Warehouse Design Process: From Order Profile to Layout

Eighteen months after opening, the new distribution centre is already renting overflow space, and every afternoon trailers queue in the yard for four outbound doors. The layout looked generous on the drawing. It was sized to last year's average inventory and an average day's dispatches, so the first peak season exposed both shortfalls at once.

This article sets out the warehouse design process in the order a design team should work: data, order profile, design-year volumes, storage media, doors, flow pattern, space, lift trucks, fire and egress limits, simulation and phasing. It includes reusable formulas for sizing pallet positions and dock doors, costed in dollars and rupees, plus the US and Indian fire rules that most often force a redraw. The subject is the physical building; for the software that later runs it, see our WMS architecture guide.

Collect the data a warehouse design actually depends on

A warehouse design is only as good as the 12 to 24 months of transaction history behind it. Before anyone opens a drawing tool, extract order lines, receipts, inventory snapshots and item dimensions, and agree which year the building must serve.

Most of the data already sits in the current WMS or ERP. The problem is quality: missing carton dimensions, units of measure that changed halfway through the year, and returns booked as negative orders. An hour spent cleaning the item master saves weeks of arguing about why a simulation disagrees with the floor.

• Order line history: order number, date, time, item, quantity and unit of measure for every line, ideally covering at least one full peak season.
• Receipt history: inbound lines by date with supplier, pallet count and whether pallets arrived as single-item or mixed loads.
• Inventory snapshots: end-of-day or end-of-week on-hand quantity per item, so storage demand can be measured rather than guessed.
• Item master: case and pallet dimensions, weights, units per case, cases per pallet, and flags for hazardous, temperature-controlled or high-value goods.
• Business plan: expected growth by channel, new customers or product lines, and service promises such as same-day dispatch cutoffs.

Agree a design horizon at this stage, commonly five to ten years depending on the lease or build cost. Every later number in the process, from pallet positions to door count, is calculated for that design year rather than for today.

Order profile, ABC and cube-movement analysis before any drawing

The order profile tells you how goods leave the building, and it drives the picking area far more than total volume does. Measure lines per order, units per line and how picks concentrate in a small share of items before choosing any storage or picking method.

ABC analysis by pick lines, not by sales value, shows which items generate most of the walking. A cube-movement view adds the second half: an item that sells fast in small cubes belongs in carton flow near pack, while a slow item with large cube may belong in reserve pallet racking. Our article on deciding what deserves a forward pick face applies the same idea at slot level.

Look at distributions rather than averages throughout. A site averaging three lines per order may in fact ship mostly one-line e-commerce orders plus a few hundred large store orders, and the right design treats those as two separate flows. The comparison in our guide to batch and wave picking shows how much the method choice depends on that split.

Setting the design year, growth rate and peak factor

Size the building for the design-year peak, not for today's average. The two numbers that matter are a growth rate agreed with the business and a peak factor measured from your own history.

Growth is usually compounded from the base year to the design year. The peak factor compares the busiest period with the average. Measure it separately for storage (peak inventory month against average month) and for throughput (peak week or peak day against the average), because a business can have a quiet inventory profile and a sharp dispatch peak, or the reverse.

Design-year storage demand
  P_design = I_avg x F_peak x (1 + g)^n
  N_pos    = ceil( P_design / U_target )

Dock doors per direction
  T_day    = ceil( Q_peakday / L_trailer )
  D        = ceil( T_day x h_door / ( H_window x U_door ) )

Where:
  I_avg     average pallets on hand in the base year
  F_peak    peak-month inventory / average-month inventory (measured, e.g. 1.10 to 1.40)
  g         agreed annual growth rate (decimal)
  n         years from base year to design year
  U_target  planned storage occupancy, below 1.00 so slots exist for receipts and moves
  Q_peakday design-year pallets on the peak day for that direction
  L_trailer average pallets per trailer for that direction
  h_door    hours a trailer occupies a door, including spotting, checking and release
  H_window  operating hours in which that direction can use doors
  U_door    planned door utilisation, below 1.00 to absorb late arrivals

Treat the peak factor range in the code block as an example of what to measure, not as a default. Use your own seasonal pattern, and run a sensitivity case with a higher growth rate so the board can see what an extra year of runway costs.

Matching storage media to each SKU class and handling unit

Choose storage media item class by item class, using pallets per item and pick frequency, not one rack type for the whole building. The right mix usually combines selective racking for range, denser systems for deep stock, and flow or shelving for fast each-picks.

The decision turns on two numbers per item: how many pallets you hold, and how often you touch them. An item with one or two pallets needs direct access to each pallet. An item with thirty pallets of one lot can sit in a deep lane because the pallets are interchangeable, as long as the lot and date rules allow last-in, first-out or the system supports first-in, first-out flow.

