A generator room ventilation design often arrives as a louvre schedule: two openings, a free area and a note that the radiator is ducted. It says nothing about the pressure the fan works against, the air the engine actually breathes, or what happens when an acoustic louvre is added later. This page sets out the air budget, the louvre calculation, the pressure path and the temperature rise that turns a nominal rating into a site rating.
Start with the heat and the air paths, not the louvre size
A generator room is a series circuit. Air enters through an intake, crosses the room, passes through the radiator or leaves through a discharge path, and exits through a second opening. Every element adds resistance, and the radiator fan can only move the airflow its curve allows at the pressure it sees. A louvre is one term in the same pressure budget as the duct, the damper and the discharge hood.
There are two air streams, and treating them as one is the most common planning error. Combustion air is consumed by the engine and must be clean, cool and available at the intake. Cooling air passes through the radiator core and carries the cooling-system heat outside. Combustion air is a small fraction of the total; radiator air is the dominant flow, and a room with ample combustion air can still overheat because the core cannot get enough cooling air.
There is a third path that is not designed at all: leakage and recirculation. Hot air leaving the discharge can be drawn back into the intake if the openings are close, face each other, or share a corner. The room then runs several kelvin hotter than the outdoor air, and the set derates for a reason no datasheet shows. Opening layout is part of the ventilation design, not an architectural afterthought.
| Path | What it carries | Sizing input | Symptom when wrong |
|---|---|---|---|
| Combustion air | Air consumed by the engine, plus a small allowance for ventilation of the engine bay | Engine datasheet combustion-air figure, about 0.10-0.15 m3/s per 100 kW | Black smoke, high exhaust temperature, rough running at load |
| Radiator cooling air | The heat rejected by the coolant and the charge-air cooler | Radiator airflow at the site ambient, from the radiator datasheet | Coolant temperature climbs at load; the set derates or trips |
| Residual room heat | Radiant heat, exhaust system heat and the part of the radiator heat not ducted out | About 5-10% of rated heat where the radiator is ducted; more when it is free-blowing | Room temperature rises above the outdoor ambient even when the radiator is ducted |
| Leakage and recirculation | Hot discharge air returning to the intake, or untreated air entering through gaps | Separation of openings, wind direction, building geometry | Hot room at light load; unstable temperature readings; dust ingress |
The air budget: combustion air is small, cooling air is the load
The engine datasheet gives combustion air as a flow at rated power. The published planning range on this site is about 0.10 to 0.15 m3/s per 100 kW, which puts a 500 kVA set at roughly 0.4 to 0.6 m3/s and a 1000 kVA set at 0.8 to 1.2 m3/s. Those figures are small compared with radiator flow, which is why an intake sized only for combustion air will not cool the room.
The radiator airflow is the larger number. The planning budget used here is 0.02 to 0.03 m3/s per kW of set rating at a 10 K room rise, which gives about 8 to 12 m3/s for a 500 kVA set and about 16 to 24 m3/s for a 1000 kVA set. These are first-pass figures for checking a layout before the radiator datasheet is available; the datasheet airflow at site ambient governs the fan, louvre and duct.
Two consequences follow. First, the cooling-air opening is several times the combustion-air opening in free area, and it is the one that sets the louvre size. Second, the combustion-air intake should draw clean, cool air and not compete with the radiator for the same pool. If the engine intake sits in the hot discharge plume, the engine breathes air hotter than the room and derates earlier than the room temperature alone suggests.
| Item | 500 kVA set (about 400 kW) | 1000 kVA set (about 800 kW) | What governs the final figure |
|---|---|---|---|
| Combustion air | 0.4-0.6 m3/s | 0.8-1.2 m3/s | Engine datasheet at rated power and site ambient |
| Radiator cooling air | 8-12 m3/s | 16-24 m3/s | Radiator datasheet airflow at the site ambient |
| Residual heat left in the room | About 5-10% of rated heat when ducted | About 5-10% of rated heat when ducted | Ducting arrangement, radiant surfaces and exhaust routing |
| Room temperature rise target | Below 10 K | Below 10 K | Site ambient, radiator air-on limit and project specification |
| Reference ambient | 40 °C | 40 °C | Site design ambient and manufacturer derating data |
Turning airflow into louvre area
A louvre is quoted by its gross opening, but the air passes only through the free area between the blades. The ratio between the two depends on the blade profile, the pitch and the accessories, and it is a manufacturer’s figure. A weather, acoustic or sand louvre with the same gross dimensions can have very different free areas, so a schedule quoting only the opening size is not a ventilation design.
