AFWAL-TR-88-2 149 PROBABILISTIC FINITE ELEMENT ANALYSIS OF DYNAMIC STRUCTURAL RESPONSE R. A. Brockman F. Y. Lung N W. R. Braisted C University of Dayton Research Institute N 300 College Park S Dayton, OH 45469 D March 1989 Final Report for Period October 1985 - October 1987 Approved for public release; distribution is unlimitel DTIC O LCTE 8 S9D AEROPROPULSION AND POWER LABORATORY AIR FORCE WRIGHT AERONAUTICAL LABORATORIES AIR FORCE SYSTEMS COMMAND WRIGHT-PATTERSON AIR FORCE BASE, OHIO 45433-6563 ON&Af -~ 10% ^
216
Embed
S9D DTIC · Appendix D. PATRAN Interfaces (PATPRO/PROPAT) ... 26 Amplitude Sensitivities for Cantilever Beam 124 27 Displacement Amplitude Variance versus Frequency 126
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
AFWAL-TR-88-2 149
PROBABILISTIC FINITE ELEMENT ANALYSIS OF DYNAMICSTRUCTURAL RESPONSE
R. A. BrockmanF. Y. Lung
N W. R. Braisted
C University of DaytonResearch Institute
N 300 College ParkS Dayton, OH 45469
D March 1989
Final Report for Period October 1985 - October 1987
Approved for public release; distribution is unlimitel
DTICO LCTE 8S9D
AEROPROPULSION AND POWER LABORATORYAIR FORCE WRIGHT AERONAUTICAL LABORATORIESAIR FORCE SYSTEMS COMMANDWRIGHT-PATTERSON AIR FORCE BASE, OHIO 45433-6563
ON&Af -~ 10% ^
NOTICE
When Government drawings, specifications, or other data are used forany purpose other than in connection with a definitely Government-relatedprocurement, the United States Government incurs no responsibility or anyobligation whatsoever. The fact that the government may have formulated orin any way supplied the said drawings, specifications, or other data, is notto be regarded by implication, or otherwise in any manner construed, aslicensing the holder, or any other person or corporation; or as conveyingany rights or permission to manuiacture, use, or sell any patented inventionthat may in any way be related thereto.
This report is releasable to the National Technical Information Service(NTIS). At NTIS, it will be available to the general public, includingforeign nations.
This technical report has been reviewed and is approved for publica-tion.
JOM D. REED, Aerospace Engineer MARVIN F. SCHMIDT, ChiefPkbpulsion Integration Engine Integration & Assessment BranchEngine Integration & Assessment Branch
FOR THE CO1AndER
ROBERT E. HNDERSOmDeputy foc TechnologyTurbine Engine DivisionAero Propulsion & Power Laboratory
If your address has changed, if you wish to be removed from our mailinglist, or if the addressee is no longer employed by your organization pleasenotify WRDC/OTA, WPAFB, OH 45433-_k _ to help us maintain a currentmailing list.
Copies of this report should not be returned unless return is required bysecurity considerations, contractual obligations, or notice on a specificdocument.
UnclassifiedSECURITY CLASSIFICATION AUTHORITY 3. OISTRIBUTION/AVAILABILITY OF REPORT
Approved for public release;2.. OECLASIFICATION/OOWNGRAOING SCHEDULE distribution is unlimited.
4. PERFORMING ORGANIZATION REPORT NUMBER(S) S. MONITORING ORGANIZATION REPORT NUMSER(S)UDR-TR-87-130
AFWAL-TR-88-2149
k NAME OF PERFORMING ORGANIZATION b. OFFICE SYMBOL 7s. NAME OF MONITORING ORGANIZATIONUniversity of Dayton If aIpcable, Air Force Wright Aeronautical Labs.Research Institute Aeropropulsion and Power Lab. (AFWAL/64. ADDRESS (City. State and ZIP Code) 7b. ADDRESS (City. State and ZIP Codel300 College Park Wright-Patterson Air Force Base, OHDayton, Ohio 45469 45433-6563
Se. NAME OF FUNDING/SPONSORING 6b. OFFICE SYMBOL 9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBERORGANIZATION (If applcabl) F 33615-8 5-C-25 85
ac. AOORESS (City. Sete and ZiP Cod) 10. SOURCE OF FUNDING NOS.
PROGRAM PROJECT TASK WORK UNITELEMENT NO. NO. NO. NO
62203 F 3066 12 21W, Tfffrf Tfof, Element Analysis fDynamic Structural ResPnse
12. PERSONAL AUTHORIS)Brockman, R. A., Lung, F. Y., Braisted, W. R.
134. TYPE OF REPORT |13b. TIME COVERED 14. DATE OF REPORT (Yr., .4o.. 7y) 15. PAGE COUNTFinalI FROM nCmRr TOOCm March 1989 217
16. SUPPLEMENTARY NOTATION
17. COSATI CODES 18. SUBJECT TERMS (Continue on neuerm if necehsary and identify by block number;FIELD GROUP SUB. GR. Finite Elements Sensitivity Analysis20 11 Plates and Shells Structural Dynamics21 05 g Probabilistic Analysis Vibration
lt. AISTRACT lConlnue on reverse if necesary and identify by block number)
This report describes techniques for the probabilistic dynamic analysis ofplate and shell structures. Statistical variables, which may include materiproperties, thicknesses, or arbitrary geometric parameters, are treated asdiscrete random parameters with normal distribution. Structural responsesensitivities and variance estimates for statistical variables are used toestimate the variances of response variables such as displacement, stress, onatural frequency. Basic solutions and sensitivity analyses are performedusing finite element techniques. The methods described require very littleinformation beyond that needed for a deterministic analysis, but can be usedto develop useful probabilistic data for large models at very low cost.Several key developments discussed in the report contribute to theeffectiveness of the probabilistic analysis method, but have potential
(continued)
20. OISTRIUTION/AVAILAILITY OF ABSTRACT 21. ABSTRACT SECURITY CLASSIFICATION
UNCLASSIFIEO/UNLIMITED [ (A& E AS APT 0 OTIC USERS . Unclassified22. NAME OP RESPONSIBLE INDIVIDUAL 22b TELEPHONE NUMBER 22c OFFICE SYMBOL
Joh Ree hndud. Area Code)John Reed (513) 255-2081 AFWAL/POTC
UNCLASSIFIEDSCURITY CLASSIFICATION OF TMIS PAGE
application in other areas of structural mechanics. The problem of stabil-
izing low-order elements with reduced order quadrature for use in dynamicproblems is addressed; a potential source
of instability is identified and a
mass formulation which produces a stable and accurate element is presented.Layered elements are considered using a shear flexibility correction whichhelps to account for large differences in layer moduli; this device isdemonstrated for layered composites and sandwich wall construction. Verygeneral sensitivity relatiozships are developed for isoparametric elements,for sensitivity parameters which may affect both the nodal element of anelement and the relationship between local and global coordinate axes. Thesesensitivity formulas require much less computation than others in commonuse, and have potential application in shape optimization.
I. &
r!
iUNCLASSIFIED
SECURITY CLASSIIFICATION OF THIS PAGE
FOREWORD
The work described herein was performed between October 1985
and October 1987 at the University of Dayton Research Institute
(UDRI), Dayton, Ohio. This task, "Stochastic Analysis of Bladed
Disk Systems", is part of the program conducted under contract
F33615-85-C-2585, "Structural Testing and Analytical Research
(STAR) of Turbine Components," for the Air Force Wright Aeronau-
tical Laboratories Aero Propulsion and Power Laboratory, AFWAL/
POTC, Wright-Patterson Air Force Base, Ohio.
