This elevator group dispatcher features load weighing hall call bypass, load weighing dispatch (reduction of the headway time at the main floor for loaded cars) and preference for cars to serve hall calls if they have a coincident car call. However, it has some dispatching inefficiencies due to shortcomings in the algorithm when compared to an actual modern microprocessor-based dispatcher. This is interesting because virtually all of the obsolete relay-based dispatchers which KJA has analyzed had mistakes or inefficiencies such as those seen here sometimes, which are not always obvious to owners and tenants. Microprocessor-based dispatchers which KJA has analyzed had no really obvious or significant errors or inefficiencies because far more resources are invested in those than in this simulated demonstration, of course.

This is a simulation of a group of 6 elevators of speed 500 feet per minute and capacity 3500 pounds serving 16 floors.

Press the hall call buttons to create passengers. This is all you can do at present.

The persons are created with random height and girth, ostensibly varying from 4'10" to 6'6" tall with a waist measurement varying from 27" to 49". The weight of the 5'8" clone person with a 38" waist is taken as 180 pounds in order to create passengers weighing between 107 pounds and 269 pounds with a simple calculation. However, person-production is heavily weighted towards producing individuals who are close to the typical clone in height and girth. Load-weighing hall call bypass is set at 50% (1750 pounds) so a car will run by hall calls when its passenger load exceeds this amount. Load-weighing dispatch is set at 60% (2100 pounds) so a car will immediately close its doors and leave the main floor when its passenger load exceeds this amount (normally a car will remain at the main floor for 20 seconds after its door opens).

It will not necessarily reflect closely the operation of any actual similar group because your single P.C. CPU (for most visitors) must perform the duties of all car controllers, the group dispatching controller, the door operators and a small part of the mind of each passenger, as well as showing them on the screen. For this reason and for practical reasons of software production economy, termed "value engineering", many features of this simulation are downgraded in comparison to reality, such as:

The speed control has an infinite "jerk rate" which means that the full acceleration of  3.2 feet per second per second occurs instantly when the car starts, the transfer to deceleration when stopping is instantaneous and the deceleration is constant until stopped at the floor. This would be felt as noticeable "bumps" starting, stopping and especially in the middle of a trip of only one floor. Real speed controls have a smooth curve with various adjustment capabilities.

The digitizing rate of the speed control (and also of all motion of persons) is 0.1 seconds and will be even coarser if your CPU speed is taxed. Real speed controls have a significantly faster response rate than this.

The group dispatching is fairly basic compared to a modern microprocessor elevator dispatcher, although it will outperform most old-fashioned relay dispatchers after we wring a few "bugs" out of it and add dynamic hall call reallocation. Although many old-fashioned relay dispatchers are sophisticated and have features not included in this software simulation, the reason why this would often outperform them is because they are mechanical and thus they often degrade over periods of years, especially if their dispatching circuits have such sophisticated features that the practical realities of economics, available time and mechanic expertise do not permit all of the electrical parts (sometimes many hundreds) to remain in their original working order and interconnectedness.

This simulation features load-weighing hall call bypass, load-weighing dispatch, preference for cars having car calls coincident with hall calls, high call reversal and low call  reversal (of course) and some basic (not really precise) estimated-time-of-arrival calculation.

It does not have any "peak" features, which would position idle cars based on recent traffic patterns and possibly allow hall calls at upper floors to be without service slightly longer in preference to serving the main floor earlier if an "up peak" traffic pattern is detected (heavy passenger loads and numerous car call placements experienced at the main floor).

The speed control of the doors is virtually non-existent. The panels move instantly between only four positions, which include fully-open and closed. The door opening time is set at 1.5 seconds and the door closing time is set at 3.0 seconds, which are fairly reasonable times, but the actual times might be greatly different because of the vagaries of your computer and operating system and the Java language used by Web Browsers, which is inherently rather slow. The control of actual elevator doors includes adjustments for speed and torque at a couple of positions or more, and the positions can also be  adjusted, to give a quiet start and "soft stop" with fast door opening and safe door closing.