Skip to main content

Full text of "NASA Technical Reports Server (NTRS) 19870017835: SINDA-NASTRAN interfacing program theoretical description and user's manual"

See other formats


NASA Technical Memorandum 100158 


SINDA-NASTRAN Interfacing 
Program Theoretical Description 
and User’s Manual 


Steven R. Winegar 
Lewis Research Center 
Cleveland, Ohio 


August 1987 


(N AS A-Tfl- 100158) SINBA-NASIBAI INTERFACING 
FECGBAfl THEORETICAL DESCRIPTION AND OSEB'S 
BANDA! (NASA) 31 p Avail: I3IS HC 
A03/HE A01 CSC! 20K 

G3/39 


N87-27268 

Dnclas 

0092890 




E- 3720 


SINDA-NASTRAN INTERFACING PROGRAM THEORETICAL DESCRIPTION AND USER'S MANUAL 


Steven R. Winegar 

National Aeronautics and Space Administration 
Lewis Research Center 
Cleveland, Ohio 44135 


SUMMARY 

Standard practice in analyzing thermal deformations or stresses in a 
structure often entails generation of finite difference thermal models, using 
a program such as SINDA (Systems Improved Numerical Differencing Analyzer), to 
predict temperatures in the structure, and generation of finite element struc- 
tural models, using a program such as MSC/NASTRAN (NAsa STRuctural ANalysis), 
to predict thermal deformations and stresses in the structure. The task of 
converting SINDA model temperature results into NASTRAN model thermal loads 
can be very labor intensive if there is not one node-to-one element, or system- 
atic node-to-element, correlation between models. 

This paper describes the SINDA-NASTRAN Interfacing Program (SNIP), a 
FORTRAN computer code that generates NASTRAN structural model thermal load 
cards given SINDA (or similar thermal model) temperature results and thermal 
model geometric data. SNIP generates thermal load cards for NASTRAN plate, 
shell, bar, and beam elements. 

The paper describes the interfacing procedures used by SNIP. SNIP uses a 
geometric search routine and a numerical coding scheme to relate thermal model 
nodes to structural model elements. SNIP then calculates element temperatures 
based on the weighted average of temperatures of the thermal nodes related to 
each element. User controlled input parameters provide control over node-to- 
element correlation. 

Sections on program set up and operation discuss the mechanics of setting 
up and running the program. Input parameters and input files are described. 
Interpretation of output file results is discussed. 

Sample cases are included to demonstrate use of the program and show its' 
performance under a variety of conditions. SNIP can provide structural model 
thermal loads that accurately reflect thermal model results while reducing the 
time required to interface thermal and structural models when compared to 
other methods. 


INTRODUCTION 

Predicting thermal distortions and thermal stresses in a structure 
requires the generation of both thermal and structural models of the structure 
under consideration. Standard practice often entails generation of finite dif- 
ference thermal models, using a program such as SINDA (Systems Improved Numeri- 
cal Differencing Analyzer), to predict temperatures in the structure, and 
generation of finite element structural models, using a program such as MSC/ 
NASTRAN (NAsa STRuctural ANalysis), to predict thermal deformations and stres- 
ses in the structure. 



Difficulty may arise, however, in converting thermal model temperature 
results into structural model thermal loads for large models. The task of 
relating specific thermal nodes to specific structural elements can be very 
labor intensive if there is not one node-to-one element, or systematic node-to- 
element, correlation between models. 

This paper describes the SINDA-NASTRAN Interfacing Program (SNIP). SNIP 
is a FORTRAN computer program that generates NASTRAN structural model thermal 
loads from thermal model temperature results. SNIP correlates thermal nodes 
to structural elements to interface SINDA (or similar) finite difference ther- 
mal models with NASTRAN finite element structural models made up of plate, 
shell, bar, and beam elements. Node-to-element correlation includes de|ermin- 
ing which SINDA nodes should be related to each NASTRAN element and calculat- 
ing a weighting factor for temperatures associated with each element-related 
thermal node. 

SNIP uses thermal model geometry, information that must be combined with 
standard thermal model temperature results, and structural model geometry, 
information that is available from standard NASTRAN input, to search through 
three-dimensional space around each structural element for the nearest thermal 
nodes. Thermal and structural models must both be defined in the same, single 
Cartesian coordinate system. The thermal nodes located nearest each element 
are used to determine element temperature for thermal distortion and stress 
analysis. Shaping and user-controlled sizing of the three-dimensional search 
region, along with numerical coding of thermal node and structural element num- 
bers, provide for separation of substructures during correlation. 

SNIP can provide structural model thermal loads that accurately reflect 
thermal model results while reducing the time required to interface thermal 
and structural models when compared to other methods. 


DESCRIPTION 

Use of SNIP requires the generation of SINDA (or similar thermal ana- 
lyzer) and NASTRAN models of the hardware under study. Note that good deal of 
coordination is required between analysts generating the two models. 

Input to the program is a file of thermal model temperature results and 
physical location of each thermal node in three-dimensional space, combined in 
a SNIP-unique format. The thermal node physical location data required by SNIP 
is not a standard thermal model output, and must be generated independently. 

If temperature gradients exist through the structural elements (determined per- 
haps by the use of multiple thermal nodes in the thermal model), the gradients 
must be calculated prior to input to SNIP as gradients at specific thermal node 
locations. Note that in order to determine the proper sign and orientation of 
temperature gradients in the structural elements the thermal analyst must be 
familiar with the structural model geometry. 

Also input to SNIP is a standard NASTRAN input deck for a model made up 
of plate, shell, beam and bar elements. SNIP supports the CTRIA, CQUAD, CBAR, 
and CBEAM elements of NASTRAN. Input parameters, adjusted in the program 
source code by the user, control the node-to-element correlation. 


2 


SINDA-NASTRAN Interfacing Program (SNIP) is, to some degree, a misnomer. 
Though the program was originally written to use SINDA temperature results, 
temperature data from any source can now be used as long as it is entered in 
the correct, SNIP-unique format. 

For NASTRAN plate and shell elements SNIP searches through three- 
dimensional space to find the thermal nodes nearest each element in each of 
four quadrants in the element plane. For NASTRAN bar and beam elements SNIP 
searches through three-dimensional space to find the thermal nodes nearest 
each element end (grid point) in each of the two directions along the element 
axis. The distance from an element, or element end, to each of the nearest 
thermal nodes is used to determine a weighting factor for temperatures at 
those nodes. 

Output from SNIP are NASTRAN element temperature load cards for each ele- 
ment and NASTRAN case control cards for each temperature load set. Also out- 
put by SNIP is a list of elements with the numbers of the SINDA nodes related 
to each element, and the weight given each node in temperature calculations. 


PROGRAM PROCEDURES 

The program uses a combination of techniques to relate thermal nodes to 
structural elements. The primary technique is to search in space for the 
nodes located nearest each element. However, since closely spaced nodes are 
not necessarily related (they may be nodes of two closely spaced, but separate, 
parts of a structure), a technique to shape and reduce the three-dimensional 
search region is used. The reduced, shaped search regions are called "qualifi- 
cation" regions and have a characteristic shape for each element type. In 
addition to qualification regions, a numerical coding scheme provides for keep- 
ing node-to-element correlation in order. 

The correlation routine logic is depicted in figure 1. Every SINDA (ther- 
mal) node is checked against each NASTRAN element for correlation. If the 
node and element numbers are coded to allow a correlation, then the geometric 
search routine is employed to determine whether a thermal node is within the 
element qualification region and whether the thermal node is the nearest node 
in the region segment. Once node-element correlation has been determined for 
every element a weighting scheme is used to calculate element temperatures 
from thermal node temperatures. Qualification regions, numerical coding of 
thermal node numbers and structural element numbers, and the temperature 
weighting scheme are described below. 


Qualification Regions 

Qualification regions define the shape and size of the space in which the 
program will search around an element for thermal nodes. Thermal nodes out- 
side the qualification region for an element are considered too far from the 
element to be useful in calculating its temperature. Qualification regions 
differ for different element types (i.e., plate and shell element qualifica- 
tion regions are shaped differently than bar and beam element qualification 
region) . 


3 


For plate and shell elements, the objective of the correlation is to 
determine the average temperature of, and the temperature gradient through, 
each element, to be output on a NASTRAN TEMPP1 load card for each element 
(ref. 1). Thermal nodes may carry some thermal gradient data, which must have 
the correct sign with respect to the plate or shell elements to which the 
nodes will be related . 

