SECTION 4 DIAGNOSING ERRORS - MODEL CHECKOUT

COMMON TYPES OF ERRORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 - 1 COMMON MODELING ERRORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 - 3 DIAGNOSIS OF A NEW MODELPARAMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 - 4 DIAGNOSIS OF A NEW MODELDIAGs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 - 5 DIAG 8 OUTPUT - MATRIX TRAILERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 - 6 .f04" FILE OUTPUTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 - 7 SPECIAL OUTPUTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 - 8 GRID POINT STRESS AND STRESS DISCONTINUITIES . . . . . . . . . . . . . . . . . . 4 - 9 SURFACE DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 - 10 MINIMUM RECOMMENDED MODEL CHECKS . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 - 11 STIFFNESS MATRIX CHECKS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 - 18 OUTPUT FROM SAMPLE 1A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421 DESCRIPTION OF SAMPLE 1A OUTPUT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427 SAMPLE 1BMODEL WITH A BAD ELEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . 430 OUTPUT FROM SAMPLE 1B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432 RESULTS OF SAMPLE 1B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433 SAMPLE 1CMODEL WITH A BAD MPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434 SAMPLE 1CIMPROPER MPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435 OUPTUT FROM SAMPLE 1C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436 RESULTS OF SAMPLE1C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438 HOW TO AVOID SERIOUS MODELING MISTAKES . . . . . . . . . . . . . . . . . . . . . . . 439

SECTION 4 DIAGNOSING ERRORS - MODEL CHECKOUT (CONT)

CHECK FOR BAD MODES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440 SOME ADDITIONAL DEBUGS FOR DYNAMICS . . . . . . . . . . . . . . . . . . . . . . . . . . 441 SAMPLE OF SHRINK PLOTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442 SOME RECOMMENDATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443

COMMON TYPES OF ERRORS

S

Mistakes in engineering judgment

Approximations to physical behavior

S S S S

Engineering theory Finite element theory Finite element implementation Modeling

S S S S

Bolted connection Welded connection Corners Transitions

41

COMMON TYPES OF ERRORS (Cont.)

S

Modeling errors
S

Connections
S S S

Beam to plate Beam to solid Plate to solid

S S S

Beam orientation Beam releases Loading (how well do you know the loading yourself?)

Finite element error

Round-off error (can be disastrous when it occurs)

S

Computers use binary arithmetic (If you enter .1, internally it may be .099999999998)

Program bugs (nobodys perfect)

S

A list of known errors is maintained and distributed

Eternal Vigilance Is the Price of a Good Analysis

42

COMMON MODELING ERRORS

S

Plates not lining up = zipper

Any connections depending on in-plane rotational stiffness of plates, or any rotational stiffness on solids

Instabilitiesfor example, releasing both ends of a beam in torsion

Offsets of elements in wrong coordinate system (should be in the output coordinate systems of the grid points for Bars and Beams)

Member properties wrong (beam membrane onlyleft out bending

orientation)also

plates

Beam end releasesare they local or global? (in beam coordinates)

Element force output is normally in element coordinate system

43

DIAGNOSIS OF A NEW MODELPARAMs

PARAM AUTOSPC,EPPRT,MAXRATIO Operation Check relative magnitudes of matrix terms Solve superelements individually Statics = fixed-boundary solution Dynamics = calculated component modes Energy error Check mass, CG Print set tables Strain energy fractions (superelementsSOL 103) Minimum percentage value of element strain enery for printout (Values not printed are not available for post-processing)

FIXEDB

IRES GRDPNT USETPRT SESEF

TINY

44

DIAGNOSIS OF A NEW MODELDIAGs

DIAG 8 14 15 56 Print matrix tailers Print DMAP listing Print table trailers

Operation

List Qualifier changes as teh solution progresses also, list all DMAP statements executed in the .f04 file (normally only modules are listed)

MSC/NASTRAN DATA BLOCK NAME CONVENTION FOR MATRICES

KYIJ

Y IJ

where Y

= type: A, D, 4

I,J = col, row sets K M B G D U P Q = stiffness = mass = viscous damping = transformation = rigid body transform = displacement = load = force of constraint

45

DIAG 8 OUTPUT - MATRIX TRAILERS

S

Prints matrix trailers as the matrices are created

DAY TIME ELAPSED I/O SEC DEL_I/O CPU SEC DEL_CPU SUB_DMAP/DMAP_MODULE MESSAGES

16:56:39 16:56:40

0:37 0:38

2.9 2.9

.0 .0

8.9 9.5

.0 .6

SEPREP2 SEPREP2

17 17

GP1 GP1

BEGN END

Module Name

DMAP Sequence ID subDMAP

Elapsed Time for Job (used for time" limit)

File Operations
Wall Clock - Elapsed Seconds

Time of Day

46

.f04" FILE OUTPUTS

Sample printout using DIAG 8

ELAPSED 0:16 I/O MB 18.6 DEL_MB .0 Cols 1 .0 Cols 48 CPU SEC 3.5 Rows 300 3.5 Rows 48 F T 6 2 F T 2 2 .0 DEL_CPU .0 SUB_DMAP/DMAP_MODULE MESSAGES SEMG 28 Density EMG BlockT 3 111 EMA BlockT StrL 3 3 BEGN StrL 300 BEGN NbrStr 192 BndAvg 22 BndMax 45 NulCol 24 *8** NbrStr 1 BndAvg 300 BndMax 300 NulCol 0 *8**

