Professional Documents
Culture Documents
NASA CR-178818
A COMPUTER CODE FOR THREE-DIMENSIONAL INCOMPRESSIBLE FLOWS USING NONORTHOGONAL BODY-FITTED COORDINATE SYSTEMS
By Y. S. Universities
Dynamics
Laboratory
Final
Report
March
1986 [NASA-CR-178818) THREE-DIMENSICNAL NONOR_HOGONAL Final Report Association) A C_POTEE CODE FOR INCCMPRESSIEL FLOWS OSING BCD-FI_ED COC_DINATE SYZI_MS [Uni.ersities 5_ac 86 _ HC A05/M_ A01 Research CSCL 20D G3/34
N86-26544
Unclas 42915
Prepared
REPORT 3, RECIPIENT'S
STANDARD CATALOG
TITLE NO.
PAGE
NASA
4. TITLE
CR-178818
AND SUBTITLE
1
5, REPORT DATE
*Y.
9.
S. Chen
ORGANIZATION NAME AND ADDRESS 10, WORK UNIT. NO.
PERFORMING
George Marshall
2. SPONSORING
35812
Report
Space
,._. SPONSORINC AGENCYCODE Space Division, Research Systems Association Dynamics Laboratory, Science
Sciences
In this report, a numerical method for solving the equations of motion of threedimensional incompressible flows in nonorthogonal body- fitted coordinate (BFC) systems has been developed. The equations of motion are transformed to a generalized curvilinear coordinate system from which the transformed equations are discretized using finite difference approximations in the transformed domain. The hybrid scheme is used to approximate the convection terms in the governing equations. Solutions of the finite difference equations are obtained iteratively by using a pressure-velocity correction algorithm (SIMPLE-C). Numerical examples of twoand three-dimensional, laminar and turbulent flow problems are employed to evaluate the accuracy and efficiency of the present computer code. The user's guide and computer program listing of the present code are also included.
7.
KE'/
WORDS
18.
DISTRIBUTION
STATEMENT
N avieI
- Stoke
s Solver Coordinate
Unclassified-Unlimited
Code
19.
SECURITY
CLASSIF.
(o thl=
report_,
SECURITY
CLASSIF.
(of
rid=
I_ie)
Unclassified
MSFC - Form 3292 (May 1969)
Unclassified
For sale by National Technical Information
NO.
OF
87PAGES
]22.
NTIS PRICE
Virginia 22151
Service,
Springfield,
TABLE
OF
CONTENTS
Page INTRODUCTION TRANSFORMATION DISCRETIZATION SOLUTION NUMERICAL A. B. C. D. CONCLUSIONS REFERENCES APPENDIX APPENDIX APPENDIX A. B. C. ............................................................. OF OF THE THE EQUATIONS EQUATIONS OF OF MOTION MOTION ........................ .......................... 1 2 5 7 12 12 15 18 18 23 24 AND USER'S GUIDE ......... 27 31 41
PROCEDURES EXAMPLES
.................................................... ......................................................
2-D Laminar Driven Square-Cavity Flows ......................... 2-D Laminar Flows Over a Backward-Facing Step ................. 2-D Turbulent Flows Over a Backward-Facing Step ............... Developing Laminar Flow Inside a 90-Deg-Bend Square Duct .............................................................. ................................................................ COMPUTER LIST OF CODE FORTRAN LISTING STRUCTURES SYMBOLS
......
................................
PROGRAM
..........................................
III
,oo
LIST
OF
ILLUSTRATIONS
Page
Locations
where
the
variables the
are
3.
where pressure
11
Physical geometry and wall boundary conditions for laminar flows inside a wall-driven square cavity ............................ Mesh Velocity Comparisons the square
1
12 13 14
systems vector
used plots
for
driven
cavity
problem
.......................
6. 7.
............................................... mid-section of
of velocity profiles along the cavity .................................................. history for the driven cavity
Convergence
9.
i0.
17 17
over
18
19
14.
turbulent flow over a backward-facing ratio ....................................... energy of turbulent 2:3 expansion ratio flow over .................... duct
19
15.
Contours of turbulent kinetic a backward-facing step with Geometry developing Primary square Secondary Primary and mesh laminar
19
16.
20
17.
velocity patterns of laminar duct ........................................................ velocity velocity patterns profiles for of a laminar 3-D
a 90-deg-bend 21
flow
inside
a square
...
22 23
90-deg-bend
iv
NOMENCLATURE
Symbol A A link link link link link link coefficient coefficient coefficient coefficient coefficient coefficient heat of of for for for for constant model model model coefficient energy of the kinetic conductivity in the term added term term term term of of of the the the u-equation v-equation w-equation per fluid for per metric finite the the the the time at difference node at
the
Au Av Aw Ao Cv C 1 C2 C
B
u-equation v-equation w-equation marching constant = 1.44 = 1.92 = 0.09 the pressure (jour/kg) correction equation scheme volume
specific turbulence turbulence turbulence diffusion internal Jacobian turbulence thermal pressure
D
e
mass
J k k' P P r
energy of
(N/m the
production energy source source source source temperature time velocity (sec) in
turbulent volume
Q
S Su Sv Sw T t
U
unit
(K)
direction
in in
X Y Z
Greek I"
diffusion turbulent difference vsriable solution turbulence turbulence curvilinear curvilinear molecular
coefficient kinetic operator of at general the model model transport time equation level = 1.0 = 1.3 energy dissipation rate (m2/s 3)
A _o k
g E
coordinate coordinate viscosity eddy viscosity (kg/m over 3) all values around a grid node P (N-S/m viscosity 2) (N-S/m 2)
;t ;eff
P Z
(N - S/m 2)
i
n
coordinate
value quantity
vi
Superscript
0
previous current
time solution
level
solution
correction
quantity
vii
CONTRACTOR REPORT A COMPUTER CODE FOR THREE-DIMENSIONAL INCOMPRESSIBLE FLOWS USING NONORTHOGONALBODY-FITTED COORDINATE SYSTEMS INTRODUCTION With the currently increasing computer capability and various flow solvers developed, numerical simulations of three-dimensional incompressible flow problems using Reynolds-average Navier-Stokes equations are now becoming more feasible in many engineering design and analysis applications. In many real world flow problems, the boundary geometries are complex such that it is more accurate to describe the geometries using body-fitted coordinate (BFC) systems. Especially for internal flow problems with complex geometries such as those of the hot gas manifold (HGM) of the Space Shuttle Main Engine (SSME), the use of nonorthogonal BFC systems for numerical solutions can be beneficial in many aspects. It is not only the boundary geometries that can be represented more closely using BFC systems, but also grid-refined solutions can be obtained without increasing an excessive amount in computer memory. In addition, once a particular flow problem has been set up, the redesign or optimization process of the boundary shapes can be performed very easily using BFC systems. Several numerical methods [ 1,2, 3,4,5,6] has been developed for solving the incompressible Navier-Stokes equations in 3-D BFC systems. The main difference between these methods lies in the way of finding a pressure field such that the flowfield can be as close to divergence-free as possible (i.e. to satisfy the mass conservation equation). This is the main feature and difficulty of solving the incompressible flow problems. Numerical methods of References 1, 2 and 3, for instance, have employed the pseudocompressibility approach and time-iterative scheme to generate the pressure field so that the continuity equation is satisfied when a steady state solution is reached. In these methods, artificial smoothing techniques must be used to obtain a strong coupling between the velocity and pressure fields. Methods of References 4, 5 and 6, on the other hand, have utilized a successive pressurevelocity correction scheme by using a Poisson's equation for pressure correction derived approximately form the continuity and momentum equations. For these latter methods, grid staggering between the v_locity vectors and the pressure nodes must be used to ensure stability of the numerical solutions. There are several possible methods of grid staggering associated with different features in solving the pressure correction equation. These grid staggering methods were discussed in Reference 6, from which one of the methods was shown to be the most promising arrangement (i.e. with the velocity vectors located at the faces of a volume which contains the pressure and other scalars at its center). But, this method has one drawback, that the velocity components are solved using different control volumes. It is for this reason that a grid staggering system similar to the one used by Vanka et al. [4] is developed in the present study. The present method of grid staggering and pressure correction equation that was described by Vanka [4[ and Maliska [6]. Also, using the present method, the same control volume is used for the velocity components and scalar quantities.