Qual if i cation regions for plate and shell elements (fig. 2) are disc- 
shaped regions centered at the element centroid. The central plane of the 
disc-shaped region is the plane defined by the element centroid and the first 
two corners (structural nodes) listed on the NASTRAN element connectivity card 
for the element. The disc is separated into four quadrants (qualification 
region segments) using the element coordinate system. The nearest thermal 
node in each region quadrant is determined during the search operation and any 
empty region quadrant is noted. 

The radius and thickness of the qualification regions are the same for 
all plate and shell elements, and are set up by user adjustable program input 
parameters. Parameter ALWDR in the program sets the radius and parameter 
ALWDTH sets the half-thickness of the search region. 

For bar and beam elements the objective of the correlation is to deter- 
mine the average temperature of, and temperature gradients in the element Y 
and Z directions (fig. 3) of, each end of each bar and beam element, to be 
output on a NASTRAN TEMPRB load card for each element (ref. 1). Gradients 
input as thermal node data must have the correct sign with respect to the bar 
or beam elements to which the nodes will be related. 

Qualification regions for bar and beam elements (fig. 4) are box-shaped 
regions, one for each end of the element, centered at the centroid of the ele- 
ment end. One side of the box-shaped region is parallel to the element 
X-axis. Each box is separated into two halves (qualification region segments) 
at the end of the element. The nearest node in each half is found during the 
search operation and any empty half is noted. 

The X, Y, and Z dimensions of the box-shaped qualification regions are 
the same for both ends of all bar and beam elements, and are set up by program 
input parameters. Parameters ALWDR2, ALWDY, and ALWDZ set the half-size of 
the search region in the element X, Y, and Z directions respectively. 

Qualification regions allow the analyst to size the search region for all 
elements. However, there is still potential for temperature data from a ther- 
mal node meant to be related to one part of a structure to be used to deter- 
mine temperature of a different part of the structure. For example, a plate 
structure supported by backing rib stiffeners may lead to overlapping qualifi- 
cation regions (fig. 5). Should a thermal node lie in the overlapped region 
it may be related to both a plate element and a bar element by the program. 

This situation does not create a problem if no temperature gradient data is 
passed from SINDA and the nodal temperature data is truly meant to be used for 
both the surface and backing structure. However, if the node is specifically 
to be related only to plate or only to bar elements (carrying appropriate tem- 
perature gradient data) a potential problem exists. Numerical coding can alle- 
viate this problem. 


4 


Numerical Coding 


Numerical coding is the use of thermal node numbers and structural ele- 
ment numbers to separate substructures. Appropriate numbering of thermal 
nodes in input data and structural elements in the input NASTRAN model identi- 
fies substructures. If thermal nodes are not numbered for correlation with a 
particular element, SNIP will not determine the thermal nodes physical (geomet 
ric) relation to the element to consider the node for correlation. 

Numerical coding of input thermal model node numbers and NASTRAN element 
numbers is depicted in figure 6. Four input parameters, NVAR1 , NVAR2, NVAR3, 
and NVAR4 control the coding operation. SINDA nodes numbered above NVAR1 are 
related only to plate and shell elements numbered above NVAR2. Nodes numbered 
below NVAR1 are related only to plate and shell elements numbered below NVAR2. 
SINDA nodes numbered below NVAR3 are related only to bar and beam elements num 
bered below NVAR4. Nodes numbered above NVAR3 are related only to bar and 
beam elements numbered above NVAR4. The coding scheme allows the analyst to 
separate substructures from each other for correlation. 

One potential way to separate the supporting beam structure from a plate 
structure is depicted in figure 7. First, number all plate elements below 
NVAR2 and all supporting beam structure elements above NVAR4. Second, set 
NVAR1 and NVAR3 equal. Third, number SINDA nodes for the plates below NVAR1 
and nodes for the beams above NVAR3. This will result in complete separation 
of the backing and surface structures (fig. 8) during correlation. 

Overall, the numerical coding scheme and the qualification region search 
technique combine to give the analyst good control of the node to element 
correlation. 


Weighting Scheme 

Once the appropriate thermal nodes have been found for each structural 
element, and the characteristic distances for each node determined, a weight- 
ing factor for each node related to each element is calculated. The character 
istic distances are element centroid to thermal node location for plate and 
shell elements, and centroid of element end to thermal node location for each 
end of bar and beam elements. 


The formula used to calculate weighting factors is: 


weighting factor No(Je x 



where 

R x is the characteristic distance for node x, one of the nodes related to 
the element under consideration 

Rj is the characteristic distance for node i, each of the nodes related to 
the element under consideration 


5 


N is the number of nodes related to an element or element end (= number of 

qualification region segments) 

N = 4 for plates and shells; N = 2 for ends of bars and beams. The sum of 
the nodal weights for each element or element end is 1.0. For qualification 
region segments in which no correlating thermal node is found, the distance 
to an imaginary node 0 is set very large so that the weighting factor 
for the node is approximately 0.0. The default characteristic distance if no 
thermal node is found is set by parameter XLARGE, which must be three orders 
of magnitude larger than the largest characteristic distance. This is varia- 
ble in order to avoid machine memory problems that may be encountered when 
working with a variety of machines. 

The NASTRAN input file must contain a standard NASTRAN input deck in the 
single field format. All grid point definition and element connectivity must 
be done with separate, explicit cards (i.e., column duplication and generation 
commands cannot be used). Only plate, shell, beam, and bar elements (CTRIA3, 
CTRIA6 , CQUADA4 , CQUAD8, CBEAM, and CBAR) can have temperature load cards gen- 
erated by SNIP. Also, columns one and ten of the cards must have their charac- 
ter fields left justified. 

SNIP places some further requirements on the thermal and NASTRAN models. 
The NASTRAN model must be defined in a single Cartesian coordinate system. In 
addition, the thermal model results must be described in the same single Carte- 
sian coordinate system. 

Input and output data device numbers are defined in the data section of 
the program code, and can be changed there by the analyst. NASTDECK is the 
device number for the NASTRAN input deck, INSINDA is the device number of for 
the SINDA input file, and ISCRTCH1, ISCRTCH2, ISCRTCH3, and IGRDHLD are scratch 
files for intermediate data storage. 


Output Files 

Outputs include a file of NASTRAN case control cards, a file of NASTRAN 
temperature load cards, and a file of node-element correlation information. 
Reference 1 describes use of the NASTRAN case control and temperature load 
cards. The case control file holds a subcase card, a label card, and a temper- 
ature load set number card for each thermal load subcase. Up to 10 thermal 
load subcases can be processed by the program in a single run. The tempera- 
ture load card file holds temperature load cards for each element for each 
case. TEMPP1 cards are for plate or shell elements and TEMPRB cards are for 
bar or beam elements. The correlation information file (fig. 9) lists each 
structural element by element type. It lists the thermal nodes related to 
each element, and the weight given temperature and temperature gradients at 
each node for each element. 

IOCHECK is the device number of the output file of node-element correla- 
tion and weighting factors to be checked by the analyst. INASSUB is the 
device number of the output file of NASTRAN case control cards, and INASTEMP 
is the device number of the output file of NASTRAN temperature load cards. 

Results of the weighting factor calculations are used to weight tempera- 
tures calculated for an element or element end as: 


6 


N 

Temperature e|(!Bent or end of e , ement - E <W-F>„ ode , * <temperature> Node , 


where 

Node i are the nodes related to the element under consideration 
W.F. is the weighting factor 

N is the number of nodes related to an element or element end (= number 

of qualification region segments) 

N = 4 for plate or shell elements; N * 2 for bar or beam element ends. The 
same equation is used to calculate element temperature gradients, with tempera- 
ture gradient substituted for temperature in the equation. 

A table of the NASTRAN elements with the related SINDA nodes and the 
weighting factors for those SINDA nodes is output for the analyst to check 
(fig. 9). A warning message is included for any element for which no thermal 
nodes were found in the correlation routine. This file shows explicitly the 
node-element relationships used by SNIP to generate NASTRAN thermal load cards. 


