Eleven Words in a Forty-Page Report
The fire strategy for the refurbishment runs to forty pages, and the part that lands on your desk is eleven words: existing Category M system to be extended to L2 throughout. Every other recommendation in that report has a drawing behind it — compartment lines coloured on a plan, travel distances dimensioned, a stair marked as pressurised. Detection has a letter, a number, and an assumption that somebody downstream knows what to do with them.
You are that somebody, and those eleven words have to become a device schedule, a loop count, a zone plan, a cable specification, a battery and a tender figure. The category tells you what the system is for and roughly where it goes. It does not tell you that the third floor carries a 700 mm downstand every 6 m, that the void over the atrium link is 1.4 m deep and full of cable basket, or that the plant room ceiling is 9 m up.
The two halves of this job fail differently. Choosing the category too low is an argument — held in a meeting, before anything is bought, and the worst case is that the scope grows. Getting the grid wrong is a defect: a system that is the right category on the certificate and has holes in it, found by a commissioning engineer with a can of test aerosol in the last week of the programme, or by nobody at all.
M, L and P Are Answers to Three Different Questions
BS 5839-1 does not describe a non-domestic system as a number of detectors or a percentage of floor area covered. It describes it by category, and the categories sort by purpose rather than by extent. Category M is manual: call points, sounders, a panel, and no automatic detection at all, depending entirely on somebody seeing the fire and breaking a glass. The L categories protect life. The P categories protect property, and they exist because an insurer's interest in an empty building at three in the morning is not an occupant's interest in it at eleven.
Two things about the L range catch people reading it for the first time. The numbering is not a ladder — L5 sits outside the sequence, covering whatever a fire engineering assessment determined against a stated objective, which is why an L5 designation means nothing on its own and always arrives with a document behind it. And the categories combine: a real building routinely carries L2 for life plus P2 over the server room, because two parties are paying for two different failures and their coverage maps are not the same map.
Who chooses is worth asking out loud, because the answer is rarely the designer. In England and Wales the Regulatory Reform (Fire Safety) Order 2005 puts the duty on the responsible person; the category ought to come out of the fire risk assessment or the fire strategy, be agreed with the enforcing authority and be checked against the policy wording, with Approved Document B volume 2 and BS 9999 the route for a building designed to guidance. A category inferred by an installer from the shape of the building is one nobody has agreed to fund or certify.
None of these letters mean anything in North American practice. There, whether a building needs a system at all and whether detection is complete or partial comes out of the occupancy chapters of NFPA 101 and Chapter 9 of the International Building Code; NFPA 72 then says how the system is built, powered, spaced and tested. A specification saying L2 to a US contractor is asking for something their code has no name for.
| Category | What is covered | The failure it accepts |
|---|---|---|
| M | Manual call points and sounders only — no automatic detection | A fire starting in an unoccupied part of the building burns until somebody walks into it. Adequate only where the risk assessment says occupancy and management do the detecting |
| L4 | Circulation areas and spaces forming the escape route | A fire in any room is found only once smoke reaches the corridor, by which time the route is degrading. It protects the walk out and nothing else |
| L3 | The escape route, plus every room opening onto it | A fire in a room opening onto something other than the escape route — a cellular office off an open-plan floor — is still found late |
| L2 | L3 plus specified areas of high hazard or high consequence | Coverage stops at the boundary of whatever the risk assessment listed, so the list is the design. An unnamed store room is an uncovered store room |
| L1 | All areas of the building, with the standard's stated exceptions | Little in coverage terms; the trade is cost, false alarm exposure, and the burden of keeping every head reachable and testable for the life of the building |
| L5 | Whatever a fire engineering assessment determined, against a stated objective | Nothing implicitly — but the objective and the assessment are the specification, and an L5 label with no document behind it is not a category at all |
| P1 / P2 | All areas, or defined parts, for protection of property | Driven by the insurer rather than by escape, and frequently combined with an L category whose coverage map is different |
The Radius Is Not the Area a Head Covers
Once the category has settled which spaces are in, the rest is one piece of geometry applied over and over. The instinct is to take a floor area, divide by an area per detector, and call the result a count. The instinct is sound. The figure people reach for is not.
