Conveyor System Design for Assembly Lines and Material Flow
The new assembly line’s conveyor announced itself on day one as a machine that had never met a box it could carry without drama. One product toppled halfway round a bend because the chain speed was tuned to a different package, another jammed at the transfer, and the line lost its promised takt time before lunch. The conveyor had been purchased as a catalog assembly of standard components and wired to a PLC by a logic that had invented, in software, a line the hardware did not provide. Conveyor design is a material-flow problem wearing a mechanical-parts costume, and the parts are the easy half. This article works through what has to be decided for assembly and material-flow conveyors: the duty, the drive, the layout, the controls and, above all, the interfaces where products change direction.
Start With the Product and the Flow Rate, Not the Catalogue
Every conveyor decision traces back to the product and the flow it must carry. Write the product’s footprint, weight, base stability, and whether it topples at a given acceleration. Then write the required rate: pieces per hour, a takt, the worst-case burst, and whether the line ever reverses or splits. From those two inputs come the speed, the width, the belt or chain pitch, and the motor power. Selecting a conveyor before the product and rate are on paper is selecting a machine for a job nobody has written, and the first artifact it will reject is the actual product.
Speed deserves particular honesty. The belt speed that looks smooth in a demo video is the acceleration that launches a tall, top-heavy product off a bend, or leaves a rolling product drifting between stations. Calculate the maximum permissible acceleration of the product from its geometry, then size the drive and the electrical ramp so the conveyor stays under it. A conveyor that “only does 0.5 m/s” can still impose a violent jerk at start and stop if the motor is direct-coupled and the ramp is missing, and product stability is governed by acceleration, not by cruising speed.
Belt, Chain or Roller: Matching the Surface to the Product
The three great conveyor families, belt, chain and roller, each have a narrow band of honesty. Belt conveyors carry a continuous flexible surface, good for fragile or small items that must not fall between gaps, cheap for long straight runs, and limited at bends and when oil or sharp edges are present. Chain conveyors carry product on a series of links or slats, durable for heavy or hot product, tolerant of bending round a long radius, and they impose the chain’s motion characteristics and its wear and lubrication needs on the line. Roller conveyors carry product on free or powered rollers, excellent for accumulation because product can queue with minimal pressure, and they need containers with a reasonably stiff, smooth underside that will not sag between rollers.
The choice is driven by the product’s underside and the accumulation strategy. A line that must accumulate, queueing product without stopping the feed, wants rollers or a low-pressure belt method. A line carrying thin, flimsy or irregular bottom surfaces wants a belt with enough support. A line carrying heavy castings or hot parts wants chain. The design sin is picking the family from habit and then compensating with tracking rollers, pallets, or a dozen guidance rails that make the standard part expensive and fragile. Choose the family from the product, then buy the standard components that fit that family.
Straight Runs Are Easy; Transfers, Merges and Diverges Are the Design
The failure concentration on any conveyor line is not the long straight where the parts cruise, but the places where the product changes direction or lane: bends, transfers, merges, diverges, and the interface to a machine. A bend works for chain and roller families, but a belt conveyor bends badly unless it is a powered corner, and a product passing a bend must be stable under the lateral acceleration that the change of direction generates. Transfers between conveyors are where product drags, tips and jams; every transfer needs a defined handoff geometry, a speed match on the two sides, and a small overlap so the product is never unsupported.
Merge points multiply the failure rate again: two streams meeting at a convey collision, and a merge without buffering and sequencing is a jam generator wearing a productivity costume. Design the merge with a sensorised buffer that lets one stream queue while the other crosses, and verify that the PLC sequencing handles the worst case of simultaneous arrivals on both streams. Diverges have the same problem in reverse: a product that must be diverted carries a momentum that must be shed into a spur without toppling or jamming, which is an acceleration problem all over again, in the divert direction.
The Drive: Motor Selection, Speed Control and Torque Ripple
Drive sizing starts from the effective pull the conveyor needs: the mass of the goods distributed along the loaded length, the mass of belt or chain, and the friction factors of the idlers, returns and bearing losses, all converted to an effective tensile force, then multiplied by speed to get power, with a start-up margin for the extra force of overcoming static friction and inertia. A motor sized for steady-state cruising is undersized at start if the line is loaded at rest, and a line that stumbles every Monday morning under a full load is a line whose start-up torque was ignored in the catalogue math.
Speed control is the modern default. A variable-frequency drive on the motor gives a soft start and stop that protects the product from the jerk mentioned earlier, lets the line match the station rate, and provides the ability to slow for manoeuvring zones and speed for empty return. The drive’s output ramp becomes the product-stability control, so set the acceleration limit from the toppling constraint and enforce it in the drive parameters. Where precise positioning at stations is required, the conveyor needs a means of accurate stop, a sensor position, a brake or servo indexing, not a general-purpose drive hoping to hit a mark with a lurch.
Framework, Tracking and the Day-to-Day Health
The mechanical details decide whether a conveyor degrades gracefully or irritates daily. Belt tracking, the tendency of a belt to wander off the crowned pulleys, is the classic ongoing fight; proper crown, installed rollers square to the frame, and periodic adjustment keep a belt running true, while a mis-tracked belt is the source of edge wear, product wander and spillage that nobody remembers assigning a cause. Chain conveyors carry the same burden as chain stretch and wear over a long run; the tensioner and the lubrication schedule are not accessories.
The frame itself is often the quiet spec: a long straight conveyor needs a frame stiff enough that it does not sag in the middle and throw the product off-level, and the level matters to product stability and to accumulation behaviour. Simple as it sounds, the frame’s levelness and squareness at installation, verified with a spirit level and a straightedge, prevent a hundred downstream mysteries. A conveyor installed level and square is a conveyor whose belting, tracking, and product behaviour are all more predictable for the rest of its life.
