Data center white space footcandle verification field guide

What a footcandle is and what the verification proves

A footcandle is one lumen per square foot, the unit of illuminance for the amount of light landing on a surface. The metric form is the lux, one lumen per square meter. Whitespace footcandle verification is the acceptance step where you measure the light actually reaching the floor, the work plane, and the rack faces, and compare it against the lighting design. It proves the room was built to deliver the light the techs need, not just that the fixtures were hung and energized.

The verification answers a narrow question. Does the as-built room put the designed illuminance where the work happens, and is it even enough across the space to be usable. A fixture count and a one-spot reading under a luminaire answer neither. You read a grid, you read it horizontal and vertical, and you check uniformity.

The reason this is an acceptance item and not a courtesy is simple. A tech reads a port label, a cable tag, and a serial sticker on the vertical face of a rack, often in a cold aisle with the rack doors closed in front of dark equipment. Light the floor to spec and leave the rack face dark and you have passed a number while failing the job the lighting exists to support.

Footcandle vs lux, and the conversion

Footcandles and lux measure the same thing in different units, so the only trap is the conversion. One footcandle equals about 10.76 lux. A drawing set on a US job usually calls out footcandles, while the meter, the fixture cut sheets, and any IES recommendation often read in lux. Mix the two and a 500 lux target reads as if it were 500 fc, which is roughly ten times the real number and an automatic false fail.

Carry the round figures. 500 lux is about 46 fc, which is why a 50 fc design and a 500 lux design are the same room described two ways. 200 lux is about 18.6 fc, call it 20 fc. 300 lux is about 28 fc. To go from footcandles to lux, multiply by 10.76. To go the other way, divide.

Set the meter to the unit your acceptance report uses and leave it there for the whole survey. Switching units mid-grid is how a transcription error gets into the record, and the report is the part that outlives everyone on the job.

Footcandle (fc)

Illuminance of one lumen per square foot, the US unit on most drawings

Lux (lx)

Illuminance of one lumen per square meter, the metric unit on most cut sheets

Why measure the rack face and not just the floor?

Horizontal illuminance is the light on a flat surface facing up, read with the meter laid flat. Vertical illuminance is the light on a surface standing up, read with the meter held against the rack face. They are different numbers in the same spot, and a white space can pass one and fail the other badly.

The rack face is vertical, and that is where techs read labels, trace cables, and seat connectors. Overhead fixtures throw most of their output straight down, so the floor reads high while the front of a rack, shadowed by its own doors and by the rack across the aisle, reads low. In a tight cold aisle the vertical reading at the middle of the rack height can be a fraction of the horizontal reading two feet away on the floor.

This is the single most skipped part of the survey. A crew reads horizontal at the work plane, hits the design, and signs off, and the techs who move in find they cannot read a port without a headlamp. The design usually calls out both numbers for this reason. If the spec gives only a horizontal figure, ask the lighting designer for the vertical target before you accept the room, because the rack face is the surface the room exists to light.

What footcandle levels does a data center white space target?

Common data center white space designs target roughly 50 fc (500 lux) horizontal at the work plane and about 20 fc (200 lux) vertical on the rack face, with aisles and circulation often a step lower and support spaces set by their own task. Those are design recommendations, not a single mandate, and the project lighting design controls the number you verify against.

The recommendation comes from a couple of places, and both are advisory. The IES (Illuminating Engineering Society) recommended practice for data and equipment spaces gives illuminance ranges for the white space and its support areas, commonly landing around 300 to 500 lux horizontal with a separate vertical figure near 200 lux. The TIA-942 data center standard carries lighting guidance in the same band, often described as staged levels keyed to whether the space is unoccupied, in transit, or occupied for maintenance, with the occupied maintenance level commonly cited near 500 lux horizontal and 200 lux vertical.

Treat these as the basis of design, not as the acceptance limit on their own. Verify the actual numbers against the project lighting design and confirm which IES or TIA-942 edition the design referenced, because the figures and the staged-level scheme are revised across editions and a spec can be tighter than the recommendation. Where the contract calls out a footcandle figure, that figure wins over the published recommendation.

Where do you measure?

You measure on a grid, not at convenient spots, because the average and the uniformity only mean something if the points are laid out evenly across the space. A common approach divides the room into a regular grid of measurement points, takes a reading at each, and keeps the points away from the immediate spill directly under a single luminaire so the average is not stacked with hot spots. The grid spacing and the work plane height come from the design and from the test method the spec references.

