Views: 0 Author: Site Editor Publish Time: 2026-11-19 Origin: Site
Warehouse peak throughput is the highest sustained rate the building must handle inside a defined time window for a defined set of processes — receiving, put-away, replenishment, picking, packing and shipping — and that rate, not the daily average, should size an automated system. Averages hide the short intervals where work actually queues, and those intervals cluster around events such as shift changes, when HSE's warehousing guidance advises assessing the greater risk that comes with more vehicles and pedestrians on traffic routes.
Fix the unit and the process scope first, then rebuild demand from timestamped transactions at short intervals, aggregate into candidate windows, choose the governing window for each process, and label every input as measured or assumed. Warehouse flow is a time-and-space problem — the layout that saves floor space can lengthen a picking route — which is why Warehouse and Distribution Science treats storage, picking method and flow as one planning problem rather than as a single capacity number.
Fix the unit and the process boundary. Choose one unit per process — pallets received per hour, order lines picked per hour, cartons shipped per hour — and name where each process starts and ends, so two planners measuring the same flow produce the same number.
Harvest timestamped demand. Pull transaction timestamps for a period covering a normal week, a month-end, a promotion and every seasonal event you trade through, and record the extract's cut-off date.
Aggregate into candidate windows. Roll the same data into 15-, 30- and 60-minute buckets and into shift buckets. The highest bucket for each window length is your candidate peak at that resolution.
Choose the governing window per process. Do not average across processes: receiving, picking and shipping often peak at different times, so each needs its own governing rate.
Label every input as measured, forecast or assumed. The assumed group is what a supplier will argue about, and the one a reviewer will ask you to defend.
Data quality limits the profile. Transactions posted in batches at shift end, timestamps rounded to the nearest hour, or manual moves that never reach the system will flatten exactly the intervals you are trying to size. Reconcile the system extract against a week of physical counts and yard records before you publish any rate, and label which figures came from a count and which from a system.
Match the window to the decision it supports: short windows size equipment, buffers and transfer points; an hour sizes dock and labour plans; a shift or a day only sizes volume. Count each process in the unit its equipment actually moves.
Candidate window | What it exposes | Use it for | Watch out for |
|---|---|---|---|
15 minutes | Bursts inside the hour and queueing at a single interface | Sizing buffers, conveyors, lifts and transfer points | Distorted by one late receipt or one batched posting |
30 minutes | The sustained working rate most systems must ride through | The headline rate stated in the requirement | Can still hide a 15-minute burst that causes congestion |
60 minutes | The overall rhythm of one process within the hour | Labour, dock and inbound scheduling | Can be met by a system that fails the half-hour rate |
Shift | How volume interacts with breaks, handovers and traffic | Staffing and traffic management windows | Averages away the changeover period when risk rises |
Peak day | Campaign and inbound arrival patterns | Storage and inbound planning | Is a total, not a rate; never substitute it for an hourly figure |
Write every rate as a rate: pallets put away per hour, order lines picked per hour, cartons despatched per hour. A peak-day total can conceal an hourly rate far above the daily average, so a supplier reading only the total will size for the total. Where two units describe the same work, keep the conversion factor in the document.
Carry three separate lines instead of one blended figure: a routine peak measured from your own interval data, a seasonal or promotional peak measured from a comparable past period, and a growth assumption that names its horizon and its source. Blended, nobody can defend the number, and a supplier cannot tell you which part of the design pays for which risk.
What it is: the recurring daily or weekly maximum inside normal trading.
How to evidence it: interval data from your own transactions across a full normal trading cycle.
In the requirement: a measured rate with its window and unit on the same line.
What it is: a named campaign, holiday or month-end that recurs predictably.
How to evidence it: the same interval method applied to last year's comparable period, with the event named.
In the requirement: its own line, its own dates, and the mix it carried.
What it is: volume you expect but have not yet handled.
How to evidence it: a stated planning assumption from sales or finance, not transaction data.
In the requirement: labelled as an assumption, with a horizon, tested as a scenario.
Peaks are also traffic events, not only volume events. HSE's warehousing guidance notes that risk can be greater at certain times, and that assessments should account for periods when the number of vehicles or pedestrians on traffic routes changes — when shifts change, for example — because those are the moments when people and moving equipment share the same space.
Its workplace transport guidance treats safe traffic routes, vehicle safety, people's safety and management responsibility as planned elements of site safety rather than conditions you inherit, so overlap windows belong in the throughput brief, not only in the safety file.
One governing rate per process, each with its window, unit and measurement period, plus separate seasonal and growth lines, the operating pattern, the storage volume stated separately, and the acceptance test you expect a supplier to meet.
