An Non-Iterative Method For Inverse Heat Conduction Problem And Its Applications

Inverse Heat Conduction Problem (IHCP) could be found widely invarious fields of nature science and engineering, especially the applicationsin the realm of nuclear security system, heat treatment of materials andaerospace. After years of development, iterative methods for IHCP arerelatively mature, such as conjugate gradient method, regularizationmethod etc. However, these methods are time-consuming and have intensevibration during the calculation process. In particular, influenced by theill-posed nature of inverse problem, subtle fluctuation or slight error of theoriginal conditions might cause severe shock or complete distortion duringthe calculation process. This is one of the biggest challenges during thestudy of inverse problem as well as its practical applications in many fields.Therefore, many scholars continue to seek for new methods andoptimization ideas to overcome the impact of the ill-posed problem.This thesis established a non-iterative model and solution method forIHCP based on Finite Element Method (FEM). On the basis of the finalmatrix equation of Direct Heat Conduction Problem (DHCP), unifiedlinear transformation was carried out where matrix elements of heat fluxunknown quality were transformed as a basic unit. This processsuccessfully avoided high-level matrix inversion, so as to preserve a highdegree sparsity of the matrix. The method needs no iterative optimization.After solving the final matrix equation of IHCP, the real solution ofboundary conditions of IHCP could be obtained directly. Therefore, it ispossible to achieve the correction of systematic errors of the measuredtemperature, improve the ill-posed problem of IHCP, raise the computationefficiency and reduce the vibration of the results. Designed IHCP program with Fortran, used a simplified model of theIVR of nuclear security system and given a default value of the boundarycondition to form a DHCP, after calculation of the DHCP through Fluent,temperature of the computational domain could be gained. Take part of thetemperature values as experimental measurements, an IHCP could then beformed and calculated through non-iterative method. Through comparisonbetween the default value of boundary condition used by Fluent and thecomputing value of IHCP, the correctness of the non-iterative model andthe IHCP program could be verified. In addition, the influence from thelocation of the temperature measuring points, mesh density and time stepto the results was further studied, and the common credibility criterion ofthe computing results of IHCP was then put forward.This non-iterative method was applied to the calculation of IHCP ofthe laminar cooling process of the hot-rolled steel plate. In order to get thetransient temperature curve of some fixed positions, embedding nickelchromium-nickel silicon thermocouple was adopted during the experiment.The transient temperature curve was used to calculate convection heattransfer coefficient, heat flux and boundary temperature. Throughcomparison of the convective heat transfer coefficient obtained by bothconjugate gradient method and the non-iterative method, the result of thenon-iterative method was shown to be more stable and had a higheraccuracy, which could be a better way to control the intense vibration ofthe result. Especially in air-cooling area, the result of non-iterative methodwas more believable. In this thesis, the non-iterative method could be aneffective way to realize the real-time monitoring of the thermal boundaryconditions, which has widely application prospect in the field of nuclearsecurity system and heat treatment of metal materials etc.