Multidisciplinary Reliability Design And Optimization Of The Twin Web Disk Using The Non-Probabilistic Method

As the continuously rising of the turbine inlet temperature,the turbine disk with single web with limits of heat transfer and weight loss does not meat the requirement of the modern aircraft engine.Twin-web turbine disk(TWD)with a hollow inner cavity is the future trend due to its advantages on heat transfer strengthen,temperature stress reducing and weight losing.The design process of the TWD disk is a typical multidisciplinary problem involving several disciplines,such as convective heat transfer,structure and vibration.Besides,as a result of the existence of uncertainty,reliability analysis is also necessarily required.However,the probabilistic model has many difficulties appling in the reliability analysis of the novel TWD due to the lacking of the probabilistic distribution information.studies on the the theory and method for the rational design of this novel structure is still lacking.Therefore,the reliability analysis of the TWD using non-probabilistic model was studied in this paper.Through the researches on the non-probability methods,a series of method was proposed to conduct the MDO and reliability based MDO for TWD and proposed novel TWD structure with guided fins in this dissertation.The main contributions are summarized as follows:(1)Individual disciplinary analysis of the typical TWD and the novel TWD with guided fins structure proposed in this paper were conducted,including the convective heat transfer,structural strength and fatigue life.Contents include: Based on the heat transfer experimental data,the numerical analysis of its convective heat transfer model was established with an acceptable accuracy;By convective heat transfer analyzing,temperature distribution of TWD was then obtained;When analyzing the structural strength,the data was transferred from temperature field to structural field using interpolation techniques like inverse distance weighted method;Then the fatigue life was obtained by Morrow mean modified Mason-Coffin equation.(2)A new evidence theory based fuzzy set method was proposed in the paper for nonprobabilistic modeling.Then the uncertainty modelling of the TWD was established using this method.Contents include: The evidence theory and fuzzy sets were studied and a method was proposed based on both theories.Compared with other methods for non-probabilistic modeling,fuzzy set is also used in the description of limit state responses and evidence data in both General and Bayesian structure can be described in this method;At the same time,we also compared the difference among varies of data structure of evidence theory on description of the uncertainty in several examples;Then we proposed a method named evidence theory based fuzzy set for describing non-probability model;Its feasibility was demonstrated by the following example.Finally,the uncertain model of the Twin-web disk was established using this method.(3)A fuzzy set reliability analysis method based on evidence theory was proposed and established for the non-probabilistic modeling of the TWD.Contents include: Based on Monte Carlo method,first order reliability method(FORM),a fuzzy set reliability analysis method based on evidence theory was proposed;A cantilever beam example was given to compare the differences between the description of the uncertainty of probability model and non-probability model;The reliability analysis of the TWD was conducted using this method and reliability results were obtained in different section level;The influence of parameters in memebership fuctions of fuzzy sets was also studied.(4)A developed self-adaptive Kriging surrogate model,which could increase the efficiency of design and analysis while keeps the required accuracy,was proposed for the complicated multidisciplinary analysis process of the TWD.Contents include: Focused on selfadaptive surrogate model,methods and concepts,including the initial sampling method,modeling method,the additional point selection method and convergence criterion were investigated and well-reviewed;A modified quadrangular grid(MQG)sampling method was carried out to increase the property and steady of sampling process;A method called Spacefilling property(SFP)was proposed to value the property of sampling;Developed Kriging selfadaptive surrogate model method was then proposed by using MQG,genetic optimization method and the criterion of the expected improvement(EI);Using this method,a surrogate model for cantilever beam was then established which proves its feasibility;Finally,the selfadaptive Kriging surrogate model of the TWD was established and the accuracy was satisfied.(5)Compared to the single web turbine disk,varies optimizations of traditional TWD and novel TWD with guided fins were conducted.Contents include: The multidisciplinary and multi-objects optimizations for a novel TWD with guided fins in its cavity was conducted with the objects of weight loss and the maximum cooling air outlet pressure using self-adaptive Kriging surrogate model;The reliability based multidisciplinary optimization of TWD was conducted to improve the reliability amd extend the design space compared with the deterministic design.As a result,a better design of TWD can be achieved.