BS 5839-1 publishes no area per detector. It publishes a radial distance: on a flat ceiling, no point should be more than 7.5 m from the nearest point smoke detector, or 5.3 m from the nearest point heat detector, measured horizontally. NFPA 72 says the same thing from the other end — a nominal 30 ft between spot smoke detectors on a smooth ceiling, a listed spacing for heat, and the 0.7S rule putting every point within 0.7 times the spacing of a detector. Those are one statement, not two: the half-diagonal of a square of side S is S over the square root of two, which is 0.707S.
Here is where it goes wrong. A 7.5 m radius traces a circle of about 176 m², and that figure appears nowhere in the standard — it is what a detector would cover if detectors were laid out in circles, which they are not, because circles do not tile. Detectors go on a rectangular grid and the binding point in each cell is its corner. The largest square whose corners sit on a 7.5 m circle has a side of 10.6 m and an area of 112.5 m². That is the real area per head: the circle times two over pi, about 64% of it. Divide by the circle and you buy two detectors where the standard asks for three.
Put a real plate through it. An open floor 30 m by 24 m is 720 m². Divided by the circle it takes five heads; by the compliant square, seven; laid out so every point genuinely falls within 7.5 m, eight — four positions at 7.5 m centres along the 30 m dimension and two at 12 m centres along the 24 m, putting the worst corner 7.07 m from a head. The grid need not be square, which is worth money on a long thin plate: the only constraint is that the two half-spacings, squared and added, stay under the radius squared.
Then treat the result as a ceiling rather than a target. A 4 m by 26 m store is 104 m², comfortably under 112.5, and still needs two heads, because one in the middle leaves both ends 13 m away. Once a room's longest dimension exceeds twice the radius, area arithmetic has stopped being relevant.
- Take the radial distance for the detector type that space is actually getting, not the type on the rest of the drawing.
- Test each candidate grid by adding the squares of the two half-spacings — the worst point is the cell corner, never the cell edge.
- Set perimeter positions no more than half the pitch from the wall, then check the room corners.
- Measure each room's longest dimension against twice the radial distance before trusting any area division.
- Only then lay the grid over the reflected ceiling plan and start moving heads off diffusers and downstands.
The count this section works by hand, for any flat-ceilinged room: the smallest grid whose cells keep their corners inside the radius — the 30 by 24 m (98 by 79 ft) floor comes back at eight, the 4 by 26 m (13 by 85 ft) store at two — with the spacing each way and the farthest point it leaves, for BS 5839-1 smoke or heat detectors or an NFPA 72 listed spacing.
Sets the coverage: BS 5839-1 gives a radius per detector type; NFPA 72 works from the detector's listed spacing.
The spacing the detector is listed for — a nominal 30 ft (9.1 m) for smoke detectors on a smooth ceiling.
The room's longer dimension, wall to wall.
The shorter dimension, wall to wall.
Detectors needed
12 detectors
The fewest detectors in rows that keep a cell's farthest corner within the coverage radius for NFPA 72 spot detectors at their listed spacing.
- Coverage radius
- 21 ft
- Detectors along the length
- 4
- Detectors across the width
- 3
- Spacing along the length
- 24.5 ft
- Spacing across the width
- 26.33 ft
- Farthest point from a detector (a cell's corner)
- 17.98 ft
- Floor area each detector covers on this grid
- 645.17 ft²
They open the calculator with your figures already in it
Fire Detector Spacing and Count Calculator (BS 5839-1 and NFPA 72): 12 detectors — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Estimated cost — your price
This site holds no price list for this material — local prices vary too much to publish honestly. Enter your supplier's price and the result is costed with it.
What this calculation does not cover
- A flat, smooth ceiling of one height. Pitched ceilings, beams deeper than a tenth of the ceiling height, and walls or partitions near the ceiling each change the layout, and each space is laid out on its own.
- Ceiling height limits apply: BS 5839-1 gives 10.5 m (34 ft 5 in) for point smoke detectors in general, 7.5 m (24 ft 7 in) for fixed-temperature heat detectors and 9 m (29 ft 6 in) for rate-of-rise; NFPA 72 reduces heat detector spacing on high ceilings.
- Detectors are kept at least 500 mm (20 in) from walls and obstructions under BS 5839-1 and away from air supply points; the grid positions are the arithmetic before those rules move them.