DAY TIME 14:16:23

*8** Module EMG 14:16:24 *8** Module EMA

DMAP Matrix 28 0:17 KELM 18.6

NzWds

600 1.00000D+00 SEMG Density

DMAP Matrix 111 KJJZ

NzWds

48 2.50000D01

Definitions
Form 1 =square, 2 =rectangle, 3=diagonal, 6 =symmetric, etc 1 = single precision real, 2 = double precision real, 3 = single precision complex, 4 =double precision complex Number of words Density, number of terms

Type

Nzwds Den Blks

(N x M)

Number of GINO blocks (1 block = buffsize" words)

47

SPECIAL OUTPUTS
Strain energy density ESE" - use to determine where to make changes most efficiently. GPFORCE" - use to trace" loads thru the structure and verify load paths, PARAMS: S1, NUMOUT, BIGER, SRTOPT GPSTRESS" GPSDCON", ELSDCON"

Grid point forces

Stress sorting Grid point stresses Stress Discontinuities

48

GRID POINT STRESS AND STRESS DISCONTINUITIES

(ALL CASE CONTROL)

Use GPSTRESS, ELSDCON, or GPSDCON to select surfaces and volumes.

Use OUTPUT (POST) or SETS DEFINITION with

S S

SURFACE data to define surfaces VOLUME data to define volumes

SURFACE Entry Definition

SURFACE {id} SET {sid} ,

ALL Z1

FIBRE Z2 , MID

Z1 Z2

MID

[SYSTEM [ELEMENT]]

BASIC CORD

,AXIS {cid}

X2 X3

X1

R X1 , , NORMAL[M]X2 X3

TOPOLOGICALTOLERANCE
GEOMETRIC

,BRANCH thetamax
0.0

, MESSAGE NOMESSAGE
MESSAGE BREAK

49

SURFACE DATA
S

Recommendations for discontinuous structures

S S S S S

Use tolerance and branch tests to handle discontinuous stresses. Use local coordinate systems for orientation. Use GEOMETRIC method when element sizes differ. Try dividing into smaller surfaces. Use several options in one run and compare them.

Remember that the elements are isoparametric, that is, ideal elements are mapped onto the real elements in the model. If the grid point stresses differ when different options are used (or if the discontinuities are too large), it may indicate any of the following conditions:
S S S

Mesh too coarse Elements badly shaped Modeling errors

4 10

MINIMUM RECOMMENDED MODEL CHECKS

Pre-Analysis

Understand the structure and the elements

S S S

Make small modelsunderstand the problem Use pilot models in areas of uncertainty If you are not familiar with using the element type or SOLution you expect to use, make simple models and compare the answers to theoretical results (with a simple model, you should be able to obtain excellent correlation with theoretical results).

Model checks before the analysis

S

Geometry
S

Preprocessor (or Undeformed plots)

S

Look at connections between different element types

S S S

Based on knowledge of elements Based on loads Look at corners (QUAD plates)

Shrink plots
4 11

MINIMUM RECOMMENDED MODEL CHECKS (Cont.)

S

Elements
S

Beam and bar

S S S S

Check that I1 and I2 have proper orientation and values Check all end releases (in member coordinates) Verify all offsets (in output coordinate system of GRIDs) Materialneed E, n (or G), and r

Plates and shells

S S S

Check aspect ratios, taper, and warpage Check orientationZ, surfaces consistent Check attachmentsespecially any depending on in-plane rotational stiffness, any corners, and shells Verify any offsets (in element coordinate system ) Materialneed E, n (or G), and r Property entrybe sure to get the correct properties. (One of the most commonly made errors is not specifying MID2 for bending plates)

S S S

4 12

MINIMUM RECOMMENDED MODEL CHECKS (Cont.)

S

Solids
S S S

Check aspect ratios Check taper Check attachments. If any attachments depend on rotational stiffness, special modeling effort is required Materialneed E, n (or G), and r

S S

Mass properties
S S

Check r on MATx entries Check NSM on property entries

S S

Bars, beams = mass/unit length Plates = mass/unit area

Submit with PARAM,GRDPNT, xxxx

where xxxx = ID of GRID point to calculate mass properties about

S

Always check center of gravity and total weight (mass) versus known values

4 13

MINIMUM RECOMMENDED MODEL CHECKS (Cont.)

S

Loadings:
S

Verify they are correct (OLOAD RESULTANT)

Constraints:
S S

Verfy that they are defined (often they are forgotten) Verify they are correct (location and orientationin output coordinate system of the GRID points) Verify that they are applied (SPC CASE CONTROL command)

Static Checks ALWAYS RUN STATICS FIRST!!!

S

Apply 1-G in X, Y, and Z directions independently

S S

Check load paths (GPFORCE) Check reactions (SPCFORCE)

S S

Does total = applied load? Are the reactions at the correct locations and do they have the correct orientation?

For dynamics, approximate frequency f ~ 1 GD 2p

where D = center of gravity displacment in direction of applied Gload

4 14

MINIMUM RECOMMENDED MODEL CHECKS (Cont.)

S

Equilibrium checkverify model is not over-constrained

S

Run free-free. Remove known constraints and check for unconstrained motion under applied loads or imposed displacements

or
S

Use DMAP Alters to check for over-constrained systems (shown later in this section)

Thermal equilibrium checkif thermal loads are to be considered

S S

Check a on MATx entries Check for unconstrained thermal expansion on a copy of your model:
S S S

Apply a determinate set of constraints Use the same a for all materials Apply a uniform DT to the structure. It should expand freely, that is, with no reactions, element forces, or stresses

4 15

MINIMUM RECOMMENDED MODEL CHECKS (Cont.)

