U-Shaped vs I-Shaped vs L-Shaped Warehouse Flow

A 3PL signs a lease on a U-shaped building for a grocery client because "U-flow is more efficient". Six months in, the early produce trucks and the first store wave both want the same strip of dock between 05:00 and 07:00, loaders and unloaders cross the same staging lanes, and store departures slip by forty minutes most mornings. Down the road, a through-flow building run by a competitor has the opposite problem: every slow-moving pallet travels the full depth of the building twice.

Neither shape is better in general. Each one wins under particular conditions that can be measured before a lease is signed: how skewed the SKU movement is, whether inbound and outbound peaks overlap, how much freight crosses the dock without being stored, and where the building can grow. With those four numbers, the choice between U-shaped, I-shaped and L-shaped flow becomes arithmetic rather than habit. This guide belongs to our logistics and supply chain library.

• What this guide covers: how each shape places receiving, storage and shipping; a travel formula with worked distances; dock door and truck court sizing; corners in L-shaped buildings; cross-dock fit; and expansion paths.
• What it adds: a timestamped dock-collision trace, a dollar and rupee comparison of the three shapes for one building, the checks that stay fixed versus those simulation assists, and a pre-lease selection checklist.

How U, I and L shapes place receiving, storage and shipping

The flow shape is defined by where the receiving and shipping doors sit relative to each other: on the same face (U), on opposite faces (I, also called through flow), or on two adjacent faces (L). Storage sits between them, and the shape decides which storage locations are convenient.

Bartholdi and Hackman's Georgia Tech text, Warehouse & Distribution Science, describes the effect clearly. In a through-flow building, every location along a line between the two docks is about equally convenient, so the design is conservative: many reasonable locations and few excellent ones. In a U-flow building, the few locations near the middle of the shared dock are very convenient for both putaway and retrieval, while the far corners become worse than any location in a through-flow plan.

Our article on the warehouse design process places this choice in the full design sequence and summarises the shapes in one section. This article goes deeper into that one decision: the travel arithmetic, the door and truck court sizes, and the cost of each option for a single building.

Travel arithmetic: when U-flow beats through flow and when it ties

U-flow saves lift-truck travel only when the moves concentrate in storage close to the dock. With single-command cycles, U-flow beats through flow exactly when the move-weighted average storage depth is less than half the building depth; with flat demand the two shapes tie.

Travel per pallet, single-command cycles, rectilinear distances
  (one putaway cycle plus one retrieval cycle, out and back)

  Through flow (I):  T_I = 2 x D + 4 x X
  U-flow:            T_U = 4 x SUM(p_k x y_k) + 4 x X
  L-flow:            T_L = 2 x SUM(p_k x y_k) + 2 x SUM(p_k x e_k) + 4 x X

    D   = depth between the receiving and shipping faces (m)
    y_k = average distance of storage class k from the receiving face (m)
    e_k = average distance of storage class k from the shipping face (m, L-flow)
    p_k = share of pallet moves in class k; SUM(p_k) = 1
    X   = average lateral offset between door and slot, one way (m)

  U-flow is shorter than through flow when SUM(p_k x y_k) < D / 2

Take a building 150 m along the dock face and 100 m deep, with an average lateral offset of 15 m. Through flow gives 2 x 100 + 60 = 260 m per pallet, wherever the pallet is stored. In a U-flow building with flat demand the average depth is 50 m, so U-flow also gives 260 m. The difference appears only once the warehouse slots fast movers near the dock.

The assumptions behind the table are illustrative. Class A sits at an average 10 m from the dock, B at 40 m and C at 80 m. For the L-flow column, receiving runs along the 150 m face and shipping along the 100 m face, with A slotted in a 30 m square at the corner. With strong skew, U-flow cuts travel by about 42 percent compared with through flow; with moderate skew the saving falls to about 18 percent.

Pallet travel in a U-flow warehouse drops below through-flow travel only when WMS move history shows a move-weighted storage depth under half the building depth, D/2. In the 100 m deep example, a 70 percent A-class share cuts travel per pallet from 260 m to 152 m, with aisle clearances per 29 CFR 1910.176(a).

U-flow has one more travel advantage that the single-command formula leaves out. Because putaway and retrieval start and end at the same dock, a lift truck can store one pallet and pick up another on the way back, a dual-command cycle that removes much of the empty return trip. In a through-flow building those two tasks begin on opposite sides, so the interleaving gain is smaller.