Sensors, Diagnostics and the Sweet Spot of Pacing
The controls layer decides how the line behaves hour to hour. Photoelectric or inductive sensors at key points, transfer zones and merges tell the PLC where product is, and the logic advances product only when the next zone is clear. This zero-pressure or low-pressure philosophy, instead of pushing product continuously against the next item, protects the product and the line from stack-ups. The sensor placement, not just the PLC code, defines the line’s pacing behaviour, so sensor positions are a design output drawn on the layout, not an afterthought at wiring.
Diagnostics deserve as much design as the happy path. A conveyor line without indication of a stalled zone, a jammed transfer or a mis-tracked belt spends its failures as hunts. Local indicator lights on each zone, a readable alarm state in the PLC, and the pinpointing of which sensor is missing turn a line-wide stoppage into a two-minute local fix. The machine that tells its operator where it hurts is the machine that keeps its takt time.
Safety, Guards and the Human Side of Flow
Conveyors are among the most dangerous equipment on a shop floor because they run continuously and attract people reaching for a stuck product. Guards over pinch points, nip points at drive and deflection pulleys, and chain or belt access, are not a checklist item; they are the difference between a spill and a caught hand. Emergency stops, local as well as line-wide, and a restart procedure that verifies visibility and all-clear before re-energising belong in the control spec before the first valve is wired.
The human interface is a design element. Access to the line for servicing, walkways and crossing points that route people clear of moving product, and the ability to jog and inch the conveyor manually for setup, all determine whether the line is maintainable. A conveyor that is fast, productive and guard-correct but impossible to service is a conveyor that gets its guards removed by frustrated technicians, and that is the defeat of the safety design.
Commissioning With Product, Not With Air
The final truth of conveyor design is that a conveyor is commissioned the moment real product runs on it, at speed, through the bends and transfers and merges, under a full line. The empty-belt test proves the drive and the tracking; it proves nothing about product stability, transfer geometry or merge sequencing. Commission with the actual product or a worst-case surrogate, with the line loaded, and run the full cycle, including the stop-and-restart under load and the worst-case burst arrival at the merge. Film it if it helps; the camera sees the topple the operator’s eye missed.
Record the operating parameters that matter: belt speeds, transfer timing, sensor maps and the acceleration ramps that keep the product stable. These become the baseline and the tuning manual. The line that staggered on day one because it was tuned to a different package was not cursed; it was commissioned empty. Conveyor design is a loop of product, flow and interface worked out before the catalogue is opened, then verified again with product on the belt. That loop, closed honestly, is what turns a conveyor from a collection of standard parts into the reliable spine of an assembly line.
The Speed Profile: Pacing Zones Instead of One Flat Speed
Many lines are built with a single belt speed because it was the number on the motor plate, and then fight the consequences in every station. A one-speed line forces the feed rate to satisfy the slowest station and wastes the throughput everywhere else, or pushes product into a station faster than it can accept, building backpressure between every pair. The mature design divides the line into zones with different speeds: a fast empty-return run, a medium feed, a slow positioning section before a machine, and a slow accumulation section where product queues. Each zone’s speed is set by the operation it feeds, and the transfer between zones is where the buffers live.
This zoning also concentrates the risk where it is manageable. A fast feed that must decelerate before a station does so over a designed buffer length, and the product topples there only if the deceleration exceeds its stability, which the zone design now controls deliberately instead of accidentally. The camera-and-stopwatch commissioning earlier is what validates each zone’s ramp against the real product. A line that documents its zone speeds and acceleration limits is a line whose tuning lives in a manual, not in the memory of the shift lead who set it once and left.
Accumulation Pressure and the Product’s Crush Limit
Products that queue on a conveyor exert a force on each other that grows with line length and friction. On a belt conveyor relying on friction, a long accumulation can build enough push to crush, dent, or destabilise the front product, or to jolt the line when the queue backs into the feed. Roller conveyors with controlled back-pressure and zero-pressure accumulation, where each zone’s rollers freewheel when the next zone is occupied, keep queue force near zero. The product’s crush limit, a number as important as its weight, decides which accumulation method the design can afford.
Set the accumulation strategy from the product, not from the conveyor vendor’s default. Fragile or soft products belong on zero-pressure accumulation with sensors gating each zone. Rigid products in rigid containers can tolerate modest back-pressure with a low-pressure belt and a controlled drive. The product’s stacking and sliding coefficient decides how many products a long queue can accumulate before the back-pressure exceeds the front product’s limit. Writing that number and that limit down, on the layout and the spec, turns a queue from a day-to-day surprise into a designed buffer.
Total Cost of Ownership and Spare Parts Discipline
The cheapest conveyor to buy is rarely the cheapest to run. A conveyor’s total cost of ownership is dominated by energy, the speed at which components wear, spare parts lead time, and the debug time of the fitters who must keep it alive. Belt and chain replacement cost, drive spares, and the frequency of sensor replacement on a dusty or oily line all belong in the selection, not as a purchase add-on. Standardisation across the line, one belt family, one drive brand, one sensor type, shrinks the spares shelf and the training burden even when an individual component is a few percent dearer.
Plan the spares as part of the design. A spare drive, a spare belt and a spare set of transfer components kept on site convert a two-week shutdown waiting for a part into a same-day repair. The line’s designers know which components are the failure-prone ones; the spare list is that knowledge written down. A conveyor system whose spares strategy matches its failure profile is a conveyor system that keeps its promised throughput, which is the whole point of designing material flow in the first place.
