MSC 1990 World Users' Conference Proceedings

The conference proceedings for the 1990 World Users' Conference are now available on-line in Adobe Acrobat PDF format. (The Adobe Acrobat Reader software is available for free download from Adobe's web site at www.adobe.com).

OVERVIEW OF MSC'S PEOPLE, PRODUCTS AND PLANS (Acrobat 679K) #0190, 6 pgs.
Joseph F. Gloudeman, President and CEO -- MSC, Los Angeles, CA

ABSTRACT: An overview of MSC's people, products and plans.

NEW DEVELOPMENTS IN ADAPTIVE FINITE ELEMENT METHODS IN COMPUTATIONAL FLUID DYNAMICS (Acrobat 2.63MB) #0290, 26 pgs.
J. Tinsley Oden -- Texas Institute for Computational Mechanics, The University of Texas at Austin

ABSTRACT: The present paper provides a general overview of adaptive methods of CFD and describes recent progress toward developing general adaptive finite element codes for broad classes of flow simulation.

NON-LINEAR LOAD AND DEFORMATION ANALYSIS OF ELASTOMER PISTON SEALS USING MSC/NASTRAN (Acrobat 1.65MB) #0390, 19 pgs.
S. A. Bradley, Senior Engineer -- Allied Signal Corporation, Bendix Automotive Systems Group (North America)

ABSTRACT: An investigation of MSC/NASTRAN's capability to model the non-linear behavior of a rubber pumping cup seal. The study limits itself to include only non-linearities caused by large displacement and multiple contact effects. The seal's material behavior is assumed to be linear. This is in accordance with empirical test data and the strain magnitudes encountered. Good correlation with available empirical data is obtained. Results are used to indicate tendencies of the design which contribute to flawed or inadequate performance. A redesign is investigated and recommended.

NONLINEAR ANALYSIS OF POINT LOADS ON AN ELLIPTICAL BULKHEAD (Acrobat 1.93MB) #0490, 30 pgs.
Howard Yang and Robert Choy -- General Dynamics, Space Systems Division

ABSTRACT: Nonlinear stress analysis was performed for point loads on an internally pressurized elliptical bulkhead. This bulkhead is the aft structure of the Centaur tank, a liquid hydrogen and liquid oxygen propellant rocket. Two 22,000 lb point loads are created by the engines on the aft bulkhead. Based on linear MSC/NASTRAN analysis, the aft bulkhead which was previously designed for two 16,500 lb engine thrust loads, required strengthening for the new upgraded engines. Subsequently, MSC/NASTRAN Solution 66 was used to account for material and geometric nonlinearities. The results of this analysis, which agreed with full-scale test results, showed no required strengthening.

ADAPTIVE ARC-LENGTH METHODS IN MSC/NASTRAN (Acrobat 1.26MB) #0590, 22 pgs.
Sang H. Lee, Steve S. Hsieh and Timothy L. Bock -- MSC, Los Angeles, CA

ABSTRACT: The static nonlinear response of a structure can not be analyzed beyond the critical limit load by the conventional Newton's method. Although the post-buckling state is not usually allowed in the structural design, the prediction of such a response is of interest in some cases. In the design process, for instance, it may be desirable to trace the response of the snap-through or post-buckling behavior. The arc-length methods allow solutions in the unstable regime for such problems. It is noted that they are not efficient solution methods, but effective methods to render solutions in the unstable regime which is not necessarily a physically viable state. An arc-length method, known as Crisfield's method, was implemented in Version 66 of MSC/NASTRAN. However, some convergence difficulties have been identified and resolved in Version 66A. In addition, a number of options for arc-length methods and the adaptive bisection algorithm have been implemented in Version 67.

ANALYZING DEPLOYMENT OF SPACECRAFT APPENDAGES USING MSC/NASTRAN (Acrobat 1.11MB) #0690, 16 pgs.
F.W. Palmieri, President -- PCC, Inc.

ABSTRACT: MSC/NASTRAN's direct linear transient dynamic solution procedure can be utilized as a powerful, low cost tool for dynamic analysis. In particular, it provides the capability to analyze the effect of rigid body translations and rotations on spacecraft appendage structural dynamic response. The fact that "NOLIN" pseudo-nonlinear elements can be utilized to model gaps and Coulomb dampers without the necessity for regenerate the stiffness matrix, increases the applicability of the procedure while reducing computational cost by at least an order of magnitude. A simple example is presented to demonstrate the requisite input data, the solution process and the resulting output. Display of the deployment motion and the residual structural dynamic motion after impacint stops is possible using PATRAN-G version 2.4 and a simple FORTRAN program for data post-processing.

ON SHELL ELEMENTAL FORCE/MOMENT IN NONLINEAR ANALYSIS (Acrobat 766K) #0790, 12 pgs.
C.H.S. Chen -- The BF Goodrich Company

ABSTRACT: An error is detected in the elemental force and moment output of the shell element in nonlinear static analysis with antisymmetric or nonsymmetric laminate. The error appears to be due to the neglecting of the membrame-bending coupling terms. A quick simple method of remedy is presented. The linear analysis is free of this error.

RELIABLE FINITE ELEMENT MODELING FOR ENGINEERING DESIGN (Acrobat 2.23MB) #0890, 17 pgs.
Mark S. Shephard, Department of Civil Engineering -- Rensselaer Polytechnic Institute, Troy, New York

ABSTRACT: Over the past thirty years the finite element method has evolved to the point that it is now a commonly applied engineering analysis tool. However, its impact on the design process is typically limited to the later stages of design, or to design verification. This restricted application of finite element techniques is not due to the cost of the computation required, it is due to the cost of performing meaningful finite element analysis, particularly on a partially completed design. The reliable use of the finite element method by design engineers, who are not analysis specialists, requires capabilities past the finite element analysis program. This paper discusses the additional capabilities needed for the reliable application of finite element methods in engineering design.

"ANALYSIS MODELING" WITH FAM (Acrobat 1.3MB) #0990, 16 pgs.
Joseph L. Walsh, General Manager -- US Operations, FEGS Limited, Duluth, Georgia

ABSTRACT: FEGS Limited introduces a new approach of modeling for Finite Element Analysis applications based on an "intelligent" CAD interface and full Geometric Associativity of analysis data. This paper describes and compares the processes related to Finite Element Pre/Post Processing, Design Modeling, and "Analysis Modeling." This paper also discusses the CAD interface technology and Geometric Associativity provided by the FAM software available from FEGS Limited.

THE SURFACE OF ULTIMATE STRENGTH OF THIN-WALLED BEAMS (Acrobat 1.6MB) #1090, 20 pgs.
Zygmunt Borowiec -- Bombardier Inc. Canadair Division, Montreal, Quebec

ABSTRACT: A MSC/NASTRAN pre- and post-processor has been developed in CANADAIR to analyse thin-walled beam-sections in elastic and plastic range. The plastic module is presented in this paper. The graphic input and the output file of the surface of ultimate strength is created and displayed in CADAM. The coordinates of a point on the surface define three ultimate allowable loads P, M1, M2 and the position of the plastic neutral axis. Any loadcase may be represented by a 3-D point in the same space. The surface would shrink if shear stresses were taken into account but the corresponding point on the shrunken surface would define six allowable loads: P, M1, M2, Q1, Q2, T.

