Nastran 2010 Doc Release

MSC Nastran 2010

Release Guide

Worldwide Webwww.mscsoftware.com

DisclaimerMSC.Software Corporation reserves the right to make changes in specifications and other information contained in this document without prior notice.The concepts, methods, and examples presented in this text are for illustrative and educational purposes only, and are not intended to be exhaustive or to apply to any particular engineering problem or design. MSC.Software Corporation assumes no liability or responsibility to any person or company for direct or indirect damages resulting from the use of any information contained herein.User Documentation: Copyright 2010 MSC.Software Corporation. Printed in U.S.A. All Rights Reserved.This notice shall be marked on any reproduction of this documentation, in whole or in part. Any reproduction or distribution of this document, in whole or in part, without the prior written consent of MSC.Software Corporation is prohibited.This software may contain certain third-party software that is protected by copyright and licensed from MSC.Software suppliers. PCGLSS 6.0, Copyright © 1992-2005, Computational Applications and System Integration Inc. All rights reserved. PCGLSS 6.0 is licensed from Computational Applications and System Integration Inc. METIS is copyrighted by the regents of the University of Minnesota. A copy of the METIS product documentation is included with this installation. Please see "A Fast and High Quality Multilevel Scheme for Partitioning Irregular Graphs". George Karypis and Vipin Kumar. SIAM Journal on Scientific Computing, Vol. 20, No. 1, pp. 359-392, 1999. MSC, MD, Dytran, Marc, MSC Nastran, MD Nastran, Patran, MD Patran, the MSC.Software corporate logo, OpenFSI and Simulating Reality are trademarks or registered trademarks of the MSC.Software Corporation in the United States and/or other countries.

NASTRAN is a registered trademark of NASA. PAMCRASH is a trademark or registered trademark of ESI Group. SAMCEF is a trademark or registered trademark of Samtech SA.

Revision 0. October 5, 2010NA:V2010:Z:Z:Z:DC-REL-PDF

CorporateMSC.Software Corporation2 MacArthur PlaceSanta Ana, CA 92707Telephone: (800) 345-2078FAX: (714) 784-4056

EuropeMSC.Software GmbHAm Moosfeld 1381829 MunichGERMANYTelephone: (49) (89) 43 19 87 0Fax: (49) (89) 43 61 71 6

Asia PacificMSC.Software Japan Ltd.Shinjuku First West 8F23-7 Nishi Shinjuku1-Chome, Shinjuku-Ku Tokyo 160-0023, JAPANTelephone: 0120-924-832 (toll free, Japan only)Mobile phone: 03-6911-1222Fax: (81) (3)-6911-1201

C o n t e n t sMSC Nastran 2010 Release Guide

MSC Nastran 2010 Release Guide

Table of Contents

Preface to the MSC Nastran 2010 Release Guide viii

List of Books ix

Technical Support x

Internet Resources xi

1 Overview of MSC Nastran 2010

Overview 2

2 Implicit Nonlinear (SOL 600)

SOL 600 Enhancements 6

3 Numerical Methods and High Performance Computing

MPI Selection 12

New Solver Available for Complex Eigenvalue Analysis 13

4 Dynamics (Noise and Vibration)

Equivalent Radiated Power (ERP) 18

Frequency Dependent Rigid Absorber Properties 28

Dynamics - Monitor Points in Dynamic Solution Sequences 35

5 Optimization

Part Superelement Optimization Enhancements 44

Optimization - Invariant DRESP3 Gradients 53


iv

Design of Monitor Points 57

Parallel Sensitivities 64

DTABLE Enhancement for Dynamic Analysis 67

Constants with DTABLE2 70

New Optimizer - IPOPT 74

Topology and Topometry Enhancements 82

Build External Servers Using the SCons Tool 91

Deactivation of Original Design Sensitivity (DSA) 94

6 Aeroelasticity

Input of Pressures on an Aerodynamic Mesh 96

Aeroelasticity - Output of Trimmed Loads 101

CSV Output of Trim Results 106

SUBCOM/SUBSEQ with Static Aeroelasticity 109

Upper Hessenberg Complex Eigenanalysis No Longer Supported for Flutter Analysis 111

7 Elements

Enhancements to Connector Elements 114

Offsets for Beams and Shells 123

8 Miscellaneous

Enhanced MONSUM 134

PARAM,NONCUP Usage Extended to SOL 111 136

Application Regions 137

New Input File Reader - IFPSTAR 138

Brake Squeal Analysis 141

Results and Output Changes 142

vContents

MSC Nastran Error List 144

A Connectors

Connectors Output 146


vi

MSC Nastran Release Guide Preface

Preface

Preface to the MSC Nastran 2010 Release Guide

List of Books

Technical Support

Internet Resources

MSC Nastran 2010 Release GuidePreface to the MSC Nastran 2010 Release Guide

viii

Preface to the MSC Nastran 2010 Release GuideThis Release Guide contains descriptions for the MSC Nastran 2010 version, and supersedes the MSC Nastran 2008 Release Guide.

ixPreface

List of BooksListed below are some of the Nastran documents. You may find any of these documents from MSC.Software at www.simcompanion.mscsoftware.com.

Both MD Nastran and MSC Nastran books are listed for your convenience.

Installation and Release Guides

• Installation and Operations Guide

• Release Guide

Reference Books

• Quick Reference Guide

• DMAP Programmer’s Guide

• Reference Manual

User’s Guides

• Getting Started

• Linear Static Analysis

• Dynamic Analysis

• MD Demonstration Problems

• Thermal Analysis

• Superelements

• Design Sensitivity and Optimization

• Implicit Nonlinear (SOL 600)

• Explicit Nonlinear (SOL 700)

• Aeroelastic Analysis

• User Defined Services

• EFEA User’s Guide

• EFEA Tutorial

• EBEA User’s Guide

MSC Nastran 2010 Release GuideTechnical Support

x

Technical SupportFor technical support phone numbers and contact information, please visit: http://www.mscsoftware.com/Contents/Services/Technical-Support/Contact-Technical-Support.aspx

Support Center (http://simcompanion.mscsoftware.com)

Support Online. The Support Center provides technical articles, frequently asked questions and documentation from a single location.

http://simcompanion.mscsoftware.com

http://www.mscsoftware.com/Contents/Services/Technical-Support/Contact-Technical-Support.aspx

xiPreface

Internet ResourcesMSC.Software (www.mscsoftware.com)

MSC.Software corporate site with information on the latest events, products and services for the CAD/CAE/CAM marketplace.

www.mscsoftware.com

MSC Nastran 2010 Release GuideInternet Resources

xii

Chapter 1: Overview of MSC Nastran 2010 MSC Nastran Release Guide

1 Overview of MSC Nastran 2010

Overview

MSC Nastran 2010 Release GuideOverview

2

OverviewMSC.Software is pleased to introduce you to the exciting new technologies in MSC Nastran 2010, the premier and trusted CAE solution for aerospace, automotive, defense, and manufacturing industries worldwide. This release includes a wide range of new features and enhancements to our nonlinear implicit (SOL 600) solution and our linear solutions in the areas of durability and NVH, Optimization, and Aeroelasticity.

Implicit Nonlinear (SOL 600)• Improved Computational Efficiency Using New Parallel Solvers

• Improved friction definition and rigid surface behavior

• Improved super element - DMIG support

• Improved dynamic integration scheme

• Automatic conversion of CHEXA, CPENTA to Solid Shell

• Support for RSSCON and RSPLINE

• User subroutines for contact and materials

• Continuous-stress contact enhancement

• Arbitrary cross section and numerically integrated beams

More information on Implicit Nonlinear (SOL 600) can be found in Implicit Nonlinear (SOL 600) (Ch. 2).

Numerical Methods and High Performance Computing (Performance)

• MPI Selection (Ch. 3)

• New Solver Available for Complex Eigenvalue Analysis (Ch. 3)

More information can be found in Numerical Methods and High Performance Computing (Ch. 3).

Noise, Vibration and Dynamics• Equivalent Radiated Power (ERP)

• Dynamics - Monitor Points in Dynamic Solution Sequences

More information can be found in Dynamics (Noise and Vibration) (Ch. 4).

Optimization• PART Superelement Optimization

• Integer Input for DTABLE

• Optimization - Invariant DRESP3 Gradients

3CHAPTER 1Overview of MSC Nastran 2010

• Design of Monitor Points

• Miscellaneous - Enhanced MONSUM

• Parallel Sensitivities

More information on these optimization enhancements can be found in Optimization (Ch. 5).

Aeroelastic Enhancements• Aeroelasticity - Output of Trimmed Loads

• CSV Output of Trim Results

Elements• Connector for Durability

• Offsets Support for Buckling

More information can be found in Elements (Ch. 7).

Future Platform SupportMSC Nastran will no longer be delivered on the SGI IRIX or IRIX64 platforms after this release.

The Linux 32 bit platform will be discontinued starting in the year 2012.

MSC Nastran 2010 Release GuideOverview

4

Chapter 2: Implicit Nonlinear (SOL 600) MSC Nastran 2010 Release Guide

2 Implicit Nonlinear (SOL 600)

SOL 600 Enhancements

MSC Nastran 2010 Release GuideSOL 600 Enhancements

6

SOL 600 Enhancements

IntroductionMSC Nastran SOL 600 is a Solution Sequence that allows MSC.Nastran users to use the familiar MSC Nastran input file format to execute complex nonlinear problems via a translation to MSC.Marc. Based upon customer input since the MSC Nastran 2008 release, a series of enhancements have been made to increase the functionality of the product.

BenefitsThis allows a broader set of Nonlinear Problems to be solved within the Nastran framework. These enhancements are outlined below:

• Added plane stress to add Marc’s capabilities – use the MRALIAS PARAM or the ALIASM Bulk Data option

• Implemented PARAM,TSTATIC for dynamic analyses

• Added friction vs other variable described by tables capability – see BCBODY and /or BCTABLE

• Better control of 2-D Rigid bodies Rx, Ry to for 2D rigid contact orientation

• Provide support for contact user subroutines ufric, ufricbbc, uhtcon, digeom, sepstr, spfor – see BCBODY and BCONUDS

• Added rigid contact rigid surface temperature and sink temperature for heat transfer simulations – see BCBODY and BCTABLE.

• Added new options for BCBOX, such that the user can specify if the complete element needs to be in the box to be put in body, or only a single node.

• Allow contact variables such as velocity to be defined using tables.

• Added conversion of CHEXA and CPENTA to solid shell (customer request)

• Add delamination capabilities – see MDELAM Bulk Data entry

• Improved VCCT crack propagation so the path along the crack tip in 3-d can be easier to describe - see VCCT Bulk Data Option – Not supported by GUI

• Add material mixtures capability – see MIXTURE Bulk Data entry

• SOL 600 now supports RSSCON

• SOL 600 now supports RSPLINE

• Added new type of RBE3 to uniformly distribute applied loads – see the RBE3U Bulk Data entry. Note that this does not distribute stiffness.

• Added generalized alpha method for dynamics, this includes support for the Hilbert-Hughes-Taylor (HHT) procedure. Use the HMOUBOLT Param and set dynamic operator to 7.

7CHAPTER 2Implicit Nonlinear (SOL 600)

• Added PARAM,MARALPHA to choose between secant and instantaneous CTE’s vs temperature with options to obtain high accuracy

• Added general capability for user subroutines. – See the MATUDS and BCONUDS Bulk Data entries.

• Added stiffness matrices in output4 and Boeing Hartwell formats. This is controlled using the MSTIFOT option.

• Added improved superelement capability (S6SUPER)

• Exposed simplified nonlinear elasticity models available in Marc NLELAST option. See MATNLE Bulk Data options

• Exposed user defined hypoelastic material model. See MATHYP

• Allow user subroutines to be used with rate dependent creep material models. See the MATVP Bulk Data entry.

• Allow nonlinear material behavior over a solid section beam of arbitrary cross section. This beam will be numerically integrated. The definition of the beam section and the integration procedure is entered through the PBMARB6 Bulk Data entry.

• Added options to improve complex loading

• Added option to keep or remove mpi services on Windows systems for a run using parallel processing with DDM

• Added an option to convert MAT1 entries with bad Poisson ratio’s to MATORT

New SOL 600 Parameters and Bulk Data EntriesTable 2-1 contains new Parameters and Bulk Data entries for SOL 600 in MSC Nastran 2010. More details can be found in the MSC Nastran Quick Reference Guide.

Table 2-1 New Parameters and Bulk Data Entries for SOL 600

New for MSC Nastran 2010 (SOL 600)

Description

Parameters

MARAUTO Determines whether NLAUTO entries for SOL 600,129 will override the default or not

MARCTOTD Determines whether for SOL 600 dynamic analyses will use full table association.

MARCTOTL Determines whether total or incremental loads are used in a SOL 600 static nonlinear analysis.

MARCTOTT Determines whether total loads, including pressures, gravity, spcd, etc., with associated tables are used in a SOL 600 static or dynamic analysis.


8

MARCL001 Determines whether Marc’s POINT LOAD (without tables) 2nd datablock, 3rd field will be honored or not. If this value is set to 1 multiple loads at the same dof in the same subcase will usually be summed.

MARFATAL Determines whether non-existent grid id’s for BCBODY entries will cause fatal errors or not.

MARTET10 Controls how to treat badly shaped 10-node tetra elements in SOL 600 if renumbering the element does not correct problems.

MARTETIN Controls whether additional information messages are output to the .f06 file or not when param,martet10 is set to a positive value.

MCORDUPD Determines the coordinates will be updated if one of the CONTINUE options is specified on the SOL 600 Executive Control statement.

MINSOUTT Determines elements that deform so much that they go inside-out in an analysis will be deactivated.

MMBOLTUS Controls how the top and bottom nodes are placed in the Marc “tying 69” input when MBOLTUS is used in a SOL 600 model.

MQUATERN Controls whether quaternions will be used for SOL 600 models with large rotation.

MREVPLST Determines whether 2D plain stress triangular element node numbers will be reversed or not.

MRDYNOLD Determines whether dynamic loads created by SOL 600 are the same as in MSC Nastran 2008 and prior releases or uses a new calculation method.

MRCONTAB Determines whether CONTACT and CONTACT TABLE for SOL 600 use table-driven form or not.

MRPLOD4R Determines how PLOAD4 pressures are treated in Marc when PARAM,MRPLOAD4,2 is set.

MRPOISCK Controls whether to check if a “bad” Poisson ratio has been entered in SOL 600 for MAT1 entries.

MRSPRVER Controls how CELAS and all other items map to the Marc input data.

MNASTLDS Option to determine complex force and/or moments using OLOAD’s as calculated from SOL 101.

MSPLINC0 This parameter controls whether to enforce C0 continuity for all spline options if any are requested by setting IDSPL=1 on any BCBODY entry.

MSTTDYNE Controls whether SOL 600 may have static and dynamic load cases in the same analysis.

MTABIGNR Determines whether tables for VCCT analyses will be ignored or used.



Description

9CHAPTER 2Implicit Nonlinear (SOL 600)

MTEMPCHK Controls how temperature-dependent properties are checked in the Marc portion of SOL 600.

MTEMPDWN Option to automatically choose the FeFp multiplicative decomposition plasticity model (PARAM,MARCPLAS,5) for plasticity problems with thermal loading when the temperature decreases (see PARAM,MARCPLAS).

MTET4HYP Controls settings for TET4 elements with hyperelasticity.

MUALLUDS Controls how material, contact and element-related user subroutines are specified in SOL 600.

MVERMOON Controls whether 5-term Mooney series or 5-constant Mooney will be used in the Marc portion of SOL 600.

Bulk Data Entries

CSSHLH Defines conversion of CHEXA elements to Solid Shell elements in SOL 600 only.

CSSHLM Defines conversion of CHEXA or CPENTA elements described by material ID to Solid Shell elements in SOL 600 only.

CSSHLP Defines conversion of CPENTA elements to Solid Shell elements in SOL 600 only.

DMIGROT Defines large rotation and other characteristics of a matrix entered using DMIG in SOL 600.

ISTRESS Defines initial stress values. This is the MSC Marc’s initial stress option used in SOL 600 only.

MATNLE The MATNLEx entries specify advanced forms of nonlinear elastic materials.

MATTUSR Specifies table variation of user defined generic materials in SOL 600 and MD Nastran SOL 400 only.

MATUDS Allows the user to provide material routines for use with enhanced material models in SOL 600.

MATUSR Specifies user-defined, generic material properties for hypoelastic material models in SOL 600 and user defined material models in MSC Nastran SOL 400 only.

MAUXCMD Defines auxiliary command to spawn on Nastran process from another Nastran process in SOL 600.

MDELAM Defines materials for which delamination may occur in SOL 600 only.

MDMIAUX Specifies the DOMAINSOLVER command to be used in conjunction with secondary spawned jobs when MDMIOUT is used. SOL 600 only.



Description


10

Additional DocumentationIn addition to the MD Nastran Quick Reference Guide the user should refer to Marc Volume A for a theoretical discussion of material models, fracture mechanics and dynamics. For additional information on the nonlinear solid section beams one should refer to Marc Volume B.

MISLAND Defines an island of connected elements that will be completely removed if the number of elements within the island becomes smaller than a specified value in SOL 600 only.

MIXTURE Defines constituents of “composite” material on original and potentiality damaged state.

PBMARB6 Defines arbitrary beam/bar cross section for use in SOL 600.

PBMNUM6 Defines four specific numerically integrated BEAM/BAR cross section for use in SOL 600.

RBE3U Defines Method to Distribute Applied Loads to a Surface in SOL 600

TABD1MD Defines how TABLED1 entries are internally modified in SOL 600.



Description

Chapter 3: Numerical Methods and High Performance Computing MSC Nastran 2010 Release Guide

3 Numerical Methods and High Performance Computing

MPI Selection

New Solver Available for Complex Eigenvalue Analysis

MSC Nastran 2010 Release GuideMPI Selection

12

MPI SelectionA single analysis executable is now provided for serial execution as well as DMP execution with all supported MPI implementations.

In previous versions:

• A separate analysis executable was delivered for DMP execution.

• If more than one MPI (Message Passing Interface) implementation was supported on a given platform, a separate analysis executable had to be delivered for each supported MPI implementation.

• An unsupported MPI implementation could not be used without creating a new, separate analysis executable

Starting with MSC Nastran 2010, a single analysis executable will be delivered which can be used for both DMP and serial jobs, and which can be used with any supported MPI implementation on a given platform. Moreover, it is possible to use an unsupported MPI implementation with the same executable.

New Keyword: mpiimplementationOn platforms which support more than one MPI implementation, a supported MPI implementation may be selected using the mpiimplementation keyword. This keyword may be abbreviated as mpiimp.

The table below lists the platforms which support more than one MPI implementation, and the available MPI implementations:

Platform : mpi implementationslinux32 : openmpi (default), hpmpilinux64 : openmpi (default), hpmpi, intelmpilinuxipf : openmpi (default), hpmpi, intelmpiwindows64: msmpi (default) , hpmpi, intelmpi

For example, to select hpmpi on linux64,

mpiimp=hpmpi

should be set on the command line or in an rc file.

The environment variable MPIIMP may also be used to select the MPI implementation. In the case where both the MPIIMP environment variable and the mpiimplementation keyword are set, the keyword value will be used.

New Keyword: mpilibraryAn alternate MPI library may be specified using the mpilibrary keyword. This keyword may be abbreviated to mpilib. The specified MPI library must be in the user’s library path on all nodes used during the computation.The environment variable MPILIB may also be used to select the MPI library. In the case where both the MPILIB environment variable and the mpilibrary keyword are set, the keyword value will be used.

13CHAPTER 3Numerical Methods and High Performance Computing

New Solver Available for Complex Eigenvalue Analysis

IntroductionThe UMFPACK linear equation solver was first incorporated into Nastran in 2004. It was embedded in the Auto-Mset capability, as well as integrated into the frequency response module (FRRD1) where it is available for complex unsymmetric solutions. In MSC Nastran 2010, UMFPACK is available in the complex Lanczos eigenvalue extraction method of the CEAD module.

More information about UMFPACK is available at the following URL: http://www.cise.ufl.edu/research/sparse/umfpack/.

BenefitsThe UMFPACK linear equation solver is more efficient than the default Nastran sparse unsymmetric solver. This is accomplished by employing contemporary matrix reordering techniques and computational kernels.

InputsTo invoke the UMFPACK solver for complex eigenvalue analysis, the UMFLU keyword must be specified as the factorization method, which must be present on the SPARSESOLVER Executive Command:

SPARSESOLVER CEAD (FACTMETH=UMFLU)

OutputsUse of UMPACK will cause User Information Message 4216 is appear in the F04 file:

The “UMFD” time stamp indicates use of UMFPACK.

11:04:19 0:25 5969.0 948.0 15.6 1.2 UMFD BGN *** USER INFORMATION MESSAGE 4216 (FACDRVI) PARAMETERS FOR IN-CORE SPARSE UNSYM. DECOMP ( MATRIX TYPE=RDP) FOLLOW MATRIX SIZE =185931 ROWS NUMBER OF NONZEROES =12587614 T NUMBER OF ZERO COLUMNS =0 NUMBER OF ZERO DIAGONAL TERMS =0 MEMORY REQUIREMENT = 147711 K WORDS MEMORY AVAILABLE = 496541 KWO MAX FRONT SIZE = 2964 NONZERO TERMS = 67880689 11:04:44 0:50 7013.0 1044.0 40.7 25.2 UMFD END

http://www.cise.ufl.edu/research/sparse/umfpack/

MSC Nastran 2010 Release GuideNew Solver Available for Complex Eigenvalue Analysis

14

Guidelines and LimitationsUMFPACK is a memory resident solution algorithm. It does not feature “spill” logic. This means that the solution is limited by available memory. If insufficient memory is available for UMFPACK, the CEAD module will produce System Warning Message 6136, an example of which is shown here:

Demonstration ExampleAn example models is shown here.

*** SYSTEM WARNING MESSAGE 6136 (CLASSD) INSUFFICIENT CORE FOR IN-CORE SPARSE DECOMPOSITION. USER ACTION: INCREASE CORE BY AN ESTIMATED 86618 K WORDS. WARNING: THE ABOVE NUMBER IS ONLY AN ESTIMATE, THE ACTUAL CORE SIZE NEEDED MAY BE HIG USER INFORMATION: AN ALTERNATVE SPARSE DECOMPOSITION METHOD WILL BE ATTEMPTED.

15CHAPTER 3Numerical Methods and High Performance Computing

Example

Analysis type: SOL 107 complex eigenvalue analysis

Number of grid points: 63,007

Number of elements: 42.137

Number of complex eigenvalues: 100

Compute platform used: Intel Linux 8664 2.8GHz

MSC Nastran 2010 Release GuideNew Solver Available for Complex Eigenvalue Analysis

16

Chapter 4: Dynamics (Noise and Vibration)MSC Nastran 2010 Release Guide

4 Dynamics (Noise and Vibration)

Equivalent Radiated Power (ERP)

Frequency Dependent Rigid Absorber Properties

Dynamics - Monitor Points in Dynamic Solution Sequences


Equivalent Radiated Power (ERP)18

Equivalent Radiated Power (ERP)

IntroductionIn automotive applications, the noise inside the passenger compartment can be caused by many sources including vehicle drive train and vibrating body panels. The Equivalent Radiated Power (ERP) calculation focuses on the vibration of body panels, which radiate acoustic power to the passenger cabin. Understanding which panels are responsible for the radiated power is important in understanding the structural behavior and acoustic consequences. The radiated power is a function of skin normal velocity, fluid density and speed of sound through the fluid.

BenefitThe ERP calculation can be used to compare laser measurements to calculated values in a quantitative way to validate calculations. ERP can also be used during the design phase to understand the effect of individual panels on the overall acoustic response. Previously the calculation was performed by in-house tools. In MSC Nastran 2010 the ERP calculation is made directly by the solver and provides convenient output in the form of a CSV file.

TheoryIn a mathematical sense, ERP squares the normal velocity and multiplies it with the element area. The sum over this product, multiplied with a constant yields the ERP over a panel. ERP values can be calculated for both structure and structure-fluid models.

where

and

for Frequency Response, 1.0 for Transient Response

ERPRLF= Radiation Loss Factor

ERPRHO= Fluid density

ERPC= Speed of sound in fluid

In MSC Nastran 2010 only direct frequency response and modal frequency response are supported. In addition to the ERP calculation, an ERPDB calculation is also performed to calculate an equivalent radiated power sound pressure level.

ERP C Vn2S

surf

panel

=

C ERPRLF ERPRHO ERPC=

ERPdB 10LOGRHOCP

ERPREFDB--------------------------------- ERPvalue =

19CHAPTER 4Dynamics (Noise and Vibration)

InputThe ERP calculation is typically requested for a group of elements defined on a SET3 Bulk Data entry. The parameters ERPRHO, ERPC, ERPRLF, ERPREFDB, and RHOCP can be defined on either the ERP Case Control command, or as PARAM entries in the Bulk Data Section. The RHOCP parameter can only be specified on the ERP Case Control command. The ERP Case Control also references an ERPPNL Bulk Data entry.

The new Bulk Data entry for ERPPNL is:

Defines one or more panels by referencing sets of elements or properties.

Format:

Example:

Remarks:

1. The SET3 entries can only refer to CQUAD4, CQUADR, CTRIA3, or CTRIAR structural elements or PSHELL or PCOMP property entries. CQUAD8 and CTRIA6 entries are ignored.

2. NAMEi are used in a Case Control SET definition defining setp to select the panels in the Case Control command ERP.

ERPPNL Equivalent Radiated Power Definition

1 2 3 4 5 6 7 8 9 10

ERPPNL NAME1 SETID1 NAME2 SETID2 NAME3 SETID3 NAME4 SETID4

NAME5 SETID5

ERPPNL ROOF 1 DOORLF 16

Field Contents

NAMEi Panel label. (CHAR)

SETIDi Identification number of a SET3 Bulk Data entry that lists the panel property entries or the panel elements. (Integer > 0)



The new ERP Case Control command is:

Requests the form and type of ERP panel participation factor output.

