Você está na página 1de 12

School of Mechanical and Manufacturing Engineering

Dublin City University


Assignment B: Forced Convective Cooling of Flat Plate

Kieran Yeow
11360006

Table of Contents
Analysis based on empirical correlations............................................................... 3
Computational Fluid Dynamics Analysis................................................................ 5
Figure 1: initial conditions................................................................................... 7
Figure2: convergence occurring......................................................................... 7
Figure 3: convergence shown at 3 separate time steps ...................................... 8
Figure 4: Heat transfer coefficient for the whole plate........................................ 8
Figure 5: skin friction coefficient for the whole plate .......................................... 9
Figure 6: Values obtained from computational results for skin friction coefficient
........................................................................................................................... 9
Figure 7: values obtained from computational results for heat transfer
coefficient........................................................................................................... 9

Analysis based on empirical correlations


Properties of air:
= 22.02 106 2 /,

Air at = 1
= ( T + T s )/ 2 = 360

= 0.0308 /,
= 0.698,

u =100m/s

= 0.995/3 ,

R e cr =5 10 5

= 1.009 / ,

T s =150C=423.15K

Material = 5.2 / ,

T =25C=298.15K

= 320 /

area700=0.7m

= 2300/3

area750=0.75m

a. The first step to finding the average coefficient of heat

transfer is to find the Reynolds number for 700mm and 750


u L

mm using R el = v

R e700 =

100 0.7
6
.00002202 =3.18 10

R e750 =

100 0.75
10 6
.00002202 =3.4

The next step is to calculate where the flow changes from laminar to
turbulent using
v

xcr = u R e cr

xcr =

. 00002202
( 5 10 5 ) =.01101 m
100

Knowing the transition of flow occurs at .01101m use the following


Nusselts number formula for both 700 and 750 using

N u L = 0.037

R e4
P r1
871
5
3

because the flow has both laminar and turbulent.

6 4

N u700 = 0.037

N u750 = 0.037

1
(( 3.18 10 )
.698
)871
5
3

1
(( 3.4 106 )4
. 698
)871
5
3

=4451.07

=4747.5

The next step is to calculate the coefficient of heat transfer for both
700mm and 750mm using h L=

Next

the

heat

N uL k
L

h700 =

4451.07 0.0308
=195.85 W/m2K
0.7

h750 =

4747.5 0.0308
=194.96 W/m2K
0.75

rate

can be calculated for 700m


q L =h L A l ( T sT ) where T s and T must be in kelvin

and

750mm

using

q700 =( 195.85 0.7 )( 423.15298.15 ) =17136.63W


q750 =( 194.96 0.75 )( 423.15298.15) =18277.86W
Next to find the q value for the module you take away the q values from each
other( q750 - q700 ) to get the average q

q a v e =18277.86-17136.63=1141.230746W

Finally the last step to find the average coefficient of heat transfer for the
q

l
module is to rearrange q L =h L A l ( T sT ) to get h on one side leaving ( A ( T T ) )
l
s

where this time h L is the average heat coefficient of heat transfer for the

module and using q L as 1141.23 and A l as .05(area of the module).


1141.23
=182.5969194W
. 05( 423.15298.15 )

To get the average coefficient for the whole plate the same procedure is followed
2

except using L as 1m instead of 700mm or 750mm and also the area is also 1 m .
The calculations are easier as its over the whole plate not just a section,
therefore not needing to find the average of each part. The following values were
calculated like above to get the average coefficient of heat transfer for the whole
plate.

R e 1000 =4.54 10 6 ,

N u1000 =6176.011

h1000 =190.2211

W/m2K

q 1000 =23777.64W
b. The first step to calculating power generation per unit volume of
the element, [/ m ] is to find the volume( lwh ) where
3

L=.05m ( l en d l 700 ),w=1m and h=.01m


Volume=.051.01=.0005 m

The final step to finding power generation is dividing q a v e from question 1 by

the volume
Power=

1141.230746
=2282461.493 / m3
0.0005

c. To find the average coefficient of skin friction C f


u L

number must be found first using R el = v

the module so l=725mm


R e725 =

u L
v

,Reynolds

at the mid point of

R e725 =

100 0.725
10 6
. 00002202 =3.29

Now the average coefficient of skin friction for the module can be found using
C f ,l =0.0592 R e

1
5

1
5

C f ,725 =0.0592 (( 3.29 10 )

) =.002943185

To calculate the average coefficient of skin friction for the whole plate so l=1m
use mixed average as its both laminar and turbulent.

C f , L =0.074 R e 1742 R e L
1

L5

1
5

C f ,1 =0.074( ( 4.54 10 )

)1742(( 4.54 10 ) ) =0.003066222


6 1

d. To calculate the boundary layer thickness at the midpoint of the


module and the whole plate use the formula

=0.37 x R e

x5

where

x is 725mm and 1m respectively.


1

725=0.37 ( 0.725 ) ( 3.29 10 6 ) 5 =.013336307m

1=0.37 ( 1 ) ( 4.54 10 6 ) 5 =0.01724905m

Computational Fluid Dynamics Analysis


1.
a. The flow for the model is mainly turbulent and this was
checked by calculating where the flow changes from
v
laminar to turbulent using xcr = u R e cr

xcr =

. 00002202
5
( 5 10 ) =.01101 m
100

as the plate is 1m long and it changes from laminar to turbulent at .