Read horizontal at the defined work plane. In a white space that is often near 30 in above finished floor, but data center work happens at the rack, so the spec frequently also fixes a floor-level horizontal reading and a vertical reading at a stated height on the rack face. Confirm the height before you start, because a reading taken at 30 in and a reading taken at the floor are not interchangeable.

Cover the real geometry. Read between the rows and in the aisle, read at the ends of rows where fixtures run out, and read the vertical face at the top, middle, and bottom of the rack, because the bottom of a rack in a cold aisle is the darkest spot in the room. Note the coordinate of every point against the rack or column grid so the report can be reproduced. A reading with no location is a number nobody can defend later.

The illuminance meter

Use a calibrated illuminance meter, and confirm two corrections before you trust a single reading: cosine correction and color correction. Cosine correction makes the meter read light arriving at an angle correctly instead of overweighting light straight on, which matters the moment you measure anything but a point directly under a fixture. Color correction, the V-lambda response, matches the sensor to the way the eye weighs colors, which matters under LED sources whose spectrum is nothing like the incandescent light old meters were built around.

Check the calibration date and record it. A meter out of calibration produces numbers that look authoritative and are quietly wrong, and the calibration sticker is the first thing a sharp reviewer asks for. Most labs calibrate these on a yearly cycle. If the sticker is expired, the survey is not acceptance data, it is a guess.

Hold the sensor flat and level for horizontal readings and plumb against the surface for vertical readings, and keep your body and clothing out of the light path. A surveyor in a white shirt leaning over a floor sensor adds reflected light. A dark sleeve over a vertical sensor steals it. Let the reading settle, and give an LED meter and the room itself the warm-up the next sections cover before you record anything.

What uniformity ratio is acceptable?

Uniformity is how even the light is across the space, expressed as a ratio, and a room can hit its average target while failing uniformity. Two ratios are common: maximum-to-minimum and average-to-minimum. The lower the ratio, the more even the light. A spec might call for an average-to-minimum no worse than 3 to 1, or a max-to-min no worse than some stated figure. The exact ratio comes from the design.

Average is a liar on its own. A grid that reads 90 fc under the fixtures and 15 fc in the dark patches between them can average right around 50 fc and still be a room the techs hate, because the work happens in the dark patches, not under the lamps. The eye adapts to the bright zone and then cannot resolve detail in the dim one, so a high-contrast patchwork reads worse than a lower but even level.

Run the uniformity check from the same grid you used for the average, not from a separate set of points. Find the minimum reading and the maximum, divide, and compare against the spec. When a room passes the average and fails uniformity, the fix is usually aiming, lensing, a missed fixture, or a controls zone that did not come up, not adding luminaires. Catch it on the grid and you catch it before the techs do.

Conditions the test depends on

The reading is only as honest as the conditions you took it under, and four of them move the number enough to flip a pass. The lamps have to be at full output, the room has to be finished, the daylight has to be accounted for, and the LEDs have to be stabilized. Get any of these wrong and you are measuring a room that does not exist yet.

Older lamp technologies needed a seasoning or burn-in period and lost output as they aged, so historical methods specified a lamp-lumen condition. LED is different but not exempt: output drifts and settles in the first stretch of operation and shifts with temperature, which the next section covers. The room finish matters because reflectance is part of the result. Bare gray deck, missing ceiling tiles, unpainted walls, and stacked cardboard all change the bounced light, so a survey run before the room is finished reads low and a survey run with packing material everywhere reads unpredictably.

Daylight is the contaminant people forget. Most white space has no windows, but support rooms, NOC areas, and loading do, and a reading taken next to a window at 2 p.m. includes sun you cannot deliver at 2 a.m. Measure the electric lighting on its own. Cover the daylight, measure at night, or measure the daylight contribution separately and subtract it, and write down which method you used. The acceptance is for the system you installed, not for the weather.

LED stabilization and the warm-up clock

LED output is not steady the instant you flip the switch. The diodes and the driver settle over a warm-up period, and the light shifts as the fixture reaches its operating temperature, which in a white space is also affected by the cold-aisle airflow blowing across the luminaires. Read too early and the number is wrong in whichever direction the fixture happens to be drifting. The fix is a clock.

