UCRL- ID- 104090

DEgl 000073

REMAP: a computer code that transfers

node information between dissimilar grids

Arthur B. Shapiro

LawrenceLivermore National Laboratory

April 1990

Lawrence Livermore National Laboratory

: Mail Stop L-194
7000 East Ave.
Livermore, CA 94550
. phone: 415-422-5066
fax: 415-422-1370

_ ,
"':.;' _ ;"_.',, i _;' '''_ " _' " ,c,¢"

" Ifip ' III

. Table of Contents

1 Introduction 1

2 Execution and Input 3

2.1 TerminalExecutionLine 3
2.2 InputFile 4
2.2.1 ControlCard 4
2.2.2 Node Point Data 5
2.2.3 ElemcntDaua7

3 Example 9
3.1 REMAPExample 9
3.2 WARNING- User Beware 15

4 Theory 17
4.1 CoarseFilter 17
4.2 FineFilter 18
4.3 Shape Functions 20
4.4 Calculating the Natural Coordinates 22
4.5 Temperature Interpolation 23

References 25

RF_MAP

..... ,,, ,, ,,,,...... ,_.... ,' _r,,

.......... ",," ,"""'r,'" '"' r, , _HTIIln,l,lll,,,
..........
 I II i II

' . 1 Introduction

I I IIII

REMAP is a computer cede that transfers the axisymmetric, two dimensiG,al planar, or three dimensional tempera-
ture field from one finite element mesh to another. The meshes may be arbitrary, as far as the number of elements and
their geometry. REMAP int',rpolates or extrapolates the node temperatures from the old mesh tc)the new mesh using
linear, bilinear, or trilinear isoparmnetric finite element shape functions. '_,EMAP is used to transfer the temperature
field from a thermal analysis mesh tc)a more finely discretized structural analysis mesh when performing a thermal
stress analysis. REMAP was designed to be used with the finite element heat transfer codes TOPAZ2D I 1] and
TOPAZ3D [21, ancl the solid mechanics codes NIKE2D [31 and NIKE3D 14]. The 1/O formats in REMAP can be eas-
ily modified to accept input f',om other codes (e.g., finite difference) and generate output files for other structural

O ccxtes. REMAP can be used to transfer any scalar field variable between dissimilar finite element meshes.
Three papers by Thomv,;on from NASA Lewis and Maffeo form GE on the TRANCITS code 15-71were found useful
in developing REMAE Their idea of a coarse filter followed by a fine filter tc>determine which element from the old
mesh contains a node point from the ne',v mesh was used. The coarse fiIter determines a subset of elements from the
old mesh that may contain the new node point. The line filter determines the element that contains the new node
point. REMAP uses the ray_surface intersection algorithm developed for the FACET [8] code for the fine filter. This
algorithm has the added capability to determine which element the node is closest to ii the ncxte l-x)intlies outside the
perimeter (or surface) of the old mesh. Once an element from the old mesh has been identified as containing o_ closest
tc)the new node poink the natural coordinates for the node point are calculated. The isoparametric finite element
shalx_ functions are calculated next. These shape functions are then used to interpolate or extrapolate the temperatures
from the ncxles comprising the ,:_lctelement to the new node point.

Chapter 2 contains information on preparing in input file and running REMAR Example problems are presented in
Chapter 3. The mathematical theory and numerical implementation are presented in Chapter 4.

O
REMAP 1/_

I ' N rlqf' til_

 I I III _,

Execution and Input '

I I

2.1 Terminal Execution Line

REMAP is kept in the public library MDGLIB on ali Cray machines. To obtain a copy of REMAP in your private
disk space, type

exe mdglib remap x.

Tile terminal execution line for REMAP Is

O remap i=inf g=tpf o=nkf p=pfl

where

• inl=text input file contain|ng control information, new node l×)int coordinates, and new element connectivity.
• tpf=absolute input [lie containing old node point coordinates, temperature,,_,anti old element connectivity. This
is a TOPAZ plot file created from running TOPAZ for the thermal analysis. If you de not need to remap, then
this liIe ix the temperature input file tc)be used with NIKE for a thermal stress anab,sis.
• nkf=absolute outpt|t file containing the new node point coordinates, temperatures, and element connectivity.
This is the temperature input tile to be used with N1KE for a thermal stress analysis on the new mesh.
(default- nkplot)
• pfl=text output file listing the input inf()rmation and temperatures at tlm n<×lepoints of the new mesh.
(default=printtl)

File name dropouts are permitted, for example

remap i=inf g=tpf

The two output files, nkf anti pfl, will be created with thc default names nkph _tand printtl, respectively.

O
REMAP 3
 JiIih ,wdwiiIW , ,,dl_l

I I I III II III I II

Execution and Input

2.2 Input_File

NOTE' Values entered are for the new mesh

2,2.1 Control Card

column ,format quantity

1-5 15 number of materials

,,

6-10 15 number of node points

11-15 ' 15 number of elements

16-20 15 geometry type (l=axisvmmetric, 2=2D planar, 3=3D)

21-25 15 temperatUrecards
as the flag
(0=default;initial
condition)l=use temperatures on the nodal O

O
4 REMAP
 I H,, ,,iii I II I I[I I I I I IIIII I I --

, ' Execution and Input

2.2.9. Node Point Data

The node point data is formatted according to NIKE2D or NIKE3D, Supply one card for each node point with the fol-
/ lowing information:

NIKE2D -Axisymmetric and 2D-planar Mesh

' Column format ,quantity

1-5 15 .node point number, ni

6-10 F5.0 displacement boundary condition .code, (see NIKE2D users

,manual [3])

" 11-20 El0.0 ri coordinate "

21-30 El0.0 zi coordinate

31-35 15 node generation increment, k

: qp 36-45 El0.0 node point temperature - if not equal to zero, thenthis temper-
ature will overwrite the initial temperature in the TOPAZ tem-
_: perature file

Node point cards de not nccd to be in order. However, the,highest node point number must terminate the node data.
When node dam is missing, node numbers arc generated according to the sequence

, ni , ni+k, ni+2k,..., nj
t

i where ni and nj arc the no(lc numbers defined on two consecutive cards, trod k is taken from the first card. Linear in-
_I tcq)olation is used to obtain the coordinate of the generated nodes and the node temperatures. The boundary condition
_I code of generated data is set to zero whenever the boundary condition code of ni differs from that of nj..

t
t
i

-I

REMAP j, 5

.... [li ..... _1, ...... ?l_ ' II_ ..... [I a , ,lr II n_l IIT II......... lm III III f'
 I ....... J III I I I ...... II I ........

F,xecution and Input

NIKE3D - 3 Dimensional Mesh 0

column ,format quantity

1-5 15 node point number, ni

6-10 15 displacement boundary condition code (see NIKE3D users

manual [4])

11-30 E20.0 xi coordinate

31-50 E20.0 yi coordinate

51-70 E20.0 zi coordinate

71.-75 15 node generation increment, k

76-80 15 rotational boundary condition code (see NIKE3D users

manual [4])

6 REMAP

:1 lr, ...... Ilii,, _," I_lr '11111" W" II_IMII,,,, "ll_,"r,,fl,lllPll_,,"llillr I#1_ qll_TIHIlll'llll'Wpl' lll_,llrl't_l' tml .... II'l, _,ln ,,,,r, I_ "' _,,q,I ,, ,l[qperl,,q'_llll_
 'III I I III I I I IIIII rl I II I II IIII

Execution attd Input

2.2.3 Element Data

The element data is formatted according'to NIKE2D or NIKE3D. Supply one card for each element with tile follow-
.ing information:
r

,,

NIKE2D - hxisymmetric and 2D-planar Mesh

column format , ,
quantity

1-5 15 element number

6-10 15 node ni

11-15 15 node n2

16-20 15 node n3

21-25 15 node n4

26-30 15 material number

31-35 15 element node point generation increment, k

36-80 Data in th_._secolumns are not read by REMAP. However, this

information may be required for the NIKE2D simulation.

Element cards arc assumed to be in element number sequence. The highest element number must terminate the data,
Omitted data is automalically generated with respect to the first Card prior tc)the omitted data by
ni+1 = ni + k. The material Property number and element node point generation increment arc taken from the first
card. The convention for numbering node points is shown below,

triangle quadrilateral
3

1
n] I"12 133n 3
2

4 n 1 n 2n Sri 4

REMAP 7 ,
 ..... I I [ II I I I
t

Execution t_,zd Input

NIKE3D - 3 Dimensional Mesh using 4, 6, or 8 node solid elements

column fermat quantity

1-5 15 element number

6-1 0 15 material number

11-15 15 element node point generation increment

16..20 15 node n1

21-25 15 node n2

26-30 15 node n3

31-35 15 nod_ n4

36-40 15 node n5

41-45 15 node n6

46-50 15 node n7

51-55 15 node n8

Thc convention Ibr numbering 4, 6, and 8 node solid e,lcmcnts arc ,_hownbelow.

1 3 4 8 7

131n2 I'13134n4 n4 134n4 n n 131132133rl4 135n6 n7 n8

8 REMAP

"I

f
 III II I I I I II

O
3 Example

lip . I I III III III II i

3.1 REMAP Example

The only intent of the following hylx)thetical problem is to demonstrate lhc procedure in using REMAP. There is no
physical basis to the problem definition. The objective is to perl'orn_ a thermal analysis on one mesh and then map the
temperature field onto another mesh for the stress analysis. The TOPAZ2D input file for this problem is shown in
Fig 1. The input file defines a 2-dimensional slab, 2 units long by 0.6 units thick, with temperature boundary condi-
tions of 0F and 2F applied to the vertical end surfaces. The lateral surfaces are adiabatic. The material density, heat
capacity, ;rod thermal conductivity are ali given a value of unity, The finite element mesh is shown irt Fig 2 and iso-
thtr, ms of the calculated node point temperatures are shown in Fig 3. The post pr(xzessing code ORLON { 10l was used
to generate the mesh and isotherm plots using the output plot file, T2PL(.)T, created by TOPAZ2D.

1 e×amp]e problem
2 1 6 2 2
3 4
4 8
5 0 1, 1
6 0
7 marl1 1 1 1
8 i I,
9 i
18 8
ii 8
12 0
13 1 8 ,2 8, 2
14 5 2 ,2 8,
15 2 8, ,8 0, 2
16 6 2, ,8 8,
17 1 1 3 4 2 t 1 2
18 I i I 0 0,
lg 5 1 1 8 2,

• FIGURE 1: TOPAZ2D input file defining the heat trans_r problem,

0
REMAP 9
' ' '"'"|iii • III IIII II II III I IIIIII III I III IIII I I I II II I I I glm

Example . ,

',

i, ,,,

e,×arnl:_ le pr-oblern IL_

dsF = 8,188e4-.81
t _mP = 8, :k88e-l-81

t,4O

t,;_O

t,BO

O,BO

0.40

-8, 20

FIGURE 2: The thermal mesh consists of 2elements and 6 node points.

0
10 REMAP

II11
I i I i II I II Illl i III II I¸ I

Example

example problern I[b_

t _me_ (_. l.OOEIfg_+0Lcontours o; ( ompero-ture _ ....
dsF = _. 100(_e'I'_I ro|n(-)',.' O, -'-'-ql_e-30
l max(+)- O, 20e+01
contour |eve[_

i,4_ a = 0,33e+88
b _ 0.67c+00
1.20
c= O.[0e+01
d= 0. [3e+01
t,_]O
e = 0.[7e+01

0 BO

£'-').GO

U, 4 B

0,20 .................

