How skyscraper elevators work

Design the journey, not just the speed

A passenger experiences waiting, boarding, motion, intermediate stops and the walk or transfer after leaving the car. A specification quoting only metres per second describes one part of that journey. Traffic analysis evaluates the group of elevators and the demand placed on it, rather than treating each car as an isolated machine. [1]

Office arrival peaks, lunch traffic, hotel guests and observation-deck visitors produce different patterns. Peters Research distinguishes simplified up-peak calculations from dynamic simulation of more varied traffic. The occupancy assumptions, floor populations, door times and loading behaviour deserve the same attention as rated speed. [1]

Traction, counterweights and energy flow

In a traction elevator, the car and counterweight are linked through the hoisting system. The machine controls motion while the counterweight offsets part of the moving load. The energy required depends on the balance between the two sides, the direction of movement and losses in the system. [2]

A regenerative drive can return electrical energy when the mechanical system is driving the machine, for example when a sufficiently heavy car descends or a relatively light car rises. In other conditions the drive consumes energy. Regeneration recovers part of an energy flow; it does not make the elevator a net energy source. Otis illustrates these operating modes in its Gen2 documentation. [2]

Long travel makes rope mass important

As the suspended length grows, the hoisting system’s own mass becomes a larger part of the moving system. KONE’s UltraRope replaces conventional steel hoisting rope construction with a carbon-fibre core and a high-friction coating. The manufacturer presents the reduced mass as a way to extend practical travel and reduce energy demand. [3]

The engineering tradeoff extends beyond weight. A hoisting system needs appropriate traction, fatigue performance, inspection methods and behaviour under building sway. KONE’s claims concern a specific product and installation conditions; they should not be turned into a universal percentage saving for every tower. Long travel is also not the same thing as the building’s architectural height. [3]

Zoning and sky lobbies protect floor area

Editorial system explanation: if every elevator served every floor from the ground, its shaft would occupy space through levels where it may provide little local service. Zoning allocates cars to groups of floors. An express shuttle can move passengers to a sky lobby, where local elevators serve a higher zone. The plan trades some direct journeys for a more manageable distribution of shafts.

This is a traffic-planning problem with architectural consequences. Transfer space, circulation, queue storage and redundancy need to be considered with the lift arrangement. Peters’ general traffic model explicitly addresses passengers using different and overlapping zones. The schematic below is explanatory and does not reproduce the elevator plan of any featured tower. [4]

Destination dispatch changes the optimisation

Destination control asks for a passenger’s floor before boarding, allowing the controller to group people with similar destinations. Schindler’s PORT combines this with access control and options such as additional approach or door time. The interface is part of the transport system because it affects how passengers register and find their assigned car. [5]

Peters Research explains that destination control may sometimes make a passenger wait longer for a particular car while reducing the overall time to destination. A claim about shorter trips therefore should not be read as a guarantee of shorter waits in every traffic condition. Compare both metrics, together with crowding and the effect of passengers not registering their trips as assumed. [6]

Handling capacity and waiting time measure different things

Handling capacity estimates how many people a lift group can transport in a specified period. Interval describes the spacing between successive departures in a simplified repeating cycle. Neither metric is identical to a measured passenger waiting time. Traffic models must account for demand and the assumptions about loading and stops. [4]

Illustrative arithmetic: suppose four cars each complete a 120-second round trip and carry an average of 16 passengers per trip. Uniformly spaced departures give an interval of 120 ÷ 4 = 30 seconds. In five minutes, the group transports 300 × 4 × 16 ÷ 120 = 160 passengers. These are assumed inputs for a simple repeating cycle, not a prediction for a named building.

If demand exceeds that service rate, a queue can grow even when cars achieve their specified speed. Real arrivals are irregular, and a car may leave less full or make more stops than the assumed cycle. The example demonstrates why a speed record cannot substitute for a demand model.

Acceleration limits the benefit of extra speed

A car starts and ends a journey at rest. Over a short distance it may never reach its speed limit. For a simplified run with constant acceleration a and equal deceleration, reaching a speed v and braking again requires distance v² ÷ a. If the journey is longer, motion time is d ÷ v + v ÷ a; otherwise it is 2 × √(d ÷ a). These equations follow from elementary constant-acceleration kinematics.

Try the model below. At the default 300-metre distance and 1 m/s² acceleration, raising the speed limit from 10 to 20 m/s reduces idealised motion time from 40.0 to about 34.6 seconds, because the car cannot reach 20 m/s on that run. Real drives smooth changes in acceleration, called jerk, and also allow for levelling and door operation. This model excludes those effects and does not predict lift traffic.