After the Analysis

Statics
S

Check EPSILON and MAXRATIO

S S

EPSILON > 10-9 may indicate trouble MAXRATIO > 105 may indicate trouble

Check reactions. Do they equal the applied loads ( S applied loads are printed as OLOAD RESULTANT in superelement solutions)? Check load pathsuse grid point force balance to trace loads
S

Check stress contours for consistency

S S

Sharp corners indicate bad modeling Use different options (i.e., topological and geometric) and compare results Check stress discontinuities Compare values to hand calc or small model results

S S

4 16

MINIMUM RECOMMENDED MODEL CHECKS (Cont.)

Dynamicsnormal modes
S

Check frequencies. Are they in the expected range? (Did you forget WTMASS???)
S S

If free-free, are there six rigid-body (f=0.0) modes? Are there any mechanisms (f=0.0)?
S S

More than six rigid-body modes in free-free? Any rigid-body modes in constrained modes?

Check mode shapes. Identify modes

S S

Plots and/or animation Effective weight and kinetic energy (sssalter checka.v705) help to identify significant modes

4 17

STIFFNESS MATRIX CHECKS

S

The model (stiffness and mass matrices) should be checked to verify that elements are not (obviously) bad and that the model is not over-constrained
S

SampleCELASi between non-coincident points or CELASi to ground

This check can be performed at various stages during the analysis at each stage, a different potential problem is checked
S

G-setat this stage of the solution, the elements (including GENELs and K2GG) are checked for grounding N-setat this stage, the MPC equations are checked A-set(free-free only) check that the SPCs do not over-constrain the structure

S S

Although these checks can be perfromed manually, a DMAP alter will make them easier to perform
S

In the sssalter directory, the DMAP alter checka.v705 is provided for model checkout

4 18

STIFFNESS MATRIX CHECKS

S

Sample Model 1Cantilever Beam

Properties: I1 = 10 I2 = 10 J=5 A=1 E = 10,000,000. n = .3 r = .1 WTMASS = .002588

4 19

STIFFNESS (AND MASS) CHECKS (Cont.)

Input File sample1a.dat

SOL 103 TIME 5 include SSSALTERDIR:checka.v705 CEND TITLE = Practical Dynamics Seminar Sample Problem 1a SUBTITLE = Cantilever Beam LABEL = Perform Model Checks DISP(plot) = ALL SUBCASE 1 SPC = 1 METHOD = 1 BEGIN BULK PARAM GRDPNT 0 PARAM WTMASS .002588 PARAM AUTOSPC YES GRID 1 0 0.0 0.0 0.0 GRID 2 0 1.25 0.0 0.0 GRID 3 0 2.5 0.0 0.0 GRID 4 0 3.75 0.0 0.0 GRID 5 0 5. 0.0 0.0 GRID 6 0 6.25 0.0 0.0 GRID 7 0 7.5 0.0 0.0 GRID 8 0 8.75 0.0 0.0 GRID 9 0 10. 0.0 0.0 CBAR 1 1 1 2 CBAR 2 1 2 3 CBAR 3 1 3 4 CBAR 4 1 4 5 CBAR 5 1 5 6 CBAR 6 1 6 7 CBAR 7 1 7 8 CBAR 8 1 8 9 SPC 1 1 123456 0.0 PBAR 1 1 1. 10. 10. MAT1 1 1.+7 . 3 .1 EIGRL 1 1. 5 ENDDATA

0 0 0 0 0 0 0 0 0 1. 1. 1. 1. 1. 1. 1. 1. 5.

4 20

PHASE I - OUTPUT EXAMPLE A

JANUARY 7, 1999 MSC/NASTRAN 4/28/98 PAGE 9

PRACTICAL DYNAMICS SEMINAR SAMPLE PROBLEM 1 CANTILEVER BEAM PERFORM MODEL CHECKS O U T P U T F R O M

0 * * * * * *

* * * * * *

OUTPUT FROM SAMPLE 1A

4-21

G R I D P O I N T W E I G H T G E N E R A T O R REFERENCE POINT = 0 M O 1.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 1.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 5.000000E+00 0.000000E+00 0.000000E+00 1.000000E+00 0.000000E+00 5.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 5.000000E+00 0.000000E+00 3.359375E+01 0.000000E+00 0.000000E+00 5.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 3.359375E+01 S * 1.000000E+00 0.000000E+00 0.000000E+00 * * 0.000000E+00 1.000000E+00 0.000000E+00 * * 0.000000E+00 0.000000E+00 1.000000E+00 * DIRECTION MASS AXIS SYSTEM (S) MASS XC.G. YC.G. ZC.G. X 1.000000E+00 0.000000E+00 0.000000E+00 0.000000E+00 Y 1.000000E+00 5.000000E+00 0.000000E+00 0.000000E+00 Z 1.000000E+00 5.000000E+00 0.000000E+00 0.000000E+00 I(S) * 0.000000E+00 0.000000E+00 0.000000E+00 * * 0.000000E+00 8.593750E+00 0.000000E+00 * * 0.000000E+00 0.000000E+00 8.593750E+00 * I(Q) * 0.000000E+00 * * 8.593750E+00 * * 8.593750E+00 * Q * 1.000000E+00 0.000000E+00 0.000000E+00 * * 0.000000E+00 1.000000E+00 0.000000E+00 * * 0.000000E+00 0.000000E+00 1.000000E+00 *

A
JANUARY 7, 1999 MSC/NASTRAN 4/28/98 PAGE 10

PRACTICAL DYNAMICS SEMINAR SAMPLE PROBLEM 1 CANTILEVER BEAM PERFORM MODEL CHECKS

^^^ ^^^ NO VALUE PROVIDED FOR PARAMETER CHECKTOL ^^^ ^^^ CALCULATED VALUE OF CHECKTOL = 1.228800E01 ^^^ ^^^ MODEL CHECKING IS INVOKED MSC RECOMMENDS THAT A SEPARATE RUN USING PARAM,CHECKOUT,YES SHOULD ALSO BE DONE TO INSURE MODEL ACCURACY. ^^^ ^^^ RESULTS OF RIGID BODY CHECKS OF MATRIX KGG FOLLOW ^^^ ^^^ ALL 6 DIRECTIONS ARE CHECKED, ONLY THOSE DOFS WHICH FAIL WILL BE PRINTED ^^^ ^^^MATRIX KGG PASSED RIGIDBODY CHECKS. THE STRAIN ENERGY IN EACH DIRECTION WAS LESS THAN 1.228800E01 ^^^ ^^^NO MSET FOR THIS SUPERELEMENT KNN NOT CHECKED ^^^ ^^^RESULTS OF CHECK OF MGG MATRIX.