The first check is face velocity. The published planning range on this site is 2.5 to 5 m/s through the free area. At 10 m3/s, that gives 2 to 4 m2 of free area. If the louvre has a 50% free-area ratio, the gross opening is 4 to 8 m2; at 70%, it is about 2.9 to 5.7 m2. The free-area ratio and the pressure drop at design flow both come from the supplier, because two louvres with the same ratio can behave differently with rain, sand or acoustic blades.
Acoustic louvres are the usual place where the airflow disappears. The absorptive blades occupy space, the free area falls, and the pressure drop rises; the noise page on this site sets out the acoustic side, and the ventilation calculation has to use the same louvre data. A louvre chosen for its sound rating and then checked for free area is a different component from one chosen for area and expected to be quiet. Both duties must be settled together, with the pressure drop at design flow stated by the supplier.
| Step | Calculation | Example at 10 m3/s | What to confirm |
|---|---|---|---|
| Free area | Airflow divided by face velocity | 2-4 m2 | Use the velocity range in the project specification |
| Gross area at 50% free area | Free area divided by 0.50 | 4-8 m2 | Louvre supplier’s free-area ratio |
| Gross area at 70% free area | Free area divided by 0.70 | About 2.9-5.7 m2 | Blade profile and accessories |
| Pressure drop | Read from the louvre curve at the design flow | Supplier data | Rain, sand, bird screen and acoustic blades all add resistance |
| Effective area check | Apply the published 1.5-2.0 times allowance | Confirm against the opening schedule | Allowance is a planning check, not a substitute for the supplier’s data |
The pressure budget is where most rooms lose their airflow
A radiator fan is selected to move a certain airflow against a certain external static pressure. The fan curve is fixed; the room is not. Every louvre, damper, duct, bend, transition, bird screen and discharge hood adds resistance, and the sum has to stay inside the pressure the fan can develop at the required flow. If the sum is higher, the fan moves less air, the radiator sees a lower mass flow, and the coolant temperature rises at the same load. The set has not changed; the room has taken its airflow away.
The pressure budget should be written down before the openings are ordered. It is not enough to ask whether a louvre is large enough; the question is what pressure drop it produces at design flow, and whether the total leaves the fan on the right part of its curve. A ducted radiator adds the duct and its bends; a free-blowing radiator adds the discharge path. A sand filter or acoustic baffle added after fan selection changes the fan’s duty, not an accessory.
The measurement that closes the loop is a pressure reading across the room and the radiator at full airflow. If the measured external static pressure is above the fan’s limit at that flow, the airflow is below the datasheet figure even if the louvre looks generous. The fix is to remove resistance, not add a larger fan without rechecking the path.
| Element | What it adds | Data to obtain | Common mistake |
|---|---|---|---|
| Intake louvre | Resistance at the design flow, plus rain or sand protection | Pressure drop curve from the louvre supplier | Sizing by gross area only |
| Acoustic louvre or baffle | Higher resistance for the same opening | Pressure drop and free-area ratio at the design flow | Adding it after the fan and louvre are fixed |
| Intake duct and bends | Friction and dynamic losses | Duct size, length, bend radius and flow | Using a duct smaller than the radiator face |
| Discharge path | Resistance from the radiator outlet to the outside | Discharge duct, hood, louvre and separation from the intake | Discharging into a corner or against a wall |
| Dampers and screens | Resistance plus a failure point | Damper pressure drop, actuator position feedback and screen data | No proof that the damper is open before load |
| Total | Sum compared with the fan’s external static pressure capability | Radiator fan curve at the site ambient | Assuming the fan will “pull harder” |
Temperature rise: the number that decides the derate
The room temperature rise is the difference between the air entering the room and the air the radiator sees. The published design target on this site is below 10 K. At a 40 °C reference ambient, a 10 K rise puts radiator air-on at about 50 °C. The engine and radiator then operate above the ambient at which the set was rated, and the derate follows from the manufacturer’s data.
The published rule of thumb is about 1-2% of rated power for every 5 K above the 25 °C reference. A 10 K rise is a real loss, not a comfort issue, and usually the cheapest to remove because it is fixed with openings and airflow rather than a larger machine. The exact derate at 40 °C or 50 °C comes from the engine and radiator curves, not from the rule of thumb alone.
