Key takeaways
- Two 2.0 mm CRCA rails over a 900 mm span carry a 1,000 kg rack at 153 MPa and 2.5 mm deflection, which is L/366 and inside an L/200 limit.
- The same load on 1.5 mm rails reaches 204 MPa, above the 165 MPa allowed at a 1.5 factor on 250 MPa yield, so gauge is a stress decision before it is a stiffness decision.
- Deflection rises with the fourth power of span: moving from a 900 mm to a 1,200 mm rail depth takes 2.5 mm to about 7.9 mm and fails L/200 on 2.0 mm sections.
- A single M6 screw tears out of 2.0 mm CRCA at about 4 kN and out of 1.5 mm at about 3 kN, so three screws per end beat a thicker rail when the joint is the weak point.
- Use a 12 mm minimum edge distance and a 6.5 mm hole for an M6 screw; a 9 mm hole adds 2.5 mm of slip per joint and lets a square frame rack out of square.
A 19-inch rack rail is a beam with a hole pattern, and it bends like one. Put 1,000 kg of equipment into a 42U frame on two 2.0 mm CRCA rails spanning 900 mm and each rail carries 4.9 kN, sees about 153 MPa at mid-span and deflects 2.5 mm. That is L/366, inside the L/200 limit most rack drawings quote.
Change the rail to 1.5 mm and the same load produces 204 MPa and 3.3 mm. The deflection still passes L/200, but the stress is past the 165 MPa a 250 MPa yield section is allowed at a factor of 1.5. That is the first thing rack sag teaches: the gauge is a stress choice, and the stiffness check can pass while the steel is already working too hard.
The rail is rarely the first member to fail. The screws tear the post, the open C-section twists, and the clearance in a 6.5 mm hole around a 6 mm screw lets the frame rack square under load. Rack sag is the visible symptom; the fixing pattern is usually the cause.
What causes rack sag?
Three mechanisms produce the same 5–10 mm droop, and they need different fixes.
The first is ordinary bending. A rail is a simply supported beam between the front and rear posts, with equipment load applied through cage nuts along its length. Treat the load as uniformly distributed and the mid-span deflection is 5wL⁴/384EI. Because the span term is to the fourth power, a 300 mm increase in rack depth matters more than doubling the load.
The second is joint slip. Rack rails are fixed to posts with two or three M6 screws through clearance holes. A 6.5 mm hole around a 6 mm screw gives 0.25 mm of movement per screw, and a 9 mm hole gives 1.5 mm. Two joints at each end can add 1–3 mm of drop that no beam calculation predicts, and the frame loses squareness at the same time.
The third is local crippling at the cage nut. The equipment ear is a small plate pulling on a small area of 1.5 mm or 2.0 mm sheet. Before the rail bends, the sheet around the nut deforms, the powder coat cracks, and the joint loosens. That is why a rack that passed a load test in the factory can sag after a year of service.
Rack rail gauge and load limits
The table uses a 50 x 30 mm lipped C-section in CRCA, two rails sharing a 1,000 kg load, and a 900 mm span. Stress is calculated at mid-span with a uniform load, and the allowable is 165 MPa, which is 250 MPa yield divided by 1.5.
| Parameter | Typical value | Ganesh limit or note |
|---|---|---|
| Rail section | 50 x 30 mm lipped C, 1.5–2.5 mm | laser cut and formed; 200 T brake to 10 mm |
| 1.5 mm rail, 900 mm span | 204 MPa, 3.3 mm (L/274) | above the 165 MPa allowable; light equipment only |
| 2.0 mm rail, 900 mm span | 153 MPa, 2.5 mm (L/366) | the standard choice for a 1,000 kg rack |
| 2.5 mm rail, 900 mm span | 122 MPa, 2.0 mm (L/450) | used where the rack is 1,200 mm deep |
| 2.0 mm rail, 1,200 mm span | 272 MPa, 7.9 mm (L/152) | fails both stress and deflection; step up or add a mid-support |
| Section modulus, 2.0 mm | about 3,600 mm³ | measured on the formed section, not the flat |
| Second moment, 2.0 mm | about 90,000 mm⁴ | bending about the strong axis only |
| Screw size and hole | M6 in 6.5 mm hole | 12 mm minimum edge distance |
| Screws per rail end | 2 minimum, 3 preferred | third screw cuts prying tension by half |
| Screw tear-out, 2.0 mm | about 4 kN per screw | 1.5 mm sheet gives about 3 kN |
| Post-to-post spacing | 482.6 mm nominal | hold to plus or minus 0.5 mm |
| Frame diagonal | 2 mm on a 42U frame | check before coating, on a fixture |
| Rail hole pitch | 44.45 mm | plus or minus 0.3 mm; cumulative error breaks panel fit |
| Rail flatness | 1 mm/m | a twisted rail loads only two of its four screws |
| Finish | 60–80 µm polyester powder | seven-tank pretreatment, 18 ft x 14 ft booth |
The row that surprises buyers is the 1,200 mm span. At 1,000 kg, a 2.0 mm rail is at 272 MPa, which is past yield before any safety factor, and it deflects almost 8 mm. A deeper rack needs a 2.5 mm rail, a deeper section, or a mid-span support tied to the side panel. Adding screws does not fix a bending failure.