- The system category, which decides which rooms are covered at all, comes first and is not part of this count.
Everything on the Ceiling Is an Obstruction, Including the Ceiling
A clean grid survives contact with a reflected ceiling plan for about ten minutes. BS 5839-1 asks detectors to stand at least 500 mm clear of any wall or obstruction; NFPA 72 keeps a ceiling-mounted spot detector back from the sidewall and at least 3 ft — about a metre — from an air supply diffuser in a ducted system. That last is the rule coordination breaks most reliably, because the diffuser layout and the detection layout are drawn by two consultants who each own a layer and neither owns the overlap. A head 400 mm from a supply grille sits in clean moving air and reads the room as smoke-free for as long as the fans run.
Structure is the bigger reduction, because a downstand does not merely interrupt the ceiling — it dams the smoke layer. BS 5839-1 treats a downstand deeper than about a tenth of the ceiling height as a wall, so a 700 mm downstand under a 3 m soffit has turned one room into a row of separate spaces, each detected on its own boundaries. NFPA 72 reaches the same place through beam depth and spacing thresholds that differ for smoke and for heat, with solid joist ceilings taking their own reduction across the joists.
Height is the third, and past a point it stops being an adjustment and becomes a different design. Both standards cap the height at which a point detector may be relied on. Above it the answer is not a tighter grid — a point head at 12 m is not slow, it is blind, because the plume has spread and cooled before it arrives — but optical beam detection to BS EN 54-12, aspirating detection to BS EN 54-20, or an engineered scheme. Atria add stratification, where hot smoke stops rising against a warmer layer under a glazed roof; the fix is detection at more than one level, not more detection at one.
Finally, environment picks the detector type, and the type resets the geometry you have just finished. A loading bay with diesel movements, a kitchen, a plant room with a steam relief, a dusty store — each is somewhere an optical head produces unwanted alarms until a zone gets isolated, and an isolated zone is an uncovered zone. BS 5839-1 gives real space to limiting false alarms, and what it points at is type selection, siting and multi-sensor detection. The cost lands in the grid: change a room from point smoke to point heat and the pitch falls from 10.6 m to 7.5 m, the area per head halves, and the count roughly doubles.
| Condition | What it does to the design | Where the number comes from |
|---|---|---|
| Supply diffuser within a metre of a head | The detector samples conditioned supply air rather than the room, and reports nothing while the fans run | NFPA 72's separation from air supply diffusers; BS 5839-1 on siting near ventilation |
| Downstand deeper than about a tenth of the ceiling height | One space becomes several. Each bay is detected on its own boundaries, not on the room's | BS 5839-1 downstand provisions; NFPA 72 beam depth and beam spacing thresholds |
| Solid joists, close-spaced | Spacing reduces across the joists while staying as it was along them, so the grid stops being square | NFPA 72 joist construction provisions, which differ for smoke and for heat |
| Ceiling above the height limit for a point detector | Point detection stops being a valid method at any spacing. Beam, aspirating or an engineered scheme replaces it | Ceiling height limits in BS 5839-1 and NFPA 72, plus the manufacturer's published data |
| Sloping or pitched ceiling | Smoke collects at the apex, which permits an increase in spacing — capped, and measured outward from the apex | The sloping ceiling provisions of both standards; neither lets the increase run unlimited |
| Room switched from smoke to heat for false alarm reasons | Grid pitch drops by nearly a third, from 10.6 m to 7.5 m, and the head count in that room roughly doubles | The 7.5 m and 5.3 m radial distances in BS 5839-1; the listed spacing on the heat detector's approval |
One Dot on the Drawing, Up to Three Detected Spaces
A detection layout shows one symbol at a position and it is easy to price as one device. On a commercial ceiling that symbol often stands for two or three, because the room is not the only space at that coordinate. Above it is a ceiling void that may be deep, may be full of combustible cable, may be doing duty as a return air plenum, and very often runs over the top of a compartment wall that stops at the tile. Below, there may be an access floor void carrying the same argument in reverse.