THE CTETRA(10) ELEMENT (Acrobat 1.29MB) #1190, 15 pgs.
D.L. Dewhirst and P.M. Grinsell -- Ford Motor Company

ABSTRACT: Several commercial solid modeling software packages are capable of automatically generating tetrahedron elements within arbitrary solid three-dimensional shapes. This "free meshing" of arbitrary shapes can significantly reduce project turnaround time when compared with conventional "mapped" meshing techniques. Figure 1a is a solid model of an automotive rocker arm which was modelled independently by two individuals. The process of creating this solid model and generating a CTETRA mesh, Fig 1b, required 3 man-days. An analogous CHEXA mesh created using conventional map-meshing techniques required 15 man-days. References typically describe ratios of approximately 2/1 in favor of free meshing. The motivation for using the CTETRA element lies in the fact that it can dramatically reduce turnaround time when used in conjunction with free meshing. However, there seems to be a consensus among finite element modelers that the CTETRA is either less accurate or less efficient than CHEX. Indeed, reference claims that a tenfold compute penalty exists for the use of the parabolic CTETRA compared with a linear CHEXA. The two following benchmark problems establish the relative accuracy of the CTETRA versus CHEXA elements and determine effects of element distortion, particularly aspect ratio.

A COMPARISON OF MSC/NASTRAN WITH SDRC/I-DEAS MODEL SOLUTION (Acrobat 1.55MB) #1290, 22pgs.
Tsay-Hsin G. Hu and Moses S. Huang -- Lockheed Engineering and Sciences Company, Houston, Texas

ABSTRACT: This study is an effort by NASA/JSC and LESC to compare the structural/stress analysis and problem solutions obtained using various solvers and post processing methods. The investigations include solutions for real-life problems and test cases. Solution results compared include NASTRAN output, post processing of NASTRAN output using SDRC/I-DEAS Finite element package and Model Solution in SDRC/I-DEAS. Significant differences in solution results are revealed. For consistent results care must be used in choosing the elements, the order of these elements as well as in the application of loads to these models. Also the post processing of the solution results may in certain cases by different. This is because different solvers report information in different ways which are not always directly comparable. This study is intended to find where the differences may occur and what steps may be taken to obtain comparable answers.

IMPLEMENTATION OF THE IRS DYNAMIC REDUCTION METHOD IN MSC/NASTRAN (Acrobat 976K) #1390, 11 pgs.
Christopher C. Flanigan, Director, Aerospace Projects -- SDRC Engineering Services Division, Inc., San Diego, CA

ABSTRACT: The Improved Reduced System dynamic reduction method is a recent extension of the Guyan reduction method. The IRS method includes mass effects in the development of the matrix reduction transformation matrix. This improvement can significantly increase the matrix reduction accuracy for certain types of models. The IRS method can be especially useful for test/analysis correlation to accurately reduce the FEM matrices to DOF corresponding to accelerometer locations. The IRS dynamic reduction method is implemented in MSC/NASTRAN using a rigid format to a normal modes analysis.

FINITE ELEMENT ANALYSIS OF THE DYNAMIC BEHAVIOUR OF AN ENGINE BLOCK AND COMPARISON WITH EXPERIMENTAL MODAL TEST RESULTS (Acrobat 1.65MB) #1490, 15 pgs.
Walter Ott, Hans-Jurgen Kaiser and Jurgen Meyer -- Ford-Werke AG, Koln, West Germany

ABSTRACT: The combination engine is the main source of noise and vibration in modern vehicles. Today the finite element method is the most important CAE tool to assist the engineer in the early stage of the development of an engine structure. The generation of the model is the most time consuming part of the FEM analysis. For reliable predictions a good compromise must be found between modelling effort and necessary accuracy of results. In this paper, two shell-solid element models and two solid element models of a typical 4-cylinder in-line engine block are compared. The natural modes for these four models were calculated up to 2kHz with MSC/NASTRAN SOL 3 and SOL 63 respectively. The analytical results have been compared with modal test data. The assessment of the calculated eigenvectors was based on computer animation of the mode shapes and the evaluation of the MAC values. The shell-solid models show a good agreement with the measured frequencies, whereas the solid models produce too high frequencies especially for the torsion modes.

ANALYSIS OF COUPLED NATURAL FREQUENCIES OF THIN-WALLED BEAMS WITH OPEN CROSS SECTIONS USING MSC/NASTRAN (Acrobat 1.04MB) #1590, 9 pgs.
M. Fouad Ahmad -- Civil Engineering and Mechanics, University of Wisconsin-Milwaukee
Carl W. Guile -- MSC, Los Angeles

ABSTRACT: Analysis of vibration modes and natural frequencies of thin-walled beams with open cross sections is performed for studying the validity of the cross section contour indeformability assumption. The thin-walled beam basic assumption and governing differential equations are presented first. Numerical examples have been solved using a finite element program for the general analysis of thin-walled beams as well as MSC/NASTRAN. Shell element models of these beams were developed to check the results of the thin-walled beam analysis. For the first few mode shapes, the differences between the results obtained using thin-walled beam models and those of the shell models are insignificant. However, this is not the case for higher modes where the values of natural frequencies obtained using thin-walled beam models differ from those obtained using shell models. The mode shapes obtained using shell models show that the beams at higher modes behave like true shells where cross sectional deformation is observed. The study recommends that for cases where higher vibration modes are important, three dimensional shell models should be developed.

INTEGRATED MODELING AND ANALYSIS OF A SPACE-TRUSS TEST ARTICLE (Acrobat 5.09MB) #1690, 27 pgs.
Alan E. Stockwell -- Lockheed Engineering and Sciences Company, Hampton, Virginia
Sharon E. Perez and Richard S. Pappa -- NASA, Langley Research Center, Hampton, Virginia

ABSTRACT: MSC/NASTRAN is being used in the Controls-Structures Interaction program at NASA Langley Research Center as a key analytical tool for structural analysis as well as the basis for control law development, closed-loop performance evaluation, and system safety checks. The objective of CSI research is to develop and validate the technology needed to design, verify and operate large space structures in which the structure and the control interact beneficially to meet the requirements of the 21st-century NASA and DoD missions. Guest investigators from academia and industry are performing dynamics and control experiments on a flight-like deployable space truss called Mini-Mast to determine the effectiveness of various active-vibration control laws. MSC/NASTRAN was used to calculate natural frequencies and mode shapes to describe the dynamics of the 20-meter-long lightweight Mini-Mast structure, predicting 153 modes below 100 Hz. Gravitational effects contribute significantly to structural stiffness and are accounted for through a two-phase solution in which the differential stiffness matrix is calculated and then used in the eigensolution. The frequencies of the first five modes calculated by MSC/NASTRAN are within five percent of the experimentally derived frequencies and analytical frequency response functions show good agreement with experimental FRF's. Reduced modal models are extracted for control-law design and evaluation of closed-loop system performance. Predicted actuator forces from controls simulations are then applied to the 153-mode model to predict member loads and stresses. These pre-test analyses reduce risks associated with the structural integrity of the test article, which is a major concern in closed-loop control experiments due to potential instabilities.

AN ITERATIVE METHOD FOR EIGENVECTOR DERIVATIVES (Acrobat 1.44MB) #1790, 20 pgs.
Gerald D. High -- MSC, Los Angeles, CA

ABSTRACT: An iterative method has been developed for calculating eigenvector derivatives. The method, called the Iterative Modal Method, combines the modal method of Fox and the iterative method of Rudisill and Chu. Only one decomposition is required, regardless of how many design variables or eigenvectors there are in the problem. IMM has been implemented in Version 66 of MSC/NASTRAN. A comparison is made with Nelson's method in two example problems. IMM is shown to be cheaper than Nelson's method on the larger of the two example problems.

