Conveyor Roller Ultimate Guide (2026): Types, Grades and How to Choose

A belt conveyor in a quarry, a cement plant or a port stockyard lives or dies on parts that cost a fraction of the machine. The conveyor roller is the clearest example. A 1,600 mm limestone line moving 2,000 t/h runs over a few thousand of them, and when one conveyor roller seizes it does not stop politely. It wears a flat on the shell, the skirt rubber tears, the belt starts to wander, and by the end of the shift you are pricing belt instead of a bearing.

We have built belting since 1988. Today we run ten lines, eight for fabric carcass and two for steel cord, and we supply more than 1,500 industrial customers in mining, ports, cement, steel, power and EPC work. Idlers came with that territory. When a customer calls about tracking, spillage or a belt that failed early, the cause is often upstream of the belt itself — the troughing geometry, the spacing, or a seal that gave up six months ago. If you buy the belt from a conveyor belt manufacturer who also builds the components, you get one throat to choke when the line still will not behave.

Send Us Your Conveyor Duty Data for a Roller Selection Review

Search for this subject in 2026 and most of what you find describes package-handling roller beds inside a warehouse, where a 50 mm roll carries a 20 kg carton at walking pace. We counted the top nine results on the English SERP: seven belong to that world. Only two come from belt-conveyor people, and both of them list types without ever showing a selection calculation. That is the gap this guide fills. Specification, life and failure, in that order, for heavy-duty troughed belts carrying rock, clinker and ore.

01What a Conveyor Roller Actually Does on a Heavy-Duty Belt

Four jobs, and only one of them is about support

A roll in a troughed set has to carry the load, hold the belt in line, absorb the cyclic stress that the belt sees as it passes, and survive whatever the environment throws at it. Everybody designs for the first job because it is the easiest to satisfy. The other three are where the money goes.

Put numbers on it. A 1,400 mm line at a 35° trough, idlers on 1,200 mm centres, carrying 1,500 t/h of limestone at 3.0 m/s. Belt mass is about 15 kg/m once you count a 6 mm top cover and a 4.5 mm bottom cover on an EP 800/4 carcass. Material mass works out at 1,500 ÷ (3.6 × 3.0) = 139 kg/m, so a single set of three rolls sees roughly 185 kg of belt and rock. Split that across a 35° set and the centre roll takes about 43 percent, the two wing rolls about 28 percent each. The centre roll is then carrying 80 kg, which is 40 kg at each bearing. Under 0.4 kN of radial load.

The order of decisions matters more than the answers

Duty sets the trough angle. Angle decides how the load splits between the rolls. Sag under the tension you actually have on that stretch of line decides spacing, and spacing decides running resistance, spillage and how badly the belt fights itself at the transition. Diameter matters last, and mostly because it fixes the bearing size and the rotational speed at the seal lip. Change one of those in isolation and you move the problem instead of solving it.

We have watched this play out on a port stockyard line handling coal at 1,200 t/h. The maintenance team replaced 89 mm steel rolls with 152 mm polymer rolls on the return side to cut noise and weight. Noise dropped by about 6 dB. But the shell mass fell and the belt ride height rose, so the return rollers now sat further from the belt, the self-cleaning effect of a heavy shell disappeared, and the build-up under the line got worse. A decision taken for one symptom created a second problem six weeks later.

What each service environment punishes

Four environments cover most of our order book, and each one attacks different parts of a roll. In a primary crusher discharge the enemy is impact and lump size, which crushes shells and fractures end caps. In a cement plant the enemy is fine, alkaline dust that migrates through the smallest gap and turns grease into a grinding paste. At a port it is salt spray, rain and washdown water, which rusts shafts and emulsifies grease. On a wet quarry face the enemy is clay and slurry that wraps the shell and destroys tracking.

Where this guide sits against everything else you will read

02Conveyor Roller Types and Where Each One Belongs

Steel rolls, and the case for a thicker wall

Seamless or ERW steel tube in 3.0, 3.5, 4.0, 4.5 or 6.0 mm wall, with a machined end cap welded to the shell and a through shaft in 20, 25, 30, 40 or 50 mm diameter. Steel is still the default on abrasive duty because nothing else holds a straight shell as long against sharp material. Wall thickness is where cheap specifications hide. A 3.0 mm wall is fine on a clean, light, well-supported line; put it under a 1,600 mm belt in granite and the shell will dish between the bearings and the belt will start to dish with it.