BS 5839-1 gives tests rather than a blanket answer: how deep the void is — 800 mm being the depth it names as the point where a void stops being negligible — whether it carries significant fire load, whether it is a plenum, whether it forms a route past a compartment boundary. A void that fails those tests earns its own head, its own address and a remote indicator on the room face beneath it, because a responding engineer under a sealed ceiling cannot otherwise tell which space alarmed, and will take down the wrong tile.
What never appears in the void arithmetic is access, and it is the line that becomes a maintenance liability: a detector above a plasterboard ceiling with no hatch will be tested once, at commissioning, then recorded as tested for fifteen years. Put the access panels on the same drawing as the heads. And one device gets miscounted the opposite way — a duct smoke detector, fitted because NFPA 90A or the local equivalent wants the plant to shut down on smoke, samples a duct rather than a room, so counting it towards the space the duct crosses leaves a hole in the plan.
When the Building Has Flats In It
Mixed use is where two standards meet on one site and the interface is the whole design. Retail at ground floor with four storeys of flats over it is not one system with a residential wing. It is a BS 5839-1 system covering the commercial units and the common parts, plus a set of independent BS 5839-6 systems, one inside each dwelling, with BS 9991 informing how the residential part was designed in the first place. The letters, the certificates and often the contractors differ on the two sides of the flat entrance door.
The question answered wrong is strategic rather than technical: does a detector inside a flat signal to the common parts panel? In a building on a stay-put strategy, generally not. Detection inside the dwelling sounds inside the dwelling so that household leaves; the common parts system covers the escape route and evacuates the block when the route is threatened. Wire flat detection into the block panel without changing the strategy and you turn out forty households every time one of them grills bacon — and a building that does that for a month is one whose occupants stop leaving. Where simultaneous evacuation has been adopted the answer changes, and what changes it is the fire strategy, not the wiring diagram.
The quantities behave differently too. The common parts loop might carry sixty devices across the corridors, the stair, the bin store and the plant. The flats carry more than that between them and are not on the loop at all: interlinked heads inside each dwelling, bought as a residential package and certificated separately. A takeoff that counts the loop and forgets the dwellings has priced perhaps a third of the detection in the building.
This is a dwelling tool, so use it as one: run it per flat TYPE — the one-bed, the two-bed, the duplex — and multiply by how many of each the block has. What comes out is the residential device total, which sits in a different package, on a different standard and often in a different contract from the loop you have been laying out, and which is the line most often missing from a mixed-use schedule altogether.
Each bedroom needs its own detector.
Each level of the home needs at least one detector, including a basement whether or not it is finished.
Smoke detectors needed
6 detectors (minimum)
This is a simplified rough minimum, not a code compliance check. NFPA 72 and your local building code have specific, more detailed placement rules (including for hallways, large bedrooms, and rooms with sloped ceilings) — always verify against your local fire code and manufacturer instructions, and consider interconnected detectors so all units sound together.
They open the calculator with your figures already in it
Smoke Detector Placement Calculator: 6 detectors (minimum) — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Estimated cost — your price
This site holds no price list for this material — local prices vary too much to publish honestly. Enter your supplier's price and the result is costed with it.
What this calculation does not cover
- Floor area never enters the arithmetic. A 40 m² (431 ft²) storey and a 200 m² (2,153 ft²) open-plan one both get a single alarm, but NFPA 72 and local codes limit how far an alarm can be from the space it protects, so a long hallway, a large open level, or a floor split into separated wings needs more than the one per level this count allows.
- Only one alarm is added outside the sleeping areas, whatever the plan. The calculation cannot see how the bedrooms are grouped, so a home whose bedrooms sit in several separate clusters has to have the total raised by hand.
- This is a device count, not electrical design and not a load calculation. It does not size or check the circuit the alarms sit on, count the interconnect conductors between them, or choose between mains-wired units with battery backup and sealed ten-year battery units — and which of those is permitted usually turns on whether the work is new build or a retrofit.
- Smoke alarms only. Carbon monoxide alarms, a heat alarm in a garage or loft, and detection for anything beyond a single-family home — flats with shared escape routes, houses in multiple occupation, commercial premises — are separate requirements designed to a detection standard, not counted from bedrooms and floors.
- Nothing here accounts for what is already on the ceiling. Alarms carry a service life printed on the unit and are commonly retired ten years from the date of manufacture, so existing units offset this total only if they are in date and of a type your code still accepts.