In the following sections, basic elements for establishing the present computer code for solving the curvilinear Navier-Stokes equations in three-dimensional space (CNS3D) will be described. These are followed by a series of standard numerical examples used to evaluate the accuracy and efficiency of the present numerical method. The numerical exampres include laminar flow driven-cavity problem, cases of laminar and turbulent flows over backward-facing steps, and 3-D laminar flows inside a 90-deg-bend square duct. Applications of the present code to the internal flow problems of SSME will be included in future publications. A user's guide Appendix B contains the computer program to the present CNS3D code is provided in Appendix a list and definitions of all the major fortran symbols which is listed in Appendix C. A. used
in
TRANSFORMATION
OF
THE
EQUATIONS
OF
MOTION
For equations
Newtonian as:
fluid,
the
continuity,
momentum
and
energy
Ut
+ Ex
+ Fy
+ G z
= S
(1)
where
(x,y,z)
represent
the
Cartesian
coordinates,
and
ep pu U=_ pv pw pe-Q
E
"pu
_/u x ]_v x
- ]J w - k'T
pw pwuG
pvupv
_Uy
pu
_W
pWV-
_Vy
p WW ]J W z
(PUx)x (UUy)
S
+ (PWx)z
- Px Py z + Uy) 2 + (w y )2 + (u + Wx )2
(_Uz)x u[2(Ux
2 + Vy 2 + w z 2) + (Vx
VZ
_ 2 3 (u x
+ vy
+ Wz )2]
e = the
internal added
energy per
per unit
unit volume
mass
= CvT
for
perfect
gas
conductivity
of the
fluid
_),
to
a general
curvilinear
coordinate
system
(_,
n,
U t + E_
_x
+ En
nx
+ E_
_x + F_
_y
+ Fn
ny
+ F_
_y (2)
+ G_ where
_z + GT] _z
+ G_
_z = S
_x
= J(Yn
- y_
z n)
_y
= -J(x
n z
- x
z n)
_z
= J(x
- x
yn)
nx
= -J(y_
- y_
z_)
my = J(x_
- x
z_)
n z = -J(x_
y_
- x
y_)
_x = J(Y_ Zn 5y =-J(x_ zn
Yn z_) Xn z_)
The transformation coefficients, _x' _y' _z' nx' ny, nz, _x' _y' and _z' are computed numerically using second order central differencing. In the transformed domain, the grid sizes (i.e., A_, An, and A_) are set to be unity. This simplifies the calculation of the transformation coefficients. For turbulent flow computations, the present code has employed the standard k- turbulence model [7] to provide the turbulent eddy viscosity ut" The standard k- turbulence model (which consists of a turbulent kinetic energy equation, kequation, and a turbulent kinetic energy dissipation rate equation, E-equation) is given as :
(Pk)t
puik
_ok eff k x i )
x i = P(Pr
_)
(3)
(PE)t
Puik
Po eff
x i)
x.
1
= P k
(CIPr
- C
(4)
where
the
effective
viscosity
Ueff
is calculated
from:
Peff
= p + ut
= _
+ p C
t.l
k2/E
and
the
turbulent
kinetic
energy
production
term,
Pr'
is
defined
as:
Pr
= Cu
k2 --_ [(Uy
2 + v x) + (v z + Wy.)
2 + (w x + u z)
2 + 2(u x
2 Y
+ w 2)]
Z
The
turbulence
model constants
are:
C C
= 0.09 = 1.44
, '
ok = 1.0 C 2 = 1.92
= 1.3
Also,
the molecular
(i) is replaced
by
In order to save the computational efforts, the widely used wall function approach [8] is employed to provide the near wall boundary conditions for the momentum and energy equations and the k-_ turbulence model. This approach avoids the requirement of integrating the governing equations up to the wall which requires a large number of additional grid points near the wall. Equations (2), (3), and (4) form a closed set of nonlinear partial differential equations governing the fluid motion. This set of equations are to be solved by means of finite difference approximations which are performed in the transformed domain. For treating the convection terms, the hybrid scheme [9] is employed for simplicity (although other more elaborate schemes such as central differencing plus artificial dissipation scheme, QUICK scheme, or skew upwind differencing scheme, etc. can be implemented [10]). These are described in the following sections.
DISCRETIZATION
OF
THE
EQUATIONS
OF
MOTION
In this section, finite difference approximations are used to discretize the governing equations, equations (2), (3), and (4). Second-order central differencing is used for the diffusion terms and the source terms. The hybrid differencing scheme [9] is employed to approximate the convection terms in the governing equations. The finite difference discretizations are performed in the transformed domain. The solution procedure for the discretized equations using a velocity-pressure correction algorithm (SIMPLE-C) of References II and 12 will be described in the next section. The governing equations of motion can transport equation in which denotes all the is the diffusion coefficient.. be represented by dependent variables the following respectively model and
(P)t
+ [pu + [pu
I'(_ r(_
_x _x
+ Cn nx + Cn nx
+ _r + _
_x)]_ _x)]n
_x nx
+ [pu,
r(g
gx _y
+ Cn
nx
+ _
_x)]
Cx _y
+ q_n ny + #_ _ _y) ]
- F(_ F(qb
_y + Cn _Y + _n _z + Cn
ny + _ ny
_Y)]n qy)]_ _z )]
ny _y _z
+ _
+ [pw
r(_
n z + _
+ [pw
- r(
Cz + Cn
n z + _ _z)] n nz
+ [pw
r(_
_z + Cn
n z + _ _z)]
_z = S
(5) 5
Discretization of equation (5) is performed using finite difference approximations in the transformed domain. The second order central differencing is used for approximating the diffusion terms. For the convection terms, the hybrid differencing scheme [9] is employed (i.e., using central differencing for cell Peclet number less than or equal to 2 and switching to upwind differencing when the cell Peclet number is greater than 2). The finite difference equation is arranged by collecting terms according to the grid nodes around a control volume as shown in Figure i. The final expression is given by equation (6) in which A represents the link coefficients between grid nodes P, E, W, N, S, T, B, NE, NW, NT, NB, SE, SW, ST, SB, ET, EB, WT, and WB as shown in Figure i.
Ap
_p = A E
_E + AW
CW
+ AN
_N
+ AS
S + AT
CT
+ AB
CB
+ Sl
(6)
where
S I = S + Ap + ASE
_pO
+ AN E NE _SW
+ ANW + AST
_NT SB
+ ANB
_NB
q_SE + ASW
+ AET
_ET
+ AEB
_EB
+ AWT
_WT
+ AWB
_WB
Ap
= A E + AW + A N + A S + AT
+ A B + Ap
Ap = pp/At
o denotes is employed
at the
the time
previous dependent
A fully problems.
implicit
ET
TJ
WB
Figure
1.
Three-dimensional
grid
structure
and
labeling
around
grid
node
P.
Thus, the nonlinear equations of motion are approximated by a system of linear algebraic equations which have the form of equation (6). Only one program subroutine is designed to calculate the link coefficients and the source terms. The number of algebraic equations depends on the number of interior grid points. For a grid size of 10 x 10 x 10 the number of algebraic equations to be solved would be around 512. This large system of equations are preferred to be solved by some iterative methods, such as Gauss-Seidel iteration, line-underrelaxation method [13] or Stone's method [14], etc., rather than using direct methods such as Gaussian elimination method. Only a few (6 to 10) iterations through the whole computational domain are needed and a complete convergence of the system of algebraic equations is not required. Since equation (6) is only a linearized version of the governing equations which are nonlinear and coupled in nature, solutions of the equations of motion must be obtained through global iterations among the equations. A tentative solution to equation (6) will not affect the final results significantly. On the other hand, if too many iterations are used to get a better solution of equation (6), then a great deal of computing time would be virtually wasted. However, the above argument can not be applied when the pressure correction equation (which will be derived in the next section) is solved. Since during each global iteration it is desirable to retain a divergence-free velocity field, better solution of the pressure correction equation would in effect promote the convergence of the whole numerical scheme. Therefore, more iterations are usually used to solve the pressure correction equation. SOLUTION PROCEDURES The governing equations used in the present analysis are nonlinear and strongly coupled. Iterative procedures are employed to drive the equations to a converged solution. It is particularly important for incompressible flow to make the flow field satisfy the continuity equation and the momentum equations at the same time. This requires a correct pressure field associated with a divergence-free velocity field. A velocity-pressure correction procedure is developed in the present study to drive the pressure field and the velocity field to be divergence free. This kind of procedure requires grid staggering between the velocity components and the locations where the pressure is estimated and stored such that the velocity field and the pressure field will not be uncoupled. In the present study, staggering grid systems as shown in Figure 2 (for 2-D case) are used. The velocity components, u and v, are solved and stored at the grid nodes and the pressure, p, is located at the corners of the control volume of u and v. In this way, solutions of u and v can be solved using the same control volume and coupling between u, v and p can also be enforced. To estimate the pressure field, a pressure correction equation is derived approximately from the discretized momentum and continuity equations. The velocity and pressure fields are then corrected using the solutions of the pressure correction equation. First, as. ApUup * = _
i
A.1 u u.* = PX * + SU 1
(8a)
X X
X
O
u, v AND PRESSURE
SCALARS
X X
Figure
2.
Locations (staggering
are
stored
Ap v
Vp*
= E i
A'v I
vi*
= P
* + S
(8b)
Ap w Wp*
= Z i
A'w I
w.* I
* + SW
(8c)
where u*, v*, w*, and p* represent satisfy the continuity equation the the following relations:
(Sb). To according to
u = u* v = v* w = w*
P = P* + P'
can be w, and
constructed p:
approximately
using
the
Ap u
Up
= E i
A'u 1
ui
- PX + SU
(10a)
Ap v Vp
= __ i
A.1v v. 1 - P y + S v
(lOb)
Ap w Wp
= E i
A'wl w.1 - Pz + Sw
(10c)
By subtracting equations (8a) through (8c) from respectively, the following equations result:
equations
(10a) through
(10c),
Ap u Up' = Z i
A'ul
U-'l
- Px'
(lla)
Ap v Vp' = E i
(llb)
Ap w
Wp' = E i
A'w I
wi' - Pz '
(llc)
According to be:
to SIMPLE-C
algorithm
[11], equations
(11a) through
(Ap u
- E i
AiU)
Up'
= E i
A'ul
(ui'
- Up')
- Px'
(12a)
(Ap v -_
AiV)
Vp' = i
. v (vi' - Vp') i
- P
'
(12b)
(Ap w
E i
Ai w)
Wp' = E i
Ai w
(w i' - Wp')
p z
(12c)
terms the
side
of
equations
(12a)
through
(12c)
are
neglected
'
(13a)
Vp, = _(ApV
_I
A'V)l
PY' =-
Dv
Py'
(13b)
WP' = - IA 'pW
_ i
A.Wl )
P zv = - D w P z
"
(13c)
decompositions as:
of
equations
(9a)
through
(9c),
the
continuity
equation
can
u x + Vy
+ w z = (Ux*
+ Vy*
+ Wz *)
+ (Ux'
+ Vy'
+ Wz ')
= 0
(14)
Substituting correction
(13c)
into
equation
(14),
the
following
pressure
-[(DuPx')
x + (DvPy')y
+ (DwPz')
z]
= -(Ux*
+ Vy*
+ Wz *)
(15)
Equation (15) is a Poisson's equation with the source term equal to the local divergence of the flow field. To enforce the coupling between the velocity and pressure fields, the source term of equation (15) is first evaluated at the control volumes centered between the velocity nodes as shown in Figure 3. An averaged source term is then calculated at the cell center of p node for solving equation (15). In this way, the difficulties in solving the pressure correction equation, as described by Vanka [4] and Maliska [ 6], are eliminated. Coupling between the velocity and the pressure field is also assured. According following solution 1) 2) 3) 4) i0 Guess Solve Solve Solve to the steps: above analyses, the present numerical method contains the
and field
pressure using
transport correction
pressure
equation,
equation
(15).
