| Since only the cloud information of each node was needed to calculate the spatial derivatives, the gridless method, better than the traditional numerical methods, had great flexibility and superiority in the flow fields involving complex geometrical structure or moving rigid body. The gridless method for numerical simulation of chemical reacting flow field and its application were studied in this dissertation.Firstly, a weighted-point filling strategy was developed to generate a uniform or non-uniform node distribution of reacting flow field. Some key technologies were investigated at first, such as the weight setting of boundary point, the calculation of weight and position of the ideal advancing point, selection criteria of the best advancing point, etc. The cloud establishment and point filling were carried out simultaneously. The results of several tests showed the strategy could achieve an ideal filling effect, and its robustness was standout, which implied that this strategy can be used in the following numerical simulation of reacting flows.Secondly, the gridless algorithm for numerical simulation of chemical non-equilibrium flow was accomplished based on 2D multi-component Euler equations with chemical sources. The spatial derivatives were approximated by least-square fits in the local cloud. The multi-component HLLC (Harten-Lax-van Leer-Contact) scheme and AUFS (Artificially Upstream Flux Vector Splitting) scheme were introduced to calculate the convective flux. In order to improve the precision, the MUSCL method was used to restructure the flow parameters on both left and right sides of the middle point between the centre point and satellitic point. The Strang splitting method was used to deal with the stiff problem arising from computation of chemical reacting flow. The flow equations were advanced by four-stage Runge-Kutta method, and the chemical ordinary differential equations were solved using the finite rate chemistry model and implicit linearization method. Subsequently, this algorithm was extended to the 3D space. The flow fields of shock-induced combustion, wedge-induced detonation, and cellular detonation in the rectangular tube were simulated, and the results showed its feasibility and accuracy adequately.Thirdly, a rapid local cloud restructuring technique was proposed to treat the malformed cloud due to the rigid body movement. The judgment basis for whether a point should be deleted was established. A new succinct method was suggested to build restructuring cavity. The weighted-point filling strategy above was utilized to stuff new points into the cavity, and the flow parameters of new added points were interpolated by a linear interpolation. Whereafter combined with previous studies, the gridless algorithm for simulation of reacting flows involving moving boundaries was developed based on multi-component Euler equations of ALE (Arbitrary Lagrangian-Eulerian) form. Given that the restructuring process resulted in excess work, the parallel arithmetic on the basis of SPMD (Single-Program Multiple-Data) model was discussed in order to improve the computational efficiency and enlarge the computational scale. Messages were transferred between processes by making use of MPI (Message Passing Interface). Lastly, the flows around the cylinder and the muzzle flow field of ideal shoot for a 7.62 mm gun were simulated to validate in respect of efficiency.Finally, in order to test the application capability of the gridless algorithm in practical engineering problems, the denotation wave phenomena induced by supersonic projectile with different conical angles and velocities were simulated; the internal and external flow fields in a single-cycle pulse detonation engine without and with diverging nozzle were computed respectively; the muzzle flow fields were numerically studied exhaustively, and the influences of propellants components, muzzle attachment, muzzle pressure, atmospheric environment on the muzzle flash were investigated separately. |
PDF Full Text Request