Zones Exist So That Somebody Can Find the Fire
Addressable systems tempted a generation of designers into thinking zoning was obsolete. It is not, because a zone is not a wiring concept — it is a search instruction. At two in the morning a crew arrives, reads a panel and the zone plan beside it, and walks. BS 5839-1 sizes zones around that walk: no more than 2,000 m² of floor area, no more than 60 m of search distance, no zone beyond a single fire compartment, and none covering more than one storey — with the exceptions the standard names, principally a building not exceeding 300 m² in total, and vertical shafts, which are zones in their own right.
An address refines that; it does not replace it. The panel can say which head, and the zone plan still decides whether that is usable by somebody who has never been in the building. A repeat panel at a second entrance, a mimic in a security office, remote indicators outside closed rooms — those get installed with the loop, not retrofitted after the first false alarm takes twenty minutes to trace.
The other half of zoning is cause and effect, the deliverable most likely to be missing on the day of witnessing. It is a matrix: for each zone or device, what happens. Which sounder circuits operate and in what pattern, whether a phased evacuation applies, which air handling units shut down, which dampers close, which lifts return to a designated floor, which hold-open devices release the doors they are holding — those doors being a joinery and ironmongery subject that sits on somebody else's schedule — and what signal, if any, leaves the building. An untested matrix is an assumption with a certificate stapled to it.
The Panel Does Not Get a Way Off the Lighting Circuit
The supply to the control and indicating equipment is a short specification with an unusually low tolerance for improvisation. BS 5839-1's power supply provisions want a dedicated final circuit from a dedicated protective device at or near the origin of the installation, serving nothing else, labelled so nobody switches it off in good faith, and not behind a residual current device shared with something that can trip it. NFPA 72 and Article 760 of NFPA 70 land in the same place in different words: a dedicated branch circuit, identified at the panelboard, accessible only to qualified personnel. Both are asking for exclusivity, not for headroom.
That distinction is the one to hold on to, because on a refurbishment the board is full and the temptation is arithmetical. Somebody notes that the corridor lighting circuit draws four amps on a sixteen amp way, concludes there is room, and puts the panel on it. The circuit passes every load check you can run and the installation is still wrong, because a lighting circuit is switched, isolated during lamp changes, and grouped under protection something else can trip. The domestic standard in the same family, BS 5839-6, expressly permits a dwelling's alarms on a regularly used lighting circuit — and it can, because a house has an occupant who notices when the lights stop working. An empty office at 3 a.m. has nobody performing that function.
So the load arithmetic still earns its place, pointed the other way. Where there is no spare way, the case for creating one is made on the circuits you already have: go through the lighting ways one at a time, establish what each carries from driver input wattage rather than nominal lamp ratings, and find the two lightly loaded circuits that can be consolidated onto a single way. Merge those, and the way you free is the one the panel takes — a documented argument a contractor can act on, which a request for a spare way is not.
Two things travel with the supply and get priced with it. Detection and alarm circuits are specified for fire survival, with BS 5839-1 distinguishing standard and enhanced grades and NFPA 70 Article 760 setting out equivalent circuit integrity requirements. And where that cable crosses a compartment line it becomes a firestopping job with its own tested detail, covered in its own guide here.
Run it on the circuits you already have rather than the one you are adding, once per existing lighting way. You are not looking for a way with room to spare — the panel may not share one however much room it has. You are looking for the two circuits whose combined load still passes, because merging those is what frees a way for a dedicated supply in a board that has none.
The total number of light fixtures on this circuit.
The rated wattage of each fixture (or lamp/driver combination).
The nominal voltage supplying the circuit.
The rating of the protective device on this circuit.
Total connected lighting load
720 W
The circuit's current draw is 6.0 A, at or below 16.0 A — 80% of the breaker rating, which NEC 210.19 and 210.20 set for a continuously-loaded circuit. Under BS 7671 the design check is instead Ib ≤ In ≤ Iz — load under device, device under cable capacity — with no 80% derate for lighting, so this comparison is on the cautious side of the British rule rather than a statement of it. Being inside the rating on this one check settles nothing about the rest of the circuit — the conductors, the overcurrent device, and the work as installed are all outside it.
