This article was downloaded by: [North Carolina State University]

On: 02 October 2012, At: 09:30

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Computational Fluid Dynamics

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/gcfd20

FINITE ANALYTIC NUMERICAL SOLUTION OF NAVIER-

STOKES EQUATIONS
a a
D. XU & G. WU
a
Power Engineering Department, Nanjing Aeronautical Institute, Nanjing Jiangsu, 210016, The
P. R. China

Version of record first published: 27 Feb 2007.

To cite this article: D. XU & G. WU (1994): FINITE ANALYTIC NUMERICAL SOLUTION OF NAVIER-STOKES EQUATIONS, International
Journal of Computational Fluid Dynamics, 2:3-4, 243-252

To link to this article: http://dx.doi.org/10.1080/10618569408904491

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone
is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
 Camp. Fluid Dyn., 1994, Vol. 2, p. 243-252 © 1994 Gordon and Breach Science Publishers, S.A.
Reprints available directly from the publisher Printed in the United States of America
Photocopying permitted by license only

FINITE ANALyrrIC NUMERICAL SOLUTION OF

NAVIEJR-STOKES EQUATIONS

D. xu and G. WU
Power Engineering Department, Nanjing Aeronautical Institute
Nanjing Jiangsu, 210016, The P. R. China
Downloaded by [North Carolina State University] at 09:30 02 October 2012

(Received 15 November 1992; in final form 18 June 1993)

SUMMARY

The finite analytic method is used in the present study to calculate the turbulent flow field described with
Navier-Stokes equation in body-fitted curvilinear coordinate system The finite analytic method invokes
the analytic solution of governing partial differential equation in formulating the algebraic equation that
relates a nodal value in an element to its neighbour nodal values according to the direction and the
magnitude of convection. It is shown that the finite analytic method has good numerical stability and
accuracy. The turbulent flow fields through a single and a tandem cascades of airfoil are numerically
simulated by using finite analytic method respectively in this paper. The k - e turbulence model and wall
function are employed in the present study. The agreement of numerical solution with experiment result
is quite good.

1. INTRODUCTION

The accurate prediction of the turbulent flow field inside a cascade of airfoil is very
important for both scientific and engineering purposes. The excellent design and
development of turbomachine depend on the knowledge of flow field in the cascade
of arifoil. There are two ways to get such knowledge: the experiment and
calculation. As so far, detailed and accurate measurement of real flow field of a
cascade is impossible. The flow field in a cascade is typical turbulent flow with
disturbance of wall and now separation, therefore one can not calculate the
Navier-Stokes equations directly nowaday. So, in the present study, the averaged
Navier--Stokes equations and the k - E: turbulence model have been used. And a
newly developed method, finite analytic method, is adopted in this paper.
The basic idea of the finite analytic method 1,2 is the incorporation of local
analytic solutions in the numerical solution of partial differential equations. In the
finite analytic method, the whole region of the problem is divided into the small
elements in which the governing equation is locally linearized and solved
analytically. An algebraic equation which approximates the governing equation is
then obtained for numerical calculation. The feature of the finite analytic method
is that it remains the non-linear character of governing partial differential equation
and it has ability to simulate automatically the upwind influence of neighbour nodal
values according to the direction and magnitude of convection. So it needs not use
243
 D. X U AND G . W U

upwind scheme as finite differential and control volume numerical methods d o

when Reynolds number is very large. .
The 2-dimensional turbulent flow through a single and a tandem cascades of
airfoils with Reynolds number more than 2 x 10' are numerically calculated by
using finite analytic method in body-fitted curvilinear coordinate system. Compared
with experimental results, the numerical solution has good accuracy.
The k - f turbulence model and wall function are adopted in the present study.
The numerical tests show that these methods are still useful for turbulent flow
calculation in curvilinear coordinate system.
Downloaded by [North Carolina State University] at 09:30 02 October 2012

2. FINITE ANALYTIC FORMULATION O F GOVERNING EQUATION

For steady-mean turbulent flow, the time-averaged continuity, Navier-Stokes and

turbulence equations are expressed as follows

Table I

equation
continuity

x-momentum

y-momentum

turbulent kinetic energy k

"+
- Y,
G -c
Uk

turbulent kinetic Y + Y, c cZ
energy dissipation rate c
- cl G - - c 2 -
uc k k

where

Normalizing governing equations and translating them from x-y coordinate system
onto calculating plane by introducing new independent variables I and q, where
5 = [(x, y); q = q(x, y), we can rewrite governing equation in general curvilinear
 FINITE ANALYTIC SOLUTION 245

coordinate system as [31

where the coefficients A, B, C and D are generated by coordinate transferring, g

is regarded as source term.

