Specifying a pallet rack system is a structural problem first and a purchasing decision second: the 4–6 governing numbers are unit load (kg), pallet footprint (mm), fork-truck mast tilt and collapsed height (mm), clear-aisle width for the truck (mm), bay count per fire-sprinkler node, and the seismic/wind zone the building sits in [S1]. Get any of those wrong and you either under-size the upright frames (frame collapse risk) or over-size the project (10–25% wasted steel and floor area).
The end use drives almost every other choice. Supermarket back-of-house shelving, industrial bulk storage rack in a 12 m clear warehouse, mobile drying racks for a screen-print shop, and Red Hat Ceph-optimized server racks each follow different sizing paths [S1][S2][S4]. Picking the rack family before locking those six numbers is the most common spec error in the field.
Governing Load and Geometry Inputs
Unit load and pallet footprint set beam length and beam capacity. A standard EUR pallet (1200 × 800 mm) needs a beam pair of at least 2700 mm clear for two pallets back-to-back, while a US GMA pallet (1219 × 1016 mm) needs 2743 mm; 3600 mm and 3658 mm beams are the common two-pallet SKU [S1]. Beam capacity is normally given per pair — typical light-duty beams run 1000–2500 kg/pair, medium 2500–4500 kg/pair, and heavy structural beams exceed 4500 kg/pair, with a published safety factor of 1.67 (RMI/ANSI MH16.1 family of tests).
Bay depth is set by the pallet count along the aisle-perpendicular axis — 2-pallet deep (2400 mm), 3-pallet deep (3600 mm), or 4-pallet deep (4800 mm) are the most common cells. Double-deep and push-back layouts use the same beam but add a second row behind; drive-in/drive-through racks trade selectivity for cube density and need a different upright gauge.
Clear-aisle width is dictated by the lift truck: counter-balanced fork trucks need ~3600–4000 mm, reach trucks ~2700–3000 mm, turret trucks ~2400–2700 mm, and order pickers under 2400 mm.
Upright Frame, Beam, and Deck Selection
Frames are the columns; beams are the horizontal load arms. Three structural decisions flow from the load inputs: frame section (roll-formed vs structural), beam profile (step beam, box beam, teardrop), and decking (wire mesh, wood, steel pan). [S1]
Roll-formed teardout frames dominate the selective-rack market because the teardrop connector accepts a wide range of beam heights from different makers, but structural C-channel frames with bolted beam connectors are mandatory for fork-truck impact columns, high-rise installations above ~9 m, and most seismic zone D+ builds [S1]. A typical roll-formed frame section runs 75–100 mm wide with 2.0–3.0 mm steel; structural frames start at 76 mm × 76 mm × 6.4 mm C-section and go up to 127 mm × 127 mm × 9.5 mm for the heaviest configurations.
For lighter, non-pallet use — supermarket stockrooms, mobile drying carts, archive shelving — the rack type shifts to a storage cage, boltless shelving, or cantilever-style rack. AWT/Saturn mobile drying racks, for example, use 25–50 mm square-tube frames with casters rated 150–300 kg per shelf, sized for screen frames rather than forklift pallets [S2]. That distinction — pallet-rated structural rack vs. hand-loaded utility rack — is the line that most spec errors cross.
Comparison of Main Rack Families
Four rack families cover roughly 90% of warehouse applications; the right pick is a function of SKU count, selectivity, and cube density needed. [S2]
Selective pallet rack is the baseline: every pallet is accessible from the aisle, selectivity is 100%, cube density is the lowest. Drive-in/drive-through rack uses rails the truck drives into, selectivity drops to ~33–50%, and cube density climbs 60–75% over selective — best for low-SKU, high-quantity loads like raw material and finished goods at the end of their life. Push-back rack uses nested carts on inclined rails, gives 2–6 pallets deep with LIFO discipline, and sits between selective and drive-in on both axes. Pallet flow rack uses roller or wheeled tracks with brake wheels, gives FIFO at 2–10 pallets deep, and is the standard pick for date-sensitive goods.
