A numerical reconstruction is implemented according to a combination of MUSCL(Monotonic Upstream-Centered Scheme for Conservation Laws) and THINC(Tangent of Hyperbola for Interface Capturing) methods. And a new constructed solver is derived by the combination. The main principle of this reconstruction solver is to keep the variation diminishing of every cell boundary to a low level. And a BVD(Boundary Variation Diminishing) principle is set according to the left and right boundary states in MUSCL method, as well as which in THINC methods. To obey the BVD principle, an alternative choice between the MUSCL and THINC is needed here. Thanks to the BVD solver, the discontinuous section becomes smooth and the oscillation is significantly controlled. The numerical performance of BVD reconstruction is then tested by 1D and 2D numerical examples.

In this paper, our main concern is the effects of the mitigation layer in the underwater explosion problem. As the explosive gaseous products, water, mitigation layer, and so on constitute a fluid mixture, the physical states of these components are covered by a uniform Mie-Grüneisen equation of states(EOS). Then the Mie-Grüneisen mixture model is applied to simulate the complicated interaction among gaseous product, water and solid layer. In the calculation, the particular parameters of EOS are used to make a distinction among the physical properties of fluid components and the volume fractions are used to identify the interfaces. In addition, the inlet and outlet boundary conditions are implemented here. During the calculation process, the effect of the mitigation layer is investigated here. According to the investigation, it is found that the effect of the layer is decided by the shock impedance of layer material. The pressure of the shock wave will decrease when it penetrates a new medium with lower impedance. On the other side, the penetrated shock wave in the layer is reflected by the structure and produces a second shock wave. During this process, the affection of layer thickness and distance is also studied here. But both two factors are unimportant under the mitigation mechanism.

We focused on effects of treatment of mixture model as the interface is in a diffused style. In order to seek a convenient way to treat the interface and keep strong adaptability, Mie-Grüneisen mixture model is used. The original Mie-Grüneisen mixture model in terms of volume fraction is modified. In the model, the mass fraction is employed as color function, and the optimization of sound speed is achieved with the help of the density of fluid components. The optimized sound speed reduces time steps under the convergence mechanism and promotes efficiency of the calculation. Meanwhile, the introduction of limiter make the pressure curves stable. Numerical tests show that the modified mass fraction model reduces the number of time steps under different numerical schemes. Effects of this reduction depend on physical conditions of case.

To solve the problem that the forced evolutionary random walk algorithm (RWCE) falls into local optimization and reduces search ability in the later stage of optimization,a strategy of combining periodic dominance structure extraction with search path enhancement is proposed. Firstly,population of the system is preliminarily optimized,and the dominant individuals are extracted in certain period.Then these dominant individuals are replicated by multiple paths to other individuals. Finally,according to the search mechanism,they are spread all over the whole solution domain. It shows that the multi-path search strategy centered on dominant individuals improves accuracy of local optimization,increases diversity of population,enhances global search ability,and improves efficiency and quality of optimization.