Professional Documents
Culture Documents
QUICK Scheme
QUICK Scheme
n u dA n grad dA S dV
A
CV
Apply the central differencing method of obtaining discretised equations for the
diffusion?
Not the same, since the diffusion process affects the distribution of a transported
quantity along its gradient in all directions, whereas convection spreads influence
only in the flow direction (i.e. the grid size is dependent on the relative strength
of convection and diffusion)
In the absence of sources, the process in a given one-dimensional flow field u is:
d
u d d
dx
dx dx
uA e uA w
dx
dx
Define the convective mass flux per unit area, F, and diffusion conductance at
cell faces, D:
F u
with
Fw u w
Fe u e
Dw
w
xWP
De
e
x PE
P E
2
and w
P W
2
Fw
Fe
Dw 2 De 2
Fw
Fe
P Dw 2 W De 2 E
or
Fw
Fe
Fw
Fe
Dw 2 De 2 Fe Fw P Dw 2 W De 2 E
where
aW Dw w
2
a E De e
2
a P aW a E Fe Fw
Examples: one-dimensional steady states convection and diffusion
A property is transported by means of convection and diffusion through the onedimensional domain. The equation governing is:
d
u d d
dx
dx dx
and
L 0
xL
Using 5 equally spaced cells and the central differencing scheme for convection and
diffusion, calculate the distribution of as a function of x for,
(1) Case 1: u = 0.1 m/s
(2) Case 2: u = 2.5 m/s
(3) Case 3: u = 2.5 m/s with 20 grid nodes
L = 1.0 m, = 1.0 kg/m3, = 0.1 x 10-3 kg/m/s.
Ans:
Divide the length into 5 equal control volumes. This gives x = 0.2 m.
where
aW Dw w
2
a E De e
2
a P aW a E Fe Fw
For the boundary nodes 1 and 5, from:
Fe
P E Fw W P De E P Dw P W
2
2
Fw
P W DB B P Dw P W
2
a PTP aW TW a E TE S u
with
a P aW a E Fe Fw S P
we have
Node
aW
aE
SP
Su
D-F/2
-(2D+F)
(2D+F)A
2,3,4
D+F/2
D-F/2
D+F/2
-(2D+F)
(2D-F)B
0
0
0
1.55 0.45
0.55
1.0
0.45
0
0
0
0.55
1.0
0.45
0
0
0.55
1.0
0.45
0
0
0
0
0.55 1.45
1
1.1
0
2
3 0
4
0
5
0
1 0.9421
0.8006
2
3 0.6276
4 0.4163
5 0.1579
ux
1
0
x
x
x
( 3 )
( 2 )
( 3 )
e 3 4
w3 3
q B w 4 4
x
x
x
qB q A
e1
Flux consistency ensures conservation of over the entire domain for the
central difference formulation of the diffusion flux.
In consistent flux interpolation formulae give rise to unsuitable schemes that do
not satisfy overall conservation.
The flux values calculated at the east face of control volume 2 and the west
face of control volume 3 may be unequal if the gradient of the two curves are
different at the cell face. If this is the case, the two fluxes do not cancel out
when summed and overall conservation is not satisfied.
This should not suggest that the quadratic interpolation is entirely bad. A
8
Boundedness
When solving a set of algebraic equations, normally iterative numerical
techniques are used. Scarborough (1958) has shown that a sufficient condition
for a convergent iterative method can be expressed in terms of the values of the
coefficient of the discretised equations:
a
a P
nb
1 at all nodes
aP is the net coefficient of the central node P (i.e. aP SP) and the summation
in the numerator is taken over all the neighbouring nodes (nb).
If the differencing scheme produces coefficients that satisfy the criterion, the
resulting matrix of coefficients is diagonally dominated.
For diagonally dominance, we need large values of net coefficient (aP SP).
(i.e. the linearisation practice of source terms should ensure that SP is negative.)
Diagonal dominance states that in the absence of sources the internal nodal
values of the property should be bounded by its boundary values.
