Sizing Exits, Doors and Travel Distance for a Floor

Eighteen hundred square metres, and not one of them is a person

The plate usually arrives as a landlord's lease drawing: an area in the title block, a core hatched in roughly the right place, two stairs the shell-and-core architect drew four years ago, and a demise line that has moved twice since. The first question anyone asks it is how many desks fit. The second, asked weeks later and usually by somebody else, is whether those two stairs are enough. Both are answered from the same rectangle by completely different arithmetic, and it is the second one that decides whether the floor can be let at the density the first one assumed.

Order matters because the two answers harden at wildly different speeds. A desk layout is a Tuesday afternoon. An exit is a hole through every slab in the building, a shaft wall, a discharge at ground level and a piece of the landlord's asset. If the occupant load a fit-out wants exceeds what the existing stairs can discharge, nothing inside the fit-out drawing fixes it: the available answers are fewer people, more suppression, or a third stair nobody is going to build.

The chain has four links and runs one way only: area to occupant load, occupant load to a number of exits, exits to a width for each, and then a check that nobody standing anywhere on the plate walks further than the adopted document permits to reach one. Every link has a named document behind it, and the numbers in those documents are not interchangeable between editions or across jurisdictions. What follows is the shape of the arithmetic and where each figure lives — not a substitute for the edition your authority has actually adopted, which is the only place the figures themselves are safe to read.

Counting people who are not there yet

Occupant load is not a forecast of attendance. It is a credible worst case derived from floor area by a factor keyed to use, and it is deliberately unkind, because that is its job. In International Building Code jurisdictions it comes from Section 1004, Occupant Load, and its table of maximum floor area allowances per occupant. NFPA 101, Life Safety Code, does the same work in Chapter 7, Means of Egress, with a factor table of its own. In England and Wales the equivalent is the floor space factor table in the appendices to Approved Document B, Volume 2, and BS 9999 offers a risk-profile route to the same figure for buildings that can justify it.

The trap in every one of those tables is the single word gross or net sitting against each entry. A gross factor is applied to the whole area inside the demise — corridors, washrooms, cupboards, internal wall thicknesses and all. A net factor is applied only to the area the use itself occupies. Swapping one for the other moves the answer by a fifth or more, which is enough to change the number of exits on its own, and the mistake survives review because both figures look entirely plausible on the page.

Then come the additions, which are not averaged into the main rate. Fixed seating is counted by seats rather than by area, with a stated length per person where a bench is fitted. A room whose use has its own denser factor — a training suite, a canteen, a town-hall space carved out of an office floor — is counted at that factor and added on top. A mezzanine puts its load into the storey below. And where an owner intends to hold more people than the table produces, the building official can accept a higher figure provided the exits are sized for it, which is why an assembly room ends up with a number screwed to the wall by the door.

The most expensive property of the number is that it may or may not stay on your floor, and which of the two it is has to be established rather than assumed, because the documents genuinely differ. The model codes size a stair on the occupant load of each storey considered individually, and add loads together only where routes from above and below converge at an intermediate level. Approved Document B and BS 9999 size an escape stair on the load accumulated from the storeys it serves, which is a different and usually larger number. Ask which of those two rules the adopted document applies, then ask for the design occupant load of every storey each stair serves, before assuming the shell-and-core figure still holds. A building let at one person per eight square metres, when its cores were designed at one per ten, has already spent capacity in a shaft that belongs to somebody else.

One exit, two exits, and the diagonal that settles it

How many exits a storey needs is a function of occupant load and, in several documents, of height and use as well. The threshold questions come first, and they are cliff edges rather than gradients: below a stated load and within a stated common path allowance, a space or a storey may have a single exit, and one occupant over the line it needs two. The IBC sets this out in Section 1006, Number of Exits and Exit Access Doorways, with separate tables for spaces and for storeys; NFPA 101 places it in the occupancy chapters; Approved Document B frames the same decision as whether escape in a single direction is acceptable at all, and for how far.

Two exits on a plan is not two exits in the code's sense. They have to be remote from one another, and remoteness is measured rather than judged: draw the diagonal of the area served, and the exits must be separated by at least half its length, reduced to a third where the building is sprinklered throughout. The separation is taken between the nearest edges of the two doorways, in a straight line, whether or not walls stand between them. Two doors at the same end of a plate satisfy a count and fail the diagonal, and that failure is dangerous precisely because the drawing looks compliant. The reason behind the rule outranks the arithmetic wherever no formula is written down: two exits exist so that one fire cannot take both. A pair either side of a single open kitchen, or a second exit reached only by walking past the first, is a scheme an assessor will push back on even where the diagonal passes comfortably.

1. Fix the area the code counts, and state whether it is gross or net so it matches the factor about to be applied.
2. Apply the factor from the adopted edition, adding fixed seating and any denser sub-use as separate figures.
3. Read off the number of exits the load and the common path allowance require, for each space and then for the storey.
4. Draw the diagonal of the area served and measure the separation between the two furthest-apart exit doorways against it.
5. Ask what occupant load every stair is already carrying from the storeys above and below yours.

Width is sold by the occupant, and the stair charges more

With the load fixed, egress width becomes a rate. The model figures in IBC Section 1005, Means of Egress Sizing, are 0.3 inches of clear width per occupant for stairways and 0.2 inches per occupant for every other component of the route — doors, corridors, ramps, level passage. Where a building is sprinklered throughout to NFPA 13 and fitted with an emergency voice/alarm communication system, those rates drop to 0.2 and 0.15. NFPA 101 states the same base pair in its capacity factor table, but its reduced factors are keyed to occupancy rather than granted generally for sprinklers, so read them there rather than carrying the IBC reduction across by analogy. Approved Document B reaches the same place by a different road: a table of minimum widths of escape routes and exits against the number of people served, which becomes a millimetres-per-person rate above the tabulated thresholds.