01101m this shows its not laminar and also not transitional as it
would need to be change to turbulent at the middle of the plate not
the very start.
6

b. The velocity inlet boundary condition defines an inflow


condition based on the flow velocity, this boundary
condition usually is for incompressible flows and is used to
find the velocity of the flow at the inlet. The flow inlet
velocity can be defined by the velocity value, inlet flow
rate or the free stream. When using the velocity inlet
condition at least one pressure boundary condition must
be specified for stability. Translational wall conditions are
used when a linear translational motion occurs like in the
question as the air is flowing over a rectangle plate this is
why its suitable for a translating wall condition. A wall or
no slip boundary condition is used for velocity components
at the wall its also the most common boundary condition.
The thermal condition is used for heat transfer
calculations when solving the energy equation and is at
flow inlets and exits. Temperature is used for thermal
condition and is 423.15K at the bottom. The velocity
condition for the bottom wall is a stationary wall as the
wall is fixed. A velocity of 0 means the wall is stationary.
c. From assignment 1 the residuals should be 1e-06 as they
have a better accuracy than 1e-03 which was also tried in
assignment 1.1e-03 took longer to converge.
d. The fluid is air and the density remains constant as it is
incompressible. To check this the mach number must be
less than 0.3.using the mach number formula m=u/c
where u is speed of fluid and c is speed of sound and in
this question the speed of the fluid is 100m/s and speed of
sound is 340.29m/s so m=.293

To find if natural convection is significant check Richardsons number .if


its <<1 then free convection effects can be neglected, if its >>1 then
forced convection effects can be neglected, if its =1 then free and
forced convection effects have to be considered. The formula for
g LT

Richardsons number is R i= U
0
T f=

T s +T
where = T and T f = 2
f

423.15+ 298.15
=360.5 K
2

1
Tf

=0.00277

R i=

.00277 9.81 1 125


100 2

=.00034

As this is <<1 natural convection can be neglected. Gravity may also be


neglected as the relative simple geometry of the shape doesnt have any dips
which gravity would affect.
e.

Figure 1: initial conditions

The initial conditions are shown above in figure 1 the temperature affects the
convergence along with the turbulent kinetic energy and turbulent dissipation
rate which must be either both 1 or 0. The reason why the x velocity is 0 is
because when the empirical value is calculated you assume it starts at 0 and not
100.if you used 100 it will not converge as it should. However any value from 098 it will converge
2.

Figure2: convergence occurring

From the picture above the convergence can clearly be seen. The straight line
shows convergence and this occurs roughly after 1400 when zoomed in on the
graph. An initial 100 was used then 1000 and for both a straight line could not be
seen so a higher time step of 2000 was used to clearly display the convergence
(straight line).
Taking time step of 1800, 1900 and 2000 as shown below they all converge

Figure 3: convergence shown at 3 separate time steps

3.

Figure 4: Heat transfer coefficient for the whole plate

Figure 5: skin friction coefficient for the whole plate

3.

10

Figure 6: Values obtained from computational results for skin friction


coefficient

a. The average coefficient for skin friction for the whole plate from
fluent is .0037006643 and the theoretical average coefficient for
skin friction worked out in q3 turned out to be .003066222,
working out the % difference it turns out to be 10.9%.
b. The average coefficient for skin friction for x=l+b/2 which is
0.725m from fluent is .0034007877 and the theoretical average
coefficient for skin friction worked out in q3 turned out to be .
002943185, working out the % difference it turns out to be
15.5%.

Figure 7: values obtained from computational results for heat transfer


coefficient

c. The average coefficient of surface heat transfer for x=l+b/2


which is 0.725m from fluent is 184.92014 and the theoretical
average coefficient of surface heat transfer worked out in q1
turned out to be 182.5969194, working out the % difference it
turns out to be 1.27%.
d. The average coefficient of surface heat transfer for the whole
plate from fluent is 208.47212 and the theoretical average
coefficient of surface heat transfer worked out in q1 turned out
to be 190.2211, working out the % difference it turns out to be
2.9%.
4. The correlation results are the most accurate but could be off due to
assumptions of the properties of air and to the formulas being used for each
calculation however the empirical results uses less assumptions and therefore
the answer will be more accurate than the computational results. The
computational results are accurate however could be off due to wrong values put
into the fluent or choosing the wrong values when initializing the solution. Any
wrong value for these initial conditions will results in convergence being
different. Using k epsilon in the computational results presumes the whole plate
is turbulent however in fact we know that the start of the plate to .01101m

(taken from question 1a) is laminar. This is why the empirical value
calculated is a better accuracy than the computational.
From the computational results for surface heat transfer coefficient at 0.725m
the answer was 184.92014 and the answer from the correlations was
182.5969194 this shows a % difference of 1.27% which is very small. This was
11

used at .725m (average of the module) because it was a more accurate result
than over the whole plate.
For the skin friction coefficient for the whole plate the answer from the
computational results is 0.0037006643 and the results from the correlations
were 0.003066222 which gives a % difference of 10.9%. This was for the whole
plate because it was a more accurate result than over the average module for
skin friction.

12

Você também pode gostar