Nobody gets excited about conveyors. The filler, the robot cell and the palletiser get the attention and most of the budget, and the conveyor that links them together is often treated as plumbing. Then the line goes in, and it turns out the plumbing decides how much of the time those machines actually get to run.
Most of what makes a conveyor system work well is not the steel or the belts. It is the control logic: how product is held when something downstream stops, how it gets routed to the right place, and how the conveyor talks to every machine it connects. This post explains the three things that matter most, accumulation, sortation and integration, and where conveyor projects tend to go wrong.
Accumulation: why lines need somewhere to wait
Every machine on a line stops now and then. A case packer needs a reel change, a labeller jams, a palletiser waits for an empty pallet. If the machines are joined by conveyor with no room to hold product, one machine stopping stops everything upstream of it within seconds.
Accumulation is the space and the logic that lets machines keep running through each other's short stoppages. It decouples them, so a two minute stop on one machine stays a two minute stop on that machine rather than spreading along the whole line.
The main approaches
| Type | How it works | Best suited to |
|---|---|---|
| Zero pressure | The conveyor is split into zones, each with its own sensor and drive, so product waits in separate zones without touching | Fragile, unstable or mixed products, and anywhere you need to track individual items |
| Minimum pressure | Product is allowed to queue and gently touch, with the drive force reduced so the line pressure stays low | Robust packs and cases where some contact is acceptable |
| Line pressure | Product simply backs up against itself on a running belt or roller | Short queues of tough products where cost matters more than control |
| Buffer tables and stores | A dedicated area holds a larger quantity of product and releases it when the downstream machine restarts | Covering longer stoppages between high speed machines |
Sizing the buffer
The arithmetic is simple, and it is surprising how often it gets skipped. If a filler runs at 120 packs a minute and the case packer after it routinely stops for three minutes, you need room for 360 packs between them to keep the filler running through that stop. Anything less and the filler stops too.
The part people miss is getting the buffer empty again afterwards. If the downstream machine can only run at exactly the same speed as the upstream one, a full buffer never drains, and the next stoppage hits a line with no headroom left. Downstream machines generally need to be able to run a little faster than the ones feeding them so the queue can clear. That is what line balancing is really about, and it has to be designed in from the start rather than tuned on site.
Key point: Size buffers against the stoppages that actually happen on your line, not the nominal speeds on the machine data sheets. A buffer sized on paper speeds will look fine on day one and fall short as soon as real life arrives.
Sortation: getting product to the right place
Sortation is any point where the conveyor has to make a decision: send this case to lane two, reject that one, merge three lines into one, split one line across two palletisers. The mechanical options are well established and the right one depends on the product, the rate and how gently it needs handling.
- Pushers and paddles knock product sideways onto another lane. Simple and robust, but not gentle
- Pop up wheel or roller sorters rise between the rollers and steer product off at an angle. A common choice for cases and totes
- Swing arms and diverter gates guide product across as it travels. Good for moderate rates and mixed sizes
- Sliding shoe, cross belt and tilt tray sorters handle high rates and large numbers of destinations, typically in distribution and parcel handling
Whichever mechanism you choose, the hard part is knowing which item is which by the time it reaches the divert. Product is usually identified once, by a barcode scanner, a vision system or a checkweigher, and then tracked along the conveyor until it reaches the point where the decision is acted on. If the tracking slips, the right decision gets made about the wrong item.
The questions sortation logic has to answer
A sorter that works on the happy path is easy. The control design is really about the awkward cases:
- What happens to an item the scanner could not read?
- What happens when the destination lane is full?
- Does the system confirm a reject actually left the line, or just assume it did?
- If an item is pushed off by hand mid-conveyor, does the tracking recover or does every item after it get sent to the wrong place?
Rejects deserve a particular mention. On a line with a checkweigher or a metal detector, a reject that is commanded but not confirmed is a product safety issue, not just an efficiency one. Reject confirmation and a lockable reject bin are worth insisting on.
A conveyor that works when everything goes right is not difficult. The engineering is in what it does when something goes wrong.
Integration: where most conveyor problems really live
A conveyor on its own does nothing useful. Its whole job is moving product between machines, so it is only as good as its conversations with them.
Machine handshakes
Each machine on the line needs to tell the conveyor whether it is ready, running, blocked or starved, and the conveyor needs to respond sensibly to each. That sounds obvious, but those signals often come from different machine builders who have each implemented them a little differently. Agreeing exactly what every signal means, and testing it, is a large part of a conveyor controls job.
Robots and palletisers
Where a conveyor feeds a robot cell or a palletiser, product has to arrive where the robot expects it, oriented the right way and at the right pitch. Getting that right is a shared problem between the conveyor and the robot, and it is far easier when the same team is programming both ends.
Site systems
On larger systems the conveyor also has to report upwards to SCADA, or take instructions from a warehouse management system telling it where each item should go. These links are where projects quietly slip, because the conveyor supplier and the IT side each assume the other has defined the interface.
Fault recovery
The test of a well integrated system is what happens after an emergency stop. Product is left mid-transfer, some zones are full, and tracking data may no longer match what is physically on the belt. A good system restarts with product still on it, rebuilds its picture of where everything is, and carries on. A poor one needs someone to clear the line by hand every time, which costs production and puts people next to moving equipment far more often than they should be.
Planning a conveyor project or upgrade?
We design and commission conveyor control systems that coordinate flow across the whole line, including accumulation, divert logic, line balancing and integration with your machines and site systems.
Conveyor Automation →Where conveyor projects go wrong
Most problems we are asked to fix fall into a handful of patterns:
- "Controls by others." The mechanical conveyor arrives with no control logic, and the control scope is bolted on late without a proper specification
- Sensor positions decided on site. Photocells end up wherever there is a bracket, rather than where the zone logic needs them
- Buffers sized on nominal speeds, so they fill and stay full
- No recovery logic, so every stop needs a person to sort out the line by hand
- Changes made live. Logic modified directly on a running system because there was no offline way to test it
The last one matters most on existing lines. When we reconfigured the routing logic on an automotive skid buffer to take a new long wheelbase variant, all of the PLC and HMI changes were developed and tested offline before any of it touched the live system, and the line started up without production issues. You can read how that was done in our skid buffer case study. For bigger systems, virtual commissioning takes the same idea further by testing the real PLC code against a model of the whole conveyor network.
Questions to ask before you commit
Whether you are buying a new system or upgrading an old one, these will tell you a lot about how the project will go:
- How have the buffers been sized, and against what stoppage data?
- How is product tracked between identification and the divert point, and what happens on a no-read?
- How are rejects confirmed?
- Who defines and tests the signals between the conveyor and each machine?
- How does the system restart after an emergency stop with product on the line?
- How will changes be tested before they reach a live line?
- What documentation will you get, and can your own maintenance team support it?
How we approach it
At Duke we treat the conveyor as part of the line rather than something between the interesting machines. We design the PLC control, HMI, accumulation and divert logic with the whole line in mind, on Siemens, Rockwell or whatever platform your site already runs, and we integrate it with the robots, palletisers, checkweighers, scanners and site systems it has to work with. Because our conveyor engineers sit alongside our robot programmers, the handover between conveyor and cell is designed by one team rather than negotiated between two.
We do new systems and we upgrade existing ones, including obsolete conveyor controls on live lines, with the work developed and tested offline first. If you have a line where the conveyor is holding everything else back, we are happy to take a look.