Why the fixing pattern fails before the rail does
Take the 4.9 kN rail load and split it between two ends: 2.45 kN at each post connection. With two M6 screws, each takes 1.23 kN in shear. The equipment sits about 25 mm forward of the rail web, so the load also creates a moment of 61 kN·mm at the joint. A 30 mm vertical screw spacing turns that into 2.0 kN of prying tension per screw.
A single M6 in 2.0 mm CRCA tears out at about 4 kN with a 12 mm edge. The joint is therefore at roughly half its capacity in tension and a third in shear, which sounds acceptable until vibration and repeated installation cycles loosen it. At 1.5 mm the tear-out falls to about 3 kN, and the same joint is at two-thirds of capacity with no margin for a reamed hole.
Three screws over 60 mm change the geometry. The prying couple is shared over a longer lever, so tension per screw falls to about 1.0 kN, and the joint runs at a quarter of its tear-out capacity. That is why a third screw is worth more than a 0.5 mm gauge increase on a rail that is already inside its stress limit.
Two more details decide whether the pattern holds:
- Edge distance. Keep the hole centre at least 12 mm from the post edge, or 2d. A hole 8 mm from the edge tears at about 2.5 kN in 2.0 mm sheet, whatever the screw grade.
- Hole clearance. Use 6.5 mm holes for M6, not 8 mm slots. If site adjustment is needed, slot one end only and use a hardened washer, so the other end stays a datum.
Cage nuts are not a structural fastener. They are a service fastener. Where the rail is expected to carry more than about 300 kg per fixing point, use a tapped insert or a through-bolt with a load-spreading washer, and say so on the drawing.
A point load doubles the bending moment
The uniform-load assumption flatters a rack. A 1U server is a point load, not a distributed one, and the difference is a factor of two in stress.
Take the same 4.9 kN on a 900 mm rail. Spread evenly, the moment is 551 kN·mm and the 2.0 mm rail sees 153 MPa. Put all of it at mid-span and the moment becomes PL/4, or 1,102 kN·mm, which is 306 MPa — past yield, with a deflection of 3.9 mm. The rail has not become weaker; the load has moved to the worst place.
Real installations sit between the two. Equipment ears transfer load at the front and rear posts, and a 4-post mounting shares the mass across four points rather than two. That is why a 1,000 kg rack with equipment spread over 20U behaves better than the same mass in one 4U chassis at the middle of the rail.
The drawing rule is simple: state the heaviest single item, its mounting points and its position along the rail. A rail rated for 500 kg uniformly is rated for about 250 kg as a single mid-span point.
Where the sag actually shows
Sag is not evenly distributed, and the pattern tells you which mechanism is at work.
A rail that droops at mid-span with the ends still tight is a bending problem. Check the span and the section before touching the screws. A rail whose front end has dropped while the rear stays level is a joint problem: the front screws have torn or the cage nuts have pulled. A rail that has twisted so the mounting face is no longer vertical is a forming problem: the C-section was formed from a blank with an uneven bend line, or the rail was straightened after coating.
A frame that has become a parallelogram is neither. It is a bracing problem. A 42U frame with moment connections in 2.0 mm steel has very little lateral stiffness, and a 1 kN push at the top moves it several millimetres. Fit a diagonal, a rear panel or a welded gusset, and the frame starts to act as a shear panel.
Check the rack on a flat floor with the levelling feet loaded, not on a pallet. A rack standing on four feet with one foot 2 mm short will twist, and the twist will be blamed on the rails.
How to check a rack that has already sagged
Measure before you change anything, because the fix depends on which of the three mechanisms is running.
- Rail straightness. Lay a 1,000 mm straightedge along the mounting face and measure the gap with a feeler gauge. Record it at mid-span and at each end. A curve that peaks at mid-span is bending; a straight rail that sits low at one end is a joint.
- Post spacing. Measure between post faces at the top, middle and bottom of the frame. A difference of more than 1 mm over the height means the frame has racked, and the rail is following it.
- Diagonal. Measure corner to corner on the front face. A 2 mm difference on a 42U frame is acceptable; 5 mm is a frame that needs bracing, not new rails.
- Screw witness marks. Look for cracked powder around the screw heads and elongated holes. Both mean the joint has moved, and the rail may be undamaged.
