Pressure-Driven Modeling of Cyber-Physical Attacks on Water Distribution Systems

Abstract

Water distribution systems are becoming increasingly vulnerable to cyber-physical attacks as they are further augmented with otherwise helpful monitoring and control devices. Simulating the hydraulic responses of networks to attacks is an important first step toward understanding and mitigating their potential impacts. To date, the tools available for this have either been small-scale, bespoke models or have relied on demand-driven analysis. In this work, we improve the suite of modeling tools currently available by extending the capabilities of epanetCPA—a Matlab toolbox that uses the demand-driven hydraulic engine of EPANET—to simulate pressure-driven simulations of cyber-physical attacks on water distribution systems. The proposed modeling approach is scalable and, by adding pressure-driven capabilities, vastly increases the range of possible failure scenarios that can be reliably simulated. The approach was tested and verified on a town-scale benchmark network through multiple attack scenarios. These showed that the approach can be used to gain helpful insights into the behavior of networks, including the relative vulnerability of different sections of networks, the suddenness with which supply is fully cut off, and the window of opportunity for responding to attacks. Such insights can then be used to improve network resiliency and aid in the detection of attacks.