AERODYNAMIC DERIVATIVES FACTORING SCHEME FOR THE MSC/NASTRAN DOUBLET LATTICE PROGRAM (Acrobat 1.71MB) #1890, 20 pgs.
Emil Suciu, John Glaser, and Rosemary Coll -- Boeing of Canada Ltd., de Havilland Division

ABSTRACT: A factoring scheme based on correcting a full set of subsonic trip aerodynamic derivatives calculated with the MSC/NASTRAN Doublet Lattice Method is presented. Verification of the procedure was performed by comparing the calculated flutter speed with existing experimental flutter speeds for an unswept wing with control surface at low speed. The generalized aerodynamic forces are calculated using factored and unfactored aerodynamic derivatives.

A TRANSIENT PLASTICITY STUDY AND LOW CYCLE FATIGUE ANALYSIS OF THE SPACE STATION FREEDOM PHOTOVOLTAIC SOLAR ARRAY BLANKET (Acrobat 1.58MB) #1990, 14 pgs.
Sasan C. Armand and Mei-Hwa Liao--National Aeronautics and Space Administration, Cleveland, OH
Ronald W. Morris--NASA Lewis Research Center Group, Cleveland, OH

ABSTRACT: The Space Station Freedom Photovoltaic solar array blanket assembly is comprised of several layers of materials having dissimilar elastic, thermal and mechanical properties. The operating temperature of the solar array, which ranges from -75 C to 60 C, along with the material incompatibility of the weld points of the assembly. The weld points are secondary structures in nature, merely serving as electrical junctions for gathering the current. The thermal mechanical loading of the blanket assembly operating in low earth orbit continually changes throughout each 90 min orbit, which raises the possibility of fatigue induced failure. This paper describes the MSC/NASTRAN transient plasticity analysis and the low cycle fatigue calculations for the photovoltaic solar array blanket assembly.

FLUTTER SIMULATION FOR BRIDGES (Acrobat 1.02MB) #2090, 13 pgs.
Hitoshi Nakamura and Ikuko Ogaki--Century Research Center Corporation

ABSTRACT: Since the overall stiffness in long span bridge like suspension bridges and cable-stayed bridges is relatively small compared with medium and small span bridges, it is especially important to evaluate such aerodynamically unstable phenomena as vortex shedding, flutter and so on. In the current design procedure for long span bridges, the evaluation of these phenomena is carried out in wind tunnel tests.

In the non-streamlined cross sections like bridge girders which are mainly subjected to wind force, the aerodynamic characteristics are considerable complex due to the separation of air flow or the generation of vortices. Therefore it is difficult to determine the unsteady aerodynamic force analytically. On the contrary, in streamlined sections like aircraft wings, the unsteady aerodynamic force can be relatively easily determined based on the potential theory. The methods for analyzing aircraft flutter based on aeroelastic theory have been applied in practical use.

Scanlan et al showed the formulation of the unsteady aerodynamic force acting on non-streamlined cross sections by using the aerodynamic coefficients determined by wind tunnel tests as a function of reduced frequency. Since the aerodynamic coefficient includes the effect of separation of air flow and vortex shedding, it can be used to properly evaluate the aerodynamic behavior of bluff bodies like the cross section of bridge girders.

In this work, the method of flutter analysis for actual bridges was investigated byapplying the experimental formulation to three-dimensional finite element analysis. The effectiveness was evaluated by performing the flutter analysis for the model of the Tacoma Narrows Bridge.

NONLINEAR TRANSIENT ANALYSIS OF A SHOCK ISOLATED MECHANICAL FUSE (Acrobat 1.66MB) #2190, 20 pgs.
Martin J. McNamee and Parviz Zavareh--United Technologies, USBI, Alabama

ABSTRACT: MSC/NASTRAN Solution 99 was used to simulate a mechanical fuse in a severe dynamic load environment. The analysis model demonstrates the application of compression and tension GAP elements with a nonlinear strain hardening shock isolator. Results of the analysis were applied to redesign the mechanical fuse. The mechanical fuse is critical to the proper function of a debris retention system on the Solid Rocket Boosters (SRB) Space Transportation System (STS) 28 incorporated the redesigned mechanical fuse which performed exactly as predicted by the analysis.

UNSTEADY PRESSURE DISTRIBUTION OF A WING-BODY CONFIGURATION: MSC/NASTRAN RESULTS COMPARED WITH WIND TUNNEL TESTS (Acrobat 940K) #2290, 18 pgs.
Arne Vollan, LOreto Alati, and Gino Ronti--OMEGA GmbH, Germany

ABSTRACT: Although the doublet-lattice method implemented in MSC/NASTRAN give good results for the aeroelastic analysis with SOL 75, there are configurations where test data is required because of the very important safety criterion of flutter stability for an airplane design.

In the present paper, a new method of unsteady pressure measurement was adopted. With this method, the surface of the model is covered by a special coating which changes the reflection properties of a laser beam with pressure. With this method, no internal pressure tubes are necessary, which facilitates the pressure measurements considerably.

The measured pressure distribution is compared to the results of MSC/NASTRAN using the doublet lattice method combined with slender body and interference elements.

MSC/PROBE: AN OVERVIEW (Acrobat 1.82MB) #2390, 15 pgs.
Barna A. Szabó, Albert P. Greensfelder, and Blanche Y. Greensfelder--Center for Computational Mechanics, Washington University

ABSTRACT: The technological base of MSC/PROBE, is discussed from the point of view reliability and quality assurance in the engineering decision-making process. Some aspects of the computer implementation are discussed and an example is presented.

USING MSC/XL TO PERFORM MSC/NASTRAN RESULTS PROCESSING (Acrobat 2.2MB) #2490, 20 pgs.
Larry Brown and Rohit Thukral--MSC.Software Corporation

ABSTRACT: MSC/XL Version 1 allowed the user to access analysis results data directly from the MSC/NASTRAN-generated graphics database and produce a variety of results processing displays. Version 2 extends the display capabilities and supports the use of external files. In addition to deformation, line contour, fringe contour and fill color displays, MSC/XL now generates arrow (vector) displays and XY plots. Results processing data can now be imported from external sources. Therefore, the user is not limited to MSC/NASTRAN results databases for the purpose of generating results processing displays.

TAKING ADVANTAGE OF ADVANCED USER INTERFACE TOOLS IN MSC/XL (Acrobat 2.66MB) #2590, 37 pgs.
Daniel Bryce, Natalie Jaco, and Mark Kenyon--MSC.Software Corporation

ABSTRACT: Version 2 of MSC/XL has many new capabilities. Future releases of MSC/XL will continue to expand both horizontally, to support other MSC analysis packages (as Version 2 supports MSC/EMAS), and vertically, to support each analysis package in more detail. However, no matter how many options are added to MSC/XL, users will always request additional capabilities in order to help them solve their problems more effectively. This paper describes a number of external mechanisms users can take advantage of in order to extend the capabilities of MSC/XL.

This paper uses several simple examples to focus on the introduction of ways in which users may tailor MSC/XL to suit their own needs. A detailed discussion of the underlying components is beyond the scope of this paper. For more information on these concepts refer to the MSC/XL User's Manual or consult your MSC regional support staff.

WHAT STRUCTURAL ENGINEERS SHOULD KNOW ABOUT MSC/EMAS (Acrobat 1.15MB) #2690, 15 pgs.
Bruce E. MacNeal--MSC.Software Corp.

