but not the way it first appears. Grinding does not move the roll's axis. The axis is held by the bearings and their housings, and if those are not touched, the axis does not move. What changes is the position of the roll surface, which recedes inward by half the diameter reduction at every point around the circumference. On a 400 mm roll ground down to 392 mm, the surface sits 4 mm closer to its own centerline.

Where the axis does move is afterwards, and by an amount the adjustment mechanism sets rather than the grinder. The gap between two rolls is the center distance minus the two radii. Losing 4 mm of radius would open the gap by 4 mm, so an operator restoring the same gap has to bring the rolls 4 mm closer, and that adjustment moves the housing and with it the axis. The coupling at the drive end sees the result of that adjustment, not the result of the regrind, and the two are only equal if the machine is being set back to the same gap. A calender re-set to a different gap, or a roll changed for a spare of a different diameter, moves the drive end by whatever the new setting requires. A coupling that was comfortably inside its angular rating at the old setting can be outside it at the new one, and nothing about the drive was touched to cause it.

Thermal growth is not uniform. Rolls run hot; housings and frames run less hot; the gearbox runs at its own temperature. Each element grows by its own amount in its own direction, and the resulting displacement at the coupling is a vector sum that is not predictable from any single component's expansion. A machine aligned cold and left at that is a machine whose alignment at production temperature is unknown.

It helps to reduce the geometry to one number, because the coupling only sees the distance between two shaft ends. Take a roll driven through a cardan shaft from a gearbox output, and suppose for the sake of the arithmetic that it moves 40 mm toward the gearbox to close the gap; the actual stroke comes from the adjustment mechanism, not from this figure. The coupling's joint angle changes by an amount that depends on the shaft length, and the rate of change is not constant through the travel: the same 40 mm early in the range is a small angle change, and late in the range it can be a larger one. The offset at the roll end also changes, because the roll axis has moved and the gearbox shaft has not. Neither figure is separately useful. What the coupling rating is compared against is the combined displacement at the most extreme position, and that position is set by the adjustment stroke, not by the production gap.

The regrind case has a second layer that is easy to miss, and it is about which surface the machine treats as its datum. Take the same 400 mm roll ground to 392 mm, with the bearing housing left where it was. The surface has come in 4 mm; the axis has not moved. If the machine sets its gap from the roll surface, the gap has opened by 4 mm and the operator closes it by moving the housing 4 mm, which moves the axis 4 mm and moves the drive end with it. If the machine sets its gap from the roll center, the gap is whatever the housing positions say it is, and the operator sees a product that has changed thickness with no adjustment having been made. Same 4 mm of material, two completely different maintenance conversations, and which one you get depends on the drawing convention rather than on the machine.

The Bearing Load That Follows

Because misalignment generates restoring forces, and because cardan shafts generate bending moments, the gearbox output bearings are the components that pay for a geometry that has moved out of range.

A calender gearbox specification that considers only torque will produce a unit whose bearings are adequate for the torque and marginal for the radial and moment loads from the coupling. The symptom is bearing failure at the output shaft, which gets investigated as a bearing problem, then as a lubrication problem, and eventually, if anyone measures the alignment across the roll gap range, as a geometry problem.

The measurement that resolves it is a shaft alignment check at each roll position, not just at the setup position, with the coupling disconnected from the load so that the true shaft positions can be read. That is a morning's work on a stopped machine and it directly bounds the problem.

What the Drawing Stage Can Rule Out

Most of the bad outcomes on a calender drive are decided before the gearbox is ordered, and they are decided by omission rather than by error. Nobody chooses a coupling that cannot take the angle at the threading position. What happens is that nobody checks the threading position, because the drawing shows the production setting and the production setting is where the machine spends its shift.

Three questions at the drawing stage eliminate most of the candidates.

What is the displacement at each end of the adjustment range, and at the single-sided open position used for crown adjustment? Those are the candidate extremes, and which one governs has to come from the geometry rather than from an assumption about width. If the answer is not on the drawing, it is not in the specification, however complete the specification looks. This one goes first because it is the only one of the three that can eliminate a candidate outright: a coupling that cannot take the angle at the governing position is not a candidate, whatever else it can do.

What changes when a roll is reground? Every roll in a calender is a wear item, and the diameter it will be ground to is a planned event, not an accident. The surface comes in by half the diameter reduction, so restoring the same gap requires the housing, and therefore the axis and the drive end, to move by that amount. A coupling chosen against today's diameter, without asking where the housing will sit after the next regrind, is chosen against a condition that will not exist in three years.

And what is the geometry at operating temperature? Rolls run hot, the gearbox runs at its own temperature, and the frame runs cooler than either. The displacement at the coupling at temperature is not the sum of those expansions because they act in different directions from different anchors. If the only alignment record is a cold setup reading, the running condition is unmeasured.

Answer those three and the drive train can be specified against a range rather than a point. A calender has no single alignment condition, so a gearbox and a coupling chosen for one condition are being asked to hold a range they were never rated for, and the failure will appear at whichever position was never checked.

The three answers, plus the frame and roll center drawing, the minimum and maximum gap, the coupling and intermediate shaft dimensions, the operating temperatures, and the drive arrangement for each roll, are what a drive train quotation has to be built from. Send that set and the range becomes the specification instead of a footnote to it.