Format:

Examples:SET 17 = 10.,20.,30.,40.,80.,100. $ A list of frequenciesSET 25 = ROOF, DOORLF $ A list of ERP Panel names

$ from a ERPPNL Bulk EntryERP ( PRINT,PUNCH,SOLUTION=17,KEY=frac ) = 25

ERP Equivalent Radiated Power Panel Participation Factor Output Request

Describer Meaning

SORT1 Output is presented as a tabular listing of ERP panels for each frequency.

SORT2 Output is presented as a tabular listing of frequency for each ERP panel.

PRINT Output is written to the .f06 file

PUNCH Output is written to the .pch file

PLOT Results are computed and placed on the ERP table but not output.

SOLUTION Keyword to select frequencies

setf Identifier of Case Control SET command defining frequencies.

ERPSORT2

SORT1

PRINT, PUNCH

PLOTSOLUTION

ALL

setf= ,

KEYfrequency

fraction

= FILTER0.01

_valuereal

= ,

ERPRHO1.0

_valuereal

= ERPC1.0

_valuereal

=

RHOCP1.0

_valuereal

= ERPRLF1.0

_valuereal

=

ERPREFDB1.0

_valuereal

= CSV unit= ALL

setp

NONE

=


Remarks:

1. ERP is required to produce any ERP output.

2. Output is generated in SORT2 by default. Unlike other Case Control requesting SORT2 format, the ERP command does not force all other output into SORT2 format.

3. FILTER has no effect on PUNCHed, CSV or OP2 output.

4. In addition to individual panel output a summary named ALLPANEL is produced. If there are multiple subcases, the panel name is formed from the serial subcase number (1-nsubc) and the characters ‘ALLP’ as in ALLP0002 unless the ERP command request output for ALL panels across the Subcases. In this case, the summary panel name ALLPANEL is retained.

5. Selectable frequencies are dependent on the presence of an OFREQ Case Control command.

6. ERPRHO, ERPC, ERPRLF, RHOCP, and ERPREFDB are actually PARAM,name,value entries.

ALL If associated with SOLUTION, all frequencies are selected. If associated with setp, all ERPPNL entries are selected.

KEY Keyword selecting the output item used to sort the printed output. The default produces output sorted on either frequency (SORT2) or ERP panel name (SORT1). KEY=fraction produces output sorted in descending order of the fractional ERP value of total ERP.

FILTER Keyword specifying the value of a filter to b e applied to the printed output only. ERP values are printed only if the fractional ERP value of total ERP exceeds the filter value.

ERPRHO Fluid density for Equivalent Radiated Power (ERP) analysis. This item is actually an MSC Nastran parameter.

ERPC Phase speed of the fluid for Equivalent Radiated Power (ERP) analysis. This item is actually an Nastran parameter.

ERPRLF Radiation loss factor. In frequency the scale factor C = ERPRLF * (½ERPRHO * ERPC). In transient the scale factor C = ERPRLF * (ERPRHO * ERPC).

RHOCP Scale factor used in dB computation. This item is actually an MSC Nastran parameter.

ERPREFDB Scale factor used in dB computation. This item is actually an MSC Nastran parameter.

The dB calculation is ERPdB = 10 log .

CSV Results will be written to a .csv file.

unit Unit of the .csv file as used on the required ASSIGN statement.

setp Identifier of Case Control SET command defining NAMEi entries from an ERPPNL Bulk Data entry defining panels.

NONE No ERP output is produced.

Describer Meaning

RHOCPERP

ERPREFDB------------------------------



7. The filter process avoids printing ERP for cases where ERP/ERPMAX is less than the FILTER value. ERPMAX is the maximum ERP value across all frequencies for a panel.

8. If output to a .csv file is requested, the file must be assigned with logical key USERFILE and FORM=FORMATTED, e.g.,

ASSIGN USERFILE = myfile.csv UNIT=50 FORM=FORMATTED STATUS=NEW

The SET3 Bulk Data entry is also necessary to define.

Set 3 Examples:

• Set3,id,prop,pshellid1,pshellid2,etc.

• Set3,id,elem,elemid1,elemid2,etc.

• Set3 prop, can be exchanged between acoustic and ERP panels

See Bulk Data entry SET3 (p. 3119) in the MD/MSC Nastran Quick Reference Guide. Also see the Bulk Data entry ERPPNL (p. 1816) in the MD/MSC Nastran Quick Reference Guide.

Example Input The following input is typical for ERP calculation including CSV output.

File Management

ASSIGN USERFILE=’myerp.csv’ UNIT=30 FORMATTED NEW DELETE

Case ControlERP(PUNCH,Filter=0.0,rhocp=2.0E9,ERPRHO=1.189E-12,ERPC=3.43E5,CSV=30)=ALL

Example ERP Panel Definition Bulk DataERPPNL,ROOF1,103,ROOF2,203,ROOF3,303set3,103,prop,100set3,203,prop,200set3,303,element,114,124,134,214,224,234,,314,324,334

OutputThe results are available in the OP2, MASTER, Print, Punch, and CSV formats. The output includes ERP, Fraction, and ERP(dB). Note that the fraction is not based on the entire ERP of the model, just the ERP that is calculated and there is no check for overlapping or missing elements. There is also a summation of total ERP. Both SORT1 and SORT2 options are available.


Figure 4-1 Representative ERP results for a complicated system.

Guidelines and Limitations1. ERP is calculated currently for linear elements 3 and 4 node shells only. If desired the user can

generate a layer of linear shells on top of quadratic solids.

2. PSHELL and PCOMP are supported

3. ERP is supported in direct and modal frequency response only.

4. There is no Direct Results Access (DRA) support

5. ERP is not supported in Optimization.

6. No limits on coordinate systems

Test CasesThe following test cases are available in the TPL in directory /tpl/erp_mdr4:

erp_1000.dat, erp_base1_frac.dat, erp_base2_frac.dat, erp_c_param.dat, erp_erpx3.dat, erp_fs.dat, erp_rhocp.dat, erp_soln.dat, erp_base1.dat, erp_base2.dat, erp_c.dat, erp_def.dat, erp_frac_c.dat, erp_ofreq.dat, erp_rho.dat

TPL Example Problem erp_base1.dat

Test problem erp_base1.dat is a simple fluid bounded by two panels. The excitation is on one panel and the ERP is measured.



Figure 4-2 Example erp_base1.dat geometry.

The input for erp_base1.dat is a standard modal frequency response with a pressure loading and including fluid-structure interaction. The case control and bulk data entries required for ERP calculation are as follows:

Case ControlERP(PRINT,PUNCH,FILTER=0.0)=ALL

Example ERP Panel Definition Bulk DataERPPNL,ERPX0,103,ERPX3,203,erpeid3,303set3,103,prop,100set3,203,prop,200set3,303,element,114,124,134,214,224,234,,314,324,334

Listing 4-1 TPL example erp_base1.dat Output in SORT1 format.

FREQUENCY = 8.000000E+00 E Q U I V A L E N T R A D I A T E D P O W E R PANEL ERP FRACTION ERP(dB) AREA ERPX0 2.702487E-02 7.543413E-04 -1.568236E+01 1.000000E+01 ERPX3 4.871353E-03 3.356652E-04 -2.312350E+01 9.000000E+00 ERPEID3 4.871353E-03 3.356652E-04 -2.312350E+01 9.000000E+00 ALLPANEL 3.189623E-02 6.336370E-04 -1.496261E+01 1.900000E+01


Listing 4-2 TPL example erp_base1.dat Output in SORT2 format.

To obtain CSV output, an ASSIGN statement is added and the ERP case control is modified as follows:

ASSIGN USERFILE='MYERP.CSV' UNIT=30 FORMATTED NEW DELETE

ERP(PRINT,PUNCH,SORT1,FILTER=0.0,CSV=30) = ALL

The resulting CSV file is easily manipulated into a graph using Microsoft Excel, or other programs that understand CSV format. Note that the graph shown in Figure 4-3 is based on a modified erp_base1.dat file that has a FREQ1 entry with more output frequencies. Note that the panels that contribute to the overall ERP switch at the frequencies of 31 and 33Hz.

Listing 4-3 TPL example erp_base1.dat Output in CSV format (partial listing)

PANEL = ERPX3 (AREA = 9.000000E+00) E Q U I V A L E N T R A D I A T E D P O W E R FREQUENCY ERP FRACTION ERP(dB) 2.000000E+00 4.220276E-03 2.908022E-04 -2.374659E+01 4.000000E+00 1.497942E-03 1.032172E-04 -2.824505E+01 6.000000E+00 1.099377E-01 7.575362E-03 -9.588533E+00 8.000000E+00 4.871353E-03 3.356652E-04 -2.312350E+01 1.000000E+01 2.019563E+00 1.391599E-01 3.052574E+00 1.200000E+01 1.467790E-01 1.011395E-02 -8.333361E+00 1.400000E+01 1.451253E+01 1.000000E+00 1.161743E+01 1.600000E+01 2.543595E-02 1.752689E-03 -1.594552E+01 **ERP MAX** 1.451253E+01

Subcase, 1000" EQUIVALENT RADIATED POWER IN PANELS OF QUAD4S "" ALL IN 1 SUBCASE "" FIRST SUBCASE (1000) SUBCASE 1000 "Equivalent Radiated Power , ERP , ERP , Fraction , Fraction , ERP(dB) , ERP(dB) Area , 1.00000E+01, 1.90000E+01, 1.00000E+01, 1.90000E+01, 1.00000E+01, 1.90000E+01Frequency , ERPX0 , ALLPANEL , ERPX0 , ALLPANEL , ERPX0 , ALLPANEL 2.00000E+00, 3.60158E-03, 7.82186E-03, 1.00530E-04, 1.55386E-04,-2.44351E+01,-2.10669E+01 4.00000E+00, 3.66840E-04, 1.86478E-03, 1.02395E-05, 3.70450E-05,-3.43552E+01,-2.72937E+01 6.00000E+00, 9.09168E-02, 2.00855E-01, 2.53775E-03, 3.99009E-03,-1.04136E+01,-6.97118E+00 8.00000E+00, 2.70249E-02, 3.18962E-02, 7.54341E-04, 6.33637E-04,-1.56824E+01,-1.49626E+01 1.00000E+01, 5.50420E+00, 7.52376E+00, 1.53638E-01, 1.49464E-01, 7.40694E+00, 8.76435E+00 1.20000E+01, 2.83194E-01, 4.29973E-01, 7.90474E-03, 8.54165E-03,-5.47917E+00,-3.66559E+00 1.40000E+01, 3.58258E+01, 5.03383E+01, 1.00000E+00, 1.00000E+00, 1.55420E+01, 1.70190E+01 1.60000E+01, 2.49801E-02, 5.04161E-02, 6.97266E-04, 1.00154E-03,-1.60241E+01,-1.29743E+01



Figure 4-3 TPL example erp_base1.dat plot in Microsoft Excel.

ERP for a Complicated Automotive Assembly

The example shown in Figure 4-4 is used to demonstrate a more complicated system level automotive example. The loading is based on an engine event and the Equivalent Radiated Power is calculated for various panels that connect directly to the passenger compartment. Note that this example does not perform an acoustic response, but the ERP calculations provide insight into which panels would contribute to an acoustic response at various frequency levels.

Figure 4-4 System Level ERP Example


GUI SupportNeither SimXpert nor Patran currently support pre or post processing of ERP. However, the CSV output provides a convenient interface for users who want to generate plots using Microsoft Excel.


Frequency Dependent Rigid Absorber Properties28

Frequency Dependent Rigid Absorber Properties

IntroductionThe capability to model basic rigid skeleton porous absorber properties in acoustic response analysis was introduced in MSC Nastran 2007. It allows modeling some types of absorbent material such as vehicle seat structures or lining materials with stiff carcasses.

The absorber material is described taking into account an equivalent fluid analogy and is modeled as standard fluid elements using:

• Standard fluid solid elements (CHEXA, CPENTA or CTETRA)

• Connecting grid points with CD field defined as -1

• Referenced PSOLID entry with option PFLUID defined in field 8

• Referencing MAT10 entry where the ‘normalized admittance coefficient’ is defined in field 7 and equivalent values are used for density and bulk modulus

The limitation for MSC Nastran 2007 implementation is that the normalized admittance coefficient cannot be defined as frequency dependent.

In MSC Nastran 2010 the frequency dependency for this coefficient has been implemented. The new option FFLUID has been added for field 8 of PSOLID entry. Furthermore the user must take care to

define the normalized admittance coefficient in the MAT10 entry properly calculated at .

BenefitsThe new capability introduced in MSC Nastran 2010 allows defining an automatic calculation of a different value for the normalized admittance depending on the value of the excitation frequency.

The major benefit for the user is the possibility to describe in a very simple way the right absorbing behaviors of the rigid porous material at the different excitation frequencies.

Note the user interface chosen to define the frequency dependency of the porous absorber allows:

• Maintaining the backward compatibility (Frequency independent porous absorbers can still be modeled)

• Defining one set of fluid elements to have frequency dependent normalized admittance coefficient and another set to be frequency independent

Theory

The porous absorber properties are described by complex parameters (density and bulk modulus). The general implementation allows introducing complex material properties for elements in the fluid which

1.0=


represent region where sound energy is absorbed. It implies that if the complex density and bulk modulus are constant:

Mass density

Bulk Modulus B

Damping coefficient GE

The normalized admittance coefficient is a function of the frequency:

INPUTAs already mentioned, the equivalent fluid analogy allows using the same entries used to describe a standard fluid region. PSOLID and MAT10 entries are affected by this implementation.

PSOLID Entry

A new option for field 8 (FCTN) of the PSOLID entry has been introduced

1 2 3 4 5 6 7 8 9 10

PSOLID PID MID CORDM IN STRESS ISOP FCTN COROT

FAC

FCTN Fluid element flag. (Character: “FFLUID” indicates a fluid element with frequency dependent rigid absorber properties, “PFLUID” indicates a fluid element, “SMECH” indicates a structural element; Default = “SMECH.”)

r= ii+ e

r2 i

2+

r-------------------=

B Br= iBi+ Be

Br2

Bi2

+

Br--------------------=

GEi

r-----=

Bi

Br----- 2f

Bi

Br-----= =



All the elements which refer to a PSOLID entry where the option FFLUID has been selected will be considered as rigid porous absorber with frequency dependent normalized admittance coefficient.

MAT10 Entry

No modification has been done in the format of this entry and no new options have been added. The only remark that has to be done is relative to the meaning of field 7 in case of frequency dependent rigid porous absorber.

In fact:

• If the MAT10 entry is referenced in a PSOLID entry where FFLUID option is selected, the value

in the 7th field (ALPHA) is considered as the normalized admittance coefficient calculated at

unit circular excitation frequency ;

• If the MAT10 entry is referenced in a PSOLID entry where PFLUID option is selected, the value defined in field 7 (ALPHA) has no special meaning but it is only the normalized admittance coefficient calculated at the most appropriate excitation frequency (defined in order to have good results in the range of interest.

The use of a nonzero value in field 7 of the MAT10 entry causes the generation of a damping matrix, because the normalized admittance coefficient is multiplied by the imaginary operator i. Consequently, the use of modal methods on the fluid are not appropriate and frequency response analysis must be carried out using the direct method, at least for the fluid.

OUTPUTThere is no additional output are generated for the elements used to describe the frequency dependent rigid absorber region.

Test CasesThere are many test cases available in the TPL in subdirectory /tpl/fdr_absorb

TPL problem fh8pr10.dat

1 2 3 4 5 6 7 8 9 10

MAT10 MID BULK RHO C GE ALPHA

1.0=


Consider the following unbounded fluid (air) and porous absorber medium domains as in Figure 4-5. An acoustic source is placed at the location indicated and the acoustic response (pressure) at the center of the fluid is monitored.

Figure 4-5

Using experimental methods the following properties have been determined.

The equations illustrated above have been used to calculate the equivalent properties to be used in the MAT10 entries. Two different calculations have been executed to check the effect of the new frequency dependent porous absorber properties implementation.

1. Frequency independent materials have been considered and frequency of 250 Hz was selected to calculate the values of alpha for air and the porous absorber.

Air Material

Porous Absorber Material

Density Speed of Sound Bulk Modulus

Air

Porous Absorber

MID BULK RHO C GE ALPHA

MAT10 1 141652.5 1.225 0.0 31.41907

PID MID CORDM IN STRESS ISOP FCNT

PSOLID 1 1 0 PFLUID


MAT10 2 -232389. 56.16948 -3.67804 -939.196

1.225 0.0i+ 340.0 3.4i+ 141595.8 2832.2i+

3.8663 14.2204i+ 92.7076 70.2854i+ 171190.0– 102356.3+



2. Frequency dependent materials have been considered. The normalized admittance coefficients for air and porous absorbers have been calculated for .

AirMaterial

Porous AbsorberMaterial

Both the analyses have been executed using 2 different models in which 8 HEXA and 20 HEXA elements have been used. Notice that the values of bulk modulus, GE damping coefficient and alpha are all negative; this is a normal characteristic of the implementation. The response at the centre of the air domain is calculated and the results compared with the same model run in Actran. Both HEXA-20 and HEXA elements are compared.


PSOLID 2 2 0 PFLUID


MAT10 1 141652.5 1.225 0.0 0.020002


PSOLID 1 1 0 FFLUID


MAT10 2 -232389. 56.16948 -3.67804 -0.59791


PSOLID 2 2 0 FFLUID

1.0=


The results using frequency dependent rigid absorber properties fit completely with those from Actran. In fact the increasing differences obtained using the original implementation for porous material properties departing from the reference excitation frequency (250 Hz in the example) disappear.

GUI Support

Currently neither Patran or SimXpert support the preprocessing definition of the FFLUID option in field 8 of the PSOLID entry.

The post-processing capability of Patran and SimXpert is not affected by this implementation.

Additional Information and References

Additional documentation regarding the implementation of rigid porous absorbers can be found in the following references:

1. M.E. Delany and E.N. Bazley, Acoustical Characteristics of Fibrous Absorbent Materials, National Physics Laboratory, Aerodynamics Division, NPL Aero Report Ac 37, March 1969.



2. J. Wandinger, Possible Implementations of Porous Absorbers in Nastran, MSC internal memo, April 2006.

3. M. Etchessahar, Caracterérisation mécanique en basses fréquences des matériaux acoustiques, Thèse de Doctorat, Université du Maine, 2002.

4. MSC Nastran Quick Reference Guide

5. MSC Nastran 2007 Release Guide


Dynamics - Monitor Points in Dynamic Solution Sequences

IntroductionMonitor points is a generic name for four types of user requests. The MSC Nastran R1 Release Guide provided the most comprehensive discussion of these inputs. Briefly,

1. MONPNT1 – The MONPNT1 provides integrated loads at a user defined point in a user defined coordinate system.

2. MONPNT2 – The MONPNT2 provides element results (e.g., Stress, Strain, Force)

3. MONPNT3 – The MONPNT3 provides a summation of grid point forces at a user specified integration points.

4. MONDSP1 – The MONDSP1 allows for the sampling of a displacement vector to create a blended displacement response at a user specified point and coordinate system.

Prior to MSC Nastran 2010, monitor points were only available in SOLs 101, 103, 144 and 146. Design of Monitor Points, 57 discusses the MSC Nastran 2010 implementation in SOL 200. These sections discusses their application in the linear dynamic response solution sequences; i.e., SOLs 108, 109, 111 and 112.

BenefitsMONPNT1 was first introduced in MSC.Nastran 2001 and provides the user with a way to extract the applied loading for a specified set of structural nodes (or aerodynamic elements for static or dynamic aeroelasticity). This enables the batch calculation of VMT (shear, moment and torque) data for user specified regions and locations.

MONPNT2 provides a way of pinpointing a particular response for output, as opposed to finding it in a large OFP listing.

MONPNT3 provides a summation of the internal loads and therefore useful in calculating resultant forces at a cut in the structure.

The MONDSP1’s ability to provide an averaged displacement is seen as providing a qualitative assessment of the elastic deflection of a vehicle.

Input

There is no change to the input required to define the monitor points, only the solution sequences which are supported. The MSC Nastran Quick Reference Guide provides guidance on specifying the MONPNT1, MONPNT2, MONPNT3 and MONDSP1 Bulk Data entries. The MONITOR Case Control command must be used to obtain output results in the dynamic solution sequences. The command provides options for frequency response results in terms of either the Real/Imag or MAG/Phase form.


Dynamics - Monitor Points in Dynamic Solution Sequences36

Output

Guidelines and Limitations1. Dynamic monitor points are not available as design responses in SOL 200.

2. Inertia results are available for the MONPNT1 but have not been implemented for the MONPNT3.

3. The MONSUM feature can be used but is of limited utility when the MONSUM spans monitor types as described in Connectors (Ch. A) of this guide.

4. In Frequency Response analysis, the monitor point output is in SORT2 format.

Test Cases

The following test cases are available in the TPL in directory /tpl/ue6_09a. There are four TPL files with the name sXXXm13d and four with sXXXm2 where the XXX is one of 108,109,111 or 112. The m13d files contain MONPT1, MONPNT3 and MONDSP1 entries while m2 files contain MONPNT2 entries.

TPL example problem s111m13d.dat

Example problem s111m13.dat is a modal frequency response model that contains MONPNT1, MONPNT3, and MONDSP1 entries. There are two subcases for a central load; the 1st subcase is shown in Figure 4-6.

Figure 4-6 TPL example s111m13d.dat

A set of MONDSP1s are generated to define virtual point displacement results using an RBE3 derived from the GRIDs defined on the SET1 lists referenced on the AECOMP entry:

MONDSP1 DISP2 THIS IS A DISPLACEMENT MONITOR POINT 123456 PLATE3 52 1. 2. 3. AECOMP PLATE3 SET1 3 4


SET1 3 110901 110902 110903 110904 110905 SET1 4 110801 110802 110803

MONDSP1 DISPREF THIS SHOULD MATCH GRID 110902 123456 POINT1 4. 18. 0. 123456AECOMP POINT1 SET1 110902 SET1 110902 110902

A set of MONPNT1s are generated to define integration load monitor points; the integration occurs over the GRIDs associated with the SET1 entries defined on the AECOMP entry:

MONPNT1 MPT11 THIS IS THE FIRST MONPNT1 123456 PLATE1 52 1. 2. 3.0 AECOMP PLATE1 SET1 1 2 SET1 1 110000 110010 111010 111000 SET1 2 110505

MONPNT1 MPT12 THIS IS THE SECOND MONPNT1 123456 PLATE2 20. 20. 0.0 AECOMP PLATE2 SET1 1 SET1 1 110000 110010 111010 111000

Finally, a set of MONPNT3s are generated to sum Grid Point Forces defined on the GRIDSET and ELEMSET.

MONPNT3 MPT31 THIS IS THE FIRST MONPT3 123456 5 6 1. 2. 3. SET1 5 110901 119992 110903 110904 110905 SET1 6 1100081 9999980

MONPNT3 MPT41 THIS IS THE SECOND MONPT3 123456 3 4 1. 2. 3.SET1 3 110901 110902 110903 110904 110905 SET1 4 110801 110802 110803

Typical output for each output is shown below:

STATIC LOAD SUBCASE 1

S T R U C T U R A L M O N I T O R P O I N T D I S P L A C E M E N T S (REAL/IMAGINARY)

MONITOR POINT NAME = DISP2 COMPONENT = 123456 GENERAL SUBCASE NO. 1 LABEL = THIS IS A DISPLACEMENT MONITOR POINT CP = 52 X = 1.000000E+00 Y = 2.000000E+00 Z = 3.000000E+00 CD = 52

FREQUENCY T1 T2 T3 R1 R2 R3 ------------ ------------ ------------ ------------ ------------ ------------ ------------ 1.000000E+02 9.331630E-06 7.714264E-06 1.142239E-06 3.345541E-07 -4.731310E-07 4.621815E-07 -5.560066E-08 -4.663981E-08 -6.488654E-09 -1.900483E-09 2.687689E-09 -3.033787E-09 2.000000E+02 9.721770E-06 8.044613E-06 1.186316E-06 3.474638E-07 -4.913880E-07 4.847471E-07 -1.203002E-07 -1.010250E-07 -1.398631E-08 -4.096495E-09 5.793319E-09 -6.610803E-09 3.000000E+02 1.044643E-05 8.658742E-06 1.267942E-06 3.713716E-07 -5.251988E-07 5.268791E-07 -2.072620E-07 -1.743763E-07 -2.394549E-08 -7.013472E-09 9.918547E-09 -1.152342E-08 4.000000E+02 1.165478E-05 9.684080E-06 1.403449E-06 4.110607E-07 -5.813276E-07 5.976722E-07 -3.416177E-07 -2.881529E-07 -3.912291E-08 -1.145883E-08 1.620524E-08 -1.929935E-08



Torquebox example problem

This example is under construction.

GUI Support

Patran

Patran supports Monitor Point creation via the Flight Loads application. To access Flight Loads, it needs to be installed during the Patran installation and the current analysis type must be Aeroelasticity. The figures in this section provide a general description of how to create the various Monitor Points described in this chapter. After the Monitor Points are created, the user can export them to a bdf file for subsequent inclusion in a non-aeroelasticity solution. Finally, the user will have to change the Analysis Type back to Structures. Currently Patran does not support the post-processing of Monitor Point results.

