Discipline: Mathematics and Statistics
Subcategory:
Xaimarie Hernandez Cruz - University of Puerto Rico Mayaguez
Co-Author(s): Saylisse Dávila and Norbert Franqui, University of Puerto Rico, Mayaguez, PR
Most pedestrian evacuation models have restrictive assumptions that hinder the adequate simulation of the variability in the behaviour of individuals in a flood or tsunami event. By correcting some of these presumptions the evacuation model is expected to better explain the behaviour of individuals in an evacuation scenario. The first assumption that most models have is assigning the same velocity for each individual, not contemplating the effect of factors like age and discapacities. The second assumption is assigning a constant velocity throughout the complete evacuation process, not taking into consideration that fatigue might somehow reduce this speed over time. The last assumption identified is the lack of consideration of the individual’s reactions when they identify the risk signals (e.g. evacuate immediately vs. gather dependants before evacuating). To establish different velocities for individuals, data from a 5k is analyzed and segregated by gender and age. Using parametric and non-parametric confidence intervals and hypothesis tests for the mean and variance of the velocities, age groups that had similar tendencies are grouped together and are assigned the probability distribution that best describes them. With percentiles calculated from the probability distributions, different velocities are calculated to represent slow, intermediate and fast velocities. Considering that each individual will not maintain a constant velocity throughout the evacuation process an estimation of a fatigue factor is investigated with the purpose of simulating the reduction of the velocity after a period of time has elapsed. A similar analysis to the estimation of velocities is implemented, with the exception that the fatigue factor (M) is calculated using the 5k data and an equation that relates time, distance and fatigue established by Peter S. Riegel (1981). For the simulation model to correctly represent the reaction of people, human behavior rules need to be created and implemented into the model. To create the rules data from a questionnaire were examined to identify patrons of evacuation. Results of these analyses include identifying and grouping combinations of age groups and gender that have similar tendencies in velocities and fatigue factor, determining slow, intermediate and fast velocities for each group, and extracting the structural patterns that define how humans would react in the event of an evacuation. Future research includes incorporating the findings of this research in an agent-based pedestrian evacuation model within QGIS.
References: AllSportCentral.com. N.p., n.d. Web. 25 Apr. 2016.R Core Team (2015). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
Funder Acknowledgement(s): Funding was provided by: Sea Grant, NOAA
Faculty Advisor: Saylisse Davila, saylisse.davila@upr.edu
Role: My participation in this research project consists of the processing of the 5k marathon data and the analyses of the confidence intervals, hypothesis tests and the determination of the probability distributions for each group in order to establish the velocities in the pedestrian evacuation model as well as the fatigue factor. I was also involved in the distribution of the questionnaire for the recollection of data necessary to establish the human behavior rules.