Polymer shells on wet and corrosive lines

HDPE and UHMWPE shells in 89 to 194 mm diameter cut mass by roughly a third against steel and never rust. They shed sticky material better than steel, which is why they win in fertiliser, salt and wet coal. The trade-off is temperature and impact. Above about 80 °C continuous, or at a transfer point where 300 mm lumps land, a polymer shell is the wrong answer and a steel or rubber-disc impact roll is the right one.

Impact and buffer rolls at the transfer point

Under the chute, the roll is not supporting a steady load; it is catching a stream. Spacing there drops to 300–450 mm, shell walls go up to 6 mm, and rubber discs or a thick rubber sleeve are used to absorb the strike. The bearing in an impact roll also takes the axial component of a bouncing lump, so shaft diameter jumps to 30 or 40 mm even when the radial load does not require it.

On a 900 t/h stone line we rebuilt after repeated impact failures, the fix was not a bigger bearing. It was a set of five 152 mm impact rolls on 350 mm centres with a 12 mm rubber sleeve and a proper skirt, which spread the strike over three rolls instead of one. Failure stopped. The load per roll fell by more than half without changing the bearing size at all.

Self-aligning and training rolls

These are the rolls that steer rather than support. A self-aligning roller carries its own pair of angled end rolls, either pivoting about a centre or about a vertical axis, and turns the belt back toward the centre whenever it drifts. They are installed in the zone where misalignment starts, and the useful ones are the ones with enough pivot friction to correct a belt without snapping it sideways.

Two mistakes are common. Mounting a self-aligning set within three metres of a transition area, where it fights the trough geometry. And using self-aligning sets as a substitute for fixing a skewed frame, an off-centre load point or badly hung skirt rubber. Training rolls correct drift; they do not correct a bad installation, and they will chew belt edges while you wait for them to do the impossible.

Troughing idlers: two-roll, three-roll and five-roll

The carrying side is where most of the money sits. Three-roll troughing idlers dominate from 500 mm to 2,000 mm belt width, with the centre roll horizontal and two wing rolls at the trough angle. Two-roll sets survive in narrow belts and in confined transfers. Five-roll sets, usually 35° or 45° on the outer pair and 20° on the inner pair, appear on very wide belts and where a deep load section is needed without over-stressing the belt edge.

There is a real difference between a troughing idler that is built as a set with a welded frame and one assembled from three rolls bolted onto a bent stringer. On heavy duty the set is the only version worth buying, because frame stiffness holds the trough profile. A frame that flexes under load lets the wing rolls splay, and the belt then runs on a shallower profile than the design assumed, so it carries less, spills more and mistracks.

Return rolls, spiral and disc types

Those cost indices are relative to a standard 3.5 mm steel carrying roll of the same frame class, and they are the right way to think about the decision. Choosing a polymer roll to save rust can double component life in a port, and choosing the same roll at a crusher discharge can halve it. The type is only correct in the context of one location.

Three-roll troughing set with sealed end caps, ready for a 1,600 mm quarry line.

03Trough Angle, 20° to 45°, and What It Does to Belt Edges

Trough angle is the single most consequential number in a carrying set, and it is usually copied from a standard drawing without a second thought. It decides how much the belt can carry, how much extra strain the edge zone sees on every pass, how the load splits between the three rolls, and how much trouble you will have with spillage at the skirt. Get it wrong and no bearing upgrade will save the belt.

Where the load actually goes

A three-roll set with the wing rolls at 35° carries more than the same set at 20°, because the load section is deeper. It also shifts where the load lands. At 20° the centre roll takes roughly 30 percent of the vertical load, at 35° about 43 percent, and at 45° closer to 55 percent. Maintenance teams that swap trough angles often without checking pressure on the centre roll end up replacing centre rolls on a two-to-one ratio against the wings.

Capacity rises more slowly than most people expect. Against a 35° set at the same belt width, our load-section ratios run about 0.80 at 20°, 0.89 at 25°, 0.96 at 30°, 1.00 at 35°, 1.06 at 40°, 1.12 at 45° and 1.28 at 60°. Those figures assume a 20° material surcharge angle; a sticky material with a 10° surcharge cuts the gain substantially. Run the numbers against CEMA 5 for your own material before you buy a wider belt to solve a capacity shortfall.