PROGRAM OPERATION 

Use of SNIP requires the generation of SINDA, or similar thermal, and 
NASTRAN models of the hardware under study. Coordination is required between 
the analysts generating the two models. Adjustment of input parameters in the 
SNIP source code allows the SNIP analyst to control the node-to-element corre- 
lation performed by the program. The output of SNIP are set of temperature 
load cards of each thermal load case, case control cards for each case, and a 
table of node-element correlation and associated weighting factors. 


Input Parameters 

Input parameters are used to control node-to-element correlation as des- 
cribed in PROGRAM PROCEDURES. Two additional parameters, IDIMI and IDIM2, 
dimension the arrays used in SNIP. These allow for changing memory require- 
ments depending on problem size (problem size is limited on the PC). IDIMI 
sets the dimensions on most NASTRAN-related arrays, and should be set equal to 
the larger of the number of NASTRAN grid points and the number of NASTRAN ele- 
ments in the input NASTRAN file. IDIM2 sets the dimensions on most thermal 
model -related arrays, and should be set equal to the number of thermal nodes 
in the input thermal model results file. 

To change the parameters one must change the parameter settings in the 
PARAMETER statements of the SNIP source code and recompile the source code. 
Program comments in the code describe each parameter. Appendix A shows the 
variables and parameters lists in the comments section of the SNIP source file 
and shows the first 35 lines of the SNIP program. The first 35 lines contain 
all program PARAMETER and DATA statements. 


7 


Input Files 


Inputs to the program include a file of thermal model temperature results 
in a SNIP-unique format, and a standard NASTRAN input deck in the single field 
format . 

The thermal model results input file must contain, for each case, the run 
title, a time associated with the results, the number of node data cards for 
the case, and the node data cards. The run title and time are used only to 
generate NASTRAN subcase labels. Node data cards must contain, in order, the 
thermal node number, average node temperature, temperature gradient in the 
element Z direction at the node, X, Y, Z coordinates of the center of the 
node in space, and temperature gradient in the element Y direction at the 
node. For nodes with no associated temperature gradients a 0.0 gradient must 
be entered. At present no standard SINDA subroutine exists that produces tem- 
perature results in the SNIP-unique input format. Note that in order to deter- 
mine temperature gradients in the element Y and Z directions the thermal 
analyst must be familiar with the structural model geometry. Figure 10 shows 
the SINDA results input file and formats required by SNIP. 


SAMPLE CASES 

Four sample cases show various features of SNIP and its operation. Case 1 
shows the basic operation of the program for a simple temperature pattern in a 
structure. Case 2 shows the results of program operation for a fairly complex 
temperature pattern. Cases 3 demonstrates use of coding to separate substruc- 
tures. Case 4 demonstrates us of qualification regions to separate 
substructures . 


Case 1. - Basic Performance for a Simple Temperature Pattern 

Case 1 shows SNIP performance for a simple temperature pattern in a struc- 
ture. Case 1 consists of a square plate under a uniform temperature gradient 
across the plate, from corner to corner. Figure 11 shows the NASTRAN grid, a 
simple 10 by 10 grid of square 1 by 1 in. elements. Figure 12 shows the SINDA 
node layout on the plate. The locations of the SINDA nodes are their physical 
centers (i.e., node 1 is located at x = 1.67 in., y = 1.67 in., z = 0.0 in.). 

Table I is the thermal model results input file for the temperature pat- 
tern of figure 13. There are no temperature gradients through the plate for 
Case 1. The relevant input parameters for Case 1 are: ALWDR = 10 in., ALWDTH 

= 0.01 in., X LARGE = 1x10$ in., NVAR1 « 500, NVAR2 - 500, NVAR3 = 500, NVAR4 = 
500. Figure 14 shows the temperatures of the structural elements, as calcu- 
lated by SNIP. One would expect elements along lines parallel to a line from 
the upper left-hand corner at element 91 to the lower right-hand corner at ele- 
ment 10 to have approximately equal temperatures calculated by SNIP. SNIP cal- 
culates 75° temperatures for elements 10, 19, 28, 37, 46, 55, 64, 73, 82, and 
91, as expected. A line from element 61 to element 7 should show a nearly con- 
stant temperature sightly greater than 50°and does. SNIP generates a gradient 
along a line from element 2 to 100, as expected. SNIP produces a good tempera- 
ture pattern in the structural model, representative of the temperature pat- 
tern from the thermal model results for Case 1. 


8 



Figure 15 depicts the physical node-element relationships for element 
number 21 in Case 1. Qualification region segment 1 contains nodes 4 to 9. 

Node 4 Is the closest of those in segment 1 and, therefore, is the node used 
for element temperature calculation. Similarly, only node 1, of nodes 1 to 3, 
which all lie in segment 4, is used for element temperature calculation. Ri 
and R 4 are the characteristic distances from element 21 to the nodes in seg- 
ments 1 and 4 respectively. No nodes are found in segments 2 and 3, and so 
R 2 and R 3 to an imaginary node 0 are set equal to parameter XLARGE. Note 
that node 1 is closer to element 21 than node 4, and is, therefore, given more 
weight in element temperature calculation (table II). 

Note that ALWDR is set large in Case 1 so all thermal nodes will be con- 
sidered in the search routine performed for each element. If ALWDR had been 
set smaller, 2.2 in. for example, only closely located nodes would have been 
used for temperature calculation at each element. Table II shows the node- 
element correlation and weighting factors for some elements with ALWDR = 10 in. 
for Case 1. Table III shows the node-element correlation and weighting fac- 
tors for the same elements with ALWDR = 2.2 in. for Case 1. Figure 16 shows 
the temperatures of the structural elements calculated by SNIP for Case 1 with 
ALWDR = 2.2 in. The result for a small ALWDR is that temperature calculations 
are dominated by the nodes near each element. 


Case 2. - Performance for a Complex Temperature Pattern 

Case 2 demonstrates SNIP performance for a complex temperature pattern in 
a structure. Case 2 consists of the same structure and thermal nodes as Case 
1 under a unique temperature pattern. Figure 17 shows the temperature pattern 
of the thermal model for Case 2 . The pattern represents a case with a heat 
source at node 6 and a heat sink at node 4. Input parameters for Case 2 are 
the same as for Case 1, with ALWDR = 10 in. 

Figure 18 shows the temperatures of the structural elements as calculated 
by SNIP for Case 2. One would expect a constant temperature line at x = 5 in., 
and nearly constant temperatures slightly greater than 50° from element 6 to 
element 96. SNIP generates expected results. Figure 19 shows what could be 
considered isothermal lines based on the thermal model input. Figures 18 and 
19 are consistent. The Case 2 results show that SNIP can generate reasonable 
temperature patterns in a structural model for complex input temperature pat- 
terns. 


Case 3. - Use of Coding to Separate Substructures 

Case 3 shows how coding can be used to separate substructure. Case 3 
adds to the structure of Case 1 and 2 a line of bar elements along y = 5 in., 
offset from the plate 0.5 in. in the +Z direction (fig. 20). SNIP input 
thermal nodes 1001, 1002, and 1003 represent the bar structure and are cen- 
tered 0.5 in. above nodes 4, 5, and 6 , respectively. Figure 21 shows the ther- 
mal model temperature pattern. Input parameters for Case 3 are the same as 
those for Casel and 2 with the addition that ALWDR2 = 10 in., ALWDY = 0.5 in., 
and ALWDZ = 1 in. ALWDZ is purposely set large enough to emcompass some plate 
thermal nodes. 


9 



Parameters NVARl , NVAR2 , NVAR3, and NVAR4 are all set equal to 500, which 
allows the numbering of the bar elements and the bar thermal nodes to com- 
pletely separate the bar and plate structures during node-to-element correla- 
tion. As a result thermal nodes 1001 to 1003 are related only to element 501 
to 510, and thermal nodes 1 to 9 are only related to elements 1 to 100 in the 
correlation scheme, though some nodes lie in overlapping qualification regions 
for Case 3. The temperatures along the bar structure, as calculated by SNIP, 
are: element 501 (x = 0 in. end) = -20°, element 503 (x = 2 in. end) = -15°, 

element 505 (x = 4 in. end) = 15°, element 507 (x = 6 in. end) = 45°, element 
509 (x = 8 in. end) = 75°, and element 510 (x = 10 in. end) = 80°. The output 
temperature pattern of the plate is identical to that of Case 2, as expected. 

The temperature cards generated by SNIP for Case 3 accurately reflect the 
thermal model results fed to the program. 


