Flows With More Than One Dependent Variable - 2D Example Juan M.

Download Report

Transcript Flows With More Than One Dependent Variable - 2D Example Juan M.

Flows With More Than One Dependent Variable - 2D Example

Juan M. Lopez Steven A. Jones BIEN 501 Wednesday, April 4, 2008

Louisiana Tech University Ruston, LA 71272

Recall - Generalized Newtonian

T

p

I

 2   

D

where   2 tr 

D

D

 Recall that:

D

 1 2  

v

  

T

 tr stands for “trace,” which is the sum of the diagonal elements. Tr(

T

)=

T ii

While the expression looks complicated, it will look much simpler once a given form for  is found.

Louisiana Tech University Ruston, LA 71272

Generalized Newtonian

T

p

I

 2   

D

where   2 tr 

D

D

 Recall that:

D

 1 2  

v

  

T

 tr stands for “trace,” which is the sum of the diagonal elements. Tr(

T

)=

T

ii

T

  

P

   

v

   1   0 0 0 1 0 0 0 1              

u

1  

x u

1 2 

x

3 2 

u i

x

 1 

u

  

x u

1 3 

x

2 1 

u

2 

x

1 

u

2 

x

3 2  

u

2 

u

x

1 2 

u

 2 

u

3 

x

2 

u

3  

x u

1 3 

x

2 2   

u

3   

u x u

x

1 3 2 3 

x

3       

Louisiana Tech University Ruston, LA 71272

Parallel Plate Poiseuille Flow

Given: A steady, fully developed, laminar flow of a Newtonian fluid in a rectangular channel of two parallel plates where the width of the channel is much larger than the height, h, between the plates.

• • • • •

Find: The velocity profile and shear stress due to the flow.

Assumptions:

Entrance Effects Neglected No-Slip Condition No vorticity/turbulence

Louisiana Tech University Ruston, LA 71272

Additional and Highlighted Important Assumptions

• The width is very large compared to the height of the plate.

• No entrance or exit effects.

• Fully developed flow.

• THEREFORE… – Velocity can only be dependent on vertical location in the flow (

v x

) – (

v y

) = (

v z

) = 0 – The pressure drop is constant and in the x direction only.

 p 

x

 Constant   p

L

, where

L

is a length in

x

.

Louisiana Tech University Ruston, LA 71272

Boundary Conditions

• No Slip Condition Applies – Therefore, at y = -h/2 and y = +h/2,

v

– z = -w/2 and z = +w/2,

v

width of the channel.

= 0 • The bounding walls in the z direction are often ignored. If we don’t ignore them we also need: = 0, where w is the • For this problem we include this, and make the width finite to make this dependent on two variables.

Louisiana Tech University Ruston, LA 71272

Incompressible Newtonian Stress Tensor

Adapted from Table 3.3 in the text.

τ

           2  

u y

x

u

x z

u

x x

 

u

y x

 

u

z x

     

u y

x

   2  

u

y z

 

u

y x

u y

y

 

u y

z

        

u

x z

u

y z

2    

u

z z

u

z x

u y

z

Now, we cancel terms out based on our assumptions.

This results in our new tensor:        

τ

           0 

u

y x

u

z x

  

Louisiana Tech University Ruston, LA 71272

   

u

y x

  0 0  

u

z x

0 0      

Navier-Stokes Equations

In Vector Form:  

v

t

 

v

 

v

  p    2

v

 

g

Which we expand to component form from table 3.4: x component :    

v x

t

v x

v

x x

v y

v

y x

v z

v x

z

     p 

x

     2 

x v x

2   2 

y v x

2   2 

z v x

2    

g x

( Eq.

