Sealed enclosure temperature rise: when you need a fan

A closed, painted steel enclosure indoors sheds about 5.5 W per square metre of effective surface per kelvin by the Rittal and Schneider method, so a 1200 x 800 x 300 mm box against a wall gets rid of about 128 W at 10 K and 256 W at 20 K. nVent's curve for the same box gives about 85 W and 185 W, read by eye, so plan on the lower figure unless you can test. For critical applications nVent adds a 25% margin to the rise from its curve, which for a 10 K limit means about 70 W on this box. If your losses are higher, first move heat out of the box (Schneider's example is drive braking resistors at 500 W to 3.5 kW) or use a bigger box, then consider outside air, a heat exchanger or a cooling unit.¹ A filter fan needs about V = 3.1 x P / ΔT in m³/h at sea level, with P in watts and ΔT in kelvin, and about 13% more at 1,000 m.

These are estimates from makers' constants, and they are not temperature-rise verification under BS EN IEC 61439.

How much surface the box really has

The back of a box against a wall, or the side of a box in a suite, does much less work than an open face, so the makers weight each face.

The first row is Rittal's worked example.² The second is inferred: Schneider prints no factors, but these reproduce its 3.55 m² and 5.07 m² for two back-to-wall enclosures exactly.¹ The third is AutomationDirect's,³ and the last combines those on our own reading, because AutomationDirect's "ground and wall" row gives no reduction for the wall.

The most a sealed box can shed

Compare your loss budget with this table. The weighted columns use k = 5.5. The nVent columns are its painted curve, read by eye, on exposed faces (a wall box loses the back and a floor box the bottom).⁴

A single 5.5 kW drive at 174 W in ABB's table already exceeds what the 800 x 600 x 300 box can shed at 20 K by either method.⁵ Rise = P / (k x A). Rittal calls this the maximum rise and nVent calls its curve an average.⁶⁴ Treat it as an average, because Schneider's illustration shows the top about 10 K above a mixed box.¹

Where the makers' coefficients part company

All figures are in W/m²K. Keep a coefficient with the area basis it was published for.

Stainless runs from 1.94 to 4.71 depending on finish and document. Bare aluminium is about 12 in two AutomationDirect tables, 1.94 in a third and about 2.2 to 2.6 on nVent's curve, and nVent says unfinished aluminium runs hotter than painted steel. The white paper's 12 sits in a heater-sizing table, where a high figure is the safe side, on our inference, so do not use it to size cooling unless that enclosure's maker gives it for that product. Plastic is about two thirds of steel in Schneider and two AutomationDirect tables, and equal to painted steel in nVent and the quick method. A two-film estimate of our own, with assumed convection and emissivities, supports the low figures: 4.88 for a painted box, 2.62 for bare stainless, 2.10 for bare aluminium.

Set the limit from the weakest part in the box

Wiring and breakers count as much as electronics. Schneider's table, which is supplier guidance, gives recommended and maximum air temperatures: drives 35 and 50 °C, PLCs 35 and 40 to 45 °C, power supplies 35 and 40 °C, contactors and breakers 45 and 50 °C.¹ It also recommends a mean of 35 °C inside. Against that, 50 °C inside is a maximum with a risk of malfunction, not a design target. Breaker and cable ratings also fall as the air warms, so take the derating from each maker and size conductors for the air they actually sit in. BEAMA notes that temperature-rise verification assumes conductors at no more than 70 °C.⁸

Which assumption moves the answer most

Base case: 2000 x 800 x 600, free-standing, painted steel, 500 W, room at 30 °C, with a 40 °C limit (Schneider's power supply maximum). The table compares that limit with the mean air. A power supply in the top of the box sees hotter air than this, so keep it out of the top or stir the air, and check the air at its inlet.

For the rise alone, finish moves it most, then material, then mounting, and ambient not at all. For the decision you actually face, ambient wins: 5 K more in the room halves the heat you can shed and 10 K removes all of it, which no mounting, material or finish change does. So spend your effort on the hottest realistic temperature where the panel will stand. Indoors, paint colour matters much less than having paint at all; nVent treats colour as a solar question, and warns that the sun can take an outdoor enclosure well above these estimates.⁴

Fan airflow, with the altitude correction

A fan replaces hot air with cooler air, so V = f x P / ΔT, with f = 3600 / (ρ x cp) in m³·K/(W·h) and ΔT measured, as Rittal defines it, from supply air to exhaust air.⁶

Rittal's factors track air at about 35 °C on the standard atmosphere, on our reading. The US constants are the same physics with cooler air: AutomationDirect's 3.17 CFM·°F/W converts to 2.99,³ and Pfannenberg's 1.82 CFM·K/W to 3.09.⁹ A sealed box hardly changes with altitude: our film estimate gives about -2% at 1,000 m.

