Key takeaways
- An open 1U space is a 0.020 m² orifice; at 10 Pa it passes about 176 m³/h (104 CFM), which returns roughly 0.9 kW of heat to the cold aisle at a 15 °C rise.
- 1.2 mm CRCA is the right gauge for a solid 1U panel: 1.0 mm drums under fan airflow and 1.5 mm makes the 44.45 mm fold harder without adding stiffness where it is needed.
- Steel beats aluminium on a blanking panel because mass is not the point; a 1.5 mm aluminium panel is 0.09 kg against 0.25 kg for 1.2 mm steel, and the lighter panel rattles against the rail instead of seating on it.
- Keep the coated width inside 482.6 mm +0/−0.4 mm and flatness within 0.8 mm; powder at 70 µm per face adds 0.14 mm, which is most of the fit budget in a stacked rack.
- Perforated panels only help where bypass is wanted; a 55% open pattern on 1.2 mm sheet passes about half the air of an open unit, so specify open area as a number rather than as "vented".
An empty rack unit is a hole with a price on it. At a 10 Pa pressure difference across the rack, a 1U opening of 0.020 m² passes about 176 m³/h — roughly 104 CFM — of air that has already cooled the servers below it. Multiply by the twelve or twenty open units in a partly filled rack and the bypass flow is a serious fraction of the rack's total airflow, all of it returning to the cold aisle at the temperature of the hot aisle.
A data centre blanking panel is the cheapest control on that path. It is also a part with almost no room for error: 482.6 mm wide, 44.45 mm high, two folds, a latch and a powder coat, stacked against twenty identical panels in a live rack. The material and finish decisions are what decide whether it seals, rattles or jams.
This page covers the airflow arithmetic, the steel-versus-aluminium-versus-plastic choice, the coating and masking that keep the fit, and the laser and press routes that produce the panel at different volumes. The feature-level DFM rules sit in the blanking panel design guide, and the step-by-step shop route in how a data centre blanking panel is manufactured.
How much air does an open 1U space actually leak?
About 104 CFM at 10 Pa, which is enough to matter in a rack drawing 2,000–3,000 CFM. The calculation is an orifice, not a guess.
Take the opening as 450 mm wide by 44.45 mm high, which is the clear space between the mounting flanges: A = 0.450 × 0.04445 = 0.020 m². For a sharp-edged opening the discharge coefficient is about 0.6. At a pressure difference of 10 Pa and air density of 1.2 kg/m³, velocity is √(2 × 10 ÷ 1.2) = 4.08 m/s, so volume flow is 0.6 × 0.020 × 4.08 = 0.049 m³/s, or 176 m³/h, or 104 CFM.
The heat that flow carries is the number that gets attention. Mass flow is 1.2 × 0.049 = 0.059 kg/s, and at a 15 °C difference between hot and cold aisle the recirculated heat is 0.059 × 1,005 × 15 = 886 W. Call it 0.9 kW per open unit at that rise. Twenty open units in a row return heat equivalent to a small in-row cooler running backwards.
The pressure difference is the assumption that matters, and it varies with the rack, the perforated tile and the fan speed. At 5 Pa the same opening passes 74 CFM and at 20 Pa it passes 147 CFM, because flow goes with the square root of pressure. The conclusion does not change: seal the units you are not using, and seal them with a panel that actually contacts the rail.
Steel, aluminium or moulded plastic?
Steel for most racks, aluminium where weight or corrosion matters, plastic where the panel must be non-conductive. The table sets out what each choice changes.
| Parameter | 1.2 mm CRCA (standard) | 1.5 mm aluminium | Moulded PC/ABS | Ganesh limit or note |
|---|---|---|---|---|
| Panel mass | 0.25 kg | 0.09 kg | 0.12–0.18 kg | Check the flat pattern on the weight calculator |
| Stiffness in a 1U face | Good with 12 mm returns | Adequate, needs deeper returns | Good, ribbed by the tool | 1.0–1.5 mm steel; 1.0 mm drums, 1.5 mm folds poorly |
| Forming | Press brake or progressive tool | Press brake, softer springback | Injection moulding | We form steel and aluminium; we do not mould |
| Finish | Powder, 60–80 µm | Powder or anodising | Moulded-in colour | Seven-tank pretreatment; anodising is bought in |
| EMC contact | Conductive at masked ears | Conductive at masked ears | Needs a conductive insert | We mask; gasket bonding is outside our process |
| Corrosion | Coated steel, C3 interior | Good without coating | Good | Coastal halls favour aluminium or a C4 coating system |
| Tooling | Progressive tool once volume justifies it | Laser and brake | Mould, high upfront cost | Tooling quoted separately from piece price |
| Recyclability | High | High | Low | Relevant to tender scoring, not to the drawing |
The steel panel wins on stiffness per rupee and on coating match to the rack, which matters because a blanking panel is a visible part and a shade difference between January and July stock is obvious in a white hall. Aluminium wins where the rack is specified light or the hall is coastal, but the gauge has to go up to keep the same stiffness: 1.5 mm aluminium has a bending stiffness of about 30% of 1.2 mm steel, so the returns need to be deeper or the panel needs a swage.