Dock door counts and truck court depth for each flow shape

Size doors from the busiest hour, not the daily total: door count equals peak trucks per hour times average hours at the door, divided by a target utilisation such as 85 percent. A shared U-flow dock needs fewer doors than two separate faces whenever inbound and outbound peaks fall at different times.

In the example building, inbound peaks at 12 trucks an hour between 06:00 and 10:00 with one hour at the door, and outbound peaks at 10 trucks an hour between 16:00 and 20:00 with 1.25 hours at the door. Off-peak, about 3 inbound trucks arrive during the outbound peak and 2 outbound trucks load during the inbound peak.

• Through flow, inbound face: 12 x 1.0 / 0.85 = 14.1, so 15 doors.
• Through flow, outbound face: 10 x 1.25 / 0.85 = 14.7, so 15 doors, giving 30 in total.
• U-flow, inbound peak window: (12 x 1.0 + 2 x 1.25) / 0.85 = 17.1, so 18 doors.
• U-flow, outbound peak window: (10 x 1.25 + 3 x 1.0) / 0.85 = 18.2, so 19 doors are needed.
• U-flow if the peaks overlap: (12 x 1.0 + 10 x 1.25) / 0.85 = 28.8, which 19 doors cannot absorb.

The truck court in front of each door face must let the longest vehicle using the site back square onto a door. In the US, 23 CFR 658.13(b)(1) bars States from limiting semitrailers to less than 48 feet on the National Network, and Bartholdi and Hackman note that standard large trailers are 48 or 53 feet. This example assumes a 40 m truck court in the US and 35 m in India; confirm both with a swept-path study for the actual vehicles.

Dock door counts for a shared U-flow dock come from yard management appointment data: peak trucks per hour times average door hours, divided by an 85 percent utilisation target. Truck court depth follows trailer length, and 23 CFR 658.13(b)(1) bars US States from limiting semitrailers below 48 feet on the National Network.

Through flow doubles the truck court and the gatehouse traffic pattern, but it also keeps inbound and outbound drivers apart, which matters at sites with heavy live unloading. If most inbound trailers are dropped rather than live-unloaded, the yard rather than the dock absorbs the peak; our comparison of drop and hook versus live load shows how that changes the door hours used in the formula.

L-shaped flow and the hidden cost of a corner

An L-shaped dock pays for its corner twice: doors are lost at the inside corner, and freight congests at the outside corner. Choose it when the site forces it, not as a design preference.

Bartholdi and Hackman explain both losses for cross-dock terminals, and the same geometry applies to any building with doors on two adjacent faces. On the outside of a corner, each door has less floor in front of it to stage freight, which creates congestion. On the inside of a corner, trailers interfere with each other in the yard, so door positions are lost and the dock must be longer for the same door count, which lengthens freight travel.

• Where L-flow is forced: a corner plot with road frontage on two sides, or an existing building whose only usable truck courts sit on adjacent faces.
• Where it performs well: strongly skewed demand slotted near the corner, as the 181 m strong-skew figure in the travel table shows.
• Where it performs badly: flat demand, because storage far from the corner is far from both docks; the flat-demand figure of 310 m is the worst in the table.
• Door count check: at an assumed 4 m door spacing, a 100 m face offers about 25 door positions before the inside-corner loss, so count usable positions on the drawing rather than dividing face length by spacing.

Cross-dock fit: when through flow beats a shared dock

For buildings where most freight crosses the dock without being stored, a long, narrow I-shaped dock is the default. Bartholdi and Hackman describe the typical cross-dock as a long rectangle shaped like the letter I, generally about 120 feet (36.6 m) wide, and that most docks have 80 to 120 doors, with few above about 200.

Their width figure has a logic worth reusing. About 100 feet covers space for the trailer on each side of the dock to stage its freight, and two 10-foot travel aisles run along the length; docks as narrow as 80 feet work only when most freight can be loaded without staging. They also advise yard parking for two trailers per door, so each destination can have a trailer at the door plus one full and one empty in the yard.

Cross-dock buildings are usually I-shaped, about 36.6 m (120 ft) wide, with 80 to 120 doors, according to Bartholdi and Hackman's Warehouse & Distribution Science. L-shaped and U-shaped docks lose door positions at inside corners, and every option must keep the safe loading-dock clearances 29 CFR 1910.176(a) requires.

• Deep storage run as through flow: every cross-dock pallet crosses the full depth, 100 m in the example, whereas on a shared U-flow dock it travels along the face.
• Long docks and central doors: traffic past the middle doors grows fast as doors are added, because central doors are the best destinations for everyone; with 2n doors and equal flows, the flow passing between doors i and i + 1 is i(2n − i).

