Unsteady flow is the dominant flow state in real life; thus, the simulation of it is of vital importance, especially in engineering, for example, the flutter or buffeting of the aerofoil. In the past decades, the progress in computational science greatly paced the development of computational fluid dynamics (CFD), providing powerful tools for simulating unsteady flow via numerical methods. However, the unsteady flow state depends on more variables than a steady flow, including the external conditions in different time moments and the flow's properties that vary with time. The calculation is still too massive, even using CFD. Therefore, CFD algorithms with higher efficiency and less reduction in accuracy are still needed to optimize the technique. This paper reviews the main CFD computational methods that have been maturely developed and proven effective, including direct numerical simulation (DNS), classic turbulence models and reduced order model (ROM), illustrating the main mechanisms and displaying their features. The paper also sheds light on these methods' latest research progress.