Compare the result with the fan's flow through its filters and grille, not the free-air figure.⁷⁹ Rittal's condition for a filter fan at all is that the ambient air is clean and below the required internal temperature.⁶

A filter fan ends the sealed enclosure. On our reading the assembly can be no better than the fan and filter's IP rating (Schneider's fans are IP54 and Rittal's SK 3245 diagrams show IP51),¹⁶ and whoever declares the panel must verify its rating with the openings cut. Put the fan in the lower third and the exhaust near the top, as Pfannenberg advises.⁹ Flow falls as filters load: if it fell 30%, as an illustration, a 10 K rise would become 14.3 K, so size on a part-loaded filter and schedule changes, which Schneider suggests every four weeks for critical workshops.¹ Fit an over-temperature alarm, because a stopped fan leaves a box that cannot shed its load.

A worked illustration

The losses and the efficiency here are invented round numbers. Two drives at 200 W each, a 180 W power supply at an invented 90% (20 W lost), a PLC and I/O at 30 W, and switchgear and wiring at 50 W make 500 W in a 1200 x 800 x 300 box against a wall, in a 30 °C room, with 40 °C as the limit.

Sealed, the rise would be 39.1 K. The surface sheds 128 W at the allowed 10 K, and staying sealed would need 9.1 m² against the 2.33 m² available. A heat exchanger would need at least 37.2 W/K. A fan at 0 m needs 156 m³/h ignoring the surface, or about 136 m³/h net of it, because the surface sees only about a 5 K mean rise; at 1,000 m those become 176 and 154. Size on the first figure. In a 25 °C room it falls to 104 m³/h, in a 35 °C room it rises to 313 m³/h, and at 40 °C no fan can hold the limit.

Replace the round numbers with real ones. ABB's ACS580 catalogue has a heat dissipation column by type code, for example 174 W for the ACS580-01-12A6-4 at 5.5 kW, with 50 m³/h as the drive's own airflow.⁵ For a power supply, take the efficiency from its data sheet: loss = output x (1 / efficiency - 1). Schneider says the enclosure fan's flow should be at least the drive's flow, to avoid air locks.¹

Why the average is not the temperature your component sees

nVent says that near a device producing much of the heat the temperature can be much higher than its curve, and credits an internal circulating fan with removing hot spots and improving dissipation by as much as 10 percent.⁴ Schneider calls the top of the enclosure the hottest part.¹ Do not put a heat-sensitive part directly above a hot one, compare each device's limit with the air at its inlet, and leave the 100 to 200 mm air column and 100 mm of top and bottom space Schneider asks for.¹ Follow the drive maker's cabinet instructions first.

ABB rates ACS580 frames R0 to R3 up to +50 °C and frames R4 to R9 up to +40 °C, sends you to the hardware manual for derating above those,⁵ and in a related manual gives 1% per 100 m above 1,000 m.¹⁰ A drive that derates at 40 °C inside a panel that averages 45 °C has lost output before anyone opens the door. For a machine panel, which standard you are working to comes first.

Notes

1. Schneider Electric, Control Panel Technical Guide, How to reduce damage to components through effective thermal management, reference CPTG001_EN, 2017. Supplier guide, read in a copy hosted away from se.com. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰

2. Rittal technology library, Standard-compliant switchgear and controlgear production: application of IEC 61439, 2013, worked example and conditions in the section on verification of temperature rise using calculations. Supplier guide, which predates the 2021 edition of the standard. ↩

3. AutomationDirect, How to Select and Size Enclosure Thermal Management Systems, white paper, 2017. Supplier guide. ↩ ↩² ↩³

4. nVent Hoffman, Heat Dissipation in Electrical Enclosures, Spec-00488. Supplier guide. nVent's example text says about 30 °F for 7.1 W/ft²; its graph reads about 33 °F. ↩ ↩² ↩³ ↩⁴ ↩⁵

5. ABB, General purpose drives ACS580 catalogue, revision G. Supplier data. Later revisions exist, so use the current one for your drive. ↩ ↩² ↩³

6. Rittal, Technical System Catalogue: Cooling with ambient air, 2024, pages 12 and 13. Supplier guide. ↩ ↩² ↩³ ↩⁴ ↩⁵

7. AutomationDirect, Enclosure Cooling selection overview. Supplier guide. Values given there in BTU/(h.ft².°F) and W/(°F.ft²), converted here. ↩ ↩² ↩³

8. BEAMA, Specifying a Low-Voltage and Controlgear Assembly, edition 1, July 2026, section 8. ↩

9. Pfannenberg USA, product catalogue, 2019, section on the technology of cooling. Supplier guide. ↩ ↩² ↩³

10. ABB, ACS580-0P packaged drive user's manual, 3AXD50000617333 rev B, ambient requirements table. Supplier data for a related product; check the hardware manual for your frame. ↩