OUTPUT FROM SAMPLE 1A (CONT)

4-22
CSP INT 3.863988E+02 0 0.000000E+00

0 MATRIX WGHTG (GINO NAME 101 ) IS A DB PREC 6 COLUMN X 6 ROW SQUARE 0COLUMN 1 ROWS 1 THRU 1 ROW 1) 1.0000D+00 0COLUMN 2 ROWS 2 THRU 6 ROW 2) 1.0000D+00 0.0000D+00 0.0000D+00 0.0000D+00 5.0000D+00 0COLUMN 3 ROWS 3 THRU 5 ROW 3) 1.0000D+00 0.0000D+00 5.0000D+00 0COLUMNS 4 THRU 4 ARE NULL. 0COLUMN 5 ROWS 3 THRU 5 ROW 3) 5.0000D+00 0.0000D+00 3.3594D+01 0COLUMN 6 ROWS 2 THRU 6 ROW 2) 5.0000D+00 0.0000D+00 0.0000D+00 0.0000D+00 3.3594D+01 0THE NUMBER OF NONZERO TERMS IN THE DENSEST COLUMN = 2 0THE DENSITY OF THIS MATRIX IS 25.00 PERCENT.

MASSWT

GRDPNT

A
JANUARY SUBCASE 1 7, 1999 MSC/NASTRAN 4/28/98 PAGE 13

PRACTICAL DYNAMICS SEMINAR SAMPLE PROBLEM 1 CANTILEVER BEAM PERFORM MODEL CHECKS

OUTPUT FROM SAMPLE 1A (CONT)

4-23

^^^ ^^^ RESULTS OF RIGID BODY CHECKS OF MATRIX KAA FOLLOW ^^^ ^^^ ALL 6 DIRECTIONS ARE CHECKED, ONLY THOSE DOFS WHICH FAIL WILL BE PRINTED ^^^ ^^^ ^^^ NO VALUE PROVIDED FOR PARAMETER CHECKTOL ^^^ ^^^ CALCULATED VALUE OF CHECKTOL = 1.228800E01 ^^^ ^^^LARGEST STRAIN ENERGY OF 6.144000E+08 EXCEEDS PROVIDED LIMIT OF 1.228800E01 ^^^ ^^^ IF THE G OR N CHECKS HAVE FAILED, THEY SHOULD BE RESOLVED BEFORE LOOKING AT THE A SET CHECKS ^^^ ^^^ THE FOLLOWING DIRECTIONS HAVE FAILED THE TEST ^^^ ^^^ ^^^DIRECTION 1 HAS STRAIN ENERGY = = 8.000000E+06 ^^^ ^^^DIRECTION 2 HAS STRAIN ENERGY = = 6.144000E+08 ^^^ ^^^DIRECTION 3 HAS STRAIN ENERGY = = 6.144000E+08 ^^^ ^^^DIRECTION 4 HAS STRAIN ENERGY = = 1.538462E+07 ^^^ ^^^DIRECTION 5 HAS STRAIN ENERGY = = 3.200000E+08 ^^^ ^^^DIRECTION 6 HAS STRAIN ENERGY = = 3.200000E+08 ^^^ ^^^ IF THE MODEL HAS PASSED THE PREVIOUS G AND N CHECKS POSSIBLE REASONS FOR FAILING THIS TEST ARE ^^^ ^^^ 1) THE STRUCTURE IS NOT INTENDED TO BE FREEFREE ^^^ IF THIS IS THE CASE (A CONSTRAINED STRUCTURE) FAILING THIS CHECK IS A SIGN THAT THE STRUCTURE ^^^ IS PROPERLY CONSTRAINED TO GROUND ^^^ ^^^ 2) PARAM,GRDPNT IS LOCATED FAR FROM THE CG OF THE MODEL. MSC RECOMMENDS THAT GRDPNT SHOULD BE LOCATED ^^^ AS CLOSE AS POSSIBLE TO THE GEOMETRIC CENTER OF THE MODEL (SEE GPWG OUTPUT) ^^^ ^^^ 3) AUTOSPC CONSTRAINING NEARSINGULAR DOF. ^^^ POSSIBLE CORRECTIVE ACTIONS: ^^^ INCLUDE PARAM,K6ROT,100. OR SET PARAM,SNORM,20. ^^^ CURRENT MSC RECOMMENDATION IS TO USE PARAM,SNORM

A
JANUARY SUBCASE 1 7, 1999 MSC/NASTRAN 4/28/98

PRACTICAL DYNAMICS SEMINAR SAMPLE PROBLEM 1 CANTILEVER BEAM PERFORM MODEL CHECKS

^^^STRAIN ENERGY FOR DIRECTION 1 = 8.000000E+06 ^^^NORMALIZED REACTIONS FOR THAT CHECK FOLLOW ^^^ONLY TERMS LARGER THAN 1.000000E01 ARE PRINTED ^^^ ^^^ USE PARAM,PRINTOL TO CHANGE THIS VALUE

