Print Email Facebook Twitter A ship egress analysis method using spectral Markov decision processes Title A ship egress analysis method using spectral Markov decision processes Author Kana, A.A. (TU Delft Ship Design, Production and Operations) Singer, D.J. (University of Michigan) Contributor Nielsen, U.D. (editor) Jensen et al, J.J. (editor) Date 2016 Abstract This paper introduces a means of performing a ship egress analysis by applying eigenvalue analysis to the ship-centric Markov decision process (SC-MDP) framework. This method focuses on how people egress, the decisions they make under uncertainty, and the interaction between the individuals and the layout of the vessel. The objective is to understand the implications of uncertain decision making of people on general arrangement design. One metric is introduced defined as the ratio between the largest eigenvalue and the second largest. Thisdecision metric is used to identify and quantify changes in decisions, as well as to help identify system attributes driving those changes in decisions. A case study is presented showing the utility of this method on a ship egress problem. Sensitivity studies are performed examining the affect of uncertainty and rewards on individuals’ decision making behavior Subject Ship designdecision makingegress analysisMarkov decision processeigenvalue analysis To reference this document use: http://resolver.tudelft.nl/uuid:a64363f0-863d-4589-b418-0a9d043640c7 Publisher PRADS Organising Committee, Copenhagen, Denmark ISBN 978-87-7475-473-2 Source Proceedings of the 13th International Symposium on Practical Design of Ships and Other Floating Structures (PRADS2016) Event International Symposium on Practical Design of Ships and Other Floating Structures, PRADS'2016, 2016-09-04 → 2016-09-08, Copenhagen, Denmark Part of collection Institutional Repository Document type conference paper Rights © 2016 A.A. Kana, D.J. Singer Files PDF ID047.pdf 722.77 KB Close viewer /islandora/object/uuid:a64363f0-863d-4589-b418-0a9d043640c7/datastream/OBJ/view