Let the lighting run before you record. A common practice is to energize the room and let the fixtures stabilize for a stretch of minutes before the first reading, and to keep the air handling running because the moving air changes the fixture temperature and therefore the output. The exact stabilization time comes from the fixture manufacturer and the test method, so confirm it rather than guessing.

This is the rookie failure on LED acceptance. The crew shows up, energizes, and starts reading in the first minute to save time, and the survey records a room that has not warmed up. The reading and the design disagree, everyone chases the fixtures, and the real problem was the clock. Start the warm-up, do something else, then read.

Emergency and egress lighting verification

Egress and emergency lighting is a separate verification with its own numbers, and it is life-safety, not task lighting, so it does not trade against the white space footcandle target. It proves people can get out when normal power fails. You verify it on backup power, with the normal lighting killed, which means simulating the power loss and confirming the emergency source actually carries the load.

The numbers come from NFPA 101, the Life Safety Code, and the building code (IBC), and they are measured at the floor along the path of egress. NFPA 101 calls for emergency illumination that provides an initial average of not less than 1 fc and not less than 0.1 fc at any single point along the egress path at floor level, sustained for at least 90 minutes, with the level permitted to decline by the end of that period to a lower stated average and minimum. It also caps the maximum-to-minimum ratio along the path at 40 to 1 so there are no dark gaps between bright spots. Confirm the exact figures against the adopted edition and the AHJ.

Verify it the way it has to work. Drop normal power, time the transfer so the lights come on without an unacceptable gap, walk the egress path with the meter at floor level, and confirm the level holds for the full duration on battery or generator. Emergency fixtures and the inverter or unit equipment are commonly listed to UL 924, so check the listing and the monthly and annual test provisions the code requires. This is the part where blunt is the right tone: if the egress path goes dark on power loss, nothing else about the room matters.

Drivers, dimming, and occupancy controls

Modern white space lighting is dimmable and on occupancy controls, and the controls decide what level the room is actually at when you measure. Read the room in the wrong control state and you verify a number nobody will ever see in operation. Before the grid, put the system into the lighting scene the design assigns to occupied maintenance, confirm it is at full programmed output and not held down by a daylight or energy setpoint, and lock it there for the survey.

Occupancy and vacancy sensors are the common trap. A white space often runs at a low standby level when empty and ramps to full when a tech enters, which is how the design meets the energy code. During the survey the sensors will see the surveyor and ramp the local zone, but a zone you are not standing in may have dropped back to standby, so you read a hot spot under yourself and dim everywhere else. Override the sensors to full for the acceptance reading, then verify the automatic behavior separately as a controls test.

Dimming drivers also affect the lowest levels and the warm-up. A driver dimmed to a setpoint puts out less than its rated lumens by design, and some drivers shift color slightly as they dim. Verify at the design operating level, record the control state you measured in, and keep the controls functional test as its own line item so the lighting acceptance and the controls acceptance do not get tangled into one ambiguous pass.

Color temperature and CRI for tech work

Footcandles tell you how much light there is, not how good it is for the work, and two color metrics decide the second part: correlated color temperature (CCT) and color rendering index (CRI). They are not acceptance pass or fail in the footcandle sense, but they belong on the survey because a room can hit the illuminance target and still be a poor place to read a color-coded cable.

CCT is the warmth or coolness of the light in kelvin. Data center white space commonly uses a neutral to cool source, often in the 4000K to 5000K range, because it reads as crisp and alert for detailed work. The exact CCT is a design choice, and consistency matters as much as the number, since mismatched fixtures from different batches make a room look patchy even at the right footcandles.

CRI is how faithfully the light shows color against a reference, on a scale to 100. Cable management and connector work lean on color, so a low-CRI source that washes out the difference between a blue and a green jacket is a real problem on the floor. Many designs call for a CRI in the 80s or higher. Confirm the delivered fixtures match the specified CCT and CRI, because a substitution that holds the lumens but drops the CRI passes the footcandle check and fails the people.

Reflectance and room finish

Reflectance is the fraction of light a surface bounces back, and it is part of the illuminance result, not a side detail. The lighting design assumes specific reflectances for the ceiling, walls, and floor, and the calculated footcandles depend on that bounced light. Finish the room differently than the design assumed and the measured number moves, often enough to matter.

White space tends to run light ceilings and reflective floor finishes precisely to bounce light onto the vertical rack faces the fixtures cannot reach directly. A dark raised-floor tile, a deck left bare instead of painted, or a wall color darker than the design erodes that bounce and pulls the vertical readings down first, because the vertical face depends most on reflected light. The horizontal floor reading is more forgiving because it gets direct downlight.

