1: \begin{abstract}  

2:   In [Z. Hu, R. Li, and Z. Qiao. Acceleration for microflow

3:   simulations of high-order moment models by using lower-order model

4:   correction. J. Comput. Phys., 327:225-244, 2016], it has been

5:   successfully demonstrated that using lower-order moment model

6:   correction is a promising idea to accelerate the steady-state

7:   computation of high-order moment models of the Boltzmann

8:   equation. To develop the existing solver, the following aspects are

9:   studied in this paper. First, the finite volume method with linear

10:   reconstruction is employed for high-resolution spatial

11:   discretization so that the degrees of freedom in spatial space

12:   could be reduced remarkably without loss of accuracy. Second, by

13:   introducing an appropriate parameter $\tau$ in the correction step,

14:   it is found that the performance of the solver can be improved

15:   significantly, i.e., more levels would be involved in the solver,

16:   which further accelerates the convergence of the method. Third,

17:   Heun's method is employed as the smoother in each level to enhance

18:   the robustness of the solver. Numerical experiments in microflows

19:   are carried out to demonstrate the efficiency and to investigate the

20:   behavior of the new solver. In addition, several order reduction

21:   strategies for the choice of the order sequence of the solver are

22:   tested, and the strategy

23:   $m_{\sss l-1} = \lceil m_{\sss l} / 2 \rceil$ is found to be most

24:   efficient.

25: 

26: \vspace*{4mm}

27: \noindent {\bf Keywords:} Boltzmann equation; High-order moment model;

28: % Linear reconstruction;

29: Lower-order moment model correction; Multi-level method; Microflow

30: \end{abstract}

31: