Cyber-attacks on a cyber-physical power system could lead to significant data failure, false data injection and cascading failure of physical power system components. This paper proposes an advanced approach based on a ternary Markovian model of cyber-physical components interactions to capture the subsystem layers' interactions of the cyber-physical power system and to quantify the interdependency impacts on physical power system security. The approach models cyber-physical interactive operation based on interactions and characteristics of three subsystem layers of the system with the presence of random and unforeseen contingencies, load demand variations and then quantify the impacts with Monte Carlo simulation. The viability of the approach is investigated by simulating a set of scenarios, representing realistic physical power system operating conditions with the cyber network interactions. Findings justify the presence of cyber-attacks in a cyber-physical power system components operation could lead to severe insecurities. However, the impacts on physical power system security does not always correlate with the severity of cyber-attacks.