H = Ref,
Downloaded by [North Carolina State University] at 09:30 02 October 2012

The governing equation (2) should be linearized firstly and then be solved on grid
element shown in Figure 1. Let

U=U,+U', V=Vp+V1 (3)

 D. XU A N D G . W U

(a) (b)
Downloaded by [North Carolina State University] at 09:30 02 October 2012

Figure I Grid element (a) physical plane (b) calculating plane.

where Up and V, are velocity components at node P, and U ' and V' are the small
deviation of velocity at neighbour nodes from center node P. They are quite small
in general if the element is reasonably small. Introducing equation (3) into
equation (2). we get

where

For convenience, we still use A, B, C, D, instead of A,, B,, C,, D,. Therefore
equation (4) can still be written as follows

If A, B, C, D and G are supposed to be constants, the source term G can be

removed by introducing

Introducing equation (6) into equation (S), we get

In order to solve equation (7). one must specify boundary conditions firstly. An
exponential and a linear functions are employed to approximate boundary
condition for the local element, because these functions are also the solutions of
equation (7). For example, the northern boundary function of an element as shown
in Figure 1 can be expressed as
 FINITE ANALYTIC SOLUTION 247

Here, coefficients a, b, c are determined by three nodal values in the NW, NC, NE

aN = (&NW+ &NE - ~ & N c )[/4 s i n h 2 ( ~ h / C ) ]

b~ = [QNE- &NW - (&NW+ &NE - 2 & ~ ~ ) ~ 0 t h ( A h / C ) ] / ( 2 h )

CN = &NC

The first derivative term in equation (7) can be removed, if we suppose

Downloaded by [North Carolina State University] at 09:30 02 October 2012

&(€, 7) = WE, v)exp(AE/C + &ID) (9)

Introducing equation (9) into equation (7), we get

= ( A ~ / c +B ~ / D )w
CWEF+DWqq

The boundary condition (8) will become as follows

Equation (10) with its boundary condition (11) can be solved analytically by the
method of separation of variables. Problem (lo) can be divided into four small
problems with simple corresponding boundary conditions.

Here, we still use northern as example

The other three problems are similar. From method of separation of variables,
introducing

into equation (13). we get

CX" Y + DXY" =(A2/c + BZ/D)XY

 D. XU A N D G. W U

Solving equation (15) with corresponding boundary conditions, we will get the
analytic solution of equation (13). Using similar method, we can solve other three
problems. Finally we can get analytic solution of original governing equation. Its
center node value is expressed as follows
Downloaded by [North Carolina State University] at 09:30 02 October 2012

3. NUMERICAL SOLUTION O F GOVERNING EQUATIONS

3.1. Pressure Correction Equation
The main difficulty for solving governing equation with original variables, that are
u, v and p, is incorporation of velocity with pressure. In the present study, pressure
correction method is used to treat this problem. The basic idea of pressure
correction is that momentum equations are solved t o find a pseudo-velocity field u*
and v* under assumed pseudo-pressure field p*, then pseudo-velocity is introduced
into continuity equation t o get a pressure correction equation

By solving equation (17), we get pressure correction field p ' , then we can correct
both velocity and pressure fields.

3.2. Wall Function

The wall function is adopted to determine the dependent variables (u, v, k and c )
near wall in this paper. The purpose of introducing the wall function is to minimize
grid number without reducing numerical accuracy a t the least C P U time and to deal
with boundary layer near the wall where the k - c turbulent model is unadoptable.