"ROLVOLUME
Figure
3.
Control volumes where the mass conservation is evaluated for solving the pressure correction equation. pressure fields using equations (13a) through
to step (2) until solution converges. when the following criterion is met: 2 f
converged
solution is obtained
Error
= (IAU]max -4
lAY[max
[AWlmax)
/ Ure f +
]P'[max/PUre
3x
i0
where Au, Av, and Aw represent velocity changes solutions of the momentum equations. In of about solving 0.6 is the momentum equations With this, in step in (b)
during
each
iteration due
to the
above,
factor are
recommended.
Ap's
equations
modified according to the underrelaxation factor. For the correction of velocity field, no underrelaxation is required. But the correction of pressure field should be underrelaxed slightly (around 0.9) when the grid nonorthogonality is strong. This is different from that suggested by References 11 and 12 (which recommend no underrelaxation for pressure correction).
11
NUMERICAL EXAMPLES In this section, several numerical examples are employed to demonstrate the efficiency and accuracy of the present numerical method. To serve this purpose, 2-D and 3-D, laminar and turbulent flow cases are included. These cases are: (a) 2-D laminar driven square-cavity flows; (b) 2-D laminar flows over a backward-facing step; (c) 2-D turbulent flows over a backward-facing step; (d) 3-D developing laminar flow inside a 90-deg-bend square duct. Detailed descriptions and results of the computation of the above cases are included as follows. A. 2-D Laminar Driven Square-Cavity Flows
The first test case is concerning laminar recirculating flows inside a square cavity. Only one side of the walls is moving at a constant speed tangent to that wall. This case has been studied extensively by Burggraf [15] and has often been used as one of the standard testing cases for numerical methods in solving the incompressible Navier-Stokes equations. Physical geometry and wall boundary conditions are illustrated in Figure 4. Reynolds number of the flow (based on the cavity size and the moving wall velocity) studied in the present analysis is 400. Two different mesh systems, as shown in Figure 5, are used to study the effect of grid non-orthogonality on the accuracy of the present method. The grid system of Figure 5(a) is uniform and orthogonal while the grid system of Figure 5(b) is non-uniform and non-orthogonal.
UPPERWALL,
II
v= o
//
__
"/f////////////1111/I///I/1111//I/I/111////I//111/11/1111,
lj lj lj
cj rA vA vA v_ vj IA
,/
vj rA tA
U=O V=O
W
U=O_ V = 0'/
v_ IA
tj //
/,
f, /j
i c
r, /A g*
0
U=O V=O
"illllllllllll/illllillllillilllillllilllllll////lllliJ
Figure
boundary square
conditions cavity.
for
12
c_
o
r-i
_'_
o o
_z
_ o m bl_
b_
_ I
13
Results of the computations are shown in Figures 6 and 7. Velocity vector plots of the predicted flow fields are compared in Figure 6 for the mesh systems shown in Figure 5. Detailed comparisons of the predicted velocity profiles along the mid-section of the cavity are illustrated in Figure 7. Predicted results of Burggraf [15] are also included. Good agreements between the present calculations and those
,T.tfc_7..-_,_- '_\\ t . . _--r" .....
i I I I t t # #
----
----' -_
----0_
'-
,.
Ili't.,r_
_\
,_rrtttt'_','.zl/.. , ,1 r I i i_;llt_,'--,,
' ",,
J ) t \ \ \\\\\-
:,,
"--..'."/l"ll/li'l,
, , _ '_ t \ \ ,,,..._--_
_//'J/Vl6
'/
(a)
Figure 6. Velocity (b)
I I
grid.
1.0
0.8
0.6
--
ylw 0.4
------
PRESENT
(ORTHOGONAL
0.2
0 -0.4
-0.2
0.2
0.4
0.6
0.8
1.0
U/Uupperwall Figure 7. Comparisons of 14 of velocity the square profiles cavity. along the mid-section
of Burggraf [15] are also included. Good agreements between the present calculations and those of Burggraf are shown in Figure 7. Discrepancies between the present predictions and Burggraf's results are mainly due to the hybrid differencing scheme used in the present method. The upwind part of the hybrid scheme produces large numerical diffusion which tends to reduce the strength of the vorticity inside the cavity. Effects of differencing schemes in approximating the convection terms on the predicted results will be studied in the next test case. Convergence history of the computation of the present case using uniform grids is given in Figure 8 which shows that the present numerical method is quite different. Almost identical convergence rates were found for the non-orthogonal case. 0
I I I U V
-1
LOG (AUmax_ \ Uref /, (AVmax._ LOG \ Uref / OR LoG/APmax_
-2
-3
--4 0 100 200 ITERATION Figure 8. Convergence history for the driven cavity problem, Re = 400. 300 400
B.
2-D
Laminar
Flows
Over
a Backward-facing
Step
This test case concerns 2-D laminar recirculating flows over a backward-facing step with 1:2 expansion ratio. The dependence of the size of the recirculation region (characterized by the reattachment length) on the Reynolds number (based on the inlet bulk velocity and twice of the inlet channel width) of the flow is of major concern. The physical domain and boundary conditions are illustrated in Figure 9 in which a fully developed laminar flow velocity profile is imposed at the flow entrance. A non-uniform grid of 45 x 45 was used for numerical computations. Several cases with different Reynolds numbers from i00 to 800 have been studied. An experimental and theoretical study about this problem, which results will be used as the basis of data comparisons, has been provided by Amaly et al. [16].
15
U = O. V = 0 IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII,
h
du
_'x=O dv _--_= o
h
U=O/ y=O//
,/1/1/1/1//////////////////////////////////////////////////////////////////,
J
Xr
"I
Figure
9.
(1:2
To save computational efforts, the solution of one case with Reynolds number i00 is obtained in the first run. Then, a series of cases with increasing Reynolds numbers (i.e., 100, 200, 300, 400, 600, and 800) are calculated using the preceding results of lower Reynolds number as the initial guesses of the flow field. In this way, an average of 500 iterations for each case were needed to obtain converged solutions. Two different differencing schemes in approximating the convection terms are employed to demonstrate the effects of the differencing schemes on the predictions. One of the schemes is the widely used hybrid scheme [9]. The other scheme employs the central differencing scheme plus an artificial dissipation term used to stabilize the solution which is similar to the one used by Rhie [ 17]. The artificial dissipation term becomes effective only when the cell Peclet number (or cell Reynolds number) exceeds 10. Results of the present predictions using two different differencing schemes are compared with the experimental measurements [16] and other predictions as shown in Figure 10. It can be seen clearly from Figure i0 that the present method with hybrid scheme gives results similar to those predicted by TEACH code [16] while the present method with central differencing and artificialdissipation reveals predictions close to those predicted by INS3D [18] and the method of Kim and Moin [19]. This is reasonable since the TEACH code and the present method (with the first scheme) use the hybrid scheme which introduces large numerical dissipation by its upwind part (for cell Peclet number greater than 2). This tends to reduce the reattachment length for Reynolds number greater than 400. The second scheme, which is similar to the ones used in INS3D and the method of Kim and Moin, has the numerical accuracy close to second order by setting the artificialdissipation to be as small as the solution stability permits such that better accuracy of the predictions is expected.
16
16
I I I PRESENT CODE WITH CENTRAL DIFFERENCING SCHEME PRESENT SCHEME INS3D [18] CODE WITH HYBRID __e ,/_ _**"
14 12
s
6 4
/
0 I 200 1 400 NUMBER Reynolds (1:2 the I 600 REYNOLDS length backward-facing of the versus step predictions using two
_
800 number expansion). for laminar differencing schemes for 11 that the hybrid scheme of the step side
Figure
Stream
Reynolds number 600 are compared in Figure 11. It is shown in Figure second scheme gives a smooth shape of the recirculation zone while the gives a sudden change in the shape of the recirculation region upstream reattachment point. Also, larger sizes of the separation regions on the wall and along the upper wall are predicted using the second scheme.
I 23h
(a)
X (b) Figure (l: 11. Streamline 2 expansion). plots for (a) Hybrid plus artificial laminar flow Scheme. dissipation
I 23h
step
17
C.