The price you pay in the edge zone

Every bend at the hinge between the centre roll and a wing roll puts the belt carcass into a compound shape, and the outer 100–150 mm of the belt takes the worst of it. The edge must stretch and compresses differently from the centre, which is why edge cracking, edge delamination and feathering almost always start in the troughing zone rather than at a pulley.

As a working rule we treat the added edge bending strain as scaling with the square of the trough angle at a fixed belt width. Going from 35° to 45° then adds roughly 65 percent to that edge strain while buying you 12 percent more capacity. On a stiff belt, say a 6 mm cover heat-resistant carcass or a steel cord belt, that trade is usually bad. On a soft, thin EP carcass in a limestone plant it can be fine. The belt has to be able to take the shape, which is what the troughability test in the belt standards measures. A sample that will not sag enough under its own weight will bridge between the rolls, lift off the wing roll and concentrate the load somewhere the designer never intended.

Spillage and the skirt line

Shallow troughs spill because the load section is shallow and any surge rolls over the edge. Deep troughs spill because the gap under the skirt rubber changes as the belt profile changes, and the material finds the gap. The practical answer on a quarry line is usually 35°, a correctly set skirt with a flexible seal, and a load point that is centred rather than dumped to one side.

Transition distance: the part nobody schedules

The belt cannot go from flat over the tail pulley to a full 35° trough in one metre. It needs a transition length, and if that length is too short the edge zone is forced into a shape it cannot take, which is where you get the classic crack pattern that starts at the edge and runs diagonally into the carcass. Fabric belts on heavy duty normally want something of the order of 1.2 times the belt width of free run for a 35° trough; steel cord belts with thick covers want more, and going up to 45° wants more again.

When a customer sends us photographs of edge cracks and asks which belt to buy instead, the honest first question is whether the transition was ever right. A conveyor belt supplier who will not ask that question is selling you a replacement part rather than a longer life. We ask for the drawing, the trough angle and the measured transition distance before we will quote a belt for a line with edge cracking.

Choosing the angle for your material

Our default for mining, quarry and cement work is 35°. Step up to 45° only when belt width is at least 1,600 mm, the carcass is genuinely troughable and you have measured the transition zone, or when you need capacity at a width you cannot change. Step down to 20° or 25° for thin belts, very stiff high-temperature carcasses, thick steel cord belts, difficult splice geometry and lines with frequent short reverses.

On the industrial conveyor belt side the rule of thumb is that a belt can be troughed more deeply if it has a thinner carcass and adequate cover adhesion. A 1,000 mm EP 630/3 belt with a 4 mm top cover will happily run at 40° in a cement plant. A 1,000 mm steel cord belt with 8 mm covers will not, and forcing it will cost you the splice long before it costs you a roll.

Read that table as a set of trade-offs, not a menu. The 60° row exists because some enclosed and pipe-conversion designs need it, and it comes with nearly three times the edge strain of a 35° set. If you are not building an enclosed conveyor, that row is not for you. If you buy a generic wholesale conveyor belts carcass and then push it to 45°, expect the belt to advise you of the mistake within a year.

04Diameter, Spacing and Sag: One Decision, Not Three

Spacing is where most selection guides give a table of ranges and stop. That is not useful on a real line, because the correct spacing depends on the tension in the belt at that point, and tension changes along the conveyor. Here is the calculation we actually use.

The sag rule and the two percent line

Sag is expressed as a percentage of the idler spacing, and it is what turns a belt into a hammock between two rolls. Keep it under 2 percent and the belt runs flat enough that material sits still, running resistance stays low and the cyclic stress at each roll is small. Go past 3 percent and you get material bouncing, spillage at the edges, higher indentation resistance, and a belt that flexes enough to work the carcass and the splice. Some short, slow conveyors live happily at 3 percent. Nothing on a 2,000 t/h line should.

