Elevator Scheduling

1
Elevator Scheduling

• Ingrid Cheh
• Xuxu Liu
• 05/05/09

2
Elevator Scheduling Problem

• Elevator as a control system
• Response time and behavior depends on programmed
  algorithm(s)
• Different solution depending on building type and
  number of elevators working together
• In algorithm, assignment of job
• External elevator request
• Internal floor request

3
Our Problem Context

• Setting
• 15 floor school administration building
• 2 elevators
• Goal
• Analysis of possible elevator scheduling
  algorithms through simulation
• Find most optimal in given problem
• Suggest future directions

4
Tools

• Basic graphical user interface software
• Programmed in Pascal in Delphi programming
  environment
• Strategies, labeled A, B, C D

5
User Interface Software
6
Strategy A

• Elevators calculate most requested floor
• Each elevator heads to most requested floor,
  unless
• Other elevator already heading there
• Other elevator contains passenger who want to go
  there
• If no passengers in elevator, elevator will
  position itself on floor 8

Reacting to modal distribution of requests
7
Strategy B

• If no passenger in elevator, it waits and goes to
  the floor where earliest request is made
• Otherwise, elevator will head to floors requested
  in order of passenger entry into elevator
• If elevator passes a floor where current
  passenger has requested to get off, elevator
  stops and picks up new passengers on direction of
  travel

External requests in FIFO manner
8
Strategy C

• Both elevators do the same
• If elevator is empty, it heads towards the
  earliest external request
• Otherwise, elevator would head towards desired
  destination of occupants
• When elevator opens, it picks up passengers in
  either direction

Internal before external requests in FIFO manner
9
Strategy D

• Elevator A
• Initially goes to floor 15 and descends one floor
  at a time picking up passengers who are going
  down
• Heads straight back up to floor 15 when reaches
  floor 1
• Elevator B
• Initially goes to floor 1 and ascends one floor
  at a time picking up passengers who are going up
• Heads straight back down to floor 1 when reaches
  floor 15

Round Robin scheduling method
10
Performance Metrics

• w time request submission to final destination
  arrival
• c total passengers carried
• s total passengers who gave up elevator wait and
  used stairs
• p c/(cs)
• e energy efficiency level in elevator operation
• chart running profile of customers served in
  different time brackets

11
Inputs to Simulation

• Available Inputs
• Scenario
• Particular pattern of passengers waiting and
  choice of floor
• Peak Hours
• Choice of more or less passenger traffic
• Simulation Length
• Total simulated of each simulation run
• Boredom Level
• How much time before switching to stair use

• Our Selection of Inputs
• Scenario
• 30 different scenarios
• Peak Hours
• Peak chosen
• Simulation Length
• 5 minutes
• Boredom level
• Level of 20

12
Results
73.70
38.90
41.07
34.77
84.27
75.93
79.93
61.40
34.57
29.70
27.60
24.67
19.27
19.73
20.33
70.57
12.07
61.20
57.53
57.27
13
Analysis of Results

• Strategy A is the best strategy
• Least average waiting time
• Least number of passengers using stairs
• Maximum passengers carried
• Greatest percentage of passengers satisfaction
• Ranking of strategies from best to worst
• A, C, D, B

14
Limitations of Strategies

• Strategy A
• Efficiency may be related to its position on 8th
  floor
• Overlooks the potential heavy skew of requests
  around the lower floors
• Difficult for Elevators A and B both to calculate
  the most requested floor
• Interlacement of strategies B C necessary
• Smart strategies are less power efficient

15
Extensions within Problem Context

• Investigation into other variables
• Simulation length
• Peak hours to non-peak option
• Boredom level
• Scenario
• Uniform probability distribution is not realistic
• Poisson arrival processes might be more reflected
• More requests assigned to ground level
• Different setting
• Floors
• Elevators

16
Inspiration from Literature

• Passenger behavior (Susi, Sorsa and Sikonen)
• Modeling of diverse traffic flows with passenger
  compositions that incorporate physical and
  behavioral characteristics
• Complex controllers (Bartz-Beielstein, Preuss and
  Markon)
• Fujitec
• Neural network structure to determine control
  strategy
• Zoning Policy (Chu, Lin and Lam)
• Set of floors divided into blocks
• Goal of increasing overall handling capacity

17
Summary

• Insight into elevator scheduling simulation for a
  particular setting
• Compromise between running efficiency and power
  efficiency
• Very simple and limited model
• Extensions possible, as shown by literature
  research

18
Thank You!