In this thesis, the difference between DC and AC modeling of network in determining key equipment and calculating the amount of damages caused by key equipments outage is investigated. For this purpose, the conventional attacker-defender (AD) model based on DC modeling of network is represent. Then a new model for determining the key equipments based on AC modeling of network via conic optimization technique is proposed. Then, the two mentioned models are implemented for the test network and compared in terms of key equipments selection, occurred damages, and computational performance. Due to considering voltage and reactive power in the AC model, the numerical results show better accuracy of the AC model for determining key equipments and damages. In contrast, as the AC model is a second order model, the computational performance of the AC model is weaker than that of the DC model. Operating condition of the power system before occurring an event has a significant impact on amount of damages. Considering a safety margin in the operating conditions of a power system, helps the operator to reduce the damages caused by an event. In this thesis, firstly, the worst (critical) operating condition is defined and then a new model is proposed to determine the worst operating condition. From the attacker's point of view, the worst operating condition is the best opportunity to attack the power system. Proposed model is based on the bi-level attacker-defender (AD) model. In the proposed model, the operation variables, including the power output of generators, flow of transmission lines and planned outage for power plants and transmission equipments, are determined before events, while the load shedding amount due to the attack is maximized. This model implemented on test networks to evaluate it performance. The Numerical results show the proposed model is capable to show the dependency of the operation conditions and the attack damages and also determining the optimal attack plan in according to the worst operating condition. Keywords: Power system vulnerability, Deliberate attack, critical equipment, worst operation condition, bi-level programing, conic optimization