Technical direction and support for this project were pro-vided by Messrs. William A. Stange and John D. Reed (AFWAL/POTC).
The effort was conducted within the Structures Group (Blaine S.West, Group Leader) of the Aerospace Mechanics Division (Dale H.
Whitford, Project Supervisor). The UDRI Principal Investigator
was Mr. Michael L. Drake.
The authors also wish to acknowledge the contributions ofseveral individuals who made essential contributions to this
work. Dr. Anthony K. Amos of AFOSR made numerous suggestions on
the overall direction of the effort. Mr. Robert J. Dominic of
UDRI provided day-to-day support and encouragement, as well as
technical suggestions and experimental data. Mr. Thomas W. Held
(UDRI) lent expertise in computer operations and communications
whenever it was needed. Dr. Ronald F. Taylor, formerly Group
Leader, Analytical Mechanics, provided administrative and techni-
cal guidance through much of the project.
Aooession For
NTIS GRA&I RDTIC TAB 0Unianounced 11
Juzt!:!Iation
Distribution/Ava1lab Ity Codes
Avail and/orDiet Special
TABLE OF CONTENTS
Page
INTRODUCTION .......... ................. 1
2 ANALYSIS PROCEDURES ....... ............. 5
2.1 LINEAR STATIC SOLUTION ..... .......... 52.2 NATURAL FREQUENCY SOLUTION .... ........ 72.3 STEADY-STATE HARMONIC SOLUTION ... ...... 9
3 SENSITIVITY ANALYSIS ... ............. . 11
3.1 SENSITIVITY FORMULAS FOR ISOPARAMETRIC . . 11ELEMENTS
3.1.1 Shape Functions and the Jacobian . . 12Determinant
7.3.1 Static Analysis of a Tension Strip . 1047.3.2 Statics of a Cantilever Beam . . . . 1047.3.3 Orientation Sensitivity of a Beam . 1077.3.4 Frequency Sensitivity of a Flat Strip 1127.3.5 Frequency Sensitivity of a Beam . . 1157.3.6 Twisted Plate Frequency Sensitivity 117
7.4 PROBABILISTIC ANALYSIS EXAMPLES ..... . 121
7.4.1 Forced Vibration of a Cantilever Beam 1217.4.2 Natural Frequencies of a Twisted Blade 132
REFERENCES ...... .................. . 139
Appendix A. PROTEC Input Data Descriptions ...... .. A-1Appendix B. POSFIL Results File Description . I . I I B-iAppendix C. LAYSTR Layer Stress File Description . . . C-i
Appendix D. PATRAN Interfaces (PATPRO/PROPAT) . . .. D-IAppendix E. DISSPLA Interface (PRODIS) .. ........ . E-i
vii
LIST OF FIGURES
FirPae
1 Bladed Disk 22 Truss Member with One Geometric Variable 203 Local Coordinate System Definition 234 Local Coordinates for Quadrilateral Element 285 Circular Arc with Variable Radius 416 Graphical Interpretation of the Distribution 47
Function #(z)7 Percentile Values of Natural Frequency for a 49
Plate with Thickness Variation8 Bilinear Mindlin Plate Element 559 Hourglass Displacement Pattern 57
10 Combined Hourglass-rotation Mode 6711 Hourglass-rotation Mode in a Regular Mesh 6912 Slender Strip Geometry and Properties 8313 Corner-supported Square Plate 8714 Semi-infinite Plate with Sinusoidal Pressure 9215 Transverse Shear Stresses in Unsymmetric Plate 9316 Square [0/90/0] Plate under Pressure Load 9417 Circular Sandwich Plate 9718 Moment Resultants in Circular Sandwich Plate 9819 Shear Forces in Circular Sandwich Plate 9920 Clamped Sandwich Panel under Uniform Pressure 10121 Rectangular [0/90/0) Laminate 1022d Cantilever Bcam with Tip Load 10623 Cantilever with Specified Angular Orientation 11024 Twisted Cantilever Plate 11825 Frequency Response of Cantilever Beam 12226 Amplitude Sensitivities for Cantilever Beam 12427 Displacement Amplitude Variance versus Frequency 12628 Tip Displacement versus Frequency and Confidence 127
Level29 Displacement-Frequency-Confidence Level Surface 12830 Moment Amplitude Variance versus Frequency 12931 Root Moment versus Frequency and Confidence 130
Level32 Moment-Frequency-Confidence Level Surface 13133 Finite Element Model of 45-Degree Twist Blade 13334 Twist Profiles versus Twist Parameter "C" 13435 Blade Frequencies as Functions of Twist 137
Parameter36 Frequency Variances for Twisted Blade 138
viii
LIST OF TABLES
Table -Page
1 Number of Standard Deviations versus Percentile 48Level
2 Shear Factors for Graphite/Epoxy Laminates 783 Comparison of Results for Planar Vibration of 84
Thin Strip4 Vibration Modes of Thin Strip (Four-Element 86
Solution5 Natural Frequencies for Corner-Supported Plate 886 Natural Frequencies for Free-Free Plate 907 Normalized Stresses for Square [0/90/0) Plate 968 Natural Frequencies of [0/90/0] Plate 1039 Sensitivity Data for Simple Tension Problem 105
10 Displacement Sensitivity Data for Cantilever 108Beam
11 Force Sensitivity Data for Cantilever Beam 10912 Results for Angular Orientation Problem (8=0) i113 Results for Angular Orientation Problem 113
(8=26.565')14 Frequency Sensitivities for Axial Vibration 114
Problem15 Frequency Sensitivities for Cantilever Beam 11616 Frequency Comparison for 30" Twisted Plate 11917 Frequency Sensitivities for 32* Twisted Plate 12018 Natural Frequencies for 45" Twisted Plate 136
ix
CHAPTER 1
INTRODUCTION
Turbomachinery components exhibit more diverse and complex
structural behavior than most classes of engineering structures.
Stress analysis of rotating propulsion system components has been
a driving force in the development of many of the most sophisti-
cated numerical methods in common use: substructuring and cyclic
1Note that the variable e is defined in radians.2 w/ae is computed directly by the sensitivity solution.3Aw/AO is computed by differencing values at 31.90 and 32.10.
120
The relatively low accuracy of the sensitivities in torsionare thought to be an artifact of the bilinear element used in the
frequency calculations. Since the element is integrated with a
single point, the twisted undeformed geometry is not reflected in
the stiffness computation (other than in "nodal offsets" which
occur at each node, and which are taken into account). High
accuracy for twisting modes (and presumably for the corresponding
sensitivities) therefore requires a relatively fine mesh.
7.4 PROBABILISTIC ANALYSIS EXAMPLES
Probabilistic solutions obtained with the methods described
herein are described in this Section. It should be recognized
that the finite element calculations performed in the probabil-
istic analyses are limited to the basic (deterministic) solution
and the sensitivity analyses, so that the numerical behavior
reported for the sensitivity analyses of the previous Section is
typical of the probabilistic solutions as well. The additional
steps of performing variance and percentile calculations complete
the process.
7.4.1 Forced Vibration of a Cantilever Beam
The cantilever beam shown in Figure 22 (see Sections 7.3.2and 7.3.5) is subjected to a uniform pressure load which varies
sinusoidally in time, q=-0.01.sin(wt). The finite element model
used is the same as Mesh 1 of Section 7.3.5, so that the first
resonant frequency is at w=202.3 Hz. We consider forcing fre-
quencies in the range 200:wS205, to determine the steady-state
response behavior of the beam near its first mode.