ABSTRACT: Electromagnetic forces, as described by electric and magnetic fields, are important to design of many mechanical structures. CSC/AMASS is a new program which solves general electromagnetic field problems using the finite element method. The unique formulation of MSC/EMAS produces a matrix equation that is identical to the matrix equation solved in structural analysis, e.g., by MSC/NASTRAN. Because the equations are identical, the advanced solution methods and numerical algorithms used in structural analysis are also applicable to field analyses. Thus, finite element technology from MSC/NASTRAN was used extensively to develop MSC/EMAS. The MSC/EMAS formulation also establishes an analogy between field analysis and structures, which is useful to structural engineers.

REPORT FROM THE MSC/NASTRAN USERS' ADVISORY COMMITTEE (Acrobat 693K) #2790, 6 pgs.
A. Mera-- Boeing Computer Services

ABSTRACT: The Users' Advisory Committee (UAC) was founded by a group of experienced MSC/NASTRAN users during the 1989 MSC/NASTRAN World Conference The reasons leading up to its formation are discussed in the framework of developments and trends in Computer Aided Engineering. The objectives of the committee are stated and a few activities are described.

MSC/NASTRAN V66 OPTIMIZATION AS IMPLEMENTED IN A PRODUCTION ENVIRONMENT USING THE CAPABILITIES OF THE IBM 3090 VECTOR FACILITY (Acrobat 2.36MB) #2890, 22 pgs.
Scott R. Bird and Curtis J. Hoff--Automated Analysis Corporation

ABSTRACT: Structural optimization is a tool that has been used in academic circles and research facilities for a number of years. It is now being recognized as a tool for use in production environments by design engineers and analysts. MSC/NASTRAN has included an optimization capability in Version 66 that claims to put this tool in the hands of the engineer. This paper will evaluate the tool from three different perspectives. The first is the accuracy and features contained in the code itself. Secondly, from a larger point of view, the capability is reviewed as a tool in a production environment. The last perspective is purely from a hardware point of view. The implementation on the IBM 3090 Vector Facility is reviewed in terms of CPU, disk access, and parallel processing.

SHAPE SENSITIVITY ANALYSIS AND OPTIMIZATION USING MSC/NASTRAN (Acrobat 1.93MB) #2990, 24 pgs.
R.J. Wang--Ford Motor Company

ABSTRACT: Major software improvements have been made by MSC in the area of structural optimization. Grid point sensitivity is available in version 67 (alpha( This paper tests this capability by comparison with analytical solutions. Suggestions for further improvements are identified as well as recommendations for its use. The new capability has been integrated into a optimization system for component design. Several examples including a cantilever beam, a simplified engine connecting rod, a cantilever plate, and an upper suspension control arm are analyzed and optimized using this system.

OPTIMUM DESIGN OF SPACECRAFT STRUCTURES USING MSC/NASTRAN ( Acrobat 968K) #3090, 13 pgs.
R.A. Oldham and F.H. Chu--General Electric, New Jersey

ABSTRACT: In the design and analysis of the complicated structures that exist in the aerospace industry, there is, without exception, the necessity for structural optimization with respect to weight. In addition, there will almost always be constraints placed on this optimization such as stress restrictions, or, significant to the scope of this paper frequency requirements.

What follows is an account of the application of an optimization technique which employs MSC/NASTRAN to the stiffness design of a typical spacecraft built at General Electric/Astro Space Division (GE/ASD). It is hoped that the results of this technique, which was developed at GE/ASD, may eventually be compared to those generated by the application of MSC/NASTRAN's optimization capability for normal modes analysis when it becomes operative in Version 66A.

DYNAMIC OPTIMIZATION APPLIED TO ENGINE STRUCTURE (Acrobat 1.06MB) #3190, 16 pgs.
Masanori Ishikawa, Mitsuo Iwahara--Isuzu Motor Ltd.
Akio Nagamatsu--Tokyo Institute of Technology, Japan

ABSTRACT: With the remarkable advances in the field of electronic computers, it has become possible to carry out vibration analysis of large structure using the FEM. Shifting the natural frequency of a vibration mode that creates problems by some effective structural change is very important in engine designing.

Till now, experience of the designer and analyzer has played a major role in making the structural changes. However, with structures becoming more and more complex and an increased necessity for taking action against any arbitrary vibration modes that may create problems, a situation has arisen where experience can no longer deal with the problem effectively.

This time, the author employed the super element method for carrying out vibration analysis of the cylinder body, bearing caps, crank shaft and flywheel systems, and shifted a given natural frequency to the designated frequency range efficiently with the use of the dynamic optimization program developed by Isuzu Motors Ltd., using the the sensitivity of eigenvalues of the residual structure (cylinder body + bearing caps in this case).

OPTIMAL DESIGN OF COMPUTERED TOMOGRAPHY SCANNER STRUCTURE (Acrobat 974K) #3290, 11 pgs.
Y. Seki nad H. Arakawa--Toshiba Corporation
K. Koide and M. Iwata--Toshiba CAE Systems, Inc.

ABSTRACT: Recently, in order to save the cost and to reduce the space for medical electronic equipments, it has become important problem to minimize the weight of the structure, subject to various constraints in operation and manufacturing.

We have studied the feasibility weight reduction for the computed tomography (CT) scanner. In order to obtain the optimal values of many design parameters effectively, the sensitivity analysis capability in MSC/NASTRAN (V65A) and mathematical programming technique (SAMT) were combined. As a result, 25 percent weight reduction was attained.

This paper shows the effectiveness of the present approach for practical size problems.

NASTRAN BINARY OUTPUT PROCESSOR (Acrobat 2.11MB) #3390, 21 pgs.
Richard T. Wigginton, Andrew LeBlanc, and Vijay Vasani -- Electronic Data Systems Corporation

ABSTRACT: The NASTRAN Binary Output Processor (NBOP) was developed to provide an effective, flexible system to expedite postprocessing results from MSC/NASTRAN. The NBOP system consists of ISPF[1] functions, FORTRAN programs and DMAP ALTERS. The current version of NBOP is designed to run in a computing environment consisting of multiple large scale MVS/XA IBM mainframes using an IBM 3270 network for time sharing (TSO) access by engineers. NBOP provides a user a friendly automatic system to rebuild MSC/NASTRAN binary output datasets generated by OUTPUT2 and OUTPUT4 DMAP statements. The system provides facilities to read an existing binary output dataset and extract specific data blocks. NBOP then creates a new binary output dataset placing the extracted data blocks in the new dataset in any desired order and optionally changing the data block names.
The NBOP system was originally developed at the request of the users of ODYSSEY[2] an optimization system developed by GMR/EDS. ODYSSEY utilizes MSC/NASTRAN through a custom DMAP program that provides multiple load, boundary and analysis capabilities. This custom DMAP, because it combines the features of statics and normal mode solutions sequences, by necessity, uses data block names some consider non-standard. ODYSSEY users wanted to access a variety of programs to postprocess ODYSSEY/NASTRAN results stored in the NASTRAN data block input requirements. NBOP now provides the capability to create new binary output datasets acceptable as input to almost any appropriate MSC/NASTRAN postprocessing program. The NBOP system has proven very useful to the ODYSSEY user community and provides facilities often helpful to the general NASTRAN user. The NBOP design and implementation compromises documented in this paper have significance to anyone integrating MSC/NASTRAN into existing design systems.

INTERFACING CAD/CAM TO A GRAPHIC PRE- AND POSTPROCESSOR TO PERFORM A NONLINEAR BUCKLING ANALYSIS OF A COMPOSITE STRUCTURE (Acrobat 1.54MB) #3490, 25 pgs.
Mike Farley -- McDonnell Douglas Helicopter Company

ABSTRACT: The objective is to demonstrate in a VAX environment, the working relationships interfacing a 3-dimensional CAD/CAM model to a graphic pre and post processor. Upon developing an all composite preprocessed finite element model. The preprocessed fill will thus be translated into a readable MSC/NASTRAN bulk data file. At which point a subsequent nonlinear buckling analysis will be performed employing MSC/NASTRAN. The MSC/NSTRAN output, namely the eigenvectors, will be postprocessed to illustrate graphically the buckled mode shape by means of color contour fringe plots.

