مشکل مکانی نقاط مجموعه تصادفی
Abstract
1. Introduction
2. Literature review
3. Modelling approach and problem formulation
4. Solution approach
5. Computational study
6. Conclusion
Appendix. Supplementary materials
References
Abstract
In this paper, a Casualty Collection Points (CCPs) location problem is formulated as a two-stage robust stochastic optimization model in an uncertain environment. In this modelling approach, the network design decisions are integrated with the multi-period response operational decisions where the number of casualties with different levels of injuries coming from the affected areas is uncertain. Furthermore, the transportation capacity for the evacuation of casualties to CCPs and hospitals is also uncertain. To solve this complex problem, a robust sample average approximation method with the feasibility restoration technique is proposed, and its efficiency is examined through a statistical validation procedure. We then evaluate the proposed methodology in the backdrop of a hypothetical case of Bhopal gas tragedy (with the same hazard propagation profile) at the present day. We also report the solution robustness and model robustness of 144 instances of the case-study to show the proficiency of our proposed solution approach. Results analysis reveals that our modelling approach enables the decision makers to design a humanitarian logistic network in which not only the proximity and accessibility to CCPs are improved, but also the number of lives lost is decreased. Moreover, it is shown that the proposed robust stochastic optimization approach converges rapidly and more efficiently. We hope that our methodology will encourage urban city planners to pre-identify CCP locations, and, in the event of a disaster, help them decide on the subset of these CCPs that could be rapidly mobilised for disaster response.
Introduction
Severe weather events and natural disasters have displaced approximately 32 million people globally in 2012 and numbers are projected to continue rising (IPCC, 2014). According to the Centre for Research on the Epidemiology of Disasters, over the past ten years, natural disasters affected almost 1.7 billion people, including 0.7 million killed, and resulted in 1.4 trillion dollars in damages worldwide (Guha-Sapir, Hoyois & Below, 2015). Similarly, manmade disasters have human, environmental and economic consequences. Examples of such disasters include stampede, nuclear or chemical plant explosion, emergencies resulting from incorrect handling/transportation of hazardous materials, water contamination and oil spill. Man-made disasters happen mainly due to accidents, negligence or incompetence. With the global increase in the number and severity of the disasters, researchers from different disciplines are increasingly paying attention to disaster management problems. Alerts and early warning systems are among the tools available to city planners for dealing with emergencies. These inform the population of an impending disaster, e.g., tsunami warning system of the Japanese Meteorological Agency (Tatehata, 1997) and COBRA alerts in the UK (Thunhurst, Ritchie, Friend & Booker, 1992). Although these are useful for the advance warning, it is also essential to have, in place, existing strategies for humanitarian logistic network design that could be initiated after a disaster occurs.