Figure 25 shows the amplitude of the end deflection versus
forcing frequency. Note that the tip displacement and the forces
are in phase for frequencies lower than the natural frequency,
121
Fk'quency Repneof Cantilever Beam
40-
" 20
0-
~ 20-
-40 ,20 201 202 203 204 205
Arci'ng Preqwunc"y
Figure 25. Frequency Response of Cantilever Beam.
122
and out of phase after crossing the resonance. In the neighbor-
hood of the natural frequency, a step of 0.125 Hz. has been used
for the forcing frequency in generating the results of Figure 25.
Statistical parameters selected for this analysis are the
elastic modulus (E=xl07 ; OE=lXlO 5), mass density (p=2.54x10-4
3488'.585 14241.83 20943.01NFITURAL FREQUENCY MODE
Figure 36. Frequency Variances for Twisted Blade.
138
REFERENCES
1. A. V. Srinivasan, "Vibrations of Bladed-Disk Assemblies -- ASelected Survey," J. Vib. Acous. Stress Reliab. Des. 106,165-168 (1984).
2. D. J. Ewins, "Vibration Characteristics of Bladed Disk Assem-blies," J. Mech Enqnq. Sci. 15(3), 165-186 (1973).
3. D. Hoyniak and S. Fleeter, "Forced Response Analysis of anAerodynamically Detuned Supersonic Turbomachine Rotor," J.Vib. Acous. Stress Reliab. Des. 108, 117-124 (1986).
4. N. A. Valero and 0. 0. Bendiksen, "Vibration Characteristicsof Mistuned Shrouded Blade Assemblies," J. Enqnq. Gas Turb.Power 108, 293-299 (1986).
5. L. E. El-Bayoumy and A. V. Srinivasan, "Influence of Mistun-ing on Rotor-Blade Vibrations," AIAA J. 13(4), 460-464(1975).
6. K. R. V. Kaza and R. E. Kielb, "Effects of Mistuning onBending-Torsion Flutter and Response of a Cascade in Incom-pressible Flow," AIAA J. 20(8), 1120-1129 (1982).
7. W. A. Stange and J. C. MacBain, "An Investigation of DualMode Phenomena in a Mistuned Bladed Disk," J. Vib. Acous.Stress Reliab. Des. 105(3), 402-407 (1983).
8. R. A. Ibraham, "Structural Dynamics with Parameter Uncertain-ties," Appl. Mech. Rev. 40(3), 309-328 (1987).
9. T. Belytschko, "A Review of Recent Developments in Plate andShell Elements," in A. K. Noor (ed.), Computational Mechanics-- Advances and Trends, AMD Vol. 75, 217-231, ASME, New York(1986).
10. , PATRAN II Release Notes, Version 2.1, PDA Engineer-ing, Santa Ana, California, 1986.
11. , DISSPLA (Display Integrated Software System andPlotting LAnguaQe) User's Manual, Version 10.0, ISSCO, Inc.,SanDiego, California, 1985.
12. K. J. Bathe and E. L. Wilson, Numerical Methods in FiniteElement Analysis, Prentice-Hall Co., New Jersey, 1976.
13. 0. C. Zienkiewicz, The Finite Element Method, McGraw-HillCo., New York, 1977.
14. R. D. Cook, Concepts and Applications of Finite ElementA, John Wiley and Sons, New York, 1981.
139
15. R. H. Gallagher, Finite Element Analysis: Fundamentals,Prentice-hall Co., New Jersey, 1975.
16. A. Ralston, A First Course in Numerical Analysis, McGraw-HillCo., New York, 1965.
17. C. A. Felippa, "Solution of Linear Equations with Skyline-Stored Symmetric Matrix," Comp. Struc. 5, 13-29 (1975).
18. K. J. Bathe and E. L. Wilson, "Solution Methods for Eigen-val.ue Problems in Structural Mechanics," Int. J. Num. Meth.Engng. 6, 213-226 (1973).
19. N. S. Khot, L. Berke, and V. B. Venkayya, "Minimum WeightDesign of Structures by the Optimality Criterion and Projec-tion Methods," Proceedings of the AIAA/ASME/ASCE/AHS 20thStructures, Structural Dynamics, and Materials Conference,St. Louis, Mo. (1979).
20. U. Kirsch, "Multilevel Approach to Optimum Structural De-sign," Proc. ASCE, J. Struct. Div. 101, ST1 (1975).
21. R. T. Haftka and B. Prasad, "Programs for Analysis and Resiz-ing of Complex Structures," Comp. Struc. 10, 323-330 (1979).
22. U. Kirsch, M. Reiss, and U. Shamir, "Optimum Design by Parti-tioning and Substructures," Proc. ASCE, J. Struct. Div. 98,ST1, 249-267 (1972).
23. 0. C. Zienkiewicz and J. S. Campbell, "Shape Optimization andSequential Linear Programming," in R. H. Gallagher and 0. C.Zienkiewicz (eds), Optimum Structural Design, John Wiley andSons, 1973.
24. C. V. Ramakrishnan and A. Francavilla, "Structural ShapeOptimization Using Penalty Functions," J. Struct. Mech. 3(4),403-422 (1974).
25. G. K. Smith and R. G. Woodhead, "An Optimal Design Schemewith Applications to Tanker Transverse Structure," Engineer-ina Optimization 1, 79-98 (1974).
26. U. Kirsch and G. Toledano, "Approximate Reanalysis for Modif-ications of Structural Geometry," Comp. Struc. 16, 269-277(1983).
27. S. Wang, Y. Sun, and R. H. Gallagher, "Sensitivity Analysisin Shape Optimization of Continuum Structures," Comp. Struc.20, 855-867 (1985).
28. L. C. Rogers, "Derivatives of Eigenvalues and Eigenvectors,"AIAA J. 8, 943-944 (1970).
140
29. H. Benaroya and M. Rehak, "The Decomposition Method in Struc-tural Dynamics," AIAA Paper 85-0685, 26th AIAA/ASME/ASCE/AHSStruc. Dyn. Mat. Conf., 266-281 (1985).
30. F. S. Wong, "Stochastic Finite Element Analysis of a Vibrat-ing String," J. Sound. Vib. 96(4), 447-459 (1984).
31. W. K. Liu, T. Belytschko, and A. Mani, "Random Field FiniteElements," Int. J. Num. Meth. Engna. 23, 1831-1845 (1986).
32. M. Shinozuka, "Simulation of Multivariate and Multidimen-sional Random Process," J. Acous. Soc. Amer. 49(1), Pt. 2,357-367 (1971).
33. J. E. Freund, Mathematical Statistics, 2nd ed., Prentice-HallCo., New Jersey, 1971.
34. J. E. Ashton and J. M. Whitney, Theory of Laminated Plates,Technomic Publ. Co., 1970.
35. , Stock Catalog and Metals Handbook, American Brassand Copper Co., Oakland, Calif., 1966.
36. R. D. Mindlin, "Influence of Rotatory Inertia and Shear onthe Bending of Elastic Plates," J. Appl. Mech. 18, 1031-1036(1951).
37. T. J. R. Hughes, M. Cohen, and M. Haroun, "Reduced and Selec-tive Integration Techniques in Finite Element Analysis ofPlates," Nucl. Eng. Des. 46, 203-222 (1978).
38. D. Kosloff and G. Frazier, "Treatment of Hourglass Patternsin Low Order Finite Element Codes," Num. Anal. Meth. Geomech.2, 52-72 (1978).