0 POINT VALUE POINT VALUE POINT VALUE

POINT

REACCOLA VALUE

POINT

VALUE

COLUMN

2 T1

1 1.00000E+00

OUTPUT FROM SAMPLE 1A (CONT)

4-24
VALUE POINT 2 R3 6.25000E01 VALUE POINT 2 R2 6.25000E01

^^^STRAIN ENERGY FOR DIRECTION 2 = 6.144000E+08 ^^^NORMALIZED REACTIONS FOR THAT CHECK FOLLOW ^^^ONLY TERMS LARGER THAN 1.000000E01 ARE PRINTED ^^^ ^^^ USE PARAM,PRINTOL TO CHANGE THIS VALUE 0 REACCOLA POINT VALUE POINT VALUE POINT

VALUE

POINT

VALUE

COLUMN

2 T2

1 1.00000E+00

^^^STRAIN ENERGY FOR DIRECTION 3 = 6.144000E+08 ^^^NORMALIZED REACTIONS FOR THAT CHECK FOLLOW ^^^ONLY TERMS LARGER THAN 1.000000E01 ARE PRINTED ^^^ ^^^ USE PARAM,PRINTOL TO CHANGE THIS VALUE 0 REACCOLA POINT VALUE POINT VALUE POINT

VALUE

POINT

VALUE

COLUMN

2 T3

1 1.00000E+00

^^^STRAIN ENERGY FOR DIRECTION 4 = 1.538462E+07 ^^^NORMALIZED REACTIONS FOR THAT CHECK FOLLOW ^^^ONLY TERMS LARGER THAN 1.000000E01 ARE PRINTED ^^^ ^^^ USE PARAM,PRINTOL TO CHANGE THIS VALUE 0 REACCOLA POINT VALUE POINT VALUE POINT VALUE POINT VALUE POINT

VALUE

COLUMN

2 R1

1 1.00000E+00

^^^STRAIN ENERGY FOR DIRECTION 5 = 3.200000E+08 ^^^NORMALIZED REACTIONS FOR THAT CHECK FOLLOW ^^^ONLY TERMS LARGER THAN 1.000000E01 ARE PRINTED ^^^ ^^^ USE PARAM,PRINTOL TO CHANGE THIS VALUE 0 REACCOLA POINT VALUE POINT VALUE POINT VALUE POINT VALUE

POINT

VALUE

COLUMN 2 R2 4.16667E01

1 2 T3 1.00000E+00

OUTPUT FROM SAMPLE 1A (CONT)

4-25
VALUE POINT 2 R3 4.16667E01

^^^STRAIN ENERGY FOR DIRECTION 6 = 3.200000E+08 ^^^NORMALIZED REACTIONS FOR THAT CHECK FOLLOW ^^^ONLY TERMS LARGER THAN 1.000000E01 ARE PRINTED ^^^ ^^^ USE PARAM,PRINTOL TO CHANGE THIS VALUE 0 REACCOLA POINT VALUE POINT VALUE POINT

VALUE

POINT

VALUE

COLUMN

2 T2

1 1.00000E+00

21

OUTPUT FROM SAMPLE 1A (CONT)

4-26

^^^RESULTS OF CHECK OF MAA PRACTICAL DYNAMICS SEMINAR SAMPLE PROBLEM 1 JANUARY 7, 1999 MSC/NASTRAN 4/28/98 PAGE CANTILEVER BEAM 0 PERFORM MODEL CHECKS SUBCASE 1 0 MATRIX WGHTA (GINO NAME 101 ) IS A DB PREC 6 COLUMN X 6 ROW SQUARE MATRIX. 0COLUMN 1 ROWS 1 THRU 1 ROW 1) 9.3750D01 0COLUMN 2 ROWS 2 THRU 6 ROW 2) 9.3750D01 0.0000D+00 0.0000D+00 0.0000D+00 5.0000D+00 0COLUMN 3 ROWS 3 THRU 5 ROW 3) 9.3750D01 0.0000D+00 5.0000D+00 0COLUMNS 4 THRU 4 ARE NULL. 0COLUMN 5 ROWS 3 THRU 5 ROW 3) 5.0000D+00 0.0000D+00 3.3594D+01 0COLUMN 6 ROWS 2 THRU 6 ROW 2) 5.0000D+00 0.0000D+00 0.0000D+00 0.0000D+00 3.3594D+01 0THE NUMBER OF NONZERO TERMS IN THE DENSEST COLUMN = 2 0THE DENSITY OF THIS MATRIX IS 25.00 PERCENT. 0 MASSWT CSP 3.863988E+02 0.000000E+00 0 GRDPNT INT 0

DESCRIPTION OF SAMPLE 1A OUTPUT

S

Grid Point Weight Output (GPWG module)

S

The scale factor entered with parameter WTMASS is applied to the assembled element mass before GPWG. The GPWG module, however, converts mass back to the original input units that existed prior to the scaling effect of the parameter WTMASS GPWG is performed on the g-size mass matrix, which is the mass matrix prior to the processing of the rigid elements, MPCs, and SPCs Any masses at scalar points and fluid-related masses are not included in the GPWG calculation GPWG for a superelement does not include the mass from upstream superelements. Therefore, GPWG for the residual structure includes only the mass of the residual (not any upstream superelements). The center of gravity location is also based on the mass of the current superelement only The output from the GPWG is for information purposes only and is not used in the analysis The rigid-body mass matrix [M0 ] is computed with respect to the reference grid point in the basic coordinate system. The Grid point to be used is specified using PARAM, GNDPNT For further information see the MSC/NASTRAN Linear Static Analysis Users Guide (V68), Appendix B

4-27

DESCRIPTION OF SAMPLE 1A OUTPUT (Cont.)

