Columbia’s International Affairs Building has fifteen floors and four elevators which ave what seem to me to be really crappy software. While you’re waiting for the elevators on the 4th floor (which happens to be street level; the campus is on a hill), there are readouts showing where each elevator is currently located and whether it is going up or down. Sometimes there will be several elevators coming down at once, and then the one that’s closest will turn around at 5, leaving us waiting. When an elevator finally comes, everyone has to cram in. Other times, the elevators seem to be chasing each other around and are never where you want them to be. (I’m part of the problem myself, taking the elevator just three floors from 4 to the political science department on 7.)
But maybe the elevators are programmed the best they can be, given the pattern of demand. I don’t know.
What I do think would be cool would be to use these ‘vators as an engineering class project: the students could first get some information on the technical specifications of the elevators and their current software, then they could gather some data on the customers (here, I’m thinking of at least two surveys: first, simply going by the different floors at randomly-sampled times and counting how many people are waiting, for how long, and where they’re going to; second, a survey asking people if they’re satisfied with the elevator service and, if not, what bothers them), then they could create a computer simulation and play with various algorithms, and ultimately they could reprogram the elevators and perform an evaluation (comparing customer satisfaction, waiting time, etc., before and after).
Is this the sort of thing they do in the industrial engineering and operations research department? It seems like a group of students could learn a lot from this.
In the situation that you describe — an elevator approaching from above, but reversing course just before reaching you — perhaps the elevator discharged its last passenger at floor 5, and "intended" to continue to your floor, but someone got on and pushed the button for a higher floor. At that point the vator has no choice (under normal conventions of elevator behavior) but to go up again.
A couple of times in the past year I have been in buildings with multiple elevators, where the elevators don't have floor buttons inside; instead, you select your floor while you are _outside_ the elevator. The system then tells you which elevator to board. I think this is an excellent approach, allowing much higher efficiency…although until this becomes common, it will be confusing for many riders. (It certainly was for me, the first time).
actually… there are systems now where you type your destination on a keypad when you arrive at the elevator. every floor has a keypad. the software can then instantly optimize where to send the elevators. very slick… and it works.
Yes, now that you mention it, I've seen that keypad thing. I guess that changes everything: for one thing, once it's implemented, it can collect data on when people go from point A to point B. I still like my idea of the class project, though.
If the building has 15 stories, and you are typically going from the 4th to the 7th story, you will typically find the first elevator goes the wrong way.
This is the famous elevator paradox.
http://mathworld.wolfram.com/ElevatorParadox.html
or
http://en.wikipedia.org/wiki/Elevator_paradox
Suggestion: if two of you are waiting to go from 4 to 7, point out to your colleague that the first elevator is more likely to be in the other direction. You will likely be greeted with odd looks, maybe disbelief.
Andrew inquires, "Is this the sort of thing they do in the industrial engineering and operations research department? "
I don't know about IE, but it is definitely something that OR students do, usually in a simulation course.
Bruce
Professor
Department of Decision Sciences
Drexel University
Well, the next generation of lifts software will have mood sensors!(True)