2-D
Turbulent
Flows
Over
a Backward-Facing
Step
In order to demonstrate the applicability of the present method to turbulent flow case, one of the standard test cases presented in the Stanford Conference [20] is selected here (i.e., turbulent flow over a 2:3 expansion backward-facing step). The standard k-c turbulence model was used to provide the eddy viscosity for the transport equations. The physical geometry and boundary conditions imposed are shown in Figure 12. The calculation domain extends upstream of the expansion plane by 4 step heights and downstream of the expansion plane by 30 step heights to assure a fully developed velocity profile at the exit. A uniform velocity profile is located at the inlet plane. A 45 x 42 grid was used in the computation. 300 iterations were required to obtain converged solutions. Only hybrid differencing schemes were used in this case.
U=O,V--O
T '
_-4h_ Figure 12. Physical over geometry and a backward-facing 24h u = o, v = o conditions of (2:3 expansion). turbulent
m
du
_'=o dv _-_'=o
dk de
boundary step
flows
Results of the computation are shown in Figures 13, 14, and 15. These results are compared with the experimental measurements [20]. The under-prediction of the reattachment length is mainly due to the fast development of the mixing layer downstream of the expansion plane which is the characteristics of the standard k-_ turbulence model. Numerical diffusion provided by the hybrid scheme also contributes some part to the discrepancies between the predictions and measurements.
D.
Developing
Laminar
Flow
Inside
90-Deg-Bend
Square
Duct
This test ease simulates a three-dimensional developing laminar flow inside a 90-deg-bend square duct as illustrated in Figure 16(a). The symmetry plane is located at z = 0 where the symmetric boundary conditions are imposed. A fully developed velocity profile of laminar flow inside a straight square duct is prescribed at the entrance which is 2.8 duct widths upstream of the bend, A zero pressure 18
METHOD
(k-e MODEL)
1.0
.8
.6 y/h .4
0 0 2 4 x/h
l 6
I 8
Figure
13. turbulent
Locus flow
for
5.4h Figure 14. Stream line pattern of turbulent flow over step with 2:3 expansion ratio. a backward-facing
i_=--------_---=
- .........
02
.01
Figure
15.
Contours over
of
turbulent
kinetic step
(k/Uo expansion
2)
of
turbulent
flow
a backward-facing
ratio. 19
(a)
(b)
Figure 16. Geometry and developing mesh system laminar of flow a 90-deg-bend problem. square duct
gradient exit is imposed. and the inlet computations. Figure 16(b). comparisons. Velocity plane) are the by
(which is 4.5 duct widths downstream of the bend) boundary condition The Reynolds number of the flow (based on the duct hydrolic diameter bulk velocity) is 790. A 21 x 18 x 10 grid was used for numerical The front view and side view of the mesh system are illustrated in Experimental measurements of Humphrey et al. [21] are used for data
x-y
vector shown
plots on in Figure
the
flow at are 22 x
directions several
(i.e., stations
on
across obtained
of
to those 15 x ll
by half 19.
Rhie and Vanka, respectively. of their grid numbers, gives between the measured and the
numerical to general
is
believed problems
that with
the good
present
r/l
b_
II
0.-I
r/l ,.-..,
0 _
bl
0_,
II
b_
r-I
21
....
t t/e
...,11111 \\\....._._..__.....-...--....-....-....-.till
A
.Q
v
/,
. , , , , _ _111
ob_
A
1I '
\\...,...____.__._..-.-,,..If/ill
tf
1.1
coo
11
l,,.-'l.,-'-'--"--
. , , r llll
3
""
""
"" "" \
%,,
_m
,"_ II
o r_ tl
A v
N_
0_-,I
22
(a)
r/d
0.5
(b)
O0
r/d
0.5
1.0 0 0.5 1.0 1.5 U/Uref Figure 19. square Primary duct.
I_
profiles = 0.0.
for (b)
CONCLUSIONS
A numerical
method
for
solving
the
steady
or
transient
incompressible
Navier-
Stokes equations in three-dimensional body-fitted coordinate systems has been developed. In the present paper, the basic numerical algorithms and grid arrangements have been described in detail. A brief user's guide to the present computer code (CNS3D) has been included in Appendix A. A program listing has also been attached in Appendix C. Several numerical testing examples of 2-D and 3-D, problems included in the present work have demonstrated code is efficient and robust, and can be used as a reliable design turbulent and analysis applications. flow problems of the Applications SSME will be of the presented laminar and turbulent flow that the present computer tool for engineering code future to the internal publications.
present in the
23
REFERENCES
Chorin, Problems. A. J. : A Numerical Journal of Comp. L. and Kutler, P.: of Vortex Wakes. Method Physics, for Solving Incompressible Vol. 2, 1967, pp. 12-26. Viscous Flow
Steger, J. Computation 58 I- 590. Kwak, D., Incompressible Coordinate Nev., Jan. Vanka, Procedure Numerical S.
4,
C.,
Shanks, S. P., and Chakravarthy, S.: An Flow Solver in Three-Dimensional Curvilinear Primitive Variables. AIAA Paper 80-0253, Reno,
P., Chen, B. C.-J., for Flows Described Heat Transfer, Vol. S. S., and a Body-Fitted 99-113.
in 3,
Correa, S. Coordinate
Flow Transfer,
Maliska, C. R., and Raithby, G. D.: A Method for Computing sional Flows Using Non-Orthogonal Boundary-Fitted Coordinates. J. for Numerical Methods in Fluids, Vol. 4, 1984, pp. 519-537. Launder, Academic B. E., Press, and Spalding, London, 1972. D. B.: Mathematical Models of
Three DimenInternational
Turbulence.
of Turbu269-289. Book
S. V.: 1980.
Heat
McGraw-Hill
10.
Saadat A., Chiappetta, L. M., and - Final Report. NASA CR-174776, Raithby, Fluid G. D.: Flows.
Gosman, January
A.
Reduction
Ii.
Van Doormal, J. P., and for Predicting Incompressible 1984, pp. 147- 163. Latimer, Solution Anderson, Mechanics Stone, Partial Burggraf, Separated B. R. and Algorithms. D. and H. L. : Differential
12.
of
pp.
13.
Pletcher, Book
14.
Iterative Solution of Implicit Approximations Equations. SIAM J. Numer. Anal., Vol. Analytical and Fluid Mech., Numerical Vol. 24, Studies Part l,
15.
O. R.: Flows.
J.
Steady
24
16.
Armaly, B. F., Durst, F., Pereira, J. C. F., and Sch}_nung, B.: Experimental and Theoretical Investigation of Backward-Facing Step Flow. J. Fluid Mech., Vol. 127, !983, pp. 473-496. Rhie, C. M. : A Pressure Based Navier-Stokes Solver Using the Multigrid Method. AIAA Paper 86-0207, Reno, Nev., January 1986. Rogers, S. E., Kwak, D., and Kaul, U.: On the Accuracy of the Pseudocompressibility Method in Solving the Incompressible Navier-Stokes Equations. AIAA Paper 85-1689, AIAA 18th Fluid Dynamics and Plasmadynamics and Laser Conference, July 16-18, 1985, Cincinnati, Ohio. Kim. J. and Moin, P.: Application of a Fractional-Step Method to Incompressible Navier-Stokes Equations. J. of Comp. Phy., Vol. 59, 1985, pp. 308-323. Kline, S. J., Cantwell, B. J., and Lilley,
and Conference III. on
17. 18.
19. 20.
G. M. (Ed.):
Turbulent
AFOSR-HTTM-Stanford University, I, II
Complex
21.
J. A. Duct
K.,
Flow 1977,
Laminar 1985,
Passage Flow Analysis Method Aimed and Fluids, Vol. 13, No. 4, 1985, pp.
Chieng, C. C. and Launder, B. E.: On the Calculation of Turbulent Heat Transport Downstream from an Abrupt Pipe Expansion. Numerical Heat Transfer, Vol. 3, 1980, pp. 189-207. Amano, R. S.: Development of a Turbulence Near-Wall Model and Its Application to Separated and Reattached Flows. Numerical Heat Transfer, Vol. 7, 1984, pp. 59- 75. Chen, Y. S. : tations. AIAA Applications of a New Paper 86-0438, Reno, Wall Function to Turbulent Nev., January 1986. Flow Compu-
25.
26.
25
of program control parameters that specify the maximum number of iterations, the type of flow (i.e., laminar or turbulent), number of iterations for solving the pressure correction equation (typically 10), and underrelaxation factors for solving the transport equations, etc. This is followed by the definitions of all the program constants including turbulence model constants (these constants are subject to change according to the user's specific flow problem). Next, the program asks for inputs of the initial flow field guess from a restart file (LU = 8) which contains the grid system coordinates and flow field data that may be created by the user (including grid generation) or obtained from the previous solutions. Format of this data file is also subject to change according to the user's preference. Next, wall boundary control parameters, boundary grid normal distance to the wall, and wall boundary direction cosine are calculated in subroutine DIRCOS. Subroutine TRANF is then invoked to obtain the grid transformation coefficients. Before the solution procedure starts, the inlet mass flow rate is calculated which will be used to control the outlet mass flow rate to enhance mass conservation. The solution procedures consist of a series of subroutine calls to SOLVEQ starting from the solutions of the velocity vectors, u, v, and w, and then the solutions of scalar quantities (including the energy equation and the turbulence model equations) and finally the solution of the pressure correction equation to update the velocity and pressure field such that a divergence-free flow field can be retained. After each global iteration of the solution procedures, the numerical of iterations and the maximum flow field corrections are checked with the initial settings. If the convergence criterion is satisfied or the number of iterations reaches the prescribed value then the solution procedures stop and the flow field solutions will be written on the pre-assigned disc file (LU = 7). For instance, if a steady-state laminar flow problem (Reynolds number of 600) is of interest and a converged solution is expected within 300 iterations and the number of iterations for solving the pressure correction equation is I0 and the underrelaxation factors are 0.5 and 0.95 for transport equations and pressure correction equation, respectively, the first inputs from LU = 5 would be: Line 1. 2. 3. 300 0.5 600. 1 0.5 0.0 input sequence (i.e., from restart file), the program reads in data records. See Figure A-2 for grid structures. Notice that variable dimensions of (L+I, M+I, N+I) for solving the pressure It is important to check the COMMON table for proper variable 10 0.5 1 0.95 0.5 0.5 0.5 0.5
0
m
O0
E bn o
(,/)
u., _:
Z
[]
o o o to o
_o_
_Z
_z
re_
d _a_
I
\_z_-
I
_z
_0
o N_
0_,-
w_
ee--
z_
_aa
a
ez_
_o_g
28
(I = 1, J = 1, K = N),\
(I=L,J=I,K=I) (I=I,J=M,K=I)
(I=1,J=1,
K=1)
Figure
A-2.