The span between two rolls under a uniformly distributed load and a belt tension T settles into a parabola, so the sag is

y = (m × g × s²) ÷ (8 × T)

with m as belt plus material mass in kg/m, g as 9.81 m/s², s as the spacing in metres and T as the belt tension in newtons at that location. If you want to hold the sag at 2 percent of the spacing, y = 0.02 s, and the equation rearranges to something you can do in your head on site.

s_max = 0.16 × T ÷ (m × g)

That single line is the spacing criterion, and it explains why the spacing in a standard table only works at the tensions the table assumed. It also explains why the loading zone and the tail need closer spacing than mid-span. Tension at the tail on a short conveyor can be a fraction of the tension at the head pulley, and the sag penalty goes up as the square of the spacing.

Two worked examples

Take the 1,400 mm limestone line again. m is 154 kg/m, and at mid-span the running tension is a healthy 40 kN. Then s_max = 0.16 × 40,000 ÷ (154 × 9.81) = 6,400 ÷ 1,511 = 4.24 m. The sag rule is nowhere near binding, so the spacing will be set by standard frames, by shell stiffness and by how much belt you want hanging between supports. We would still use 1.4 m rather than 4 m, because a belt with 4 m of unsupported span flaps in the wind and will not stay in the trough.

Now move to the loading point on the same line. The tension there might be 10 kN, and m rises because the chute is adding material continuously. s_max = 0.16 × 10,000 ÷ 1,511 = 1.06 m. That is already tighter than the 1.4 m we were using at mid-span, and it is the reason the first four sets after a chute deserve attention. Under a crusher discharge with 300 mm lumps we would not stop at 1.06 m either; impact rolls go in at 350–450 mm centres regardless of what the sag formula says, because the load is a strike rather than a static weight.

What diameter changes

Diameter does not change capacity or sag. It changes rotational speed, and rotational speed is the multiplier that turns a bearing rating into a calendar interval. A 152 mm roll on a 3.0 m/s belt turns at 377 rpm. Drop to 89 mm and the same belt spins it at 644 rpm, nearly double the revolutions for the same distance travelled, so a bearing with an identical rating life in revolutions reaches its limit in half the time.

Minimum and maximum spacing in practice

Two more limits sit outside the formula. Very wide spacing on the carrying side lets the belt lift off the wing rolls between sets, so the trough profile collapses and material runs to the edge; heavy duty carrying sets rarely go beyond 1.6 m for that reason. Very close spacing on the return side makes the belt work harder than it needs to, and every extra conveyor roller is another seal to fail. A conveyor belt factory that also builds the frames will usually be able to tell you what spacing the structure and its conveyor brackets were designed for and where it can be stretched.

05Bearings and Seals: What Actually Sets Service Life

A 6204 deep groove ball bearing carries a basic dynamic rating of 13.5 kN. On the 1,400 mm limestone line, the end bearing of the centre conveyor roller sees about 0.4 kN. That is three percent of rating, and at that point fatigue is no longer the mechanism that ends the bearing's life. Grease ageing, fit, clearance and contamination take over. It is the reason two plants can fit the same bearing and see eighteen months in one and eight years in the other.

The seal is the life-limiting component

The working gap at a seal lip is a fraction of a millimetre. Quarry and cement dust runs from 0.05 mm to 0.5 mm, which means grit is not merely fine enough to pass the gap, it is comfortably inside the size band where it migrates. A single contact lip slows that migration and wears itself out doing so. A metal labyrinth lengthens the path and adds no friction, but it needs a grease-filled chamber at the outer stage to block the last gap.

Water is a different enemy from dust. Rain, hose-down and slurry destroy a single lip because water emulsifies grease into a white paste that no longer carries a film. On port stockyards and wet quarry faces we specify a multi-stage labyrinth with a lip at the outer stage and a grease chamber behind it, and we ask for the actual construction in writing rather than a marketing grade name. Because we press our own rubber conveyor belt carcass in the same plants, the people who choose a cover compound for abrasion also choose the seal stack, and that has kept us honest on both sides.

One cement plant came to us after replacing 1,800 bearings in fourteen months on a clinker line. The bearings were fine. The rolls were single-lip with no outer grease chamber, so the air-borne clinker dust reached the ball track within weeks of each change. We moved them to a multi-stage labyrinth with a lip, kept the same bearing brand and the same frame, and the replacement rate dropped below 200 a year. Nothing about the bearing changed. Everything about the seal did.

