Why The Lift Never Comes At Lunchtime

Lifts that behave impeccably at ten in the morning become useless at half past twelve. The building has not changed, and neither has the equipment. The traffic pattern has.

Capacity is sized around a five-minute window

Lift provision is calculated from how many people the system can move in a five-minute period, measured against the building population. That single figure decides how many shafts get built.

Shafts are enormously expensive, because every one of them removes lettable floor area on every storey. Developers therefore buy enough capacity for a normal peak, not a generous one.

The design target is a tolerable wait during the worst five minutes of a typical day. Anything beyond that is money spent on empty air.

Every extra stop costs more than the ride

Travelling between floors is quick. Stopping is not. A stop means decelerating, levelling, opening, holding the doors, waiting for people to shuffle, closing, and accelerating again.

That sequence takes a surprisingly large slice of a round trip, and round-trip time is the number that governs everything. Two additional stops can lengthen a circuit by more than a whole extra floor.

So a lift that seems to be dawdling is usually not slow. It is stopping more often than it did an hour ago, and each stop is expensive.

Midday is the worst pattern there is

Morning traffic is nearly all upward from one entrance, which is easy to schedule. Evening traffic is nearly all downward. Both are predictable and both can be served efficiently.

Lunchtime is two-way, with people leaving, returning and moving between floors at once. A car heading down keeps being interrupted by requests going up, and the reverse.

Interfloor traffic is the pattern that fills every car with stops while carrying few passengers per trip. It is the least efficient thing a lift system can be asked to do.

The dispatcher is optimising a different number

Modern group controllers assign calls across all cars to minimise average waiting time across the building, not the waiting time of whoever happened to press most recently.

That routinely means sending a car past your floor because collecting you would delay four other people. The algorithm is behaving correctly and it feels like an insult.

Destination systems, where you enter your floor in the lobby, exist to reduce exactly this problem by grouping passengers going to the same place before the doors even open.

Mirrors are part of the engineering

Perceived waiting time falls sharply when there is something to look at, which is why lift lobbies acquired mirrors long before anybody used them to check their hair.

Door dwell times are similarly tuned. Shortening them improves throughput and increases the number of people who get clipped by a closing door, so the setting is a compromise.

None of this makes the lift arrive faster. It makes the interval feel shorter, which is considerably cheaper than digging another shaft through twenty floors of building.