Stiffness Check Output

S

These checks are performed by multiplying the stiffness matrix by a set of rigid-body vectors(Rb) which are based on the geometry (calculated about PARAM,GRDPNT) The rigid-body strain energy checks are calculated as (note that the factor of 1/2 is not included in the calculation)

R b TK R b
S

+ CHKKii

This check is performed on the G-, N-, and A- set matrices (i in CHKKii is the set being checked) If any term in the resulting CHK matrix exceeds the value of PARAM,CHECKTOL (default value is calculated base on the stiffness of your model), the results of the check are printed Reaction forces are calculated, normalized to a maximum of 1.0, filtered, and printed (if CHECKTOL is exceeded )

K Rb
S

+ REACi

Note that full data recovery is not performed, and if a dof which does not belong to the remaining set is constrained, the nearest point (by connection) in the remaining set is indicated. See results for CHKKAApoint 1 is constrained, but does not belong to the A-set, therefore, the constraint shows up at point 2

4-28

DESCRIPTION OF SAMPLE 1A OUTPUT (Cont.)

Mass Check Output

S

These checks are performed by multiplying the mass matrix by a set of rigid-body vectors(Rb) which are based on the geometry (calculated about PARAM,GRDPNT) The calculation is similar to that performed on the stiffness matrix The results at the G-set should match Grid Point Weight Generator The checks at the N- and A-set check if MPCs and constraints remove (or re-distribute) mass

S S S

4-29

SAMPLE 1BMODEL WITH A BAD ELEMENT

Same model as before, only now connect a CELAS2 element between dof 2 at Grid Points 2 and 3 (this will cause grounding), as the direction of the stiffness terms is not along the line connecting the GRID points)

Samples of CELASi elements which cause grounding

Connected to GROUND" Geometric mismatch Kq to ground

q
r1 r2

d1
2

d1 d2

0 d grounded motion

4-30

STIFFNESS CHECKS (Cont.)

Input File sample1b.dat

ID MSCXL, MSCNASTRAN SOL 103 TIME 5 include SSSALTERDIR:checka.v705 CEND TITLE = Practical Dynamics Seminar Sample Problem 1b SUBTITLE = Cantilever Beam LABEL = Perform Model Checks bad CELAS2 element in file DISP(plot) = ALL SUBCASE 1 SPC = 1 METHOD = 1 BEGIN BULK PARAM GRDPNT 0 PARAM WTMASS .002588 PARAM POST 0 PARAM AUTOSPC YES $ $ add a CELAS2 which is incorrectly specified $ CELAS2 99 1000. 2 2 3 2 $ GRID 1 0 0.0 0.0 0.0 0 GRID 2 0 1.25 0.0 0.0 0 GRID 3 0 2.5 0.0 0.0 0 GRID 4 0 3.75 0.0 0.0 0 GRID 5 0 5. 0.0 0.0 0 GRID 6 0 6.25 0.0 0.0 0 GRID 7 0 7.5 0.0 0.0 0 GRID 8 0 8.75 0.0 0.0 0 GRID 9 0 10. 0.0 0.0 0 $ CBAR 1 1 1 2 1. CBAR 2 1 2 3 1. CBAR 3 1 3 4 1. CBAR 4 1 4 5 1. CBAR 5 1 5 6 1. CBAR 6 1 6 7 1. CBAR 7 1 7 8 1. CBAR 8 1 8 9 1. SPC 1 1 123456 0.0 PBAR 1 1 1. 10. 10. 5. MAT1 1 1.+7 .3 .1 EIGRL 1 1. 5 ENDDATA

4-31

PHASE I - OUTPUT EXAMPLE A

JANUARY 7, 1999 MSC/NASTRAN 4/28/98 PAGE 10

PRACTICAL DYNAMICS SEMINAR SAMPLE PROBLEM 1B CANTILEVER BEAM PERFORM MODEL CHECKS BAD CELAS2 ELEMENT IN FILE

OUTPUT FROM SAMPLE 1B

4-32
3 T2 1.00000E+00

^^^ ^^^ NO VALUE PROVIDED FOR PARAMETER CHECKTOL ^^^ ^^^ CALCULATED VALUE OF CHECKTOL = 1.228801E01 ^^^ ^^^ MODEL CHECKING IS INVOKED MSC RECOMMENDS THAT A SEPARATE RUN USING PARAM,CHECKOUT,YES SHOULD ALSO BE DONE TO INSURE MODEL ACCURACY. ^^^ ^^^ RESULTS OF RIGID BODY CHECKS OF MATRIX KGG FOLLOW ^^^ ^^^ ALL 6 DIRECTIONS ARE CHECKED, ONLY THOSE DOFS WHICH FAIL WILL BE PRINTED ^^^ ^^^LARGEST STRAIN ENERGY OF 1.562500E+03 EXCEEDS LIMIT OF 1.228801E01 ^^^ ^^^ THE FOLLOWING DIRECTIONS HAVE FAILED THE TEST ^^^ ^^^DIRECTION 6 HAS STRAIN ENERGY = = 1.562500E+03 ^^^ ^^^ ^^^ SOME POSSIBLE REASONS FOR THIS ARE ^^^ ^^^ 1) CELASI ELEMENTS CONNECTING TO ONLY ONE GRID POINT ^^^ ^^^ 2) CELASI ELEMENTS CONNECTION TO NONCOINCIDENT POINTS ^^^ ^^^ 3) CELASI ELEMENTS CONNECTING TO NONCOLINEAR DOF ^^^ ^^^ 4) IMPROPERLY DEFINED DMIG MATRICES ^^^ ^^^ ^^^STRAIN ENERGY FOR DIRECTION 6 = 1.562500E+03 ^^^ ^^^REACTION FORCES FOR THIS CHECK FOLLOW ^^^ THESE TERMS ARE NORMALIZED TO A MAXIMUM VALUE OF ^^^1.OE+0. ONLY TERMS LARGER THAN 1.000000E01 ARE PRINTED ^^^ ^^^ USE PARAM,PRINTOL TO CHANGE THIS VALUE ^^^ 0 REACCOLG POINT VALUE POINT VALUE POINT VALUE POINT VALUE POINT