39. D. Flanagan and T. Belytschko, "A Uniform Strain Hexahedronand Quadrilateral with Orthogonal Hourglass Control," Int. J.Num. Meth. Engng. 17, 679-706 (1981).
40. T. Belytschko, J. I. Lin, and C. S. Tsay, "Explicit Algor-ithms for the Nonlinear Dynamics of Shells," Comp. Meth.Appl. Mech. Engna. 42, 225-251 (1984).
41. R. H. MacNeal, "A Simple Quadrilateral Shell Element," Comp.Struct. 8, 175-183 (1978).
42. T. J. R. Hughes and T. E. Tezduyar, "Finite Elements Basedupon Mindlin Plate Theory with Particular Reference to theFour-node Bilinear Isoparametric Element," J. Appl. Mech. 48,587-596 (1981).
43. R. L. Taylor, "Finite Element for General Shell Analysis,"5th Intl. Seminar on Computational Aspects of the FiniteElement Method, Berlin, August 1979.
141
44. T. Belytschko, C. S. Tsay, and W. K. Liu, "A StabilizationMatrix for the Bilinear Mindlin Plate Element," ComD. Meth.A221. Mech. Engng. 29, 313-327 (1981).
45. T. Belytschko and C. S. Tsay, "A Stabilization Procedure forthe Quadrilateral Plate Element with One-point Quadrature,"Int. J. Num. Meth. Enana. 19, 405-419 (1983).
46. W. K. Liu, J. S. Ong, and R. A. Uras, "Finite Element Stabil-ization Matrices - A Unification Approach," Comp. Meth. Ap21.Mech. Engna. 53, 13-46 (1985).
47. W. K. Liu, E. S. Law, D. Lam, and T. Belytschko, "Resultant-Stress Denerated-Shell Element," Comy. Meth. A2D1. Mech.E 55, 259-300 (1986).
48. K. C. Park, G. M. Stanley, and D. L. Flaggs, "A UniformlyReduced, Four-noded CO Shell Element with Consistent RankCorrections," Comp. Struct. 20, 129-139 (1985).
49. J. M. Whitney, "Shear Correction Factors for OrthotropicLaminates under Static Load," J. A2)1. Mech. 40, 302-304(1973).
50. T. J. R. Hughes, "Recent Developments in Computer Methods forStructural Analysis," Nucl. Ena. Des. 57, 427-439 (1980).
51. R. M. Jones, Mechanics of Composite Materials, Scripta BookCo., Washington, D. C., 1975.
52. R. A. Brockman, "Current Problems and Progress in AircraftTransparency Impact Analysis," in S. A. Morolo (ed.), Proc.14th Conf. on Aerospace Transparent Materials and Enclosures,AFWAL-TR-83-4154, Air Force Wright Aeronautical Laboratories,Wright-Patterson Air Force Base, Ohio, 1058-1082 (1983).
53. R. A. Brockman, "On Vibration Damping Analysis Using theFinite Element Method," in L. Rogers (ed.), Vibration Damping1984 Workshop Proceedings, AFWAL-TR-84-3064, Air Force WrightAeronautical Laboratories, Wright-Patterson Air Force Base,Ohio, pp. 11-1:11-10 (1984).
54. A. N. Palazotto and W. P. Witt, "Formulation of a NonlinearCompatible Finite Element for the Analysis of Laminated Com-posites," Comput. Struct. 21, 1213-1234 (1985).
55. H. P. Huttelmaier and M. Epstein, "A Finite Element Formula-tion for Multilayered and Thick Shells," Comput. Struct. 21,1181-1185 (1985).
56. S. T. Mau, P. Tong, and T. H. H. Pian, "Finite Element Solu-tions for Laminated Thick Plates," J. Comp. Mat. 6, 304-311(1972).
142
57. R. L. Spilker, "Hybrid-stress Eight-node Elements for Thinand Thick Multilayer Laminated Plates," Int. J. Num. Meth.Engnaj 18, 801-828 (1982).
58. R. L. Spilker, "An Invariant 8-node Hybrid-stress Element forThin and Thick Multilayer Laminated Plates," Int. J. Num.Meth. Engng. 20, 573-582 (1984).
59. R. L. Spilker and D. M. Jakobs, "Hybrid Stress Reduced-Mindlin Elements for Thin Multilayer Plates," Int. J. Num.Meth. Engna. 23, 555-578 (1986).
60. J. M. Whitney and N. J. Pagano, "Shear Deformation in Hetero-geneous Anisotropic Plates," Trans. ASME, J. Appl. Mech. 37,1031-1036 (1970).
61. N. J. Pagano, "Exact Solutions for Composite Laminates inCylindrical Bending," J. ComR. Mater. 3, 398-411 (1969).
62. N. J. Pagano, "Exact Solutions for Rectangular BidirectionalComposites and Sandwich Plates," J. Comp. Mater. 4, 20-34(1970).
63. J. J. Engblom and 0. 0. Ochoa, "Finite Element FormulationIncluding Interlaminar Stress Calculations," Comp. Struct.23(2), 241-249 (1986).
64. P. Sharifi, "Nonlinear Analysis of Sandwich Structures,"Ph.D. Thesis, Univ. California, Berkeley, 1970.
65. H. P. Kan and J. C. Huang, "Large Deflection of RectangularSandwich Plates," AIAA J. 5, 1706-1708 (1967).
66. G. R. Monforton, "Discrete Element, Finite DisplacementAnalysis of Anisotropic Sandwich Shells," Ph.D. Thesis, CaseWestern Reserve Univ., 1970.
67. R. A. Brockman, "MAGNA: A Finite Element System for Three-Dimensional Static and Dynamic Structural Analysis," Comput.Struct. 13, 415-423 (1981).
68. T. R. Tauchert, Energy Principles in Structural Mechanics,McGraw-Hill Co., New York, 1974.
69. R. E. Kielb, A. W. Leissa, and J. C. MacBain, "Vibrations ofTwisted Cantilever Plates - A Comparison of TheoreticalResults," Int. J. Num. Meth. Engnc. 21, 1365-1380 (1985).
143
APPENDIX A
PROTEC INPUT DATA DESCRIPTIONS
The computer program in which the analysis techniques re-
ported herein are implemented is called PROTEC (probabilistic
Besponse Of Turbine Engine Components). PROTEC is written in
ANSI FORTRAN 77, and has been executed successfully on CDC Cyber,
CRAY X/MP, and DEC VAX machines. This Appendix summarizes input
requirements for PROTEC. The remaining Appendices of this report
describe PROTEC file output (Appendices B and C), data conversion
between PROTEC and the PATRAN10 modeling package (Appendix D),
and plotting of probabilistic data using DISSPLA (Appendix E).
Input to PROTEC is arranged in a series of input "blocks".
Each input block begins with a header line identifying the block,
followed by the data, and ends with a blank line signifying the
IBEG = First node number to be constrained.IEND = Last node in a series of nodes to be constrained.INCR = Node number increment.ID1-ID6 = List of nodal degrees of freedom to be fixed; the
numeric values 1-6 refer to u, v, w, 8 9 , 8respectively.
Notes:
" If IBEG, IEND, and INCR are all present, each of thenodes IBEG, IBEG+INCR, IBEG+2*INCR, ..., IEND areconstrained.
" If IEND is omitted, the default is IBEG (single node).
" If INCR is omitted, a default of INCR = 1 is assumed.
" Only one degree-of-freedom value (IDn) is required.
