Honeycombing in Pallet Racking: Measuring Lost Space

The occupancy report says 91% of pallet positions are full, yet receiving has 40 pallets parked on the dock floor and the putaway screen keeps returning "no location found". Walk the aisles and the reason is visible: drive-in lanes holding one pallet of a slow SKU, and 1.1-metre loads sitting in openings built for 1.7-metre loads. The positions are taken; the space is not used.

That gap is honeycombing, and it can be measured as precisely as occupancy. With the location master, the pallet heights captured at receiving and a daily inventory snapshot, a warehouse engineer can express vertical and horizontal honeycomb as two separate percentages, convert them into lost positions and money, and decide which fix pays back: re-pitching beams, changing lane depth, or changing the rules that assign SKUs to lanes.

• What this guide covers: definitions you can measure, the vertical air-gap formula, the average loss in a deep lane, the square-root rule for lane depth, and a snapshot query that computes both losses.
• What it adds: a worked 4,800-position layout priced in US dollars and Indian rupees, an incident trace, a split between fixed WMS rules and assisted decisions, and a redesign checklist.

Air above the load and empty slots behind it: two separate losses

Treat honeycombing as two numbers, because they have different causes and different fixes. Vertical honeycomb is the height inside an opening that no load can use because the load below the beam is shorter than the opening. Horizontal honeycomb is the pallet positions inside a deep lane that sit empty but cannot take another SKU until the lane clears.

Vertical loss lives in selective racking and in the top tier of block stacks; it is fixed with beam elevations, opening sizes and slotting by height. Horizontal loss lives in drive-in, push-back, pallet-flow and floor-stacked lanes; it is fixed with lane depth, lane assignment rules and how many lots may share a lane. A single "space lost" figure hides which lever to pull.

Our article on space utilization versus cube utilization explains why occupancy and cube fill are different ratios. This one goes a step further: it isolates the honeycomb part of that gap, measures it from system data rather than a walk-through, and prices the fixes. Rack families themselves are compared in our pallet racking types guide.

Calculating vertical honeycomb from beam pitch and load height

Vertical honeycomb is the share of usable opening height that sits above the load and below the next beam after allowing for lift clearance. Measure it per opening, then sum, because averaging percentages across mixed opening sizes misstates the loss. The inputs are the clear opening height from the layout drawing, the measured load height and the lift-off clearance your rack supplier specified.

Take a selective rack bay with 1,800 mm beam pitch and 120 mm deep beams, giving a 1,680 mm clear opening. A 1,150 mm load plus a 100 mm lift clearance needs 1,250 mm, so 430 mm of each opening is air: 25.6% of the clear height. Across 1,800 such openings, each holding two pallets, that adds up to 774 metres of unused height, but the air only becomes capacity if a whole level can be added.

The practical question is therefore how many whole levels fit. Using the same bay: four levels at 1,800 mm pitch occupy 7,200 mm of upright. At a 1,370 mm pitch, which is 1,250 mm of need plus a 120 mm beam, the same height holds five levels with 350 mm to spare. That is one extra level per bay, a 25% gain in those bays, subject to sprinkler clearance, floor loading and frame capacity checks.

Vertical honeycomb in selective pallet racking equals clear opening height minus load height minus lift clearance, summed over every opening in the WMS location master. In a 1,680 mm opening holding a 1,150 mm load with 100 mm clearance, 25.6% of height is air; recovery requires manufacturer approval and a revised load plaque.

Why a deep lane averages (k − 1) / 2 empty positions

In a lane k pallets deep that holds one SKU and empties at a steady rate, the average number of positions that are empty but locked to that SKU is (k − 1) / 2 over the lane's life. The first slot freed stays unusable for the longest time, the last one for no time at all, and the average of that staircase is half of one less than the depth.

The derivation comes from Bartholdi and Hackman's open textbook, Warehouse and Distribution Science, which charges each lane its storage positions plus half the aisle in front of it. A 6-deep drive-in lane therefore wastes 2.5 positions on average, and a 2-deep push-back lane wastes 0.5. The rule holds whether pallets leave whole or are picked down carton by carton, because the slot only frees when the pallet in it is gone.

