The Hidden Science Behind Elevators
https://www.popularmechanics.com/technology/infrastructure/a20986/the-hidden-science-of-elevators/
Initially, elevators were manually controlled by human operators, who stopped wherever they would see a person. By the 1950s, electric switches took over and engineers had to pre-define rules for elevators. They established a schedule with certain starting floors and ending floors at specific timings – similar to a bus schedule. However, this was extremely inefficient. By 1965, elevators began utilizing an algorithm called “collective control.” The basic idea was that the elevator would continue to move in one direction until it exhausted all of the requests in its current direction. This algorithm caused several problems in larger buildings as the waits at the very bottom and very top were nightmares. To handle larger set-ups, engineers developed several tricks: they had elevators communicate to one another or they had some elevators assigned to specific floors depending on traffic prediction. With the emergence of more and more strategies, engineers faced a new problem: choosing the most optimal algorithm. One of the most successful approaches has been letting the computer decide. Using machine learning techniques, engineers can specify what the optimal scenarios look like, and let the elevator controller experiment on its own in simulation. With these more sophisticated policies, even the people who built the software often don’t know why it’s doing what it’s doing!
From the graphical perspective, the journey of an elevator can be thought about as a directed graph, where nodes could represent all of the floors in a building and a directed edge from floor A to floor B could represent the elevator having the ability to go from floor A directly to floor B. This network of nodes must be connected and consist of a single component; if it’s not, then at least one floor is unreachable from the elevator, which defeats the purpose of the elevator itself. In the 1950s, elevator engineers had to spell out rules for when elevators should go where; some elevators did not have the capability to go from the 1st floor directly to the 10th floor; rather, the elevator had to go from the 1st floor to the 5th floor to the 10th floor. Clearly, the network was connected but not complete, as it was preset to start at certain floors and end at certain floors – not every node (floor) was reachable by every other node (floor). However, as technology advanced, elevators undeniably stopped being preset to specific routes and gained the ability to go from any floor to any other floor. The directed graph of a modern elevator’s journey is connected as well as complete, consisting of 2 directed edges (facing in opposite directions) going from every node to every other node. Particularly, the graph contains 2*(f choose 2) directed edges. All in all, the advancement in elevator technology through time can be clearly analyzed from the two different graphical representations. The change in the graph going from incomplete to complete demonstrates the growth in efficiency and technology, as all floors are reachable from all other floors, making people overall wait less.
Interesting article, and unique analysis.