A - 3
COORDINATE
Input Block
COORDINATE Input Block: Nodal Coordinate Data
Header: COORFormat:
5 10 20 39 40NODEl INCR XCOORD. YCOORD ZCOORDI
Example:
1 121 1 10.281 -67.428751 1.257E-
Variables:
NODE = Current node number.INCR = Increment for node number generation.XCOORD = Cartesian coordinate X at the current node.YCOORD = Cartesian coordinate Y at the current node.ZCOORD = Cartesian coordinate Z at the current node.
Notes:
" Valid node numbers are from one to the maximum number inthe model. Intermediate node numbers may be omitted, butmust be constrained.
o INCR is used for generating a series of nodes along aline from two successive lines of data. For example, theinput
DERIVATIVES Input Block: Coordinate Derivatives for SensitivityAnalysis
Header: DERI <value>Format:
I NODEI INCRI XDERIVl YDERIVI ZDERIVl
Example:
i 51 1 0.51 0.01 0.21
Variables:
<value> = Integer value in header line, specifying IDENT(parameter i.d., see PARA input block) for thesensitivity parameter being defined.
NODE = Current node number.INCR = Increment for node number generation.XDERIV = Derivative of Cartesian coordinate X at the
current node, with respect to this parameter.YDERIV = Derivative of Cartesian coordinate Y at the
current node, with respect to this parameter.ZDERIV = Derivative of Cartesian coordinate Z at the
current node, with respect to this parameter.
Notes:
o This block is used to define a geometric parameter foruse in sensitivity analysis (that is, a parameter whichcontrol.; .hc placement of nodes in the model). A DERIblock is required for each such parameter; multiple DERIblocks are distinguished from one another by the <value>appearing in the header line.
o NODE numbers are as defined in the COOR input block, andINCR is used to generate data exactly as the COOR block.
o If the sensitivity parameter is p, then XDERIV = ax/ap,the derivative of coordinate X.
0 Nodes for which all derivatives are zero for the currentparameter may be omitted.
A - 5
DIAGNOSTICS
Input Block
DIAGNOSTICS Input Block: Selection of Diagnostic Output Options
IDSWi = Number of a diagnostic output switch to beactivated during the present analysis.
Notes:
" Continue input on additional lines until all selectionshave been made. Each input line may contain from one tosixteen switch values.
" Valid diagnostic output options are as follows:
IDSWi Description of Diagnostic Output
1 Element data (nodes, properties, coordinates)2 Element stiffness matrices3 Element mass matrices4 Element harmonic stiffness matrices (K-XM)5 Element stabilization (artificial) forces6 Element transformation matrices7 Element local coordinates8 Element shape functions9 Element strain-displacement matrices10 Element stress-strain matrices11 Element local displacements12 Element displacement sensitivities
A - 6
ELEMENT
Input Block
ELUENT Input Block: Finite Element Connections and Properties
Header: ELENFormat:
5 19 1 29 25 39 4 45 50
ETYJ! IDEL MATLI IPRIINGENIIEGENI N11 N21 N31 N4r
Example:
ISHELLI 211 11 2 1 1 2641 2881 2951' 276
Variables:
ETYP = Mnemonic for element type.IDEL = Element number for current element.MATL = Material number for current element. A negative
value refers to a laminate number, as defined inthe LAMI input block.
IPR = Physical property set number for this element.INGEN = Node increment for element generation.IEGEN = Element increment for element generation.Nl-N4 = Nodes connected to the current element, listed in
counterclockwise order around the boundary.
Notes:
" At present, the only acceptable element type mnemonic(ETYP) is "SHELL", designating the bilinear, 24-D.O.F.Mindlin plate/shell element.
o Valid element numbers are from one to the total number ofelements in the model.
" Elements may be generated in any pattern which involvesequal increments in all node numbers Nl-N4. INGEN andIEGEN appear on the second input line of a pair, andspecify node and element number increments, respectively.For instance, the data
ISHELLI 201 1I 11 1 1 101 141 161 121ISHELLI 261 11 11 31 21 I 1i {generates the element data:
FORCE Input Block: Imposed Nodal Forces, Moments, Displacements,and Rotations
Header: FORCFormat:
19 20 2NODEI KODEI VALUE' ICA&
Example:
5 281 1I 279.5 11
Variables:
NODE = Node at which load or displacement is specified.
KODE = Code for type and direction of prescribed value:
1 =F; 2 =F; 3 =F; 4 = Mx; 5 = My; 6 = Mz;
7 = x 8 = uy; 9 = uz; 10= $x; 11= y 12= z
VALUE = Value of prescribed force, moment, displacement,or rotation.
ICASE = Static load case number.
Notes:
O The first three values must be provided. There are nodefault values for NODE or KODE. ICASE, if omitted, isassumed to be 1.
o If a nodal displacement or rotation is set to zero inthis input block, the effect is the same as a constraintspecified in the BOUNDARY input block.
A - 8
GRAVITY
Input Block
GRAVITY Input Block: Self-Weight Body Force on Entire Model
Header: GRAVFormat:
19 2230GX1 GY Gal
Example:
0.01 0.0 -386.L
Variables:
GX,GY,GZ = Cartesian components of gravity vector, definingboth the magnitude and direction of the localgravitational acceleration.
Notes:
" The gravitational force per unit volume at any point isdetermined from F = p(GXi + GYj + GZk), in which p is thematerial density at the point.
" By default, gravity loads become part of load case 1.
A - 9
LAMINATE
Input Block
LAMINATE Input Block: Laminate Definitions for Layered Shells
Header: LAMIFormat:
(1) Sizing Data (one per laminate):
(2) Layer Data (one per layer):
1 MATLI THICK ANGL
Example:
I if 31
2 0.5001 45.0EA 0.0601 0.0
Variables:
LAM = Laminate number.NLAY = Number of layers in current laminate.MATL = Material number for a specific layer.THICK = Layer thickness.ANGLE = Angle from local 'x' axis of an element to the
material '1' axis (fiber direction).
Notes:
o Laminate definitions must be numbered sequentially andinput in ascending order.
o The layers of a laminate are numbered from bottom (layer1) to top (layer NLAY).
o MATL may reference either an isotropic or orthotropicmaterial, as defined in the MATERIAL input block.
o ANGLE is positive counterclockwise when viewing anelement from the top.
o ANGLE is measured in degrees.
A - 10
MATERIAL
Input Block
MATERIAL Input Block: Material ?roperties Data
Header: MATEFormat:
(1) For isotropic materials (one line/material):
MA' 10 20 30 49 50. . .// El xNUI RHOI : SYI
(2) For orthotropic materials (two lines/material):
MAT = Material number for current material.E = Extensional modulus.XNU = Poisson's ratio.RHO = Mass density.SY = Yield stress.El = Extensional mudulus in material direction '1'.E2 = Extensional modulus in material direction '2'.XNU12 = Major inplane Poisson's ratio.G12 = Shear modulus in material (1,2) plane.G13 = Shear modulus in material (1,3) plane.G23 = Shear modulus in material (2,3) plane.Cl, C2 = Failure stress constants.
Notes:
o Materials may be entered in any order, but should benumbered from 1 to the total number of materials, withfew gaps.
o The relationship E = 2G(l+v) is assumed for isotropicmaterials.
o Mass densities must be entered in units consistent withforce, length, and time units used elsewhere in input.
0 Constants Cl, C2 are currently not used.
A - 11
OPTION
Input Block
OPTION Input Block: Selection of Solution Options
Header: OPTIFormat:
(Enter keywords and values as described below.All input in this block may be in free format.)
