| As the technology develops,the performance and reliability of aero-enginesdraw wide concern and high requirements.Twin-web turbine technique is a key technologyin advanced aero-engines,which can improve the performance of aero-engines and decrease their weight effectively.However,the researches on the structure design and fatigue analysisof twin-web turbine disk are very limited.Besides,the probabilistic fatigue life prediction accuracy of high temperature alloysfor the stress characteristics of turbine disk is unsatisfactory,for being lesseffective toconsider the influence of mean stress effect caused by unsymmetical loads,notch effect caused by inconsecutive geometitry features,and size effect caused by large detrimental volume on fatigue life.Thus,the twin-web turbine disk made of high temperature alloyisstudied as an object.In this dissertation,probabilistic fatigue life models were proposedwith the considerations of the influence of mean stress effect,notch effect and size effect on fatigue life,theautomatic optimization platformwas established based on lifereliability,and the design methodof simulated specimenswas presented with equivalent fatigue life in elastoplasticity cases.Firstly,by analyzing the influenceof the mean stress/strain effect on fatigue life,it is noticed that both of thematerial and stress level have unnegligible influence on mean stress effect.Therefore,the modified Walker stress-life model and a new strain-life model were proposedseparately,both ofwhich can effectively take consideration of theaforementioned two factors.The modified Walker stress-life model gives the minimum mean values and standard deviations of prediction errors for all the nine sets of fatigue test data from smooth bars.While for the new strain-life model,it provides the fatigue life prediction results of smooth bars made of two kinds of steels in the double scatter band,as well as the maximum error for smooth bars and center-hole specimens made of GH4133 superalloy being-3.67%.It can be seen that the two proposed models show satisfactory capability to estimate fatigue lives with mean stress effect.Besides,further analysis on these two models reveals that the mean stress effect becomes more obvious with the decrease of stress level and the increase of stress ratio.Secondly,through the investigations on the internal relation between the inelastic responses in the cyclic stress-strain curve and the fatigue life,the area of cyclic stress-strain curve deviate from elastic line duo to plastic deformation in the first cycle is taken as the plastic strain energy damage parameter,whereafter it was developed to the equivalent plastic strain energy damage parameter to include the mean stress effect.Therefore,the plastic strain energy model to predict fatigue life of small and notched engineering components in view of energy is proposed,afterwards the model is expanded to the probabilistic form to consider the influence of the variability of parameters such as material properties,geometrical features and loads,especially the randomness of the cyclic stress-strain curve on fatigue life.The plastic stain energy model was employed to analyze the fatigue lives of eight sets of smooth bars and two sets of turbine disk simulated specimens,and it is showed that for smooth bars,both of the plastic strain energy model and the modified Walker stress-life model provided good predictions,however for simulated specimens,the plastic strain energy model gave more precise results,with the maximum prediction error being about 3.0%,which was much better than any other candidate models.Besides,the probabilistic plastic strain energy model was used to investigate the probabilistic fatigue lives of the aforementioned simulated specimens,and the results demonstrated that the distribution of predicted fatigue lives and the test data shared the same tendency,with the maximum prediction error for the specific reliability being less than -26.0%,showing favorable accuracy.Furthermore,the sensitivity analysis based on the probabilistic model revealed that the randomness of the cyclic stress-strain curve,especially the plastic part,had great impact on the dispersibility of fatigue life.Thirdly,by investigating the notch effect and size effect on fatigue behavior,it wasdiscovered that notch effect belongs to the generalized size effect in essence,and the weakest link theory can reasonably account for the generalized size effect in view of reliebility.The three-parameter Weibull distribution and logarithmic normal distribution were analyzed and compared,it was found that the three-parameter Weibull distribution were more suitable to describe the distribution of fatigue life,hence,a uniform three-parameter Weibull distributionwas establishedto account the scatter of fatigue life under various stress levels with the assumption that the fatigue life in different stress levels share similar distributions.Besides,based on the principle of“while the probabilistic fatigue lives are the same,the damage contained are equal”,an equivalent volume analysis method accounting for mean stress effect,notch effect and size effect was proposed,in which the detrimental volume with various stress and stress ratios in the high stress region of a notchcould be transferred to the equivalent volume of the target stress and stress ratio,and accumulated,containing equal damage.The equivalent volume analysis method was applied to predict the medium fatigue life of 10 sets of notched bars and the probabilistic fatigue lives of 160 pieces of simulated specimens of turbine disk with various notch types,temperatures,thickness and load levels,which were all made of FGH96 superalloy.The results show that the estimated medium fatigue lives of notched bars are basically in the double scatter band,and the predicted probabilistic fatigue lives of simulated specimens are better than any other candidate models in study,with the least mean error values and standard deviations,showing favorable capability to estimate probabilistic fatigue lives with good stability.Fourthly,by studying the common failure mode of turbine disks and the static strength optimization method,an automatic strength and life reliability optimization platform with ANSYS and MATLAB dual softwear for twin-web turbine disk on the base of equivalent volume analysis method was developed.The effect of random factors such as the cyclic stress-strain curve,geometrical features and loads on fatigue life were considered through the modification of the uniform three-parameter Weibull distribution.The optimized results of a certain twin-web turbine disk show that there arefive detrimental regions in the disk in total,three of which are close to the lower bounds,while the other two are also not far from the design boundary.It can be seen that the optimized twin-web turbine disk is basically designed with equal probabilistic fatigue lives.Besides,the stress safety factors for the optimization of twin-web turbine disk were recommended based on the design criteria of traditional turbine disk.Finally,an elastoplasticity specimen simulating method was proposed to ensure equal fatigue life,with the similarities in the maximum Von-Mises equivalent stress,the maximum equivalent stress gradient and detrimental equivalent volume in the critical region as design rules.Besides,an equivalent way to make up for the significant difference of detrimental volume between the specimen and the critical region of components was proposed by changing the load ratio while the maximum load remaining the same.Whereafter the specimen simulating method was employed to design the simulated specimen for the critical region in the cavity of optimized twin-web turbine disk,and the results show that the probabilistic fatigue life at P_s=500.% and P_s=99.87% are almost the same,which can ensure the similar fatigue behavior between the simulated specimen and the cavity in the twin-web turbine disk. |
PDF Full Text Request