VALUE

COLUMN

1 2 T2 1.00000E+00

RESULTS OF SAMPLE 1B
S

At the G-set, the structural matrices are grounded when the alter attempts to rotate the model about the z-axis

This is indicated by the large term in the CHKKGG matrix for dof 6

By looking at the REACGNRM matrixthis matrix represents the forces (normalized to a maximum of 1.0) preventing the model from moving as a rigid body. The column associated with dof 6 (zrotation) contains terms for dof 2 of GRID points 2 and 3, indicating that a modeling error exists in that area

This is the location of the CELAS2

4-33

SAMPLE 1CMODEL WITH A BAD MPC

S

Same model as before, only now connect an MPC between dof 2 at Grid Points 2 and 3 (since the points are not coincident, this will cause grounding)

The MPC states that the y-direction translation of Grid Point 2 must equal the y-direction translation of Grid Point 3

4-34

SAMPLE 1CIMPROPER MPC

S

Input File

SOL 103 TIME 5 include SSSALTERDIR:checka.v705 CEND TITLE = Practical Dynamics Seminar Sample Problem 1c SUBTITLE = Cantilever Beam LABEL = Perform Model Checks MPC equation DISP(plot) = ALL SUBCASE 1 MPC = 2 SPC = 1 METHOD = 1 BEGIN BULK MPC,2,2,2,1.,3,2,1. PARAM GRDPNT 0 PARAM WTMASS .002588 PARAM POST 0 PARAM AUTOSPC YES GRID 1 0 0.0 0.0 0.0 0 GRID 2 0 1.25 0.0 0.0 0 GRID 3 0 2.5 0.0 0.0 0 GRID 4 0 3.75 0.0 0.0 0 GRID 5 0 5. 0.0 0.0 0 GRID 6 0 6.25 0.0 0.0 0 GRID 7 0 7.5 0.0 0.0 0 GRID 8 0 8.75 0.0 0.0 0 GRID 9 0 10. 0.0 0.0 0 CBAR 1 1 1 2 1. CBAR 2 1 2 3 1. CBAR 3 1 3 4 1. CBAR 4 1 4 5 1. CBAR 5 1 5 6 1. CBAR 6 1 6 7 1. CBAR 7 1 7 8 1. CBAR 8 1 8 9 1. SPC 1 1 123456 0.0 PBAR 1 1 1. 10. 10. 5. MAT1 1 1.+7 .3 .1 EIGRL 1 1. 5 ENDDATA

4-35

PHASE I - OUTPUT EXAMPLE A

OUTPUT FROM SAMPLE 1C

4-36
1 R3 6.25000E01 3 T1 1.00000E+00

PERFORM MODEL CHECKS BAD MPC EQUATION ^^^ ^^^ NO VALUE PROVIDED FOR PARAMETER CHECKTOL ^^^ ^^^ CALCULATED VALUE OF CHECKTOL = 1.228800E01 ^^^ ^^^ MODEL CHECKING IS INVOKED MSC RECOMMENDS THAT A SEPARATE RUN USING PARAM,CHECKOUT,YES SHOULD ALSO BE DONE TO INSURE MODEL ACCURACY. ^^^ ^^^ RESULTS OF RIGID BODY CHECKS OF MATRIX KGG FOLLOW ^^^ ^^^ ALL 6 DIRECTIONS ARE CHECKED, ONLY THOSE DOFS WHICH FAIL WILL BE PRINTED ^^^ ^^^MATRIX KGG PASSED RIGIDBODY CHECKS. THE STRAIN ENERGY IN EACH DIRECTION WAS LESS THAN 1.228800E01 ^^^ ^^^ RESULTS OF RIGID BODY CHECKS OF MATRIX KNN FOLLOW ^^^ ^^^ ALL 6 DIRECTIONS ARE CHECKED, ONLY THOSE DOFS WHICH FAIL WILL BE PRINTED ^^^ ^^^ NO VALUE PROVIDED FOR PARAMETER CHECKTOL ^^^ ^^^ CALCULATED VALUE OF CHECKTOL = 1.228800E01 ^^^ ^^^LARGEST STRAIN ENERGY OF 1.920000E+09 EXCEEDS PROVIDED LIMIT OF 0.000000E+00 ^^^ ^^^ THE FOLLOWING DIRECTIONS HAVE FAILED THE TEST ^^^ ^^^DIRECTION 6 HAS STRAIN ENERGY = = 1.920000E+09 ^^^ ^^^ SOME POSSIBLE CAUSES OF THIS ARE ^^^ ^^^ 1) MPC EQUATIONS WHICH DO NOT SATISFY RIGIDBODY MOTION ^^^ ^^^ 2) RBE3 ELEMENTS FOR WHICH THE INDEPENDENT DOF CANNOT DESCRIBE ALL POSSIBLE RIGIDBODY MOTIONS ^^^ ^^^STRAIN ENERGY FOR DIRECTION 6 = 1.920000E+09 ^^^ ^^^NORMALIZED REACTIONS FOR THAT CHECK FOLLOW ^^^ONLY TERMS LARGER THAN 1.000000E01 ARE PRINTED ^^^ ^^^ USE PARAM,PRINTOL TO CHANGE THIS VALUE 0 REACCOLN POINT VALUE POINT VALUE POINT VALUE POINT VALUE POINT

VALUE

COLUMN

1 1 T2 1.00000E+00

3 R2

6.25000E01

A
MATRIX.