Plastic is a different supply chain. It is the right answer for a rack that must be fully insulated, and it is the wrong answer for a tender that specifies a metal panel for fire performance or for earthing. We do not mould, so a plastic panel is a bought-in item with the metal latch and the metal ear insert coming from a different supplier — which is a supplier-interface problem before it is a technical one.
What does a data centre blanking panel finish have to survive?
A powder coat on a blanking panel does three jobs: it matches the rack, it protects a 1.2 mm edge from rust in a dusty hall, and it stays put when the panel is pulled out by a technician wearing a tool belt.
60–80 µm of polyester over the seven-tank pretreatment is the standard. Below 50 µm the film thins over the laser-cut edge and the mounting ear, which is where handling damage starts; above 100 µm the film chips at the fold and the chips sit in the rack as powder flakes.
Masking is the part of the finish specification that affects function:
- Mounting ear faces where they contact the rail, so the panel seats metal to metal and the ear does not creep under load.
- The latch contact zone, because a clip gripping a coated edge loses friction and the panel walks out in transit.
- Any EMC contact strip, because powder is an insulator and a bonded gasket needs bare, correctly prepared metal underneath.
Everything else takes the full coat, including the returns and the inside face. Those are the surfaces a hand touches and the surfaces that rust first.
Adhesion is checked by cross-cut to ISO 2409, class 0–1, on the first part of each shift. The place to cut is the outside of a return, where the film is thinnest and where a chip will be visible.
Forming a 1U panel without losing the fit
The panel begins as a flat blank about 40 mm high before folding, or as a developed blank if the ears are part of the same strip. Two folds create the returns that stiffen the face; the ears carry the mounting holes.
Four forming rules protect the fit:
- Inside radius at least one material thickness. On 1.2 mm CRCA that is 1.2 mm; 1.5 mm avoids a sharp outer corner that chips.
- Holes at least 4.2 mm from the bend line on 1.2 mm sheet with a 1.2 mm radius. Closer, and the mounting hole pulls oval as the material draws.
- Fold before piercing the latch slot, or pierce in the tool if the sequence allows, so the slot stays parallel to the face rather than rotating with the bend.
- Deburr the finger slot. A laser-cut edge at 1.2 mm with a burr underneath is the defect that gets the panel rejected by the person installing it, not by goods-in.
Flatness is the fourth dimension of the fit. A panel that is 1 mm out of flat across 482.6 mm shows a light gap when stacked, and a stack of twenty shows twenty gaps. The cause is usually the brake: a ram that deflects under load, or a blank that was not flat to begin with. Bending near the centre of the bed with a stored programme per part number keeps the angle consistent, which is the same discipline used on rack frames and cabinets.
Laser route or tooled route?
Laser and brake below about 5,000 panels a year, a progressive tool above it. The punch load is not the constraint; tool life and hole quality are.
A 1U panel’s blank perimeter is roughly 1,100 mm. In 1.2 mm CRCA at 300 MPa shear strength, the punching force is 1,100 × 1.2 × 300 = 396 kN, or about 40 tf — comfortably inside the 250 T punching capacity and well inside the 300 T press. So the press is never the reason to hesitate.
What changes with volume is the economics of the tool and the consistency of the hole pattern. A progressive tool pierces the mounting holes, the latch slots and any perforation in one pass, holds the hole centres to ±0.1 mm, and produces the same panel a million times. A laser cuts a nest of panels from one sheet with no tool, holds ±0.2 mm on hole position, and suits a colour or pattern change without a tooling charge.
The practical split:
- Prototype and low volume — laser cut, press braked, coated. No tooling, and a design change costs a revised DXF rather than a new die.
- Repeat volume with a fixed design — progressive tool on the press line, with the laser as the fallback for spares and small top-ups.
- Mixed colours in one order — laser, because a tool is per part number, not per colour, and the powder batch is the changeover.
We quote both routes separately so the crossover is a decision the buyer can see rather than an assumption buried in a piece price.
The panel as part of the rack system
A blanking panel is bought with the rack, installed with the rack and judged with the rack, and that is why the surrounding parts matter. The 19-inch rack frame design guide covers the rails the panel mounts to and the hole pattern it must match. The mounting hole centres of 465.1 mm are not a suggestion: they are the EIA-310 pattern, and a panel drilled to a nominal 465 mm will fight every rail on site.