When a survey reads low on the rack face but the fixtures and the controls check out, look at the finishes before you blame the layout. Confirm the actual ceiling, wall, and floor finishes against the reflectances the design assumed. A room that was value-engineered to a darker floor after the lighting was calculated will read low and there is no fixture fix for it, only more light or a lighter finish.

Glare and screen readability in a NOC

More light is not always better, and the NOC is where that flips. A network operations center is full of screens, and the goal there is enough light to work without veiling the displays in glare and reflections. The white space wants high vertical illuminance on the racks. The NOC wants controlled, lower-glare light on the desks and walls so operators can read both their monitors and a printed runbook. Verifying the two spaces to the same instinct is a mistake.

Glare comes from luminaires too bright in the field of view and from light reflecting off screens and glossy surfaces back at the eye. In a NOC you check that the design used low-glare fixtures and positioned them to keep reflections off the displays, and you read horizontal at the desk against the NOC target rather than chasing a white-space vertical number. A NOC lit like a white space puts hot reflections on every monitor.

In the white space itself, glare matters less but does not vanish. A tech working at a rack faces the fixtures overhead, and a poorly shielded high-output luminaire directly in the line of sight makes the dark rack face harder to read, not easier. Note glare as a qualitative observation on the survey even where the spec gives no number, because a room that passes footcandles and blinds the worker has not really passed.

The grid and the acceptance report

The acceptance report is the grid, point by point, with every reading tied to a location. It is not a single average and a checkmark. The reviewer, the owner, and the next person to touch the room need to see where each number came from, so the report carries the coordinate, the horizontal reading, the vertical reading, the local uniformity, and the pass or fail against the design at that point.

Build the report from the grid you actually walked, with the room laid out the way the floor is addressed, by rack row and column. A reviewer should be able to put a finger on a low reading and know exactly which aisle and which rack face it came from. That traceability is what turns a survey into acceptance data, and it is what a tool like FieldOS captures on the floor so the grid, the readings, and the meter record stay attached to the location instead of living on a clipboard.

Include the conditions and the instrument record with the grid. The control state you measured in, the daylight handling, the warm-up time, the meter and its calibration date, and the design targets all belong on the same document as the numbers, because a footcandle with no context is not evidence. The report is the deliverable. The light is just what it measures.

Field example: a white space row that passes the floor and fails the face

Take a cold aisle between two rows of racks, design target 50 fc horizontal at the work plane and 20 fc vertical on the rack face, average-to-minimum uniformity no worse than 3 to 1. The crew lays a grid down the aisle and reads horizontal first. The floor comes in between 49 and 55 fc across the points, averaging about 52 fc. The horizontal passes clean, and on a survey that stopped there, the room ships.

Now read the rack face. At the top of the rack the vertical reads near 24 fc. At the middle it falls to about 18 fc. At the bottom, in the shadow of the closed rack doors and the equipment across the aisle, it reads 11 fc. The average vertical is below the 20 fc target and the bottom-of-rack minimum drags the uniformity past 3 to 1. The same aisle passed horizontal and failed vertical, which is the most common real result in a white space.

The cause was not too few fixtures. It was a darker floor tile substituted after the lighting was calculated, which killed the bounce onto the lower rack face, plus an occupancy zone two racks down that had dropped to standby and was not overridden. Lighter floor finish and the controls override brought the bottom-of-rack reading up to about 17 fc and the uniformity back inside the ratio. The number that mattered, the one the techs read by, was the one the floor reading never showed.

What if a zone fails the lighting check?

When a grid point or a whole zone comes in under target, work the cause in order before you reach for more fixtures, because adding light is the slowest and most expensive fix and usually the wrong one. Most failures trace to the control state, the conditions, the finishes, or a single dead luminaire, not to a layout that is short on fixtures.

Start with what is cheapest to rule out. Confirm the zone is at full programmed output and not held down by an occupancy sensor in standby or a daylight or energy setpoint. Confirm the warm-up clock was honored and the air handling was running. Check for a fixture that did not energize, a driver dimmed below design, or a lens or aiming problem throwing the light wrong. Then check the finishes against the design reflectances, since a dark substituted floor or unpainted deck pulls the vertical readings down with no fixture fix available. A loose or wrong-CCT fixture from a different batch shows up here too.

