Reducing trawl bycatch rates on non-target species is a hot topic in fisheries science. The sorting grid is widely used in trawl fisheries to reduce bycatch and improve trawl selectivity. The change in hydrodynamic force of the bycatch reduction device influences the behavioral response of fish after contact, determining the device's separation efficiency and selectivity. Therefore, understanding the hydrodynamic performance of the bycatch reduction device is critical for improving the selectivity of trawl gear. This study investigated the effects of grid sizes (Grid-0.3, Grid-0.4 and Grid-0.5) on the mesh opening degree, spatial geometry, and drag of codend under three different grid angles (A-30°, A-45° and A-60°) using flume tank experiment. The results showed that: (1) The average mesh opening angle of codend without grid was 22°±5°. The average mesh opening angle of the codend with Grid-0.4 and Grid-0.5 was 3.3% and 18.8% respectively, greater than that of the codend without grid, while the mesh opening angle of the codend with Grid-0.3 was 9.2% lower than that of the codend without grid. (2) Grid-0.4 has the least influence on the codend shape and was streamlined to accommodate it, while Grid-0.3 and Grid-0.5 distorted the codend. (3) The differences in the drag of the codend with Grid-0.3 at different grid angles were small with an average similar to that of the codend without grid. The average drag of codend with Grid-0.4 and Grid-0.5 at different grid angles was 12.7% and 24.4% greater than that of the codend without grid, respectively. Except for the codend with Grid-0.3 at A-30°, the codend drag increased with increasing grid angle and size. The findings could have a significant impact on the research on bottom trawl selectivity and the protection of demersal fishery resources in China, particularly research on devices that can effectively release bycatch and juvenile fish.