Case 4. - Use of Qualification Regions to Separate Substructures 

Case 4 demonstrates how qualification regions can be used to separate sub- 
structures. Case 4 adds to the structure of Case 3 a second line of bar ele- 
ments along y =* 4 in., offset from the plate 0.5 in. in the +Z direction 
(fig. 22). SNIP input thermal nodes 1011 to 1013 represent this additional 
backing structure. Locations of the nodes are 101 1 — (0 , 4, and 0.5 in.), 

1 01 2- ( 5 , 4, 0.5 in.), and 1 01 3-< 1 0 , 4, 0.5 In.). Temperature input data are 
the same as Case 3 with the addition of temperatures for nodes 1011, 1012, and 
1013 which are -15, 35, 85° respectively. Input parameters for Case 4 are the 
same as those for Case 3. Note that ALWDY = 0.5 In. 

Figure 23 depicts the qualification region of the x = 2 in. end of ele- 
ment 512. Note that, because of the size and shape the region, the potential 
problem of using temperature data from node 1001 for calculation of the temper- 
ature of the x = 2 in. end of element 512 is precluded. Temperature data from 
thermal node 1011 is used at the x = 2 in. end of element 512 even though ther- 
mal node 1001 is physically closer. Table 4 shows that thermal nodes 1001, to 
1003 are related only to elements 501 to 510, and thermal nodes 1011 to 1013 
are related only to elements 511 to 520 by the correlation routine. Table 5 
shows the resulting temperatures in the beam structures for Case 4. 

One note of caution: The bar element X, Y, and Z directions happen to 

correspond to the global X, Y, and Z directions in this example. ALWDR2 , 
ALWDY, and ALWDZ are the half-sizes of the box-shaped qualification region in 
the element coordinate system (figs. 2 and 3). 


PROBLEM SETUP / RESULTS INTERPRETATION / OPERATING PROCEDURES 

The analyst faces determination of qualification region sizes and choice 
of a substructure separation method (qualification regions with or without 
numerical coding) when setting up a SNIP run. 

The spacing between thermal nodes must be considered in sizing the quali- 
fication regions. Case 1 results above show that sizing qualification regions 
large enough to envelope nodes in most or all of the region segments leads to 
smooth gradients in the temperature pattern across the structural model. 

Sizing qualification regions small, such that they envelope only the nodes 

10 


nearest an element, leads to a structural model temperature pattern that 
clearly matches the layout of thermal nodes on the structure, without smooth 
gradients across the entire structure. The SNIP analyst must keep these 
results in mind when sizing qualification regions. 

The choice of a substructure separation method depends primarily on model 
geometry. In the case where model geometry is simple enough for qualification 
regions alone to keep substructures separate (i.e., there is little intertwin- 
ing or closely spaced substructure) coding should not be used. This allows 
complete freedom in node and element numbering for SINDA and NASTRAN model- 
ers. In the case where the structure is complex (i.e., there is a significant 
amount of intertwining or closely spaced substructure) coding should be used, 
in addition to qualification regions, to separate substructures. 

Potential problems the SNIP analyst faces include inappropriate qualifica- 
tion region sizing, inconsistency between model geometries, and incompatibil- 
ity in node and element numbering (coding). The SNIP node-element correlation 
file provides a warning message if any element is without any correctable 
nodes. This file holds the key to assuring that SNIP is performing well. 

Note that it is important to carefully review this file as it is the primary 
indicator of how SNIP has performed. 

If many elements have few or no related thermal nodes, there is a possi- 
bility of: (1) undersized qualification regions, (2) inconsistency between 

SINDA and NASTRAN model geometries, or (3) node and element number coding 
incompatibility. The key to good SNIP performance is good coordination between 
SINDA and NASTRAN modelers to assure geometric consistency and coding compati- 
bility. The SNIP analyst must understand both models, and understand com- 
pletely the operation of SNIP, in order to set up and control operation of 
SNIP and to provide appropriate guidance to thermal and structural modelers. 

In order to completely understand SNIP operation of the SNIP analyst 
should become familiar with the source code. The source code contains exten- 
sive comments in order to guide the user through its 1 operation. 

SNIP is written in ANSI standard FORTRAN. Two versions of the program 
currently exist. One version runs on a IBM 3270 PC-AT and the other runs on a 
CRAY X-MP. The only difference in the two programs is that OPEN statements in 
the source code of the PC version perform the function of ASSIGN statements 
that must be present in the CRAY job Control Statement File when running the 
Cray version (ref. 2). Assign statements must be included in the user- 
supplied CRAY Control Statements File for each SNIP input and output file when 
the CRAY version is run. 

For running the PC version the NASTRAN input deck must be in a file named 
NASTRAN. CDS, and the thermal model results must be in a file named SINDA. CDS. 
The PC version will write temperature load cards into a file named TMPTR.CDS , 
will write subcase control cards into a file named SUBCASE. CDS, will write the 
node-element correlation and weighting factors list into a file named 
SNIPIO.CHK. Appendix B is a SNIP operation checklist for running the PC ver- 
sion of the program. 

For running the CRAY version of SNIP, CRAY Job Control Language (JCL) 
statements must be included in the Control Statements File to put NASTRAN and 
thermal model input files into CRAY files and assign the files the appropriate 

11 


FORTRAN unit number so SNIP can access the files. NASTRAN input cards must be 
in unit NASTDECK, and thermal model input cards must be in unit INSINDA. 

CRAY JCL must also name, and assign appropriate FORTRAN unit numbers to, 
output files and intermediate data storage files. FORTRAN units ISCRTCH1, 
ISCRTCH2 , ISCRTCH3, and IGRDHLD must be available for intermediate data stor- 
age. NASTRAN temperature load cards will be written into unit INASTEMP, 

NASTRAN subcase cards will be written into unit INASSUB, and the node-element 
correlation and weighting factors list will be written into unit IOCHECK by 
SNIP. The FORTRAN unit numbers are defined in source code data statements, 
and can be changed there by the analyst (see appendix A). Appendix C is a sam- 
ple run file, including CRAY JCL, for running the CRAY version of SNIP. 

Appendix D is a SNIP operation checklist for running the CRAY version of 
the program. 


APPLICATIONS 

SNIP can be applied to large thermal and structural models (hundreds or 
thousands of thermal nodes and structural elements) made up of plate, shell, 
bars, and beam structures. Program operation presumes a NASTRAN element mesh 
equal to or finer than the thermal model node mesh. This presumption is based 
on thermal modeling limitations when considering radiative heat transfer, as 
one must when analyzing structures in the space environment. A situation with 
a thermal node mesh equal to or greater than the finite element mesh is assumed 
to be one in which one node-to-one element or systematic node-to-element corre- 
lation can be exercised. 

SNIP has been applied to thermal deformation analysis in satellite 
antenna reflectors with good results (ref. 3). The time required to feed ther- 
mal analysis results to structural analysis models using SNIP is small when 
compared to other methods. One indirect advantage of using SNIP is the high 
degree of coordination and communication that takes place between thermal and 
structural modelers using this approach. This can lead to early detection of 
errors in either or both models. 

SNIP is currently in use for analysis of solar concentrators for space 
solar dynamic electric power generation systems. Large thermal and structural 
models are required to accurately predict thermally distorted surface shapes. 
SNIP is expected to save a significant amount of time and effort in combining 
concentrator thermal and structural models. 


CONCLUDING REMARKS 

Potential future enhancements to SNIP include the addition of more diag- 
nostic messages, change in the method of parameter control, automation of qual- 
ification region sizing, addition of different weighting schemes, addition of 
the capability to use standard SINDA output, and addition of other element 
capabi 1 i ty. 

More diagnostic message capability could be added to the program to indi- 
cate potential problems and solutions. Control of SNIP parameters may be 


12 



changed from PARAMETER statements In the source code to an input file, so the 
program need not be recompiled each time parameters are changed. 

Qualification region sizing could be automated using a combination of 
SINDA node number coding information and the average SINDA node spacing. This 
could reduce the amount of work required of the SNIP analyst. 

Additional user-controlled, and perhaps even user-written, weighting 
schemes could be added. This would allow more user control of the node- 
element relationships. 

Addition of the capability to use standard SINDA output could be added. 
This would eliminate the need for reformatting SINDA output temperature 
results, though thermal model geometric data would still be added separately. 

