You are on page 1of 3

Report on Finite Element Solution of One Dimensional

Transient Heat Conduction


Rashidul Hasib
11/04/17

Abstract
One dimensional transient heat conduction follows a second order second order
derivative of temperature with respect to position and first order derivative of
temperature with respect to time. In this homework a symmetric weak formulation of one
dimensional transient heat conduction through a bar is solved by dividing the bar into
finite number of elements. An object oriented method is followed to organize different
properties of the elements and to perform finite element procedures. The classes of these
properties and procedures were supplied with this homework. Three solutions are
developed using Crank–Nicolson method for three types of boundary conditions and
three solutions are developed using Fourth order Runge-Kutta method for the same three
types of boundary conditions.

Program Description
First three solutions uses a Crank-Nicolson algorithm for time dependence. The last three
solutions uses a fourth order Runge-Kutta method for time dependence. For this another
solver is created whose definition can be found in a new class named
OneDTransientHeatTransfer_FEMSolver2.

Result and Analysis


In Crank-Nicolson algorithm if boundary conditions are changed from heat flux to
convection, there is no change in the temperature which cannot be true.
Fourth order Runge-Kutta method gives approximately same temperature as shown
below but heat flux and convection boundary conditions give the same results.
Crank-Nicolson method Fourth order Runge-Kutta method
Original Solution: temperature BC for each Temperature BC for each end of the bar
end of the bar

Heat Flux, Convection BCs gives the same Heat Flux, Convection BCs gives the same
result result
Conclusion
The result found from Crank-Nicolson method and Runge-Kutta method are
approximately comparable but heat flux and convection boundary conditions are giving
the same result. Besides the algorithms when run for variable spacing of nodes or variable
number of nodes gives erroneous result.

You might also like