| Popis: |
In mechanical structures exposed to high temperatures, such as turbine blades for power generation, cracks are detected during in-service inspections. However, there is a risk of fracture owing to the relatively rapid rate of crack propagation, even when a large safety factor is in force. Therefore, a technique is required to estimate the creep-strain distribution before crack initiation and to predict the remaining service life. An inverse analysis method, based on the eigenstrain methodology, has been proposed to estimate the three-dimensional creep-strain distribution using non-destructively measured surface displacements. Three-dimensional creep-strain distributions can be estimated even when the relationship between the creep strain and displacement is non-linear. Through iterative calculations, this method converges to a solution with relatively high estimation accuracy, however, a method for estimating actual complex creep-strain distributions is needed. In this study, a method was proposed to improve the convergence of the iterative calculations. Its effectiveness was demonstrated by numerical analysis using a torsional geometry model that could be used to describe an actual turbine blade. Many unknown values were assumed so that relatively complex creep-strain distributions could be estimated. Convergence calculations were performed to improve the estimation accuracy through iterative calculations, even when displacement measurement errors were considered. The estimation accuracy was improved by using more actual measurements in the inverse analysis. |
