A chemical plant protection game incorporating bounded rational attackers and distribution-free uncertainties. (September 2021)

• Record Type: Journal Article
• Authors: Zhang, Laobing; Reniers, Genserik; Chen, Bin; Qiu, Xiaogang
• Is Part Of: Journal of loss prevention in the process industries. Volume 72(2021)
• Publication Date: 2021-09
• Subjects: Chemical plant protection -- Game theory -- Bounded rationality -- Distribution-free uncertainty; Chemical industries -- Safety measures -- Periodicals; 660.2804
• DOI: 10.1016/j.jlp.2021.104545
• Languages: English
• ISSNs: 0950-4230

Abstract

Abstract: The chemical industry has an important role in our modern society. Due to the existence of hazardous materials and possible extreme producing conditions, chemical facilities are also considered dangerous. Research has pointed out that a successful attack on chemical plants may cause mass casualties, in the United States. Game theory has been employed to improve the protection of chemical plants, and current literature on chemical plant protection games assume a 'rational' attacker. The present paper studies a game-theoretic model, which is played by a rational defender and a 'bounded rational' attacker, for improving chemical plant protection. The attacker modeled in this paper is assumed to play higher payoff strategies with higher probabilities, which is innovative from the current chemical security literature. Attackers in the current chemical plant protection games would always play the strategy with the highest payoff (probability of 100%). Distribution-free uncertainties on attacker's parameters are also integrated into the model. An algorithm for solving the game presented in this paper is also proposed. A case study reveals that although a bounded rational attacker would reduce the defender's expected payoff, the defender's equilibrium strategy from the present model is robust to different attacker behaviors. Highlights: This is the first game theoretic paper modelling bounded rational attackers in the chemical security domain. The attacker's behavior is based on a very relaxed assumption, that is, he will play higher payoff strategies with higher probabilities. With this paper, the CPP game is now able to deal with bounded rational players as well as payoff uncertainties. case study results show the solution is robust.