FINITE ELEMENT ANALYSIS ON THE MACINTOSH PERSONAL COMPUTER (Acrobat 533K) #3590, 6pgs.
Ken Ranger -- The MacNeal-Schwendler Corporation

ABSTRACT: MSC has upgraded it Macintosh version of the MSC/pal 2 finite element program to include solid elements, thermal stress analysis, improved performance, smaller databases, and a new user interface with more "MAC-like" menus and dialog boxes. Two versions are now available: a 500-mode version for the Macintosh Plus or SE and a 2000-node version for the Macintosh II. Two examples of analysis are presented to show the improved finite element tools that are available to the Macintosh personal computer owner.

LAPCAD, A MSC/NASTRAN PRE/POSTPROCESSOR ON THE MACINTOSH (Acrobat 1020K) #3690, 14 pgs.
Gert M. Lundgren -- LAPCAD Engineering

ABSTRACT: LapCADII Version 4 now offers both preprocessing as well as postprocessing on the Macintosh microcomputer. The preprocessing features make it possible to create a new model entirely from scratch, using low-level tools for generation of nodes and elements, as well as automated mesh generating macros. Models can also originate in AutoCAD and VersaCAD, and then be imported into LapCAD via its built-in DXF and TWGES translators. Once a model has been created, it can be altered using LapCAD's built-in editing tools. These tools allow instant conversion of single elements into an alternate mult-element pattern, selectable via a click of the mouse on a candidate icon. These editing tools exist for surface elements as well as for solid elements. In addition to the editing tools, LapCAD also provides numberous features for rotation, stretching, moving, scaling and alignment. The preprocessing features also include the application of boundary constraints, enforced displacements, and external forces, pressures, and temperatures. The created MSC/NASTRAN input file is complete, with an Executive Control, Case Control, and Bulk Data. Once the model has been run, the resulting output can be retrieved with LapCAD, and a deformed shape can be displayed. Internal stresses can also be superimposed directly on the model. LapCAD offers the user an interactive graphics interface, where each major group of tasks ha been collected in a single window. The user selects the desired task by clicking on an icon with its design being related to the task at hand. Once the user has selected a given task or icon in one of the tool windows, a dialog box is brought up, which contains user controlled variables relating to the selected task. After the user has verified or altered these variables, he proceeds with the completion of the task, which may amount o only the clicking on an implementation button, or lets the user complete the task with the mouse, such as connecting a HEXA element with the mouse.

INTERFACING CAD TO MSC PROGRAMS (Acrobat 1.01) MB #3790, 11 pgs.
Deepak Brar, Robert Lentz, Jenny Fang, and Ken Blakely -- MSC

ABSTRACT: This paper describes interfacing CAD programs to MSC's structural finite element programs MSC/NASTRAN, MSC/XL, MSC/pal2, and MSC/mod. Interfacing is done by reading a CAD geometry text file and writing an MSC/NASTRAN or MSC/pal 2 model file, and MSC/XL input file, or an MSC/mod element file. In particular, this paper shows how to interface AutoCAD geometry to MSC products via the microcomputer programs MSC/mod and CAD2MSC. Discussion is presented about which CAD entities can be successfully transmitted to MSC's programs. Several examples conclude the paper.

THE P-VERSION OF THE FINITE ELEMENT METHOD IN MSC/PROBE (Acrobat 1.24MB) #3890, 18 pgs.
John E. Schiermeier -- MSC, St. Louis, Missouri

ABSTRACT: Structural analysis can generally be divided into two areas: global analysis, where general behavior, such as load paths, or a structure is important; and local analysis, where the specific response, such as the detailed stress state, of a single component is important. The former is suited well to the h-version, which is sued in MSC/NASTRAN, while the latter can take advantage of the p-version, which is incorporated in MSC/PROBE.
In this paper, the h-version and the p-version will be described. The convergence rates of the different extension processes will be discussed, and the quality control procedures for the p-version in MSC/PROBE will be presented. Then a sample problem for detailed stress analysis in MSC/PROBE will be analyzed, and the quality control procedures will be applied to verify that the solution is good.

MSC/PROBE P-VERSION APPLICATION: PARAMETRIC FINITE ELEMENT ANALYSIS (Acrobat 1.18MB) #3990, 15 pgs.
Michael J. Heskitt -- MSC, St. Louis, Missouri

ABSTRACT: Many recurring problems encountered in industry, though not exactly the same, are similar in nature. These can often be grouped into classes where simple dimensional changes accurately define the problems within a class. The need to change parameter values for an analysis may arise in the design phase for sensitivity and optimization studies or in the production phase to analyze manufacturing discrepancies. Simple parametric analyses are also useful when several groups or individuals analyze similar parts. The parametric approach in FEA using MSC/PROBE is discussed and illustrated with a 3-D solid model of a bathtub fitting. Model construction techniques are included and simple, front-end preprocessors are proposed. Other parametric applications are also suggested.

PROBE UTILIZATION AT THE MCDONNELL AIRCRAFT COMPANY (Acrobat 1.43MB) #4090, 19 pgs.
Donald R. Ladwig -- McDonell Aircraft Company, St. Louis, Missouri

ABSTRACT: The PROBE finite element analysis program has been widely used at the McDonnell Aircraft Company since 1985. Element formulations within PROBE utilize shape functions of varying polynomial "p-level" order. By increasing the P-level accuracy of the solution can be improved without mesh refinement. A single analysis utilizing multiple P-levels can also provide unique checks for measuring solution accuracy. McDonnell Aircraft has used PROBE to obtain highly accurate solutions in structures containing stress concentrations and singularities. A sampling of various 2D and 3D McDonnell Aircraft structural models is presented. The problems encountered with using MSC/NASTRAN internal loads for input to a probe analysis are also discussed.

NONLINEAR MODELS FOR FASTENED STRUCTURAL CONNECTIONS BASED ON THE P-VERSION OF THE FINITE ELEMENT METHOD (Acrobat 2.04MB) #4190, 25 pgs.
J. Bortman and B.A. Szabo -- Washington University

ABSTRACT: Structural fatigue cracks are initiated and propagated in areas of high stress concentration, such as fastened joints. Often one or more fastener holes are sites of crack initiation. Hence, a large amount of effort is devoted in the aeronautical industry to the evaluation of the fatigue life of fastened joints. The first step in this analysis is determination of the load distribution between the fasteners and the stress field. Because of the complexity of the problem, certain modeling assumptions have to be made. In finite element models fasteners are usually idealized as one-dimensional springs or rigid links which connect nodal points between two elastic bodies. The elastic bodies are usually idealized as membranes or plates. While such practices are intuitively plausible, they are inconsistent with formulations based on the principle of virtual work, and are therefore conceptually incorrect. As a result, it will be demonstrated that the computed fastener forces and the stresses in the vicinity of the fasteners are entirely discretization-dependent. This gives the motivation to formulate a new model. An efficient and convenient technique is therefore suggested for modeling load transfer through fasteners, based on the p-version of the finite elements. The interaction between the fastener and the two-dimensional elastic body are modeled by normal displacements imposed on distributed springs. Friction is imposed as a weak condition (external tractions). Each fastener is represented by a nonlinear relation between the transferred force and the relative displacements. This relation may be obtained from a detailed three-dimensional analysis or from tests. After condensing out all linear degress of freedom, the nonlinear equations are solved.