Two features of the rate catch people out. The first is that a stair costs half again as much width per person as a level component does, which is why the stair rather than the door is nearly always the binding constraint on an upper floor. The second is that the calculated width is a floor, not a target: absolute minimum dimensions sit on top of it and routinely govern in smaller buildings, so a corridor serving a dozen people is not two and a half inches wide.

There is also a rule that stops the whole capacity being spent in one place. Where more than one exit is required, the loss of any single exit must not reduce the remaining capacity below half of what the storey requires. On a two-stair floor that means each stair carries half the load and neither may be sized to the leftovers. It rules out the scheme that meets the total by making one stair generous and the other minimal — which is exactly what happens when one stair is inherited from the shell and the new one is being squeezed into whatever corner is free.

Capacity is measured at the narrowest point of each component, and the narrowest point is rarely drawn. Handrails, an extinguisher cabinet, a surface-mounted closer arm, a column the finishes were drawn around, a radiator on a corridor wall: each takes width at the height where width counts. The model codes permit a stated projection from certain elements and none at all from others, and those allowances come from the adopted document rather than from the corridor looking generous when you stand in it.

Corridor, common path and dead end are three tapes, not one

Three different measurements get called travel distance in meetings, and they are measured from different points to different endpoints against separate limits. Common path of travel is the stretch where the occupant has no choice of direction, from the most remote point until two independent routes genuinely become available. A dead end is the length of corridor beyond the last point at which somebody could have turned the other way. Exit access travel distance is the whole walk from the most remote point to the exit itself. A corridor can clear the travel distance limit with room to spare and still fail on the dead end at its far end, and the two failures have entirely different fixes.

Corridors carry requirements alongside width. A fire-resistance rating tied to occupancy and sprinkler status sits in IBC Section 1020, Corridors, with equivalents in NFPA 101 and Approved Document B; that rating has to be continuous above the ceiling and through every penetration; and where a corridor acts as an exit passageway, or as the exit access from a stair discharge, tighter rules apply again. None of it changes the width arithmetic, but all of it changes what the corridor costs, because a rated corridor is a different partition build-up, ceiling and firestopping schedule.

The dead end is the one that catches fit-outs rather than base builds. Shell floors are usually handed over as a clean loop or a straight run between two cores, with nothing that could be called a dead end anywhere on them. Then a tenant layout drops a cellular block at one end and leaves a nine-metre spur to a meeting room. That spur is a dead end whether or not anybody names it one, and its limit is a small number keyed to occupancy and to whether the building is sprinklered.

Measuring the walk the way an assessor measures it

Exit access travel distance is measured along the natural and unobstructed path of travel, on the floor, from the most remote occupiable point to the entrance to the exit — which in a multi-storey building means the door into the protected stair enclosure, not the street outside. That endpoint is the most misunderstood part of the whole exercise, and it is misunderstood in both directions: designers include the descent and conclude the plate fails when it does not, or stop the tape at the corridor and miss the last leg to the enclosure door.

The path bends. It goes around fixed furniture, around the reception desk, around the demountable partition run sitting in the tenant package, and along the centre of the route rather than through walls. It follows the aisle a bench layout actually leaves, not the diagonal across an empty floor. Where no layout is fixed at the time of the check, the honest method is to measure the worst plausible arrangement and write down which arrangement was assumed — that assumption is the thing a future fit-out will breach without ever knowing it existed.

The limit itself sits in a table keyed to occupancy group and to sprinkler protection: IBC Section 1017, Exit Access Travel Distance; NFPA 101's occupancy chapters read with the measurement provisions in Chapter 7; and the travel distance tables in Approved Document B, which distinguish escape in one direction from escape in more than one. The sprinkler increase is the largest single lever in the calculation, and on most commercial plates it has already been spent — the floor is only compliant because the building is sprinklered, which quietly makes any proposal to value-engineer, defer or partially omit suppression an egress decision rather than a suppression one. Where the walk still exceeds the allowance, the remaining levers in ascending order of cost are to move an exit, add an exit, add a protected corridor so the measurement stops earlier, or reduce the occupied depth of the plate.

What the drawing has to say when it leaves your desk

Every number in this exercise rests on an assumption that a later drawing can break without anyone noticing. The occupant load rests on a factor and on whether the area behind it was gross or net. The travel distance rests on a furniture layout that has not been bought yet. The corridor width rests on nothing ever being mounted on its walls. The exit count rests on a sprinkler system being installed, commissioned and maintained. The refuge rests on a landing dimension surviving the finishes package.

So the deliverable is not the plan. It is the plan plus a short, dated schedule of those assumptions, naming the document and the edition each figure came from. That schedule is what the building control body reads, what the next fit-out designer reads, and what protects the work when a tenant puts a store cupboard across a dead-end limit three years later. Two further packages are then sized directly off the fixed route, and both are far cheaper to coordinate now than to retrofit: escape lighting along the route and at the points the standard calls out irrespective of spacing, and exit signage with sight lines that survive the ceiling and the eventual partition layout. Neither changes a single width. Both fail an inspection just as effectively as a missing exit does.