EIGENVALUE . . . . Selects natural frequency solutionFREQUENCY <values> Defines forcing frequencies for steady-
state harmonic solutionHARMONIC ....... .. Selects steady-state forced harmonic
vibration solutionLOAD CASES .... Defines number of static loading casesMODES. ........ .Requests a specified number of natural
frequencies in an eigenvalue analysisSENSITIVITY <name> Requests sensitivity analysis following
a basic solution, to determine responsederivatives
SSITERATIONS . . . Defines the maximum number of iterationcycles for eigen, alue solutions
SSTOLERANCE . . Defines the relative accuracy toleranceused to test eigenvalue convergence
STATIC ....... .. Selects linear static solution
Notes:
" Linear static analysis normally requires the STATIC andLOADCASES options.
o Steady-state harmonic analysis normally requires the useof HARMONIC and FREQUENCY options.
o Natural freauency analysis normally requires the use ofEIGENVALUE and MODES options.
o Valid names for the SENSITIVITY option are: STATIC,HARMONIC, and EIGENVALUE.
o The first four characters of each keyword (shown in boldabove) must be present.
A - 12
OPTIONInput Block(Continued)
Defaults:
0 LOADCASFS= 1, if STATIC option is specified.
0 SSITERATIONS= max( 2*NODES, 10 ) if EIGENVALUE specified.
0 SSTOLERANCE= 1.E-6, if EIGENVALUE specified.
A - 13
PARAMETERSInput Block
PARAMETERS Input Block: Definition of Control Parameters forStatistical or Sensitivity Analysis
Header: PARAFormat:
(1) SizinQ Data (one line only):
I NPARI,(2) Control Parameter Data (one line/parameter):
5 10 15 251 IPAR ITYPEIIDENT! STDDEV
Example:
21 1 4 100000.2 4 999 0.051
Variables:
NPAR = Number of control parameters to be defined.IPAR = Sequence number of current parameter.ITYPE = Parameter type: 1 = modulus; 2 = density; 3 =
thickness; 4 = geometric.IDENT = I.D. of material, property set, or other data
corresponding to the current parameter.STDDEV = Standard deviation of current parameter.
Notes:
o Valid sequence numbers IPAR are from 1 to NPAR; numbersoutside this range are ignored.
" For ITYPE = 1,2,3, the parameter being defined is simplya property value defined elsewhere in the MATERIAL dataor PROPERTY data. When ITYPE = 4, the parameter controlsthe positions of nodes in the model, and requires someadditional data for its definition (see DERI block).
o IDENT refers to a material number if ITYPE = 1 or 2. IfITYPE = 3, IDENT refers to a physical property number asdefined in the PROPERTY input block.
A - 14
PARAMETERSInput Block(Continued)
o When ITYPE = 4, the geometric parameter is defined by thederivatives aX/ap, 8Y/8p, az/ap of coordinates at certainnodes. These derivatives must be specified in a DERIVAT-IVE input block, with the value of IDENT specified in theblock header.
o STDDEV is unnecessary for sensitivity analysis alone, butmust be defined when probabilistic information about theresponse is to be computed.
O The units of STDDEV must be the same as those of the meanvalues defined elsewhere (e.g., a modulus value definedin MATERIAL data). For ITYPE = 4, STDDEV might have thesame units as the nodal coordinate data (if the parameteris a key dimension), or different units (if the parameteris an angle, for instance).
A - 15
PRESSURE
Input Block
PRESSURE Input Block: Element Pressure Loading
Header: PRESFormat:
IEBEG 10 1 E?IEEND I IENCR| PRESSIICAE4
Example:
51 351 21 - 50.0 0
Variables:
IEBEG = First element number to which the specifiedpressure is to be applied.
IEEND = Last element to which pressure is applied.IENCR = Element number increment.PRESS Surface pressure, positive outwardICASE = Static loading condition number.
Notes:
o Pressures are applied to elements IEBEG, IEBEG+IENCR,IEBEG+2*IENCR, ... , IEEND.
o If IEEND is not given, its default is IEBEG (one elementloaded).
o If IENCR is not specified, the increment is set to one(all e.ements from IEBEG to IEEND loaded).
o The "outward" direction for an element is determined bythe ordering of its nodes. When the element is viewedfrom the top (nodes Nl-N4 arranged counterclockwise), apositive (outward) pressure acts upward, toward theviewer.
o If ICASE is omitted, load case 1 is assumed.
A - 16
PROPERTY
Input Block
PROPERTY Input Block: Element Thicknesses and Areas
Header: PROPFormat:
1 IPRI VALUE
Example:
41 0.375
Variables:
IPR = Property set number.VALUE = Property value (area for 1-D elements, thickness
for 2-D elements and shells).
Notes:
0 Property sets may be entered in any order, but should benumbered from 1 to the total number of distinct elementproperties (or with few gaps).
A - 17
TITLE
Input Block
TITLE Input Block:
Header: TITLFormat:
Example:
I Sensitivity Analysis of Blade with Variable Twist
Variables:
TITLE = Alphanumeric problem title.
Notes:
o TITLE may include any valid alphanumeric characters.
A - 18
APPENDIX B
POSFIL Results File Description
This Appendix documents the results file output written from
PROTEC. The results file POSFIL is a formatted, card-image file
whose structure is rigid (and therefore simple to read from other
programs). The PATRAN translator PROPAT (see Appendix D) is an
example of a program which reads this results file and transmits
data to other programs for analysis and display.
Data on POSFIL are arranged in blocks, similar in concept to
the input data blocks (Appendix A). Each data block begins with
a header line identifying the block, followed by the data, and
ends with an empty line signifying the end of the block. Types
of data blocks generated as output include:
Block Naxe Description
BOUN Nodal boundary conditionsCORD Nodal coordinatesDISP Nodal displacementDSEN Nodal displacement sensitivitiesELEM Element connectionsESEN Eigenvalue (frequency) sensitivitiesFREQ Harmonic forcing frequencies or system natural
frequenciesLOAD Nodal forces and prescribed displacementsMATL Material propertiesMSEN Mode shape sensitivity coefficientsPATR Patran neutral file titlePVAR Sensitivity parameter variancesREAC Nodal force reactionsSSEN Element stress sensitivitiesSTRS Element stress resultantsTITL Alphanumeric problem title
Formats for the individual data blocks and block headers are
ICASE L,;,d case/mode numberNUMNOD Number of nodesIPARAM Sensitivity parameter numberNODE Node number 18, 8X,DISP(6) Nodal displacement and rotation 3E16.8,/,
NUMELT Number of elementsELTYPE Element type A8, 718IELT Element numberMATLNO Material numberIPROP Property numberNCON(4) List of connected node points
ESEN 'ESEN' Block identifier A8, 218NUMMOD Number of vibration modesNUMPAR Number of sensitivity param's.MODE Mode number 218,IPARAM Sensitivity parameter number 2E16.8FREQ Natural frequency
NUMMOD Number of vibration modesNUMPAR Number of sensitivity param's.MODE Mode number 318, E16.8IPARAM Sensitivity parameter numberINDEX Coefficient numberCOEFF Mode shape sensitivity coeff-
icientPATR 'PATR' Block identifier A8
DATE Date neutral file generated A12, A8,TIME Time neutral file generated A10PATVER Patran version number
PVAR 'PVAR' Block identifier A8, 18NUMPAR Number of sensitivity param's.IPARAM Sensitivity parameter number 318, E16.8ITYPE 'I' = material modulus
'2' = material density'3' = thickness'4' = geometric parameter
IDENT Material, property, or DERIVblock i.d. for this parameter
ICASE Applied loading case numberNUMNOD Number of nodesNODE Node number 18, 8X,FORC(6) Nodal reaction force 3E16.8, /,
and nodal reaction moments 16X.3E16.8SSEN 'SSEN' Block identifier A8, 318
NCASE Number of load cases or modesNUMELT Number of elementsNPARAM Number of sensitivity param's.IDEL Element number I8,A8,318ELTYPE Element typeICASE Load case/mode numberIANAL '4' = Static sensitivity
'5' = Frequency sensitivity'6' = Harmonic sensitivity
STRS 'STRS' Block identifier A8, 218NCASE Number of load cases or modesNUMELT Number of elementsIDEL Element number 18,A8,218ELTYPE Element typeICASE Load case/mode numberIANAL 'I' = Static solution
'2' = Natural frequency'3' = Steady state harmonic
SSIGVM(3) von Mises stressesTITL 'TITL' Block identifier A8
TITLE Alphanumeric problem title 80Al
B - 4
The example below shows the POSFIL output for a very simplefinite element model. For larger models, the nodes, elements,degrees of freedom, and other data are repeated as required inthe same formats.