Grid
mesh
structures
for
3-D
calculations.
If the flow problem involves symmetric or cyclic boundary conditions, then the user can look into the subroutine SYMOUT to specify the appropriate boundary conditions (the conditions shown in the program listing of Appendix C are for symmetric boundary conditions at K = 1). For cyclic boundary conditions at K = 1 and K = N, data at K = 2 and K = N-1 can be used to obtain boundary conditions at K = 1 and K = N by requiring same gradients across K = 1 and K = N. This method is simple but will lag the boundary conditions by one iteration. A direct method without lagging the boundary conditions can also be employed by modifying the subroutine of linear algebra solver LINERX such that the boundary conditions can be part of the solution of the TDMA (tridiagonal matrix) solver. In case of incorporating different wall functions for turbulent flow problems (e.g., References 24, 25, and 26), subroutine BOUNC and WALFN can be modified according to the user's method of wall treatments. The set of wall functions given in the program listing of Appendix C are derived from the conventional wall law and the equilibrium turbulent kinetic energy relations [8]. When additional source terms are to be added to the transport equations due to flow problem requirements, modifications to the source term calculation section in the subroutine SOLVEQ can be carried out. Notice that in the subroutine SOLVEQ source terms for the velocities v and w are included in the u-source section. Purpose of this is to save some computing time since these source terms use similar calculation routines. Some times it is required included in Appendix C. to solve more To modify the transport program equations other than the basic to incorporate more equations,
ones
29
several changes are necessary. First, new variables must be added to the COMMON table (this can be easily done through the computer editor session). Then, new source term sections are added in the SOLVEQ subroutine. Finally, subroutines WALFN and SYMOUT are modified to incorporate the new variables into the boundary condition setting routines.
3O
B SYMBOLS
A(K) AB(I,J,K) AE(I ,J ,K) ALC ALE ALK ALP ALU ALV ALVIS ALW AN(I ,J,K) ANAB ANVI(I) ANWl(I) AP(I ,J ,K) APO(I,J,K) ARDEN AREA AS(I ,J ,K) AT(I ,J,K) AW(I ,J,K) B(K) BB(I ,J,K) BOUNC
of
a tridiagonal the the for for for for for for for for the
through through factor factor factor factor factor factor factor factor through coefficients
a control
= Underrelaxation = Underrelaxation = Underrelaxation = Underrelaxation = Underrelaxation = Underrelaxation = Underrelaxation = Underrelaxation = Link = Sum coefficient of the wall wall the link
cyclic
boundary
c-equation k-equation pressure u-equation v-equation effective w-equation north face faces for for a control direction in physical domain v-equation w-equation volume of a control volume viscosity correction equation
at link link
all
= Modified = Modified = Sum = Link = Area = Area = Link = Link = Link = Matrix of
link
of
a control
a control a control
of
= Coefficients
Stone's getting
31
of
through through
bottom east
face face of
of
a control
volume volume
a control
Karman
constant, model
= Turbulence = CMU**0.25 = CMU**0.75 = Convective = Convective = Convective = Convective = Grid = Grid = Grid = Matrix = Diffusive = Diffusive = Diffusive = Diffusive = Diffusive = Diffusive = Turbulent = DE at the
constant,
= 0.09
of of
a control a control
a control a control
of
transformation transformation transformation elements flux flux flux flux flux flux of
coefficient, coefficient, coefficient, a tridiagonal the the the the the the bottom east north south top west
CY(I,J,K) CZ(I D(K) DDB DDE DDN DDS DDT DDW DE(I ,J ,K) DEO(I,J,K) DEN(I,J,K) DENO(I,J,K) DENC ,J,K)
matrix face face face face face face of of rate, of of of of a control volume volume volume volume volume volume
a control a control
dissipation level
the at
time of a
level surface
32
= Initial
value
of for
density calculating
of
the the
value k level
of
dissipation
rate
previous size, coefficient coefficient coefficient constant, criterion correction correction correction correction correctlon correctlon correctlon
DW(I ,J ,K) E EREXT ERRE ERRF ERRK ERRM ERRU ERRV ERRW EX(I ,J ,K) EY(I,J,K) EZ(I,J,K) F(I ,J ,K) FO(I ,J,K) FI(I ,J,K) FLOW FLOWIN GEN(I,J,K)
= Convergence = Maximum = Maxlmum = Maximum = Maximum = Maxlmum = Maximum = Maximum = Grid = Grid = Grid
transformation transformation transformation variable previous quantity mass mass flow flow kinetic of
= Tentative = F at the
= Turbulent
production
rate
33
HINUM I IBC(I) IE IG
= Large = Index
number,
= 1.E30
along the _ grid lines grid index for the transport equations flow and =2 for turbulent
assigned
control parameter,
=1 for laminar
IITO IITY IJLO(I,J,K) INIT INPRO INSOE INSOK INSOP INSOT INSOU INSOV INSOW IS ISWE ISWK ISWP ISWU ISWV ISWW IT ITT J
of wall boundary
grids
= Logical parameter = Logical parameter = Logical parameter = Logical parameter = Logical parameter = Logical parameter = Logical parameter = Logical parameter = Starting = Number = Number = Number = Number = Number = Number
for solving the k-equation for solving the p'-equation for solving the T-equation for solving the u-equation for solving the v-equation for solving the w-equation
value of I of the solution domain of sweeps of sweeps of sweeps of sweeps of sweeps of sweeps for solving the E-equation for solving the k-equation for solving the p'-equation for solving the u-equation for solving the v-equation for solving the w-equation
= Last value of I of the solution domain = Number = Index of time steps along the n grid lines
34
= Boundary = Starting = Last = Index = Boundary = Starting = Last value value along
grid value
solution lines
domain
the
solution
domain
of K of dimension
solution grid
domain in I direction
system
point of for
of
blockage region
region in
in
I direction
blockage solving
I direction equations
= Subroutine =L1
algebraic
dimension
of
grid
system
in J
direction
point of
of
region in
in J
direction
blockage region
J direction
blockage 1
parameter
dimension
of
grid
system
in
K direction
of
region in
in
K direction
K direction viscosity
effective iterations
of
(relative) P 35
gradient,
PDUV PEDA PP PPBLK PSCI PTA PW RENL SIGE SIGK SINX(I) SINY(I) SINZ(I) SMNUM SOC1 SOC2 SOC3 SOLVEQ SP(I ,J ,K) SPK(I,J,K) SU(I ,J ,K) SUK(I,J,K) SX(I ,J,K) SY(I ,J,K) SYMOUT SZ(I
,J,K)
parameter p rl p_
for
link
coefficients
boundary value
source control
= Reynolds = Turbulence = Turbulence = Wall = Wall = Wall = Small = Source = Source = Source = Subroutine = Linear = Secondary = Constant = Secondary = Grid = Grid part
constant, constant,
direction direction direction 1.E-30 due due due for of linear to to to solving the source part of the of shear shear shear
term
part
source part of
constant
term
conditions
TAUN(I)
36
TIMT TJO(I,J,K) TM(I ,J,K) TMO(I,J,K) TMULT TRANF TXXE(I,J,K) TXXW(I,J,K) TXYN(I,J,K) TXYS(I,J,K) TXZT(I,J,K) TXZB(I,J,K) TYYN(I,J,K) TYYS(I,J,K) TYXE(I,J,K) TYXW(I,J,K) TYZT(I,J,K) TYZB(I,J,K) TZZT(I,J,K) TZZB(I,J,K) TZXE(I,J,K) TZXW(I,J,K) TZYN(I,J,K) TZYS(I,J,K) U(I ,J,K) UO(I ,J,K) UC UCXI
= Total = Jacobian
= Temperature = TM at = Wall the previous stress for coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient coefficient calculating for for for for for for for for for for for for for for for for for for east west north south top the face face face face face grid transformation flux flux flux flux flux flux flux flux flux flux flux flux flux flux flux flux flux flux coefficients time level
shear
= Subroutine = Metric = Metrlc = Metric = Metric = Metric = Metrlc = Metrle = Metrlc = Metric = Metric = Metric = Metrlc = Metrlc = Metric = Metric = Metric = Metric = Metric = U-velocity = U at = Velocity = U-velocity the
diffusive diffusive diffusive diffusive diffusive face diffusive diffusive diffusive diffusive diffusive diffusive face diffusive
previous at the
time center u_
level of a surface
gradient,
37
= U-velocity = Velocity
gradient, at
outlet u ux Uy u
Z
plane
VO(I,J,K)
VISC VISE(I,J,K) VCXI VEDA VSCI VX VY VZ W(I ,J ,K)
at the previous viscosity, viscosity, gradient gradient gradient gradient gradient gradient
time level
= Molecular = Effective = V-velocity = V-velocity = V-velocity = V-velocity = V-velocity = V-velocity = W-velocity = W at the
Ueff v_ v v vX v v Y
Z
WO(I,J,K)
WALLFN WALVAL WCXI WEDA WSCI WX WY WZ X(I ,J ,K)
w_
W
W X W
Y
W Z
38
from
the
last
grid
= Nondimensionalized = Z-coordinate
YN,
= u y/v
39
APPENDIX C PROGRAMLISTING
p_
N_'_
41
J
;IF.. ,le-
,11
42
r'_
--) II
_,
_J
_,
un
-1_
I[
U_
_.v
Z3) UJ
I1
II
:1 II
II
_
o..) T-..'_
,-,_;
I-._ _
.... ., ,, ,,,_
cO oEI "-_ >OOO
UJ_MMM_NII_H _ H _ H H H 0 } _ _ _ U _ uJ _ _
._Z_ZZ_ZIL_Z_ _ _ _t H _4 H H H H H _ H U u I I I !
