Lecture 4
One Dimensional Fluid Flow
FEM Dr. Riessom 1
Fluid Flow in Pipes
We now consider the steady-state irrotational flow of an incompressible
and inviscid fluid.
For the ideal fluid, the fluid particles do not rotate; they only translate, and
the friction between the fluid and the surfaces is ignored.
Also, the fluid does not penetrate into the surrounding body or separate
from the surface of the body, which could create voids.
The equations for this fluid motion can be expressed in terms of the stream
function or the velocity potential function.
The velocity v of the fluid is related to the velocity potential function by
FEM Dr. Riessom 2
Fluid Flow Equation
In the absence of sources or sinks Q, conservation of mass in two
dimensions yields the two dimensional differential equation as
At an impermeable boundary, the flow velocity and thus the derivative of
the velocity potential normal to the boundary must be zero.
The one-dimensional finite element formulation of the fluid-flow problem
by now realizing that the fluid-flow problem is analogous to the heat-
conduction problem
FEM Dr. Riessom 3
One-Dimensional Finite Element Formulation
The fluid velocity potential function for the temperature function T, the
vector of nodal potentials denoted by {p}for the nodal temperature vector
{t}, fluid velocity v for heat flux q, and permeability coefficient K for flow
through a porous medium instead of the conduction coefficient K. If fluid
flow through a pipe or around a solid body is considered, then K is taken
as unity.
The basic two-node element is again used, as shown in Figure, with nodal
fluid heads, or potentials, denoted by p1 and p2
FEM Dr. Riessom 4
Finite Element Formulation Cont’d
We choose the potential function similarly to the way we chose the
temperature function.
where p1 and p2 are the nodal potentials (or fluid heads in the case of the
seepage problem) to be determined, and are again the same shape functions
used for the temperature element. The matrix [N] is then
The hydraulic gradient matrix {g} is given by
FEM Dr. Riessom 5
Finite Element Formulation Cont’d
where [B] is identical to the strain relating matrix
and
The velocity/gradient relationship based on Darcy’s law is given by
where the material property matrix is now given by
That is, the fluid-flow stiffness matrix is analogous to the conduction part
of the stiffness matrix in the heat-transfer problem. There is no comparable
convection matrix to be added to the stiffness matrix.
FEM Dr. Riessom 6
Finite Element Formulation Cont’d
Consider the fluid element shown in Figure with length L and uniform
cross-sectional area A.
In the fluid-flow problem, we define the stiffness matrix to relate nodal
volumetric fluid-flow rates to nodal potentials or fluid heads as f=k*p.
Therefore,
defines the volumetric flow rate f in units of cubic meters or cubic inches
per second.
In scalar form; g is given in explicit form by
FEM Dr. Riessom 7
Finite Element Formulation Cont’d
Applying at nodes 1 and 2, we obtain
and
where f1 is directed into the element, indicating fluid flowing into the
element ( p1 must be greater than p2 to push the fluid through the element,
actually resulting in positive f1), whereas f2 is directed away from the
element, indicating fluid flowing out of the element;
In matrix form, we have for flow through a porous medium.
The stiffness matrix is then
FEM Dr. Riessom 8
Finite Element Formulation Cont’d
In general, the basic element may be subjected to internal sources or sinks,
such as from a pump, or to surface-edge flow rates, such as from a river or
stream. To include
these or similar effects, consider the element of Figure shown below now
to include a uniform internal source Q acting over the whole element and
a uniform surface flow rate source q acting over the surface, as shown in
Figure. The force matrix terms are
where Q will have units of m3/(m3 s), or 1/s, and
FEM Dr. Riessom 9
Finite Element Formulation Cont’d
The above equation indicate that one-half of the uniform volumetric flow
rate per unit volume Q (a source being positive and a sink being negative)
is allocated to each node and one-half the surface flow rate (again a source
is positive) is allocated to each node.
We assemble the total stiffness matrix [K], total force matrix {F}, and total
set of equations as
The assemblage procedure is similar to the direct stiffness approach, but it
is now based on the requirement that the potentials at a common node
between two elements be equal.
FEM Dr. Riessom 10
Example
Determine (a) the fluid head distribution along the length of the coarse
gravelly medium shown in Figure, (b) the velocity in the upper part, and (c)
the volumetric flow rate in the upper part. The fluid head at the top is 10 in.
and that at the bottom is 1 in. Let the permeability coefficient be Kxx=0.5
in./s. Assume a cross-sectional area of A=1 in2.
Using three elements, each 10 in. long.
We calculate the stiffness matrices for each element as follows:
FEM Dr. Riessom 11
Solution
In this example Q=0 (no sources or sinks) and q*=0 (no applied surface
flow rates). Therefore, the assembly of the element stiffness matrices.
Known nodal fluid head boundary conditions are p1=10 in. and p4=1 in.
The resulting solution is given by
The volumetric flow rate Qf in element 1
This volumetric flow rate is constant throughout the
length of the medium.
FEM Dr. Riessom 12
