Abstract: To solve the problem of low resource recovery efficiency in the deep mining area of Jinch⁃ uan No.2 Mining area，numerical simulation was used to conduct research on the optimization of deep min⁃ ing drift parameters. Firstly，6 m × 4 m large section drift stope model and 5 m × 4 m conventional section drift stope model were established. Secondly，FLAC3D software was used to simulate the mining method of cemented backfilling with downward layered horizontal drift，and the stress values and deformation laws of large section and conventional section drift during the mining process were analyzed and studied. The re⁃ sults show that during the mining process，the maximum subsidence of the conventional section and en⁃ larged cross-section drift roof is 12.55 and 17.2 mm，respectively. And the maximum convergence displace⁃ ment values of the two sides are 12.28 and 21.56 mm，respectively. Compared to others，the displacement values of roof and two sides have increased after enlarging the cross-section，but the increase is not signifi⁃ cant，indicating that the rock movement in the surrounding area caused by the excavation of the two types of cross-section sizes is extremely close. Besides，the maximum compressive stresses in the vertical direc⁃ tion of the conventional section and enlarged cross-section drift roof are 0.389 1 and 0.457 8 MPa，respec⁃ tively. The maximum tensile stresses in the horizontal plane perpendicular to the direction of the drift are 0.254 2 and 0.297 8 MPa，respectively，which are lower than the ultimate tensile stress of the top plate of the backfill by 0.53 MPa. It can be seen from this that after enlarging the cross-section of the downward drift，the maximum tensile stress borne by the roof backfill increased by 17.2%，indicating that the width of
the drift has a significant impact on the maximum tensile stress value of its false roof backfill，expanding the
drift-sectional width of the panel to 6 m is reasonable and feasible.

Key words: downward drift, deep stope, section optimization, numerical simulation, roof stability