- Loaded deflection. Measure the rail with the rack empty, load it to the rated mass, measure again, then remove the load and measure a third time. The difference between the first and third readings is permanent set. Anything above 0.5 mm means the steel has yielded and the rail should be replaced.
If the rail has yielded, a gauge increase is the answer. If it springs back, the rail is elastic and the problem is the span, the load position or the joint. Replacing an elastic rail with a thicker one hides a joint problem for a year and then returns it.
What Ganesh runs for rack frames and rails
Rails and posts are cut on the 12 kW fibre laser with the 40 ft x 16 ft bed, which keeps a 2,400 mm rail and its hole pattern in one setup. Forming runs on the 200 T, 5-axis brake with the 4,450 mm bed, so a full-length rail is bent in one pass rather than in two. Frames are welded on three CLOOS robotic MIG cells or at the 30+ manual stations, and the checking fixture is built in the tool room, so post spacing and diagonal are measured before the frame reaches the paint line.
Coating is seven-tank pretreatment into the 18 ft x 14 ft powder booth, with DFT and cross-hatch adhesion recorded per batch. The same route produces 19-inch frames, mounting rails, PDU brackets and blanking panels, which is why a rack and its internals can arrive from one supplier. The rack frame design guide covers the hole pattern, the bending guide for rack frames covers the forming limits, and the data centre rack range lists the frame sizes.
What to put on the drawing
- Loaded mass, the number of load points and the mass per point, not just a total.
- Rack depth and the rail span between post faces.
- Rail section and gauge, with the section modulus if the rail is a special form.
- Screw size, grade, hole diameter, edge distance and the number of screws per end.
- Post spacing, diagonal and rail hole pitch, with the inspection fixture called out.
- Deflection limit under the stated load, expressed as L/200 or a millimetre value.
- Finish system and the coating thickness range, plus any masked earthing points.
- Whether the rack is free-standing or wall-tied, because that changes the bracing.
When a heavier rail is the wrong answer
A 2.5 mm rail is not a universal upgrade. On a 600 mm deep rack it adds mass and cost for a deflection that was already inside L/500, and it makes the cage nut seating worse because the sheet is stiffer under a standard nut. If the load is concentrated at one point rather than distributed, a local doubler is cheaper than a gauge increase.
Ganesh does not certify seismic performance, run IEC 61587-1 dynamic tests in-house or design the room cooling. Where a project needs a tested seismic rack, the frame has to go to a lab with the anchorage design, and we build to the drawing that comes back. Send the load schedule and the rack depth with the enquiry, and the quotation will state the rail section, screw pattern and deflection it assumes.
Rack sag is fixed at the drawing stage: one span, one gauge, one screw pattern, written down. If the rail schedule says "2 mm rail, 2 screws", the drawing has not decided anything yet.
Frequently asked questions
How much weight can a 19-inch rack rail hold?
A 2.0 mm CRCA rail formed as a 50 x 30 mm lipped C-section holds about 500 kg per rail over a 900 mm span at a 1.5 factor on yield, which is 1,000 kg for a two-rail rack. A 1.5 mm rail holds about 400 kg per rail and a 2.5 mm rail about 600 kg on the same span.
Why does my rack sag even though the rails are rated?
The rating is usually for a pair of rails at a stated span and load pattern. Sag appears when the span is longer than the rating assumed, when the load sits between fixing points rather than over them, or when the screws have torn or slipped. Check the span and the screw group before changing the rail.
Should rack rails be 1.5 mm or 2.0 mm?
Use 2.0 mm for a 900 mm deep rack carrying more than about 700 kg, and 1.5 mm only for light equipment below about 400 kg per rail. On a 1,200 mm deep rack, 2.0 mm is marginal even at 1,000 kg because deflection scales with span to the fourth power.
Do blanking panels add stiffness to a rack?
Very little. A 1.2 mm blanking panel fixed with two screws at each end acts as a diaphragm only if it is positively connected to both posts and the rack is square. Treat it as an air seal, not as a structural member, and size the rails for the load without it.
What tolerance should a rack frame be built to?
Hold the post-to-post spacing at plus or minus 0.5 mm, the diagonal at 2 mm on a 42U frame and the rail hole pitch at plus or minus 0.3 mm. A frame within those limits accepts a standard 482.6 mm panel without reaming holes on site.
Sources and further reading
- IEC 61587-1 — Mechanical structures for electronic equipment: environmental requirements and test set-up for cabinets, racks and subracks · reviewed 3 Nov 2026
- ISO 2768-2 — General tolerances for geometrical tolerances on features without individual tolerance indications · reviewed 3 Nov 2026
- ISO 12944-2 — Corrosion protection of steel structures: classification of environments · reviewed 3 Nov 2026
Have a rack drawing with a rail schedule?
Send the loaded mass, rack depth, rail section and fixing pattern. We will check the rail stress and deflection, then quote the frame, rails and coating as one build.
Request a quote