Adjoint-based shape optimization for radiative transfer using level-set function and volume penalization method. (15th August 2023)

• Record Type: Journal Article
• Authors: Liu, Ming; Hasegawa, Yosuke
• Is Part Of: International journal of heat and mass transfer. Volume 210(2023)
• Publication Date: 2023-08-15
• Subjects: Radiative transfer -- Inverse design -- Adjoint method -- Level-set function -- Volume penalization method; Heat -- Transmission -- Periodicals; Mass transfer -- Periodicals; Chaleur -- Transmission -- Périodiques; Transfert de masse -- Périodiques; Electronic journals; 621.4022
• DOI: 10.1016/j.ijheatmasstransfer.2023.124158
• Languages: English
• ISSNs: 0017-9310

Abstract

Highlights: A novel adjoint-based shape optimization scheme for radiative heat transfer is proposed. Through comparisons with existing reference cases for simple 2D geometries, it is demonstrated that the present optimization algorithm can achieve similar or better performances. The proposed scheme is further applied to a more complex problem of optimizing 3D geometry and show its superiority in large-dimensional optimization problems. Abstract: A inverse design of a radiative heat transfer surface is established based on an adjoint analysis combined with a level-set function and a volume penalization method. The level-set function is employed to represent an arbitrary complex geometry embedded in a Cartesian grid system, whereas the volume penalization method is applied to simulate a forward radiative transfer problem. Then, the sensitivity of a current shape with respect to a prescribed cost functional is obtained from adjoint analysis, and it is used to update the shape. The entire inverse design procedures including the forward and adjoint analyses as well as the shape update based on the adjoint analysis are implemented to an open-source CFD solver, OpenFOAM. Two pure radiation problems from the existing literature are considered for verifying the proposed method. The resultant optimal geometries in both problems show good agreement with the reference data. Finally, the present method is applied to a more complex three-dimensional problem to validate the strength of its applicability to complex geometries.