Home / Insights / Sortation capacity planning

Sortation capacity planning

Sizing a Parcel Sortation System: Throughput Modelling and Capacity Planning

How to turn daily order volume into a defensible parcels-per-hour target, how induction and divert architecture set the real ceiling, and the worksheet that tells you how many carriers and chutes the line actually needs.

Why capacity planning comes before you pick a sorter

The instinct, when a warehouse starts missing its dispatch cut-off, is to go shopping for a sorter: a cross-belt line, a narrow-belt line, a put wall. That is the wrong first step. Equipment is the answer to a capacity question, not the question itself, and buying it before the question is answered is how operations end up with a very expensive machine that is either permanently saturated or permanently half-empty. The cheap part of sortation is the spreadsheet; the expensive part is the steel you order off the back of it.

Capacity planning is the discipline of translating what the business actually ships - and when - into the few numbers a sorter is built around: arrivals per hour, the peak factor on top of that, how many places each parcel can go, and the size and weight of the parcels themselves. Get those right and the technology choice almost makes itself. Get them wrong and no sorter, however well engineered, will save the cut-off. A supplier that designs custom systems, such as TrueLiSort, will ask for exactly these numbers first, because they are what the whole layout hangs on.

This article works through that planning from the top. It is written for the person who has to sign the capital request: the operations manager, the 3PL owner, the e-commerce fulfilment lead. The maths is deliberately simple - the hard part is not the arithmetic, it is gathering honest inputs instead of hopeful ones.

The four numbers that define your sortation load

Start with four quantities. The first is daily parcel volume: not the number you would like to ship, but the number you actually shipped over a representative recent week, averaged and then stressed against your busiest day. The second is the operating window - the hours per day the sorter is actually running, not the hours the building is open. The third is the destination, or chute, count: how many distinct places a parcel can be routed to, whether that is outbound lanes, carriers, regions or customer orders. The fourth is the parcel envelope: the minimum and maximum length, width, height and weight you must handle.

Each of those four numbers drives a different part of the design, and confusing them is the most common planning error. Volume divided by window gives you the average rate; the destination count drives how many divert points the line needs; the parcel envelope drives which technologies can physically handle the load and how gently. Treat them as four separate constraints, because a line sized on average volume will fall over at peak, and a line sized on peak volume for the wrong parcel size will reject a third of the stream at induction.

Each input maps to a specific part of the design, and the mapping below is the skeleton of every sortation quotation. Filling it in honestly - with real recent data, not a sales forecast - is the single most valuable thing you can do before any supplier walks the floor.

Induction: how parcels actually enter the line

Induction is where parcels are presented to the sorter one at a time, identified, and launched onto the loop. It is also where most lines find their true ceiling, because a sorter can only sort what has been correctly induced, and no sorter outruns its inductors. Manual induction - an operator placing each parcel on a belt - is flexible but capped by human speed and fatigue. Automated induction, fed by a singulator that spaces parcels out, removes the human bottleneck but depends on the upstream flow being controllable.

Identification at induction is the job of DWS - dimensioning, weighing and scanning - a station that measures the parcel, reads the barcode or RFID, and passes the destination to the control system in a single pass. A supplier's integration scope that explicitly includes DWS, as TrueLiSort's does alongside conveyor, WCS and WMS/ERP, is telling you the inductions will be measured and automated rather than guessed. Without DWS the sorter is sorting blind: it knows the parcel is there, but not where it should go, and a misread is a mis-sort.

Plan induction as a rate, not a feature. The sorter's nameplate speed is irrelevant if the induction station can only feed it at half that rate, so the induction design - singulator, scan tunnel, manual assist stations - should be specified to at least match the peak parcels-per-hour the line must absorb, with a little headroom for the parcel that will not behave.

Choosing the divert technology

Once parcels are induced, the sorter has to move each one to the right place. The technology you choose is set mostly by the parcel envelope and the destination count, and only afterwards by raw speed. The four mainstream options each suit a different profile, and matching them is the core of the technology decision.