Deciding which receipts should cross-dock at all is a separate, order-level question. Our article on cross-dock eligibility covers matching received stock to confirmed outbound demand; the building shape decides how cheaply the eligible freight can then move from one door to another.

Expansion paths each flow shape leaves open or closes

A U-flow building can grow on three sides without touching its dock; a through-flow building grows best along its length, and adding depth raises travel for every stored pallet. Decide the expansion direction before fixing the shape, because it determines where land must be reserved.

The arithmetic is quick. Adding 30 m of depth to the through-flow example raises travel from 260 m to 2 x 130 + 60 = 320 m per pallet, an increase of about 23 percent for every pallet stored. Adding the same 30 m to the back of the U-flow building and placing only C-class stock there moves the C average from 80 m to about 95 m; with strong skew, travel rises from 152 m to about 158 m, roughly 4 percent.

• U-flow growth: extend the back or either end wall; the dock and truck court stay as they are, but door count does not grow unless the dock face is lengthened.
• Through-flow growth: extend both dock faces lengthwise, which adds doors on both sides and keeps travel per pallet unchanged; this needs land at one or both ends.
• L-flow growth: extend the two walls without doors; lengthening either dock face pushes the far end of that face further from the corner.
• Land reserve: mark the expansion zone on the site plan and keep drainage, fire hydrant mains and utility routes out of it.

A shared dock at 05:40 when inbound and outbound peaks collide

A U-flow dock sized for separate peaks fails the morning the peaks overlap, as the trace below shows. The site, trailer numbers and system transactions are illustrative; the door arithmetic follows the example above.

1. 04:50, late inbound arrivals: four produce trucks booked for 03:00 arrive after a highway closure. The yard system records gate-in G-4471 to G-4474 and assigns doors 3, 5, 6 and 9, which are still free.
2. 05:30, store wave release: the WMS releases store wave SW-218 needing 12 outbound doors; only 8 are free, so store trailers T-5502 to T-5505 wait in the yard.
3. 05:40, staging conflict: unloaders from doors 3 to 9 and loaders for the store wave share the same 20 m staging strip. Pallet jacks queue, and putaway of the produce stops to clear a lane.
4. 06:55, late departures: the last store trailer leaves 45 minutes after its 06:10 plan, and three stores miss their 07:00 delivery windows.
5. Controlled version: a time-band rule reserves doors 1 to 12 for outbound between 05:00 and 08:00. The yard system diverts late inbound loads to drop spots, and they are unloaded from 08:00 when outbound demand falls.

The fix needed yard space more than doors. A time-band rule only works if late arrivals have somewhere to wait, so the trailer pool and drop spots must be sized with the dock; our guide to trailer pool size calculation sets out that sizing. Without them, the yard fills and drivers wait on the road outside the gate.

Comparing the three layouts in dollars and rupees for one DC

For the example building with strong demand skew, U-flow costs least to run and least to build, provided inbound and outbound peaks stay apart. The figures below are illustrative and assume 1,800 pallets stored and 1,800 retrieved a day for 300 days a year.

Travel hours come from the travel table at an effective lift-truck speed of 1.8 m/s, including acceleration and turns; handling time is the same in every shape and is left out. Hourly rates combine operator and truck cost: $46 in the US and ₹440 in India. Door costs cover the door, leveller, seal and vehicle restraint; truck court costs cover paving only.

Read the table with its conditions attached. With flat demand, the U-flow travel advantage disappears and the comparison turns on doors, paving and congestion risk. If the two peaks overlap, the U-flow door count rises to 29 and most of its build saving goes. A fair business case therefore shows each shape under at least two demand profiles and two arrival patterns, not one.

Who decides the flow shape and which checks software can run

The flow shape is an operations decision with real-estate and finance consequences, so the inputs come from different owners and the final call needs both operations and finance sign-off. Fixed design checks run on every option; simulation and AI tools help generate and test options but do not choose.

Layout selection checklist before signing a lease or design brief

Complete every line below for at least two candidate shapes before a lease or design brief is signed. A line that cannot be completed is a data gap to close, not a reason to fall back on a rule of thumb.

If you are comparing candidate buildings and want the move-weighted depth, door count and peak-overlap tests run on your own WMS and yard data, share the site drawings and twelve months of movement history through our request for proposal page, and we will set out the comparison for your operations and finance teams.