Explosion risk analysis (ERA) is known as one of the dedicated safety studies for offshore installations and its purpose is to evaluate the explosion design accidental loads (DALs) on offshore topside structures and facilities. In general, ERA is more likely to be implemented in a probabilistic manner because it has a problem that needs to deal with a large number of explosion scenarios. In the probabilistic ERA, flammable gas cloud frequency distribution is a kind of intermediate result, which can be obtained by integrating the results of frequency analysis and gas dispersion modeling. In general, this distribution is applied to investigate a certain number of appropriate explosion scenarios to evaluate exceedance curves, which are commonly used to determine the DALs. The use of Computational Fluid Dynamics (CFD) to perform the gas dispersion and associated ignition probability modeling has become a trend in recent offshore projects. In most cases, however, the gas cloud frequency distribution has not yet fully benefited from the CFD models due to the high computational cost. Generally, the distribution is derived only using some particular values rather than reflecting the overall results of the CFD simulations. These particular values are likely to be steady-state results calculated using constant leak rates, which reduces the benefits of the CFD simulations and time dependent ignition probability model (TDIIM). In addition, when using these values, the effect of time-varying leak rates has only been considered with some simplified methods and the accuracy of these methods is still controversial. In the current study, the main effort is focused on proposing a new type of gas cloud frequency distribution that can reflect the overall results of the CFD simulations performed with the time-varying leak rates. One aspect to keep in mind is that every instantaneous gas cloud can pose a hazard of explosion, regardless of whether the leak rate varies with time or not. Corresponding to that, a specific method is further proposed which takes into account the entire transient process of gas cloud propagation, i.e. the time history of gas cloud size and ignition probability. This paper also presents case studies in which the gas clouds are both partially and fully selected in CFD simulations, and the final results are investigated in terms of overpressure exceedance curves, The case studies have shown that the transient process of gas cloud propagation is important for obtaining an accurate ERA, and that the proposed distribution dose not results in a significant increase in total computational cost.