• Uniform depletion assumption: the (k − 1) / 2 average assumes steady demand; a SKU that sells half its lane in a week and then stalls loses more.
• Single-SKU lanes: mixing SKUs in a lane removes horizontal honeycomb but adds double handling to reach a buried pallet.
• Lot-pure lanes: where first-expiry rules forbid two lots in one lane, every lot behaves as a separate SKU, so losses rise with the number of lots on hand.
• Stack height: in floor stacking, the depth loss applies per column, so stacks three high multiply each empty floor position by three pallets of unusable cube.

Deeper lanes save aisle space and lose more inside the lane, so depth is a trade-off, not a maximum. That trade-off has a closed-form answer, which the next section works through. For the operating rules that decide which pallet goes into which lane in the first place, see our guide to directed putaway rules.

Setting lane depth with the square-root rule from aisle width

The most space-efficient lane depth for a SKU is the square root of aisle width times pallets received, divided by twice the stack height. Bartholdi and Hackman derive it by adding the in-lane waste to the shared aisle cost and minimising the sum; their Theorem 6.1 gives k* = √(a·q / 2z) for one SKU.

Here a is the aisle width measured in pallet depths, q is the pallets in a typical receipt, and z is how many pallets high the SKU stacks. With a 3.4-metre aisle and 1.2-metre-deep pallets, a is about 2.8. A SKU received in lots of 48 pallets stacked three high gives k* = √(2.8 × 48 / 6) = √22.4 ≈ 4.7, so a 4-deep or 5-deep lane.

Honeycomb measurement and lane-depth formulas (illustrative units: mm, pallets)

Vertical honeycomb share for a set of openings:
  VH = SUM_i max(0, H_i - h_i - c) / SUM_i H_i
    H_i = clear opening height of location i (beam pitch minus beam depth)
    h_i = measured height of the load in location i
    c   = lift-off clearance from the rack layout drawing (example uses 100 mm)

Horizontal honeycomb share across deep lanes on a snapshot day:
  HH = SUM_l (k_l - n_l) * [n_l > 0] / SUM_l k_l
    k_l = depth of lane l in pallet positions
    n_l = pallets currently in lane l (lane holds one SKU or lot)
    [n_l > 0] = 1 if the lane is active, else 0 (empty lanes are free, not lost)

Expected in-lane loss with steady depletion:  (k - 1) / 2 positions per active lane

Space-efficient lane depth (Bartholdi and Hackman, Theorem 6.1):
  k* = SQRT( a * q / (2 * z) )
    a = aisle width in pallet depths, q = pallets per receipt, z = stack height
Shared depth for n SKUs (Theorem 6.2):
  k* = SQRT( (a / (2 * n)) * SUM_j (q_j / z_j) )

Few buildings let each SKU have its own depth, so the shared version averages q/z across the SKUs assigned to a zone. If a zone holds three SKUs with q/z values of 16, 10 and 18, and a is 2.8, the shared depth is √(2.8 / 6 × 44) ≈ 4.5. Fast, large-lot SKUs pull the answer deeper; many small lots pull it shallower.

Lane depth in floor-stacked or drive-in storage should follow Bartholdi and Hackman's square-root rule k* = √(a·q/2z), not the deepest lane the building allows. A 2.8-pallet aisle with 48-pallet receipts stacked three high gives about 4.7 deep, and each active lane then loses (k − 1)/2 positions.

Turning a daily WMS snapshot into honeycomb percentages

Measure honeycomb from data every day rather than from an annual walk-through, because it swings with receipts and seasonality. A snapshot at a fixed time, such as 06:00 before the first wave, gives one vertical and one horizontal figure per zone; a four-week average is the number to manage against.

Three data sets are needed. The location master must hold clear opening height and lane depth for every location, which many WMS installations only hold as a "location type" code. The inventory table must hold the pallet height measured at receiving, or at least the item-master standard height. And deep lanes must be modelled as lanes with a position count, not as single locations with a capacity of six.

• Height capture: a dimensioner or a height gauge on the receiving forklift mast gives a measured height; item-master heights drift as suppliers change pack patterns.
• Lane modelling: store lane ID, depth and level for each position so the query can tell an active lane from an empty one.
• Snapshot table: write location, SKU, lot, pallet count and height to a daily table so trends survive inventory moves.
• Zone tagging: tag every lane with its storage mode, because a drive-in zone and a selective zone need different targets.