3.3. Numerical Procedure

The sequence of numerical operation can be summarized as follows
 FINITE ANALYTIC SOLUTION 249

A. read in all boundary and other constants about flowing field, calculate grid
parameter, distribute a n initial flowing field.
B. calculate finite analytic coefficients of momentum equation, solve this
algebraic equation by using iterative numerical method t o get pseudo-velocity field.
C. calculate pressure correction equation, get pressure correction field p ' and
velocity correction u' and v' .
D. correct pseudo-velocity and pseudo-pressure fields.
E. calculate turbulent kinetic energy k and turbulent kinetic energy dissipation
rate c, calculate new turbulent viscosity.
F. check convergence criterion, if it is not achieved, go back step B until it is
Downloaded by [North Carolina State University] at 09:30 02 October 2012

achieved.
G , print final results, stop.

T o accelerate convergence and to promote numerical stability, the under relaxation,

source term linearizing and line by line iterative method are used in the present
study.

4. NUMERICAL SOLUTION

The 2-dimensional turbulent flow fields in a curvilinear coordinate system through

both a single and a tandem cascads of airfoil are numerically calculated in the
paper.

4 . 1 . Flow Field through a Single Cascade o f Airfoil

The single cascade of airfoil is shown in Figure2. Its geometrical and

Figure 2 Single cascade

 D. XU AND G . WU
Downloaded by [North Carolina State University] at 09:30 02 October 2012

0.0 0.2 0.4 0.6 0.8 LO rlc

Figure 3 Surface pressure coefficient.

aerodynamical data are obtained from reference [4]. The Reynolds number and
inlet angle of flow are 2.8 x lo5 and 30" respectively.
The flow field is divided by using body-fitted grid generating technique5 with total
grid number 51 x 17. On the inlet boundary, u, v, p, k and c are defined. On the
outlet boundary, the first derivatives of dependent variables with respect to are
assumed to be 0. On the wall of blade, the wall function is used to determine wall
condition. The under relaxation factors for u, v, p, k and E are 0.7, 0.7, 0.3, 0.5,
0.5 respectively. The comparison of calculated and measure surface pressure
coefficients is shown in Figure 3. Here surface pressure coeffient Cp is

Except somewhere at front and rear of blade, the agreement of calculated results
with measured data is very good.

4.2. Flow Field through a Tandem Cascade of Airfoil

The second calculation example is about the flow field through a tandem cascade
of airfoil. The geometrical and aerodynamical data are from reference [3]. The
Reynolds number and incidence angle of inlet flow are 2.0 x lo5 and 0"
respectively. The tandem cascade consists of two rows of single cascade of airfoils
as shown in Figure 4.
The flow field is also divided by using body-fitted technique. The grid number is
69 x 21. Boundary arranging is the same as in single cascade. Figure 5 shows
calculated and measured surface pressure coefficients. Here surface pressure
coefficient is
 FINITE ANALYTIC SOLUTION
Downloaded by [North Carolina State University] at 09:30 02 October 2012

Figure 4 Tandem cascade

Figure 5 Surface pressure coefficient.

Subscript 1 means upstream condition, and 2 downstream. Although some

discrepancies are apparent, the agreement is considered to be good.

5. CONCLUSIONS

The turbulent flow fields through a single and a tandem cascade of airfoils are
numerically studied by using finite analytic method in curvilinear coordinate
system. The two numerical examples indicate that finite analytic method has good
numerical accuracy and stability. The agreement of calculated results with
measured data is good.
 252 D. XU AND G . WU

References

1. Chen, C. J. and Li, P. (1980) The finite analytic method for steady and unsteady heat transfer
problems, ASME Paper 80-HT-86.
2. Chen, C. J . (1981) Finite analytic solution of heat transfer in two-dimensional cavity flow, Numerical
Heat Transfer, Vol. 4, No. 2.
3. Xu. Diao (1988) The analytic and experimental studies of tandem cascade, Ph.D. dissertation,
Nanjing Aeronautical Institute.
4. Felix, A. R. and Emery, J. C. (1957) A comparison of typical national gas turbine establishment and
NACA axial-now compressor blade sections. NACA TN-3937.
5. Thompson, J . F. (1977) Boundary fitted curvilinear coordinate system for solution of partial
differential equations on fields containing any number of arbitrary two-dimensional bodies, NASA
CR-2729.
Downloaded by [North Carolina State University] at 09:30 02 October 2012