Denis Lemonnier

In this article, an inverse mathematical model was developed to achieve fast and simultaneous estimation of thermophysical and geometric parameters for thermal barrier coating (TBC) systems in high-temperature environments with measurement errors. First, considering the convective and radiative heat transfer between the TBC system and the external environment, a one-dimensional unsteady conduction–radiation coupled heat transfer model was originally developed in high-temperature environments. Subsequently, this forward model was solved using the finite volume method (FVM), and the grid independence as well as the accuracy were validated. Thereafter, an accelerated inverse model, which adopts a principal component analysis (PCA)-optimized artificial neural network (ANN) to fit and substitute the forward model and employs the particle swarm optimization (PSO) algorithm for estimation, was developed based on the inverse method. Finally, under three different noise conditions, two situations were used to perform simultaneous estimation studies: a two-parameter case (top coat thermal conductivity and thermally grown oxide (TGO) layer thickness) and a three-parameter case (adding the thickness of an additional existing debonding defect layer). The results show that the PCA-optimized ANN-PSO-based accelerated inverse model is approximately 111–130 times faster than the traditional PSO-based inverse model, and the maximum relative errors of the arithmetic means of 20 estimations for two-parameter and three-parameter situations under three noise cases are on the order of 2.4% to 4.3%. Overall, the proposed accelerated inverse model achieves a favorable balance between speed and accuracy in multi-type parameter simultaneous estimation for TBC systems at high temperature, providing a methodological basis for addressing parameter estimation in more complex situations in future research.

Radiative heat transfer in buildings can be determined through the resolution of the radiosity equations. This technique requires the evaluation of geometric couplings between the elements of the scene called view factors. Obtaining them is particularly challenging, especially when obstructions are involved. In addition, the system to be solved can be cumbersome, since each surface is likely to interact with all the others and the number of nodes involved in the description of a complex scene is large. We present a solution method that refines the mesh of a scene while simultaneously constructing a representation at different scales of the view factors between the contributing elements, to avoid calculating all the transfers at the highest resolution. It thus considerably reduces the calculation time and makes the tool suitable with a routine use in a design perspective.

The influence of surface radiation on the transition to the unsteady state in natural convection is studied numerically. The configuration of the differentially heated square cavity with adiabatic horizontal walls is chosen to generate an internal natural convection flow. It is known that radiative transfers reduce the temperature difference between the adiabatic walls, which consequently reduces the thermal stratification of the central zone and increases the velocity flow. Many studies have focused on the stationary regime, but few of them have investigated the transition to unsteady flow. For this purpose, the effect of the wall emissivity on the critical Rayleigh number and the associated critical frequency was studied for a given cavity length. The cavity length and mean temperature of isothermal walls are set for the whole study. The results show that all these values are between the values obtained without radiation and those obtained for perfectly conducting horizontal walls. The critical Rayleigh number decreases with emissivity while the associated frequency increases. Moreover, the symmetry of fluctuating properties of the flow is changed when the radiation is taken into account.

In this article, a multilayer artificial neural network (ANN) identification model is developed to simultaneously identify the thermal conductivity and the effective absorption coefficient of semitransparent materials from flash-type experimental measurements. Firstly, the ANN is trained by means of data generated by a 2-D axisymmetric heat transfer model whose radiative part is treated via the P1 approximation. A sensitivity study is then used to prove the theoretical feasibility of the identification strategy. Several training data distributions (uniform or Gaussian types) are tested on synthetic data, and on noisy ones for checking the robustness. Finally, the efficiency of this estimation approach is investigated using experimental data obtained by flash method on a PMMA sample. The estimated thermal conductivity and the effective absorption coefficient are compared with values obtained from the literature and other measurements.

Following previous theoretical development based on the assumption of a rank correlated/comonotonic spectrum, the Locally Correlated SLW (LC-SLW) method is outlined. The relationship between the LC-SLW method and the RC-SLW method is established, and their performance in prediction of radiative transfer is compared. Despite the fact that the RC-SLW model demonstrates better overall accuracy, the LC-SLW model generally proves more accurate in regions of higher gas temperature.