- Circuit current draw
- 6 A
- Maximum continuous load (80% of breaker)
- 16 A
They open the calculator with your figures already in it
Lighting Circuit Connected Load Calculator: 720 W — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Add the equipment this sizes
This result is a specification — 720 W — not a quantity. Put the thing it sizes into your project: how many, what you call it, and your supplier’s price.
What this calculation does not cover
- Watts divided by volts is the current only at unity power factor, and lighting is not at unity. LED drivers and ballasts pull current out of phase with the voltage, so the real figure is watts divided by volts times power factor — at 0.9 that is 11% more current than shown here, and uncorrected budget drivers sit nearer 0.5, which roughly doubles it. The breaker responds to current, not to watts.
- Steady-state load says nothing about the moment of switch-on. Electronic drivers charge their input capacitors in the first milliseconds at many times running current, and enough drivers on one circuit will trip a B-curve MCB or a thermal-magnetic breaker every time the lights are turned on, on a circuit that clears this check with room to spare. That is a driver-count and breaker-curve question, and it is decided nowhere on this page.
Twenty-Four Hours of Nothing, Then Half an Hour of Everything
The standby supply is where a system grown over three refurbishments quietly stops complying. BS 5839-1 asks the batteries to hold the system quiescent for 24 hours and then drive it in full alarm for a further 30 minutes, with a longer standby period where the arrangements for attending a mains failure do not justify 24. NFPA 72 requires 24 hours of standby then 5 minutes of alarm for a fire alarm system, and a longer alarm period for emergency voice and alarm communication. The reason it fails later is that thirty devices joined the loop over ten years and nobody went back to it.
What makes the calculation less obvious than it looks is two stages of completely different shape. The quiescent stage is a small current for a long time, dominated by detector standby draw and panel electronics. The alarm stage is a large current for a short time, dominated by sounders and beacons — the loads nobody counts, because they are not detection. Both then derate for battery age and for temperature, since a riser cupboard in February is not the test lab. The charger carries its own requirement on top: recharge within a stated time while still supplying the quiescent load.
The tool below does the first stage honestly and will not do the second. Convert the battery from amp-hours at the panel's nominal voltage into watt-hours, enter the quiescent load in watts, and read the hours. The alarm stage, the derating factors and the recharge check come from the panel manufacturer's published calculation sheet, and no general-purpose runtime figure substitutes for them. Treat any answer close to the required period as a fail, because that period has to be met on a battery that has aged.
Multiply the battery's amp-hour rating by the panel's nominal voltage to get watt-hours, enter the total quiescent load in watts, and read the standby hours against the period the standard requires. Keep the efficiency figure honest — it is carrying both conversion losses and the usable capacity of an aged battery — and treat this as the quiescent stage only. The alarm-load stage is a separate and much larger calculation from the panel manufacturer's own sheet.
The total energy storage capacity of the battery bank.
The total wattage of everything running off the battery.
Inverter losses AND the battery usable depth of discharge, multiplied together.
Estimated runtime
5.67 hours
Real-world runtime varies with battery age, temperature, and how deeply it's safely discharged (especially for lead-acid batteries, which shouldn't be fully depleted) — check your specific battery manufacturer's guidance.
- Usable energy
- 1,700 Wh
They open the calculator with your figures already in it
Battery Backup Runtime Calculator: 5.67 hours — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
What this calculation does not cover
- Assumes the load is constant for the whole discharge. Anything that cycles - a fridge compressor, a well pump, a heat pump - draws its full watts only part of the time, so a runtime worked out from summed nameplates comes out far shorter than the bank really delivers, while one worked out from an average hides the peaks. The inverter's own standby draw also continues whether or not anything is switched on, and it is not in the figure unless you add it to the load.
- Runtime is an energy answer and says nothing about whether the system can carry the load at all. Whether the inverter and the battery can supply the continuous watts, and the starting surge of a motor or compressor, is a separate power question this calculation does not touch.
- Takes the bank as fully charged with nothing recharging it. No solar, generator or grid contribution is in the figure, and a bank already part-discharged when the outage began gives proportionally less than the runtime shown.
