The geometric deviation of a precision casting low-pressure turbine blade is studied by using the method of probability and statistics. The main conclusions are as follows:
(1) Through the eigenvalue decomposition of geometric deviation, it can be seen that there are many sources of blade geometric deviation, mainly including overall geometric deviation (positional error, torsional error, etc.) and local deviation (blade profile error). Taking all geometric deviations as local deviations and evaluating the process accuracy will significantly reduce the blade qualification rate. It is necessary to decompose the geometric deviation and evaluate its accuracy separately.
(2) Using low precision blade geometric description to minimize the maximum absolute distance from the measured point to the design blade, supplemented by one-dimensional linear search within the scope of the global optimal solution, can significantly improve the optimization efficiency while ensuring the optimization accuracy. Using the optimization strategy in this paper, the calculation time of geometric deviation decomposition is reduced by 73.7%.
(3) Compared with before the decomposition of geometric deviation, the statistical mean and standard deviation of blade profile profile after decomposition have decreased significantly in the whole leaf surface, the profile error has also decreased significantly, and the blade qualification rate has increased significantly, which can meet the current industrial requirements of high-precision grade in China. On the whole, the proposed geometric deviation decomposition and statistical method can be used to evaluate the profile accuracy of precision casting turbine blades.
(4) The probability density functions of the overall geometric deviation parameters of precision casting turbine blades are close to the Gaussian distribution, and their statistical mean values show that there are obvious blade body deflection and torsion on each section, and obvious blade body deflection along the blade height direction. The probability density function of blade profile at several key positions on the blade surface is also very close to the Gaussian distribution. The leading edge of the blade is thinner, but more concentrated; The average contour of the tail edge is small, but the shape is relatively scattered; The suction surface and pressure surface become thicker, but the shape of the suction surface is more dispersed. The above research can provide a useful reference for the statistical modeling of the overall geometric deviation and blade profile error of precision casting turbine blades.
(5) The statistical modeling of key design parameters such as the radius of front and rear edges, inlet and outlet wedge angles, and inlet geometric angles of precision casting turbine blades and their effects on the aerodynamic performance of the blades need to be studied.