Measurement matters because the room is not isothermal. Air can stratify, the radiator can draw from a hot pocket near the ceiling, and the engine intake can sit in a different zone from the temperature sensor. Readings should be taken at the intake louvre, at the radiator air-on face, at the radiator outlet, and at working height in the room, at steady load. A single sensor on the wall can read comfortable while the radiator is breathing air several kelvin hotter.
| Reading | Where | What it tells you | Action if outside the design |
|---|---|---|---|
| Outdoor ambient | Shaded, away from the discharge | The reference the room rise is measured against | Recheck the site design ambient |
| Intake air | At the intake louvre, inside the room | Whether the intake is drawing its own hot discharge | Separate or reorient the openings |
| Radiator air-on | At the radiator face, several points | The air temperature the cooling system actually sees | Increase airflow or remove recirculation |
| Radiator air-off | At the discharge, before dilution | The heat the radiator is rejecting | Compare with the radiator datasheet |
| Room working height | About 1.5 m above the floor | Stratification and the environment for service | Check path and mixing, not only louvre size |
| Room rise | Radiator air-on minus outdoor ambient | Whether the below-10 K target is met | Fix the airflow before accepting the installation |
Free-blowing and ducted radiators: what the ventilation has to remove
When the radiator is ducted outside, most cooling-system heat leaves through the duct, and the residual in the room is about 5-10% of rated heat plus radiant heat from the engine and exhaust. That residual still has to be removed, which is why a ducted room is not a sealed room. The residual opening is smaller than the radiator opening, but it is not zero, and it must work at full load on a hot day.
When the radiator is free-blowing inside the room, the room itself is the discharge plenum. The full radiator airflow has to enter the room and leave through the discharge opening, and the discharge path becomes part of the pressure budget. A free-blowing layout can work well, but it needs more room volume, a clear path from intake to radiator to discharge, and a discharge that does not short-circuit to the intake. Ducting is not automatically better; it moves resistance into the duct and can worsen the fan duty if undersized.
The rest of the plant has to be considered as well. A data centre with a PUE of 1.4 rejects 1.4 kW of heat for every 1 kW of IT load, but the generator room sees only the generator’s own heat when the set runs; the two heat balances are separate. What the room shares with the data hall is the need to keep controls and dampers powered. The UPS covers the seconds between mains loss and the generator carrying load, but not a failed fan or a damper that never opened.
| Item | Ducted radiator | Free-blowing radiator | What to verify |
|---|---|---|---|
| Heat leaving through the duct | Most of the radiator heat | None; the room is the plenum | Duct airtightness and discharge location |
| Heat left in the room | About 5-10% of rated heat plus radiant heat | The full radiator heat plus radiant heat | Residual ventilation and room temperature rise |
| Pressure path | Intake louvre, duct, radiator, discharge duct | Intake louvre, room, radiator, discharge opening | Total against the fan curve |
| Recirculation risk | Lower, if the duct discharges away from the intake | Higher, because the discharge is near the room | Separation and wind direction |
| Service access | Duct and radiator clearance | Clear space around the set and the discharge | Maintenance manual clearances |
Commissioning and the annual load test
The ventilation design is proved at load, not at idle. A cold start proves the fan turns and the damper opens; it does not prove the room can reject full-load heat at 40 °C. The annual load test is written as 50% to 100% of rating, and ventilation readings should be taken at each step: airflow, external static pressure, room rise, radiator air-on and air-off, coolant temperature and damper position. The load bank page sets out repeatable load points; the commissioning checklist sets out the handover record.
The fuel figures show why the air demand follows load rather than average demand. A 500 kVA set burns about 101 L/h at full load, 80 L/h at 75% and 60 L/h at 50%, at 0.24-0.29 L/kWh; a 1000 kVA set is about 200 L/h, or roughly 4800 L in 24 hours, at full load. The ventilation system must be sized for full-load heat rejection even if the site usually runs at part load. Sizing for the average leaves the room short on the day the building demands the rating.