STATIC LOAD SUBCASE 1

S T R U C T U R A L M O N I T O R P O I N T I N T E G R A T E D L O A D S (MONPNT1) (REAL/IMAGINARY)

MONITOR POINT NAME = MPT11 COMPONENT = CX GENERAL SUBCASE NO. 3 LABEL = THIS IS THE FIRST MONPNT1 CP = 52 X = 1.000000E+00 Y = 2.000000E+00 Z = 3.000000E+00 CD = 52

FREQUENCY INERTIAL EXTERNAL FLEXIBLE GUST TOTAL TOTAL INCREMENT AERO ------------ ------------ ------------ ------------ ------------ ------------ ------------ 1.000000E+02 -4.455447E-05 1.306395E+01 0.000000E+00 1.306390E+01 2.311694E-07 0.000000E+00 0.000000E+00 2.311694E-07 2.000000E+02 -1.835128E-04 1.306395E+01 0.000000E+00 1.306376E+01 1.957264E-06 0.000000E+00 0.000000E+00 1.957264E-06 3.000000E+02 -4.345861E-04 1.306395E+01 0.000000E+00 1.306351E+01 7.308116E-06 0.000000E+00 0.000000E+00 7.308116E-06 4.000000E+02 -8.349048E-04 1.306395E+01 0.000000E+00 1.306311E+01 2.028260E-05 0.000000E+00 0.000000E+00 2.028260E-05

STATIC LOAD SUBCASE 1 S T R U C T U R A L I N T E G R A T E D F R E E B O D Y M O N I T O R P O I N T L O A D S (MONPNT3) (REAL/IMAGINARY)

MONITOR POINT NAME = MPT31 COMPONENT = CMY SUBCASE NO. 1 LABEL = THIS IS THE FIRST MONPT3 CP = 0 X = 1.000000E+00 Y = 2.000000E+00 Z = 3.000000E+00

FREQUENCY RESULTANT ------------ ------------ 1.000000E+02 -3.689061E-01 -3.930317E-05 2.000000E+02 -3.910366E-01 4.639984E-04 3.000000E+02 -4.319148E-01 2.253416E-03 4.000000E+02 -4.992898E-01 6.649540E-03


Figure 4-7 Setting Analysis Type to Aeroelasticity for Monitor Point Access via Flight Loads

Figure 4-8 Flight Loads icon enables the flight loads menus

Figure 4-9 Monitor Point Action-Object-Type menu



Figure 4-10 Example of Creating a MONPNT1






Figure 4-13 Example of Creating a MONDSP1

Figure 4-14 Exporting Monitor Points from Flight Loads

SimXpert

SimXpert does not currently support Monitor Point Creation.

Chapter 5: Optimization MSC Nastran 2010 Release Guide

5 Optimization

Part Superelement Optimization Enhancements

Optimization - Invariant DRESP3 Gradients

Design of Monitor Points

Parallel Sensitivities

DTABLE Enhancement for Dynamic Analysis

Constants with DTABLE2

New Optimizer - IPOPT

Topology and Topometry Enhancements

Build External Servers Using the SCons Tool

Deactivation of Original Design Sensitivity (DSA)

MSC Nastran 2010 Release GuidePart Superelement Optimization Enhancements

44

Part Superelement Optimization Enhancements

IntroductionPart Superelements are a method for defining superelements by having a completely independent bulk data for each Superelement. One advantage of Part Superelements is that bulk data for various components or subassemblies can be easily assembled without the need to renumber GRIDs, properties, etc. Another advantage is that parameters are independent for each superelement. The method for defining Part Superelements in MSC Nastran is by the delimiter BEGIN SUPER=seid, where seid is the user-defined Superelement Identification number. Note that the residual structure is defined in the main Bulk Data Section (BEGIN BULK, or BEGIN SUPER=0). The residual structure is also designated as a Part Superelement with seid=0. The residual structure includes the effects of all the upstream Part Superelements and is used to calculate the solution vector. For more information on Superelements and Part Superelements, refer to the MSC Nastran 2001 Superelement User’s Guide.

Part Superelement Optimization was introduced in MSC Nastran 2008 with the limitation that the design variables, responses, and constraints for SOL 200 optimization were required to be defined in the main Bulk Data Section. With the MSC Nastran 2010 release, Part Superelement Optimization extends the design model in SOL 200 to allow design of upstream Part Superelements in addition to the residual structure.

BenefitsPart Superelement technology is widely used in the Aerospace and Automotive industry to assemble bulk data models from various sources. Some advantages of Part Superelement technology include automatic connections based on geometry searches and the ability to have completely independent bulk data. Extending the design model to Part Superelements will allow users great flexibility in assembling models and performing design optimization using Part Superelement technology. Design variable, response, and constraint definition for Part Superelement Optimization can include the residual structure and all Part Superelement(s). Design variables can be linked across Part Superelements. In addition synthetic responses can include responses from different Part Superelements are supported with new Bulk Data entries, namely SEDLINK, SEDRSP2, and SEDRSP3.

InputEach Partitioned Superelement may contain traditional Solution 200 bdf entries such as DESVAR, DVPREL1, DRESP1, etc. The new bulk data entries are introduced to support cross boundary Part Superelement design variable linking and synthetic responses. These new bulk data entries are for Part Superelement only and must involve quantities from more than one Part Superelement. See additional comments and remarks for Bulk Data entries SEDLINK, SEDRSP2, and SEDRSP3 (p. 3096) in the MD/MSC Nastran Quick Reference Guide. These bulk data entries must be specified in the residual Bulk Data Section.

45CHAPTER 5Optimization

1 2 3 4 5 6 7 8 9 10

SEDLINK ID DSEID DDVID C0 CMULT ISEID1 IDV1 C1

ISEID2 IDV2 C2 ISEID3 IDV3 C3

ISEID4 IDV4 C4 -etc.-

1 2 3 4 5 6 7 8 9 10

SEDRSP2 ID LABEL EQID or FUNC

REGION METHOD C1 C2 C3

“DESVAR” DVSEID1 DVID1 DVSEID2 DVID2 DVSEID3 DVID3

DVSEID4 DVID4 -etc.-

“DTABLE” LBSEID1 LABL1 LBSEID2 LABL2 LBSEID3 LABL3

LBSEID4 LABL4 -etc.-

“DRESP1” R1SEID1 NR1 R1SEID2 NR2 R1SEID3 NR3

R1SEID4 NR4 -etc.-

“DNODE” NDSEID1 G1 CMP1 NDSEID2 G2 CMP2

NDSEID3 G3 CMP3 -etc.-

“DVPREL1 P1SEID1 DPIP1 P1SEID2 DPIP2 P1SEID3 DPIP3

P1SEID4 DPIP4 -etc.-

“DVCREL1” C1SEID1 DCIC1 C1SEID2 DCIC2 C1SEID3 DCIC3

C1SEID4 DCIC4 -etc.-

“DVMREL1” M1SEID1 DMIM1 M1SEID2 DMIM2 M1SEID3 DMIM3

M1SEID4 DMIM4 -etc.-

“DVPREL2” P2SEID1 PDI2P1 P2SEID2 DPI2P2 P2SEID3 DPI2P3

P2SEID4 DPI2P4 -etc.-

“DVCREL2” C2SEID1 DC12C1 C2SEID2 DC12C2 C2SEID3 DC12C3

C2SEID4 DCI2C4 -etc.-

“DVMREL2” M2SEID1 DM12M1 M2SEID2 DMI2M2 M2SEID3 DM12M3

M2SEID4 DMI2M4 -etc.-

1 2 3 4 5 6 7 8 9 10

SEDRSP3 ID LABEL GROUP TYPE REGION

“DESVAR” DVSEID1 DVID1 DVSEID2 DVID2 DVSEID3 DVID3

DVSEID4 DVID4 -etc.-

“DTABLE” LBSEID1 LABL1 LBSEID2 LABL2 LBSEID3 LABL3

LBSEID4 LABL4 -etc.-

“DRESP1 R1SEID1 NR1 R1SEID2 NR2 R1SEID3 NR3

R1SEID4 NR4 -etc.-

“DNODE” NDSEID1 G1 CMP1 NDSEID2 G2 CMP2

NDSEID3 G3 CMP3 -etc.-


46

For user convenience, specifying PARAM,PSENPCH,YES will write the updated bulk data entries into separate ‘.pch’ files for each Part Superelement, and each design cycle. Note that the number of design cycles that are output is dependant on the value of PARAM,DESPCH. The advantage of specifying PARAM,PSENPCH,YES is that each of the ‘.pch’ files with the updated design bulk data can be used to replace the original model with an ‘INCLUDE’ entry after the appropriate ‘BEGIN SUPER=seid’. If PARAM,SPENPCH,YES is not specified, the updated bulk data entries for all Part Superelements are written to a single ‘.pch’ file which will require the user to manually extract each Part Superelement model from the .pch file and place it in the appropriate ‘BEGIN SUPER=seid’ section of the Bulk Data Section.

OutputFor SOL 200 with design models in each Part Superelement, the .f06 output is similar to non Part Superelement optimization jobs. There are some minor differences that are specific to Part Superelements only.

“DVPREL1” P1SEID1 DPIP1 P1SEID2 DPIP2 P1SEID3 DPIP3

P1SEID4 DPIP4 -etc.-

“DVCREL1” C1SEID1 DCIC1 C1SEID2 DCIC2 C1SEID3 DCIC3

C1SEID4 DCIC4 -etc.-

“DVMREL1: M1SEID1 DMIM1 M1SEID2 DMIM2 M1SEID3 DMIM3

M1SEID4 DMIM4 -etc.-

“DVPREL1” P2SEID DPI2P1 P2SEID2 DPI2P2 P2SEID3 DPI2P3

P2SEID4 DPI2P4 -etc.-

“DVCREL2” C2SEID1 DC12C1 C2SEID2 DCI2C2 C2SEID3 DCI2C3

C2SEID4 DCI2C4 -etc.-

“DVMREL2” M2SEID DMI2M1 M2SEID2 DMI2M2 M2SEID3 DMI2M3

M2SEID4 DMI2M4 -etc.-

“USRDATA” String

-etc.-

PARAM,PSENPCH Default=NO. Setting PSENPCH to YES causes updated bulk data entries of a Part Superelement for a design cycle punched to a separate file named as follows

JOBNAME_psexx_yy.pch Where xx is for Part Superelement ID and yy is for design cycle.

Note that PARAM, PSENPCH has no effect for non-Part Superelement run.


Comparison Between Input Property Values from Analysis and Design Models

This section of the .f06 will repeat for the design model of each Part Superelement. A sample for a Part Superelement is shown as follows:

Please note that the Part Superelement ID shows up in the title line of a page. In addition, residual Part Superelement will get two versions of above output, first and last. If there are differences between design and analysis model for residual Part Superelement, the differences will show up in the first version.

1 DOUBLE FLYSWATTER MODEL $ ANALYSIS USING PART SUPERELEMENTS JANUARY 13, 2009 MSC Nastran 1/12/09 PAGE 39 S.E. STATICS - MULTIPLE LOADS SUPERELEMENT 4 0

----- COMPARISON BETWEEN INPUT PROPERTY VALUES FROM ANALYSIS AND DESIGN MODELS -----

----------------------------------------------------------------------------------------------------------------------------- PROPERTY PROPERTY PROPERTY ANALYSIS DESIGN LOWER UPPER DIFFERENCE SPAWNING TYPE ID NAME VALUE VALUE BOUND BOUND FLAG FLAG ----------------------------------------------------------------------------------------------------------------------------- PSHELL 4 T 5.000000E-02 5.000000E-02 5.000000E-03 1.000000E+20 NONE

1. THE DIFFERENCE FLAG IS USED TO CHARACTERIZE DIfFERENCES BETWEEN ANALYSIS AND DESIGN MODEL PROPERTIES: IF THE FLAG IS NONE, THEN THERE IS NO SIGNIFICANT DIFFERENCE BETWEEN THE TWO VALUES. IF THE FLAG IS WARNING, THEN THE USER IS ADVISED THAT DIFFERENCES EXIST AND THE DESIGN MODEL IS BEING USED TO OVERRIDE THE ANALYSIS MODEL. IF THE FLAG IS FATAL, THEN THE DIFFERENCES ARE GREATER THAN 1.00000E+35 AND THE RUN WILL BE TERMINATED. 2. THE SPAWNING FLAG (*) INDICATES THAT THE SPAWNED PROPERTY IS DERIVED EITHER FROM THE BEAM CROSS SECTION LIBRARY OR FROM A PBEAM ENTRY. THE PROPERTY ID FOR THE SPAWNED PROPERTY IS IDENTICAL TO ITS PARENT.


48

Updated Bulk Data Entries

The updated bulk data entries are punched into either a single or multiple ‘.pch’ file(s). It can be controlled by ‘PARAM,PSENPCH,YES’as described previously. A sample punched updated bulk data entries for a Part Superelement is shown as follows:

Listing 5-1 Partitioned Superelement punch output for PARAM,PSENPCH,YES

Guidelines and Limitations1. For each Part Superelement, the design model specifying design variables, responses, and

constraints must be located in the corresponding 'BEGIN SUPER=seid' section.

2. DESVAR, DVxRELi, and DRESPi IDS can be reused in superelements - i.e. each Part Superelement may define DESVAR,1 DRESP1,1, etc. Note that DESIGN VARIABLE HISTORY output will not distinguish the SEID, therefore, it is suggested that unique ids be used whenever practical.

3. SEDLINK, SEDRSP2, and SEDRSP3 must be placed in the main bulk data section before 'BEGIN SUPER=seid' (for seid>0).

4. For Part Superelement, DCONSTR entries can reside in each individual Part Superelement Bulk Data Section starting with ‘BEGIN SUPER=seid’. If DCID is different from a Part Superelement to the next, DCONADD in the main Bulk Data Section can be defined to group DCONSTR entries together for reference by DESSUB. Note that DCONADD entries in ‘BEGIN SUPER=seid’ where seid>0 will be ignored.

5. Part Superelement optimization does not support topology (TOPVAR), topography (BEADVAR), or topometry (TOMVAR) optimization.

$ ******************************$ * *$ * PART SE 2 *$ * *$ ******************************$$ *************************************************************$ * *$ * CONTINUOUS DESIGN CYCLE NUMBER = 5 *$ * *$ *************************************************************$$ ******************************$ * *$ * PART SE 2 *$ * *$ ******************************$$ UPDATED DESIGN MODEL DATA ENTRIES$DESVAR * 102T2 1.00000001E-01 1.00000001E-01+D 1V*D 1V 1.00000000E+01$ ******************************$ * *$ * PART SE 2 *$ * *$ ******************************$$ UPDATED ANALYSIS MODEL DATA ENTRIES$PSHELL* 2 2 5.00000035E-03 2** 1.00000000E+00 2 8.33333313E-01 0.00000000E+00** 0 **


6. Design Responses for Image Part Superelements (copies, mirrors, etc. via SEBULK Bulk Data entry) are supported in the design model.

Test CasesThe following test cases are available in TPL library tpl\pse_200:

d200pse1, d200pse2, d200pse3, p200pse6, d200pse7, d200pse8, d200psea and d200pseb

TPL Problem d200pse1.dat

The double-headed fly swatter model will be used to demonstrate Part Superelement Optimization with the design model including DESVAR from each Part.

Figure 5-1 Example problem d200pse1


50

The case control is similar to case control and design model definition is similar to non Part Superelement optimization files. Note that the DESSUB in the case control points to the DCONSTR in each Part Superelement.

Listing 5-2 Partial input for file /pse_200/d200pse7.dat

$ tpl problem d200pse7.dat$ Case control$ANALYSIS = STATICS $DESOBJ(MIN)=1001 DESSUB = 100 $ all DCONSTR with ID = 100 will be considered $

$ tpl problem d200pse7.datBEGIN BULK$ design model for se 0DCONSTR 100 801 -7.00 7.00 DCONSTR 100 802 -7.00 7.00 DESVAR 110 T10 1.0 .1 10.0 DRESP1 801 RESG12Z DISP 3 13DRESP1 802 RESG23Z DISP 3 23DRESP1 1001 WEIGHT WEIGHT DVPREL1 10 PSHELL 10 T .005 110 .05

BEGIN SUPER=1$ design model for SE 1DCONSTR 100 101 -7.00 7.00 DCONSTR 100 102 -7.00 7.00 DESVAR 101 T1 1.0 .1 10.0 DESVAR 1001 GE_1 1. 0.1 10. DRESP1 101 S1G57Z DISP 3 57 DRESP1 102 S1G93Z DISP 3 93 DVMREL1 1001 MAT1 1 GE .005 ++0000031001 .05 DVPREL1 1 PSHELL 1 T .005 ++000001101 .05 $ each BEGIN SUPER has additions to the design model


The output is similar to output from non Part Superelement Optimization runs, as an example, the Design Variable History for d200pse1.dat is:

Listing 5-3 Design Variable History Partitioned Superelements

TPL Problem d200pse7.dat

TPL problem d200pse7.dat provides an example of defining a synthesized response using design components that span multiple Part Superelements. The SEDRSP2 entry is similar to the DRESP2 entry with the exception that an SEID qualifier is required for each design component that is used in the synthesized response.

Listing 5-4 Example of SEDRSP2 in file /pse_200/d200pse7.dat

GUI Support for Part Superelement Optimization

Pre Processing

Patran supports Part Superelement and Optimization, however, all the design data will be written to the main bulk data section. The user will have to manually adjust the location of the design entries. Alternatively, the user can use the Patran group option to write out each Part Superelement to an individual bulk data file and then use INCLUDE files to assemble the final models and design models.

$ tpl problem d200pse7.dat$ design model for se 0 (continued)$ synthesized response across SESEDRSP2 818 AVERD 108 $ SEID DVID SEID DVID DESVAR 2 102 5 105 $ SEID const SEID const DTABLE 3 CONST3 6 CONST6 $ SEID RESPID SEID RESPID SEID RESPID DRESP1 0 801 1 101 3 301 6 601 $ SEID PRELID SEID PRELID DVPREL1 4 4 7 7 $ SEID MRELID DVMREL1 1 1001


52

There is no direct Patran support for the SEDLINK, SEDRSP2, and SEDRSP3 entries, nor PARAM,PSENPCH. These entries could be added via Direct Text Input, or the bdf can be modified before job submission.

SimXpert does not currently support SOL 200 Design Optimization.

Post Processing

With PARAM,POST,-1 the objective, maximum constraint, and design variable history data will be written to the OP2 and available for post-processing by Patran as normal.

SimXpert does not currently support Optimization.


Optimization - Invariant DRESP3 Gradients

IntroductionOne component of the design model in gradient based design optimization is the design response definition. A design response can be used as an objective or a constraint can be placed on the response to bound the optimization space. A simple example of an objective is to minimize the weight. Constraint can be as simple as “von Mises stress must be less than 20,000 psi,” or as complex as a subroutine that calculates multiple margins of safety based on stress, stability, empirical, and manufacturing considerations. In MSC Nastran, there are three types of responses:

1. Direct responses are output quantities generated by a typical MSC Nastran analysis. Examples are Weight, Frequency, Stress, Displacement, etc. Direct responses are specified for the design model by specifying DRESP1Bulk Data entries.

2. Equation based responses can be defined by the user based on current design model values and constants. The equation based responses can be quite extensive and allow the user great flexibility in defining a synthesized response. The DRESP2 is used to define equation based results.

3. External responses are available by calling a user supplied subroutine. This type of response gives the user ultimate flexibility by allowing the designer to write his own subroutines that can include empirical based tables, conditional clauses, loops, etc. in defining the response. The DRESP3is used to define the inputs and external response server used to calculate the responses.

Sensitivity analysis is performed to calculate the gradient of each response with respect to each design variable. Additional information on design optimization in MSC Nastran can be found in the MD Nastran Design Sensitivity and Optimization User’s Guide.

The MSC Nastran 2008 release enhanced DRESP3 by returning multiple responses, providing for analytic gradients, and using a more efficient algorithm when there are many more DRESP1’s in the DRESP3 than there are design variables. The MSC Nastran 2010 release has added a feature that allows the user to specify that gradients of the DRESP3 with respect to the design are to be considered invariant during the approximate optimization task.

BenefitsIn the approximate optimization task, DRESP3 gradients have been calculated using central differencing techniques. For a DRESP3 with 100 arguments, this entails 200 calls to the server for a single gradient calculation. This can be a performance burden if the server call is non-trivial. For this reason, it was decided to implement invariant gradient approach which assumes the gradient is not a strong function of the individual arguments. The approximate optimizer does not have to make any calls to the server.

InputThe format of the DRESP3 Bulk Data entry is unchanged. The user is required to modify the R3SGRT server subroutine that supplies MSC Nastran with the information on the number of responses and the

MSC Nastran 2010 Release GuideOptimization - Invariant DRESP3 Gradients

54

type of gradient evaluation desired. Prior to MSC Nastran 2008, the only gradient option that was supported was finite difference so there was no need for the user to specify the method to be used. MSC Nastran 2008 provided two options for the gradient type and this was specified in the R3SGRT subroutine by specifying a GRDTYP for each response. With MSC Nastran 2010, the number of options is four:

• GRDTYP = 1 specifies that analytic gradients will be provided by the user and they will be computed explicitly during the approximate optimization task

• GRDTYP = 2 specifies that analytic gradients will be provided by the user and they will be considered invariant during the optimization task.

• GRDTYP = 3 specifies that finite difference techniques will be required to compute gradient information and they will be computed explicitly during the approximate optimization task.

• GRDTYP = 4 specifies that finite difference techniques will be required to compute gradient information and they will be considered invariant during the approximate optimization task.

The user provided R3SVALD subroutine that evaluates the DRESP3 response does not need to be modified as a result of this enhancement.

OutputThere is no change to the existing output formats.

ExamplesThere are two tests cases in the TPL subdirectory /tpl/edresp3_08 that demonstrate the invariant dresp3 capability:

dresp3fi.dat – This file calculates all the gradients using finite difference methods and considers the gradients to be invariant during the approximate optimization task (i.e., GRDTYP = 4).

dresp3fo.dat – This file calculates all the gradients using all four methods. The R3SGRT subroutine in this case has the following specification:

Listing 5-5 Updated input for subroutine R3SGRT

PARAMETER(NTYPES=5) CHARACTER*8 R3TYPE(NTYPES)C DATA R3TYPE/'TYPE88 ','TYPE91 ', 'TYPE92 ', 'TYPE93 ', 1 'TYPE94 '/ ERROR = 0 nresp = 1 DO 100 ITYPE = 1, NTYPES IF (TYPNAM .EQ. R3TYPE(1)) THEN grdtyp(1) = 3 go to 200 else IF (TYPNAM .EQ. R3TYPE(2)) THEN grdtyp(1) = 1 go to 200 else IF (TYPNAM .EQ. R3TYPE(3)) THEN


grdtyp(1) = 2 go to 200 else IF (TYPNAM .EQ. R3TYPE(4)) THEN grdtyp(1) = 3 go to 200 else IF (TYPNAM .EQ. R3TYPE(5)) THEN grdtyp(1) = 4 go to 200 else ERROR = BADTYP END IF100 CONTINUE

It is seen that the user is able to specify the gradient type as a function of the response name.

Guidelines and Limitations

Modifying Existing Server Subroutines

The enhanced capability does not require any changes in the MSC Nastran input files that have been developed to utilize the DRESP3, but it does require changes in the R3SGRT server subroutine relative to the MSC Nastran 2008 capability (see the MSC Nastran 2008 Release Guide to see the changes in this subroutine due to the analytic gradient and multiple response enhancements). To retain the current capability for an existing DRESP3, the changes required in the R3SGRT subroutine are to:

1. Change existing GRDTYP(i) = -2 to GRDTYP(i) = 3

2. Change existing GRDTYP(i)= 2 to GRDTYP(i) = 1

No changes are required to the R3SVALD subroutine relative to the MSC Nastran 2008 capability.

Other GuidelinesIn R3SGRT, GRDTYP needs to be defined for all NRESP responses and the values must be either 1, 2, 3 or 4. It is an error if any other value is used.

MSC currently does not have test cases that allow an evaluation of the relative performance of the various gradient options. However, one can make the following recommendations:

1. If the DRESP3 evaluations are cheap and simple, GRDTYP = 1 (analytic, variant) is recommended.

2. If the DRESP3 evaluations are simple, but expensive, GRPTYP = 2 (analytic, invariant) is recommended. It would seem that this option would be rarely needed.

3. If the DRESP3 evaluations are not simple, but still cheap, GRDTYP = 3 (finite difference, variant) is recommended.

4. If the DRESP3 evaluation are complex and expensive, GRDTYP = 4 (finite difference, invariant) is recommended.

MSC Nastran 2010 Release GuideOptimization - Invariant DRESP3 Gradients

56

The invariant gradients will not provide as accurate a calculation as the variable ones so it becomes a trade-off between the time spent performing the optimization and possible more design iterations to reach the final design. It is expected that this new capability will be used sparingly, but will provide a dramatic improvement in CPU time in special circumstances.

The current limitation that all GRDTYP’s for a particular TYPE be the same is retained for this project. The GRDTYP’s do not need to all be the same for all the DRESP3’s in an input file. That is, one can specify analytic variant gradients (GRDTYP = 1) for one TYPE and finite difference, invariant gradients (GRDTYP = 4) for another type.

Additional InformationBuilding and Using DR3SERV (p. 253) in the MD Nastran 2010 Installation and Operations Guide has information on installing and running the DR3SERV, the server associated with the DRESP3.

GUI Support for Invariant DRESP3 Gradients

Pre Processing

Since Invariant DRESP3 Gradients is invoked by modifying the DRESP3 fortran server routine, there is no need for GUI support.

Post Processing

There are no additional post-processing requirements associated for Invariant DRESP3 Gradients.


Design of Monitor Points

IntroductionMonitor points is a generic name for four types of output requests. The MSC Nastran R1 Release Guide provided the most comprehensive discussion of these inputs. Briefly,

1. MONPNT1– The MONPNT1 provides integrated loads at a user defined point in a user defined coordinate system.

2. MONPNT2– The MONPNT2 provides element results (e.g., Stress, Strain, Force)

3. MONPNT3– The MONPNT3 provides a summation of grid point forces at a user specified monito points.

4. MONDSP1– The MONDSP1 allows for the sampling of a displacement vector to create a blended displacement response at a user specified point and coordinate system.

With the release of MSC Nastran 2010, each of the MONPNT1, MONPNT3 and MONDSP1 quantities can now be specified as design response quantities on the DRESP1entry. The MONPNT2 capability for element results effectively duplicates existing DRESP1 response quantities, so MONPNT2 is not supported.