El'. 00

-0.20

'-8, 40

_--.... _1 1 t J ! 1 ] 1 t L_

FIGURE 3: Shown is the steady state temperature field. The isotherms are vertical because the heat flow
is 1-dimensional from the right surface to the left surface. Also note that the temperature gradient varies
linearly between the end surface boundary conditions of OF and 2E

REMAP 11
 'l I , II

•. v.n:xam"ne "

The REMAP inputfile, REMAPIN, cletining the new mesh for the stress problem is shown in Fig 4. The execution
line for REMAP using the files REMAPIN which cletines the stress mesh, and T2PLOT which detines the thermal
mesh is

remapi=remaping=t2plot/ t v
RE.MAP creates a liie with the default name NKPLOT. Thi_ contains the new mesh data and remappect temperatures
I to be used as the temperature input file to NIKE2I). ORION can be used to display the new mesh (Fig 5) and iso-
therms of the remappecl temperatures (Fig 6). I advise using ORION to plot isotherms on the r_ewmesh and compar-
ing ',hem to isotherms on the old mesh to set ii they make sense before performing the thermal stress simulation.

1 I 8 2 2 0
2 1 0, ,0 2
3 5 2, ,4
4 2 0, 1, 2
5 6 2, ,15 _b,
6
7
1
2
1
3
3
5
4
6
2
4
1
1
U

i FIGURE 4: The REMAP input file defining the stress mesh.

1
i
i
0
12 HEMAP

I
 I IIII II I I II I

Example

0 i

example problem
dsf _ _,100e+01 L_
lime= 0.000e+00

1.40

1.20

-0.20

-0. 40

I I 1 __. I I' I I I I I

FIGURE 5: The new mesh to be used for the stress problem. The temperatures at node points 1 and 2 will
require extrapolation from the thermal mesh (Fig 2), while temperatures at node points 3, 4, 5, and 6 will
require interpolation.

_ ,' '

i REMAP _ 13
 III II ' I _ I I

Example

exa_p le problem l_L_

lit, e= _. i001_l_+Olcontours oT i_mperalur_
d_+" = 0. 10000e+_]i mln(-.)- [3.40e-BO
max(+)., 0.20e+81

contour ]eve[_

i.4_ a= 8.B3e+00

b _ 0.67e+DO

1,20 c _ 0. 10e+01

d= 0. 13e+01
l,B8
"-_---"_--_--____ e= 0, lTe+Dl

""--,-.-...,___,_

lD. 60

0,4@

0, 20 ---/

-El. 2B

-8.40

I I I I I I I I I I

FIGURE 6: Shown is the steady state temperature field mapped onto the stress mesh using the
temperature field data from the thermal mesh (Fig 3).

@
;I

I
 I I I II I II

Example

3.2 WARNING - User Beware _.

The h:apping ot"the temperature field from the thermal mesh (Fig 2) to the stress mesh (Fig 5) is done by interpolation
or extrap(clarion. The governing physics is ignored. Two dimensional heat transfer must be considered to accurately
calculat.e the temperature lien for the drastically different shanc of the stress mesh. Figure 7 shows the isotherms on
the stre,_¢smesh when 2-dimensional effects arc conside:ed. In comparing Fig 7 with Fig 6, note that the isotherms arc
curved aad are shifted to"the right. Although this is an extreme example, you should be aware of the physics o1"the
problem that is being remapped.

exa_4p 1e I_L_
_,]rn_,= [_. [(_Ol_Oc+pj[ contours of lempcwaature
dsf = 8. _P_qOSi_e+[31 rain(-) _ 0.07e-02
max(_) = 8.20e+81
contour ]euets

1.4El O= [_, 33e+88

b = 0,67_+00
1.20
c = _. tOe+Oi

d = 0. [3e+01
i, 88
---"'- .... --_._
e = 0 17e+81

_,E,_ ........ _..........

8. 68

• ']

0.2B

_. _10 ......

-8.20

.i
--8. 40
.

i 1 1 .L ........ __L._.. I i--__--_A [ J _[

I _ ,:_ ,:,.4 6 d, ,-, ,4 ,4 ,4 .4

I
t .
1|
! FIGURE 7: Shown are the steady state isotherms on the stress mesh when 2-dimensional heat transfer ts
.; considered. Note the difference in shape and position of the isotherms between this figure and Fig 6.
)

.i

P.Ev_,
P 15II t,,
w

ii il lilt' II Irl H 'l ' ,i,I]l_,=, rli......