Hence, in a steady state conduction problem without sources and with
boundary temperatures of 500C and 200C all interior values of A should be
less than 500C and greater than 200C.
Another essential requirement for boundedness is that all coefficients of the
discretised equations should have the same sign.
Transportiveness
Define the non-dimensional cell Peclet number as a measure of the relative
strengths of convection and diffusion:
Pe
F u
D
x
Two extreme cases: (1) no convection and pure diffusion (Pe = 0) and (2) no
diffusion and pure convection (Pe )
It is very important that the relationship between the magnitude of the Peclet
number and the directionality of influencing, known as the transportiveness, is
borne out in the discretisation scheme.
aW
aE
aP
Dw+Fw/2
De-Fe/2
aW+aE+(Fe-Fw)
Pe
F u
2
D
x
For given values of and , it is only possible to satisfy the condition if the
velocity is small, hence in diffusion-dominated low Reynolds number flows, or
10
11
uw > 0 and ue > 0 (Fw > 0 and Fe > 0), the upwind scheme sets:
w W and e P
the discretised equation is:
Fe P Fw W De E P Dw P W
re-arranging
Dw De Fe P Dw Fw W De E
or
Dw Fw De Fe Fw P Dw Fw W DeE
12
uw < 0 and ue < 0 (Fw < 0 and Fe < 0), the upwind scheme sets:
w P and e E
the discretised equation is:
Fe E Fw P De E P Dw P W
re-arranging
Dw De Fe Fe Fw P DwW De Fe E
aE
Fw > 0 Fe > 0
Dw+Fw
De
Fw < 0 Fe < 0
Dw
De-Fe
or
aW
aE
Dw + max(Fw,0)
De + max(0,-Fe)
13
with
a P aW a E Fe Fw S P
We have:
Node
aW
aE
SP
Su
-(2D+F)
(2D+F)A
2,3,4
D+F
D+F
-2D
2DB
14
15
The error is large for the coarsest grid and that refinement of the grid can, in
principle, overcome the problem of false diffusion.
The upwind differencing scheme is not entirely suitable for accurate flow
calculations.
16
The hybrid differencing scheme uses pieces formulae based on the local Peclet
number to evaluate the net flux through each control volume face:
1
2
1
2
1
W 1
P
Pew
2
Pew
2
q w Fw
q w Fw AwW
for Pew 2
for Pew -2
q w Fw Aw P
aE
max Fw , Dw w ,0
2
max Fe , De e ,0
2
17
with
a P aW a E Fe Fw S P
We have:
Node
aW
aE
SP
Su
-(2D+F)
(2D+F)A
2,3,4
-2D
2DB
Comparison:
a P P aW W a E E a S S a N N a B B aT T
with
a P aW a E a S a N a B aT F
18
(u)wAw
w
Aw
xWP
(u)eAe
e
Ae
x PE
(u)sAs
s
As
y SP
(u)nAn
n
An
y PN
(u)bAb
b
Ab
z BP
(u)tAt
t
At
z PT
q w Fw W w P W
where
1 0.1Pew
q w FwW
Pew
for Pe > 10
19
The coefficients of the one-dimensional discretised equation for steady onedimensional convection-diffusion are given by:
central coefficient
a P aW aE Fe Fw
and
aW
aE
5
The scheme has proved to be useful in practical flow calculations (Fluent V4.22)
8. Higher order differencing schemes for convection-diffusion problems
Higher order schemes involve more neighbour points and reduce the
discretisation errors by bringing in a wider influence.
Quadratic upwind differrencing scheme: the QUICK scheme
The quadratic upstream interpolation for convective kinetics (QUICK) scheme
of Leonard (1979) uses a three-point upstream-weighted quadratic
interpolation for cell face values.