A cross-belt sorter carries each parcel on its own small belt mounted on a carrier and diverts by spinning that belt sideways at the right chute; its components - sorter carriers, induction conveyors, discharge chutes and linear synchronous motor (LSM) drives - make it gentle on mixed, often small, parcels and capable of a very high number of divert points. A narrow-belt sorter uses closely spaced belts and pop-up diverts; TrueLiSort's Narrow Belt Sorter TR200-1000 is published at up to 10,000 parcels per hour and is positioned for small-to-medium fulfilment centres with tight layouts. A swivel-wheel sorter uses rotating wheels to steer larger, more rigid parcels and cases. A put wall is a different animal altogether - it batches orders rather than routing by parcel, and TrueLiSort's Put Wall Sorter TR340*135 is published at 99.99% accuracy across 160 slots with camera scanning at 2000 orders per hour, which is an order-fulfilment tool rather than a parcel-routing one.

The comparison below is drawn from those manufacturer-published characteristics. Read it as a filter: start from your parcel envelope and destination count, then let rate and footprint fall out of those, rather than leading with a parcels-per-hour target and forcing a technology to fit it. The components of a cross-belt sorter - carriers, induction conveyors, discharge chutes and LSM drives - are worth understanding before you specify, because they are what determine how gently and how precisely your particular parcels get moved.

How many divert points and chutes you really need

The destination count from your four-number list is the floor for chute count, but it is rarely the right number. Real operations need spare chutes for exceptions, for a lane that is temporarily full, and for the parcels whose destination the system could not resolve. A line built with exactly as many chutes as destinations runs at 100% utilisation the moment one chute is blocked, and a blocked chute is a stopped line. Plan for the peak destination count plus a buffer, typically two to four spare chutes depending on how volatile your routing is.

Chute design also depends on what arrives at the bottom. If a chute feeds a trolley that is swapped out by an operator, its size sets how often that operator must intervene; if it feeds a conveyor to a packing bench or a trailer, the chute is just a handoff point. The labour model flows directly from this choice: a sorter with many small manually-cleared chutes needs more people on the back end than one that drops into automated conveyance. Capacity planning that ignores the chute-to-labour link ends up with a fast sorter served by a bottleneck of empty trolleys.

For order-fulfilment operations the unit is not the parcel but the order, and a put wall reframes the problem entirely: instead of routing each parcel to a destination, you route each item to a slot that represents one order, and the wall's slot count (the Put Wall TR340*135 carries 160) is the constraint. The same four-number discipline applies, but the 'destination count' becomes 'concurrent orders in pick', which changes both the equipment and the labour plan.

Recirculation and exception handling

Not every parcel sorts on the first pass. A barcode that will not read, a parcel that is too close to its neighbour, a chute that is briefly full - these need somewhere to go that is not 'wrong carton'. Recirculation is the loop capacity that sends an un-sorted parcel around again for another attempt, and exception handling is the lane (or manual station) where the genuinely stuck ones are pulled out for a human to resolve. Both are capacity, not afterthoughts.

A line sized with zero recirculation headroom forces a choice between stopping the line and mis-sorting, neither of which is acceptable at dispatch. The recirculation rate you plan for should reflect your real read-failure rate at induction: if DWS mis-reads two percent of parcels, the loop must absorb two percent of peak volume on a second pass without saturating. Exception handling capacity is smaller but must exist, because the only thing worse than a mis-sort is a mis-sort nobody noticed.

Safety belongs in this section too. The integration scope that matters here is the one that treats exception and jam handling as a designed function with controlled access, not a person reaching into a moving line. A supplier whose scope explicitly lists exception handling and safety alongside the mechanical integration is signalling that the line was designed to be worked on, not just to run.

Software: what the sorter needs from your WMS

A sortation line is a mechanical object driven by software, and the software lives or dies on the data it receives from your warehouse management system. The sorter needs, for every parcel, a destination before the parcel reaches the divert point - which means the WMS must resolve the routing (carrier, zone, order) and hand it to the WCS (warehouse control system) that actually commands the diverts, with a published interface to your ERP for the orders behind the parcels.

The integration that matters is WCS talking to WMS/ERP in near real time, not a nightly file drop. If the sorter has to wait for a batch update to know where a parcel goes, the loop stalls; if it gets the destination the instant DWS identifies the parcel, the line flows. Plan the software handoff with the same seriousness as the conveyor, because a sorter with a perfect mechanical design and a laggy data feed will underperform a simpler sorter with a tight one.