Finally, the capability to handle additional NASTRAN elements, such as 
solid elements, could be added. This would make SNIP applicable to a larger 
number of thermal -structural problems. 

SNIP is, however, a very powerful analytical tool as it stands today. 

SNIP is useful for combining thermal and structural analysis for which there 
is not one node-to-one element, or systematic node-to-element correlation 
between models. SNIP can provide structural model thermal loads that accu- 
rately reflect thermal model results through the use of a geometric search rou- 
tine and a numerical coding scheme. 

SNIP requires the addition of geometric data to standard thermal model 
results in order to relate thermal and structural models. Though the addition 
of geometric data requires additional effort on the part of the thermal ana- 
lyst, the overall effort required to use SNIP is, for most large models, sig- 
nificantly smaller than that required to interface the models manually. 


13 


nnnnonnnnnnnnnnnnnonnnnnonnnnnnnnnnnnnnnonnnnnnnnnnonnnnnnn 


APPENDIX A 


SNIP Source Code Variable and Parameter Lists, and PARAMETER and DATA Statements 


XXXXKXXXXXXXKXXXXX VARIABLE LIST xxxxxxxxxxxxxxxxxxxxxxxxxx 
CARDS -ARRAY CONTAINING THE INPUT NASTRAN CARDS 
IQUADS -ARRAY CONTAINING FIELDS 2 THROUGH 7 OF CQUAD4 
AND CQUAD8 CARDS FROM NASTRAN 
ITRIAS -ARRAY CONTAINING FIELDS 2 THROUGH 6 OF CTRIA3 
AND CTRIA6 CARDS FROM NASTRAN 

BARS -ARRAY CONTAINING FIELDS 2 THROUGH 10 OF CBAR AND CBEAM 
CARDS FROM NASTRAN 

IDS -ARRAY CONTAINING THE BEAM AND BAR ELEMENT NUMBERS 
PLUS -ARRAY CONTAINING FIELDS 1 AND 4 THROUGH 9 OF CONTINUATION 
CARDS FROM NASTRAN 

IGRIDS, GRIDS -ARRAY CONTAINING FIELDS 2 THROUGH 6 OF GRID 
CARDS FROM NASTRAN 

IPTRQ, ITPTRT, IPTRA, IPTRB -POINTER ARRAYS CONTAINING NODE NUMBER 

OF SINDA NODES ASSOCIATED WITH EACH 
QUAD, TRIA, A END, AND B END OF NASTRAN 
ELEMENTS 

QFACTOR, TFACTOR, AFACTOR, BFACTOR -ARRAYS CONTAINING WEIGHTING 

FACTORS ASSOCIATED WITH SINDA 
NODES IN POINTER FILES ABOVE 

CNTROID -X, Y , Z POSITION OF PLATE AND SHELL ELEMENT CENTROIDS 
T, DELT -TEMPERATURE AND TEMPERATURE GRADIENT FILES FOR PLATE 
AND SHELL ELEMENT TEMPERATURE CALCULATION 
VECT1 , VECT2, VECT3 -VECTORS USED FOR CHECKING SINDA NODE 
QUALIFICATION 

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx 


xxxxxxxxxxxxxxxxxxxxxxPARAMETER LISTxxxxxxxxxxxxxxxxxxxxxxxx 
NVAR1 - CODING PARAMETER FOR SINDA NODES TO BE RELATED TO 
PLATE AND SHELL ELEMENTS 

NVAR2 - CODING PARAMETER FOR NASTRAN PLATE AND SHELL ELEMENTS 
NVAR3 - CODING PARAMETER FOR SINDA NODES TO BE RELATED TO 
BAR AND BEAM ELEMENTS 

NVAR4 - CODING PARAMETER FOR NASTRAN BAR AND BEAM ELEMENTS 
ALWDR - PLATE AND SHELL ELEMENT QUALIFICATION REGION RADIUS 
(OF DISC-SHAPED REGION) 

ALWDTH - PLATE AND SHELL ELEMENT QUALIFICATION REGION 
HALF-THICKNESS COF DISC-SHAPED REGION) 

ALWDR2, ALWDY, ALWDZ - BAR AND BEAM ELEMENT QUALIFICATION 

REGION HALF-SIZES IN THE X, Y, AND Z 
DIRECTIONS RESPECTIVELY 

XLARGE - DEFAULT RADIUS TO IMAGINARY NODE NUMBER "0" WHEN NO 
NODE IS FOUND IN A PARTICULAR QUALIFICATION REGION 
SEGMENT (THIS IS VARIABLE FOR MACHINE ADAPTATION 
PURPOSES) 

IDIM1 , IDIM2 - ARRAY DIMENSIONING PARAMETERS. THESE ALLOW FOR 
CHANGING MEMORY REQUIRMENTS DEPENDING ON 
PROBLEM SIZE. IDIM1 SETS DIMENSION ON MOST 
NASTRAN-RELATED ARRAYS. IDIM2 SETS DIMENSION 
ON SINDA-RELATED ARRAYS. 

XXXXXXXXXXXXXXX XX XXX XXXXXXXXXXXXX XX XX XX xxxxxxxxxxxxxxxxxxxxxxxxxx 


PROGRAM INTRFACE 

PARAMETER (NVAR1=200,NVAR2=2Q0 ,ALWDR=10 . , ALWDTH= . 001 ) 

PARAMETER ( ALWDZ= . 5, ALWDY= . 5, NVAR3=50Q , NVARA=5Q0 , IDIM1=300 ) 
PARAMETER (ALWDR2=10 . , IDIM2=300,XLARGE=1 . E+5) 

DIMENSION CARDS(3XIDIM1,10) 

DIMENSION SINCD(IDIM2,7), IPTRQC IDIM1 , 4) , IPTRT ( IDIM1 , A ) , NUMBER! 4 ) 
DIMENSION GRIDS(IDIM1,5),IQUADS(IDIM1,6),ITRIA$(IDIM1,5) 
DIMENSION IGRIDS(IDIM1,2), CNTROIDC 3 ) , VECT1 C 3) , VECT2( 3 ) , UNORMC 3) 
DIMENSION ARE( A ) , Q FACTOR ( I DIM1 , <♦) , TFACTOR ( I DI Ml , A) 

DIMENSION T(‘4),PLUS(IDIM1,7),BARS(IDIM1,9), ZAX( 3 ) , YAX( 3 ) , DELTC A) 
DIMENSION IPTRA( IDIM1,2),IPTRB(IDIM1,2), AFACTOR ( IDIM1 , 2) 
DIMENSION BFACTOR ( IDIM1,2),IDS(IDIM1), XAXC 3) , VECT3C 3 ) 

DIMENSION I$INCD( IDIM2 ) 

INTEGER 0 

CHARACTER BARX8 , BEAMxg , QUAD4X8 , QUAD8*8 , TRIA3X8 , TRIA6#8 , GRID*8 
CHARACTER ASCX8 , CARD$*8 , BARSX8 , PLUS*8 , SINNAMEX8 , TIME#8 
EQUIVAL ENCEC SINCD(1,1),ISINCD(D) 

DATA BAR/ 'CBAR •/ 

DATA BEAM/ 'CBEAM '/ 

DATA QUAD4/ ' CQUAD4 V 
DATA QUADS/ 'CQUAD8 '/ 

DATA TRIA3/ 'CTRIA3 '/ 

DATA TRIA6/ ' CTRI A6 '/ 

DATA GRID/ 'GRID V 
DATA NASTDECK/ 30 / 

DATA INSINDA/ 25 / 

DATA ISCRTCH1/ 50 / 

DATA ISCRTCH2/ 60 / 

DATA ISCRTCH3/ 90 / 

DATA IGRDHLD/ 80 / 

DATA IOCHECK/ 19 / 

DATA INASSUB/ 45 / 

DATA INASTEMP/ 46 / 

1 = 1 
M = 1 
N = 1 
0 = 1 


ORIGINAL PAGE IS 
OE POOR QUALITY 


14 



APPENDIX B 


SNIP Operations Checklist - PC Version 

1 - Place input NASTRAN Bulk Data Deck in a file named NASTRAN . CDS 

2 - Place input thermal model results in a file named SINDA.CDS 

3 - Adjust input parameters in SNIP source code PARAMETER statements (see 

appendix A) 

4 - Compile source code 

5 - Link and run SNIP code 

6 - Review resultant node-element relationships in SNIP output file named 

SNIPIO.CHK (If relationships are not acceptable, review and revise inputs 
to 1 , 2, and 3 above as appropriate and run SNIP again) 

7 - NASTRAN temperature load cards are in the SNIP output file named TMPTR.CDS 

and NASTRAN case control cards are In the SNIP output file named 
SUBCASE. CDS 


15 


APPENDIX C 


SNIP Sample Run File - CRAY Version 


JOB, JN s SNIP RUN.T=1 5.MFLs6nnnnn 

account , APM=MHB. 

