Inverse design of an irregular-shaped radiant furnace using neural network and a modified hybrid optimization algorithm. (1st December 2020)

• Record Type: Journal Article
• Authors: Sepahvand, P.; Abdizadeh, G.R.; Noori, S.
• Is Part Of: Thermal science and engineering progress. Volume 20(2020)
• Publication Date: 2020-12-01
• Subjects: Heat transfer -- Radiant furnace -- Optimization -- Inverse method -- GA-SQP; Heat engineering -- Periodicals; Heat engineering; Thermodynamics; Periodicals; 621.402
• DOI: 10.1016/j.tsep.2020.100730
• Languages: English
• ISSNs: 2451-9049

Abstract

Highlights: Inverse estimation of temperature and location of two heaters as well as the location of the design body in an irregular 3-D radiant furnace has been performed. Non-gray calculations have been performed by the spectral line based on the WSGG model. Therefore, the updated coefficients of the high temperature spectral databases (HITEMP2010) applied for non-gray gases. Using a modify form of GA-SQP algorithm reduces the time required for designing the radiation furnace. The GA-SQP reaches to optimal point faster than other algorithms. According to validation results, WSGG model gives good results, with the average relative error, less than 2.1% compares to the LBL. The difference between the optimized and the desired heat flux in the worst case is very negligible, about 0 ∙ 1249 k w / m 2 which is under one percent. Abstract: In this study, the inverse solution technique is applied to design a three-dimensional heating furnace where radiation is the dominant part of heat transfer. The heat transfer mechanism is fully radiative and the furnace medium is non-gray with a mixture of 71.2% N2, 17.7% CO2, and 11.1% H2 O. The design parameters include the temperature and location of the two heaters as well as the location of the design body. For the purpose of estimation of the design parameters to achieve a desired uniform heat flux at the design body, the inverse problem is considered as an optimization problem. The discrete ordinate method (DOM) is used to solve the radiation heat transfer equation (RTE) and non-gray calculations are performed by the spectral line based on the weighted-sum-of-gray-gases (WSGG) model. Additionally, the updated coefficients of the high-temperature spectral databases (HITEMP2010) are applied for non-gray gases. This study employs a modified hybrid method based on the artificial neural network (ANN) and genetic algorithm (GA) coupled with sequential quadratic programming (SQP) to optimize design parameters. In the present work, the results of three optimization algorithms including GA alone, ordinary GA-SQP and a modified form of GA-SQP are compared to find the best method to optimize the furnace problem. The final optimization results demonstrate a negligible difference between the optimized and the desired heat flux which is less than one percent. The consequences prove that the modified form of the GA-SQP algorithm (present work) is faster than others.