Chapter 5
Numerical Methods
(Steady and Unsteady)
Finite Difference Method
In general, the temperature in a body is a function
of location and time. In rectangular coordinates,
T = T (x , y , z , t ). The parameters x , y , z , t are continuous
variables over their respective ranges. For the finite
difference method, the parameters are not continuous.
They are only defined at specific locations or times:
nodes for the spatial parameters and time steps for
the time parameter.
The first step in a finite-difference analysis is to
determine the grid and nodal points for the object.
1
, Nodes for a Large Wall
The wall is "L " wide. The surfaces temperatures are known,
and we want to determine interior temperatures. We decide
to have 11 nodes, equally-spaced, with two of the nodes at
total length L
the surfaces. x = =
number of divisions 10
Nodes for a Rectangular Plate
The cross-hatched nodes show the extent of Node 1, a corner node;
Node 7, an interior node; and Node 15, an edge node. This 16-node
grid could be replaced with the 28-node square grid on the next slide.
2
, Nodes for a Rectangular Plate (Cont’d)
With the 28-node grid, we will get better information about
the temperature distribution in the plate than we would have
gotten with the 16-node grid.
Nodes for a Circular Plate
There are 5 nodes, with r = ro /4. Node 1 has a circular area
of radius r/2. Nodes 2, 3, and 4 have ring areas of r width.
Boundary Node 5 has a ring area of r / 2 width. Node 4
is cross-hatched in the figure.
3