 L

k I I IIIII II '

• 4 Theory

IIIII II I ii II Ill i I

REMAP transfers information (e.g, node po;nt temperature) from an old mesh to a new mesh. To accomplish this, the
element from the old mesh containing or closest to a node from the new mesh must be identified. This is achieved by
an element search comprised era coarse filler followed by a fine filter. The cotu'se filter uses a simple distance algo-
rithm that ,identities a subset of candidate elements. The line iilter determines the one element that contains or is ties-
' esl to the new nc_e point. Next, the tinite element natural coordinates for the new node point are calculated with
reference to the element, l'ollowed by the calculation of the isoparametric shape functions. Using these shape func-

O tions and the values of the field variable at the element's node points, the value at the new node point is calculated.
Following is a detailed description of the above procedure for two dimensional planar meshes. The two dimensional
planar results are directly applicable to an axisymmetric mesh. Only the iinal equations are given for 3-dimensional
meshes. For case of prcscntation, the tollowing terminology will be fi)llowed:

• element refers to the linite elements that comprise the old mesh
• stress point refers to the node points that comprise the new mesh

4,1 Coarse Filter

For every stress point (i.e., node point)in the new mesh, the coarse Iilter searches through ali the elemenLs in the old
mesh to determine a subset of elements that may contain the stress points. A simple coordinate location check is pcr-
formed. First, the extreme values oi"the ro×lc point coordinates comprising the element are determined. For a 4 node
quadrilateral

Xmi_ = min ( x 1,x 2, x 3,x 4)

Xma x "- ??'l(1X (Xi, X2, X3, X4)

Ymin = min (Yl, Y2, Y3, Y4) (EQ 1)

• .y,,,, = max (Yl, Y2, Y3, Y4)

'0
REMAP !.7

iI_I _1r ' . iIr ..... II

 _.,,--_ I I I I I I III' [ ' I I I

Theory

Next the stress point coordinates (x s , Ys) are compared tc)the extreme values of the node txfints. The stress point is
contained by the element ii"

Xmi n _ X s _ Xma x
(EO2)
Y.,in <-Y,; <-Ymax

II"the (EQ 2) relationships arc satisticd, then the element is included in the subset of possible elements that contain the
stress point. As shown in Fig 8, the (EQ 2) condithms are necessary butnot st|fficient to identify the one element that
c:()ntains the stress Ixfint. , '

(X4, Y4)

stress point (xs , Ys) 0 _

' " "" (x2,)'2) /

FIGURE 8." A stress point satisfying (EQ 2) but located outside of the element

The z-coordinate must bc included in (EQ 1) and (EQ 2) for three dimensional meshes. Additionally, lhc coordinates
()1"ali the element nodes (i.e., 6 for a wedge, 8 for a hexahedral) must be include(I in determining the extreme values
by (EQ 1).

4.2 Fine Filter

The fine filter searches through the subs ;t ()l'elements idcntitied by tile coarse filter to lind the one element that con-
tains the stress point. The syml×)ls used in the following equations are delined in Fig 9.

V,, = V t + c,_Vl2 + (x2Vt4

V, -- V 3 -F _3V32 + _4V34 (EQ 3)

:==1=,_
O
18 REMAP
 ,, q

I II I III I ,

Theory

(Xs,Ys) -- 3_7

-7

FIGURE 9: 7his sketch illustrates the symbols used in (EQ 3)

Each o1'these vector equations can be expanded in Icmls o1'their components, This provides the following two sets o1

O simultaneous equations to besolved for (x1, o_2, c_3, and ct.4,

17,v2.....xi x4 .... xi

III
.... :1
(_,l =: ,v,..--.x

(Ea4)
- - X3 X_, - ):3

IJY'I Y3Y,_
X3 X4 Y_ [%1
0_3 =:: Y, y

If c,1, o_2, (:*3,and cx4 are greater than or equal to zero, then the stress point is within or on the surface of the elemenl.,
A way of mentally visualizing this algorithm is that ii"the stress point is within both ot"the angles/_412 and Z234,
then ii is within lhc element, This algorithm is easily extended into three di|nensions, The vector equations for an 8
node hexahedral ele|nent (Fig 10) are:

