•A high level of complexity and automation has brought new challenges to risk management.•An extended FRAM method is applied to find necessary safety barriers in the system.•FRAM can give a quick way to check the adequacy of safety barriers.•Extended FRAM is applied to assess safety based on the performance of the barriers.•Comparison is made among FRAM, FRAM-STPA, Bow-tie, and Bayesian network. Safety barriers are used in the system to prevent unwanted events and accidents. Traditional approaches like fault tree or bow-tie method use linear accident models without considering complex interactions of failures of safety barriers. The present paper presents an extended FRAM model to identify required safety barriers and proposes a safety analysis method to predict the system's safety. The initial step of the method is to identify the necessary main and auxiliary functions to achieve the system goal. The later step is to determine the necessary safety functions to execute the main functions to achieve the system goal and to resist variability in performing the main and related auxiliary functions. A simple mathematical model is proposed to assess system safety based on the performance of existing barriers. The method is described with the help of a case study, the LNG ship-to-ship transfer process. The paper compares the extended FRAM method with other methods such as Bow-tie, FRAM-STPA, and Bayesian network. Analysis shows that FRAM can qualitatively, quantitatively, and dynamically assess system safety. The most vital point of FRAM lies in its capability of effective qualitative evaluation, which considers coupling between functions and related aspects, can be presented graphically, and future actions can be taken accordingly.