Seismic, Fire, and Code Constraints
Seismic design is the single largest hidden cost in rack specification. US practice follows the RMI/ANSI MH16.1 specification with seismic design per ASCE 7, which most commonly uses SDS (design spectral response acceleration at short periods) to set the anchorage force and frame bracing pattern. Zone A/B frames can ship with light bracing; zone D+ projects require heavier base plates, more anchors per foot, and often x-bracing on both faces [S1].
Fire code is the second sleeper. NFPA 13 (sprinkler) and FM Global Data Sheet 8-9 typically require in-rack sprinklers when the rack exceeds 12 m, when commodity class is Group A plastics, or when flue spaces are not maintained at 150 mm transverse and 75–100 mm longitudinal. Storage height of 7.6 m is a common breakpoint where ceiling-only sprinkler design is no longer accepted for Class I–IV commodities, pushing the spec to in-rack.
Anchorage is the third sleeper: a single 13 mm wedge anchor in 2000 psi concrete carries roughly 35–50 kN tension in static load, but seismic and impact loads can multiply the design tension by 1.5–2.0 — the project engineer of record almost always re-specs the anchor count and embedment, and the rack supplier's "standard" anchor pattern is rarely the final one.
Who a Pallet Rack Is For — and Who It Is Not
Pallet rack is the right answer for warehouses handling 500+ pallet loads in a multi-SKU environment with standard GMA or EUR pallets, where selectivity matters and forklift reach is already on the floor. A retailer running 200 SKUs at 30 pallets each is a textbook case, and the system will pay back in 18–36 months on labour and cube utilization vs. block stacking. [S3]
Pallet rack is the wrong answer for: operations with fewer than 200 pallet positions (cantilever or shelving is cheaper), mixed-shape loads (a linear guide or conveyor-integrated workstation is more flexible), date-critical FIFO where pallet flow is overkill, and any project where the floor slab is not rated for the loaded rack weight. A loaded 4-high selective rack cell of 1000 kg pallets in a 4 × 3 bay puts roughly 12 t on a 12 m² floor footprint — that's 1 t/m², and a 125 mm slab on poor subgrade will crack without a slab-by-engineer review.
Inspection, Re-Spec, and Lifecycle Levers
Annual inspection is the cheapest insurance on a rack. OSHA 1910.176 and the RMI/ANSI MH16.3 guideline define three damage levels: minor (paint off, no measurable deformation — record and keep in service), moderate (bracing bent 5–25° or beam end connector lip bent — unload and repair within 30 days), and severe (upright twist, cross-aisle brace torn, anchor pulled — unload immediately and replace). A typical pallet rack has a 15–20 year design life with periodic re-spec; impact from lift trucks is the dominant degradation mode, not fatigue. [S4]
2.0 mm → 2.5 mm) at the same depth — that adds roughly 8–12% to the frame cost and extends the frame's capacity by 30–50%. The second lever is the beam connector: locking tabs and safety pins cost a few dollars per beam and prevent the most common beam-displacement accident during load shifts.
Sourcing reality: lead time for a roll-formed selective rack in 2026 is 4–6 weeks for standard beams and 8–10 weeks for frames with custom colour or galvanized finish. Structural rack and seismic-zone frames run 10–14 weeks. Projects that try to compress that to 3 weeks pay a 15–30% expedite premium or accept partial shipments. The 4-week engineering review (seismic calc, anchor pattern, slab check) is the schedule bottleneck and should start before PO release.
When project timing, budget, and warehouse layout align, the typical next step is a 3–5 day site survey that confirms pallet mix, lift-truck fleet, clear heights, and slab condition — those five data points convert a catalogue into a working bill of materials. If your project also touches an adjacent workflow like overhead crane selection (shared building clear height) or dock leveler sizing (shared apron geometry), run those surveys together so the structural and bay-width numbers stay consistent.