There have been many elevator upgrade projects with the aim of saving energy and time for the managers and occupants of buildings. I suspect this has to do with the increase in height and volume of people moving in and out of buildings, and one of the trends I have noticed is the system put in place for people to select the floor they want to go to before entering the elevator (so no up or down buttons any more) and being allocated a specific elevator number to enter.

Theoretically the system would be able to calculate the most efficient way to dispatch the elevators if it knows all the floors that people want to exit the elevator. Practically I suspect this depends on the degree which the path of the elevator (i.e. which floors it opens at) can be disrupted with additional requests as it traverses the floors.

I am curious as to whether there have been studies of elevator dispatch algorithms combined with human behaviour to optimize such systems, as I see the factors at play include:

  • The distance between the floor selection control and the elevator door
  • The number of floors in the building
  • The number of elevators in the building
  • The distribution of the occupants in the building that have a need to go to other floors (other than the shared floors like Ground and car park)
  • The amount of disruption allowed to the path of the elevator

So to me this seems like a very complex problem to which a single solution seems to have been proposed that will apparently cater for all these different types of scenarios.

How does the system take into account that this is not the usual behaviour for people used to pressing the call button and selection the floor once they are inside the lift? Are there specific design strategies that makes it more 'user-friendly'?

Additionally, is there also a higher cost for people who change their mind or get off at the wrong floor and have to try and get back to the correct floor?

But I guess most importantly, as with most designs, people also adjust their behaviour when they are exposed to a new system, so that they end up not necessarily using it the way it is intended.

UPDATE: I tried to do a little bit more research, and apparently I should be looking at Destination Dispatch Systems. There are some companies that explain the systems in more details available:

  • 2
    Not an answer but if you want to experiment at writing your own algorithm there's a great Javascript game: play.elevatorsaga.com Commented May 22, 2016 at 1:17
  • Interesting topic. I've also noticed what @AmeenAkbar refers to in his answer (i.e. elevators only servicing certain floors) although not always in an odd/even divide. Can I ask though, what exactly are you asking? I was going to edit your question to bold I am curious as to whether there have been studies of elevator dispatch algorithms combined with human behaviour to optimize such systems but then wasn't sure if that was too restrictive, especially considering the questions in your last two paragraphs?
    – Monomeeth
    Commented May 24, 2016 at 1:02
  • @Monomeeth I posted a few questions that got put on hold for being either too broad or too restrictive, so I am just hedging my bets here :p
    – Michael Lai
    Commented May 24, 2016 at 1:53
  • 1
    An interesting algorithm and summary of considerations in a related post: ux.stackexchange.com/a/36508/44275
    – Jason C
    Commented Feb 24, 2017 at 6:29
  • Btw I have seen, in practice, a number of systems in high rise buildings (here in NYC) where the call panel contains a keypad and you enter the floor you are going to as part of the call. It will then make the optimal choice based on current pending calls and elevator positions and tell you which elevator to get in to. When idle the elevators wait near learned high traffic floors for that time of day. See also this company.
    – Jason C
    Commented Feb 24, 2017 at 6:38

2 Answers 2


As a starting point for you I can help answer the following part of your question: I am curious as to whether there have been studies of elevator dispatch algorithms combined with human behaviour to optimize such systems

According to Cortés, Larrañeta and Onieva (of Seville University), researching elevator systems is only a fairly recent phenomenon:

The elevator system research is quite recent and has followed the technology development. The late eighties and the nineties decade can be considered as the start point of the industrial investigation, especially in USA and Japan. After that the research experimented the impulse of the largest multinational companies. By the end of the nineties the research in vertical transportation was a reality and the collaborations among the private companies and the research centres were reinforced, some examples are the Systems Analysis Laboratory in the Helsinki University of Technology with the KONE Corporation, the Konrad-Zuse-Zentrum für Informationstechnik of Berlin or the Seville University with MAC PUAR, S.A.

Above quoted from the paper Genetic algorithm for controllers in elevator groups: Analysis and simulation during lunch peak traffic.

I used to have a neighbour some years ago who worked in the elevator industry. My understanding is that around the 2010 mark there was a renewed focus in terms of researching algorithms for better time management, driven primarily by the need for so-called green buildings (ie. buildings that were much more environmentally friendly). However, any changes to elevator algorithms to make them more time efficient, regardless of the reason, is still going to benefit users.

Below are some links to various research that may be useful to anyone interested in this topic:

There is also a well respected book on this topic readily available in print or as an eBook, aptly titled Elevator Traffic Handbook: Theory and Practice. I haven't read this book, but you can view the Contents to determine how much research may have taken on board human behaviour to optimise these systems.

While this may not answer your question entirely, I hope it helps you hedge your bets, so to speak. :)

  • +1 It has definitely provided some very useful references, and the keyword search term of Destination Dispatch. Interestingly it has mostly been explored from the engineering and physics perspective though. However, most of the papers and studies are still only looking at the traditional hybrid or full dispatch systems.
    – Michael Lai
    Commented May 24, 2016 at 2:44
  • You may also find this blog post interesting, along with the first comment from San. And yes, often it's about knowing what search term we should be using!
    – Monomeeth
    Commented May 24, 2016 at 2:53
  • Please note that the link: "Strategies for an elevator dispatch system" is broken. A working link is at: citeseerx.ist.psu.edu/viewdoc/…
    – Rob
    Commented May 29, 2019 at 2:21
  • Thanks Rob! I've edited the answer to replace the broken URL with the one you provided. :)
    – Monomeeth
    Commented May 29, 2019 at 23:56

In some elevator lobbies, I've noticed where 1 lift will only service odd number floor and one will only handle the even numbers for instance. i believe this would optimize time taken and efficiency in the elevators.

Another way to approach this is by numbers. have some kind of tracking set-up to record the frequency of which floors were serviced per elevator. then come up with some methods of optimization based on these numbers.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.