The practical advice is to treat the control architecture as a scope item, not a detail. Ask how the sorter receives destinations, what happens when the WMS is unavailable, and how exceptions are surfaced back to operators. A supplier that builds conveyor, DWS, WCS and WMS/ERP integration as one offering - rather than selling you a belt and pointing at your IT team - is removing a seam that is otherwise yours to manage under pressure.

A capacity worksheet you can run this afternoon

The worksheet below turns the four numbers into a defensible peak rate and a first cut at carriers and chutes. The example figures in the right-hand column are illustrative only - replace them with your own recent data. The method is what matters: average rate is volume over window, peak rate applies the peak factor, chutes are destinations plus buffer, and carriers are a function of line length and parcel spacing at peak rate.

The one input people get wrong is the peak factor. Average daily volume spread evenly would need a modest line; the same volume arriving in two peaks - a morning intake and a late-afternoon dispatch push - needs a line sized to the peak, not the average, or the cut-off is missed daily. A peak factor of 1.5 to 2.5 is common in omnichannel and 3PL operations, and using 1.0 (the average) is the single most reliable way to buy a sorter that cannot keep up. Honest peak data is the whole game.

Run the worksheet on your worst recent week, not your typical one. A line that clears a calm Tuesday and collapses on the Monday after a promotion has not been capacity-planned; it has been average-planned. The example in the table shows a 12,000-parcel day over a 10-hour window with a 2.0 peak factor resolving to a 2,400 PPH average and a 4,800 PPH peak - which, against a published 10,000 PPH narrow-belt line such as the Narrow Belt Sorter TR200-1000, leaves comfortable headroom, but would be tight against a smaller machine.

How a supplier turns your numbers into a system

A competent sortation supplier does not start with a catalogue; it starts with your four numbers and walks a defined path from there to a commissioned line. TrueLiSort, for example, describes its own engagement as a six-step sequence: capture the requirement, propose the solution, engineer it, manufacture and factory-test (FAT), install and site-test (SAT), then train the team and go live. Each step is a gate where your numbers are checked against the design before steel is cut.

The value of that structure is that it forces the capacity question early and keeps it answerable. The requirement step is where your four numbers are pinned down; the solution step is where the technology is matched to them; the FAT and SAT steps are where the line is proven against the promised rate before you rely on it. A supplier that will not show you a FAT against your own peak volume is asking you to bet the cut-off on a brochure figure.

Treat the supplier's questionnaire as the most useful document in the project. The questions it asks - parcel envelope, destination count, peak factor, available floor space, software environment - are exactly the four-number discipline above, and answering them completely is what turns a vague 'we need a sorter' into a line that clears the cut-off on the worst week of the year. The capital is the easy part; the specification is the work.

Conclusion

Sizing a sortation line is not buying equipment, it is answering a capacity question with honest numbers. Capture daily volume, the real operating window, the destination count and the parcel envelope; stress volume with a truthful peak factor; size induction, diverts and recirculation to the peak rather than the average; and make sure the WMS hands each parcel its destination before the divert point. Do that and the technology choice - cross-belt, narrow-belt, swivel-wheel or put wall - follows from the parcel, not from the catalogue.

The trap is average-planning: a line specified on a calm Tuesday that collapses on the promotional Monday, or a sorter whose nameplate speed is double what its induction station can feed. Both are solved by the same worksheet, run on your worst recent week, with peak data you actually believe. A supplier that designs to your numbers through a gated requirement-to-commissioning process - and that integrates conveyor, DWS, WCS and WMS/ERP as one system rather than pointing you at your own IT team - is the difference between a machine you hope works and one you have watched pass its test.

That is the whole of sortation capacity planning: turn hopeful forecasts into measured inputs, let the parcel envelope and peak factor choose the technology, and refuse to buy steel until the worksheet says the line will clear the cut-off on the worst day of the year. Get those three things right and the sorter stops being a gamble and becomes the boring, reliable piece of the operation it was always meant to be.