3.3.25) y component :    

v y

t

v x

v y

x

v y

v y

y

v z

v y

z

     p 

y

      2

v y

x

2   2

v y

y

2   2

v y

z

2     

g y

z component :    

v z

t

v x

v z

x

v y

v

y z

v z

v

z z

     p 

z

     2 

x v z

2   2 

y v z

2   2 

z v z

2    

g z

Louisiana Tech University Ruston, LA 71272

Reducing Navier-Stokes

x component :    

v

t x

v x

v

x x

v y

v x

y

v z

v

z x

     p 

x

     2 

x v x

2   2

v x

y

2   2 

z v x

2    

g x

y component :    

v y

t

v x

v y

x

v y

v y

y

v z

v y

z

     p 

y

      2

v

x

2

y

  2

v y

y

2   2

v y

z

2     

g y

z component :    

v

t z

v x

v

x z

v y

v

y z

v z

v

z z

     p 

z

     2 

x v z

2   2 

y v z

2   2 

z v z

2    

g z

Louisiana Tech University Ruston, LA 71272

Reducing Navier-Stokes

• This reduces to: Modified 0    p 

x

     2 

y v

2

x

  2 

z v

2

x

  • Including our constant pressure drop: 0    p

L

     2

v x

y

2   2

v x

z

2   • Oops! Now we have a nonhomogenous higher-order differential equation that is inseparable. How do we deal with it?

Louisiana Tech University Ruston, LA 71272

DiffEq Assumptions

• We will assume this solution is a combination of simple parallel plate Poiseuille flow plus some perturbation that is dependent on the walls and finite width.

• Extracting the 1D Poiseuille flow, we can rewrite the equation as: 0  where 

v x

 p

L

     

v

V x x

2

v

y

2

x

  2

v

z

2

x

        Therefore : 0    p

L

   

d

2

V x dy

2        

y

2  2   2  

z

2  

Louisiana Tech University Ruston, LA 71272

DiffEq Solution - Setup

• We can separate this into two equations, each of which equals zero.

• Why?

– 0=0+0 0    p

L

   

d

2

V x dy

2        

y

2  2   2  

z

2   Separated : 0    p

L

   

d

2

V x dy

2   0      

y

2  2   2  

z

2  

Louisiana Tech University Ruston, LA 71272

DiffEq Solution - Poiseuille

 p

L

   

d

2

V x dy

2   because this is the simple Poiseuille solution, we can see from Eq.

2.7.18

that the above equation is equivalent to :  p

L

   

d

2

V x dy

2     

d

yx dy

 

Louisiana Tech University Ruston, LA 71272

DiffEq Solution - Poiseuille

From our stress tensor definition , 

yx

   

du x dy

  Therefore, we can follow the solution from Section 2.7.2

to end up with :

u x

  p

h

2 8 

L

  1  4

y

2

h

2   Now we can focus our remaining efforts on the perturbation function.

Louisiana Tech University Ruston, LA 71272

Perturbation Function - Reduction

0       2  

y

2   2  

z

2    Where 

y

,

z Y y Z z

Therefore 0 0   0       2

Y

y

2   2

Y

z

2      

Z

  

Y

1     2

Y

2

Y

y

1  

Louisiana Tech University Ruston, LA 71272

• We can approach this perturbation function by a separation of variables method, as it is homogeneous.

   

Y y

1        

Z

 2

Y

y

2

y Y y

2

Z

z

2

z

  

Perturbation Function - Separation

• Because each term is independent of the other term, the ONLY way this can be true is if each of the expressions is equal to a constant. Thus we define a constant as follows:  2  

Y

1  2

Y

y

2 

Z

1    2

Z

z

2 We can now use our standard homogeneou s general

Y Z

      solution

A

1

B

1 sin sinh   

y

:    

A

2

B

2 cos   cosh

y

  • Now we can use our boundary conditions to solve for these constants.

Louisiana Tech University Ruston, LA 71272

Perturbation Function – B.C.’s

At y  0, we have a symetric region in our flow (the point of maximum velocity on our parabola) 

dY dy

 0 |

y

 0

dY

A

1  cos 0 

A

2  sin 0

dy

Because cos  1 ,

A

1 must be 0.

At the walls (y   / h/2) :  0

dY

 0  cos  

h

2  

A

2  sin  

h

2   0

dy

To be a nontrivial Therefore, sin  

h

solution, 2   0

A

2 cannot (

error in text

 0 ?) Thus  can only have values of 

n

  2

n

 1  

h

Louisiana Tech University Ruston, LA 71272

Perturbation Function – B.C.’s

For the Z portion of our separated function : 

dZ dz

 0 |

z

 0 Because cosh 

dZ

 1 ,

dz B

1 

B

1 

n

must cosh be 0.

 

B

2 

n

sinh   0

dZ

 

B

2  sinh  0

dz

To be a nontrivial solution,

B

2 cannot  0 We can now combine our equation for Y and our equation for Z to give us  .

   

Y

  

Z

  

n

   1  

A

1 cos  

n

 

B

2 cosh  

n

  Because constants are just constants, they combine  

n

   1 

A n

cosh 

n

n

Louisiana Tech University Ruston, LA 71272

Perturbation Function – B.C.’s

We use our Boundary Conditions one more time to obtain : 

y

,  

y

, 

w

2   

y

, 

w

2    

n

   1 

A n n

   1 

A n n

   1 

A n

cosh cosh  

n

cosh  

n

n

n

w

2  cos  

n

 

u x w

2  cos  

n

   p

h

2 8 

L

  1  4

y

2

h

2   We now have an equation purely in terms of one variable (y).

We can integrate to solve for the coefficien cos  2

m

 1  

y

t

h A n

.

At this .

This point the textbook multiplies makes both sides of the equation both sides of the equation appropriat ely periodic.

by This

n

   1 

A n

solution cosh  

n

is nontrivial

w

2  cos 

n

only when  cos  2

n

n  m, so this can be rewritten  1  

y

h

 p

h

2 8 

L

  1  4

y

2

h

2   cos as :  2

n

 1  

y h

Louisiana Tech University Ruston, LA 71272

Perturbation Function – Integration

We can now

A n

n     0   

h h

/ / 2 2 integrate cosh  

n w

2 :  cos  

h h

/ / 2 2  p

h

2 8 

L

  1  4

y

2

h

2 

n

  cos   cos  2

n

 2

n

  1  

y h

1  

y dy h dy

 Rearrangin

A n

 g, we can re n     0   

h h

/  

h h

/ / 2 2 / 2 2 cosh  p

h

8 

L

2  

n

write   1  2 4 with the cos

h y

2 2    

n

  cos coefficien  2

n

cos  2

n

h

1  

y

h

t isolated 1  

y dy dy

: Which results in :

A n

 1

n

   8 p

h

L

2  2 cosh    

n

2

h

32  2

n

  1  

w

1  3 

Louisiana Tech University

3

Ruston, LA 71272

 

for n

 0 , 1 , 2 ,..., 

DID YOU CATCH THAT?

This is a form of the Fourier Transform.

Express a function as a series of sin and cosine terms, and then you can integrate and

Perturbation Function – Integration

We can plug this into our original

v x

  p

h

2 8 

L

  1  4

y

2

h

2     p

h

2 8 

L

equations n    0 32 1 :

n

 2 cosh

n

 1  3  2  3

n

h

1  

z

 2 cos

n

  1   2

n

w

cosh 2

h

 1  

y h

The textbook covers a way of calculating the shear stress. However, we have the stress tensor, so we can go to this tensor directly to calculate this from our equation

τ

above.

           0 

u

y x

u

z x

      

u

y x

0 0    

u

z x

0 0       You should be able to start spotting the similarities between our velocity equation, above, and the stress tensor on the left.

Louisiana Tech University Ruston, LA 71272

Discussion

• Why would it be useful to run an analysis like this?

– Helps select critical design dimensions for a flow channel.

– If there is a controlling dimension, we can design a workaround.

• Where else do you think they run this type of analysis in engineering?

Louisiana Tech University Ruston, LA 71272

Announcements

• Office hours today, let me know if you need them • Tutorial lab tonight…will go over more problems and answer questions about the current assignment.

• New assignment to be posted soon.

Louisiana Tech University Ruston, LA 71272

• QUESTIONS?

Louisiana Tech University Ruston, LA 71272