BenefitsThe user of this capability is likely to be an investigator who wants to control a load path in an aeroelastic analysis. This is a sophisticated application that is related to aeroelastic tailoring, implying that it will be done in conjunction with composites. It’s likely that our users will find other applications that are currently undefined.

InputThe existing DRESP1 entry now has 5 additional options. The following is extracted from the description of the entry DRESP1 (p. 1644) in the MD/MSC Nastran Quick Reference Guide:

Response Type ATTA ATTB ATTi

STMONP1 Component (see Remark 36.) Blank Blank

STMOND1 Component (see Remark 36.) Blank Blank

MONPNT3 Component (see Remark 36.) Blank Blank

AEMONP1 Component (see Remark 36.) Blank Blank

AEMOND1 Component (see Remark 36.) Blank Blank

MSC Nastran 2010 Release GuideDesign of Monitor Points

58

36. For monitor point responses (RTYPE = STMONP1, STMOND1, MONPNT3 AEMONP1 or AEMOND1) the ATTA field specifies the components to be extracted. These can be any subset of the integers 1 through 6 that appear on the monitor quantity with the NAME provided in the PTYPE field. All of these responses can be invoked in a static aeroelastic (ANALYSIS=SAERO) subcase. STMONP1, STMOND1 and MONPNT3 can be invoked from a static (ANALYSIS=STAT) subcase. The responses are not available in a dynamic response or normal modes subcase. The response types have the following meaning:

a. STMONP1 – A structural MONPNT1

b. STMOND1 – A structural MONDSP1

c. MONPNT3 – A MONPNT3

d. AEMONP1 – An aerodynamic MONPNT1

e. AEMOND1 – An aerodynamic MONDSP1

For all but the STMONP1, the response is the elastic monitor point value. For the STMONP1, it is the elastic “minus” inertial “plus” elastic applied load value.

OutputThe various response outputs have been modified to clearly identify Monitor Points.


Listing 5-6 Sample Monitor Point Response output identification.

Guidelines and Limitations These new responses are only available for static and/or static aeroelastic subcases (ANALYSIS=STATIC or ANALYSIS=SAERO). The extracted response value is typically the “elastic” value printed in the .f06 file. An exception is the STMONP1. In this case, the response value is the sum of “elastic restrained” minus the “inertial” plus the “restrained applied” values.

----- DESIGN CONSTRAINTS ON RESPONSES -----

(MAXIMUM RESPONSE CONSTRAINTS MARKED WITH **)

--------------------------------------------------------------------------------------------------------- INTERNAL EXTERNAL INTERNAL INTERNAL DCONSTR RESPONSE DRESPx RESPONSE L/U REGION SUBCASE ID ID ID ID TYPE FLAG ID ID VALUE --------------------------------------------------------------------------------------------------------- 1 100 10 3456 STMOND1 UPPER 0 10 2.3412E-01 2 100 11 4567 MONPNT3 LOWER 0 10 -2.7536E-01 3 100 8 5678 GPFORCE LOWER 5678 10 -2.7536E-01 4 100 8 5678 GPFORCE UPPER 5678 10 2.6273E-01** 5 100 9 5679 GPFORCE LOWER 5679 10 5.7648E-02 6 100 9 5679 GPFORCE UPPER 5679 10 -6.6978E-02 7 100 1 6543 EQUA UPPER 6543 10 6.6703E-02

---------------------------------------------------------------------- | R E S P O N S E S IN D E S I G N M O D E L | ----------------------------------------------------------------------

----- MONITOR POINT RESPONSES -----

------------------------------------------------------------------------------------------------------------ INTERNAL DRESP1 RESPONSE NAME COMPONENT LOWER UPPER ID ID LABEL NO. BOUND VALUE BOUND ------------------------------------------------------------------------------------------------------------ 10 3456 STIP SWTIP 3 N/A 3.0853E+00 2.5000E+00 11 4567 SMP3 SPOINT 3 1.0000E+00 1.2754E+00 N/A

**************************************************************************** * * * D E S I G N S E N S I T I V I T Y M A T R I X O U T P U T * * * * * * R E S P O N S E S E N S I T I V I T Y C O E F F I C I E N T S * * * **************************************************************************** ---------------------------------------------------------------------------------------------------------------- DRESP1 ID= 3456 RESPONSE TYPE= STMOND1 NAME = SWTIP COMP NO= 3 SEID= 0 SUBCASE RESP VALUE DESIGN VARIABLE COEFFICIENT DESIGN VARIABLE COEFFICIENT ---------------------------------------------------------------------------------------------------------------- 10 3.0853E+00 1 INBD -8.0550E-01 2 SNDD -6.8049E-01 ---------------------------------------------------------------------------------------------------------------- DRESP1 ID= 4567 RESPONSE TYPE= MONPNT3 NAME = SPOINT COMP NO= 3 SEID= 0 SUBCASE RESP VALUE DESIGN VARIABLE COEFFICIENT DESIGN VARIABLE COEFFICIENT ---------------------------------------------------------------------------------------------------------------- 10 1.2754E+00 1 INBD 3.6495E+00 2 SNDD -3.9502E+00


60

Test CasesThree test cases are provided in the tpl subdirectory /tpl/ue_mdr4 that demonstrate this new capability:

TPL problem dmpa.dat

TPL problem dmpa.dat is a design sensitivity analysis with a single static aeroelastic subcase. The model is a classic test case of a half-span 15 degree swept wing studied in NASA TN D-1824. The boundary conditions are a wind tunnel mount and the ANALYSIS=SAERO is performed at a Mach Number of 0.45. Further description of the model and aeroelastic capabilities of MSC Nastran can be found in the Aeroelastic Analysis User’s Guide, example problem ha144c.dat.

Figure 5-2 Example problem dmpa

The subcase invokes constraints on each type of monitor responses with constraints contrived to force their retention. This test case is used to test the quality of the sensitivities. A part of the bulk data file which contains the monitor points, the design response and their constraints is shown here:

Listing 5-7 Design Response definition for Monitor Points

mondsp1 swtip transverse disp and twist at wing tip 35 swtip 2.515 5.525aecomp swtip set1 840set1 840 8 16 24 32 40 23


mondsp1 awtip pitch and plunge at the wing tip 123456 acaero 2.3 4.7 aecomp acaero aelist 101124aelist 101124 101 thru 124 monpnt3 spoint contains results at a single grid 3 5 6 1.88124 3.1572 0. set1 5 21 set1 6 12 19 set1 50 5 13 21 29 37set1 60 5 12 19 26set1 15 2 10 18 26 34set1 16 2 9 16 23set1 25 1 9 17 25 33 set1 26 1 8 15 22 monpnt1 outbd contains from the three outboard strips 123456 otbd 1.88124 3.1572aecomp otbd set1 1234set1 1234 5 thru 8 13 thru 16 37 21 thru 24 29 thru 32 38 39 40monpnt1 aoutbd contains the three outboard aerodynamic strips 35 aotbd 1.88124 3.1572aecomp aotbd aelist 1234aelist 1234 113 thru 124dresp1 1234 oaero aemonp1 aoutbd 3 dresp1 1235 tiptran aemond1 awtip 3 dresp1 1236 ostru stmonp1 outbd 3 dresp1 3456 stip stmond1 swtip 3 dresp1 4567 smp3 monpnt3 spoint 3dresp1 6789 tipdis disp 3 24 8 16 32 40 23$constr 200 1234 1.0 1.01dconstr 200 1234 1.0 3.0 dconstr 200 1235 1.0 3.0 dconstr 200 1236 1.0 3.0 dconstr 200 3456 1.0 3.0 dconstr 200 4567 1.0 3.0 dconstr 200 6789 1.0 3.0 doptprm desmax 20 p1 1 p2 15 iprint 7 delb .01

It is seen that five monitor points are constructed, one for each of the available monitor types. In this example, only the “TZ” (component 3) of each of the monitor points is being designed even though multiple components are available from monitored quantities. A number of displacement constraints are also specified to provide a qualitative assessment of the MONDSP1 results and their sensitivities.

TPL Problem dmoncants.dat

TPL problem dmoncants.dat is a variation of the TPL problem dmpa.dat and contains both a static aeroelastic and a static subcase. In this example, constraints are placed on monitor point results that require redesign in order to provide an optimal design. The weight is minimized while satisfying all the imposed constraints. The bulk data sample shown in Listing 5-8 indicates that aeroelastic subcase (ANALSYIS=SAERO with DESSUB=200) has placed upper bound limits on structural and aerodynamic MONPNT1 responses, an aerodynamic MONDSP1 and a lower bound on a structural MONDSP1. The static subcase (ANALYSIS=STATIC with DESSUB=100) applies a single, more stringent limit on the structural MONDSP1.


62

Listing 5-8 Partial listing of TPL problem dmoncants.dat

mondsp1 swtip transverse disp and twist at wing tip 35 swtip 2.515 5.525aecomp swtip set1 840set1 840 8 16 24 32 40 23 mondsp1 awtip pitch and plunge at the wing tip 123456 acaero 2.3 4.7 aecomp acaero aelist 101124aelist 101124 101 thru 124 monpnt3 spoint contains results at a single grid 3 5 6 1.88124 3.1572 0. set1 5 21 set1 6 12 19 set1 50 5 13 21 29 37set1 60 5 12 19 26set1 15 2 10 18 26 34set1 16 2 9 16 23set1 25 1 9 17 25 33 set1 26 1 8 15 22 monpnt1 outbd contains from the three outboard strips 123456 otbd 1.88124 3.1572aecomp otbd set1 1234set1 1234 5 thru 8 13 thru 16 37 21 thru 24 29 thru 32 38 39 40monpnt1 aoutbd contains the three outboard aerodynamic strips 35 aotbd 1.88124 3.1572aecomp aotbd aelist 1234aelist 1234 113 thru 124dresp1 1234 oaero aemonp1 aoutbd 3 dresp1 1235 tiptran aemond1 awtip 3 dresp1 1236 ostru stmonp1 outbd 3 dresp1 3456 stip stmond1 swtip 5 dconstr 200 1234 6.5dconstr 200 1235 2.0dconstr 200 1236 6.5dconstr 200 3456 -2.5 dconstr 100 3456 -2.0

TPL Problem dmp3.dat

TPL problem dmp3.dat is another variant of TPL problem dmpa.dat that provides an example of using monpnt3 as a design quantity.

Listing 5-9 Partial listing of TPL problem dmpa.dat

MONPNT3 SPOINT CONTAINS RESULTS AT A SINGLE GRID 3 5 6 1.88124 3.1572 0. DRESP1 4567 SMP3 MONPNT3 SPOINT 3

Reference DocumentsThe MSC Nastran Quick Reference Guide provides a description of the input required for these new responses while the Design Sensitivity and Optimization User’s Guide is a good resource for learning about design optimization in MSC Nastran.


GUI Support for MONPNTi Optimization

Pre Processing

Neither Patran nor SimXpert supports MONPNTi definition or optimization.

MSC Nastran 2010 Release GuideParallel Sensitivities

64

Parallel Sensitivities

IntroductionDesign sensitivity and optimization in MSC Nastran requires sensitivity calculation of the design responses w.r.t. each design variable. Design sensitivity calculations can be a very costly portion of a SOL 200 run for models with large numbers of design variables and large numbers of design responses. The Design Sensitivity and Optimization User’s Guide provides great detail about design sensitivity calculations in MSC Nastran.

In MSC Nastran 2010, design sensitivity calculations have been enhanced to be performed in a distributed parallel (dmp) environment in SOL 200 of MSC Nastran. The parallel implementation divides the sensitivity task across a number of processors so that each processes a subset of the total number of design variables. Following the sensitivity analysis and before optimization, the separate sensitivity data are appended into a global sensitivity set.

BenefitsParallel Sensitivity Analysis is aimed at users who have design optimization tasks that spend significant time in the sensitivity calculation phase. This typically occurs for models with large dof, or there are many (perhaps thousands) of design variables and the adjoint method of sensitivity analysis is either unavailable or still time consuming.

Inputs1. Specify the DOMAINSOLVERExecutive Control statement with the new DSA keyword. For

example:

DOMAINSOLVER DSAOther DOMAINSOLVER options like ACMS, FREQ, and MODES may also be specified or modified along with DSA. For example,

DOMAINSOLVER DSA ACMSNote: DOMAINSOLVER options MODES and FREQ are defaults on with dmp keyword on the Nastran submittal command, but the DSA option must be explicitly specified.

2. Specify the dmp=n keyword on the Nastran submittal command; where n is the number of available processors.

OutputsThere are no new outputs. It should be noted that with dmp=n, by default, only the master processor’s output (f04 and f06) is saved. “slaveout=yes” may be specified on the Nastran submittal command to request the slaves’ output.


Test CasesTPL problem /ugdesopt/dsoug7.dat involves a frequency response sizing optimization problem. As an example, the Executive Control Section is modified as follows:

domainsolver acms dsa SOL 200cend

and the job is submitted with:

nastran dsoug7 dmp=2

The job converges after five design cycles and compares exactly with the serial results demonstrating that the parallel sensitivity calculations provide the same answers.

Listing 5-10 Design Variable History for TPL problem ugdesopt/dsoug7.dat

Testing on larger models has demonstrated that the implementation of parallel sensitivity scales well for statics. Similar performance is expected for other analysis disciplines and on all platforms that support dmp.

Figure 5-3 Example problem dspdsa1 run with MEM=2Gb on IBM Power5

DESIGN VARIABLE HISTORY ---------------------------------------------------------------------------------------------------------------------------------- INTERNAL | EXTERNAL | | DV. ID. | DV. ID. | LABEL | INITIAL : 1 : 2 : 3 : 4 : 5 : ---------------------------------------------------------------------------------------------------------------------------------- 1 | 1 | T1 | 8.0000E-02 : 9.6000E-02 : 9.6301E-02 : 9.9607E-02 : 9.8070E-02 : 1.0022E-01 : 2 | 2 | T2 | 8.0000E-02 : 7.7291E-02 : 8.4733E-02 : 8.3555E-02 : 8.5617E-02 : 8.3472E-02 : 3 | 3 | T3 | 8.0000E-02 : 6.6729E-02 : 6.6797E-02 : 7.0123E-02 : 6.9962E-02 : 6.8825E-02 : 4 | 4 | T4 | 8.0000E-02 : 6.7694E-02 : 5.6433E-02 : 4.8899E-02 : 4.5751E-02 : 4.3705E-02 : 5 | 5 | T5 | 8.0000E-02 : 7.3645E-02 : 6.4645E-02 : 5.6491E-02 : 5.1688E-02 : 4.9985E-02 : 6 | 6 | T6 | 8.0000E-02 : 7.9524E-02 : 7.4968E-02 : 7.1886E-02 : 7.3299E-02 : 7.4792E-02 : 7 | 7 | T7 | 8.0000E-02 : 8.5560E-02 : 8.6633E-02 : 8.6667E-02 : 8.9316E-02 : 9.1213E-02 : 8 | 8 | T8 | 8.0000E-02 : 9.4564E-02 : 1.0857E-01 : 1.1484E-01 : 1.1281E-01 : 1.1295E-01 : 9 | 9 | T9 | 8.0000E-02 : 9.5853E-02 : 1.1502E-01 : 1.3803E-01 : 1.5202E-01 : 1.5574E-01 : 10 | 10 | T10 | 8.0000E-02 : 9.6000E-02 : 1.1416E-01 : 1.3621E-01 : 1.6346E-01 : 1.8296E-01 : *** USER INFORMATION MESSAGE 6464 (DOM12E) RUN TERMINATED DUE TO HARD CONVERGENCE TO AN OPTIMUM AT CYCLE NUMBER = 5.

MSC Nastran 2010 Release GuideParallel Sensitivities

66

GUI Support for Parallel Sensitivity

Pre Processing

Neither Patran nor SimXpert supports the DOMAINSOLVER command for Parallel Sensitivity. Direct text input can be used to specify DOMAINSOLVER DSA and the job can be submitted with the additional command line argument DMP=n.

Post Processing

There are no additional post-processing requirements associated for the parallel sensitivity enhancement.


DTABLE Enhancement for Dynamic Analysis

IntroductionAs discussed in Optimization - Invariant DRESP3 Gradients, 53, one component of the design model in gradient based design optimization is the design response definition. The advanced response definitions are available via the equation based response DRESP2, or the external program response DRESP3. In addition to using design variables and direct responses, the user may want to define constants to use in the advanced response equations. These constants can be defined with a DTABLEentry for subsequent use by the DRESP2 or DRESP3. Additional information on design optimization in MSC Nastran can be found in the MD Nastran Design Sensitivity and Optimization User’s Guide.

The DTABLE enhancement for MSC Nastran 2010 allows the DTABLE entry to reference a TABLEDi entry. In MSC Nastran 2010, when the DTABLE encounters an Integer input for VALUi, it will use the Real values found on TABLEDi. When the corresponding LABLi is used in a synthesized response DRESP2, the data defined on the TABLEDi entries will be used.

BenefitsThe DTABLE enhancement benefits dynamic optimization problems that use the MATCH function by greatly simplifying the input. The integer input for DTABLE in MSC Nastran 2010 reduces the input to 1 TABLEDi, 1 DRESP1 with ATTB field left blank and a DRESP2 with a single LABLi under DTABLE and a single NRid under DRESP1.

Prior to this enhancement, the user would need to define

1. DTABLE with LABLi for each forcing frequency

2. DRESP1 for each forcing frequency

3. A DRESP2 with MATCH function with DTABLE calls out each LABLi and DRESP1 ID involved

For a frequency response SOL 200 job with, say, 50 forcing frequencies, one will need DATBLE with 50 (LABLi,VALUi) pairs, 50 DRESP1 each with a single frequency specified and a DRESP2 which has 50 LABLi under DTABLE flag and 50 NRid under DRESP1 flag. The old-style vs. new-style simplified input will be demonstrated in the sample section.

InputThe DTABLEBulk Data entry now accepts integer input for a VALUi. The integer value invokes a TABLEDi which carries (freq,value) or (time,value) pairs for a number of frequencies or times, respectively. The TABLEDi will be interpolated for frequency or time values that are not explicitly defined in the table.

MSC Nastran 2010 Release GuideDTABLE Enhancement for Dynamic Analysis

68

Output No new output.

Guidelines and Limitations1. Integer input for DTABLE can be utilized with RTYPE of FRxxxx, PSDxxxx, ACxxxx and

Txxxx when the MATCH function selected on DRESP2.

2. For dynamic analysis with DRESP2 selecting MATCH function, spawning of DRESP2 of single frequency (or time) will not be performed. Instead, DRESP2 will include all responses of DRESP1 ID specified.

3. Multiple DTABLE entries are allowed and LABLi on DTABLE must be unique among all DTABLE entries.

Test CasesThere are two tests cases in the TPL subdirectory /tpl/dtabl200 that demonstrate the DTABLE enhancement

TPL Problem d200tbi1.dat

TPL problem d200tbi1.dat is a design optimization job including normal modes (ANALYSIS=MODES) and modal frequency response (ANALYSIS=MFREQ) subcases. For the MFREQ subcase, there are 61 forcing frequencies. To use MATCH function on DRESP2 prior to Nastran 2010, the following table (abridged) shows the input entries involved:

Listing 5-11 Old-style DTABLE input for MATCH function

DTABLE c30 1.e+10 c31 1.e+10 c32 1.e+10 c33 1.e+10 c34 1.e+10 c35 1.e+10 c36 1.e+10 c37 1.e+10 c38 1.e+10 c39 1.e+10 c40 1.e+10 c41 1.e+10 c42 1.e+10 c43 1.e+10 c44 1.e+10 c45 1.e+10 c46 1.e+10 c47 1.e+10 c48 1.e+10 c49 1.e+10 c50 1.e+10 c51 1.e+10 c52 1.e+10 c53 1.e+10 c54 1.e+10 c55 1.e+10 c56 1.e+10 c57 1.e+10 c58 1.e+10 c59 1.e+10 c60 1.e+10 c61 1.e+10 c62 1.e+10 c63 1.e+10 c64 1.e+10 c65 1.e+10 c66 1.e+10 c67 1.e+10 c68 1.e+10 c69 1.e+10 c70 1.e+10 c71 1.e+10 c72 1.e+10 c73 1.e+10 c74 1.e+10 c75 1.e+10 c76 1.e+10 c77 1.e+10 c78 1.e+10 c79 1.e+10 c80 1.e+10 c81 1.e+10 c82 1.e+10 c83 1.e+10 c84 1.e+10 c85 1.e+10

c86 1.e+10 c87 1.e+10 c88 1.e+10 c89 1.e+10 c90 1.e+10

1 2 3 4 5 6 7 8 9 10

DTABLE LABL1 VALU1 LABL2 VALU2 LABL3 VALU3 LABL4 VALU4

LABL5 VALU5 LABL6 VALU6 LABL7 VALU7 LABL8 VALU8

-etc.-


$dresp1,130,f130,frvelo,,,1,30.,701001dresp1,131,f131,frvelo,,,1,31.,701001dresp1,132,f132,frvelo,,,1,32.,701001.. (55 DRESP1 removed).dresp1,188,f188,frvelo,,,1,88.,701001dresp1,189,f189,frvelo,,,1,89.,701001dresp1,190,f190,frvelo,,,1,90.,701001$dresp2,330,f330,MATCH+,dtable,c30,c31,c32,c33,c34,c35,c36,+, ,c37,c38,c39,c40,c41,c42,c43,+, ,c44,c45,c46,c47,c48,c49,c50,+, ,c51,c52,c53,c54,c55,c56,c57,+, ,c58,c59,c60,c61,c62,c63,c64,+, ,c65,c66,c67,c68,c69,c70,c71,+, ,c72,c73,c74,c75,c76,c77,c78,+, ,c79,c80,c81,c82,c83,c84,c85,+, ,c86,c87,c88,c89,c90+,dresp1,130,131,132,133,134,135,136,+, ,137,138,139,140,141,142,143,+, ,144,145,146,147,148,149,150,+, ,151,152,153,154,155,156,157,+, ,158,159,160,161,162,163,164,+, ,165,166,167,168,169,170,171,+, ,172,173,174,175,176,177,178,+, ,179,180,181,182,183,184,185,+, ,186,187,188,189,190

With the DTABLE enhancement, the input for MSC Nastran 2010 is shown as follows

Listing 5-12 MSC Nastran 2010 DTABLE enhanced input for MATCH function

tabled1 530 20. 1.e+10 2000. 1.e+10 endtDTABLE const 1.e+10 const2 530$ dresp1,130,f130,frvelo,,,1, ,701001 $dresp2,330,f330,MATCH+,dtable,const2+,dresp1,130

Note that this example happens to have a constant matching function. Usually, the matching function in dynamic analysis is not constant and TABLED1 entry may have many more physical lines to define the response curve.

GUI Support for DTABLE Enhancement

Pre Processing

Neither Patran nor SimXpert supports the MSC Nastran 2010 DTABLE Enhancement.

Post Processing

There are no additional post-processing requirements associated with the DTABLE Enhancement.

MSC Nastran 2010 Release GuideConstants with DTABLE2

70

Constants with DTABLE2

IntroductionAs discussed in Optimization - Invariant DRESP3 Gradients, 53, one component of the design model in gradient based design optimization is the design response definition. The advanced response definitions are available via the equation based response DRESP2, or the external program response DRESP3. In addition to using design variables and direct responses, the user may want to define constants to use in the advanced response equations. These constants can be defined with a DTABLE entry for subsequent use by the DRESP2 or DRESP3. Additional information on design optimization in MSC Nastran can be found in the MD Nastran Design Sensitivity and Optimization User’s Guide.

Historically, the DTABLE entry was used to associate a real constant to a label for subsequent use in design property relations (DVCREL2, DVMREL2, DVPREL2) or advanced design responses (DRESP2, DRESP3). The DTABLE has a simple input that is a paired label / constant (LABLi/VALUi).

The DTABLE2 Bulk Data entry extends this capability to “lookup” constant values defined on property, connectivity, and material entries.

BenefitsDTABLE2 provides a direct access to fields with real value on property, connection and material entries. This allows the user to change the input file properties without having to redefine all of the values associated to the properties that are defined on a DTABLE entry. The LABLi on DTABLE2 entries can be used interchangeably with LABLi on DTABLE for level 2 property relations: DVCREL2, DVMREL2, and DVPREL2as well as advanced responses: DRESP2and DRESP3.

InputThe new Bulk Data entry for DTABLE2 is:

Format:

Example:

1 2 3 4 5 6 7 8 9 10

DTABLE2 LABL1 PNAME1 PID1 FNAME1 LABL2 PNAME2 PID2 FNAME2

LABL3 PNAME3 PID3 FNAME3

DTABLE2 PTHK10 PSHELL 10 T MAT1E MAT1 38 E

CBARX1 CBAR 3888 X1


OutputNo new output for DTABLE2.

Guidelines and Limitations1. LABLi on DTABLE2 and DTABLE must be unique among all DTABLE and DTABLE2 entries.

2. LABLi on DTABLE2 can be referenced under DTABLE flag of DVxREL2 (where x=P, M or C) /DRESP2/DRESP3.

3. Value for FNAMEi field of PNAMEi Bulk Data entry with the ID of PIDi are taken from analysis model before updating analysis values with the designed values. If the designed value is desired, use DVxREL2 flag on DRESP2 or DRESP3 entries instead.