Why the seal beats the bearing brand

Buy through a conveyor belt distributor who never asks about your environment and you may not learn which seal you received until the second replacement. Put the construction in the specification instead. The same argument applies to anything that spins in grit; we are a transmission belt manufacturer and a V-belt manufacturer as well, and a V-belt that loses cord to abrasive dust fails in a way nobody blames on the rubber.

Spacing check and seal inspection on a quarry carrying run, 1,200 mm belt.

06L10 Bearing Life and How to Convert the Load

Rating life is the only number in a roller specification that claims to be predictive, so it is worth understanding what it actually predicts. It predicts fatigue. It says nothing about the grease, the seal, the fit or the water.

The rating life equation

L₁₀ = (C ÷ P)^p × 10⁶ revolutions

C is the basic dynamic rating from the bearing catalogue, P is the dynamic equivalent load at the bearing, and p is 3 for ball bearings and 10/3 for roller bearings. L10 is the life that 90 percent of an identical population reaches. In theory. Convert to hours by dividing the revolutions by 60 times the rotational speed in rpm, and remember that rotational speed is set by shell diameter and belt speed.

Building P from the idler load

Start with the load on the set, from belt mass plus material mass times spacing. Split it by trough angle using the shares in section 03. Halve the roll load to get the load at one bearing, then add the roll's own weight, which is not negligible: a 152 mm steel roll 530 mm long weighs about 7 kg, contributing 34 N to each end bearing continuously.

Then apply an application factor. We use 1.0 for a clean, well-aligned, indoor duty, 1.3 for normal outdoor dust, and 1.5 for severe duty where the belt is misaligned, the chute is off centre or the belt runs on top of spillage. The factor is not a fudge; it stands in for the axial thrust, shock and misalignment that the simple radial model ignores. Where the wing roll sees a real axial component, use P = X·F_r + Y·F_a with the bearing maker's X and Y for your F_a/F_r ratio rather than pretending the load is purely radial.

A worked check on a 1,600 mm iron ore line

Parameters: 2,000 t/h at 2.5 m/s, 1,600 mm steel cord belt at 32 kg/m, 35° trough, 1,500 mm carry spacing. Material mass is 2,000 ÷ (3.6 × 2.5) = 222 kg/m, so the set carries (32 + 222) × 1.5 = 381 kg. The centre roll takes 43 percent, which is 164 kg, so 82 kg at each bearing, or 0.80 kN. Add 0.03 kN for the roll, then apply 1.5 for a dusty transfer-adjacent zone and P = 1.25 kN.

With a 6204 at C = 13.5 kN, L₁₀ = (13.5 ÷ 1.25)³ × 10⁶ = 1.26 × 10⁹ revolutions. A 152 mm roll on a 2.5 m/s belt turns at 314 rpm, so the rating life comes out near 67,000 hours, or seven and a half years of continuous running. Step up to a 6305 at C = 22.5 kN and the number goes past 300,000 hours, which should tell you the bearing is not the constraint.

Why the real interval is a tenth of that

Field data from heavy duty installations puts decent idlers at 20,000 to 40,000 hours before replacement, and poor ones at 8,000. The gap between that and 67,000 hours is contamination, grease life and fit, none of which appears in the L10 equation. Modern rating practice acknowledges this through a life modification factor derived from lubrication condition and contamination level, and for a severe dust environment that factor can be as low as 0.2. Multiply the clean-world answer by 0.2 and you land at 13,400 hours, close to what a cement plant actually reports.

07Failure Modes and How to Diagnose Them on Site

Diagnosis is where most maintenance budgets are lost, because the visible damage is usually downstream of the cause. A flat on a shell, a cracked belt edge and a torn skirt can all come from one seized conveyor roller, and replacing the roll without asking why it seized buys you the same failure on the next shift.

Shell, cap and bearing inspection before a carrying set leaves the plant.

Seized and flat-spotted rolls

A conveyor roller that stops turning does not announce itself. The belt simply slides over it, and because the belt is moving at 2–3 m/s the shell polishes a flat in hours. Walk a carrying run and look for a shiny flat strip on a shell that has lost its original finish; that is a seized roll, and there are usually three or four more in the same boat on the same line. Spin each suspect by hand and feel for notchiness, then check the shell temperature with an infrared gun. A bearing that has lost its grease runs 15–25 °C above its neighbours long before it locks up, which is why we recommend a thermal walk on any line that has just been rebuilt.

