13th International Conference on Fracture June 16–21, 2013, Beijing, China -2- [4] could manage various models so that it could balance the cost and the precision of the MDO [5], and get a feasible optimal solution and avoid numerical noise. Nevertheless, the VCM also exist some shortage in the organization, so some improving methods are required. Recently the MDO has made a great progress in some technology, but a lot of theory and methods which still stay in academic research stage without verification on complex engineering system design could have distance with the practical application. Therefore, this paper, in view of the aero-engine turbine and the high precise coupled analysis, will fully discuss the MDO method of the complex structure, which not only has very important significance and challenging but also provides certain technical support for the development of aero-engine. 2. Turbine Fluid-Solid Coupled Analysis 2.1. ALE Problem of Fluid-Solid Couple For the coupled problem, the most difficulty lies in the unified coordinate system and the coordination of two phases interface. The Solid habits using the Lagrange coordinate system with a view to the particle, while the fluid uses the Euler coordinate system with a view to the space point. The descriptive differences could not be distinguished on a little movement problem, but is very complicated for big movement and nonlinear problem. The ALE (Arbitrary Lgrangian-Eulerian) method which Hirt puts forward to describe the free liquid in 1974 has been widely used [6]. It provides an efficient way connecting the Laplace system in solid with the Euler system in fluid. The ALE coordinate system could move in the space motion by any speed. If its velocity is zero, that is Euler system. If the speed equals to the particle velocity, that is namely the Laplace system. Thus the ALE coordinate system provides a unified description for two kinds of coordinate system. This paper describes the fluid field in the ALE. The spatial domain uses the finite element discrete format, while the Navier-Stokes equation adopts the substep calculation format in the time domain. The fluid-solid closely coupled based on the ALE can combine the ANSYS with the CFX to complete iterative computation. When they solve in sequence, the interface need transfer the pressure from the fluid to the solid, and then transfer the displacement to fluid. This method considers the interaction between fluid and solid, so the accuracy is relative higher. 2.2. Dynamic Grid Technology and Coupled Step Settings Because of the moving interface, the discrete equation of the fluid field must allow the grid mobile and the grid deformation. There are two kinds of the way [7-10]. The first one is named mesh fairing method dealing with the small mobile. This operation pulls and extrudes the grid in a small range displacement. That could be realized by solving the Laplace's equation. If the grid deformation is serious, the mesh fairing method is not enough to provide the high quality grid. In other case, the new grid must be established in maybe different grid topology. The variable value will be interpolated to the new nodes next time step. Because the grid nodes are inconformity before and after remeshing, the interpolation inevitably would produce the error. The remeshing cost is huge for hundreds of thousands nodes, so the efficiency could drop. Then, if the deformation is not a lot, for example that the maximum deformation in this paper is less than 1.4 mm (analysis data), the small-scale coupled timestep could solve the dynamic grid problem with the first method. Every coupled variable transfers more smoothly by the small-scale coupled timestep. Usually, a blade employs the corresponding period of the first order natural frequency as the total timestep. The first order natural frequency is 723.522Hz. Its corresponding period is 0.00138s. Then
RkJQdWJsaXNoZXIy MjM0NDE=