Litcius/Paper detail

An arbitrary order numerical framework for transient heat conduction problems

Wenxiang Sun, Wenzhen Qu, Yan Gu, Po-Wei Li

2023International Journal of Heat and Mass Transfer34 citationsDOIOpen Access PDF

Abstract

This paper presents an arbitrary order numerical approach for achieving highly accurate solutions in transient heat conduction problems. In this framework, the arbitrary order Krylov deferred correction method (KDC) is first utilized to discretize the temporal direction of the problems, ensuring high-accuracy results by selecting an adequate number of Gaussian nodes. During each time iteration, a boundary value problem in terms of the inhomogeneous modified Helmholtz equation is generated by utilizing the KDC methodology. Subsequently, the arbitrary order GFDM is introduced to solve the boundary value problem, allowing for flexible selection of the Taylor series expansion order to improve spatial accuracy. Consequently, the present numerical framework is of arbitrary order of accuracy and very stable through integrating advantages of the arbitrary order KDC and the arbitrary order GFDM. The developed numerical framework is ultimately verified for its characteristic of high accuracy and stability through two numerical examples in two-dimensions and three numerical examples in three-dimensions. The results highlight the significant improvement in numerical accuracy achieved by the present arbitrary-order numerical approach for long-time simulations.

Topics & Concepts

DiscretizationTransient (computer programming)Boundary value problemNumerical analysisApplied mathematicsHelmholtz free energyStability (learning theory)Taylor seriesComputer scienceNumerical stabilityOrder of accuracyThermal conductionComputer simulationOrder (exchange)GaussianMathematical optimizationMathematicsMathematical analysisPhysicsThermodynamicsEconomicsMachine learningSimulationQuantum mechanicsFinanceOperating systemElectromagnetic Simulation and Numerical MethodsNumerical methods in engineeringModel Reduction and Neural Networks