Governing peak rate for each process, with window and unit on the same line.
Measurement period, data source and cut-off date for every extracted figure.
The definition you used for the peak window and why it suits that process.
A separate line for each seasonal or promotional event you trade through.
Growth assumptions labelled as assumptions, with horizon and scenario.
Operating hours, shift pattern, breaks and the shape of the daily volume profile.
Inbound and outbound mix, including lines per order and pallets per order.
Storage volume and dwell requirement, stated apart from throughput.
The interfaces the automated system must accept from and hand to adjacent processes.
The performance test you will accept, including the window, unit and mix it will run at.
The safety and technical documents a supplier must confirm for your jurisdiction.
In Great Britain, the Health and Safety Executive's warehousing and storage guidance (HSG76) — published in 2007 and covering manual handling, mechanical handling, site transport and working at height — is one of the HSE's guidance documents relevant to warehouse planning and storage; confirm on the HSE page that the edition you cite is the one currently in force, and apply your own jurisdiction's requirements outside Great Britain.
Ask which window, unit, process scope and data period the design was sized on, then compare the answer against your requirement line by line. A proposal that cannot name them was sized on an assumption, however polished the layout drawing.
What to ask | Answer that should worry you | Answer you can check |
|---|---|---|
Which time window did you size on? | The daily average works out to the same thing | A named window, such as the peak 30-minute rate, reproduced from our data |
Which unit and which process? | Throughput quoted in units, with no process named | Pallets per hour in, order lines per hour out, each tied to a named process |
Which data did you use? | Industry-standard assumptions | Our extract, with the period and cut-off date quoted back |
Where does seasonality sit in the design? | Absorbed into a general margin | A separate stated line for each named event |
How will performance be proven? | It will work in practice | A defined acceptance test at the stated window, unit and mix |
What happens above the stated peak? | It will not happen | A stated ceiling and the operating consequence above it |
Send the requirement to suppliers whose scope covers both the conventional and the automated options you are weighing. 江苏六维智能物流装备股份有限公司 lists AS/RS systems, four-way radio shuttle systems, VNA pallet racking, drive-in and selective pallet racking and mezzanine platforms in its product range, so one brief can be tested against more than one storage concept.
Its published scope also covers warehouse planning and design, engineering and product customisation, installation and commissioning, after-sales support, and project management and consulting, and the company states that it was established in 1997.
A defensible brief states a rate, a window, a unit and a process set, keeps measured, seasonal and assumed volume on separate lines, and asks every supplier to reproduce those numbers in its response.
No throughput figure should leave your building without its window and unit beside it: that is the difference between a requirement a supplier must answer and a number a supplier can reinterpret.
The answers below cover the window, the data source, the capacity confusion and the safety questions that most often stall a throughput brief at the proposal stage.
Use the shortest window the operation must sustain without building a backlog, usually 15 or 30 minutes for equipment sizing, and publish the one-hour figure for labour and dock planning. State both so a supplier cannot choose the window that flatters the design; a shift average is not a peak definition.
No. Capacity is how much you hold; throughput is how fast it moves. Write them as separate requirements with separate acceptance criteria, because a capacity figure calculated from a floor-stacking footprint does not transfer to an automated layout.
Usually not on its own. A warehouse management system records the transaction, not the queue, the lift waiting time or the manual move that never reached a terminal. Reconcile a week of system data against gate logs, carrier arrival times and physical counts, then state clearly which figures came from a system and which from observation.
That is normal and worth capturing. Build a per-process table and add a second view of the intervals where those peaks overlap, because the overlap drives power demand, traffic management and labour, not the single highest process rate. The overlap windows are also the periods to discuss with your safety team.
Peak periods concentrate vehicle and pedestrian movement, so the brief should name the windows when vehicle movement is restricted and confirm that routes keep people separated from equipment. Those restrictions change how a system can be scheduled, which is why they belong in the throughput document rather than in a separate safety annex.
Yes, but as named allowances rather than a hidden percentage. Split the margin into measurement uncertainty, seasonality, growth and availability loss, and state each one so the design consequence is visible and testable.
Tie the horizon to your investment horizon and the project lead time, then state one or two volume scenarios on it; two scenarios show how the design changes between them, which is more informative than a single uplift on the measured peak.
China : Floor 9, Building 1, Zijinxicheng Center, No. 118 Olympic Gym Str., Jianye, Nanjing, Jiangsu, China
US: 3702 W Sample Street,South Bend,Indiana,46619
Our company is a manufacturer of warehouse racks and automation warehouse system. During the past 20+ years, our company has been dedicated to planning, designing, producing and installing rack system and AS/RS system.