-T ITINatural frequencies of square plate. inptane motions onty, one eLement
This Appendix describes the data translation performed by
PATPRO (PATRAN-to-PROTEC) and PROPAT (PROTEC-to-PATRAN). PATPRO
converts a finite element neutral file from the geometric model-
ing program PATRAN into a standard input file for finite element
analysis by PROTEC. PROPAT transforms a PROTEC results file into
a PATRAN results file for postprocessing. PATRAN1 0 is a product
of PDA Engineering in Santa Ana, California.
The modeling-analysis-postprocessing cycle begins in PATRAN,
where the finite element model is generated. The completed model
is written (by PATRAN) to a PATRAN Neutral File. A Neutral File
is a card-image text file which contains geometric data, node and
element definitions, properties data, loads, constraints, and
model identification parameters. From the Neutral File, PATPRO
generates most of the data required to perform a finite element
analysis with PROTEC.
When the analysis is complete, the results file POSFIL (see
Appendix B) may be processed using PROPAT to create plotting data
files compatible with PATRAN. Results files, together with the
original PATRAN Neutral File, are then used within PATRAN for the
graphical display of stress and displacement results.
Both PATPRO and PROPAT are written in ANSI FORTRAN-77, and
are operational on the DEC VAX under VMS and CDC Cyber under NOS.
Important features and limitations for each of the programs are
noted in the paragraphs below.
PATPRO (PATRAN-to-PROTEC) : PATPRO uses the PATRAN Neutral
File to generate most of the PROTEC input needed for an analysis.
PATRAN data types which can be translated are shown in the Table
below.
D - 1
PATRAN PROTEC Notes andPacket Data Block Description Restrictions
25 TITL Problem title
26 PATR Model identification
1 COOR Nodal coordinates
2 ELEM Element connections QUAD/4 (SHELL) only
3 MATE Material p'-perties Isotropic materials
4 PROP Physical properties Element thicknesses
6 PRES Pressure loads Element avg. only
7 FORC Nodal forces
8 FORC Nodal displacements
All nodes present in the PATRAN model are translated into
PROTEC format, without resequencing. The model should be fully
equivalenced (i.e, duplicate nodes eliminated) in PATRAN before
writing the Neutral File. We also recommend the node renumbering
facilities in PATRAN, which are extremely effective; the RMS
WAVEFRONT criterion is most appropriate when the analysis is to
be performed using PROTEC.
When data blocks other than those listed above are needed,
these must be entered manually using a text editor. Examples are
the OPTIons, SENSitivity, and LAMInate input blocks.
PROPAT (PROTEC-to-PATRAN): PROPAT processes the results file
POSFIL) generated by PROTEC, and produces PATRAN-compatible
files containing nodal or element results "columns". The PATRAN
results files are binary files and cannot be listed or printed;
PROPAT will, at the user's option, generate formatted versions of
the results files for printing. For postprocessing, both the
binary results files from PROPAT and the original PATRAN neutral
file must be supplied to PATRAN. Postprocessing options include
plots of deformed geometry, stress or displacement contours, and
color-coded plots of key element or nodal results from PROTEC.
D - 2
The listings which follow demonstrate the operation of the
PATRAN interface programs, and show the types of data which are
generated at each stage of the process. The table below gives a
summary of the sample listings.
Listi Title DescriptionD.1 PATRAN Session File Keyboard input to PATRAND.2 PATRAN Neutral File Model as output from PATRAND.3 PATPRO Execution Change PATRAN data to PROTEC formatD.4 PROTEC Input Data Final PROTEC input fileD.5 POSFIL Results File Results file output by PROTECD.6 PROPAT Execution Change results file to PATRAN formatD.7 Element Results File Element results as used in PATRAND.8 Nodal Results File Nodal results as used in PATRAND.9 PATRAN Session Interactive postprocessing
D - 3
4
A-0A
z o: P ns 0 .03' 3i
.4 C. C. C 0 0 C 0 0 0 0 0 0. C C C 0o C 0 C 0 C 0 0 C ~ 0 . 0 C C 0 0
* ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ C C o c 0 C e C 0 C~
o 0 0 C ' C C 0 0 0 '4 3C 1 1. a, l C C40 0 0 0 Cr 0 L i .4 . i I 0
So00 .OC .o 0.oC0 CC 00 0, O 00 -0 000 CO 0 C C COI
C~~~~I '4 C 0 C 0 0 C . v C 4 0 0C C
00 CC C C' GO C 00 C CO 0 00 00 0 CC CL OW 0W CC OW 0'
0, aC UC 0C . 1.. 0 0 0a WC 0 C '0 .0.0 C0 00 CC 00 C0 C0 C0 00 0 C Co 0 00 0C 4 C
C .C 0 C 0 0 0 0 C C
* C C 0 0 C Li C I C 0 C0 4' C LO C 11 0 C ' i . C C IC L 0 . C00
a . , 0CO .00 0 CC C 00 C 0.)0 aC CC .00 00 - CC C CC CO CO
. .. C 0 .- C. 0 C 0 0 :o4 CC . 0 C T 0 C .. .. C 0 . 0 0 Cl C* 0 C
1 10 1 0 .0 CO
O~ C C C C C C C C C 0 0 0 0 C 0 0 00~ ~ C C 0 0 0
a LC C 00 CCC C C .0 ri LICC I 0C CC CC C COO CC 0 l 00 0 CC0 C C C Li L C 0 C 04C C 5
o~~~~~ ~~ aiC L I C L 4 C 0 L I , I L 0 0 0 0 9
4 0 4 0 C C 0 0 ' 0 0 0 '4 C CC0 0 0
0 1. 0
.3 0 0 0 * 0 0 0 0 0 0 0 0 0 0
0 * @ 0 00e
o 0 0 a 0 0 0 0 0 e oe e eo e o ee
*~ ~~ ~ 0 0 0 0 0 0 0 @ 0 0 0 @00
o 3 0 0 000 0 0 00 0e, 0 0 00 0 0
o @ ~ e 0 0 0 0 000 000 0 0 -006000 0o0 e0 0: .0 00 0 .0 0 .0 00 0 000000.400000000000* 0 0 0 3 e e 0 0 . . o e M e:o 00 0 0 0 0 0 a 0000 0 0 0 00 0
0~ ~~ ~ . 00 0 0 0 00 0 000 F 4 0 0 0
0 0
0 0 0 0 0 0 000 0 0 0 0 0 0
o 0 .: . . . . . . . 0 0 0 o e o o O O e oe
00 S . S M S ~ u u u u u M3SSS.SSSSSSo ~ ~ ~ ~ ~ ~ ~ ~ ~ o : a: 03 0 c 0 0 0 0 0
o-9 RE-;0, reel l.9. . -M . n C. freC3o C .C, t0CC 03C 3C COCO 30 0 0 a0 0 t000c 0mm r, * . .9Z 999 0 * C.99) . r . I I.1
-nn 1 4o 4
ief irn -T 10 ';1~. '!- 1. T1 C
7 . -4 CD z,4 T~
n a~- -4.9 & *C' ad~f A.9.0 rd .d r-' NC
£.: .9(9 99. CtJ C 4 0 ~ Q'i 4CI,
~ C~'4 r Ton . ZVI q-9 r. r,- -J T~ 44
CO0 OCC. 00CC C..C. CoOC 000o00 0003 300 0MM30
jY 0 . C- . d .. 0 '9 . 0 .. 4 C . C - d f 4 - ' 9 r C 0 4 - ~ . 0 4 n l 'o0.~ ~ ~ ~ ~ ~~~~~I o00n0C ~ 004 0~~8 ~ooe* ~ 9 .L. Aweeo~0U~ L' ~ L~... w w w Iwso~w ~IA