V s = V 1 -t- {XlVl2-t-(_,2V 14 "-FIX3 i5

V,,. = V:,, + (x,4V?.1 --t-(_5V3.i + (zaVt 2 (EQ 5)

V.,.::: V(, + c,TVa2+ (:_V65 + (_,_V(,-
7

V,. = V 8+ CXloV_5+ c_1lV_4 + c_l_Vs7

• Each ot' these equation can he expanded in terms of their components providing 12 simultaneous equations for the 12
unknown (x's, If ali a's are greater than or equal to zero, then the stress point is within or on the surface ot' the ele-
ment, Similar algorithms I'or a 3 node triangle, a 4 ro×lc tetrahedral, and a 6 node wedge are included in REM; _P.

REMAP 19

irl PM ........... I1'

 II I II IIIIII I I I I I I I II _

Theory

I1"the, fine filter does not find an element that contains the stress point, then the stress point lies outside the subset of
elements, The closest element tc)the stress point is then dctcrmined, The sum of the distances between lhc ,stress point
and the element nodes is calculated, The element for which this stim is a minimum is selectcd as the element closest
to the stress point,

FIGURE 10: Node point numbering for a hexahedral element

4.3 Shape Functions

q'he domain of a straight edged qtmdrilatcral c,lement is tlelined by the locations of ii.,;l'¢)t=rnode points [(xa,ya),
a= I .....41. We seek a change of coordinates which maps the given quadrilateral into the bitmit square, as depicted in
l:ig 1I. 'l'hc c{)ordinatcs (_,q) are sometimes called the "natural coordinates". The coordinates (_,rl) oi"a point in the
biunit _tlttare are related to the coordinates (x,y) by mappings of the form (see Hughes [91)

x(_,q) :: ZN.(_,q)x,",
.=
4I (EQ6)

y(_,q) = _ N,,(_,q).y,_
bl-_= [

The functions Na are the finite element shape I'tmctions, For a bilinear tluadrilatcral they are

I 1
N I : _(1-_)(l--q) N.1 --: _(1+_)(l-l-q)
(EQ7)
I 1
N:_ := _(I-t-_) (l-q) N4 = _(1 ...._)(l+q)

20 REMAP

I
 ................. i. .o.... , '-- ' I [ .... • .m "i '' , ........ IIII

Theory

lD _ ,, ,fM

FIGURE 11' Mapping of a quadrilateral Into a biunit square

II The node numbering for a triangle, 4 ni,de tetrahedral, and 6 node wedge arc shown in Fig 12, The node numbeling
for an 8 node hexahedral has previously been shown in Fig 10, The shape l'unctions for a linear triangle are

N I = _ N 2 = TI N 3 = 1 --_--'11 (EQS)

The shapefunctions l'or a linear 4 node tetrahedral are

N 2 = rI N4 = 1-_-11 ...._ (EOg)

The shape functions for a bilinear 6 node wedge are

1 1
N I : _(1-_3_ N4 : _(l-)-_)_

l :._ 1 ..'
' (EO 10)
N2 _, (.I - _)'q Ns _ (1 -)-_)'q :"
1 1
Ns: :

O
REMAP ' 21

......... _ ..... ,r, ,, ..... t' "' r, .... I1' '" ['' ..... '" 'll .... f'"tl ....... rl I I_ ] IIIII I li '1"11 li
 III I I III IIIII I • I I I I III IIIII II I ii [ I

Theory

FIGURE 12: Node point numbering for a triangle, tetrahedral, and wedge

4.4 Calculating the Natural Coordinates

For a quadrilateral, the natural c(x)rdinatcs corresponding to the sl.ross Ix)inr can be found by expanding (F_,J6) in
terms o1"lhc shape functions (EQ 7), clement node lx)int coordinatcs, and stress point coordinates, 'Fhis provides tw()
nonlinear simultaneous equations that can be solved by Newton's method l'or _ and q.

