Abstract: Dual time-stepping (DTS) with implicit lower-upper symmetric Gauss-Seidel (LU-SGS) algorithm are implemented into a finite volume time domain (FVTD) based solver to improve the computational efficiency and accuracy. The DTS is a 2nd order time stepping method and unconditionally stable, in which the selection of physical time step is only determined by the accuracy requirements. The stability in the time marching process only depends on the pseudo time step, which significantly relaxes the restriction of time step in the explicit schemes. Large Courant Friedrichs Lewyor (CFL) numbers can be used in the forward and backward sweeps, in which small storage is used and inversion of matrices are unnecessary. The typical results of 2D, 3D and complex configurations show that the computational accuracy and efficiency can be ensured by rational combinations of the physical time step, maximum sub-iterations and convergence criterion.