^^^RESULTS OF CHECK OF MGG 0 MATRIX WGHTG (GINO NAME 101 ) IS A DB PREC 6 COLUMN X 6 ROW SQUARE 0COLUMN 1 ROWS 1 THRU 1 ROW 1) 1.0000D+00 0COLUMN 2 ROWS 2 THRU 6 ROW 2) 1.0000D+00 0.0000D+00 0.0000D+00 0.0000D+00 5.0000D+00 0COLUMN 3 ROWS 3 THRU 5 ROW 3) 1.0000D+00 0.0000D+00 5.0000D+00 0COLUMNS 4 THRU 4 ARE NULL. 0COLUMN 5 ROWS 3 THRU 5 ROW 3) 5.0000D+00 0.0000D+00 3.3594D+01 0COLUMN 6 ROWS 2 THRU 6 ROW 2) 5.0000D+00 0.0000D+00 0.0000D+00 0.0000D+00 3.3594D+01 0THE NUMBER OF NONZERO TERMS IN THE DENSEST COLUMN = 2 0THE DENSITY OF THIS MATRIX IS 25.00 PERCENT.

OUTPUT FROM SAMPLE 1C (CONT)

4-37

0 ^^^RESULTS OF CHECK OF MNN 0 MATRIX WGHTN (GINO NAME 101 ) IS A DB PREC 6 COLUMN X 6 ROW SQUARE 0COLUMN 1 ROWS 1 THRU 1 ROW 1) 1.0000D+00 0COLUMN 2 ROWS 2 THRU 6 ROW 2) 1.0000D+00 0.0000D+00 0.0000D+00 0.0000D+00 5.1563D+00 0COLUMN 3 ROWS 3 THRU 5 ROW 3) 1.0000D+00 0.0000D+00 5.0000D+00 0COLUMNS 4 THRU 4 ARE NULL. 0COLUMN 5 ROWS 3 THRU 5 ROW 3) 5.0000D+00 0.0000D+00 3.3594D+01 0COLUMN 6 ROWS 2 THRU 6 ROW 2) 5.1563D+00 0.0000D+00 0.0000D+00 0.0000D+00 3.4180D+01 0THE NUMBER OF NONZERO TERMS IN THE DENSEST COLUMN = 2 0THE DENSITY OF THIS MATRIX IS 25.00 PERCENT.

MATRIX.

RESULTS OF SAMPLE1C
S

At the G-set, the structural matrices pass the rigidbody tests, since the CELAS2 which caused the problem in sample1b has been removed

Now matrix KNN fails the rigidbody test due to the incorrectlyspecified MPC equation (which does not satisfy rugudbody requirements)

This is indicated by the large term in the CHKKNN matrix for dof 6

By looking at the REACNCOL matrixthis matrix represents the forces (normalized to a maximum of 1.0) preventing the model from moving as a rigid body. The 6th column contains terms for GRID points 1 and 3, indicating that a modeling error exists in that area

This is the location of the MPC (NOTE since the test is performed on the Nset, GRID 2 dof 2 no longer exists, since it is in the Mset and has been removed)

4-38

HOW TO AVOID SERIOUS MODELING MISTAKES

S

Take the time to understand the structure and how it behaves under load. Perform hand analysis or use a simple model first.

Take the time to understand MSC/NASTRAN (particularly the elements). Run small samples each time you try something new.

Use independent checks (if available).

Estimate the cost (labor and computer costs) before you start.

4-39

CHECK FOR BAD MODES

S

Identify your modes using one or more of the following:

S

Plot your eigenvectors (either using the MSC/NASTRAN plotter or MSC/PATRAN) and identify them Try setting NORM=MAX on EIGRL entry and look at modal masses. Small values may indicate singularities or local modes (not recommended) Use alters to print kinetic energy and modal effective mass (See Practical Dynamic Analysis Seminar Notes).

Watch for warnings on orthogonality checks.

Look for extraneous low frequency modes these often indicate incorrect modeling (for example plate elements without MID2 on the PSHELL entry).

4-40

SOME ADDITIONAL DEBUGS FOR DYNAMICS

S

In dynamic analysis, use normal modes as a diagnostic tool.

Simulate statics in modal dynamic solutions and compare the results to a static solution (this is a way to determine if your modes are capable of representing the solution).
S S

In Transient analysis, apply a constant loading In frequency response, apply the load at 0.0 hz

Use sssalter modevala.vxx to see if your modes can represent the solution if the loads are applied statically (although you are looking at a dynamic solution, it is hard for the modes to represent the dynamic solution under loading if they cannot represent the static solution).

Selecting time or frequency set selection can have a major impact on the solution accuracy.
S

In Transient response, the accuracy is directly related to the integration time step (A central difference is used to calculate the velocity and acceleration). If you are using a direct solution, run using different integration time steps to see if the answers change. In Frequency Response, the peak responses occur at resonance. Use a Modal solution with FREQ3, FREQ4, and/or FREQ5 entries to guarantee that the solution is obtained at resonance.

4-41

SAMPLE OF SHRINK PLOTS

Stiffened Plate with Error in Modeling

4-42

SOME RECOMMENDATIONS

Understand the important things BEFORE you get into trouble!!!

S S S S S S

understand your structure and how you expect it to perform Understand your loading. understand your model understand how to use the program understand the limitations of the method use small sample problems to understand the solution

ALWAYS perform a static solution first, then progress to the more complicated solutions

4-43

4-44