4x,.: (x l+x 24.x3+x 4)- (x l--x 2.-x_+x 4)

- (x I +x2-x3-x,O'q+ (x I -X 2 +x3-x4)_rl
4y,_= (Yt + Y2.4-Y.3+ Y4) - (Yl - Y? - Y3 + Y4) _ (EQ12)
- (Yl + Y2 - Y3 - Y4) 11+ (Yt - Y:z+ Y3 -Y4) _q

0
22 REMAP

'1
i '

_
 I I I I II II I I I ' I I I I I I II I
r

Theory

• Note th,at (EQ 12) can be solved analytically, However, in thc_computer hnplenlenlatton of the exact solution a dect-
• sion must be made ,'tsto which of the 4 roots should be kept. This decision is easy ii' fl_estress point is within tile ele-
ment. Then the roots with Values in the range [-1, 1] are retained. No cle_u'decision can be m_lde ii' the sl_'esspoint is
outside the element. Both the exact solution and Newton's method were coded. Newton's method was found to be
" more reliable in determining the roots of (EQ 12).

The procedure I'or the linear triangle, linear tetrahedral, 6 node wedge, and 8 noc_ehexahedral are similar.

4.5 ,Tempeiature Interpolation ,

The shape functions chosen t'or the coordinate transformations (EQs 7-11) can be used to interpolate or extrapolate
ttlt element node point temlx_ratures Lothe stress point (i,e., isoparametric concept) by

n
n - 3 triangle

7',. = y_ Na!,.,,a where nn = (EQ13)

' =, 46 quadrilateral
tetrahedral
a=l .
n= 8 hexahedral

R MA' 23I 'q

II1....
 = ,

IIII II I I I II I I

. References

I =iii I II I [ I li I

1 A.B. Shapiro, "TOPAZ2D - A Two Dimensional Finite Element Code tbr lleat Transfer Analysis, Electrostatics,
and Magnetoslattc Problems", University of California, Lawrence Livermore National Laboratory, Rept
UCID-20824, 1986.

2 A. B, Shapiro, "TOPAZ3D - A Three Dimensional Finite Element Heat Transfer C(xte", University of California,
Lawrence Livermore National Laboratory, Rcpt UCID-20484, 1985,

3 J_ O. Hallquist, "NIKE2D - A Vectorized Implicit, Finite Deformation, Finite Element Code for Analyzing the
Static and Dynamic Reslxmse of 2-D Solids with Interactive Rezoning and Graphics", University of Calitk)rnia,
Lawrence Livermore National Laboratory, Rept UCID- 19677, 1986.

4 J.O. Hallquist, "NIKE3D. An hnplicit, Finite Deformation, Finite Element Code for Analyzing the Static arid
Dynamic Resl-xmse of 3-D Solids", University of California, Lawrence Livermore National Laboratory, Rept
UCID.. 18822, 1984.

5 R.L. Thompson and R. L. Maffeo, "A Computer analysis Program for Interfacing Thermal and Structural
Codes", NASA Lewis Research Center, Rept NASA-TM-87021, 1985.

6 R. Maffeo, "TRANCITS Program User's Manual", General Electric Company, Cincinnati, OH, NASA Rept
CR- 174891, 1985.

7 R. Maffeo, ,'Burner Liner Thermal/Structural Load Modeling", General Electric Company, Cincinnati, OH,
NASA Rept CR- 174892, 1984.

8 A.B. Shapiro, "FACET - A Radiation View Factor Computer Code for Axisymmetric, 2D Planar, and 3D Geom-
etries with Shadowing".. University of Calil'omilt, Lawrence Livermore National Laboratory, Rept UCID-19887,
1983.

• 9 T.J.R. Hughes, The Finite Element Method, Prentice Hall, 1987.

10 J.O. Hallquist, "ORLON- An Interactive Color Post processor h_r 2-Dimensional Finite Element Codes", Uni-
. versity of Calit'ornia, Lawrence Livermore National Laboratory, Rept UCID-19310, 1985.

i
m i i ii

REMAP 25 ] _, I,_

i i'-
lr ii , plp_