The thickness mistuning of coatings and blisks cannot be predicted before manufacturing process, and the mistuning patterns normally tend to be random, which has not been investigated in opened studies. A novel finite element model (FEM) is developed to incorporate variable thicknesses of the coatings and the blisks. The FEM requires only the sector-level node coordinates of the uncoated blisks and is validated by the ANSYS software. Nevertheless, the classical subset of nominal modes (SNM) cannot capture the accuracy of coated blisks with thickness mistuning and to solve the eigenproblem of the full-order FEM will require expensive computational cost. As a consequence, on the basis of a subspace iteration method, an extended SNM is developed to improve the computational efficiency, where the initial mode matrix of the iteration is obtained by using the Craig–Bampton method and the cyclic symmetric boundary conditions. This paper presents a numerical comparison of the classical SNM, the extended SNM and the full-order FEM. The results suggest that the developed FEM and the extended SNM are effective for the analysis with thickness mistuning of the coatings and the substrate. Compared with the solution time of the eigenproblem for the full-order FEM, it is reduced by roughly 89.5% for the extended SNM.

中文翻译：

---

带涂层和叶盘厚度不均匀的涂层叶盘标称模式方法的扩展子集

在制造过程之前无法预测涂层和叶盘的厚度不均匀，而且不均匀图案通常趋于​​随机，这尚未在公开研究中进行研究。开发了一种新颖的有限元模型（FEM），以结合可变厚度的涂层和叶盘。FEM只需要未涂覆的叶盘的扇区级节点坐标，并通过ANSYS软件进行了验证。然而，标称模式（SNM）的经典子集无法捕获带有厚度模糊的包被叶栅的精度，并且要解决全阶有限元法的本征问题，将需要昂贵的计算成本。因此，在子空间迭代方法的基础上，开发了扩展的SNM以提高计算效率，其中，通过使用Craig–Bampton方法和循环对称边界条件获得迭代的初始模式矩阵。本文介绍了经典SNM，扩展SNM和全阶FEM的数值比较。结果表明，开发的有限元法和扩展的SNM可有效地分析涂层和基材的厚度。与全阶FEM的本征问题的求解时间相比，对于扩展的SNM，它的求解时间减少了大约89.5％。结果表明，开发的有限元法和扩展的SNM可有效地分析涂层和基材的厚度。与全阶FEM的本征问题的求解时间相比，对于扩展的SNM，它的求解时间减少了大约89.5％。结果表明，开发的有限元法和扩展的SNM可有效地分析涂层和基材的厚度。与全阶FEM的本征问题的求解时间相比，对于扩展的SNM，它的求解时间减少了大约89.5％。