STRESS AND MODAL ANALYSES OF AN AUTOMOTIVE MOUNTING BRACKET USING USC/PROBE (Acrobat 993K) #4290, 22 pgs.
Joseph T. Gierer and Anirudh K. Thakore -- MSC, St Louis, Missouri

ABSTRACT: The finite element analyses of mounting brackets or assemblies which are attached to sources of excitation, such as engines or motors, often involve both a detailed stress analysis and a modal analysis. Typically, separate models are used for the two analyses. This is because a stress analysis demands a relatively high level of discretization, or more degrees of freedom, to adequately capture the behavior of the local stress quantities. On the other hand, for a modal analysis, a simpler model with relatively fewer degrees of freedom is more appropriate because the output quantities, eigenvalues and eigenvectors, converge faster than stresses and also because eigensolvers are much more computer-intensive than linear equation solvers.
With MSC/PROBE-SOLID, which has detailed stress and modal analysis capabilities, the same FE mesh can be used for both the analyses. Typically, modal analysis results of acceptable accuracy are obtained at a low polynomial order (4 or 5) whereas stresses converge at a higher p-level (6 or 7). The included example, describing detailed stress and modal analyses of a typical automotive mounting bracket, illustrates the application of these MSC/PROBE features. Also described is the use of MSC/PROBE's solid elements along with "thin solid" shell elements and the quality assessment convergence checks, which are possible because MSC/PROBE incorporates the p-version formulation.

EVALUATION OF DIRECT MODEL MODIFICATION METHODS VIA MSC/NASTRAN DMAP PROCEDURES (Abstract 1.22MB) #4390, 16 pgs.
Wen-Ren Harn and Chi-Ching Hwang--Chung Shan Institute of Science and Technology, Taiwan

ABSTRACT: Various dynamic model modification methods have been developed in the last two decades. They are used to correct the analytical models using modal test data. However, both advantages and disadvantages exist for each method. In other words, none of them are suitable for all problems. In this paper, some of the direct modification methods are examined to show the accuracy and efficiency of the methods. Also, further analyses are made after the modification, through the use of MSC/NASTRAN DMAP, to demonstrate the effects of incomplete model modification on frequency response function and the transient responses. Accordingly, some discussions and application comments are given.

CALCULATING FINAL MESH SIZE BEFORE MESH COMPLETION (Abstract 1.41MB) #4490, 14 pgs.
Richard D. Lowrey -- Lockheed Astronautic Division, Sunnyvale, CA

ABSTRACT: With the advent of truly automatic 3D solid finite element meshing on solid models, very complex meshes can be created in minutes to a few hours. However, due to CPU costs and core limitations, the size of the mesh must of course be well controlled. As soon as a mesh is started, a report of the predicted final mesh should be made to the user. Each subsequent step of the mesh process should issue an updated report to ensure continued compliance with the desired mesh goal. Manual checks can be performed as well when non-automatic software is used.

A HYBRID L-M/BFGS METHOD FOR SYSTEM IDENTIFICATION (Abstract 487K) #4590, 9 pgs.
T.Ting -- University of Bridgeport, Bridgeport, Connecticut

ABSTRACT: Structural system identification is concerned with the correlation of analysis and test results. The most practical way to improve such a correlation is to directly adjust structural parameters and, in turn, modify the mathematical model of the system. This is normally done by minimizing an error measure associated with the differences (or residue vector) between the two sets of results with respect to design variables. The most common methods employed for this purpose can be classified as least-squares methods, where the sum of squares of differences is treated as an objective function to be minimized. If the posed least squares problem is unconstrained and the residue is linear with respect to design variables, the solution to the problem can be obtained by simply solving the associated normal equations. However, the residue vector for structural problems are usually nonlinear, while the problems can always be cast into unconstrained ones. The problems should be handled by nonlinear least squares method.
Nonlinear least squares methods require a number of iterations to converge. For a large-scale math model, the response analysis and the design sensitivity analysis are very costly, so an efficient method is sought to limit the number of function and sensitivity evaluations within one iteration and, at the same time, to possess a rapid convergence rate. Normally, a good approach is to avoid using exhaustive line search to find the corresponding design variable changes in an iteration. This is usually achieved by solving a well-conditioned normal equation in an iteration. Unfortunately, many investigators have indicated that the normal equations are frequently ill-conditioned and the resulting design changes are unacceptable or the process diverges. It has been demonstrated that this type of problem may be effectively overcome (in most cases) by adopting Lenvenberg-Marquardt type modification to the normal equations. Speeding up the convergence rate remains a challenge to the researchers in this field.
Although significant progress has been made, it is clear that additional work is required to improve the robustness and efficiency of the present computational methods. One common way to improve the convergence rate is to use Newton-like second order methods. But the associated cost in evaluating the second derivatives and their unstable conditions at remote points from the solution should prohibit us to consider such methods. Instead, some well-developed quasi-Newton's methods, which require only the first-order derivatives and provide better than linear convergence rate, are worthy of investigating. This paper is to propose a hybrid L-M/BFGS algorithm for this purpose and a numerical comparison between this method and the original L-M method is given in the sequel.

IMPROVED EIGENSOLUTION REANALYSIS PROCEDURES IN STRUCTURAL DYNAMICS
(Acrobat 1.34MB) #4690, 21 pgs.
B.P. Wang -- University of Texas at Arlington, Arlington, Texas
S.P. Caldwell and C.M. Smith -- McDonnell Aircraft Company, St Louis, Missouri

ABSTRACT: An improvement to eigensolution reanalysis procedures has been developed in this paper. This new method is based on the well known Bubnov-Galerkin procedure with a set of truncated normal modes and the associated residual static modes as global approximation functions. The formulation has been implemented in DMAP for MSC/NASTRAN version 65. Numerical results showed that the inclusion of residual static modes drastically improves the solution accuracy with a minor increase in computational cost over the assumed mode method. Additionally, an interactive scheme has been introduced to further improve the reanalysis results.

KINETIC ENERGY CALCULATION AS AN AID TO INSTRUMENTATION LOCATION IN MODAL TESTING (Acrobat 2.40MB) #4790, 28 pgs.
Grant R. Parker and Ted L. Rose -- MSC, Los Angeles, CA
John J. Brown -- McDonnell Douglas Electronic Systems Company, Huntington Beach, CA

ABSTRACT: MSC/NASTRAN provides the user with pre-coded programs, called rigid formats, that can be executed without having to learn the MSC/NASTRAN programming language. However, the program allows the user to write his own programs or alter existing MSC/NASTRAN rigid formats. This capability is called Direct Matrix Abstraction Programming.
A number of papers have been written to illustrate the application of the kinetic energy calculation for mode shape identification utilizing the DMAP capability. This calculation has been utilized by industry for over ten years and has proven to be an invaluable tool. This paper expands the application to aid in identifying and selecting structural instrumentation locations for modal testing or for monitoring input/response during general and qualification vibration testing. The modal identification and instrument location procedures can be coupled to identify the modes of interest and then to select monitoring locations based on these modes. The DMAP is easily altered into the standard Normal Modes, Response, or Superelement rigid formats in MSC/NASTRAN and can be added to the standard SUBDMAP capability in Version 66 of the program.
Background on the approach plus unique (but not widely utilized) advantages offered by the kinetic energy calculation are presented along with a simple example. The power of the approach is realized in the effective display of results and the ability to quickly identify and select locations for test monitoring. This information, derived from the finite element analysis model of the structure, provides the test/analysis engineer with valuable, time-saving information.