4. DATBLE2 is accessible from IFPNEW only. IFPNEW can be turned on with ‘NASTRAN SYSTEM(444)=1’.

Test Cases

TPL Problem d200tb2b.dat

TPL problem tpl/dtabl200/d200tb2b.dat is modified to use DTABLE2 for the DRESP2 reference. The label DVP11 is associated to PBAR with ID 11, “A” for area. The PNAME “A” is taken directly from the Bulk Data entry PBAR:

Field Contents

LABLi Label for the constant. (Character)

PNAMEi Property, material or connection bulk data entry name. (Character)

PIDi ID of PNAMEi entry. (Integer > 0)

FNAMEi Field name of PNAMEi. (Character)

1 2 3 4 5 6 7 8 9 10

PBAR PID MID A I1 I2 J NSM

C1 C2 D1 D2 E1 E2 F1 F2

K1 K2 I12

Field Contents

PID Property identification number. (Integer > 0)

MID Material identification number. (Integer > 0)

A Area of bar cross section. (Real; Default = 0.0)

MSC Nastran 2010 Release GuideConstants with DTABLE2

72

Listing 5-13 DTABLE2 example input

$ with DTABLE2 $dtable2 DVP11 pbar 11 A DVP12 PBAR 11 J dvc110 cbar 1 x1 dvc111 cbar 1 x2 dvm113 mat1 1 rho dvp23 pbar 11 i1 dvc21 cbar 1 x3 dvm214 mat1 1 Edresp2 1000 rtest 1010 desvar 1 2 dtable l1 l2 dvp11 dvp12 dvc110 dvc111 dvm113 dvp23 dvc21 dvm214 dresp1 1 dresp2 999

DRESP2,1000 has 10 LABLi including 8 defined via DTABLE2. All 10 LABLi will be considered as constants during design cycles. Note that if these were considered as designed properties instead of design constants, the setup should be as shown in following table for DRESP2,1000.

Listing 5-14 Old-style input for using properties as constants

DVPREL1 1 PBAR 11 A 1 1.0DVPREL1 2 PBAR 11 J .1 12. 1.5 1 1.0DVCREL1 10 CBAR 1 X1 1 1.0DVCREL1 11 CBAR 1 X2 .1 12. 1.5 1 1.0DVMREL1 13 MAT1 1 RHO 1 1.0DVPREL2 3 PBAR 11 I1 100 DESVAR 1 2 DTABLE L1 L2DVCREL2 12 CBAR 1 X3 100 DESVAR 1 2 DTABLE L1 L2DVMREL2 14 MAT1 1 E 200 DESVAR 1DTABLE L1dresp2 1000 rtest 1000 desvar 1 2 dtable l1 l2 dresp1 1 dvprel1 1 2

I1, I2, I12 Area moments of inertia. (Real; I1 > 0.0, I2 > 0.0, I1*I2 > ; Default = 0.0)

J Torsional constant. (Real; Default = for SOL 600 and 0.0 for all other solution sequences)

NSM Nonstructural mass per unit length. (Real)

Ci, Di, Ei, Fi Stress recovery coefficients. (Real; Default = 0.0)

K1, K2 Area factor for shear.(Real or blank)

Field Contents

I122

12--- I1 I2+


dvcrel1 10 11 dvmrel1 13 dvprel2 3 dvcrel2 12 dvmrel2 14 dresp2 999

GUI Support for DTABLE2

Pre Processing

Currently, neither Patran nor SimXpert supports DTABLE2.

Post Processing

There are no additional post-processing requirements associated with DTABLE2.

MSC Nastran 2010 Release GuideNew Optimizer - IPOPT

74

New Optimizer - IPOPT

IntroductionMSC Software conducts surveys of optimizer technologies from industry and academia. This has lead to the integration of optimizer IPOPT. IPOPT implements an interior point line search filter method that aims to find a local optimal solution for large scale nonlinear optimization. It was originally developed by Carnegie Mellon University in 2002 and now is supported by IBM.

Interior point method as one of barrier methods was first proposed in the sixties. Barriers methods are used to transform a “difficult” constrained problem into a sequence of “easy” unconstrained problems. MSCADS SUMT method is one of this classical barrier method and BIGDOT is also based on this approach. Barrier methods were popular during the sixties. However, this classical barrier method has its shortcomings, practitioners of nonlinear programming lost interest and switched to newly emerging, apparently more efficient MMFD and SQP-like methods (MSCADS and DOT) in the mid-seventies and eighties.

In the mid-eighties, a modern interior-point revolution started with the well-known Karmarkar linear programming algorithm which can be interpreted as a barrier method. Since then, interior point algorithms have emerged as one of most important and useful algorithms for mathematical programming. In particular, these interior point methods provide an attractive alternative to active constraint set methods in handling problems with large numbers of design variables and inequality constraints.

IPOPT is a software package for large scale nonlinear optimization. This code has been shown to be capable of handling tens of thousands of design variables. It can be used to solve SOL 200 sizing, shape, topology, topometry, and topography problems. Currently, SOL 200 has two license options, Design Optimization and Topology Optimization (IPOPT requires the Topology Optimization license feature).

In MSC Nastran 2010, IPOPT has been made the default optimizer for topology, topometry, and topography design optimization problems. This is done because testing indicates that IPOPT provides a more robust solution than the MSCAD SUMT method. The choice of optimizers can be made with the OPTCOD feature on the DOPTPRMBulk Data entry, or with MSC Nastran System Cell OPTCOD (413)). See also notes in the .Licensing, 77.

BenefitsIn theory, the interior point method is a very robust algorithm that provides an alternative to SOL 200 existing optimizers, in particular, MSCADS SUMT method. The IPOPT optimizer not only enables performing practical topology, topometry, and topography optimization tasks but can also be used to perform standard shape and sizing optimization for design tasks.

TheoryIn this section, a very brief discussion about the interior point method implemented in IPOPT is presented. More detail can be read on paper “On the implementation of a primal-dual interior point filter

http://en.wikipedia.org/wiki/Carnegie_Mellon_University


line search algorithm for large-scale nonlinear programming”, Mathematical Programming, 106(1):25–57, 2006 by A. Wachter and L. T. Biegler.

To simplify the description of the interior point method, we consider a problem with equality constraints as

where and are the lower and upper bounds on the design variables . is the number of design

variables, and m is the number of equality constraints. The objective function and the inequality constraints are assumed to be twice continuously differentiable. Inequality constraints can be

transformed to equality constraints by introducing slack variables.

In general, gradient-based optimization algorithms have a common strategy as below:

A general optimization algorithm loop

• Start ,

• Evaluate and

• Calculate gradients of and

• Determine a search direction

• Perform a one-dimensional search to find that will minimize subject to the

constraints.

• Set

• Check for convergence. If satisfied, exit. Otherwise repeat the loop

Two critical parts of the optimization task consists of determining a search direction and finding a best one-dimensional search step. The determining a search direction is the most time consuming part and one of major difference between the interior method and SOL 200 other optimization methods.

minimize

subject to (5-1)

f x

Cj X 0= j 1 2 m =

Xi 0 i 1 2 n =

XL

XU X n

f X Cj X

k 0= X X0

=

f X Cj X

f X gj X

dk

*f X dk+

Xk 1+

Xk

= *dk+


76

As a barrier method, the interior point algorithm computes (approximates) solutions for a sequence of barrier problems

for a decreasing sequence of barrier parameters µ converging to zero. Equivalently, this can be interpreted as applying a homotopy method to the primal-dual equations,

(5-3)

with the homotopy parameter which is driven to zero. Here, and correspond to the Lagrangian multipliers for the equality constraints and the bound constraints, respectively. Note, that

Eq. (5-3) for together with “x; are the Karush-Kuhn-Tucker (KKT) conditions for the original problem Eq. (5-2). Those are the first order optimality conditions for Eq. (5-1) if constraint qualifications are satisfied Eq. (5-3).

In order to solve the barrier problem Eq. (5-2) for a given fixed value of the barrier parameter, a

damped Newton's method is applied to the primal-dual Eq. (5-3). Here, a search direction is obtained from solving a symmetric linear system

(5-4)

where Jacobian and denotes the Hessian of the Lagrangian function

The choice of scalars and is discussed in Wachter's paper.

The overall efficiency of the interior point method is dependent on solving a sparse linear system Eq. (5-4).

minimize: (5-2)

subject to:

x f x = xi ln

i 1=

n

–

Cj x 0= j 1 m =

f x c x z–+ 0=

c x 0=

XZe e– 0=

Rm

zRn

0= z 0

j

Wk k wI+ + Ak

AkT cI–

dkx

dk

jxk Akk+

c xk =

Ak c xk = Wk xx2

L xk k zk

L x z f x = c x T z–+

w c

n m+ n m+


InputThere are two ways to select the optimizer IPOPT code. One way is by modifying the Nastran system cell OPTCOD (413))as shown in Table 5-1.

The second way is by a parameter OPTCOD on a DOPTPRM Bulk Data entry that has options shown in Table 5-2.

.LicensingMSC provides two optimization license options:

1. Optimization (license file FEATURE line MD_Optimization)

2. Topology Optimization (license file FEATURE line MD_Topology_Optimization).

The default behavior is as follows:

• If both MD_Optimization and MD_Topology_Optimization licenses are found:

The default behavior is that the optimizer and METHOD will be automatically selected for a better performance based on number of design variables, number of constraints, number of active/violated constraints and computer memory.

Table 5-1 System Cell Summary

System Cell Number

System Cell Name Description and Default Values

413 OPTCOD Specifies which optimization code to be used in SOL 200 (Default = 0, automatic selection for a better performance based on number of design variaables, number of constraints, number of active/violated constraints and computer memory)3 - MSCADS4 - IPOPT Optimizer

Note: Options 1 and 2 are no longer used.

Table 5-2 DOPTPRM Design Optimization Parameters

Name Description, Type and Default Values

OPTCOD OPTCOD (Character; Default= Blank) = Blank (taken from system cell number 413)= “MSCADS”: MSCADS is used= “IPOPT”: IPOPT is used


78

• If MD_Optimization is found but MD_Oopology_Optimization license is NOT found:

The default behavior is that the MSCADS optimizer will be used for models with any sizing, shape design variables, topology, topometry, or topography design variables. The method used in MSCADS is automatically selected.

• If MD_Topology_Optimization license is found but MD_Optimization is NOT found:

The default behavior is that the IPOPT optimizer will be used for models with any sizing, shape design variables, topology, topometry, or topography design variables.

OutputIf default OPTCOD and/or METHOD is used, the program prints injobname.f06 what optimizer and method is used. For example,

***SYSTEM INFORMATIN MESSAGE 6649 (ADS9D)MSCADS METHOD = 1 (MMFD) HAS BEEN SELECTED FOR DESIGN CYCLE= 1.

***SYSTEM INFORMATION MESSAGE 6649 (DOMIP9D)IPOPT HAS BEEN SELECTED FOR DESIGN CYCLE= 1.

There are no outputs that are affected by optimizer selection with the exception that the optimizer output produced using the IPRINT > 0 parameter on the DOPTPRM entry is written to a file “msc_ipopt.out”.

Guidelines and LimitationsThe optimization results from IPOPT are expected to be comparable to those from other optimizers. Numerical results show IPOPT is a robust optimizer. However, unlike MSCADS, IPOPT does not support active constraint sets. Thus, IPOPT may be slow for problems with many constraints, in particular, many constraints are inactive. If this is the case, the OPTCOD parameter can be used to invoke the MSCADS SUMT optimizer for problems with many design variables and many constraints.

ExamplesThere are several example files in the TPL that can be used to demonstrate the new IPOPT performance.


TPL Problem /optim68/icwse01.dat

TPL problem icwse01.dat is and intermediate complexity wing shown in Figure 5-4 was solved using MSCADS and IPOPT.

Figure 5-4 Intermediate Complexity Wing Model

This composite structure is modeled with 62 CQUAD4 elements, 55 CSHEAR elements, 39 CROD elements, and 39 CONM2 elements. Two static load cases are imposed along with an eigenvalue load case. The objective was to minimization. There are 153 sizing design variables and 414 stress and failure index constraint and a lower and upper bounds on the first fundamental frequency. To use IPOPT, a parameter OPTCOD=IPOPT is added on Bulk Data entry DOPTPRM such as

Table 5-3 shows the IPOPT results for this example and all results are comparable.

DOPTPRM APRCOD 2 DESMAX 30 DELP 0.50 DPMIN 0.001delx 0.49 p1 1 p2 12 OPTCOD Ipopt

Table 5-3 Intermediate Complexity Wing Optimization Results

Initial Value Optimum IPOPT Optimum MSCADS (MMFD)

Objective 1.6892+2 1.9277+2 1.9414+2

Max Con 5.8425+2 8.7297-04 1.3169-03

# Cycles 21 20


80

TPL Problem /topography/ip3dbeam.dat

TPL problem ip3dbeam.dat, shown in Figure 5-5, is used to demonstrate IPOPT for topology optimization. The 3D beam is modeled with 48,000 six-sided solid elements (CHEXA). The geometry, mesh, and loads are shown in Figure 5-5. The structural compliance is minimized with a mass target 0.2 (i.e., 80% material savings).

.

Figure 5-5 TPL file /topography/ip3dbeam.dat Beam Finite Element Model

The input data for this example related to topology optimization model is given in Listing 5-5. A Bulk Data entry TOPVAR =1 is used to define a topological design region. Type one design responses DRESP1 = 2 and 10 identify compliance and fractional mass respectively. OPTCOD=IPOPT on the Bulk Data entry DOPTPRM selects the new optimizer IPOPT for solving this optimization problem. SMETHOD= ELEMENT is used to select CASI iterative solver that can provide a major speedup in the solution of large static analyses for solid element models.

Listing 5-15 Partial input for file /topography/ip3dbeam.dat

DESOBJ = 10DESGLB = 1ANALYSIS = STATICSSMETHOD=ELEMENT$ DIRECT TEXT INPUT FOR GLOBAL CASE CONTROL DATA


SUBCASE 1$ SUBCASE NAME : RUN1_LOAD_CASE SUBTITLE=RUN1_LOAD_CASE SPC = 2 LOAD = 3

BEGIN BULKDCONSTR 1 2 .2DOPTPRM, OPTCOD, IPOPTTOPVAR 1 PSOLID PSOLID .2 1DRESP1 2 FRM FRMASSDRESP1 10 COMP COMP

Figure 5-6 shows the topology optimized result that is smoothed and smoothed by using Patran. This optimal design is very clear without any checkerboard effect.

Figure 5-6 TPL file /topography/ip3dbeam.dat Proposed Topology Design Concept

GUI Support for IPOPT

Post Processing

Patran supports IPOPT. There are no additional post-processing requirements associated for IPOPT.

MSC Nastran 2010 Release GuideTopology and Topometry Enhancements

82

Topology and Topometry Enhancements

IntroductionTopometry optimization is a special form of sizing optimization most commonly used for shell elements whereby each shell element in a property region is allowed to change thickness independently of its neighboring element. This provides the user an insight to the optimum material distribution to achieve the objective while satisfying constraints. Additional element types commonly supported by topometry optimization are non-volume elements such as CWELD, CBUSH, and CFAST.

Topology optimization adjusts each elements’ “effectivity” by adjusting the modulus of Elasticity and density to determine a optimum material distribution. Topology optimization can be applied to solid, shell, and beam elements.

The two methods are describe in further detail in the MD Nastran Design Sensitivity and Optimization User’s Guide.

MSC Nastran 2010 contains a few enhancements for SOL 200 topometry and topology optimization capabilities. The enhancements include

1. density constraint method for topology minimum member size control

2. composite (PCOMP) topometry optimization

3. discrete topometry optimization.

4. enhanced TOPVAR entry and casting constraints

Benefits

Topology Optimization Density Constraint Method

A density constraint method is implemented for topology minimum member size control. This approach is more efficient than the filtering method for topology problems with a very fine mesh and a relatively large predefined minimum member size.

Composite (PCOMP) Topometry Optimization

This new feature enables SOL 200 to support composite ply-by-ply thickness optimization. The ply-by-ply means each ply thickness (or orientation angle) per composite element is treated as an independent design variable. The ply thickness can be linked together to support element-by-element thickness optimization. Although topometry optimization is not recommended for topology optimization tasks, it is observed topometry optimization can be used to get “similar topological results” for many cases. Since SOL 200 topology optimization does not support PCOMP entries, this composite TOMVAR can be used to decide which composite element should be retained and which composite element should be discarded from the design space.


Discrete Topometry Optimization

The discrete optimization capability is expanded to support topometry optimization. This capability enhances the simplicity of the design and hence its manufacturability.

Enhanced TOPVAR entry and casting constraints

Topology initial value XINIT has a default in MSC Nastran 2010. It is easier and recommended that the user use the default value of XINIT. Aligned mesh option is added to topology casting and extrusion constraints. When aligned mesh is used for topology designed properties with casting/extrusion constraints, a smaller tolerance is used to process casting constraints during optimization. Thus, a better/sharper topology design proposal may be produced. In addition, topology casting constraint capabilities are significantly enhanced in MSC Nastran 2010.

Input for Topometry EnhancementsA review of the topometry and topology optimization section in the MSC Nastran 2008 and the MSC Nastran 2005 r2 Release Guide is recommended if you are new to MSC.Nastran’s implementation of the topometry and topology technology.

The enhanced topometry TOMVAR Bulk Data entry format is:

Input for Minimum Member Size EnhancementsThe density constraint approach can be selected by a parameter TCHECK on Bulk Data entry DOPTPRM.

1 2 3 4 5 6 7 8 9 10

TOMVAR ID TYPE PID PNAME/FID

XINIT XLB XUB DELXV

“DLINK” TID C0 C1

“DDVAL” DSVID

Field Contents

TYPE Property entry type. Used with PID to identify the elements to be designed. (Character: “PBAR”, “PSHELL”, “PSOLID”, and “PCOMP”, etc.)

“DDVAL” Indicates that this line defines discrete TOMVAR variables

DSVID DDVAL entry identifier (Integer > 0)

“DLINK” Indicates that this line relates a ply thickness to another ply thickness

TID TOMVAR entry identifier (Integer > 0).

C0 Constant term (Real; Default = 0.0)

C1 Coefficient term (Real; no Default)


84

DOPTPRM Design Optimization Parameter: TCHECK

Input for Topology EnhancementsThe enhanced topology TOPVAR Bulk Data entry format is:

Format:

OutputThe only change in output for these features is for the Composite Topometry optimization. In this case, a element result file jobname.plyxxx (where xxx is a PCOMP ply identifier) contains the optimal design values for each composite ply. The element result file can be imported into Patran or third party post-processor to display composite topometry optimization results.

Name Description, Type and Default Value

TCHECK Topology Checkerboarding/minimum member size control option. (Integer > -1)

-1 Automatic selection of filtering or density constraint for a better result.

1 Filtering algorithm (Default)

2 Density constraint

0 No control

1 2 3 4 5 6 7 8 9 10

TOPVAR ID LABEL PTYPE XINIT XLB DELXV POWER PID

“SYM” CID MSi MSi MSi CS NCS

“CAST” CID DDi DIE ALIGN

“EXT” CID EDi ALIGN

“TDMIN” TV

Field Contents

XINIT Initial value. (Blank or Real, XLB < XINIT < 1.0 Default=blank). Typically, XINIT is defined to match the mass constraint on DRESP1=FRMASS, so the initial design does not have violated constraints. In this case, the default is set to the constraint value. If the mass (DRESP1=FRMASS or WEIGHT) is the objective, the default is 0.9. The default of XINIT is 0.6 for the other cases.

ALIGN Indicates whether the designed property finite element mesh is precisely aligned with the draw direction or extrusion direction. (Character: “YES” or “NO” or Blank; Default = blank = “NO”)


Guidelines and Limitations• DISCOD = 3 and 4 (discrete processing method) is recommended for discrete topometry

optimization tasks since topometry Optimization usually involves many design variables and DISCOD = 4 is the fast discrete processing method.

• The density constraint approach is more efficient. However, this approach may result in more unexpected intermediate density elements than the filtering approach.

Examples

TPL Problem /topography/tomex5.dat

TPL problem /topography/tomex5.dat is a 2D Composite Plate example intended to demonstrate a ply-by-ply thickness optimization using the TOMVAR entry. This composite plate has 640 CQUAD4 element as shown in Figure 5-7. The ply layup is symmetric: 0°, 90°, 45°, -45°,-45°,45°,90°,0°. The objective is to minimize structural compliance and lower/upper bounds are applied on each ply thickness. since the composite is modeled with the “SYM” option, there are 4 independent design variables. The problem is treated as a planar problem an dofs 3456 are permanently constrained on a GRDSET entry. The input data for this example pertinent to the composite lay up and topometry optimization model is given in Listing 5-16. The TOMVAR Bulk Data entries 1-4 define the ply-by-ply thickness optimization. It is noticed that all four ply thickness per element are independent variables. Thus, there are 640x4 independent design variables.

Figure 5-7 A Composite Plate example tomex5.dat

Listing 5-16 Partial Input File for tomex5.dat

$COMPOSITE TOPOMETRY OPT EXAMPLE DESOBJ = 10


86

ANALYSIS = STATICS$ DIRECT TEXT INPUT FOR GLOBAL CASE CONTROL DATASUBCASE 1$ SUBCASE NAME : RUN1_LOAD_CASE SUBTITLE=RUN1_LOAD_CASE SPC = 2 LOAD = 3

BEGIN BULKPCOMP 1 -.0105 0.0 0.65E6 TSAI SYM 70 1.000 0.0 YES 70 1.000 90. YES 70 1.000 45. YES 70 1.000 -45. YESDOPTPRM, OPTCOD, IPOPTDRESP1 10 COMP COMP$...DESIGN TOPOMETRY DESIGN DEFINITIONTOMVAR, 1 , PCOMP, 1, T1 , .5, 1.25-3, 1.0TOMVAR, 2 , PCOMP, 1, T2 , .5, 1.25-3, 1.0TOMVAR, 3 , PCOMP, 1, T3 , .5, 1.25-3, 1.0TOMVAR, 4 , PCOMP, 1, T4 , .5, 1.25-3, 1.0

In addition to the standard .f04, .f06 and .pch output files, the final ply thickness distributions are contained in files tomex5.ply0001, tomex5.ply0002, tomex5.ply0003, tomex5.ply0004. To post process these in Patran, they must be read from the tools menu, and then the results can be displayed using standard Patran fringe plots.

Figure 5-8 Importing ply topometry results in Patran


Figure 5-9 through Figure 5-12 show the optimized ply thickness distribution for all elements.

Figure 5-9 Ply 1 Thickness Distribution of 0° plies



88


Figure 5-12 Ply 4 Thickness Distribution -45° plies

The DLINK feature can be used to relate one ply thickness to another ply thickness in order to support composite element-by-element thickness optimization. The input data for this example is given in Listing 5-17. The DLINK line is used to explicitly link the thickness of plies 2, 3, and 4 to ply 1. Thus, each composite element has only one independent design variables.


Listing 5-17 Input File for tomex6.dat with DLINK

Figure 5-13 shows the optimized element-by-element thickness distribution. Note that only 1 ply output file is generated: tomex6.ply0001.

Figure 5-13 Composite Element Combined Thickness Distribution

GUI Support for PCOMP Topometry

Pre Processing

Patran supports topometry optimization, but the DLINK,DDVAL options are not yet supported. The current “Quick” topology, topometry, topography optimization setup can be read in the Patran 2008R2 Release Guide.

$...DESIGN TOPOMETRY DESIGN DEFINITIONTOMVAR, 1 , PCOMP, 1, T1 , .5, 1.25-3, 1.0TOMVAR, 2 , PCOMP, 1, T2 , .5, 1.25-3, 1.0 , DLINK, 1, 0.0, 1.0TOMVAR, 3 , PCOMP, 1, T3 , .5, 1.25-3, 1.0 , DLINK, 1, 0.0, 1.0TOMVAR, 4 , PCOMP, 1, T4 , .5, 1.25-3, 1.0 , DLINK, 1, 0.0, 1.0


90

Post Processing

Patran supports post processing of PCOMP Topometry enhancements as described previously in Output, 84.


Build External Servers Using the SCons Tool

IntroductionMSC Nastran provides the external server capability (beam library, DRESP3 and Spline servers) to allow the client to create their custom applications without modifying the Nastran program. Prior to MSC Nastran 20101, the binary server executables are built using make utilities. Starting this release, they will be built using the SCons tool.

OverviewSCons is a Software Construction tool. Its major benefits are given below. It is expected that SCons tool provide same or better experience to build an external server.

1. Automatically to resolve source code dependencies;

2. Fairly easy to build binary programs on various platforms with different operating systems. It is particular true for building external server programs on Windows;

3. Easy to develop applications that use combined Python scripts and other source codes written in C or Fortran because SCons configuration files are written in Python scripts;

4. Easy to extend the server template codes by simply adding your source codes to the target directory without the need to modify SConscript.

Although DRESP3 server is used to describe the procedures to build an external server, they can be applied directly to beam library and spline servers.

The Installation Directory for External Server programsThe location of a server directory has been changed and is placed under install_dir/msc20101/nast/ directory. Three external servers are placed in three separate directories, respectively.

The root directory for beam library server: install_dir/msc20101/nast/beamlib

The root directory for dresp3 server: install_dir/msc20101/nast/dr3

The root directory for spline server: install_dir/msc20101/nast/spline_server

The DRESP3 server directory can be located in install_dir/msc20101/nast/dr3 on UNIX and install_dir\msc20101\nast\dr3 on Windows. Its structure is shown in the figure below that is borrowed from Directory for dr3 server (p. 11) in the MD Nastran 2010 Installation and Operations Guide. It contains three SCons construction files and 3 subdirectories: include, lib and src. Notice that the library subdirectory contains a set of predefined libraries and object files that architecture dependent. The src directory contains the source codes that are used together with the library and object files to build server programs.

MSC Nastran 2010 Release GuideBuild External Servers Using the SCons Tool

92

The SCons tool requires a subdirectory be created for each server program. For example, directory dr3serv is created for server dr3serv that contains a SConscript file and Fortran files r3sgrt, r3svald, r3svals. It is convenient to use the program name as the directory name but is not required.

Data Structure of DRESP3 Server ---- dr3 (root) | --- SConopts --- SConscript --- SConstruct --- Include |-- ftncalls.h |-- stdmsc.h |-- stdsystm.h --- lib | ----- < ARCH 1> | ----- linux64 | -- libdr3srv.a, libdr3main.a | -- cnxx.o | -- initgmsrvcmns.o | -- main.o | -- semd.o ----- win64 | -- dr3srv.lib, dr3main.lib | -- cnxx.obj | -- getlserm.obj | -- initgmsrvcmns.obj | -- main.obj | -- semd.obj ------ <ARCH i> ---- src | -- SConscript | -- dr3serv | | -- SConscript | | -- r3sgrt.F | | -- r3svald.F | | -- r3svals.F | -- dr3serva | |… | -- other sample dresp3 server directories

Build a DRESP3 ServerThe simplest way to build a server is to enter the command from the install_dir/msc20101/nast/dr3 directory

scons dr3serv

where dr3serv is the program name you want to build. The command creates dr3serv on UNIX or dr3serv.exe on Windows. By default, the command saves the program in the directory dr3/Apps/arch/bin


that is architecture dependent. For example, arch=LX8664 indicates Linux 64 bit machine or arch=WIN8664 a Window 64 bit machine.

