Structural Optimization Of Turbine Foundation Based On Krigirig Surrogate Model And Modified El Criteria

As an important part among the design of large power station, the design quality ofturbine foundation will influence the running security of the whole power station directly. Fordomestic imported turbine, lack the matched method of designing foundation，and nuclearpower unit of1000MW and above always is half speed turbine, which is easier to resonatewith foundation, so if s necessary to optimize the foundation structure. The main researchcontents and conclusions are as follows:In order to reduce the cost and improve the running security, we establish themulti-objective optimization model, which is to find two kinds of design variables (structuresizes: the cross-sectional areas of beams and columns and the thickness of soleplate,geometric shapes: the coordination of nodes of columns in the foundation), such that theweight and the maximum vibration amplitude of the turbine foundation are to be minimized.Using hierarchical optimization technique, firstly to optimize structure sizes variables，then,turn to the geometric shapes variables.The general finite element analysis sotfware-ANSYS is used to obtain the dynamicdisplacements based on the Winkle-Voigt modal, then, establish the relationship betweenthese dynamic displacements and design variables by using Kriging agent model interpolationtechnology. The final optimal solution of the turbine foundation is conducted by sequentiallymodifying the surrogate model which is driven by the optimization criteria.. Winkle-Voigtmodel considers the interaction between foundation slab and subsoil, which simulatesvibration properties of subsoil with spring-damper. Kriging agent model has both flexiblefeatures of parameterized model, also overcomes the limitation of nonparametric model whileprocessing high dimensional problem.A new criterion for iniflling sample is proposed, which is based on parametric samplingevaluation (PSE) function and maximized weighted expected improvement (WEI) criterion. Aseries of Pareto solutions are obtained by maximizing WEI function, then，cohere the EIfunction by mean of these Pareto solutions. Compared with the traditional EI criterion，it canobviously reduce iteration steps and approximate global optimal solution effectively andstably.A universal program is developed to optimize the structure of turbine foundation bychanging structure sizes and geometric shapes. As examples, two practical turbinefoundations are investigated. The results indicate that the optimization model and method weadopted can be used in optimization design of turbine foundation effectively，and it couldlargely reduce the structure weight and dynamic displacement.