COMPONENT MODE SYNTEHSIS OF SPACE STATION FREEDOM USING MSC/NASTRAN SUPERELEMENT ARCHITECTURE (Acrobat 1.75MB) #4890, 19 pgs.
Steve Del Basso and Sudeep Singh -- Grumman Space Station Engineering and Integration Center
Alan J. Lindenmoyer -- NASA Space Station Freedom Program Office

ABSTRACT: The generation of a Space Station Freedom system structural model for on-orbit loads analysis presents an integration challenge. With four prime contractors and three international partners involved, the methodology employed needs to be able to handle a variety of component model representation, and still satisfy all the traditional analysis demands. An approach, which minimizes software development and offers considerable flexibility, relies on the superelement solution sequence of MSC/NASTRAN. Component models can be transferred as physical mass and stiffness matrices, NASTRAN bulk data, or in a generalized, reduced state such as Craig-Bampton transformed mass and stiffness matrices. These are then coupled within MSC/NASTRAN from which the synthesized system modes are obtained. Two such sets of system modes were developed in this study. In one case, component models were derived from an existing system finite element model of the Permanent Manned Capability configuration. This provided a test bed for validation. The other system modes were also for the Permanent Manned Capability configuration. In both instances, the approach taken proved effective.

SPECIAL APPLICATIONS OF GLOBAL-LOCAL ANALYSIS (Acrobat 672K) #4990, 5 pgs.
Dara Sabahi -- Jet Propulsion Laboratory, Pasadena, CA
Ted Rose -- MSC, Los Angeles, CA

ABSTRACT: Local regions of a structure may require more detailed analysis due to higher stress levels or gradients than can be accurately modelled by the overall finite element model (called the global superelement model in this paper). These local regions, as long as they are contained completely within a superelement, can be analyzed independently with a finer mesh by using "Global-Local" analysis. Global-Local techniques can be used to apply the forces and displacements resulting from the global model to the boundaries of the detailed local model. These Global-Local methods have been presented in an MSC/NASTRAN Application Note entitled "Applications of Superelements in Global-Local Analysis," dated September, 1887. However, certain types of detailed local analysis cannot be readily performed using standard Global-Local techniques. Boundaries of two or more superelements, local analysis that involved multiple superelements, superelements with partitioned databases, and solving for local buckling are typical examples. This paper outlines Global-Local methods which can be utilized in performing these analysis.

MSC/NASTRAN SUPERELEMENT ANALYSIS OF THE NASA/AMES PRESSURIZED WIND TUNNEL (Acrobat 3.57MB) #5090, 17 pgs.
Dara Sabahi -- Jet Propulsion Laboratory, Pasadena, California
Ted Rose -- MSC, Los Angeles, California

ABSTRACT: Under contract to NASA, Norman Engineering has performed a complete re-design of the 12 Foor Pressurized Wind Tunnel. This effort was done to fulfill the requirements of the ASME Pressure Vessel code, which required analyses for loadings varying from vacuum to being filled with approximately 55,000,000 pounds of water and pressurized. This effort required the creation of several finite element models and utilization of advanced features in MSC/NASTRAN and CADAM, including superelements and global-local analysis. This paper summarizes the effort involved with special attention to obstacles encountered and how they were overcome.

CREATION OF AND USE OF "CRAIG-BAMPTON" MODELS USING MSC/NASTRAN (Acrobat 2.59MB) #5190, 42 pgs.
Ted L. Rose -- MSC, Los Angeles, CA

ABSTRACT: As structures become larger and more complex, many programs are requiring contactors to deliver "Craig-Bampton" models of hardware instead of complete finite element models. This is identical to the default method used when performing component modal synthesis in MSC/NASTRAN using superelements. This method of component modal synthesis consists of reducing a finite element model into a set of generalized mass and stiffness matrices which can be connected to physical grid points. The generalized coordinates used to create these matrices consist of "constraint" modes and fixed-interface elastic modes of the structure. This paper presents a simple method to create the generalized mass and stiffness matrices, and shows how to use them in further analysis.

EVALUATION OF THE NEW DATABASE IN VERSION 66 (Acrobat 1.52MB) #5290, 21 pgs.
Ingo Raasch -- BMW AG, Munich, West Germany

ABSTRACT: At BMW most MSC/NASTRAN runs are single-shot runs, which save a minimum amount of data on the Data Recovery Database. Its size is about 3-5% of a restart tape or a normal superelement database. Postprocessing is initiated by an inhouse program, which reads this database and sets up a new Nastran job, which in turn creates all the necessary datablocks the user needs for further processing. Because the interfacing of a finite element program with a postprocessor is error prone, it was possible to reduce the number of unsuccessful runs considerably. This user friendly environment needs to be implemented in Version 66. Documentation of the new database does not exist, similarly to the previous version, therefore the structure of the database was extracted from many test runs and dumps. Description of the previous version of the database was met by considerable interest by the MSC/NASTRAN user community, therefore it is worthwhile to document our experience with the new database.

FINITE ELEMENT ANALYSIS OF POROSITY ON MATERIAL PROPERTIES USING MSC/NASTRAN (Acrobat 1.36MB) #5390, 17 pgs.
Paul R. Woodmansee and Howard D. Gans -- Air Force Institute of Technology

ABSTRACT: In this paper, a finite element model using MSC/NASTRAN is developed that can numerically determine the material's modulus of elasticity using the limited information from one material sample. The model is three dimensional, and simulates pores by placing small elements that are non-load bearing into the structure. These voids are randomly and unevenly distributed (using a Poisson distribution) to better simulate the response of a real porous material to a load. The porous model deformation can be used directly to calculate the porous modulus of elasticity. The model is shown to be accurate. The same finite element model is used to demonstrate the effects on material behavior of changing the pore size and distribution. From this information a trend was noticed and an empirical equation developed that predicts a material's elastic modulus based on porosity and pore size.

DESIGN OF HIGH PRECISION SANDWICH STRUCTURES USING ANALYTICAL AND FINITE ELEMENT MODELS (Acrobat 1.25MB) #5490, 16 pgs.
W. Elspass -- Swiss Federal Institute of Technology, Zurich, Switzerland

ABSTRACT: As honeycomb is regarded as a homogeneous orthotropic material, it can be modelled with the knowledge of all 9 elastic constants by use of solid elements in finite element analysis. This procedure simplifies the analysis of sandwich structures thus improving the accuracy concerning the implane stiffness of the core material. To quantify the influence of different material combination for core and facesheets, parametric studies are performed with a special finite element reference structure. For selected thermal load cases the deformation of a sandwich with parabolic shape are analyzed by calculating the contour error.

COMPARISON OF HONEYCOMB PLATE MODELS USING MSC/NASTRAN (Acrobat 1.86MB) #5590, 33 pgs.
Paul F. Martin -- MIT, Lincoln Laboratory, Lexington, Massachusetts

ABSTRACT: An evaluation of a two-inch-thick, one-way honeycomb optical bench is performed with MSC/NASTRAN. Geometry of the bench is such that its behavior may be predicted by flat-plate analysis. Aspect ratios of the bench are set such that shear deformation is not negligible. The core of the bench is a honeycomb with a 2-inch-square mesh. This core is connected to top and bottom cover plates. The evaluation of the bench is performed with five different models. In the first, the cover plates and all ribs are represented with CQUAD elements. The second and third use a single layer of CQUAD elements whose properties are given with PSHELL and PCOMP. The fourth uses a single layer of CQUAD elements combined with CBEAMS for the ribs. The fifth uses a single set of CBEAMS which lie along the longitudinal axis. Static and dynamic responses are obtained.