If you do not have the write privilege to the install_dir, you have two options:

1. Define APPS_LOCAL and SCA_OBJECT construction variables to redirect the output to another writable location. To learn more about the SCons build environment, please consult Subsection of 'Build External Servers using SCons tools' Under the External Response Section in the DS&O's User Guide.

2. Copy the entire dr3 directory to another location that you have the write access to. This option is particularly useful when you want to create your server program in a new subdirectory.

Guidelines to Build a DRESP3 Server

Building an external server requires your computer to have Software Development Kit installation (SDK) and MSC Nastran installation with the external server option.

To create a new server program, you may either work in a new subdirectory under the src to develop required SConscript and Fortran template files or work in an existing subdirectory to modify the required files.

If your server program requires additional source codes, you can simply include them in the target directory without the need change in SConscript. However, if you want to add your custom library or object file, you need to place them in the lib/ directory and to update the SConscript file to reference the library and/or object files.

To learn more about customizing the SCons build environment and advanced build scenarios, please see Build External Servers using SCons Tool (p. 294) in the Design Sensitivity and Optimization User’s Guide or SCASCons Build System (p. 403) in the SCA Framework User’s Guide.

Using Server Program

Consult the External Responses (p. 191) in the Design Sensitivity and Optimization User’s Guide for more information on how to use the server program in a Nastran job.

MSC Nastran 2010 Release GuideDeactivation of Original Design Sensitivity (DSA)

94

Deactivation of Original Design Sensitivity (DSA)MSC deployed a capability to provide design sensitivities for SOLs 101, 103 and 105 results over 20 years ago. This capability was characterized by a DVAR entry that defined design variables, an associated DVSET that defined the properties that could be varied and a DSCONS entry to define the response and the constraint limits. This capability has been absent from the Quick Reference Guide since the MSC Nastran 2004 release but has still been available in the code. With the release of MSC Nastran 2010, this capability is no longer available. The extensive Design Sensitivity and Optimization capability that is contained in MD/MSC Nastran and that is documented in the MSC Nastran 2010/MSC Nastran 2010 Design Sensitivity and Optimization User’s Guide provides all the functionality of the Original Design Sensitivity Analysis plus many other features as detailed in this document and the User’s Guide.

Chapter 6: Aeroelasticity MSC Nastran 2010 Release Guide

6 Aeroelasticity

Input of Pressures on an Aerodynamic Mesh

Aeroelasticity - Output of Trimmed Loads

CSV Output of Trim Results

SUBCOM/SUBSEQ with Static Aeroelasticity

Upper Hessenberg Complex Eigenanalysis No Longer Supported for Flutter Analysis

MSC Nastran 2010 Release GuideAeroelasticity

96

Input of Pressures on an Aerodynamic Mesh

IntroductionIn MSC Nastran 2010 the user can apply pressures at aerodynamic grid (AEGRID) points. Previously, this was not possible.

BenefitsIt is now possible to input pressures that come from an external source, such as CFD analyses or wind tunnel tests, onto an aerodynamic mesh and Nastran will transform these pressures to forces that can be included in a static aeroelastic analysis.

Feature DescriptionA new Nastran module has been provided that combines information input on the AEPRESS/DMIJ entries with geometry provided on AEGRID/AEQUAD4/AETRIA3 entries and coordinate systems data to produce forces at the aerodynamic mesh points. Spline input can then be used to transfer these forces to structural grid points so that they can be included in a standard SOL 144 static aeroelastic analysis. No new input commands or entries are required to invoke this capability.

Guidelines and Limitations1. Pressure is input at the grids of the aerodynamic mesh and is converted to forces using the same

techniques as are applied with a PLOAD4 entry. That is, the forces are computed using a combination of connected elements (AEQUAD4 and/or AETRIA3 in this case) to define the area over which the pressure is acting in conjunction with systems to determine the direction in which the forces are acting. The forces are computed at the AEGRID locations.

2. Pressure can be input in all three directions at the aerodynamic grid and they do not need to act in a direction normal to the surface. The aerodynamic coordinate system defined by the AEROS Bulk Data entry defines the directions in which the pressures act.

3. There is no aeroelastic effect associated with the AEPRESS input in this case. Only rigid aerodynamic loads are produced. In the example given above, deformations are produced at the structural grid points but there is no aerodynamic correction due to these deformations.

4. It should be apparent that an application of this technique is to input results from a CFD analysis into a Nastran static aeroelastic analysis. To do this, the user needs to define the aerodynamic mesh using AEGRID/AEQUAD4/AETRIA3, the pressure vector using a combination of AEPRESS/UXVEC and DMIJ entries and the appropriate spline methods.

5. The AEPRESS data can be used in combination with AEFORCE/AEDW input to directly input aerodynamic forces or downwashes on the same aerodynamic mesh.

97CHAPTER 6Aeroelasticity

6. It is not possible to combine aerodynamic mesh input with Doublet Lattice like aerodynamics in a single run. However, it is possible to have separate meshes using the “RIGID/Flexible Mesh” concepts introduced in MSC Nastran 2005 R3. This entails making an initial run with only the rigid aerodynamic mesh and then, in a subsequent run, attaching the database from this run and using the AERCONFIG case control command to identify the rigid mesh results in a static aeroelastic analysis that would obtain its flexible increments from a doublet lattice based analysis.

7. Currently, this technology is only available in static aeroelasticity (SOL 144 or the ANALYSIS=SAERO option in SOL 200). Support for SOL 146 (dynamic response) is not provided at this time.

Test CasesThe following test cases are available on the TPL in subdirectory /tpl/aero_asm: cyl144b.dat

TPL Problem cyl144b.dat

TPL file cyl144b.dat provides an example of the loads on a cylinder spinning in an airstream. Textbooks on introductory fluid mechanics such as Chapter 4.8 of Reference 1. provide a closed form solution for this airflow.

The aerodynamic mesh is depicted in Figure 6-1. The cylinder has a radius of 1.0 and a length of 10.0. Length units are in inches. Data are input at angles of 22.5 degrees at five spanwise stations ranging from -5.0 to 5.0.

The structural model consists of 10 CBAR elements that reference a PBARL entry that creates TUBE cross sections that have an outer radius of 1.0 and an inner radius of 0.95. From the reference, pressure per unit dynamic pressure in a uniform stream can be expressed as:

where is the rotation speed in radians/sec. is the radius of the cylinder and is the freestream

velocity. The pressure acts radially while Nastran requires input in rectangular coordinates at the grid points. With a coordinate system that has positive x in the direction of the flow and the z axis normal to the flow, the radial pressure at each point has components in the x and z directions. For irrotational flow, the net force in the streamwise direction is zero so that it is only necessary to input the z component of the pressure. Note that 3D effects of the flow are neglected so there is no spanwise component.

p q 4sin2= 2 r V sin r V 2+ +

r V


98

Figure 6-1 The Aerodynamic and Structural Model for the Flow Around a Spinning Cylinder.

A section of the pressure input is given shown in Figure 6-2. The AEPRESS entry identifies a UXVEC entry that indicates the state of the pressure input. In this case, the state includes the intercept and a CIRC value of 0.1, where CIRC is the parameter. The AEPRESS data at the UXVEC condition is input

as a vector using the DMIJ format that lists the grid and component for each real number that provides the pressure at the mesh point. Since the three dimensional aspects of the flow are ignored in this analysis, the same pressure distribution in input at the 5 spanwise cuts.

r V


Figure 6-2 A Portion of the Input Bulk Data that Provides Pressure Data on the Aerodynamic Mesh

Figure 6-3 provides an additional snippet with splining and boundary condition information. The SPLINE7 is a 6 DOF spline that uses a finite beam technique to connect all the aerodynamic mesh points to the structural grids that are defined along the axis of the tube. The GRDSET entry constrains the motion in the 1256 directions at all the structural grid points while the SPC condition invokes symmetry about the x-axis and the SUPORT entry allows the tube to move in the 3 direction.

Figure 6-3 Spline and Boundary Condition Input for the Spinning Cylinder

The goal of the analysis is to determine the rate of rotation necessary to create a lift force that is equal to the weight (URDD3 = 1.0) of the aluminum tube when immersed in an airstream moving at M = .45 (Velocity = 6031.8 inches/sec) at sea level. The .f06 file indicates that the value of the CIRC parameter

at “trim” is 1.214E-3. The rotational rate is then cycles/second.

References1. Kuethe, A.M and Schetzer, J.D., Foundations of Aerodynamics, John Wiley & Sons, New York,

Second Edition, 1959.

aepress 0.45 asymm asymm 102 circ1uxvec 102 intercpt 1.0 circ 0.1 dmij circ1 0 9 1 0 0 1DMIj circ1 1 1 1 3 .0 2 3 .2866 3 3 1.6213 4 3 3.505 5 3 4.41 6 3 3.505 7 3 1.6213 8 3 .2866 9 3 0.0 10 3 -.1694 11 3 -1.2213 12 3 -2.8221 13 3 -3.61 14 3 -2.8221 15 3 -1.2213 16 3 -.16942 101 3 0.0

$ $$ * 6DOF FINITE BEAM SPLINE * $$ $$ EID CAERO BOX1 BOX2 SETG DZaelist 101 1 thru 16 101 thru 116 401 201 thru 216 301 thru 316 402 403 404 thru 416set1 101 1001 thru 1011SPLINE7 100 1 101 101 1.0 both spc 1 1006 4suport 1006 3 grdset 1256

1.214E-3 6031.8 2 1.16=


100

GUI SupportPatran Flight Loads -

SimXpert does not currently support aeroelastic analysis.


Aeroelasticity - Output of Trimmed Loads

IntroductionMSC Nastran now supports the creation of bulk data force/moment entries for trim loads.

BenefitsA primary reason for performing a static aeroelastic analysis is to determine the aeroelastic loads acting on the free-flying vehicle. With this new capability it is now possible to output the loads from the trim solution in the familiar FORCE/MOMENT Bulk Data entry format so that these loads could be passed to the group performing detailed stress analysis. In another application, these loads could be viewed in a graphical fashion using, for example, MD Patran.

Input

The TRIMF Case Control command is used to invoke this new capability. The user can provide an ASSIGN statement to direct the results to a special purpose file, or they will go to the .pch file by default. The TRIMF command is quite flexible in that it can output load components or total loads. It is also possible to limit the output to a set of user defined grid points.

TRIMF Format:

Example:TRIMF(LOADSET=10001,LARGE)=ALLTRIMF(UNIT=59,INERTIA,NOSUM)=1

Describer Meaning

UNIT Fortran unit to which data are written. (Optional; Default = 7) (punch file).

LOADSET Load set id for output bulk data entries. If the TRIMF specification results in multiple load sets, then the defined ID will be used for the first and each subsequent load set has an ID incremented by 1. (Optional; Default = 1)

LARGE Write the output data in large field format (16 characters per field). The default is 8 characters per field.

INERTIA Write out inertial loads as a separate load set. By default, the separate load set will not be written.

TRIMF UNIT i= LOADSET n= LARGE INERTIA APPLIED AIR

NOSUM RIGID NOELASTIC QNORM , ALL

n

=


102

Remark:

1. By default, the loads are written to the punch file (Fortran unit 7). If the user specifies an alternate Fortran unit number on the TRIMF entry, by default the loads will be written to a file name that is machine specific (i.e. ‘fort.53’ on many UNIX platforms). The user may connect the Fortran unit to a user-defined file name by using an ASSIGN entry in the FMS Section of the input file. For example:

ASSIGN USERFILE='load13.inc',STATUS=UNKNOWN,FORMATTED,UNIT=532. Up to eight loads sets are available: Rigid Inertial, Rigid Applied, Rigid Air, Rigid Sum and four

more with the sum of the rigid and elastic increment.

3. Care must be taken if LOADSET is specified in a run with multiple subcases. There are no checks that the load set IDs which are generated by one subcase are not also used for another subcase. For example, consider the following Case Control commands:

SUBCASE 1TRIM = 1TRIMF(RIGID) = ALL $

SUBCASE 2TRIM = 2TRIMF(LOADSET=2) = ALL

Subcase 1 will generate two load sets with set IDs 1 and 2. Subcase 2 will also output a load set ID 2.

4. The LOADSET option should not be specified in a TRIMF entry that is located above the subcase level. For example, consider the following:

APPLIED Write out applied loads as a separate load set. By default, the separate load set will not be written.

AIR Write out aerodynamic loads as a separate load set. By default, the separate load set will not be written.

NOSUM By default, the sum of the inertial, applied, and aerodynamic loads will be written as a separate load set. This option suppresses the writing of that set of loads.

RIGID Write out rigid instances of the selected loads (Inertial, Applied, Air and/or Sum) as separate load sets. By default, the separate load set will not be written.

NOELASTIC By default, the sum of the rigid and elastic increment loads will be written as a separate load set. This option suppresses the writing of that set of loads.

QNORM Normalize the load by the dynamic pressure used in the trim analysis. By default, the loads are not normalized.

ALL Loads for all points will be output

n Set identification of a previously appearing SET command. Only loads on points with identification numbers that appear on this SET command will be output (Integer > 0)

Describer Meaning


TRIMF(LOADSET=101) = ALLSUBCASE 1

TRIM = 1SUBCASE 2

TRIM = 2

Here, both subcases will output trim loads with load set ID = 101.

OutputThe output of TRIMF are FORCE and MOMENT bulk data entries. The output is to either the punch file or user defined file. See test case below for sample output.

Guidelines and Limitations The TRIMF request can be placed above all subcases, in which case it is in effect until overwritten or at the subcase level to make a subcase dependent request.

Test CasesThe following test cases are available in the TPL in directory /tpl/ue6_09a: trimf.dat

TPL file trimf.dat is a variation of the ha144e.dat test file documented in the MSC Nastran Aeroelastic User’s Guide. The variation is to add the following trimf commands in two of the five subcases:

assign userfile='abrupt.inc', status=unknown,formatted,unit=53

...

...SUBCASE 1 TRIM = 1 $ HIGH SPEED LEVEL FLIGHTSUBCASE 2 TRIM = 2 $ HIGH SPEED ROLLING PULLOUTSUBCASE 3 trimf(unit=53,loadset=1,rigid)=all TRIM = 3 $ HIGH SPEED PULLOUT WITH ABRUPT ROLLSUBCASE 4 trimf(loadset=10,inertia,air,rigid) = 100 TRIM = 4 $ HIGH SPEED SNAP-ROLL ENTRYSUBCASE 5TRIM = 5 $ HIGH SPEED CLIMBING TURN

The test case in contrived to test a number of the features of the new capability. For example, the assign statement is used to capture the output from the third subcase while the data from the fourth subcase goes to the .pch file. It is seen that the output only goes to a set of grids in the fourth subcase. This feature could be used, for example, if only the loads on the wing are of interest.


104

Subcase 3 output is to the user defined file “abrupt.inc”-

$...............................................................................$ $ TRIM CASE: 3 $ RIGID (AERODYNAMIC + APPLIED - INERTIAL) LOADS $ FORCE 1 97 0 1.0 0.0 0.0 -14400.FORCE 1 98 0 1.0 0.0 0.0 6408.71FORCE 1 99 0 1.0 0.0 -5.97767-16280.6FORCE 1 100 0 1.0 0.0 -849.963-17073.7FORCE 1 111 0 1.0 0.0 0.0 7102.12FORCE 1 112 0 1.0 0.0 0.0 -9393.04FORCE 1 121 0 1.0 0.0 0.0 35721.3FORCE 1 122 0 1.0 0.0 0.0 -33645.6FORCE 1 211 0 1.0 0.0 0.0 6736.25FORCE 1 212 0 1.0 0.0 0.0 -2391.09FORCE 1 221 0 1.0 0.0 0.0 -20757.3FORCE 1 222 0 1.0 0.0 0.0 28125.5FORCE 1 311 0 1.0 -1.84-11-832.2 -288.FORCE 1 312 0 1.0 -1.23-11 272.795-192.MOMENT 1 98 0 1.0 -645.697 0.0 0.0MOMENT 1 99 0 1.0 -645.697 0.0 0.0$...............................................................................$ $ TRIM CASE: 3 $ ELASTIC (AERODYNAMIC + APPLIED - INERTIAL) LOADS $ FORCE 2 97 0 1.0 0.0 0.0 -14400.FORCE 2 98 0 1.0 0.0 0.0 6958.1FORCE 2 99 0 1.0 0.0 95.7669-14832.1FORCE 2 100 0 1.0 0.0 -762.642-16588.3FORCE 2 111 0 1.0 0.0 0.0 16858.9FORCE 2 112 0 1.0 0.0 0.0 -10494.FORCE 2 121 0 1.0 0.0 0.0 46718.8FORCE 2 122 0 1.0 0.0 0.0 -34359.8FORCE 2 211 0 1.0 0.0 0.0 11975.7FORCE 2 212 0 1.0 0.0 0.0 -2332.64FORCE 2 221 0 1.0 0.0 0.0 -17538.FORCE 2 222 0 1.0 0.0 0.0 28513.3FORCE 2 311 0 1.0 -1.84-11 526.924-288.FORCE 2 312 0 1.0 -1.23-11 139.951-192.MOMENT 2 98 0 1.0 -515.146 0.0 0.0MOMENT 2 99 0 1.0 -515.146 0.0 0.0

Subcase 4 output goes to the default unit (punch file) -

$...............................................................................$ $ TRIM CASE: 4 $ RIGID INERTIAL LOADS $ FORCE 10 100 0 1.0 -1.65-12-33147.4-21.0849FORCE 10 111 0 1.0 1584.39-4922.53-298.03FORCE 10 112 0 1.0 1056.26-4337.95 954.982FORCE 10 121 0 1.0 4753.18-3093.03 844.847FORCE 10 122 0 1.0 3168.79-3118.28 1716.9$...............................................................................$ $ TRIM CASE: 4 $ ELASTIC INERTIAL LOADS $ FORCE 11 100 0 1.0 -1.65-12-32947.6 44.9123FORCE 11 111 0 1.0 1584.39-4922.53-369.264FORCE 11 112 0 1.0 1056.26-4337.95 1861.03FORCE 11 121 0 1.0 4753.18-3093.03 661.133FORCE 11 122 0 1.0 3168.79-3118.28 1832.56


$...............................................................................$ $ TRIM CASE: 4 $ RIGID AERODYNAMIC LOADS $ FORCE 12 100 0 1.0 0.0 416.853-16643.4FORCE 12 111 0 1.0 0.0 0.0 9339.64FORCE 12 112 0 1.0 0.0 0.0 -16832.9FORCE 12 121 0 1.0 0.0 0.0 18282.9FORCE 12 122 0 1.0 0.0 0.0 -4692.2$...............................................................................$ $ TRIM CASE: 4 $ ELASTIC AERODYNAMIC LOADS $ FORCE 13 100 0 1.0 0.0 166.317-15698.5FORCE 13 111 0 1.0 0.0 0.0 24597.3FORCE 13 112 0 1.0 0.0 0.0 -19263.1FORCE 13 121 0 1.0 0.0 0.0 33389.FORCE 13 122 0 1.0 0.0 0.0 -5317.57$...............................................................................$ $ TRIM CASE: 4 $ RIGID (AERODYNAMIC + APPLIED - INERTIAL) LOADS $ FORCE 14 100 0 1.0 -1.65-12-32730.5-16664.5FORCE 14 111 0 1.0 1584.39-4922.53 9041.61FORCE 14 112 0 1.0 1056.26-4337.95-15877.9FORCE 14 121 0 1.0 4753.18-3093.03 19127.7FORCE 14 122 0 1.0 3168.79-3118.28-2975.3$...............................................................................$ $ TRIM CASE: 4 $ ELASTIC (AERODYNAMIC + APPLIED - INERTIAL) LOADS $ FORCE 15 100 0 1.0 -1.65-12-32781.3-15653.6FORCE 15 111 0 1.0 1584.39-4922.53 24228.1FORCE 15 112 0 1.0 1056.26-4337.95-17402.1FORCE 15 121 0 1.0 4753.18-3093.03 34050.1FORCE 15 122 0 1.0 3168.79-3118.28-3485.

GUI SupportPatran currently does not support the TRIMF case control

SimXpert does not have an aeroelastic workspace.


106

CSV Output of Trim Results

IntroductionMSC Nastran 2010 contains a new feature that allows the user to create a summary of the trim results in a CSV (Comma Separated Values) file suitable for viewing and manipulating in a spreadsheet application.

BenefitsThe CSV file provides a convenient summary of the trim results that would otherwise have to be gleaned from disparate parts of the .f06 file. This is particularly valuable when hundreds of subcases are being analyzed in a single run.

InputThe CSV feature in activated by PARAM LDSUM (Ch. 5) in the MD/MSC Nastran Quick Reference Guide. The unit the CSV file is written to is specified by PARAM XYUNIT. An assign statement such as:

assign userfile='aecsv1.csv' status=unknown form=formatted unit=52

defines the file where the results are stored. The unit 52 corresponds to PARAM XYUNIT 52.

PARAM LDSUM

Default = 0

Dictates what information is to be stored on a CSV (comma separated values) file in a SOL 144 (static aeroelasticity) task. The unit the CSV file is stored to is specified by param XYUNIT. LDSUM has the following options:

• = 0 (Default) – Do not create a CSV file for static aeroelasticity

• =1 Create a CSV file that contains for each static aeroelastic subcase:

a. Subcase ID

b. Mach number

c. Dynamic Pressure

d. Trim Values

e. Mass and CG information (mass, xcg,ycg,zcg, IXX,IYY,IZZ,IXY,IXZ and IYZ)

• = 2 Same as 1 plus net structural monitor point (MONPNT1, MONDPS1, MONPNT2, MONPNT3) results

• = 3 Same as 2 plus the output of RIGID AIR, ELASTIC RESTRAINED, and INERTIAL, RIGID APPLIED and ELASTIC APPLIED components for the structural MONPNT1 results


OutputThe output is a csv file as shown in the test case below.

Guidelines and Limitations1. PARAM LDSUM can appear in case control or in the bulk data portion of the input file. Only one

PARAM LDSUM can appear.

2. The first row in the spreadsheet provides titles for the columns that contain the results. Each subsequent row contains all the requested results for a single subcase. This means that there is a potential for more columns in the row than standard spreadsheet programs like Microsoft Excel can accommodate. This implies it may be necessary to use a special purpose spreadsheet tool to take full advantage of the new capability.

3. The PARAM XYUNIT is also used in SOL 200 to provide design optimization results. If the SOL 200 run includes static aeroelastic subcases and PARAM LDSUM is used, the resulting spreadsheet will have a row for each static aero subcase for each design iteration followed by the design optimization results.

Test CasesThe following test cases are available in the TPL in directory /tpl/ue_csv09: aecsv1.dat, aecsv4.dat

TPL Problem aecsv1.dat

TPL example problem aecsv1 uses PARAM,LDSUM,1 to provide a summary of the trim results without any monitor points results.

assign userfile='OUTDIR:aecsv1.csv' status=unknown, form=formatted unit=52

SOL 144 $ STATIC AEROCENDparam ldsum 1 $ control for csv outputparam xyunit 52 $ output unit for csv fileTITLE = EXAMPLE HA144F: FSW WITH FUSELAGE, 3 CONTROLS & 2 STOR HA144FSUBTI = UNSYMMETRIC FLIGHT CONDITIONS, DOUBLET-LATTICE AEROLABEL = FULL-SPAN MODEL WITH DISPLACED CANARDECHO = BOTHSPC = 1 $ SYMMETRIC CONSTRAINTSMPC = 10 $ CANARD/FUSELAGE STRUCTURAL CONNECTIONSDISP = ALL $ PRINT ALL DISPLACEMENTSMONITOR = ALLSUBCASE 1 TRIM = 1 $ HIGH SPEED LEVEL FLIGHTBEGIN BULK

Partial listing of the resulting CSV file:

Sub Case,Mach,Dynamic Pressure ,INTERCEPT ,ANGLEA ,P ... 1, 9.000000E-01, 1.200000E+03, 1.000000E+00 ...


108

TPL Problem aecsv4.dat

TPL problem aecsv4.dat is a more comprehensive example geometry with PARAM,LDSUM,4 to provide a summary of the trim results including comprehensive monitor point results.

Figure 6-4 TPL Problem aecsv4.dat

assign userfile='OUTDIR:aecsv4.csv' status=unknown, form=formatted unit=52SOL 144TIME 600ENDparam ldsum 4 $ control for csv outputparam xyunit 52 $ output unit for csv fileSEALL = ALLSUPER = ALLECHO = SORTMAXLINES = 999999AECONFIG = Freedom4SUBCASE 1$ Subcase name : Mach .4 Level Flight SUBTITLE=Mach .4 Level Flight DISPLACEMENT(SORT1,REAL,plot)=ALL SPCFORCES(SORT1,REAL)=ALL OLOAD(SORT1,REAL,plot)=ALL STRESS(SORT1,REAL,VONMISES,BILIN,plot)=ALLTRIM = 1AESYMXZ = AsymmetricAESYMXY = AsymmetricSUPORT1 = 1AEROF = ALLAPRES = ALL

BEGIN BULKPARAMAESMETHRITZ

GUI SupportPatran currently does not support PARAM,LDSUM

SimXpert does not have an aeroelastic workspace.


SUBCOM/SUBSEQ with Static Aeroelasticity

IntroductionThe SUBCOM/SUBSEQ commands that were previously limited to statics problems have been extended to static aeroelasticity problems (both in SOL 144 and in SOL 200 with ANALYSIS=SAERO).