(3U
OOOOCJO
L3(3
_10000
r30000
0000000
00000
O0
43
in
i_
i_
:
Ii
8
4_ _
ooo8oo_
_ Ii Ii 4_ II 41 {i O ::)
. .,
v v I _
;;_ -;
,,I v_ v I
"_
. ...........
I_
,-_
v) v1 ,_ _ uJ
: ! ....
......
T_ _. ..... ,u _ _,)
_H _ .... i__
_
_ _ "
.....
::) _ 7_ ......
. .
__
-1
-) v
b_
_,_ rJ(3 ii
ui ii
rJ II
L_
:
I I
.... ooo.
I
o ..........
I I
h_1
I
ooy
.........
_
I I
..
I
_-)
r_)
_l"
_)
ULJ
n:l
-at
UO0
000000
(1[
N_I
-_._tr3Mf_OUO
_1
O 0000
0 OOn
IE}
L_ 1)
0 OOI)
e'30
_J
rJ (_
('J
('-I r Joo
r?J Or_
t,_
r.3 O
f_O I-Jr_
000 O O ')O
r_J r J I_ I_
oo oooo
...........
tTIC_O
_o
o
O
oo _ooooooooooooo_
ooo
OO
ooo
Or
oooogg
(10 OT1
i}O
44
i_..
i_.
r_
f,,-
i_.
r.-
r _.
no
0o
o_
0o
flu
0o
o_
o,
ok
o.,
o,
o_
o_
o_
LgOOO
-1
")
Z Iu
O_
(_
"1
,.-o
C)
4_
uJ 0_
t,,_ w_Ju
uJ
A_ _ Z --') _4 -.) H Q O
uJ
_- o
>
r._
ul
,._o
tic CI -J N
4-
=2E
>-
_
IIi + _E LLI lU p
.
J L_ If]
_.
_ CJ Z ..J_ %0 __ % _ _ _ _. (j
-_,
v I-
_-. *--I
[,.... (,-_r,.-
Lii
UI r-_ -4-
.....
_ P_ ILtL v
H
LiLl kO H _ I I
z
Z O
........
H (_ OOC_O _4
H
II > II _> II
_l
rj
f_
UU
_r
_L
,"_
O0
O0 L_ I L3
0_0___000 _oOOC)(-J
O O
O r30
r_ C_
O O
O O
O O
c3 O
f_ E)
O (_
O r)
r3 O
_3 O
C) n
_3 O
L30 O O
_-) r_
C} C)
O {7)
IT) (_ _30
c_ O
O FD
O (-J
O C)
(3 C}
13 r_
O O
O O
45
H_
_ 000000
oo__oooooo_oooo
46
b4
_H
vvv_v N
?
U _lJO _Ua
_-,gg" Y
l'u
o_V_
Q.
I-
L;_
,_
.....
_
It 1-11o.{i. I I I H I-I-rn_ _I-n9
T
"1
....
II
II
II
_,
^
,,
,,
_ _ _....... ogooo3ooo
000000000
47
2,2,''
, ^ .
_,_._..,__
....
I,-I_ _
_2'_.-_
U.ltLIUJ IU
.'_.........
I--I Ml'-'ll'- I-'4M I--I_. 0,. I'$. I I I H
_'
l--4H %
''
%1,-
_....
I.-.iJ)_
" ........
I-4
_ _ H H H H H _ H H H _I _ H H _ H F_
I+-I
I.--I
+,+
I-4
i-i
i.-+
I--,I
I+-+ i-i
i.-,i
I-4
_-,,I I-I U ('__ _') I'+'+I _Ir ,.._i. (j,. ,t-1 iJJ 0 c-} ill r-) 00 LU C-,) .(:i m IU 0 C_ ,_i_ U ill r'J _ 00 0 0 _ ..+_) w IL +"_I .,0 _i_ _irj IJL C_ U0 .fat t.10 UL-) +,.J (-] U,0 L/_ P P p
ii Ul
I__ 0
(_, ,r__ 0 r.z, r-J o 0 0 C'b o r_ o 00000 _ ro o o o r-j ,o ,__ 0 o +D +-I +_ t_,-_ _ C_
{_-
t_
o,
_,
0-
_-
.el
._1
"el
_'_
.el
+n 0 0
_ C-s 0
C 000
48
_<<_ 11,,_"
_UO
__
_UO
_vv
_v
_vv
_U_
vvv
HH_
_vvfg_ll
N_HHU
_I
_4XX_C_
I! N
II _
II N _ _ _ m_ Z II O II f_(_
vc_
$_II
Hvvv_(_ll
II _f_om
II q U
_ _NHUH
_4
t_ I
_ m_
U _
U H
| n
I _ _ g
I!
II _
I_ g
II _
II _
II _
II _
II
II
II U
II _
_ U
v _
I|
I|
II
II
II _
_ U_
_I_HHH_
_CJ_m
:
!
49
IIIIII
IllIII
_;77
.......
VHHU
HHHV_I$1
IIYNNXX_
LL
U.
tJl.
U..
U.
U..
U.
U-
LL
IL
0000
OC_
',..t
%_
",,I" _t
,.,.'It ,,..r
,,._ br'_
o_
U"_ _
",0
_D
,{)
I_.
I_-
P,,.-
r',.-
O0
O0
rJO
_0
i_,.
(_P,
,_[
eli
_d[
,q'
oooo
OOOO
ooogooooooOOOoOO
OOO OOOOOOO_'JO
oo
(3
oorD
ooooooo
C._OO
OOOOOO
oooo_
C3
goooo
OOOOOOOOOEDOOO
ooooooo
oo
OO
5O
*1 U
_0 . 0
,,,,,,
NNN
:
m I_
:
m _ ** :3
I_
FIuO
N_J
_1N
....
H_
HH
......
+ _ Ul _ _ I I I _HHH_
....
vvv
_'_
.....
_ _ P U _
o
J
:3
UJ W_
o_
_-4 _ lU ul 1L;
_O_H_HH_
_HHHZ Z
_.._.
m_
,. ,, ....
v ....
, ,.............
0,. i_ (_ n_ o. o,_ oc o-
,,_
{_ ,,/)
=,_..,,,,';;,,-_,_.
V'_ C]_ CI' CI ,,,_ _..) ro L._ )C_ f] C_
......
n_l "_ U :_"
_,.-,
*_ H 0
,.-_,.-o
H _- tL el: tU 0 0
! 0 _ .
ij
_goog_oooooOOooo
O0 0000
........
______
OC) O00000 O0 _-L_
o_
51
ORIGINAL OF POOR
PAGE" IS QUALITY
oo oo
O0
o o
_
.... _:,
_
0000
_oo oo
_
0
_
0
o o
0
o o
0
o
0
o
C_
o
0
o _
oooooo _oo_oo
O0 OC)
52
O0 OOOOOOC90
'_
_ O OO
_-_
(9 C9
c_
C_
C3 C_
0 O
c_ O
(-_ O
0 C_
C_ O
Oc_ OOO
(DOOOO CDOOOO
COOOOOOO OOOOOOO
CI_ O_
53
0000
NN
N_mml_lu
000
............................................
000000000() O0 0000 OLq 000 _000 000 O0 000000
000r30
_I
"i"
_ Z _
vvvvv_v
Xx
X_
I1
II
II
#1
I111
I1
oo000o0ooooo0 O0
00000000000000000000000000000r
oOOOOoOOOr>OO
oOor_
ooo00o00o0o0ooo00ooo00000o0oo )
00000000000000000000
O0
54
nO
u_ leiii 00000
uu J 0
N_NNN
_o_ y
XX_Nr4_XXXX_NNN f_ f_J rj u_U _ _ w uJ
Z_
........
_q
......
ii: ........
_) XXX_X _UJtU NNN_N_N_N_f_gJgI_ w XX x C) O0000 C3)0( r_ 00 C_ C) C_ _u_ul Uir_ I-(_
II
II
II
II
II o_r_
wo.