FINITE ELEMENTS AND STRUCTURAL MECHANICS IN ELECTRONIC PACKAGING: PRESENT AND FUTURE (Acrobat 817K) #5690, 12 pgs.
Arturo O. Cifuentes PhD -- IBM-East Fishkill, Hopewell Junction, New York

ABSTRACT: This paper is concerned with the application of the finite element method to solve structural problems in electronic packaging. The paper presents an overview of some of the most common stress-related problems encountered in packaging, as well as their treatment using finite elements. A discussion regarding the use of commercial finite element codes in this environment is also presented.

CURVED SHELL ANALYSIS SOME PROBLEMS AND SOLUTIONS (Acrobat 630K) #5790, 9 pgs.
Gernot W. Haggenmacher, Consultant -- MSC, Los Angeles, California

ABSTRACT: This study of the behavior of curved shell analysis will concentrate on aspects of equilibrium. In fact the balance between membrane forces and bending forces plays a dominant role in curved shells. Simple curved beam models will serve to illustrate some principal problems, discussing element design and correct loads in relation to modelling strategy. Cylindrical shells are easily related to the concepts of curved beam modelling. For doubly curved shells the discussions apply in principle, but the formal treatment is considerably more involved, and will not be treated in this paper.

A SURVEY OF MSC/PISCES APPLICATIONS (Acrobat 3.16MB) #5890, 49 pgs.
A.M.A. van der Heijden and P.H.L. Groenenboom -- MSC B.V., Gouda, The Netherlands

ABSTRACT: Applications of advanced computer programs for transient analysis have for a long time mainly been restricted to defense applications and safety studies for nuclear plants. In recent years, however, applications in other industries especially, the automotive industry, have become more and more important. In this paper a selection of problems from various industries is presented which have been solved with the family of MSC/PISCES codes. The purpose of this paper is to show that these general purpose codes enable the engineer and scientist to solve a broad class of problems with the same tools.

NON-LINEAR PROPAGATION OF ELASTO-PLASTIC WAVES IN RODS (Acrobat 928K) #5990, 10 pgs.
R. Rittel -- Rheinmettal GmbH, Dusseldorf, West Germany

ABSTRACT: The investigation of elasto-plastic material behaviour of rigid bodies at extremely short loadings was not to measure exactly for a long time. For various applications e.g. simulation of car-crash, impact of meteorites, terminal ballistic problems etc. the knowledge of the dynamic behaviour of materials is necessary. G. KUSCHER from the Ernst-Mach-Institute, W-Germany, carried out the first time experiments that allow the examination of the elasto-plastic theory of wave propagation in rods, that were carried out with the MSC-PISCES-Code, and the comparison to experimental results which lead to a deeper insight into and, in connection with computational graphics, to a better visualization in material behaviour.

A THREE-DIMENSIONAL FINITE ELEMENT ANALYSIS MODEL OF AN ARTIFICIAL HIP AND BONE (Acrobat 1.13MB) #6090, 6 pgs.
J.A. Ecker and J.A. St. Ville -- The Johns Hopkins University, Laurel, MD

ABSTRACT: A three-dimensional finite element analysis model of an artificial hip implant in a cadaver femur has been developed to help further our knowledge of the stress distributions across the bone-to-implant interface in hopes of understanding the reasons for artificial hip failure (loosening) after only approximately seven to ten years of implantation. Actual measurements of the cortical and the cancellous bone were input to the model based upon computer axial tomography data. From these two-dimensional slices a three-dimension model was created representing the various layers of the bone and the implant. The cortical bone was modeled as a transversely isotropic material. Cancellous bone is known to be nonlinear and material properties are dependent upon location. This bone was modeled as totally anisotropic. Material properties for both bone types were taken from values previously reported in the literature. Different isotropic metals were modeled for the implant material. A polysulfone implant was also modeled. A standard 32mm acetabular ball was modeled at the end of the implant stem. The complete finite element model consisted of over 3100 nodal points and over 3000 isoparametric solid elements. Both hexahedron elements and pentahedron elements were used. Loading was applied over the surface of the acetabular ball to simulate pressures measured in vivo and reported previously in the literature. Abductor muscle loads were also included in the finite element analyses. This model enabled examination of the bone-to-implant interface in an analytical method. Variations in loadings, implant fixation, implant stem length and other considerations may be studied for sensitivity. We feel these types of studies may be instrumental in helping us understand the reasons for early failure of total hip arthroplasties.

SOME USER EXPERIENCES WITH MSC/DYNA (Acrobat 2.90MB) #6190, 28 pgs.
W.J.T. Daniel -- University of Queensland
Andrew Currie -- Compumod Pty Ltd, Sydney, Australia

ABSTRACT: The background to the explicit analysis techniques in MSC/DYNA is reviewed, and the inherent advantages and limitations of this approach, compared to the standard implicit MSC/NASTRAN formulation are discussed. Distinctive features of MSC/DYNA are reviewed. Examples of MSC/DYNA solutions to plate impact and penetration problems, rolling mill simulation, simulation of explosively formed particles, and modelling golf club and ball are presented. These illustrate the graphics that can be obtained, and make use of the interfaces to PATRAN and MSC/XL and make use of input files from DYNA3D models. The software available to link MSC/DYNA to PATRAN and to MSC/XL is discussed.

A USER'S PERSPECTIVE ON NONLINEAR TRANSIENT DYNAMIC ANALYSIS WITH MSC/DYNA
(Acrobat 3.03MB) #6290, 30 pgs.
Philip B. Bogert and Avivi Lev -- NKF Engineering Inc, Arlington, VA

ABSTRACT: NKF Engineering has been a user of MSC/DYNA since it was first released in early 1989. NKF engineers have become involved in a variety of analyses well suited to MSC/DYNA in response to the changing needs of its Department of Defense and commercial customers. Some of our DYNA experience has been developmental in nature, i.e., the running of fictitious sample problems in order to become familiar with the code. Recently, we have started using MSC/DYNA on "real" engineering problems. Our investigations have been much aided by the fact that we have been selected by MSC as a beta test site for their DYNA-XL converter so that we have been able to pre- and post-process our DYNA problems with MSC/XL. The purpose of this paper is to provide an overview of our DYNA and DYNA/XL experience by looking at some representative problems we have analyzed, some fictitious and some real.

SUPERELEMENT MANAGEMENT IN A NETWORK ENVIRONMENT (Acrobat 1.00MB) #6390, 11 pgs.
Keith H. Redner, President -- Danford Associates, San Pedro, California
Carl W. Hennrich, Consultant

ABSTRACT: Since the computing environment of hardware, software and model are increasingly complex and sophisticated, the typical engineer finds himself needing the help of a manger to control the use of these resources and keep track of what happens as the result of his activities in obtaining solutions to his engineering problems. The concept behind the Danford Project Application Manager is that the engineer should not have to deal with details of computer resource allocation, record keeping, job status inquiry, and a host of other computing environment details, but should be left to concentrate on the engineering aspects of the analysis activity.

SPARCE MATRIX METHODS IN MSC/NASTRAN (Acrobat 988K) #6490, 15 pgs.
Shawn Shamsian and Louis Komzsik -- MSC, Los Angeles, CA

ABSTRACT: In the past years we have been investigating the feasibility of new sparse matrix methods in MSC/NASTRAN. After careful studies we launched a two year, two phase development plan. The first phase, which recently completed for version 67, entails incorporating sparse matrix algorithms into the MPYAD, FBS, and DECOMP modules. This paper describes these new algorithms and presents the results obtained on experimental version 67 systems.