ASSIGN, DN=NASFIL E, DC=SC,A =Fm . 
ASSIGN, DN=SINFILE,DC=SC,A=FT25. 
ASSIGN, DN=SCRTCH1, DC=SC, A=FT50 . 
ASSIGN, DN=SCRTCH2, DC=SC, A = FT60 . 
ASSIGN, DN=SCRTCH3, DC=SC, A=FT90 . 
ASSIGN, DN=GRDHLD,DC=SC,A=FT80. 
ASSIGN, DN=I0CHK, DC=SC, A=FT19 . 
ASSIGN, DN = SUBCASE, DC=SC, A = FTA5 . 
ASSIGN, DN = TMPTR, DC = SC, A=FTA6 . 
COPYF, I =$IN, O s SOURCE, NF=1 . 
COPYF, I S $IN,0=NASFILE,NF=1. 
COPYF, I=$IN,0=SINFILE,NF=1. 
REWIND, DN=SOURCE. 

CFT , I=SOURCE. 

LDR . 

DISPOSE, DN=IOCHK,DC=ST. 

DISPOSE, DN=SUBCASE,DC=ST. 
DISPOSE, DN=TMPTR, DC=ST . 

EXIT. 

/EOF 




CRAY Control Statements File 


SNIP Source Code File goes here 


/EOF 


[Input NASTRAN Bulk Data deck goes here 


/EOF 


/EOF 


Input Thermal Model results file goes here 


original; page is 

OF POOR QUALITY 


16 



APPENDIX D 


SNIP Operations Checklist - CRAY Version 

1 - Adjust input parameters in SNIP source code PARAMETER statements (see 

appendix A) 

2 - Create CRAY job Control Statements File that: 


(1) ASSIGNS appropriate FORTRAN logical unit numbers to the following 
files: 

(1) input NASTRAN Bulk Data file (unit = NASTDECK) 

(2) input thermal model results file (unit ■ INSINDA) 

(3-6) four intermediate data storage files (unit = ISCRTCH1 , 
ISCRTCH2 , ISCRTCH3, and IGRDHLD) 

(7) output node-element relationships file (unit = IOCHECK) 

(8) output NASTRAN temperature load cards (unit = INASTEMP) 

(9) output NASTRAN case control cards (unit = INASSUB) 

(2) COPYs input files (NASTRAN model and thermal model results) 
into appropriate CRAY files 

(3) Compiles and runs the SNIP source code 

(4) DISPOSES the output files from the CRAY to the user (see apendix C) 


3 - Combine CRAY job Control Statements file, SNIP source code file, input 
NASTRAN Buld Data file, and input thermal model results file into a CRAY 
file 


4 - Submit CRAY run file to CRAY for batch processing 

5 - Review resulting node-element relationships in the appropriate output file 

(if relationships are not acceptable, review and revise input parameters 
and/or input data as appropriate and run SNIP again) 


17 


REFERENCES 


1. McCormick, C.W., ed.: MSC/NASTRAN User's Manual, Version 63, MacNeal- 

Schwendler Corp. , 1983. 

2. CRAY-OS Version 1 Reference Manual SR-0011, Revision L, CRAY Research Inc., 
Minneapolis, MN, July 1983. 

3. Steinbach, R.E.; and Winegar, S.R.: Interdisciplinary Design Analysis of 

Precision Spacecraft Antenna. 26th Structures, Structural Dynamics, and 
Materials Conference, Part 1, 1985, pp. 704-712. 


18 


TABLE I. - CASE 1 THERMAL MODEL RESULT 
INPUT FILE 


CASE 1 
8:00 
9 

1 . 25 . . 0 . 0 . 1 . 67 , 

2 . 50 .. 0 . 0 . 5 . 0.1 

3 . 75 . . 0 . 0 . 8 . 33 , 

4 . 50 . . 0 . 0 . 1 . 67 , 

5 . 75 . . 0 . 0 . 5 . 0.5 

6 . 100 . . 0 . 0 . 8.33 

7 . 75 . . 0 . 0 . 1 . 67 , 

8 . 100 . . 0 . 0 . 5 . 0 , 

9 . 125 . . 0 . 0 . 8.33 


. 67 , 0 . 0 , 0.0 

67 . 0 . 0 . 0.0 
. 67 , 0 . 0 , 0.0 
. 0 , 0 . 0 , 0.0 
0 , 0 . 0 , 0.0 

5 . 0 . 0 . 0 . 0.0 
. 33 , 0 . 0 , 0.0 
. 33 , 0 . 0 , 0.0 

8 . 33 . 0 . 0 . 0.0 


19 


TABLE II. - CASE 1 NODE-ELEMENT RELATIONSHIPS (ALWDR = 10.0 in.) 


QUAD 

ELEMENT 

SINDA 
NODE 1 

W.F. 

1 

SINDA 
NODE 2 

W.F . 
2 

S 1 NDA 
NODE3 

W.F . 
3 

SINDA 

N0DE4 

W.F . 
4 

21 

4 

0.342 

0 

0.000 

0 

0.000 

1 

0.658 

22 

4 

0.253 

0 

0.000 

0 

0.000 

1 

0.747 

23 

5 

0. 149 

4 

0.200 

1 

0.450 

2 

0.200 

24 

5 

0.196 

4 

0.185 

1 

0.285 

2 

0.334 

25 

5 

0.193 

4 

0.131 

1 

0.167 

2 

0.509 

26 

6 

0. 131 

5 

0.193 

2 

0.509 

3 

0.167 

27 

6 

0.185 

5 

0.196 

2 

0.334 

3 

0.285 

28 

6 

0.200 

5 

0.149 

2 

0.200 

3 

0.450 

29 

0 

0.000 

6 

0.253 

3 

0.747 

0 

0.000 

30 

0 

0.000 

6 

0.342 

3 

0.658 

0 

0.000 

31 

4 

0.533 

0 

0.000 

0 

0.000 

1 

0.467 

32 

4 

0.549 

0 

0.000 

0 

0.000 

1 

0.451 

33 

5 

0. 196 

4 

0.334 

1 

0.285 

2 

0.185 

34 

5 

0.277 

4 

0.248 

1 

0.227 

2 

0.248 

35 

5 

0.358 

4 

0.177 

1 

0.168 

2 

0.298 

36 

6 

0. 177 

5 

0.358 

2 

0.298 

3 

0.168 

37 

6 

0.248 

5 

0.277 

2 

0.248 

3 

0.227 

38 

6 

0.334 

5 

0.196 

2 

0.185 

3 

0.285 

39 

0 

0.000 

6 

0.549 

3 

0.451 

0 

0.000 

40 

0 

0.000 

6 

0.533 

3 

0.467 

0 

0.000 

41 

4 

0.706 

0 

0.000 

0 

0.000 

1 

0.294 

42 

4 

0.843 

0 

0.000 

0 

0.000 

1 

0.157 

43 

5 

0. 193 

4 

0.509 

1 

0.167 

2 

0.131 

44 

5 

0.358 

4 

0.298 

1 

0.168 

2 

0.177 

45 

5 

0.599 

4 

0.147 

1 

0.106 

2 

0.147 

46 

6 

0.147 

5 

0.599 

2 

0.147 

3 

0. 106 

47 

6 

0.298 

5 

0.358 

2 

0.177 

3 

0.168 

48 

6 

0.509 

5 

0.193 

2 

0.131 

3 

0.167 

49 

0 

0.000 

6 

0.843 

3 

0.157 

0 

0.000 

50 

0 

0.000 

6 

0.706 

3 

0.294 

0 

0.000 


20 


TABLE III. - CASE 1 NODE-ELEMENT RELATIONSHIPS (ALWDR = 2.2 in.) 