BenefitsThe motivation for this task is to allow a scaling of the results (displacements and element responses) from a static aeroelastic trim analysis as a postprocessing operation. This is particularly of benefit in an optimization task where one desires to explore applying different limit factors to the results in a design task.

User InputsThe existing SUBCOM/SUBSEQ case control commands have been activated for a static aeroelastic analysis. Typically, a single SUBSEQ coefficient is used to scale the results of the previous subcase, although multiple coefficients are supported. In a SOL 200 job, the SUBCOM subcase needs to be accompanied by an ANALYSIS=SAERO command to designate that this subcase is to be grouped with the static aeroelastic subcases.

Outputs Data recovery occurs for the SUBCOM subcases in the same way as any other subcase. The SUBCOM ID appears on the right hand side of the page to indicate that the results are associated with the SUBCOM

Guidelines and LimitationsOnly element and grid responses are affected by the SUBCOM. Trim results and stability derivative results are neither computed or output. In SOL 200, it is illegal to invoke a DRESP1 with RTYPE = STABDER or TRIM from a SUBCOM subcase that has ANALYSIS=SAERO.

ExampleA variation of the familiar HA144A test case for the forward swept wing is used to demonstrate the new SUBCOM capability in a SOL 144 test case. The test case is available in the TPL in directory /tpl/ue_csv09: subcoma.dat. The case control is shown below and it is seen that there is a SUBCOM 4 that requests output that is 50% greater than that of subcase 3.

TITLE = EXAMPLE HA144A: 30 DEG FWD SWEPT WING WITH CANARD HA14 HA144ASUBTI = SYMMETRIC FLIGHT CONDITIONS, DOUBLET-LATTICE AEROLABEL = HALF-SPAN MODEL, STATIC SYMMETRIC LOADING


110

ECHO = BOTH SPC = 1 $ SYMMETRIC CONSTRAINTS DISP = ALL $ PRINT ALL DISPLACEMENTS STRESS = ALL $ PRINT ALL STRESSES FORCE = ALL $ PRINT ALL FORCES AEROF = ALL $ PRINT ALL AERODYNAMIC FORCES APRES = ALL $ PRINT ALL AERODYNAMIC PRESSURESSUBCASE 1 TRIM = 1 $ 1 G LEVEL FLIGHT (LOW SPEED)SUBCASE 2 TRIM = 2 $ 1 G LEVEL FLIGHT (HIGH SUBSONIC SPEED)SUBCASE 3 TRIM = 3 $ 1 G LEVEL FLIGHT (LOW SUPERSONIC SPEED)subcom 4 subseq(1.5) BEGIN BULK

A snippet of the output from this job is shown below. It is seen that the displacements of SUBCOM 4 are 50% greater than those of SUBCASE 3.

1 EXAMPLE HA144A: 30 DEG FWD SWEPT WING WITH CANARD HA14 HA144A OCTOBER 21, 2009 MSC Nastran 10/20/09 PAGE 62 SYMMETRIC FLIGHT CONDITIONS, DOUBLET-LATTICE AERO 0 HALF-SPAN MODEL, STATIC SYMMETRIC LOADING SUBCASE 3

D I S P L A C E M E N T V E C T O R POINT ID. TYPE T1 T2 T3 R1 R2 R3 90 G 0.0 0.0 0.0 0.0 0.0 0.0 97 G 0.0 0.0 -5.984770E-03 0.0 -6.176078E-04 0.0 98 G 0.0 0.0 -8.086921E-04 0.0 -3.176077E-04 0.0 99 G 0.0 0.0 -8.673837E-04 0.0 3.528226E-04 0.0 100 G 0.0 0.0 -9.000074E-03 0.0 1.363732E-03 0.0 110 G 0.0 0.0 -2.312949E-03 6.497581E-04 1.981678E-03 0.0 111 G 0.0 0.0 2.641245E-03 6.497581E-04 1.981678E-03 0.0 112 G 0.0 0.0 -7.267144E-03 6.497581E-04 1.981678E-03 0.0 120 G 0.0 0.0 1.953165E-02 1.088872E-03 2.236090E-03 0.0 121 G 0.0 0.0 2.512188E-02 1.088872E-03 2.236090E-03 0.0 122 G 0.0 0.0 1.394143E-02 1.088872E-03 2.236090E-03 0.01 EXAMPLE HA144A: 30 DEG FWD SWEPT WING WITH CANARD HA14 HA144A OCTOBER 21, 2009 MSC Nastran 10/20/09 PAGE 63 SYMMETRIC FLIGHT CONDITIONS, DOUBLET-LATTICE AERO 0 HALF-SPAN MODEL, STATIC SYMMETRIC LOADING SUBCOM 4 D I S P L A C E M E N T V E C T O R POINT ID. TYPE T1 T2 T3 R1 R2 R3 90 G 0.0 0.0 0.0 0.0 0.0 0.0 97 G 0.0 0.0 -8.977155E-03 0.0 -9.264118E-04 0.0 98 G 0.0 0.0 -1.213038E-03 0.0 -4.764115E-04 0.0 99 G 0.0 0.0 -1.301076E-03 0.0 5.292339E-04 0.0 100 G 0.0 0.0 -1.350011E-02 0.0 2.045598E-03 0.0 110 G 0.0 0.0 -3.469424E-03 9.746371E-04 2.972517E-03 0.0 111 G 0.0 0.0 3.961868E-03 9.746371E-04 2.972517E-03 0.0 112 G 0.0 0.0 -1.090072E-02 9.746371E-04 2.972517E-03 0.0 120 G 0.0 0.0 2.929747E-02 1.633308E-03 3.354135E-03 0.0 121 G 0.0 0.0 3.768281E-02 1.633308E-03 3.354135E-03 0.0 122 G 0.0 0.0 2.091214E-02 1.633308E-03 3.354135E-03 0.0


Upper Hessenberg Complex Eigenanalysis No Longer Supported for Flutter AnalysisThe original implementation of the PK flutter analysis utilized an Upper Hessenberg algorithm for extracting complex eigenvalues. In MSC Nastran 2005 R3, a QZ algorithm was introduced that had the advantage of being able to accommodate a singular matrix. For several releases, the Upper Hessenberg algorithm could still have been used if NASTRAN SYSTEM(108)=1 was specified. This option is no longer available and the use of SYSTEM(108)=1 will cause a User Information Message to be printed:

*** USER INFORMATION MESSAGE 5282 (FA1PKE)SYSTEM(108) HAS A NONZERO VALUE, IMPLYING THERE IS A preference TO USE THE UPPER HESSENBERG METHOD FOR PK FLUTTER ANALYSIS.

User information: THIS IS NO LONGER SUPPORTED AND THE DEFAULT QZ ALGORITHM WILL BE USED.


112

Chapter 7: Elements MSC Nastran Release Guide

7 Elements

Enhancements to Connector Elements

Offsets for Beams and Shells

MSC Nastran 2010 Release GuideEnhancements to Connector Elements

114

Enhancements to Connector Elements

IntroductionA finite element modeler has many ways of modeling structural connections and fasteners. Spot welds, seam welds, bolts, screws, and so on can be represented, depending on the modeling goals, either with flexible springs or bars (CBUSH, CBAR), rigid elements (RBAR, RBE2, RBE3), or multipoint constraints (MPC). Though generally, these elements are sometimes difficult to use; singularities may be introduced particularly in the out-of-plane rotational direction for shells, rigid body invariance may not be assured, and data preparation and input can be a formidable task in real-world applications. Increasing mesh refinement can also introduce further stiffness errors; point-to-point connections in which effective cross-sectional areas are larger than 20% of the characteristic element lengths can often lead to significant underestimation of connector stiffness.

Connector elements are a special class of elements that were introduced to MSC Nastran in V2001. The first implementation was for elements that represent spot welds. These elements are convenient to define by the user because all that is necessary to define the elements a geometric location in space. Subsequent versions of MSC Nastran have provided many enhancements and now include elements that can model bolts and seam welds.

In MSC Nastran 2010, there have been further enhancements to the existing elements to allow data recovery in dynamic solution sequences, calculate displacements and stresses for seam weld elements and support user defined coordinate systems for spot weld elements the way that user defined coordinate systems are used for bolt elements.

BenefitsThe following features are included in MSC Nastran 2010.

1. Support auxiliary displacement output for frequency response and transient analysis for CWELD, CFAST and CSEAM elements.

2. Generate eight auxiliary grids internally and compute their associated displacements for CSEAM elements.

3. Provide stress and strain output for CSEAM elements.

4. Support user defined element coordinate system for CWELD elements.

Theory

Mathematical Model to Construct the CSEAM Auxiliary Points

The four auxiliary vertex points of the cross section at start point GS are constructed by the following

equations (see Figure 7-1), where and are tangent vectors of the element coordinate system at start t1s

t2s

115CHAPTER 7Elements

point GS, W is the width of the seam, and T is the thickness of the seam. For a continuous seam, and

vectors are adjusted to a common face for the two consecutive seam elements. Therefore, the four

auxiliary grids at the start point S of the ith seam should be coincident with the four auxiliary grids at the end point E of the (i-1)th seam, if both elements have same width and thickness.

Figure 7-1 Seam Weld Cross Section at Start Point S

The four auxiliary vertex points of the cross section at end point GE are calculated in the same way as that for the start point GS. These eight auxiliary points form an auxiliary HEXA element with the following vertex points.

1 2 3 4 5 6 7 8

GSA1 GSA2 GSB2 GSB1 GEA1 GEA2 GEB2 GEB1

t1s

t2s

xSA1 xs=W2----- t1

s–

T2--- t2

s–

xSA2 xs=W2----- t1

s–

T2--- t2

s–

xSB1 xs=W2----- t1

s–

T2--- t2

s–

xSB2 xs=W2----- t1

s–

T2--- t2

s–

t2s

t1s

GSB1 GSB2

GSA2GSA1

T/2

T/2

W/2 W/2

GS


116

InputThe only new input is related to the element coordinate system on the CWELD entry. The relevant changes are shown in the following CWELD entry (other items that did not change are not shown).

The other enhancements are to extend the solution sequences and provide additional output for the CSEAM.


Defines a weld or fastener connecting two surface patches or points. Large displacement and large rotational effects are supported when using SOL 600 and MSC Nastran SOL 400 only.

Format:

Example:

Alternate formats and examples:

Format ELPAT:

Example:

Format ELEMID:

Example:

Format GRIDID:

CWELD Weld or Fastener Element Connection

1 2 3 4 5 6 7 8 9 10

CWELD EWID PWID GS “PARTPAT” GA GB MCID

PIDA PIDB

XS YS ZS

CWELD 101 8 203 PARTPAT

21 33

CWELD EWID PWID GS “ELPAT” GA GB MCID

SHIDA SHIDB

XS YS ZS

CWELD 103 5 403 ELPAT

309 511

CWELD EWID PWID GS “ELEMID” GA GB MCID

SHIDA SHIDB

CWELD 103 5 403 ELEMID

309 511

CWELD EWID PWID GS “GRIDID” GA GB SPTYP MCID


118

Example:

Format ALIGN:

Example:

Remarks:

15. MCID = -1 or blank (Default), then the coordinate system is as defined in Remark 12.

If MCID > 0, then a “beam” like coordinate system is defined. The axis direction of the connector defined as

OutputThe results of CWELD and CFAST elements are stored in standard beam and standard bush formats respectively; while the results of CSEAM elements are displayed in hexa format with the eight auxiliary grids as vertex points. For frequency response, the output data may be in rectangular or polar format. The examples below will give a description of the output.

Test CasesThe following test cases are available in the TPL in directory /tpl/rg4_conn: r4_conn_exa.dat, r4_conn_exb.dat

GA1 GA2 GA3 GA4 GA5 GA6 GA7 GA8

GB1 GB2 GB3 GB4 GB5 GB6 GB7 GB8

CWELD 7 29 233 GRIDID QT

15 28 31 35 46 51 55 60

3 5 8

CWELD EWID PWID “ALIGN” GA GB MCID

CWELD 7 29 ALIGN 103 259

Field Contents Type Default

MCID Specifies the element stiffness coordinate system. See Remark 15.

Integer > -1 or blank

Default = -1

xelem

1xB xA–

xB xA–---------------------=


TPL example model r4_conn_exa.dat

This example demonstrates the modeling of frequency response for CWELD elements with user defined element coordinate system and the ELPAT format:

Figure 7-2 Example r4_conn_exa.dat geometry

SOL 108END

DISPL= ALLFORCE=1STRESS=1SUBCASE 1 SUBTITLE= shear the weld dload=1 method= 400 freq=11BEGIN BULK

$ Spot weldcweld, 30, 30, 300, elpat, , , , 755, +CW1+CW1, 11, 10pweld, 30, 10, 0.1$cord2r, 755, , 0., 0., 0., 0., 0., -1., , 0., 1., 0.

Note: Turning on additional diagnostics with SWLDPRM,PRTSW,1 will provide information about the auxiliary GRID locations and ids. The displacement printout shown afterwards gives the auxiliary GRID displacements.


120

The displacement, element force and stress results are shown as follows.

CWELD EID= 30 WITH ELPAT OR PARTPAT GS IS MOVED FROM ( 1.0000E+00, 4.0000E-01, 5.0000E-01) TO ( 1.0072E+00, 3.9841E-01, 2.3848E-01) AUXILIARY POINTS= ( 9.8328E-01, 3.6349E-01, 4.3080E-01) ( 1.0718E+00, 3.6307E-01, 4.2595E-01) ( 1.0714E+00, 4.5062E-01, 4.1219E-01) ( 9.8294E-01, 4.5104E-01, 4.1704E-01) ( 9.7227E-01, 3.5413E-01, 4.8613E-02) ( 1.0608E+00, 3.5407E-01, 5.3039E-02) ( 1.0608E+00, 4.4269E-01, 5.3042E-02) ( 9.7233E-01, 4.4275E-01, 4.8616E-02) NUMBER OF TIMES GS MOVES= 1 NUMBER OF TIMES DA IS REDUCED= 0 ANGLE BETWEEN TWO SHELL NORMALS= 10.77 GS=( 1.007E+00, 3.984E-01, 2.385E-01) GA=( 1.027E+00, 4.071E-01, 4.215E-01) GB=( 1.017E+00, 3.984E-01, 5.083E-02) T_BE MATRIX: -2.9113E-02 9.9958E-01 0.0000E+00 -2.3310E-02 -6.7890E-04 -9.9973E-01 -9.9930E-01 -2.9105E-02 2.3320E-02 GA ID = 101000001 GB ID = 101000002 PATCH A: EID= 11 GIDS= 111 112 113 114 0 0 0 0 EID= 11 GIDS= 111 112 113 114 0 0 0 0 EID= 11 GIDS= 111 112 113 114 0 0 0 0 EID= 11 GIDS= 111 112 113 114 0 0 0 0 PATCH B: EID= 10 GIDS= 101 102 104 0 0 0 0 0 EID= 10 GIDS= 101 102 104 0 0 0 0 0 EID= 10 GIDS= 101 102 104 0 0 0 0 0 EID= 10 GIDS= 101 102 104 0 0 0 0 0

0 SUBCASE 1 FREQUENCY = 1.000000E+01 C O M P L E X D I S P L A C E M E N T V E C T O R (REAL/IMAGINARY) POINT ID. TYPE T1 T2 T3 R1 R2 R30 101 G -7.791607E+00 -6.924270E-01 1.487187E-02 -9.710412E-02 -3.703940E-02 7.556269E-01 5.248972E-07 1.286137E-05 -3.858250E-05 8.550648E-06 -3.772603E-05 -1.167940E-050 102 G -7.795311E+00 8.285373E-01 8.895068E-02 -9.710412E-02 -3.703940E-02 7.556269E-01 -3.276013E-06 -1.135917E-05 3.686814E-05 8.550648E-06 -3.772603E-05 -1.167940E-050 104 G -8.547235E+00 -6.924270E-01 -8.223225E-02 -9.710412E-02 -3.703940E-02 7.556269E-01 1.220098E-05 1.286628E-05 -3.003202E-05 8.550648E-06 -3.772603E-05 -1.167940E-050 111 G -7.810302E+00 -6.437430E-01 1.438998E-02 -9.709817E-02 -3.749770E-02 7.554925E-01 2.649062E-06 -7.254856E-06 1.925463E-05 7.837443E-06 1.728056E-05 4.460259E-060 112 G -7.806551E+00 7.064308E-01 8.188874E-02 -9.706381E-02 -3.750144E-02 7.554878E-01 8.092410E-07 1.905125E-06 -1.219865E-05 3.714137E-06 1.772999E-05 5.025620E-060 113 G -8.705634E+00 8.381162E-01 -2.708165E-02 -9.705853E-02 -3.752504E-02 7.554884E-01 -8.982315E-06 3.584831E-06 -1.199635E-05 3.080924E-06 2.056146E-05 4.950366E-060 114 G -8.569546E+00 -6.340333E-01 -8.270818E-02 -9.709751E-02 -3.751689E-02 7.554936E-01 1.027111E-07 -8.022997E-06 2.709107E-05 7.757903E-06 1.958415E-05 4.324563E-060 300 G 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.00 101000001 G -8.114884E+00 1.247901E-01 1.340001E-02 -9.707516E-02 -3.750778E-02 7.554896E-01 -7.519017E-07 -2.020528E-06 3.345231E-06 5.076566E-06 1.849139E-05 4.807756E-060 101000002 G -8.094540E+00 8.064318E-02 1.383702E-02 -9.710412E-02 -3.703940E-02 7.556269E-01 3.244864E-06 5.525994E-07 3.173882E-06 8.550648E-06 -3.772603E-05 -1.167940E-05

0 SUBCASE 1


TPL Problem r4_conn_exb.dat

TPL problem r4_conn_exb.dat demonstrates the static analysis for CSEAM elements with displacement and stress output requests. In this case there are several overlapping plates connected with CSEAM. The CSEAM model has internally generated auxiliary grid IDs starting from 90001.

Figure 7-3 Example r4_conn_exb.dat geometry

sol 101cendload = 10spc = 10disp = allset 4 = 10001,10002,10003,10004,10005stress = 4begin bulk

...swldprm prtsw 1$cseam100011000se0elem1132141000110002cseam100021000se010elem3304401000310004cseam100031000se1010elem50300604001000510006

0 SUBCASE 1 FREQUENCY = 1.000000E+01 C O M P L E X F O R C E S I N W E L D E L E M E N T S ( C W E L D P ) (REAL/IMAGINARY) ELEMENT BEND-MOMENT END-A BEND-MOMENT END-B - SHEAR - AXIAL ID PLANE 1 (MZ) PLANE 2 (MY) PLANE 1 (MZ) PLANE 2 (MY) PLANE 1 (FY) PLANE 2 (FZ) FORCE FX TORQUE MX 30 -1.196586E-01 5.743172E-03 -3.431428E-03 3.061221E-03 -3.133458E-01 7.230475E-03 9.014460E-03 -1.499722E-02 1.436048E-02 -6.890686E-04 4.131643E-04 -3.671607E-04 3.760163E-02 -8.678561E-04 -1.081325E-03 1.800308E-03

0 SUBCASE 1 FREQUENCY = 1.000000E+01 C O M P L E X S T R E S S E S I N W E L D E L E M E N T S ( C W E L D P ) (REAL/IMAGINARY) ELEMENT AXIAL MAX STRESS MIN STRESS MAX STRESS MIN STRESS MAXIMUM BEARING ID STRESS END-A END-A END-B END-B SHEAR STRESS STRESS 30 1.147757E+00 1.278480E+03 -1.276184E+03 6.728133E+01 -6.498582E+01 1.297688E+02 3.134292E+01 -1.376786E-01 1.531557E+02 -1.534311E+02 7.810646E+00 -8.086003E+00 1.557555E+01 3.761164E+00


122

cseam100041000sl1030pshell702001000710008cseam100051000sl1030 pshell702001000910010$$23456pseam 1000 200.2 $

The displacement and stress results are shown as follows.

0 D I S P L A C E M E N T V E C T O R POINT ID. TYPE T1 T2 T3 R1 R2 R3 509 G 2.805908E-02 -2.599996E-04 -2.023823E-01 6.784551E-05 1.449939E-02 -6.875442E-07 10001 G 0.0 0.0 0.0 0.0 0.0 0.0 10002 G 0.0 0.0 0.0 0.0 0.0 0.0 10003 G 0.0 0.0 0.0 0.0 0.0 0.0 10004 G 0.0 0.0 0.0 0.0 0.0 0.0 10005 G 0.0 0.0 0.0 0.0 0.0 0.0 10006 G 0.0 0.0 0.0 0.0 0.0 0.0 10007 G 0.0 0.0 0.0 0.0 0.0 0.0 10008 G 0.0 0.0 0.0 0.0 0.0 0.0 10009 G 0.0 0.0 0.0 0.0 0.0 0.0 10010 G 0.0 0.0 0.0 0.0 0.0 0.0 101000001 G 7.372772E-03 3.151977E-03 -3.154194E-02 0.0 0.0 0.0 101000002 G 7.673701E-03 3.151946E-03 -3.261929E-02 0.0 0.0 0.0 101000003 G 6.849292E-03 -2.230086E-04 -3.259117E-02 0.0 0.0 0.0 101000004 G 6.850486E-03 -2.224528E-04 -3.151381E-02 0.0 0.0 0.0 101000005 G 7.376085E-03 -3.603303E-03 -2.985591E-02 0.0 0.0 0.0 101000006 G 7.677149E-03 -3.603285E-03 -3.093446E-02 0.0 0.0 0.0 101000007 G 6.851723E-03 -2.002655E-04 -3.090637E-02 0.0 0.0 0.0 101000008 G 6.853319E-03 -2.007576E-04 -2.982783E-02 0.0 0.0 0.0 101000009 G 2.237370E-02 3.150388E-03 -1.297074E-01 0.0 0.0 0.0 101000010 G 2.267277E-02 3.150356E-03 -1.325928E-01 0.0 0.0 0.0 101000011 G 2.469706E-02 -2.220506E-04 -1.326306E-01 0.0 0.0 0.0 101000012 G 2.469634E-02 -2.216082E-04 -1.297368E-01 0.0 0.0 0.0 101000013 G 2.237715E-02 -3.602424E-03 -1.284061E-01 0.0 0.0 0.0 101000014 G 2.267609E-02 -3.602407E-03 -1.313060E-01 0.0 0.0 0.0

0 S T R E S S E S I N S E A M E L E M E N T S ( C S E A M )0 CORNER ------CENTER AND CORNER POINT STRESSES--------- DIR. COSINES MEAN ELEMENT-ID GRID-ID NORMAL SHEAR PRINCIPAL -A- -B- -C- PRESSURE VON MISES 0 10001 0GRID CS 8 GP0 CENTER X 7.490945E+02 XY 3.996807E-02 A 7.490945E+02 LX 1.00 0.0 0.0 -2.497961E+02 7.489476E+02 Y 2.938447E-01 YZ -7.958079E-13 B 1.469501E-01 LY 0.00 0.0 0.0 Z 0.0 ZX 9.094947E-12 C 1.469501E-01 LZ 0.00 0.0 0.00 101000001 X 1.635867E+03 XY 7.365383E-02 A 1.635867E+03 LX 1.00 0.00 0.00 -6.910887E+02 1.425098E+03 Y 1.320214E+02 YZ -2.318950E-02 B 1.320214E+02 LY 0.00 1.00 0.00 Z 3.053776E+02 ZX -3.091930E-04 C 3.053776E+02 LZ 0.00 0.00-1.000 101000002 X 1.635903E+03 XY 7.365383E-02 A 1.635903E+03 LX 1.00 0.00 0.00 -6.911407E+02 1.425078E+03 Y 1.320599E+02 YZ 2.318950E-02 B 1.320599E+02 LY 0.00 1.00 0.00 Z 3.054590E+02 ZX -3.091930E-04 C 3.054590E+02 LZ 0.00 0.00-1.000 101000003 X -1.369259E+02 XY 6.282309E-03 A -1.311691E+02 LX 0.00 0.00-1.00 1.911511E+02 1.713834E+02 Y -1.311691E+02 YZ 2.318950E-02 B -3.053584E+02 LY 1.00 0.00 0.00 Z -3.053584E+02 ZX -3.091930E-04 C -1.369259E+02 LZ 0.00 1.00 0.000 101000004 X -1.370004E+02 XY 6.282309E-03 A -1.312972E+02 LX 0.00 0.00-1.00 1.912586E+02 1.714006E+02 Y -1.312972E+02 YZ -2.318950E-02 B -3.054782E+02 LY 1.00 0.00 0.00 Z -3.054782E+02 ZX -3.091930E-04 C -1.370004E+02 LZ 0.00 1.00 0.000 101000005 X 1.636985E+03 XY 7.365383E-02 A 1.636985E+03 LX 1.00 0.00 0.00 -6.919526E+02 1.425502E+03 Y 1.326195E+02 YZ -2.318950E-02 B 1.326195E+02 LY 0.00 1.00 0.00 Z 3.062530E+02 ZX 3.091930E-04 C 3.062530E+02 LZ 0.00 0.00-1.000 101000006 X 1.636935E+03 XY 7.365383E-02 A 1.636935E+03 LX 1.00 0.00 0.00 -6.918799E+02 1.425529E+03 Y 1.325717E+02 YZ 2.318950E-02 B 1.325717E+02 LY 0.00 1.00 0.00 Z 3.061331E+02 ZX 3.091930E-04 C 3.061331E+02 LZ 0.00 0.00-1.000 101000007 X -1.395096E+02 XY 6.282309E-03 A -1.322068E+02 LX 0.00 0.00-1.00 1.926500E+02 1.704930E+02 Y -1.322068E+02 YZ 2.318950E-02 B -3.062338E+02 LY 1.00 0.00 0.00 Z -3.062338E+02 ZX 3.091930E-04 C -1.395096E+02 LZ 0.00 1.00 0.000 101000008 X -1.394977E+02 XY 6.282309E-03 A -1.322486E+02 LX 0.00 0.00-1.00 1.926329E+02 1.703949E+02 Y -1.322487E+02 YZ -2.318950E-02 B -3.061523E+02 LY 1.00 0.00 0.00 Z -3.061523E+02 ZX 3.091930E-04 C -1.394977E+02 LZ 0.00 1.00 0.00


Offsets for Beams and Shells

IntroductionThe original design of the BAR, BEAM and shells included an offset feature. This feature allows you to specify a BAR/BEAM axis of shear centers offset from connected grid points; and, a shell element reference plane offset from the connected grids for triangular and quadrilateral shell elements.