0"r.r. rr ... r l, 21Cr11.1.r n,
?00 .Q.3C o330? 0 0C0'0 0000 o1 30. 0 0 a
0 M tC 0*N,9aN.. o. N.
00 000O00 0000 000r 0000 000 000a 0
6t 0 00 ?900 . 1.9 '9.10~0 .'
0C'9 90990 r-.C %0
000 0000 000 00 0 000 00 0 ? 000 ? .0 0000 ?000 0
Nfl 4 ' 3w -2 Ong999 '
n 9 4M.0 NC-fIN)S 0 A .t N9iE 4 9 i
op 00 0 0003 300 000c" 000? (0??0 0
17I "'
TII
flf-W -N1. 4No fl - -N 9o tO NU. N,1
o . - -or., ,r
'-4
-A4m 0 00 rl
-~~t trr rdo.: r r P4~rr.
Cot? 0 0 U,
O'nW M~ 4 -
0. A ~ A, ontoZ 'to In e 11 r w
000 000
000 000 000 000 000 0000 0000
10 0 XC. Cifew NIL *'Wi.tiV
4m m IM i eaw.1,v-ol0 i pram--
000 0000 00?00 000?0 0000 0000o 0000 0000C 00
,t m ,,nN ent omt Nw ort
-e-e~ 1 W etN
000~~0 000 000 00CC 00 000 CQ ?000000
8?8 ~'So 00 000C 0000, r000 o0c 0
M4! M 0N NO N9N mj ; m m Nmm o-r v o m v N ; t t omON t ot
90, -1 84 ;N t or m - -ifi -- - -
*-ew Ut f to f N-.- A" toN.t r t fi n 0 -- v .. Noo 0oo oco C, 000.0 ooc C. C
D - 18
lit '
. 5 a v l
~ 1W".. . . -w . -
PHS
,..., ' , iri Z i ,A! 9D 19
.U D - 19
. * . * .. -, oc ~ e l e+ )l + m'". . . .' l l.. .. .. .. ... ...... .........c. .* UN . 0.. - *%f *f* qCO 2 al M Dh M 11 S o El •i
AVERAGING COLUMN 5 DF ELEMENT RESULTS FILE AT NODES.DATA WIDTH - 15FILE TITLE - EXAMPLE PROBLEM
PROTEC ANALYSISEIGENVALUF. RESULTS
DATA VALUES RANGE FROM .123E+05 TO .467E 06
Listing D.9. Continued
D - 25
ASSI(GMENT? 1.AUTO 2.MANUAL 3.SEXI-AUTO 4.USE CURRENT LEVELS 5.END>1
ASSIGNED CONTOUR VALUE CODES FOLLOW:
A .4110E+06 B .3815E 06 C .3520E+06D .3224E 06 E .2929E+06 F .2634E 06G .2338E+06 H .2043E+O6 I .1748E06J .1452E+06 K .1157E+06 L .8617E+05M .5664E 05 N .2711E 05
A SINGLE COLUMN NODAL FILE CALLED "PATNOD " HAS BEEN PRODUCED.
POSTPROCESS? 1. DEFORMATIONS 2. ELEMENT QUANTITIES 3. END>RUN, HIDE, CONT
BEGINNING PHASE-II HIDDEN LINE PLOT OF ACCURACY LEVEL .20
RESTART DATA BEING WRITTEN ON 87/09/29PDA/PATRAN COMPLETED
Listing D.9. Concluded
D - 28
APPENDIX B
DISSPI INTERFACE (PRODIS)
PRODIS (PROTEC-to-DISSPLA) is an output processor for PROTEC
which performs two primary functions:
0 probabilistic (variance) computations
o presentation graphics using the DISSPLA11 library
PRODIS uses the results file POSFIL (Appendix B) to generate x-y
plots, surface plots, and histograms. Data used in PRODIS plots
also can be written to separate files for use in other programs.
PRODIS is written in ANSI FORTRAN-77, and is operational on the
DEC VAX under VMS and CDC Cyber systems under NOS.
PRODIS generally allows plotting of any quantity versus
another, although some combinations are best suited for specific
types of analysis. Quantities which can be selected for plotting
include:
o displacement components at a specified nodeo displacement magnitude at a specified nodeo maximum displacement for a collection of nodeso principal moment for a specified elemento von Mises stress for a specified elemento maximum moment or stress for a collection of elementso harmonic forcing frequency
In some cases, it is desirable to plot only one of the above
quantities for a series of load cases (static analysis) or modes
(natural frequency analysis). PRODIS will generate histograms
for such cases, which permits an easy comparison of effects from
different analysis cases. Results from steady-state harmonic
analyses, with forcing frequency as an independent variable, are
typically presented as x-y plots or 3-D surfaces.
Two modes of presentation are included in PRODIS for display
of probabilistic data. In static or natural frequency analysis,
E - 1
variance data for nodal or element results can be displayed in
histogram form. The histogram shows variances in the requested
quantity for each individual statistical parameter, and for all
parameters combined. Recall that, for any result 7 which depends
on the statistical parameters pi, the total variance is (see
Section 4.3):
nVar[r] E [I _ _2 Var[pi]
i=1 1p
In effect, the histogram displays each term in this series as
well as the total, for each of a series of loading conditions or
vibration modes. This type of plot is useful for determining
which statistical parameters contribute most to the uncertainty
in the computed result, and for comparing this data for different
modes or loading conditions.
The second mode of presentation for probabilistic results is
most often used in steady-state harmonic analysis, where forcing
frequency is nearly always an independent variable. This being
the case, one can assemble frequency response (i.e., amplitude
versus frequency) results for the deterministic response, or for
a given percentile level (confidence level). Amplitudes versus
both forcing frequency and confidence level may be presented as a
family of curves, or as a three-dimensional surface. Some plots
of this type can be found in Section 7.4.
One practical concern is the time and cost associated with
processing of results. The results which are generated by the
basic solution, sensitivity analyses, and probabilistic computa-
tions often represent a substantial amount of numerical data. We
recommend using the "searching" options (those which search for a
maximum value within a specified set of nodes or elements) with
some care. since a great deal of calculation may be required.
E - 2
In principal, PRODIS can produce output on any graphical
device which is supported by DISSPLA. However, the program has
been tested only for a relatively small subset of these devices.
At present, PRODIS is equipped to generate graphical output on: