27.11.2012 Views

FLOW AROUND A CYLINDER - istiarto

FLOW AROUND A CYLINDER - istiarto

FLOW AROUND A CYLINDER - istiarto

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

– 4.14 –<br />

in which the summation extends over the six cell faces and the non-linear terms are<br />

linearized as in the evaluation of the convection term. The diffusion across the east face is<br />

elaborated and a similar approach applies for the other faces.<br />

In evaluating the diffusive term across the east face, it is convenient to use a local<br />

coordinate system attached to the east face as shown in Fig. 4.4. Across the east face,<br />

Eq. 4.29 reads:<br />

F<br />

��<br />

D<br />

n,��1<br />

n�n�1<br />

n�n�1<br />

n,������� ��<br />

� � � � �� �S<br />

e � � � �e e ���n��<br />

e<br />

n<br />

�e n � S�e<br />

Fig. 4.4 Evaluation of the diffusion terms across the east face<br />

(4.30)<br />

The evaluation of the normal gradient presents some difficulties for its y and z<br />

components. Besides the variable at the neighbor cell E, additional ones at NE, SE, S, and<br />

N might have to be taken into consideration. This would increase the number of<br />

unknowns. To overcome this problem, the so called deferred-correction approach<br />

(Ferziger and Peric, 1997) is selected in the present model, where only the immediate<br />

neighbor cell needs to be considered. In this approach the normal gradient term is<br />

evaluated implicitly by a simple approximation and a correction is added. The correction<br />

is taken as the difference between the correct and approximate gradients; both are<br />

explicitly obtained from the previous iteration. This correction is put in the source terms<br />

at the right-hand side. The diffusion term evaluated with this approach reads (Ferziger<br />

and Peric, 1997, pp. 218-222):<br />

F D n�n�1 � � � � �<br />

e<br />

e Se<br />

��<br />

������<br />

��<br />

��<br />

����<br />

�� e<br />

��1<br />

�<br />

������<br />

�<br />

���n��<br />

e<br />

��<br />

��<br />

�<br />

�� ��<br />

��<br />

��<br />

��<br />

��<br />

����<br />

��<br />

��<br />

��<br />

��<br />

��<br />

e<br />

�� ������ �� �� �������� �� ��<br />

�<br />

� �e Se<br />

correction, explicit<br />

(4.31)<br />

In the above expression, the time index n is omitted for simplicity and a term without any<br />

index refers to the initial solution of the time step n � n �1 (for example Se is constant

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!