Agriculture Reference
In-Depth Information
and
−
∂
∂
C
z
C
C
+
i
++ +
1,
j
1
ij
,
1
v
=
v
Oz
()
(3.61)
z
where O(Δ
z
)
2
and O(Δ
z
) are the error terms associated with the above
finite-difference approximations, respectively. In Equation 3.53, the second-
order derivative (the dispersion term) is expressed in an explicit-implicit
form commonly known as the Crank-Nickolson or central approximation
method (see Carnahan, Luther, and Wilkes, 1969). This is obtained using
Taylor series expansion and is based equally on time
j
(known) and time
j
+ 1 (unknown). Such an approximation has a truncation error, as obtained
from the Taylor series expansion, in the order of (Δ
z
)
2
, which is expressed
here as
O
(Δ
z
)
2
. In Equation 3.54, the convection term was expressed in a
fully implicit form, which resulted in a truncation error of
O
(Δ
z
). In this
numerical approximation, for small values of Δ
z
and Δ
t
, these truncation
errors were assumed to be sufficiently small and were therefore ignored
in our analysis (see Henrici, 1962). Chaudhari (1971) showed that due to
numerical approximations, a correction to the dispersion term
D
is needed
such that
D
=D+
D
*
n
(3.62)
where
D
*
is the corrected dispersion and
D
n
is a numerical dispersion term that
is obtained from rearrangement of the higher-order terms of the Taylor series.
For derivatives based on central differences,
D
n
is given by (Chaudhari, 1971):
v
vt
R
=
z
−
D
(3.63)
n
2
Incorporation of
D
n
is a simple task and can yield significant improvement
to numerical approximations.
The time-dependent term of Equation 3.24 was expressed as:
−
∂
∂
C
t
CC
t
+
ij
,
+
1
ij
,
R
=
Ot
()
R
(3.64)
ij
,
where the retardation term
R
was solved explicitly as:
=+
ρ
K
f
R
1
b
C
b
−
1
(3.65)
Θ
Search WWH ::
Custom Search