A Study on Human Evacuation Behavior Involving Individuals with Disabilities in a Building Nirdosh Gaire Ziqi Song Keith Christensen Mohammad Sharifi Utah State University Logan, UT 84322, United States Anthony Chen Hong Kong Polytechnic University Kowloon, Hong Kong, China
26
Embed
A Study on Human Evacuation Behavior Involving Individuals ......Pedestrian evacuation process should be planned properly to avoid bad consequences. ! Exit doors at the public facility
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
A Study on Human Evacuation Behavior Involving Individuals with Disabilities in a Building
Nirdosh Gaire Ziqi Song
Keith Christensen Mohammad Sharifi
Utah State University Logan, UT 84322, United States
Anthony Chen Hong Kong Polytechnic
University Kowloon, Hong Kong, China
Pedestrian Evacuation
! Immediate and urgent movement of people away from the threat or actual occurrence of hazard.
! Ranges from small scale evacuation from a building to the large scale evacuation from the district.
! Reasons for evacuation: " Natural disasters " Industrial accidents " Fire " Military attacks, etc.
! Pedestrian evacuation process should be planned properly
to avoid bad consequences.
! Exit doors at the public facility plays a major role in the
evacuation process.
! Many studies found in the literature on evacuation
modeling.
3
Introduction
! Empirical studies on Individuals with Disabilities still
missing in the literature.
! Surprising because they consist of a large portion of the
population (12.6% of total population) in U.S. (Kraus,
2015).
! Evacuation models mainly being developed using Stated
Preference rather than Revealed Preference.
4
Introduction
This study is important because of two main reasons:
1. Individuals with disabilities considered in the evacuation
model.
2. Revealed Preference used for the study instead of Stated
Preference. Real life experiment used for the analysis.
5
Significance
Literature Review
! Pedestrian evacuation behavior in a room with single
or multiple exits have been investigated from
experiment and simulations.
! Exit choice from a room has been studied under
different scenarios considering different parameters.
6
Literature Review
! Studies done on the evacuation behavior of individuals
without disabilities.
! No study found on models based on individuals with
disabilities.
! Heterogeneity in population not been studied in evacuation
models.
7
Studies Done For Exit Choice Behavior During Emergency Evacuation
Author Method of study Factors considered for the exit choice
Relevant findings DE D RI F IWD HP Duives and
Mahmassani (2012)
Multinomial logit model √ √
Group behavior generally found in evacuation scenarios.
Fu et al. (2016) Discrete evacuation model √ √
Phenomenon like arching, clogging and irregular outflow seen during simulation.
Lovreglio et al. (2016)
Mixed logit model √ √ √ √
Density and distance had negative affect in the exit choice, whereas flow and room information had positive affect in exit choice.
Guo and Huang (2008)
Logit based model √ Information of exit has major role in exit choice.
Liu et al. (2009) Simulation √ √ √ Density plays an important role in exit choice. Unfamiliarity with the room features makes difficult to make exit choice.
Nilsson et al. (2008)
Unannounced evacuation experiment
√ Information like green flashlight can have positive influence in the exit choice.
Haghani et al. (2014)
Multinomial logit and mixed logit
models √ √ √
Distance, density and room information had positive affect in the exit choice behavior.
Fang et al. (2010) Experimental study √ √ √
During low density condition around exits, shortest exit chosen. During congestion, farthest exit chosen to avoid time wasting.
8 Note: DE = Distance to Exit; D = Density around exit; RI = Room Information; F = Flow at exit; IWD = Individuals with Disabilities; HP = Heterogeneous Population
Experiment Settings
! Agscience building at USU used as the research setting.
! 4 exit doors at the ground floor which were accessible for all individual types.
! Participants asked to evacuate with maximum comfortable speed after alarm went off.
9
Exit Doors
10
Radio Frequency Identification (RFID) Tracking
! Automatic identification system that consists reader and tags.
! Cost effective, small in size and capable to store more than enough information.
11
Participants wearing RFID tags lanyards
12
Radio Frequency Identification (RFID) Tracking
RFID receiver RFID tags
Experiment Settings
! Microscopic data collected using RFID tracking technology complemented by video tracking methods.
! Video cameras used for video tracking.
13
Camera
RFID signal
Participants
! 47 participants in 16 different evacuation scenarios.
14
Run Location Total IDs Number of IWDs 1 Class 40 7 2 Class 37 6 3 Class 40 8 4 Class 37 7 5 Computer Lab 41 8 6 Computer Lab 42 9 7 Computer Lab 43 10 8 Both 44 11 9 Both 40 9 10 Both 44 11 11 Class 41 7 12 Lecture hall 43 11 13 Lecture hall 31 4 14 Lecture hall 45 11 15 Computer lab 41 9 16 All places 41 11
Individuals with disabilities
13
Total participants
47 Individuals without disabilities
34
Visual disability
12
Wheelchair movement
1
RFID Data
15
! RFID data used for trajectory analysis.
! Data recorded: 2 seconds interval
Exit Doors Identification
16
Discrete Choice Model
! Evaluate alternatives measured by the utility function. ! Let Ui be the utility that determines the discrete outcome i. Where, Ui = True utility (unknown to analyst). Vi = Deterministic component (measurable).
ξ i = Stochastic component (unmeasurable).
17
Ui = Vi + ξ i
Discrete Choice Model
V↓i = ∑𝑖=1↑𝑘▒𝛽↓𝑖 𝑋↓𝑖 k = number of attributes used for the utility function.
𝛽↓𝑖 = parameter that will define the weightage of the attribute.
𝑋↓𝑖 = attribute for the selection.
Pn(i/𝐴↓𝑛 ) = Probability that individual ‘n’ will choose alternative ‘i’ from the choice set 𝐴↓𝑛 = {1,2,…, i, j, …..M}
= Prob ( 𝑈↓𝑖 > 𝑈↓𝑗 )
= Prob ( 𝑉↓𝑖 + ξ↓𝑖 > 𝑉↓𝑗 + ξ↓𝑗 )
18
Choice Probability
! If the error terms are modeled as Gumbel distribution, then we have the well known logit model:
P↓i = exp (V↓i )/ ∑𝑙∈𝐴↑▒exp (V↓l ) ! Binary choice model Probability of selecting a door from 2 doors:
𝐕↓𝐝𝐨𝐨𝐫𝟐 = BETA1 * Dd2 + BETA2 * Ke2 + BETA3 * Nd2 Dd1 & Dd2= distance of the individual’s initial position from the doors (meters). Ke1 & Ke2 = exit density at the two doors. Nd1 & Nd2 = number of individual with disabilities at doors at different time intervals.