Pallet dimensions and weights from the item master set beam lengths and load ratings. If you handle both 1,200 by 1,000 mm and 48 by 40 inch pallets, decide whether bays are sized for the larger footprint everywhere or split into zones, because that one choice changes position counts across the whole rack layout.

Sizing pallet positions and dock doors with a worked calculation

Pallet positions come from design-year peak inventory divided by planned occupancy; dock doors come from peak-day trailers multiplied by door time and divided by usable door hours. Both formulas are simple, and both fail when the inputs are averages.

The worked example uses illustrative inputs. Base-year average inventory is 6,400 pallets, the peak-month factor is 1.25, growth is 6% a year for 5 years, and planned occupancy is 90%. Design-year peak inventory is 6,400 × 1.25 × 1.06^5, about 10,706 pallets. Dividing by 0.90 gives 11,896, so the design carries 11,900 positions.

• Inbound doors: 520 pallets on the peak day at 26 per trailer is 20 trailers; at 1.5 door-hours each over a 16-hour window at 75% utilisation, 30 ÷ 12 = 2.5, so 3 doors.
• Outbound doors: 680 pallets at 22 per trailer is 31 trailers; at 1.25 door-hours each over a 10-hour window at 75%, 38.75 ÷ 7.5 = 5.2, so 6 doors.
• Cutoff check: if 40% of outbound trailers must leave in the last 3 hours, 12.4 trailers need 15.5 door-hours in 3 hours, which also needs 6 doors.
• Floor footprint: 11,900 positions over 5 beam levels is 2,380 floor positions; at an assumed 2.6 m² gross per floor position for selective racking with reach-truck aisles, the storage zone needs about 6,190 m².

The cost view below prices the result and the shortfall from ignoring peak. Unit costs are illustrative placeholders for budgeting the method, not market quotes; replace them with your own tender prices.

Pallet-position sizing in a warehouse design starts from WMS or ERP inventory history raised to the design-year peak and divided by planned occupancy below 100%. OSHA 29 CFR 1910.159(c)(10) requires at least 18 inches of vertical clearance between sprinklers and the material below, which limits the top beam level.

Dock doors deserve the cutoff check every time. Outbound trailers rarely spread evenly across the day; they bunch before carrier departures, and a door count sized on the daily total can still leave trailers waiting. Our guide to dock appointment scheduling covers how to protect those doors once the building is running.

U-shaped or through flow: picking the building's movement pattern

Pick a U-shaped flow when inbound and outbound volumes are moderate and flexibility matters; pick a through flow when volumes are high and cross-docking is significant. The pattern fixes where docks, staging and fast-moving stock sit, so it must be settled before detailed racking.

In a U-shaped layout, receiving and shipping share one building face. Doors can switch between inbound and outbound as the mix changes, and fast movers sit near the shared dock to keep travel short. In a through, or I-shaped, layout goods enter on one side and leave on the other, which separates traffic but makes every pallet travel the full building depth.

Site shape and road access often decide the pattern before operations do. A plot with access from one road only usually forces a U-shape, while a deep plot with roads front and back makes through flow possible. Check yard depth for trailer manoeuvring on each face, because a well-planned interior is useless if trucks cannot reach the doors.

Space calculation and lift truck selection for the aisle layout

The lift truck choice sets aisle width, and aisle width sets how much of the floor is storage. Decide the truck type and the rack height together, then calculate space for every functional zone, not just racking.

Counterbalance trucks need the widest aisles, reach trucks narrower ones, and very narrow aisle turret trucks the narrowest, usually with guidance rails or wires and a separate truck to bring pallets to the aisle end. Take the right-angle stacking aisle figure for the actual truck model and load from its manufacturer, then add the site's agreed operating margin.

• Storage zone: racking plus aisles, from the pallet-position calculation and the chosen truck.
• Receiving and staging: enough floor to hold the largest expected trailer load per inbound door, plus break-down space for mixed pallets.
• Forward pick and pack: shelving, carton flow and pack stations sized from the order profile and peak-hour lines.
• Outbound staging: lanes holding at least one full trailer per outbound door before its departure time.
• Support space: battery charging, returns processing, offices, welfare facilities and maintenance.

Aisle planning in a warehouse design follows the chosen lift truck, and OSHA 29 CFR 1910.176(a) requires sufficient safe clearances wherever mechanical handling equipment turns or passes, with permanent aisles marked. In the worked example, adding 0.2 m to each of 30 rack aisles consumes 6 m of building width.

