Uncertainty Analysis And Experimental Calibration Methods Of Thermal Structure Model Of Composite Materials

Multi-source uncertainties exist in the thermal structure design of composite materials,such as material properties,structural interface properties and failure criteria.In order to improve the reliability and robustness of the design structure,the structural design process based on the uncertainty analysis is proposed in this paper.Aiming at the contradiction between the accuracy and efficiency of uncertainty analysis in the design process,the efficient global sensitivity analysis method,surrogate model and dimension reduction method are studied,and a set of systematic uncertainty analysis method is developed.Aiming at the experimental calibration problems of the key uncertainty parameters and the multi-directional stress failure criterion,a set of systematic experimental calibration method is established.Based on the structural design process,a highly reliable strut structure design is achieved and verified by several repeated tests under extreme conditions.(1)Aiming at the problems of the uncertainty analysis of the thermal structure,the direct sampling-ordered permutation-Fourier amplitude sensitivity test method(DRS-OA-FAST),Fourier amplitude sensitivity test method based on single-frequency combined mapping function(SF-CMF-FAST)and the adaptive gaussian process model(AGPM)method are proposed.Compared with the FAST method,the computation efficiency of the DRS-OA-FAST method is increased by more than 95%.And the SF-CMF-FAST method reduces the calculation error of the total sensitivity index from 22% to 2%.AGPM uses a combination kernel function to improve its generalization ability,meanwhile it automatically chooses training samples that have great influences on the model precision,which reduces redundant training samples.Taking the engine nozzle thermal structure as an example,the calculation efficiency of AGPM method is more than 98% compared with the the finite element analysis.(2)Based on the active subspace method,the study of dimensionality reduction of thermal structure model is carried out to further improve the efficiency of the uncertainty analysis.First of all,a method to calculate the active subspace eigenvector based on the reduced interval is proposed.In an example of this thesis,the efficiency is improved by more than 95% compared with the traditional method.This method is applied to the engine nozzle model,and the computation efficiency is increased by more than 75% compared with the surrogate model method.At the same time,the method of data mapping between the active subspace and the complete space is established,which guarantes the realization of the AGPM in the active subspace.Finally,the problem of uncertainty calibration in active subspace is analyzed and the following conclusions are obtained: Uncertainty calibration of the combination parameters based on active subspace method can avoid the "ill-posedness" of the reverse calibration.Adding the Experimental information to the likelihood functions can improve the calibration accuracy of the uncertainty parameters.(3)Aiming at the calibration problem of the key uncertainty parameters and failure criterion in thermal structure analysis,the efficient calibration methods are established.First of all,based on the isothermal cooling process,the experimental calibration methods of the thermal expansion coefficient,total emissivity and interface thermal conductivity are established,which can be used to calibration the high temperature non-linear properties through single heating-cooling experiment,the calibration error are less than 6%,9 % and 12.7% respectively.Secondly,a calibration method of bi-modulus based on the four-point bending test is proposed,in which tensile modulus and compressive modulus are calibrated simultaneously through a single four-point bending test.Finally,a method to correct the material failure criterion under bidirectional stress state based on the dual flexure test is proposed,and the revision of the failure criterion of UHTC material in the hot-pressing direction is completed.(4)Considering the uncertainties of the material property and the interfacial thermal resistance,the uncertainty design of the fuel injection strut is carried out.Through the established uncertainty analysis method,the uncertainty distribution of structural response of quantities and the key uncertainty parameters affecting the response are obtained.The key parameters are calibrated one by one,which reduces the uncertainties of the model parameters and responses.Based on the proposed thermal structure design flow and the established method,a highly reliable(> 99%)FGLCS injection structure design is finally achieved,and verified by the repeated experiments.