A multiresolution wavelet optimised finite-difference method for simulation of thermal regenerator. (1st October 2020)

• Record Type: Journal Article
• Authors: Kumar, Gaurav; Murthy, D.S.
• Is Part Of: Thermal science and engineering progress. Volume 19(2020)
• Publication Date: 2020-10-01
• Subjects: Thermal regenerator -- Finite-difference method -- Grid adaption -- Daubechies wavelets; Heat engineering -- Periodicals; Heat engineering; Thermodynamics; Periodicals; 621.402
• DOI: 10.1016/j.tsep.2020.100669
• Languages: English
• ISSNs: 2451-9049

Abstract

Highlights: Wavelets are used for grid optimization in FDM for simulation of regenerator. Multiresolution technique for analysing data has been applied. Various forms of Daubechies wavelets are compared. Reduction of more than 85% in computational time is achieved. Abstract: The equations demonstrating the heat transfer phenomenon in a thermal regenerator were solved with the finite-difference method (FDM) and the method has been improvised by using wavelets. In the thermal regenerators, hot flue gas enters an initially cold bed, transfer all of its thermal energy in a portion of the bed and leaves at low temperature creating a heat transfer zone which traverses the bed-length with time while establishing the thermal equilibrium on both sides of the zone. The discretization process in conventional FDM takes uniform dense grids for whole domain, however, in actual physical process, the heat transfer happens only in a portion of the domain at a certain time requiring dense grids. Such problems demand a front-tracking method which can adjust the grid pattern according to the temperature-front location for which wavelets are found to be suitable. Wavelets functions have the property of being translated and dilated according to the requirement. An adaptive grid scheme has been developed using wavelets along with the multiresolution analysis. The grid patterns for different forms of Daubechies wavelets i.e. db1, db2, db3, and db4 have been compared. The method based on wavelets has been found accurate and less time-consuming.