OSHA 1910.176 also requires tiered storage to be stacked and limited in height so it stays stable, and clearance signs where clearance limits exist. Design for those conditions from the start: marked aisles and staging lanes drawn on the plan are far easier to keep clear than ones improvised after opening.

Fire, sprinkler and egress rules that reshape the warehouse layout

Fire protection and egress rules set hard limits on rack height, commodity placement, flue spaces and exit doors, so involve the fire engineer before rack elevations are frozen. Changing commodity class after racking is ordered is one of the most expensive late design changes.

In the US, NFPA 13, the Standard for the Installation of Sprinkler Systems, bases storage protection on the commodity class (Class I to IV, or Group A plastics), the storage arrangement (solid-piled, palletised, shelf or rack) and the storage height. Ask the fire engineer to classify the unit loads you will actually store, with their packaging and pallets, rather than the product name alone. The edition your authority has adopted governs the details.

• Commodity class fixed: agree the class of each storage zone with the fire engineer and insurer, including pallet and packaging material.
• Storage height capped: set the top beam level from the sprinkler design and the 18-inch clearance, not from the building's clear height alone.
• Rack spacing confirmed: show the gaps between back-to-back racks and between pallets on the drawings, and have the fire engineer check them against the sprinkler design.
• Exits counted: place exit doors and routes before racking runs are fixed, so long racking runs do not block a required path.

OSHA adds clearance rules that the layout must respect. Under 1910.159(c)(10), material must stay at least 18 inches below sprinklers. Local building and fire codes adopted by your jurisdiction then add their own requirements, so treat the federal rules as a floor rather than the whole list.

In India, the National Building Code of India 2016 Part 4, Fire and Life Safety, classifies warehouses, cold stores, freight depots and transit sheds as Group H storage buildings under clause 3.1.9. Table 7 Note 20 of the 2016 Part 4 states that Group H buildings above 15 m in height are not permitted, which caps building height on a high-bay design.

India's National Building Code 2016 Part 4 classifies warehouses as Group H storage under clause 3.1.9, and clause 6.8.1.2 requires at least two exit access doors for any storage area above 1,400 m² gross or with more than 10 persons normally present. NBCS 2026 Part F repeats both clauses.

BIS has since published the National Building Construction Standards 2026, which moves fire and life safety into Part F and states that it is guidance for state governments and local authorities. Fire NOCs are issued under state fire services acts and building bye-laws, so confirm which edition your state authority applies before fixing heights, exits and hydrant layouts.

Validating the layout with simulation and a phased build plan

Test the layout against the design-year peak week before committing capital, then phase the build so the first phase opens with enough capacity for its own peak. Simulation does not need to be elaborate; it needs to use real order and receipt data.

A spreadsheet model replaying a peak week hour by hour catches most door and staging problems. Discrete-event simulation adds value where automation, conveyors or complex pick paths interact. Either way, replay actual order lines and receipts from the history, scaled to the design year, rather than evenly spread averages. The illustrative trace below shows a missed cutoff check after go-live.

1. Day 3, 15:10 Yard queue: the yard system shows 11 outbound trailers booked into appointment block APT-OB-1700 for the 17:00 carrier departures, against 4 outbound doors built.
2. 15:40 Staging overflow: 62 picked pallets for waves W-0412 to W-0415 spill from the staging lanes into marked cross-aisle 14.
3. 16:25 Putaway stalls: reach trucks cannot pass the blocked aisle, and open putaway tasks in the WMS queue reach 140.
4. 17:55 Cutoff missed: 3 trailers leave after carrier departure time, and shipments SH-88201 to SH-88219 are rebooked for the next day.
5. Week 2 fix: two inbound doors are converted to flex doors and outbound waves are staggered, a change the cutoff check would have shown before construction.

Phasing then spreads capital across the design horizon. Build the slab, structure, sprinkler mains and dock positions for the design year, because those are costly to extend later. Racking, automation and some dock equipment can be added in stages as volume arrives.

Who owns each design decision and where software tools assist

Give every design decision one preparer and one approver, and keep calculations that carry legal weight with qualified people. Software can speed up data cleaning and option generation, but sprinkler design and egress compliance stay with licensed engineers and the authority.

The first table assigns ownership for a typical new-build or major retrofit. Titles vary by company; what matters is that finance signs the design-year volumes it will be held to, and that the fire engineer sees the commodity assumptions before racking is ordered.

The second table separates fixed calculations from places where software or AI tools help a person decide. Name the task precisely: "AI for design" means little, but "clustering 40,000 orders into profiles" is a clear, checkable job.

For more guides on warehouse operations and transport, visit the logistics and supply chain hub. If you are scoping a new facility or retrofit and want the sizing model, simulation inputs and vendor brief built from your own data, you can share your requirements with us.