Worked layout: one 4,800-position building priced in $ and ₹

Pricing honeycomb turns an engineering metric into a capital decision. The example below uses one layout, 3,600 selective positions and 1,200 drive-in positions, and prices it for a US distribution centre and a 3PL warehouse near Bhiwandi. All rents, costs and rates are illustrative, so replace them with your own lease and supplier quotes.

The selective area is 450 bays, four levels high, two pallets per level. The drive-in area is 200 lane-levels, six deep. Measured over four weeks, vertical honeycomb in the selective area is 25.6% of clear height, and 170 of the 200 drive-in lane-levels are active on an average day, holding 520 pallets in 1,020 positions.

The comparison shows why the same fix ranks differently by market. In the US example the beam change pays back in under eight months; in the Indian example cheaper space doubles the payback, so the rule change on deep lanes, which costs little in either market, should come first. Neither market has a statute that regulates honeycombing itself; the constraints are rack design, fire protection and the contract with the client.

Honeycomb cost equals lost pallet positions times the fully loaded cost per position. In the 4,800-position worked example, 900 vertical and 500 horizontal positions cost $13,300 or ₹3,64,000 a month, and re-beaming pays back in about 7.8 months in the US case and 15.7 months in India.

Incident trace: 40 pallets on the dock and 412 positions free

Honeycombing rarely announces itself as a space problem; it shows up as a receiving backlog. The trace below, with illustrative numbers, follows one inbound day at a site whose WMS reported spare capacity while the dock filled up.

1. Monday 05:40, inbound plan: The yard schedule shows four trailers, 104 pallets of a seasonal beverage SKU, 1,450 mm tall. The WMS capacity report shows 412 free positions, so the shift lead plans no overflow.
2. Monday 07:15, putaway failures begin: Putaway task PA-77314 returns "no location found". Of the 412 free positions, 236 are selective openings with 1,300 mm clear height, too low for the 1,450 mm load plus clearance.
3. Monday 08:30, deep lanes checked: Another 151 "free" positions sit behind single pallets in 31 drive-in lane-levels assigned to other SKUs. The putaway strategy forbids mixing SKUs in a lane, correctly, so they are unusable.
4. Monday 10:05, dock congestion: Only 25 truly usable positions remain. Forty pallets are staged on the floor in front of doors 6 to 9, blocking two doors for the afternoon outbound wave and creating an aisle obstruction that the safety lead orders cleared.
5. Monday 13:20, controlled response: The supervisor consolidates 18 single-pallet lanes of slow SKUs into selective openings, freeing 18 lanes, about 108 positions, for the beverage receipt. The engineer files change request RC-221 to re-slot by height and add a "tall load" location class.

The aisle obstruction in step four is not only a productivity issue. OSHA's general materials-handling rule, 29 CFR 1910.176(a), requires aisles and passageways to be kept clear where mechanical handling equipment is used, and 1910.176(b) requires stacked material to be stable and secure. Honeycomb that pushes pallets onto the floor becomes a safety finding in US buildings.

Where WMS rules decide and where software only recommends a move

A WMS can enforce fit and lane rules without judgement, but deciding whether to re-beam, re-lane or accept the loss needs an engineer and an operations manager. The table below names each honeycomb task and where it belongs.

When honeycombing is the cost of FIFO and lot control

Some honeycomb is a price worth paying. A food or pharmaceutical warehouse that must ship the earliest expiry first will keep lanes lot-pure and accept the horizontal loss, because a buried older lot costs more in write-offs than the space it wastes. The goal is to know the loss and decide it deliberately.

Honeycombing also matters in 3PL contracts. If storage is billed per pallet on hand, the 3PL absorbs the honeycomb; if it is billed per position reserved, the client pays for it. Our guide to pallet-day storage billing explains how to count chargeable pallets; this article measures the empty positions behind those pallets, which is the number a 3PL needs before agreeing a per-pallet rate for deep-lane storage.

Audit and redesign checklist for a honeycombed building

Run the work in the order data, rules, then steel. Most sites recover a large share of horizontal loss through putaway and consolidation rules within a month, and only then is a beam or lane-depth change worth its capital. The checklist below sets out each step with an owner and a finish line.

More warehouse design and storage guides are collected on the Logistics and Supply Chain hub. If you want the snapshot, height capture and putaway rules built into your WMS, or a layout review before you buy steel, send us a request for proposal with your location count, storage modes and current occupancy.