_ XN
LU
_ _
O 0
g IL
t|l
C)
."-, _U
tu O0 0000000000000000000 00000000000 0000000 [)or3 O000 I,.-L_
____o__
55
I-*_
21
_..4
L2_
.I
0 sAJ _
$o_
_
Io
_ Z
o'J
i,-i
_.
IUNmO
rJ_.l
-3
t&
c_
.J
_1_ 0 O0 0
-Jr {"_ fD
oo _
58
"oO OC) v :I r t.o ::E t/3 i.u r3 I t/1 _-4 4) _..,:3 O r*d H J
--)
)-H
v [J
_ CJ L_
"-_ _
I-4 a L) tt
_
I--_V3 tll t--I UJ 2: _,') U*(_ --_ (_ IJI
|U
UJ
If)
",_
'lj_
ItJ Z
I->-t-
_
C) at) --1 I.II I*-Z Z J UHL_ r_JZ r'a _ 0 t_ t-
.............
J_ _ (J ':> I I 0 O 0 u-
_
lU I_-I
_)_ t/)v)
_L
U.
--_
t3
_)
z3
,u
tZt c) t_ O t_
iJ
O0 O0 O0 0 0 0 0 cD w P-tu ._
_NNN_N_N_
57
O ::bOOl- O
.'IU _ H:I
_.
e--
L5 O
_o?oooo
,r- q--
........ _ _^
_ _2 "2
Z p )-NNNNN q. _--q.
-J
C:)
_--
,._OP_:!
PQ.
_J
__
>
t,,-
+++++_
_uo
*,o
_. _
_ao
_._
).-_r_r,_
_._
"-.
oo,'
r_l u, + I
.....
__XNZ I*,,h P-
_.
II I-- ,_i tL
.....
. _-,'_t.ILI Ivl V) _-_ I-P" :L ..... I%1 I--_ , _[_ i"%1 ..8 N ; II
.....
o
z
_ OZ N
_--q N N
N IM
fM l_.l v<l J
O N
r_) r-d
C_ i'd
II
......
O O E:) C30 ii II II II II Z _'I I-- I'0
..........
IJJ 31:
I
I I U U
I
I I U r_
'
I I U
nD
_1_
_n
n'l
C0
_1
(1_
_t_u
_J
UQ
Lu
[ll
[u
E)
(-'J
C)
E)
f__
0000000(7)
58
ORIGINAL OF
PAGE" iS
<,0 +,. o
......
v
},},;_+
HHHHHHX
_ =
X _
=
N N
oooooo
_+_++++_++++_++_+_+_+_+_+++
0000 00000000000000 000000000000000000000(_000000_00
OttO0
OC_O
_
0
_
0
_
0
_
c_
_
0
................
0000000(_0000000 000000000000000
59
-l-
+"_
+, T
"71 -
_ T
.` _
+ T
.` _
, _ "
-1 41 1-+ v
+_
41
iii
4-
.41
.J+
tu
i+
0_
_-_
.
UU_JU_UU_U v + + .41 .41
:_
i(
.`_HI+4
l--+.`Hl,-4l-lk--I
.`
1+--1
_I_',_
_.i_
_,,"
I-4
_I
,i +
41
II
41
II
il
"41
it
_k
41
41
41
_i
il
il
il
41
till
Ill
i_ll X_
t%l >-
_ I+-
_I+ I--
i_I _-
_i ul
O .41
r+_ I_
f70 + *t_
Cl .41
(-7 I_
(l .41
n .41
L_ iii
(L_ i-
E)
__7 t +
4-
.41
.41
I--
,=1
<It
(.+
()
.41
.41
_l
.41
li_
iu
ill
ul
Ill
+-+
c+
) ul
0 ill .41
_ t_J 4-
0 li_
_ Iu .41
C1 lu
0 Itl +
0 tU tu .41
_ _'_ _ -_ ....
-) "-) I _I I v uv ._+,')v _t li,Iii II U _i u 41 u Ii
" ....
I II IJ_)) 41 (i + _)))_4_
'-,"
.....
I + I 11 1 1
' -2.......
H H I-_ _ _I _-1 CL ]1: _ (3 0 C1
tu t
t3 _I t-LI (3 _ (1 #_-1 It r)
kl fl
i tJ
rJ r_
(i (1
I i
1_
r}
fl
(;_
tJ
()
f_
(7
(_
(i
II
II
_
II I-II it_ II I-II Ill _-+ H +_i r_ +--I
V_,I
U j U J[V I-4
UJ 2.1
g
u
I U U
I U
6O
ORIGINALPAGE I$
OF POOR QUALITY
NI_.I _
N _
NN .st _
NNNI_N IE
NN OOO If ,If
NI_N O OO ,II
N "It
I%I N
N OOOO ,Ii
f_I
+,
_ r*J
+ er_ -,r:_r'.J
OOOOOO ,I_ _
OOOOOO ,l ,k
,Ii ,I_
,I_ _
OOOO(_O ,Ii _
.Ii ,IX
,If .(I
,If .Ii
.... ' " '' ' +++ ....... .... _._ ,-, ... ,..., _,_ ?Z,?^?
"=_ I 1-I _! _
_T_
, >_
T_--_ , 7 ,_
_ _
4-
......
, , , , .
.-+
...
, , , , + .
vvvvvv
' ,
vvvv
??,_??? _.T_';';-_-
,, , ,, , "_._T_. _,.
_7 _.,..
_l
o._
c_
4.
,r-_
l_-'_-l_
l_,-
x:
I_
,,_
,,.
I-
II
II
II
II
II
II
l--l
<_L
II
lW
II
II
II
II
II
II
II
II
II
II
If
II
II
II
II
II
II
II
II
I#
#I
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
oooooooooo
0000000('3
o
-
_'3
oooooo_,_ooooooooooooooooo_oOOoOOOOOoooooooooo
0 ()00r)O0 0 ('-_ 0 (:3 E3 C3. 0 El 00 L_ 000 (-_ 0000
0000000('3
61
O0000f_O0
O0
O0 0000000 000000000000000000000000
O0
O0
62
N
+ + +
Z v t_
.N
=: ....
uJ 3r
_,_
;.-
G3
Lu
oU'_._,.
lu cO IIJ _ _ _ V W _ N _ UI _ _ _ _ v Im v III LU IU Ill _ _l N N _I N tll +_+ _ + + + _ "_ I s*l u_ Ul_ LU lU uJ I_ Z _L_ uJtlJ ,'_*LEt t._ ILl m tlJ
__,
_,
_=,.
7
UJ
.......
|1 II l! II I| II I1 II II II II II II II I1 II _ II II L._ II U II
....
II I! II II
,j
....
_ -
H II O II It l! II tl II O
Ig
It
II
II
II
II
II
II
II
II
II
II
II
el
II
II
II
II
.......
O0 O0000000000000000000000OO00000000000000000000000000000000
O0
63
CJ
cJ
'_
v I(
_ 41
I',"I U
C) LL
E) U.
0 O. "I
IJJ
" tU UJ U.I UI 3 UI Ill _J [11 LIJ _LI _ UI ILl I1_ eli U,I UJ _ _
(_ I---
-1 41
11. 4"
o. _
(_L _"
L_ |
-)0 I'-I'--
_Jf
UU
L)(-)
UU
L.P
L)
II
II
II
_1
I_
II
_1
_/)
II
I.,1
V*_
II
--1
II
II
II
_-_
, _z ...... _ ...... .......... _._. _',', .......... _':' ..... ..:'....... _o,_o
I
o ,o _o_
I !
oo_
_o , .... ?_s _
I I U
,
I L)
,,
I L)
:
I U
:
I L_
ooooo,_OOoO_Oo,.},-_o__,o
O00O 000 (7) ,_'J (_
r_
r'_
......
0 c-)
00
o ........ _, o_ooooo
r'_ _) 0 _ (_ f'} C) (-I
r)
oo,__(;,oooo
O0OOO(3
oo
O0
00000
ooOOOoOOOoOO _
E) C) 0
64
ORIGINAL OF POOR
PAGE
IS
QUALITY
C_
_-_
_-_
f-
,_)
"-)
II
II
II
II
L_I
t/t
II
II
II
II
LL
-")
_,_
II
II
II
II
II
II
If
(I
..........
H _ U_I _ 000 _ ..3 n,. ---) 13-
_, _, _ _, ......
O H _ U_ _ II 0 t 0 L'_ 0 D :;_-
__.......
_-4 _'_ LL O H (]l:30,. _ tJ_ _[
_
_ I:E
....................
O O (-} II II L) II O0 _ IL I_I./1LIJ 1--G") _) E}
_ ......
H C'," Ire 0 _ IOI. LL 0
..,
I
l
_o ,
I
o, '
-.,:I !
',
I
o: '
u'_l_ )
o
u'_
O0 000000
0_ O000EJO
O0
(_Or_ O00fJorJooorJO
00000
000 000 (_
0000 00000
0000000{30000 0000000000000
O_ 00000
65
A Ilk
"5 I,.H
0 J
gr,_
',
000000
OClOr_O0
ooooooo_oooooooooo_oooo_ooooooooo_
OOOO0000 OOOOOOO000
0000
000000000
66
N_lu_
_N
_OOu
t3_
_0_
+_
U+_
_O_UN_
m_P
O00OO
oooog_oo_gg
O0
ooooooo_ooo_ooooooooooog_goooooo
00000000000 00000000000
O00OO000OOOO00000
oooooooooo
67
T,TT
O I/%
_''" ........
! I I I I ! I I I
_ 7
1-4 _
d ....
_ _I
o d
_d ,II
44
"'4 ....
'
._?._?