- The efficiency figure is fixed for the whole run. Capacity lost to cold, capacity faded over the bank's life, and the extra loss a lead-acid bank takes when it is discharged fast all have to be built into that one number yourself - no correction for temperature, age or discharge rate is applied.
- This is not a standby-battery compliance calculation. Fire alarm, emergency lighting and similar life-safety batteries are sized by a prescribed multi-stage calculation with its own derating and recharge requirements, set by the governing standard and the panel manufacturer's own sheet, and a general runtime figure does not substitute for it.
The Paperwork Is the Product
BS 5839-1 splits responsibility across separate certificates — design, installation, commissioning and acceptance — because these are genuinely different people who can genuinely blame each other. The design certificate is the interesting one: it records the category, every variation from the standard's recommendations, and the objectives the system was designed against, including the target for unwanted alarms. Variations are legitimate. An undocumented variation is a defect, and the difference between the two is a sentence written at the right time.
Commissioning is where the grid gets audited by physics. Every detector is functionally tested with the correct stimulus for its type — smoke or an approved equivalent for optical heads, heat for heat detectors, and not a magnet at the base, which proves the address and nothing else. Cause and effect is proved line by line against the signed matrix. Audibility is measured rather than judged: BS 5839-1 asks for 65 dB(A) in accessible parts of the building, or 5 dB(A) above any ambient noise lasting more than 30 seconds, and 75 dB(A) at the bedhead where people sleep. NFPA 72 sets public mode audibility above the average ambient and reaches the same 75 dBA at the pillow.
What is handed over is a package, not a panel. The zone plan goes on the wall beside the control equipment, oriented the way somebody walking in from the entrance is facing, because a plan drawn to north and hung in a lobby facing east gets read backwards under pressure. The as-installed drawings show where the void heads and their remote indicators actually went, which is rarely where the tender drawing put them. And a soak period is worth negotiating into the programme: whatever unwanted alarms a new system will produce, it produces in its first fortnight, and finding them while the contractor is on site is the difference between an adjustment and an occupier who disables a zone.
- Get the category in writing, with whoever agreed it and the document it came from, before drawing anything.
- Survey the ceiling rather than the floor: downstand depths, void depths, ceiling heights, diffuser positions.
- Fix the detector type per space on environmental grounds, and only then lay a grid for that type.
- Schedule voids and shafts separately, with a remote indicator and an access route against each concealed head.
- Draw zones against a 60 m search and one compartment, then get the cause and effect matrix signed.
- Secure the dedicated supply way, then re-run the standby calculation against the device count as built.
- Commission with the correct stimulus device by device, measure sound levels, and record every variation.
What has to be settled before a device schedule means anything
The count is the last thing that happens, not the first. Three of these six lines are read off documents that already exist — a fire strategy report, a reflected ceiling plan, and a distribution board schedule read alongside the panel manufacturer's data sheet — and the three that are decisions all need somebody willing to sign for them. The workspace opens on the geometry, because that is the part where a wrong number stays wrong quietly.
- The category, in writing, with the document and the name behind it — M, L1 to L5, P1 or P2, and often a combination. It is a scope statement agreed with the enforcing authority and the insurer, not something to infer from the shape of the building.
- A detector type decided per space before any grid is laid — Environment picks the type and the type sets the radius. Switching a room from smoke to heat after the layout is drawn roughly doubles that room's head count.
- Ceiling geometry, taken off a reflected ceiling plan rather than a floor plan — Downstand depths against ceiling height, void depths, height at every high-bay space, and the diffuser positions the mechanical layer owns.
- A void and shaft schedule with an access route against every concealed head — Depth, fire load, plenum use and compartment crossings decide whether a void is detected. A remote indicator and a hatch decide whether it stays testable.
- Zone boundaries drawn against the search, not the drawing sheet — 2,000 square metres, 60 m of search, one compartment and one storey, with the standard's named exceptions — then the cause and effect matrix each zone drives.
- The dedicated supply way, the cable grade and the battery, priced as one item — A dedicated circuit is a requirement rather than a preference, fire-survival cable is specified by grade, and the standby calculation is redone against the device count as built.
Opens the calculators above on one screen with the dimensions from this article already filled in. Quantities only — this site publishes no price list, because local prices vary too much to publish honestly.