QUAD 

.EMENT 

S INDA 
NODE 1 

W.F. 

1 

SINDA 
NODE 2 

W.F. 

2 

SINDA 
NODE 3 

W.F . 
3 

SINDA 

NODE4 

W.F . 
4 

21 

0 

0.000 

0 

0.000 

0 

0.000 

1 

1 .000 

22 

0 

0.000 

0 

0.000 

0 

0.000 

1 

1 .000 

23 

0 

0.000 

0 

0.000 

1 

1 .000 

0 

0.000 

24 

0 

0.000 

0 

0.000 

1 

0.460 

2 

0.540 

25 

0 

0.000 

0 

0.000 

0 

0.000 

2 

1 .000 

26 

0 

0.000 

0 

0.000 

2 

1 .000 

0 

0.000 

27 

0 

0.000 

0 

0.000 

2 

0.540 

3 

0.460 

28 

0 

0.000 

0 

0.000 

0 

0.000 

3 

1 .000 

29 

0 

0.000 

0 

0.000 

3 

1 .000 

0 

0.000 

30 

0 

0.000 

0 

0.000 

3 

1 .000 

0 

0.000 

31 

4 

0.533 

0 

0.000 

0 

0 . 000 

1 

0.467 

32 

4 

0.549 

0 

0.000 

0 

0.000 

1 

0.451 

33 

0 

0.000 

4 

0.540 

1 

0.460 

0 

0.000 

34 

5 

1 .000 

0 

0.000 

0 

0.000 

0 

0.000 

35 

5 

0.545 

0 

0.000 

0 

0.000 

2 

0.455 

36 

0 

0.000 

5 

0.545 

2 

0.455 

0 

0.000 

37 

0 

0.000 

5 

1 .000 

0 

0.000 

0 

0.000 

38 

6 

0.540 

0 

0.000 

0 

0.000 

3 

0.460 

39 

0 

0.000 

6 

0.549 

3 

0.451 

0 

0.000 

40 

0 

0.000 

6 

0.533 

3 

0.467 

0 

0.000 

41 

4 

1 .000 

0 

0.000 

0 

0.000 

0 

0.000 

42 

4 

1 .000 

0 

0.000 

0 

0.000 

0 

0.000 

43 

0 

0.000 

4 

1 .000 

0 

0.000 

0 

0.000 

44 

5 

0.545 

4 

0.455 

0 

0.000 

0 

0.000 

45 

5 

1 .000 

0 

0.000 

0 

0.000 

0 

0.000 

46 

0 

0.000 

5 

1 .000 

0 

0.000 

0 

0.000 

47 

6 

0.455 

5 

0.545 

0 

0.000 

0 

0.000 

48 

6 

1 .000 

0 

0.000 

0 

0.000 

0 

0.000 

49 

0 

0.000 

6 

1 .000 

0 

0.000 

0 

0.000 

50 

0 

0.000 

6 

1 .000 

0 

0.000 

0 

0.000 


21 


TABLE IV. - CASE 4 BAR ELEMENT NODE-ELEMENT RELATIONSHIPS 


BEAM 

ELEMENT 

END: 

A 

S INDA 
N0DE1 

W.F . 
1 

SINDA 
NODE 2 

W.F. 

2 

END: 

B 

SINDA 
NODE 1 

W.F . 
1 

SINDA 
NODE 2 

W.F. 

2 

501 


0 

0.000 

1001 

1 .000 


0 

0.000 

1001 

1 .000 

502 


0 

0.000 

1001 

1 .000 


1001 

0.901 

1002 

0.09S 

503 


1001 

0.901 

1002 

0.099 


1001 

0.601 

1002 

0.399 

504 


1001 

0.601 

1002 

0.399 


1001 

0.300 

1002 

0.70C 

505 


1001 

0.300 

1002 

0.700 


1002 

1 .000 

1003 

0 . OOC 

506 


1002 

1 .000 

1003 

0.000 


1002 

0.700 

1003 

0. 30C 

507 


1002 

0.700 

1003 

0.300 


1002 

0.399 

1003 

0.601 

508 


1002 

0.399 

1003 

0.601 


1002 

0.099 

1003 

0.901 

509 


1002 

0.099 

1003 

0.901 


1003 

1 .000 

0 

0 . OOC 

510 


1003 

1 .000 

0 

0.000 


1003 

1 .000 

0 

0 . OOC 

51 1 


1011 

1 .000 

1012 

0.000 


101 1 

0.800 

1012 

0.20C 

512 


101 1 

0.800 

1012 

0.200 


101 1 

0.600 

1012 

0 . 400 

513 


1011 

0.600 

1012 

0.400 


101 1 

0.400 

1012 

0 . 600 

514 


1011 

0.400 

1012 

0.600 


101 1 

0.200 

1012 

0 . 800 

515 


101 1 

0.200 

1012 

0.800 


1012 

1 .000 

1013 

0.000 

516 


1012 

1 .000 

1013 

0.000 


1012 

0.800 

1013 

0.200 

517 


1012 

0.800 

1013 

0.200 


1012 

0.600 

1013 

0 . 400 

518 


1012 

0.600 

1013 

0.400 


1012 

0.400 

1013 

0. 600 

519 


1012 

0.400 

1013 

0.600 


1012 

0.200 

1013 

0 . 800 

520 


1012 

0.200 

1013 

0.800 


1013 

1 .000 

0 

0 . OOC 


TABLE V. - CASE 4 BAR ELEMENT TEMPERATURES 

ELEMENT TEMPERATURE , END A(X m j n ) TEMPERATURE , END B(X max ) 


501 

-20. 

-20. 

502 

-20. 

-15. 

503 

-15 . 

0 . 

504 

0 . 

15. 

505 

15 . 

30. 

506 

30. 

45. 

507 

45. 

60. 

508 

60. 

75. 

509 

75. 

80. 

510 

80. 

80. 

51 1 

-15 . 

-5. 

512 

-5. 

5. 

513 

5. 

15. 

514 

15. 

25. 

515 

25. 

35. 

516 

35. 

45. 

517 

45. 

55. 

518 

55. 

65. 

519 

65 . 

75. 

520 

75. 

85. 


22 


ORIGINAL' Page IS 
POOR QUALITY 


GET NEW NASTRAN 
ELEMENT FROM LIST 
OF ALL ELEMENTS 


GET NEW INPUT 
THERMAL NODE FROM 
LIST OF ALL THERMAL 
NODES 


CODING 
CORRECT FOR 
CORRELA- 
TION? 


NODE 

WITHIN ELE- 
MENT QUALI- 
FICATION 
REGION 


AT 

BOTTOM OF 
NODE 
LIST 


AT 

BOTTOM OF 
ELEMENT 
LIST 


NEAREST 
NODE IN RE- 
GION SEGMENT 
(QUADRANT 
OR HALF) 


STORE NODE 
NUMBER AND 
CHARACTERISTIC 
DISTANCE IN 
LIST FOR RE- 
GION SEGMENT 


GO TO WEIGHT 

CALCULATION 

ROUTINE 


FIGURE 1. - CORRELATION ROUTINE LOGIC. 


23 




FIGURE 2. - QUALIFICATION REGION FOR THE SHADED ELEMENT OF THE PLATE FIGURE 3. - BAR ELEMENT COORDINATE SYSTEM (X ALONG ELEMENT AXIS). 

ELEMENTS SHOWN. 



BAR AND BEAM ELEMENT 
QUALIFICATION REGION 
SEGMENTS (2) 

FIGURE 4. - QUALIFICATION REGION FOR END B OF THE CENTER BAR ELEMENT 
SHOWN. 


PLATE ELEMENTS 



BAR ELEMENT QUALIFICATION 
REGION 


CORRELATE CORRELATE 



NASTRAN PLATE INPUT THERMAL NASTRAN BAR AND 

AND SHELL NODE NUMBERS BEAM ELEMENT 

ELEMENT NUMBERS NUMBERS 


FIGURE 5. - OVERLAPPING QUALIFICATION REGIONS. 


FIGURE 6. - NUMERICAL CODING SCHEME. 


24 





ORIGINAL PAGE IS 

op poor quality: 


CORRELATE CORRELATE 



NASTRAN PLATE INPUT THERMAL NASTRAN BAR AND 

AND SHELL NODE NUMBERS BEAM ELEMENT 

ELEMENT NUMBERS NUMBERS 


FIGURE 7 . - POTENTIAL SETUP OF NUMERICAL CODING SCHEME. 


