This paper suggests a selection criterion of the continuous method version for a numerical solution of the Stefan problem which would allow to calculate the phase transition boundary position with a required accuracy for a long period of time and would enable generalization to multidimensional problems. Despite a large number of works deal with the solution to the generalized Stefan problem by the continuous method, the choice of the smoothing interval value for numerical feasibility is not fully clear. A comparison of the calculation accuracy of the phase transition boundary position using different versions of the continuous method was carried out on an example of the well-known 1-D plane two-phase Stefan problem which possesses an analytical solution. The dependence of the total error of the numerical calculation of the phase transition boundary position on the value of the smearing interval is determined from the comparison of numerical and analytical solutions. An analysis of the reason for increase of this error with time at any choice of a constant smoothing interval is given. A version of the continuous method with a variable interval of the delta function smoothing, in which the proposed criterion is carried out, is discussed. The position of the phase transition boundary calculated proposed version matches the analytical solution with a required accuracy over a long period of time.

Original languageEnglish
Pages (from-to)2267-2277
Number of pages11
JournalThermal Science
Volume24
Early online date10 Aug 2019
DOIs
StatePublished - 2020

    Scopus subject areas

  • Renewable Energy, Sustainability and the Environment

    Research areas

  • Choice of the delta function smearing interval, Glaciation, Numerical solutions, Stefan problem, The continuous method version

ID: 49523294