Sightlines Before Sensors: A Field Method for Scaffold PIR Placement
Photo: scaffold mounted PIR intruder detector close up on UK construction site, via play-lh.googleusercontent.com
Broad coverage is not early warning
An alarm that registers a figure standing on the deck has already conceded the argument. By that point the individual has climbed, crossed a lift, and is within reach of the material that the system exists to protect. Early detection depends less on the number of sensors than on where their beams point.
The stock answer — fit a PIR every few bays, aimed outwards, and rely on the quoted 12 m range — produces systems that detect departures rather than arrivals. The approach is what needs watching.
What the detector is actually doing
A passive infrared sensor carries a pyroelectric element behind a segmented lens. The lens divides the field of view into a fan of lobes; the electronics register a change when a warm body passes from one lobe into the next. That has two consequences for placement.
First, movement across the pattern generates the strongest signal. A person walking directly towards the detector axis perturbs relatively few lobes per metre travelled. Second, the advertised range assumes a particular mounting height and a target crossing the beams. Mount the same unit at 5 m on a scaffold and look down a narrow slot of deck, and the useful detection distance collapses — often to half the figure in the datasheet.
Neither point is new. Both are routinely ignored because the sensor is positioned for convenience of bracketing rather than for the route a thief will take.
Reading the approach lines
A survey should begin outside the hoarding, not on the scaffold. On most UK urban sites the approach lines are few and predictable:
- an adjoining flat roof, extension, bin store or single-storey rear addition;
- a party wall or shared service alley giving horizontal access at first-floor level;
- the ladder stile, loading tower or hoist landing, where the scaffold invites a vertical move;
- pallets, stillages or soil stacks left against the elevation, which convert ground level into a step;
- the pavement-to-scaffold gap where a boarded gate has been left on the latch.
Almost none of these involve a frontal climb up the elevation. The intruder arrives along a line, then steps across onto the structure. The sensor should be looking at the line, not at the structure.
Applying the 45-degree principle
A useful shorthand when setting the head: aim so that the principal approach crosses the detection pattern at roughly 45 degrees to the sensor axis. At that angle a walker crosses the greatest number of lobes in the shortest distance, which maximises signal and minimises the time between entry into the corridor and the trigger.
Two failure modes sit either side of that angle. Set the sensor nearly parallel to the approach and the target clips the edge of the pattern, generating a weak and brief event — or none at all if the person hugs the wall. Set it square to the approach and the target walks at the axis, changing few lobes until it is almost on top of the unit.
In practice a single head rarely covers a whole elevation. Zoning should give each approach line its own corridor, with overlapping patterns at the junctions — the top of a ladder, the turn from an alley onto a roof — where a thief's direction of travel changes and detection is most reliable.
Height, angle and the dead zone under the bracket
Mounting height is the most commonly misjudged variable. A high mount feels secure and looks tidy, but a downward-looking head at 5 m produces a strip of detection some distance out and a blind cone directly beneath. Anything happening at the foot of the scaffold, or on the first lift, may never cross a lobe.
On exposed scaffolds a two-level arrangement works better than one elevated head: a lower unit covering the ground-level approach and the first lift at a shallow angle, and a second unit covering the deck run along its length. Where the sensor has to look across a public footpath, a dual-technology unit — PIR with microwave — allows the microwave element to be ranged down and the PIR lobes masked, which reduces nuisance activations without abandoning the corridor.
Masking as design, not improvisation
Masking is legitimate design work. The public highway, a boiler flue, an air-conditioning condenser, a floodlight, and moving foliage all produce activation events that erode confidence in the system and generate call-outs at three in the morning.
Use the manufacturer's supplied mask or lens clip. Tape and card are common on site and are indefensible in a handover file: they fall off, they weather, and they demonstrate that the design was improvised. Whatever is masked, and why, should be recorded alongside the sensor position.
A survey method that produces a record
- Walk the perimeter and the neighbouring land at the hour when the risk sits — usually after dark in winter — and sketch the approach lines onto an extract of the site layout.
- Mark candidate head positions with temporary brackets rather than drilling first.
- Walk-test each position: approach along the line at normal pace, then crouched, then close to the wall. Note the distance at which the detector registers, using walk-test mode or the panel event log.
- Repeat after dark to confirm lighting does not alter the result.
- Photograph each head square-on, record its height, angle, mask and measured detection distance, and file it with the commissioning documents.
- Re-test after any scaffold alteration. A struck and re-erected lift moves every geometry assumption with it.
- Set review triggers: a new skip, a change of use next door, a new hoarding, a change of neighbouring occupier.
The mistakes worth naming
Heads aimed out over the scaffold edge detect departure. Corner-mounted units aimed along a wall detect the wall. High-level mounting for coverage creates dead ground. Sensors covering only the ladder stile catch the second person through, not the first. Heads raking across the public footpath trade detection for nuisance. And no record at all means the system cannot be shown to have been designed at all.
Re-aiming a detector takes an hour and costs nothing in hardware. On most sites it is the cheapest improvement available — and the one least likely to be requested.