Ride quality is a structural and aerodynamic problem

Mitsubishi Electric describes active roller guides that sense car vibration and apply counteracting electromagnetic forces. Guide-rail irregularities and aerodynamic effects can become noticeable at high speed. The firm identifies this system in Shanghai Tower’s ultra-high-speed elevators. Smooth travel is therefore an active control and mechanical-guidance problem as well as a speed specification. [7]

Editorial interpretation: the elevator and building teams must agree on movement and alignment assumptions. Rails are attached to a building that moves, while a passenger expects a quiet, stable cabin. Noise, pressure variation, vibration and acceleration should be evaluated separately. A single comfort adjective in a brochure cannot communicate the tested operating envelope.

Read speed records with their dates and conditions

Mitsubishi Electric’s May 2016 announcement described technology reaching 20.5 m/s for an elevator in Shanghai Tower. Hitachi’s September 2019 announcement subsequently described a Guangzhou CTF elevator with a rated upward speed of 21 m/s and downward speed of 10 m/s. These dated manufacturer accounts should not be merged into an undated claim that both are currently the fastest. [8] [9]

Hitachi reported approximately 42 seconds over a 440-metre journey to the hotel lobby, while listing the travel distance as 440.4 metres in its technical table. That journey time is not obtained by simply dividing distance by the maximum speed. The same release identifies one record-setting unit; it does not say that every elevator in the tower runs at that speed. [9]

Pressure changes explain different up and down speeds

Hitachi identifies air-pressure change as a distinct comfort issue on the 440-metre Guangzhou CTF run. Its development account describes cabin pressure control and a lower descent speed, alongside streamlined car geometry and active guidance. These measures address different mechanisms: pressure discomfort, aerodynamic noise and vibration. [10]

The manufacturer tested pressure-change patterns with people in a pressure simulator. That is a useful reminder that a mechanical system can meet a speed target while still needing evaluation against human response. Faster travel is not automatically a better passenger experience. [10]

Two decks change the building interface

Otis offers double-deck and adjustable-deck high-rise arrangements. Two passenger compartments share a moving assembly, so adjacent floors can be served together. Its adjustable system changes the separation between cars to accommodate different floor heights. The supplier’s reported floor-area benefits depend on the comparison and configuration. [11]

Engineering interpretation: double-deck arrangements need coordinated boarding levels, wayfinding and compatible demand at the paired floors. They can move more people per shaft, but the full journey still includes getting to the correct deck and any transfer. The extra capacity is valuable only when the architecture and passenger flows support it.

Installation, commissioning and availability

Hitachi’s development interview describes the need to plan beyond the limits of a test tower and complete evaluation at the actual project. A high-speed installation joins factory-developed machinery with project-specific shafts, controls and installation conditions. Commissioning is where those interfaces must perform as one system. [12]

Otis’ SkyBuild concept uses an internal construction elevator that climbs with the project and later converts into a permanent SkyRise installation. This illustrates another interface: vertical transport is required to build the tower before it serves the finished building. [11]

Editorial assessment: an operational review should ask how a group performs with a car unavailable, how maintenance access is arranged and how worn or obsolete components can be replaced. Peak performance with every car available is only one scenario. Planned outages and unexpected faults affect the usable capacity tenants experience.

Stopping safely is a separate engineering problem

Hitachi describes an electronic emergency terminal speed-limiting device that monitors car position and speed near the ends of the shaft. Its high-speed system also required governor development and mechanical safety systems suited to the different upward and downward speeds. Monitoring, normal motion control and emergency stopping are separate functions. [13]

Physics explains why the stopping problem becomes more demanding: kinetic energy is proportional to mass times speed squared. For equal moving mass, doubling speed means four times the kinetic energy. This is a general relationship, not a calculation of the heat or force in a particular elevator brake. Actual braking design must consider the complete moving system, deceleration limits and the applicable safety requirements.

Emergency operation needs its own design

NIST’s 2014 review examined elevator evacuation arrangements in seven building complexes, with particular attention to people with mobility impairments. Its findings concern emergency procedures and the technical challenges of integrating elevators into evacuation. It is a historical research reference, not a substitute for the standards currently adopted in a project’s jurisdiction. [14]

Passenger service, firefighter access and occupant evacuation are distinct operating cases. A designer must establish which lifts are intended for each role and what protection, power, controls and procedures support that role. An ordinary passenger elevator is not automatically an evacuation elevator. Occupants should follow the building’s emergency instructions rather than infer a safe procedure from a general technical article. [14]

What kilometre-scale towers still have to solve

KONE’s October 2025 Jeddah Tower announcement describes UltraRope and a vertical journey exceeding 630 metres. This concerns a planned system in a tower under construction, not a completed performance record. The travel height also remains substantially different from the tower’s planned architectural height above one kilometre. [15]

Editorial assessment: future towers need to reconcile long travel with multiple populations, transfer capacity, emergency planning and maintainability. Stronger or lighter ropes and faster machines address important components of the problem. The final measure is whether the building can move its occupants reliably through ordinary days, peak periods and degraded operating conditions.