/-PLATE ELEMENTS (ELEMENT 
/ NUMBERS < NVAR2) 


r OFFSET 


-BEAM ELEMENTS (ELEMENT 
NUMBERS > NVARA) 





FIGURE 8. - BEAM ELEMENTS SUPPORTING PLATE ELEMENT STRUCTURE. 


QUAD 

S INDA 

W.F. 

S INDA 

W.F. 

SINDA 

W.F. 

S I NDA W.F. 


ELEMENT 

NODE 1 

1 

NODE2 

2 

NODE3 

3 

NODE4 4 


1 


1 

1 .000 


0 

0.000 


0 

0-000 

0 0.000 


2 


1 

1 .000 


0 

0.000 


0 

0.000 

0 0.000 


3 


2 

0.342 


1 

0.658 


0 

0-000 

0 0.000 


4 


2 

0.533 


1 

0.467 


0 

0.000 

0 0.000 


5 


2 

0.706 


1 

0.294 


0 

0-000 

0 0.000 


6 


3 

0.294 


2 

0.706 


0 

0.000 

0 0.000 


7 


3 

0.467 


2 

0.533 


0 

0.000 

0 0.000 


8 


3 

0.658 


2 

0.342 


0 

0.000 

0 0.000 


9 


0 

0.000 


3 

1 .OOO 


0 

0.000 

0 0.000 


10 


0 

0.000 


3 

1 .OOO 


0 

0-000 

0 0 . OOO 


1 1 


1 

1.000 


0 

0.000 


0 

0-000 

0 0.000 


12 


1 

1 .000 


0 

0.000 


0 

0-000 

0 0.000 


13 


2 

0.253 


1 

0.747 


0 

0.000 

0 O.OOQ 


14 


2 

0.649 


1 

0.451 


o 

0-000 

0 0 . OOO 


15 


2 

0.843 


1 

0.157 


0 

0.000 

0 0.000 


BEAM 

END 

: S 

INDA 

W.F. 

SINDA 

W.F . 

END: SINDA 

W.F. SINDA W 

.F . 

ELEMENT 

A 

NODE 1 

1 

NODE 2 

2 

B 

NODE 1 

1 NODE 2 

2 

501 



0 

0.000 


1001 

1 .000 


0 

0.000 1001 

1 .000 

502 



0 

0.000 


1001 

1 .000 


1001 

0.901 1002 

0.099 

503 



1001 

0.901 


1002 

0.099 


1001 

0.601 1002 

0.399 

504 



1001 

0.601 


1002 

0.399 


1001 

0.300 1002 

0.700 

505 



1001 

0.300 


1002 

0.700 


1002 

1.000 1003 

0.000 

506 



1002 

1 .OOO 


1003 

0.000 


1002 

0.700 1003 

0.300 

507 



1002 

0.700 


1003 

0.300 


1002 

0.399 1003 

0.601 

508 



1002 

0.399 


1003 

0.601 


1002 

0.099 1003 

0.901 

509 



1002 

0.099 


1003 

0.901 


1003 

1.000 O 

0.000 

510 



1003 

1 .000 


0 

0.000 


1003 

1.000 0 

0.000 

511 



101 1 

1 .000 


1012 

0.000 


101 1 

0.800 1012 

0.200 

512 



101 1 

0.800 


1012 

0.200 


101 1 

0.600 1012 

0.400 

513 



1011 

0.600 


1012 

0.400 


101 1 

0.400 1012 

0.600 

514 



101 1 

0.400 


1012 

0.600 


101 1 

0.200 1012 

0.800 

515 



101 1 

0.200 


1012 

0.800 


1012 

1.000 1013 

0.000 

516 



1012 

1 .000 


1013 

0.000 


1012 

0.800 1013 

0.200 

517 



1012 

0 . 800 


1013 

0.200 


1012 

0.600 1013 

0.400 

518 



1012 

0 . 600 


1013 

0.400 


1012 

0.400 1013 

0.600 

519 



1012 

0.400 


1013 

0.600 


1012 

0.200 1013 

0.800 

520 



1012 

0.200 


1013 

0.800 


1013 

1 . OOO 0 

0.000 


FIGURE 9. - NODE-ELEMENT CORRELATION AND WEIGHTING FACTOR FILE LISTING. 


25 





INCHES 




FIGURE 11. - CASE 1. 2 NASTRAN ELEMENT GRID. FIGURE 12. - CASE 1. 2 SINDA NODE GRID. 


26 


ORIGINAL PAGE Is 
of POOR QUALITY 


75° 

100° 

125° 

50° 

75° 

100° 

25° 

50° 

75° 


FIGURE 13. - CASE 1 SINDA INPUT TEMPERATURE PATTERN. 


FIGURE 1A. - CASE 1 RESULTING NASTRAN TEMPERATURE PATTERN 
DEGREES (ALWDR = 10.0"). 


I - IV = QUALIFICATION REGION QUADRANTS 1 - 4 
Ro AND R, = X-LARGE 


FIGRUE 15. - NODE-ELEMENT PHYSICAL RELATIONSHIPS FOR ELEMENT 21. 


FIGURE 16. - CASE 1 RESULTING NASTRAN TEMPERATURE PATTERN 
DEGREES (ALWDR =2.2"). 





























25° 

50° 

75° 

o 

o 

50° 

100° 

25° 

50° 

75° 


FIGURE 17. - CASE 2 SINDA INPUT TEMPERATURE PATTERN. 


FIGURE 18. - CASE 2 RESULTING NASTRAN TEMPERATURE PATTERN, DE- 
GREES. 


r~ T = 25° r~T ~ 50° /~1 = 75° 

' 


\ 25° 


1/ /V 

K / / 

50° 75° 

i 

! / 

J / 


501 502 503 504 505 506 507 508 509 510 


FIGURE 19. - CASE 2 ISOTHERMAL LINES OVERLAYED ON SINDA GRID. 


FIGURE 20. - CASE 3 BAR ELEMENTS (NUMBERED) 























Report Documentation Page 


2. Government Accession No. 


NASA 

National Aeronautics and 
Space Administration 


1. Report No. 

NASA TM-1001 58 


4. Title and Subtitle 


SINDA-NASTRAN Interfacing Program Theoretical 
Description and User's Manual 


7. Author(s) 

Steven R. Winegar 


3. Recipient’s Catalog No. 


5. Report Oate 

August 1987 


6. Performing Organization Code 

481-50-32 


8. Performing Organization Report No. 

E-3720 


10. Work Unit No. 


9. Performing Organization Name and Address 

National Aeronautics and Space Administration 
Lewis Research Center 
Cleveland, Ohio 44135 


12. Sponsoring Agency Name and Address 

National Aeronautics and Space Administration 
Washington, D.C. 20546 


1 1 . Contract or Gram No. 


13. Type of Report and Period Covered 

Technical Memorandum 


14. Sponsoring Agency Code 



16. Abstract 

The task of converting SINOA finite difference thermal model temperature 
results into NASI RAN finite element model thermal loads can be very labor 
intensive if there is not one node-to-one element, or systematic node -to- 
element, correlation between models. This paper describes the SINDA NAS I RAN 
Interfacing Program (SNIP), a FORTRAN computer code that generates NASI RAN 
structural model thermal load cards given SINDA (or similar thermal model) tern 
perature results and thermal model geometric data. SNIP generates NAS1RAN 
thermal load cards for NAS1RAN plate, shell, bar, and beam elements. The 
paper describes the interfacing procedures used by SNIP, and discusses set-up 
and operation of the program. Sample cases are included to demonstrate use of 
the program and show its 1 performance under a variety of conditions. SNIP can 
provide structural model thermal loads that accurately reflect thermal model 
results while reducing the time required to interface thermal and structural 
models when compared to other methods. 


17. Key Words (Suggested by Author(s)) 

NASI RAN 
SINDA 

lhermal distortion 
Thermal deformation 


19. Security Classif. (of this report) 

Unclassif ied 


18. Distribution Statement 

Unclassified - unlimited 
STAR Category 39 


20. Security Classif. (of this page) 

Unclassified 


21 . No of pages 


NASA FORM 1626 OCT 86 


‘For sale by the National Technical Information Service, Springfield, Virginia 22161