Part-load operation has its own trap. Running at 10-20% load for long periods does not keep the engine at working temperature, and the published maintenance guidance shortens the oil interval for that pattern. Ventilation cannot fix a set that is too lightly loaded; it can only keep the air it sees cool and clean. The annual test at 50-100% confirms both the machine and the room.
| Load point | 500 kVA fuel rate | What it means for ventilation | Record at commissioning |
|---|---|---|---|
| 100% | About 101 L/h | Full heat rejection; the room must meet the design rise | Airflow, static pressure, room rise, coolant temperature |
| 75% | About 80 L/h | Lower heat, but still a useful check of the fan curve | The same readings at a repeatable load point |
| 50% | About 60 L/h | Part-load; confirms the damper and control response | The same readings, plus damper position |
| 10-20% | Below the 50% point | Low heat rejection; the engine may not reach working temperature | Note the running pattern for the maintenance interval |
| 1000 kVA reference | About 200 L/h; about 4800 L in 24 hours | Twice the example set; the air budget scales with the rating | Use the actual radiator datasheet, not a linear guess |
Send the room drawing and the radiator datasheet
How much ventilation does a generator room need?
Start with the radiator airflow at the site ambient, then add the residual room heat and combustion air. The published budget is 0.02-0.03 m3/s per kW at a 10 K rise, about 8-12 m3/s for a 500 kVA set, plus 0.10-0.15 m3/s per 100 kW of combustion air. The radiator datasheet is the final figure, and the louvre and duct must pass that flow within the fan’s pressure capability.
Can combustion air come from the same louvre as the cooling air?
It can, if the louvre is sized for the combined flow and the engine intake draws clean air rather than the radiator’s hot discharge. The risk is a combustion intake in the plume, or a shared opening sized only for the engine. A separate combustion-air path is easier to prove; a shared opening needs a layout that keeps the streams from fighting.
How do I size a louvre for a 500 kVA generator?
Take the radiator airflow, about 8-12 m3/s on the published budget, and divide by the design face velocity of 2.5-5 m/s to get the free area. That gives 2-4 m2 of free area at 10 m3/s. Divide by the free-area ratio for the gross opening: about 4-8 m2 at 50%, or about 2.9-5.7 m2 at 70%. Then check the pressure drop at design flow against the fan curve; area alone does not prove airflow.
Why is the room hotter than the outside air?
Because heat is left in the room, or because hot discharge air is coming back into the intake. A ducted radiator leaves about 5-10% of rated heat in the room plus radiant heat; a free-blowing radiator leaves the full radiator heat. Recirculation adds to either. Measure intake, radiator air-on and outdoor ambient together before changing the louvre size; the fix may be separation, not more free area.
Do acoustic louvres reduce airflow?
Yes, for the same gross opening. The absorptive blades take up space, so free area falls and pressure drop rises. If the acoustic duty is added after ventilation was sized, the fan sees more resistance and moves less air. Select the acoustic louvre and ventilation together, using the supplier’s free-area ratio and pressure-drop curve.
What should be measured at commissioning?
At steady load, record outdoor ambient, intake air, radiator air-on and air-off, room temperature at working height, external static pressure, damper position and coolant temperature. Repeat at 50%, 75% and 100% load where the load bank allows. The room rise should stay below the 10 K target, and pressure inside the fan’s limit.
Tell us the room dimensions, radiator airflow at the site ambient, louvre types and summer design temperature. The reply sets out the air budget, louvre free area, pressure path and commissioning readings, in writing.
Written and maintained by the KVA Grid sales engineering desk. The planning figures on this page are published elsewhere on this site: the room air budget of 0.02-0.03 m3/s per kW at a 10 K rise, combustion air of 0.10-0.15 m3/s per 100 kW, the 5-10% residual heat, the 2.5-5 m/s louvre velocity, the 1.5-2.0 times effective-area allowance, the below-10 K room rise and the 500 kVA airflow of 8-12 m3/s from generator room design; the 40 °C reference and derating from hot and humid climates; the fuel figures of 0.24-0.29 litres per delivered kWh and 101, 80 and 60 litres an hour for a 500 kVA set at full, 75% and 50% load from fuel consumption; the annual load test at 50% to 100% and the 10% to 20% light-load note from maintenance scheduling; the UPS bridging the seconds to generator pickup, the PUE 1.4 example and the 1000 kVA case at roughly 200 litres an hour and about 4800 litres in 24 hours from data center backup power; acoustic louvre behaviour from noise and attenuation; sand and coastal air from enclosure rating and corrosion; repeatable load points from load bank testing; and the handover record from standby generator commissioning. Louvre ratios, pressure drops and acceptance tolerances are determined by the equipment manufacturer’s data and the project specification. Nothing on this page is a quotation for a particular project. Reviewed 4 November 2026.
