Partial air cushion support catamaran (PACSCAT) is a new type of high-performance ship based on slender catamaran and supplemented by partial air-cushion support. Its bottom is flat and wide, and the air-cushion displacement volume accounts for about 30% of the whole ship. Due to the air cushion, the ship will have many nonlinear characteristics when sailing in the waves, influencing the motion performance of the hull heavily. The three-dimensional potential flow theory considers the three-dimensional effect of the flow field and it has been widely used in the field of ship hydrodynamics, so it can be effectively applied to the study of strong nonlinear motion characteristics of ships, which coincides with the study of nonlinear mechanics of hovercraft. The research was based on the rigid air seal form of the cushion system, and took into account the nonlinear characteristics of the air cushion. Therefore, in order to verify the form of numerical calculation method of PACSCAT with rigid seals, and study its movement characteristics in waves, this paper adopted the free surface method, and used overlapping grids and slip grids to analyze its performance in clam water. Furthermore, based on the best grid form, its navigation performance in regular waves was numerically simulated and compared with the experimental values, and then the calculation accuracy of the total hull resistance and the characteristics of the motion parameters under different wavelengths were studied. The results show that: overlapping grids have a higher hydrostatic resistance calculation accuracy than sliding grids, up to 5.2% when v=5.0 m/s. In the numerical calculation, the amplitude and average value of wave resistance are greater than the experimental value, and the error is the largest at the wavelength of 7.0 m. From the study of various wave motion parameters, it can be seen that there is no obvious difference between the duration curve of each motion parameter and the test value, and the calculated value of amplitude response is generally smaller than the experimental value, but the calculated value is obviously greater than the experimental value when λ/L is 2.33.