...... ......
....
--; -') Z 0 I.-4 I--"-_
........
-') -) "3 "-_ -_ "-) -3 _
, , , , , , , , .........
H H _--_ tHH _
_
"-)
"_
"-)
"_
-_%
"n
.....................
........
, .....
_II II II _ i1_ v uJ v_u_*.'_ . .,. 4vvvv u_ _ u_ u_ _
,
,,. _vvv i._ i_+_ _-, ir_
........
,r_ L_ L..', If' 'P. I"+'+ _
L_;
U".
.......
L/_ U", ill
_ +z_v,+--_
Jl: z ,tl t-(n
I
U L_
._
II II II (.+)
7
-LL)
" 222
5J .3 /3:3
H_-_HHH
II
II
II _ g
II _n 7__ lU
iii iJ 7 tlJ
II 5 g
vv jjutj*jiJ_ l, i-Iii i i
, ........... _ t/)
,v_
_>OUUU_ !
,vv
_u_Juk) ,
II
II
II ,
(Ju I, llJ II l_
C}_JIJ{J_$......... Z L./I
4*_ OZ Z I
C) _-4
CD(_
I "_
v_vvv II II
II ._Z
It m _:[
II
g, ooo_
_ ! ! I ! I U _ r3 C_l n
z
uJ r3
ill
_1
I.--
.............................. __ _1 i/J
7_
I.--
f1[}
I_I
ULI
]_
)-
l-J
I'-_
l_t
i"IL
I I
t.J
U-_L OO r'JO
U_ OO
_L g'3
U+U r3 OOC'J
tL C-"
U_U_ _'_O
II
U O C__
lt. O 1_
ll
IL
U O
IL _
If+
II C, r3
IL
tL
00(--300000000
,_r_r_ L_
r'_O C-J r_
r363
r'J
C9
(:30
r/)
r_
OO
OOO
_.]O
E]
OO
68
ORIGINAL OF POOR
PAGE IS QUALITY
.o,._+oo.o_N,.,.,_,.o,.._o.
o_...,,.,oooo_
o..-
++
,,.o
--'J
")
--'_
"
T'-' _'-' _T
....
+,
--,
:_
a=,.-
e.+.IN
U.I
31:
Z,v_,f--
n"_
.,_r
NO_-_He-'I=--+
,,::I
tLIV'I
.._v
ILl."_'um_,-_v x X
_ N
__I ._I H
I-"I-(+O_(_=_c_
_-I--
_+It
, III Iii
. LII
._,_ +. ILl _
i'D
v o. _
cl_
c'.,.
o-
l'x.
,+/,ll.rl IJL H IL II
] II
_ ;_'__,_ _ -"J
" II
" II II
_Z)I'- XHv P O Ul II
,_x I.LJ
("Jl
(-_1
Cjl
CJl
I"lJ'
I,-
C_
,'_
,l:..Jr
4_._
P_
II
'_[
It'+
.......................
I-I,-I--I,.-,,rl <_.+I_ I r_ _ _ o i_ i_ ++I-+HI*-+II,._++I_ _ t+--I '...m p P_J Pe_ _._ u+_ I u
g_o
,._J . I O U I I ,j IL_.'t
_" _
,'.I. at "_
_
":I
_
_
_
U O I_-
='_g,
-J I O I i_ e,_
69
70
e.-
r_l
oc)
co
_zO H tt.
4-
3
ul
^
m _ mo_
"2
v ! v
tu
lu
CL
I--
-')
ZO
_0
o
......... _ __-_ _ _ _. _. _ _, _ _ ....
c_ H -I_ I I + 1-4_ 411--I_ l I _ II Z_ O O O O M I O O I O II II II II p II II II O 41" 1-4 _" E H O II II II II (:30 II II II M O II II O II II II O
._Iv tu
_0 _
_,o
I 0 I _a 0 0 0
_U
u_
I1|_
o
0
ooggg
O0
gooooogggooooo
_ ooooooog
_N
000000000000000000
oooooooooooo
?1
I,_.
00_
i_,-. r,_,l
_.I
..,f
u_
,,0
UI
".dr O" 0
. 01_ C)
0_
o _iu_
_tn
u 0_12"
I.'_
0 I
i._
I ()
cl_ iiJ IL
Z 0 t_
_--
r_
_.-
t.')
72
ORIGINAL OF POOR
PAi= IS QUALITY
ooo ooo
0(_0 000 000
o o
0 _
o__ oo_oo_
000000 O0 000000 0(_
_: _oooo_
O0 0000 000000
.........
oo
O0 O0
o_ ooo_oo
__ 000000 000000
_ _
O0
73
_,
,,
_,
,-
.....
...................
LIIV
4-_4 ....
luvvv 4.
4 " ":("
ill _
................
V_ I
;_-
llJ_v
......
7_
......
............. ;. \ _. ........................
7 .....
2 _,.......
_,_
_+
m-4
,q
,uj
-,1
_--
>-
II
II
tl
-1
.....
2"
i_
_?
(17
II
I-_
I)T
_l;'_ .....
()_lOZ_l
or)
, ......
0 I I ! I ! u
g .....
Ill I I
I"11_
,_ ,,, :__ -+
_ -1{
(.)
g .....
0 I I
r,..
rt_
_ ...... _
0 L._
(_)
................
0 0 0,O, I
(_l
r3
{-_
_'J
tJ
tJ
+J
(:)
,::)
r-)
(:'_
,+_
L)
+-7'
_'-) r)
,._
r+_
C)
i_
r3
c)
C-)
C)
+':l
:::)
+-_
(-J
t.TJ
c_
+)
_"I
C)
c)
tD
?4
75
0 0
>orJI-ZOO,., .elOnO')l--I
_.1
.J__.=ll.-.I tL ,..4 Z _ 0
,_l_ rJ
uj_vv
III
o
iii
.,,_ ....
0 2_ ,x TM _ I_
v_O
.11
>-
_,_
(.j
(_l
(J
LJ
'_
tD
":>
>,
-J
UT_
U.
>-
31
i>
c_
L')
CJ
r__
L)
C,} 0 0 Q v
_ j
v _
tL J I--
.7. _-
II I.-.
II _
J _
11 I--
I! J
II ,_
-_
f_II
3 II
CI_ v)
{Z N
_'l rM
_ II
Z J
Z I _j
I|
--)
(IL 11
3 II
Z J
_'_--
II j
_') i
i_ II
Z J
Z _-
_ J
Z I.--
II i_.
J )-
C_ _,. _-
L_ ,]L
,
I
,
I
oj
,
I
,
I
8,
oo
rj
76
.J
-4'
-..Tr .4"
-*..e Mr
u'_
u'_
_r_
u_
I.r_
Lr_
It_
u_
u_
.O._O
,,04[.)
tU
I11 I--I
.11 .11
_MO:3
= ),-C)
0 3 _J 4 I'M # a C-_ cJ r_J # a rJ o_ 41UL3 Z
.__.
_ .11
L_
eL
_.,
H -"_ _.
_-_.
uul
:,,
.,_:_'_'i*:
r_3
r_3
r_3
r_
C'_
>.I I
f%l U7
'
I I L)
:
! U
ooooooooooo_,oooo+._
O000+_or'_oC3C_OC3 0 0 0 04_._ 0 0 0 0 0 0 IC}
. r_
,_(_)r_t:) 0 0 Ir_) rJ
_oooo
(-3 0 r-'JO O0 0
0 0
oo
O0
77
er_
C.)O
00C_
uJ
(31
"_, ""
1> _-I X t >I I
I1
Lu 4 I
_, I
C_
Z u_
L)
....
T_
_T_
'-
.............
_, ;,+L
H 0 0 vv v
Z 0 0
b 0
_ II OOvvv_
l-.I
v _
b_;
l'-'l 0 II J
f) _ 0 0
II U
,_ lllm
II
II _J
II _ _
II
II N _
II
_ N
lJJO,'d[
_
O0
_
0
oo _
oo ooo _oo_ooooo_
O0 00000
78
ORIGt_AL OF POOR
PAGE iS OUALITY
o, IlJ
2
J
00
too
r--i
ii
Ul
ne _ 0
lll
IP_
+-I
III
[IJ
0 I i_ 0
r%l
,,_ 0 ..J U.
U.. O0 o+'i I,I+'1U 000 llJ lj,. O +--.,It oO ,_r rO ,_ (,0 i'_l ill (0 0 ,,.1r 0 N p,t+j 0'_i co 0 iI_ OU_ eli ill I,i'I i+"i o ffl
vl _..I
OOCO01'_ _._0
O0000CD
r-boO0
00000 0
C'I
C_ 0 O0000C'd
I"_
0 0
o_-_
79
0 I-J mO v Lu
_" ZOOH
O0
;-u_ Ul J eL)
F-
vv_
f_C:l
_,-,u z_n_
L _C1_ OC-) fJ
vt.J_
n,! -1
iu
_g
_ U XH_ Z
3
b4
m .jen tu LUI_
ul
LL IU
0 m_m
n,/
_,
_ Ill j_4
iLJ I[
IL o-Ji _
r_
_,'_
.It
() (1: ,
o _ _
8O
APPROVAL
The information in this report has been reviewed for technical content. Review of any information concerning Department of Defense or nuclear energy activities or programs has been made by the MSFC Security Classification Officer. This report, in its entirety, has been determined to be unclassified.
Dynamics
Laboratory
"_'U.S.
GOVERNMENT
PRINTING
OFFICE
1986--631-058/20122
81