Panels also interact with containment. In a cold-aisle containment row, a missing blanking panel is a hole in the pressure boundary, and the containment doors will not hold differential pressure. In a non-contained row the panel still matters, because recirculation happens locally at each rack. The data centre enclosures hub collects the guides on frames, rails and panels, and the rack load calculator is the right tool for the loaded rack rather than for the 0.25 kg panel that fills its empty units.
Ganesh cuts and forms panels on the 12 kW fibre laser with the 40 ft × 16 ft bed, forms on the 200 T, 5-axis brake with the 4,450 mm bed, and coats on the seven-tank pretreatment line and in the 18 ft × 14 ft powder booth, with DFT and cross-cut checks per batch. At volume, panels run in a progressive tool built in the in-house tool room on presses from 20 T to 300 T. The plant holds ISO 9001:2015 and ISO 14001:2015, and panels sit within the data centre rack range alongside frames, rails and cable management.
What to put on the drawing
- Width over the ears at 482.6 mm +0/−0.4 mm, with a note that it applies after coating.
- Height at 44.45 mm −0.2 mm, and the return depth range that fits the rail.
- Mounting hole pattern — 465.1 mm centres, hole size, ear geometry.
- Material and gauge, with the standard if the customer specifies one: IS 513 grade D for CRCA.
- Perforation, if any: hole size, pitch, open area and the unperforated border at the fold.
- Latch type, engagement depth and retention force, typically 30–50 N.
- Coating — colour, gloss, DFT, and the masked zones listed by name.
- Tolerances written per feature, because ISO 2768-1 class m allows ±0.8 mm in the 120–400 mm band and the width needs a one-sided 0.4 mm callout instead.
- Marking and packing — part number, batch and interleaving, because panels packed face to face arrive with rub marks.
When a blanking panel is not the right answer
Four cases. A rack with a fully populated front and no empty units does not need panels, and buying them anyway is a cost with no airflow benefit. A rack that is open at the rear as well as the front needs two panels per unit, or a rear door, and one panel closes half the path.
A rack specified for EMC needs a gasket and a bonding kit, not just a metal panel; we mask and coat but do not bond conductive gaskets or test shielding effectiveness. And a rack with a non-standard width — 23-inch or a metric frame — needs a panel designed for that rail, because a 19-inch panel with an adapter bracket adds a joint that leaks and a tolerance stack that rattles.
Send the rack elevation, the panel drawing, the colour and the annual quantity, and the quotation will separate the laser route from the tooled route so the crossover is visible before the order.
Frequently asked questions
What material is best for a data centre blanking panel?
1.2 mm CRCA with a powder coat for most installations, because it is stiff enough not to drum, cheap to form and takes a coating that matches the rack. Aluminium is used where weight or corrosion resistance matters, and moulded plastic where the panel must be non-conductive and tooling is justified.
Do blanking panels need to be perforated?
Only where controlled bypass is wanted, usually at the top or bottom of a rack. A perforated panel at 55% open area passes about half the air of an empty unit, so it reduces bypass rather than eliminating it. A solid panel is the correct choice for the middle of a populated rack.
How many blanking panels does a 42U rack need?
One for every unoccupied rack unit in the front and rear mounting planes, which on a half-populated 42U rack is typically 20–24 panels. Count the open units after the equipment layout is fixed, not before, because a panel ordered for a unit that later takes a switch is scrap.
Does the powder coat stop the panel fitting?
It can. Two faces at 70 µm add about 0.14 mm to the panel thickness, and the coated width must stay inside 482.6 mm +0/−0.4 mm. Mask the mounting ear faces and any EMC contact strip, coat the returns, and measure the width after coating rather than on the bare blank.
What tolerance does a blanking panel need?
Width 482.6 mm +0/−0.4 mm, height 44.45 mm −0.2 mm, hole centres 465.1 mm ±0.3 mm and face flatness 0.8 mm. Those are tighter than ISO 2768-1 class m for a 500 mm part, so they must be written on the drawing as specific callouts.
Sources and further reading
- IEC 60297-3-100 — Mechanical structures for electronic equipment: 482.6 mm (19 in) series, basic dimensions · reviewed 3 Nov 2026
- ISO 2768-1 — General tolerances for linear and angular dimensions · reviewed 3 Nov 2026
- ISO 2409:2020 — Paints and varnishes: cross-cut test · reviewed 3 Nov 2026
Ordering blanking panels for a rack build?
Send the panel drawing or the rack elevation, the colour, the latch requirement and the annual quantity. We quote the laser route and the tooled route separately so the crossover is visible.
Send a drawing