Only after those come back clean do you treat it as a design or quantity shortfall and bring the lighting designer back in. Document the failing points with their coordinates and the cause you found, re-survey the zone after the correction, and keep both surveys. A fail that gets quietly fixed and never re-read is a fail still sitting in the record with no resolution.

Field checklist

What to document

The verification is worth exactly as much as the record behind it. Six months out, when a tech complains a row is too dark, the report is what answers whether the room was ever right and what condition it was measured under. Capture every reading with its location and capture the conditions that make the reading mean something.

Record the grid point coordinate, the horizontal footcandles, the vertical footcandles at the rack face, the local and overall uniformity, the instrument and its calibration date, the control state and warm-up honored, the daylight handling, the design target at that point, and the pass or fail against it. If a point failed and was corrected, record the cause, the fix, and the re-survey reading. Tie the whole grid to the room as it is addressed on the floor so it can be reproduced.

Common mistakes

• Reading before the LEDs have stabilized, so the survey records a room that has not warmed up.
• Measuring only horizontal at the work plane and never reading the vertical rack face.
• Hitting the average target and never computing the uniformity, so a dark patchwork passes.
• Using an uncalibrated or out-of-date meter and trusting numbers that look authoritative but are wrong.
• Letting daylight from a window contaminate a reading in a NOC, support room, or loading area.
• Leaving occupancy sensors active so zones you are not standing in read low in standby.
• Confusing footcandles and lux, reading a 500 lux target as 500 fc or the reverse.
• Surveying the room before the finishes are in, so the assumed reflectances are not there yet.
• Treating the white-space task target and the life-safety egress target as the same verification.

Standards and references

Name the body that governs the point and label whether it is a recommendation or a mandate, because the white-space footcandle target and the egress requirement sit on opposite sides of that line. The illuminance design levels are recommendations. The egress lighting requirement is enforceable life-safety law.

For the white-space and support-area levels, the IES (Illuminating Engineering Society) Lighting Handbook and the IES recommended practice for data and equipment spaces give illuminance ranges for the work plane and the rack face. These are design recommendations, not a code mandate. The TIA-942 data center standard carries lighting guidance in the same band, often as staged levels keyed to occupancy. Both are revised across editions, so confirm which edition the design referenced and treat the project lighting design as the controlling document. Do not cite a specific recommended-practice number or table unless you have confirmed it in the edition on the job.

For egress and emergency lighting, NFPA 101, the Life Safety Code, and the building code (IBC) set enforceable minimums at the floor along the path of egress, including the average and minimum footcandle figures, the 90-minute duration, and the maximum-to-minimum ratio. Emergency lighting equipment is commonly listed to UL 924. For lighting controls and energy limits, the energy code (ASHRAE 90.1 or the IECC, whichever the jurisdiction adopted) governs the occupancy sensing and the lighting power, which is why the controls fight the acceptance reading. The adopted editions and local amendments control, so confirm them with the AHJ before citing a number on a submittal.

Units, terms, and conversions

Illuminance shows up in two unit systems and under a few names, so the same target can read differently across a drawing set, a fixture cut sheet, and an IES table. Footcandles on the drawings, lux on the cut sheets, and the conversion between them is the part that has to be right.

One footcandle is one lumen per square foot and equals about 10.76 lux, one lumen per square meter. Illuminance is the light landing on a surface, measured horizontal at the work plane or vertical at the rack face. Luminance, a different word people swap in by accident, is the brightness of a surface as seen, not what this verification measures. Uniformity is a ratio of the readings across the grid. CCT is color temperature in kelvin, and CRI is the color rendering index on a scale to 100.

Footcandle (fc)

Illuminance of one lumen per square foot; 1 fc is about 10.76 lux

Lux (lx)

Illuminance of one lumen per square meter, the metric unit

Lumen

The unit of luminous flux, the total light a source emits

Illuminance

Light arriving on a surface, read horizontal at the work plane or vertical at the rack face

Uniformity ratio

Evenness of the light across the grid, as max-to-min or average-to-min

CCT

Correlated color temperature in kelvin, the warmth or coolness of the light

CRI

Color rendering index on a scale to 100, how faithfully the light shows color

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What load rating does a data center raised floor need?
• What is rack readiness in a data center?
• What is white space in a data center?
• What megger voltage do I use on busway?
• How do data centers control physical access?