The face value of is obtained from a quadratic function passing through two
bracketing nodes (on each side of the face) and a node on the upstream side:
The cell face between two bracketing nodes i and i -1 and upstream node i
6
8
3
8
1
8
-2: face i 1 i i 2
When uw > 0 and ue > 0, a quadratic fit through WW, A and P is used to
evaluate w, and a further quadratic fit through W, P and E to calculate e:
w
6
3
1
W P WW
8
8
8
20
6
3
1
P E W
8
8
8
3
1
6
Fe 8 P 8 E 8 W
3
1
6
W P WW De E P Dw P W
8
8
8
Fw
re-arranging:
3
6
6
1
3
1
Dw 8 Fw De 8 Fe P Dw 8 Fw 8 Fe W De 8 Fe E 8 FwWW
a P P aW W a E E aWW WW
with
aw
6
1
Dw 8 Fw 8 Fe
aE
3
De 8 Fe
aEE
aP
aW a E aWW Fe Fw
1
Fw
8
6 1
Dw 8 w Fw 8 e Fe
3 1 F
8 w w
aWW
1
w Fw
8
aE
aEE
3 6
De 8 e Fe 8 1 e Fe
1 1 F
8 w w
1
1 e Fe
8
where
w 1 for Fw 0 and e 1 for Fe 0
w 0 for Fw 0 and e 0 for Fe 0
21
Example: using the QUICK scheme solve the problem for u = 0.2 m/s on a 5-point
grid.
u = 0.2 m/s, F = Fe = Fw = u = 0.2, D = De = Dw =/x = 0.1/0.2 = 0.5, Pew =
Pee = F/D = 0.4.
the discretisation equation with the QUICK scheme at internal nodes 3 and 4 is
given by the previous table.
Nodes 1, 2 and 5 are all affected by the domain boundaries. Leonard (1979)
suggests a linear extrapolation to create a "mirror" node at a distance x/2 to the
west of the physical boundary.
At node 1
e at the east face of control volume 1:
6
3
1
6
3
1
7
3
2
e P E 0 P E 2 A P P E A
8
8
8
8
8
8
8
8
8
DA
9 P 8 A E
3
Fe
At node 5
The diffusive flux through the east boundary is:
DB
8 B 9 P W
3
W P WW B 8 B 9 P W Dw P W
8
8
3
8
FB B Fw
At node 2
The discretised equation at node 2:
22
3
1
3
2
6
7
P E W Fw W P A De E P Dw P W
8
8
8
8
8
8
Fe
a P P aW W a E E aWW WW S u
aWW
aW
aE
Su
1
3 8
2
2
8
De 3 D A 8 Fe 3 D A 8 Fe FA 3 D A 8 Fe FA A
3
7
1
Dw 8 Fw 8 Fe De 8 Fe
1
6
1
Fw Dw DB Fw
3
8
8
Comparison
SP
1
Fw
4
8
DB FB
3
1
Fw A
4
3 DB FB B
23
Boundedness:
If the flow field satisfies continuity, the coefficient aP equals the sum of all
neighbour coefficient which is desirable for bounded.
The main coefficients (E and W) are not guaranteed to be positive and the
coefficients aWW and aEE are negative. This give rise to stability problems and
unbounded solutions under certain flow conditions.
Since the discretised equations involve not only immediate-neighbour nodes
but also nodes further away. Tri-diagonal matrix solution (TDMA) methods are
not directly applicable.
The QUICK scheme is conditionally stable:
The Hayase et al. (1990) QUICK scheme:
w W
1
3 P 2W WW
8
for Fw > 0
e P
1
3 E 2 P W
8
for Fe > 0
w P
1
3W 2 P E
8
for Fw < 0
e E
1
3P 2E EE
8
for Fe < 0
aE
Dw w Fw De 1 e Fe
1
3 P 2W WW w Fw 1 W 2 P 3 E e FE
8
8
1
1
3W 2 P E 1 w Fw 2 E EE 3 P 1 e Fe
8
8
where
w 1 for Fw 0 and e 1 for Fe 0
w 0 for Fw 0 and e 0 for Fe 0
9. Summary
The crucial issue of discrettising the convection-diffusion equation is the
24
25