Shell wear, corrosion and water ingress

Noise, and what each sound tells you

A single high chirp at one location is almost always a dry bearing. A low rumble that appears across a group of rolls is usually a resonance excited by the belt passing frequency, and the fix may be a spacing change rather than a roll change. A rattle under the chute means an impact roll or its frame has failed and the belt is now taking the strike directly. Squealing that follows the belt around the whole circuit rather than staying at one roll is a tracking or alignment problem, and replacing rolls will not touch it. A rhythmic thump once per revolution points at a distorted shell, often a roll that was dropped during installation.

The order we check things in

08Roller Choice and Belt Choice Are One Decision

Nothing in the last seven sections is independent of the belt. Spacing decides sag, and sag is cyclic bending into a splice that will eventually fail in fatigue if the sag is too deep. Trough angle decides edge strain, and edge strain decides whether the belt completes its wear life or cracks at the edge in year two. Impact rolls and their spacing decide whether the top cover survives a crusher discharge. Roll drag is part of the conveyor's resistance, which feeds back into the tension the belt must carry, which feeds back into the sag at every set.

Send us the belt width, the tonnage, the material, the speed, the trough angle, the spacing and a photo of the loading zone, and we will come back with a component and belt recommendation that fit each other. We have built belting since 1988 and we would rather quote the right roll for your belt than sell you both twice.

Get a Roller and Belt Specification Review

09Frequently Asked Questions

How do I choose the trough angle for my belt width?

Start at 35° and stay there unless something forces you away from it. Go to 40° or 45° only when the belt is 1,600 mm or wider, the carcass passes a troughability test and the transition length can be extended. Go down to 20° or 25° for stiff steel cord belts, thick high-temperature carcasses and lines where edge cracking has already been a problem. Then check the capacity you actually need, because the gain from a deeper trough is smaller than most retrofit plans assume.

What diameter should an idler be on a heavy-duty line?

Match the diameter to the belt width and the shaft the frame was designed for. Heavy duty practice puts 89–108 mm rolls on 500–1,000 mm belts, 127–133 mm on 1,200–1,400 mm, and 152–194 mm on 1,600 mm and above. A larger shell turns more slowly, which stretches bearing life in hours, and it holds a straighter profile under load. Beyond that, the diameter is mostly a question of what the frame and the belt line can accept.

How far apart should carrying idlers be spaced?

Do not take a number from a table and stop. Calculate s = 0.16 × T ÷ (m × g) at the lowest tension on your line to hold sag at 2 percent of the spacing, where T is the belt tension in newtons and m is belt plus material mass in kg/m. If the result comes out wider than 1.6 m, use 1.4–1.6 m on heavy duty anyway, because a long unsupported span lets the belt lift off the wing rolls and lose its trough profile.

Do sealed bearings outlast regreaseable ones in dust?

It depends on whether anyone actually walks the greasing route. A purgeable design with a real schedule is the longest-lived option in a dusty plant, because fresh grease keeps pushing contamination out of the labyrinth stages. Without that schedule it is worse than a sealed multi-stage labyrinth, since the purge chamber dries and becomes a direct path to the lip. On lines with no maintenance access at all, buy sealed and put the money into the seal stack.

How can I tell a bearing failure from a shell failure?

Look at the pattern. A seized or dry bearing leaves a polished flat strip on the shell and shows rust or emulsified grease once you open the roll. A worn shell shows an even reduction in wall thickness across the whole contact band and the bearings are usually still serviceable. A distorted shell has a clean crack or a bend near a weld and its bearing often runs hot. Different causes, different fixes, and the shell pattern is the fastest evidence you have on site.

Can I change this roller type without changing the belt?

Sometimes, and only when the geometry does not move. Swapping steel for polymer at the same diameter and the same spacing changes mass, drag and wear behaviour, and the belt usually tolerates it. Changing diameter changes the belt ride height, which can put the skirt or a plough out of reach, and changing the trough angle changes edge strain, which the belt will notice immediately. If a change touches the trough profile or the belt line, treat it as a belt and component decision together.