Scenario and Sensitivity Analysis for Flooding Vulnerability using Genetic Algorithms

Edward Nataniel Apostol, Joshua Kevin Cruz

Adviser: Vena Pearl Bongolan, PhD

Abstract

We minimize flooding vulnerability for a city in the central plain of Luzon, by ‘rearranging’ the barangays (communities or neighborhoods in the city) via a genetic algorithm. The different components of flooding vulnerability were investigated, and each was given a weight, which allows us to express vulnerability as a weighted sum. This serves as the fitness function for the genetic algorithm. We also allowed non-linear interactions among related but independent components, viz, poverty and mortality rate, and literacy and radio/TV penetration. A scenario analysis was done, comparing the city's current state, and the proposed design, and the initial result was a 12 percent decrease in vulnerability with only a one percent increase in cost.

Finally, we did a sensitivity analysis on the hazard aspects to improve the fitness function, which effectively changed the initial weights, giving the much improved result of a 24% decrease in vulnerability alongside a 14% decrease in cost units.