The axis of the elastic BAR/BEAM was assumed to extend from the two endpoints offset from the connected grid points; and, the reference surface of the shell element connected points offset from the connected grids. The elastic stiffness was then calculated in the offset element coordinate system and transformed to the connected grid points using the rigid body transformation equations.

Figure 7-4 CBAR and CBEAM Element Offset Definitions

Figure 7-5 CQUAD4 and CQUAD8 Element Coordinate System Definitions

Prior to the offset project in MSC Nastran 2010, element offsets had the following limitations:

• The differential stiffness is not supported. Therefore, they are not applicable in solution sequence where differential stiffness is required, such as linear buckling analysis (SOL 105).

• The effect of offsets is not included the mass matrix, thus in dynamic analysis, the mass matrix is only an approximation for offsets.

MSC Nastran 2010 Release GuideOffsets for Beams and Shells

124

• The effect of offset is not included in computation of thermal loads, pressure loads, or gravity loads.

• For curved shell problems, the offset is defined in the z direction of the element and not the shell normal direction, this results in gaps between elements when offset geometry is applied.

• The transformation is linear; therefore, it cannot be used nonlinear analysis.

This enhancement extends the use the offsets to nonlinear solutions, linear differential stiffness, load and mass generation and maintenance of geometric continuity for adjacent curved shells along the common shell normal.

BenefitThe enhanced offset method has the following benefits:

• DIFFERENTIAL STIFFNESS - The differential stiffness is computed for the offset so that it is applicable to solution sequences that required differential stiffness. These include, for example, SOL 103, SOL 105, and SOL 400.

• MASS - The effect of offset is included in the mass matrix generation.

• LOADS - The effect of offset is included in the load generation.

• GEOMETRY COMPATIBILITY - For QUAD4, TRIA3, QUADR, TRIAR, QUAD8 and TRIA6 elements the shell normal is used as the offset direction. Thus, the new offset will enhance solution of a model in two aspects: there is no gap in the offset geometry, and the mass and stiffness matrices are computed based on the offset geometry. Figures New Offset Behavior when Angle between adjacent elements is less than SNORM and New Offset Behavior when Angle between adjacent elements is greater than SNORM demonstrate how the new offsets are considered in conjunction with SNORM.

• NONLINEAR EFFECTS - The transformations are nonlinear so that it can be used in nonlinear analysis such as SOL 400.

Feature Description

Differential Stiffness

The original offset method does not compute the differential stiffness for offset. Therefore, it is not applicable to solution sequences where differential stiffness is needed such as SOL 103 or SOL 105. The

Note: The element offset project was completed for both MD Nastran and MSC Nastran. This section includes the complete capability description for both MD Nastran and MSC Nastran as documented in the MD Nastran 2010 Release Guide. For MSC Nastran, the new offset methods are available in all the linear solution sequences that calculate differential stiffness, but the new offset capability is not available in the MSC Nastran nonlinear solutions (SOL 106 and SOL 129).


enhanced offset computes the differential stiffness for offset, using the same formulation as that of the Lagrange RBAR element. The Lagrange multipliers are eliminated internally. For the new offset method, the differential stiffness is computed for offsets for solution sequences that require it. These include SOL 105, SOL 400 and any linear solution that supports differential stiffness effects via the case control entry STATSUB.

Mass Matrix

For an element, the mass matrix is computed at the offset locations of the element. The mass matrix needs to be transformed from the offset locations to the external grid points. In performing this transformation, mass moments of inertial are created. Without this transformation, the mass matrix for model with offsets is only an approximation. With this transformation, the mass matrix for SOL 103, and other dynamic solution sequences are correctly computed. The additional terms in the mass matrix may adversely affect the solution time for dynamic solutions; therefore, a provision to disable the mass offset calculation while retaining other offset effects is available.

Load Effects

For element loads, such as thermal, pressure, and gravity load, the loads are computed at the offset locations of the element. These loads are transformed from the offset locations to the external grid points.

For thermal load, there are two types of effect due to offset:

• The location of thermal load is changed due to offset, i.e. the thermal load is first computed at the element offset locations and then transformed to the external grid points. This effect is computed for both linear solution sequences and nonlinear solution sequence SOL 400.

• The temperature load will change the length of the offset. This effect is computed for the nonlinear solution sequence SOL 400 only.

If you don’t want these effects to be calculated, a provision to disable the offset load effects is available.

Offset Direction

For beam elements, the direction of offset is given by the WiA and WiB on the connection Bulk Data entries.

For shell elements, the old offset direction is in z-direction of the element coordinate system. For curved shell model, such as cylindrical shell, this approach has two deficiencies:

• The offset geometry has gap or overlap in the structure model.

• The computations of stiffness matrix, mass matrix, and element load are based on the original geometry, which is not the same as the offset geometry.

In order to remedy these deficiencies, the new offset method for QUAD4, TRIA3, QUADR, TRIAR, QUAD8, and TRIA6 use the shell normal as the offset direction for default.


126

If you don’t want the geometric advantages of the new offset method, a provision to revert to the previous offset behavior for shell elements is available.

Figure 7-6 Unique Grid Point Normal for Adjacent Shell Elements

Figure 7-7 New Offset Behavior when Angle between adjacent elements is less than SNORM

Figure 7-8 New Offset Behavior when Angle between adjacent elements is greater than SNORM


Nonlinear Analysis

For nonlinear analysis, the following are implemented for the new offset method:

• The offsets are formulated in large rotation theory.

• The differential stiffness for offsets is computed.

• For thermal loads, the length of the offsets may change due to thermal loads. This effect has been implemented.

InputThe new offset formulation must be invoked by specifying MDLPRM,OFFDEF,option in the Bulk Data Section. The format of the MDLPRM entry and options associated with OFFDEF are shown.

MDLPRM Format:

Example:

1 2 3 4 5 6 7 8 9 10

MDLPRM PARAM1 VAL1 PARAM2 VAL2 PARAM3 VAL3 -etc.-

MDLPRM QR6ROT 2 QRSHEAR 1 OFFDEFF LROFF

Name Description, Type, and Default Values

OFFDEF Element offset definition. A flag to determine how shell elements and bar and beam elements behave when the user supplies ZOFF values on the shell connection entries (CQUAD4, CQUADR, CTRIA3, CTRIAR, CQUAD8, and CTRIA6) and WiA and WiB on CBAR, CBEAM, and CBEAM3 connection entries. (Character)

ELMOFF Standard Nastran offset method. The ZOFF rotate with the shell element. The WiA and WiB offsets for beams are fixed. MSC Nastran 2008 and earlier. (Character, Default)

LROFF Large rotation offsets. The shell normal directions are used to define the offset direction at each shell grid. New for MSC Nastran 2010. This method allows for thermal load effects on ZOFF for shells and WiA and WiB for beams. Thermal load effect for offset is computed based on the grid point or element temperature, and thermal coefficient of the element (see NOTHRM). The mass moment of inertia is computed for the offset due to the grid point location change introduced by offset. Differential stiffness is computed for the offset using the same method as that of the Lagrange formulation of the RBAR. (Character)


128

Note that there is no Case Control modification or other modifications to the input file required. By default, the old offset definition is used (MDLPRM,OFFDEF,ELMOFF)

OutputThere is no new output associated with the new offset methodology.

NODIF LROFF is used but the differential stiffness effect is turned off. (Character)

NOTHRM LROFF is used but the thermal load effects are turned off. The thermal load has two effects: 1) the location of thermal load changes due to offset and 2) the length of offset changes due to thermal load. Effect (1) is computed for all solution sequences and Effect (2) is computed for MSC Nastran SOL 400 only. Both effects are turned off by NOTHRM. (Character)

NODT LROFF is used but the differential stiffness and thermal load effect are turned off. (Character)

ELMZ LROFF is used but the element z-direction is used for the offset direction. IF PARAM, SNORM, 0.0 or the computed value for SNORM is greater then the PARAM,SNORM,value, then the LROFF option will revert to this method for CQUAD4, CTRIA3, CQUADR, and CTRIAR. (Character)

NOMASS LROFF is used but the no mass effects are included. (Character)

NDMTZ LROFF is used but the element z-direction is used for the offset direction and the differential stiffness, the thermal load effects, and the mass effects are turned off. For CQUAD4 and CTRIA3 elements this method should get similar results to the standard ELMOFF method. (Character)

Notes: This entry only effects ZOFF calculations for ZOFF specified on the shell connection entries. For ZOFF specified on the PCOMP or PCOMPG entries, the standard ELMOFF method will be used.

For CBEAM, CBAR, CQUAD8 and CTRIA6 elements, the LROFF option will revert to the ELMZ sub option.

If the computed value for SNORM is greater then the PARAM,SNORM,value and the user wishes not to change the parameter value, the Bulk Data entry SNORM can be used to override the shell normal.

Solution sequences affected: For linear - all solution sequences. For MSC Nastran - SOL 400 nonlinear only. The new method is not implemented into SOL 106 and 129.

Name Description, Type, and Default Values


Guidelines and Limitations

Solution Sequences and Element Type to be Supported by Offset

Solution sequences and element types supported by the new offset method are:

• All linear solutions sequences such as SOL 101, 103, 105, and dynamic solution sequences are supported by the new offset method.

• For nonlinear solution sequences, only SOL 400 is supported. SOL 106 and SOL 129 use the old offset method.

• For linear solution sequences, the new offset method supports the following elements: QUAD4, TRIA3, QUADR, TRIAR, BEAM, BAR, BEAM3, QUAD8, and TRIA6.

• For SOL 400, the new offset supports QUAD4, TRIA3, QUADR, TRIAR, and BEAM.

• For advanced nonlinear elements in SOL 400, in addition to QUAD4, TRIA3, QUADR, TRIAR and BEAM elements, QUAD8, TRIA6 and AXISYM are also supported.

For QUAD4 and TRIA3, the stiffness for drilling DOF’s is zero. The offset is not completely constrained by the elements. For the linear solution sequences, this will not create any problem. For nonlinear analysis, this will not impede the solution of QUAD4/TRIA3 for most cases. However, for certain type of models, especially if the model is completely flat, the zero stiffness will make the solution with offset difficult to converge in SOL 400. If this happens, a large value K6ROT, in the order of 1,000.0 - 10,000.0, may be used to resolve this problem. Another work around is to use the QUADR/TRIAR elements.

The new offset method has not been implemented for composites using the PCOMP or PCOMPG Bulk Data entries. For these entries, the user needs to transfer the offset definition to the connection entries in order to use the new offset method.

Test ProblemsThe following test cases are available in the TPL in directory /tpl/offsetmeth:

TPL Example ofsl014.dat

This example is a simple cantilever with offsets, model by QUAD4 and TRIA3 to obtain solution in the linear buckling analysis (SOL 105). There are two types of element in the file. The first subcase is preload. The second subcase is for QUAD4 and the third subcase is for TRIA3. Both loading and geometry are exaggerated to show the differential effect of offset. Bulk data input required for new offsets: “MDLPRM, OFFDEF, LROFF” The first buckling factor for both QUAD4 and TRIA3 are 1.52 with differential stiffness of offset. However, if we ignore the differential stiffness of offset (OFFDEF=NODIFF), the first buckling factor is 6087.


130

Figure 7-9 TPL example problem ofsl014.dat

TPL Example ofslq416.dat

This example is a cylindrical shell to demonstrate the effect of shell normal using linear static analysis SOL 101. Bulk data input required for new offsets: “MDLPRM, OFFDEF, LROFF” In this example, both elastic element stiffness and mass matrix are computed based on the offset geometry of the model. Since the shell normal is used as the direction of the offset, there is no gap in the geometry. For this model, the weight for shell normal model (OFFSET=LROFF) is 120.8, while for the z-direction model (OFFSET=ELMZ) the weight = 109.8. Results are obtained from grid point weight table. The displacement at T1 of grid 1 is 0.1095. However, if we use offset in z-direction (OFFDEF=ELMZ), then the corresponding displacement is 0.0996.

Figure 7-10 TPL example problem ofslq416.dat

TPL Problem ofsnbm01.dat

This example is for the nonlinear solution of SOL 400. A simple cantilever beam is modeled by BEAM with offset. Bulk data input required for new offsets: “MDLPRM, OFFDEF, LROFF”. Both geometry and loads are exaggerated to show effects of offset and geometric nonlinear. In this model, the large rotation theory and differential stiffness are computed for the offsets of beam. This model is verified by explicitly modeling the offset by RBAR element. Both offset and RBAR give the same results.


Figure 7-11 TPL example problem ofsnbm01.dat

GUI Support

Patran

Currently Patran does not support the definition of MDLPRM,OFFDEF,offdef. However, it is expected that this will be supported in Patran 2008r3.

SimXpert

SimXpert supports the definition of MDLPRM,OFFDEF,option. Figure 7-12 shows the GUI interface.

Figure 7-12 SimXpert Support for MDLPRM,OFFDEF,option


132

Chapter 8: MiscellaneousMSC Nastran 2010 Release Guide

8 Miscellaneous

Enhanced MONSUM

PARAM,NONCUP Usage Extended to SOL 111

Application Regions

New Input File Reader - IFPSTAR

Brake Squeal Analysis

Results and Output Changes

MSC Nastran Error List

MSC Nastran 2010 Release GuideEnhanced MONSUM

134

Enhanced MONSUM

IntroductionThe MONSUM Bulk Data entry was introduced in MSC Nastran 2007 to enable the linear combination of monitor results or the updating in place of a monitor result. The MSC Nastran 2007 implementation imposed the restriction that the results to be summed must be of the same type. This restriction has been relieved in the current delivery.

BenefitsThe MONSUM feature of MSC Nastran 2007 allowed to user to perform such tasks as units conversion on a monitor result or to combine similar types on monitor results. Users also felt a need to combine different types of monitor points to create a special purpose response that is meaningful in the analysis task at hand. This extension of the MONSUM provides this added capability.

Feature DescriptionThe enhanced MONSUM, 2527 Bulk Data entry is described in the MD Nastran Quick Reference Guide. “Legacy” input files are supported in that the alternate format shown in the guide is identical to the standard format of earlier releases. A remark in the MONSUM description indicates that the summed quantities must be of a similar type, specifically:

Force and moment summation monitor points: AMONPNT1, SMONPNT1, MONPNT3

Displacement monitor points: AMONDSP1 and SMONDSP1

ExampleTwo examples are provided in the TPL:

Monsum2.dat – This is a SOL 144 file that has five MONSUM entries. Three of these demonstrate the legacy feature while one combines an aerodynamic monpnt1 and a monpnt3 and another combines a structural monpnt1 with an aerodynamic monpnt1.

Monsum4.dat – This is another SOL 144 file that has a single MONSUM entry that combines a structural mondsp1 with an aerodynamic mondsp1.

Guidelines and LimitationsThe MSC Nastran 2007 Release Guide and the previous examples imply this is a SOL 144 capability. This is not the case in that it can also be used in SOLs 101, 103, 108, 109, 111, 112 and 200.

Summing a MONPNT1 and MONPNT3 in the dynamic solution sequences is currently misleading since the MONPNT1 support inertia results while the MONPNT3 does not.

135CHAPTER 8Miscellaneous

The summed results are printed as one of the monitor types with the following order of precedence: smonpnt1, smondsp1, monpnt3, amonpnt1, amondsp1. E.g., an amonpnt1 and a monpnt3 appearing on the same entry will result in a monpnt3 regardless of which appears first.

If different monitor results are being summed, the NAME appearing on the MONSUM should be unique with respect to other names. For the update in place, the name can be the same.

MSC Nastran 2010 Release GuidePARAM,NONCUP Usage Extended to SOL 111

136

PARAM,NONCUP Usage Extended to SOL 111Until now, the usage of PARAM,NONCUP was allowed only in SOL 112, not in SOL 111. In MD Nastran 2010, usage of this parameter is allowed in SOL 111 also.

In both SOLs 111 and 112, PARAM,NONCUP has the same meaning as follows:

NONCUP = -1 Use uncoupled solution if there are no off-diagonal terms in any of the modal matrices (MHH, BHH, and KHH); otherwise use coupled solution.

NONCUP > -1 Use coupled solution regardless of the existence of off-diagonal terms in the modal matrices.

NONCUP = -2 Use uncoupled solution regardless of the existence of off-diagonal terms in the modal matrices.


Application Regions

IntroductionA new set concept called SET3 has been added which allows one to group together a list of nodes, elements, properties or a list of points and associate them with a unique id.

BenefitsApplication region can be used for group the elements or nodes. It is necessary for total heat load, contact loads, super element radiation load, and primitive radiation load. It is more convenient for the translator to process the groups of FEM data.

InputSET3 entry is used to define the application regions.

MSC Nastran 2010 Release GuideNew Input File Reader - IFPSTAR

138

New Input File Reader - IFPSTARThe IFPStar component is MSC Software’s enhanced bulk data processor that is introduced in MD Nastran 2010 and MSC Nastran 2010. IFP stands for Input File Processor. The IFPStar component utilizes MSC’s Simulation Component Architecture (SCA) framework, and provides a robust mechanism to verify that the your input is both correct and accurate.

BenefitsThe benefits of the IFPStar component include:

• Faster input processing (up to 40% faster for typical large models)

• SORT is not required before interpretation

• Significantly lower memory requirements

• Clearer error messaging

• Rule based definition identifies exact entry and field of illegal data

• Identification of which INCLUDE file / line number of an illegal entry

• Uses Quick Reference Guide rule base

• Provides consistent and rigorous rule checking

• Template definition of fields provides easier implementation of new features, thus reducing development time.

• Higher precision of input tables – 64bit maintained

• The higher precision may cause slightly different answers compared to previous versions or models run with the old Input File Processor (IFP)

• Replaces IFP, RMDUPBK, and XSORT modules with a single more efficient module

• Extensible and Pluggable to in-house or external applications (future)

• Simplifies the process of adding new entries

IFPStar significantly improves developers’ ability to add or modify Nastran bulk data entries in a very short amount of time (less than 3 hours for a very complex entry, comparing to days using legacy IFP module). Rather than ad-hoc parsing rules buried in multiple layers of code, within the application, IFPStar utilizes template-based definitions for entries, which simplifies the process of adding new fields, putting proper checks in place on the fields within an entry, adding boundary conditions, etc …

Another advantage of IFStar component is that it leverages the “component” capability of the SCA framework, which means the component can be used in any SCA enabled application (in house or external), and can easily be updated in the field, without the need for updating the entire installation.

Detailed error messagesThe error messages that are created in the IFPStar component are aimed to better help engineers pinpoint their input errors. For example, let’s assume that a user makes the following mistake in a SPC entry:


SPC1 IO 12 1 3

The correct entry is:SPC1 10 12 1 3

Both IFPStar component and legacy IFP module catch the user’s error, and below is how each one issues the error messages:

Legacy IFP module:*** USER FATAL MESSAGE 315 (IFPDRV)

FORMAT ERROR ON BULK DATA ENTRY SPC1SPC1 IO 12 1

3 *** USER FATAL MESSAGE 316 (IFPDRV)

ILLEGAL DATA ON BULK DATA ENTRY SPC1SPC1 IO 12 1 3

IFPStar Component:*** USER FATAL MESSAGE 9994 (BULKPM)Term Violation for Entity: SPC1 near line 27SID:IO is an illegal integer value.

Potential Behavior DifferencesNote that IFPStar enforces the current Bulk Data rule set in the Quick Reference Guide. This means that legacy models that may have worked in previous versions may fail with IFPStar. Typical causes are illegal input that was not previously trapped, undocumented features or obsolete input entries. Note that there are several entries that are only supported by the IFPStar component in MSC Nastran 2010 and MSC Nastran 2010. These entries include:

PRJCON MAXBRG FSICTRL WETLOAD

WETSURF WETELMG WETELME MATUDS

PRPUDS BCONUDS ELEMUDS QUDS

RCPARM MAT6 MATT6 RADC

RADCT VIEWEX PCONV1 PRIM1

PRIM2 PRIM3 PRIM4 PRIM5

PRIM6 PRIM7 PRIM8 TABLEU1

TABLEU2 ENTUDS RADCOL DTABLE2

MATUSR MATTUSR

MSC Nastran 2010 Release GuideNew Input File Reader - IFPSTAR

140

In RESTART runs, the “/” bulk data has been enhanced to have more user friendly commands. Previously, the only options on the “/” bulk data entry were to remove specific sorted bulk data line numbers, but with IFPStar the delete directives are more user friendly. Please refer to the Quick Reference Guide for more details.

The sorted bulk data (print/punch) will have a slightly different look.

In future versions of MSC Nastran and MSC Nastran new bulk data entries will only be supported by the IFPStar component.

Known IssuesIn addition to the entries listed above, there are a few items that will not work with the IFPStar component. MSGMESH is no longer supported. Basic replication entries are supported, but some advanced replication applications may not work. In this case, it is recommended to use the old IFP processor with the case control command ECHO=PUNCH to generate a bulk data input without replications.

Since this is a brand new component, and Nastran has 30+ years of legacy, while all attempts have been made to support the legacy, it is possible that some client models may not work with-in the bounds of IFPStar component rules and definitions.

When this rare situation occurs with your existing models, you can get around this problem by using the old IFP by adding the following system cell nastran system(444)=0 to the top of your input file (see next paragraph for more details).

Changing the DefaultsThe NASTRAN System cell 444 is reserved to select the input file processor options. SYSTEM(444)=1 is the new default for the IFPStar component, SYSTEM(444)=0 is old IFP. These settings can be changed in the Patran Translation Parameters form and in the SimXpert Generic Solver Parameters form. In Patran 2010 and SimXpert 3.2 and earlier versions, the default can be changed in the Direct Text Input system cell section by adding "NASTRAN SYSTEM(444)=0" - without the quotes. It can also be selected by adding NASTRAN IFPSTAR=YES.

If You Find ErrorsIf you encounter an entry that follows the Quick Reference Guide rules and does not contain undocumented fields, but fails in the IFPStar component, please contact MSC support.


Brake Squeal AnalysisThe brake squeal option available in MSC Nastran 2008 was limited to (1) linear finite elements and (2) a single rotation axis. Since there are quite some applications where quadratic finite elements are preferred to linear finite elements, in MSC Nastran 2010 both linear and quadratic finite elements are supported in a brake squeal analysis. Moreover, in Sol600, the limitation of a single rotation axis has been removed, so that it is possible to model e.g. multiple gear wheel combinations in a single analysis, where per pair of contact bodies a different rotation axis can be specified. To this end, an alternate format of the BSQUEAL entry has been introduced, where the field entry "BODY" signals the start of data for a pair of contact bodies.

MSC Nastran 2010 Release GuideResults and Output Changes

142

Results and Output Changes

Change of Solution Method for Dynamic Enforced Motion Analysis Using SPC/SPCDThe total solution of a dynamic enforced motion analysis using SPC/SPCD can be regarded as a combination of a static enforced motion solution (similar to what is done in SOL 101) and a dynamic enforced motion solution that is relative to this static-based solution. Two methods are employed for obtaining the solution in Nastran. In the total or absolute (TOTAL/ABS) motion method, the program solves directly, in one step, for the TOTAL solution of the dynamic analysis which includes both the static-based solution and the dynamic solution that is relative to the static-based solution. In the relative (REL) motion method, the program obtains the total solution of the dynamic analysis in two steps, by first solving for the static-based solution and then solving separately for the dynamic solution RELATIVE to the static-based solution.

The TOTAL/ABS solution method is computationally more efficient. This is also the only method that is meaningful and that should be employed when a problem involves the use of NOLINi or NLRGAP entries. An important point to note regarding this method is that, for modal dynamic analysis, residual vectors are absolutely critical in order for this method to get correct answers.

The REL solution method, though less efficient, may be more accurate for transient solutions and for modal frequency response solutions at very low forcing frequencies. Also, for modal dynamic analysis, this method is not as critically dependent on residual vectors as the TOTAL/ABS solution method.

In pre-MSC Nastran 2010 versions of Nastran, the TOTAL/ABS method did not support modal damping and fluid structure problems. These problems are now fully resolved. With these enhancements, the TOTAL/ABS and REL solution methods both yield essentially the same results.

In earlier versions of Nastran, the REL method was the implied default solution method. However, because of efficiency and other considerations outlined above, the TOTAL/ABS method has been chosen to be the default solution method in MSC Nastran 2010. (See also related discussion below.)

New ENFMETH ParameterThere is a parameter called ENFMETH that controls the solution method when dynamic enforced motion analysis via SPC/SPCD is used in SOLs 108, 109, 111, 112, 146 and 200. This parameter was an undocumented parameter in earlier versions and was used internally with an implied default value of REL, implying the use of the REL solution method mentioned above. This parameter is now documented in the ENFMETH (p. 723) in the MD/MSC Nastran Quick Reference Guide and the default value has been changed to TOTAL (or ABS) to reflect the new default of the TOTAL/ABS solution method indicated above.

It should be emphasized here that the new ENFMETH parameter is completely separate, independent and distinct from the similarly sounding ENFMOTN parameter. These two parameters should not be confused with each other. The former controls the solution method when dynamic enforced motion analysis via SPC/SPCD is used while the latter controls how the results of such an analysis are output.


Details will be clear from the descriptions of these parameters in the ENFMETH and ENFMOTN (p. 724) in the MD/MSC Nastran Quick Reference Guide.

NEW OPTMIZER Optimization tasks that previously used the BIGDOT optimization algorithm now employ the IPOPT optimization algorithm (See The IPOPT Algorithm (App. C) in the Design Sensitivity and Optimization User’s Guide). These are typically topology optimization tasks or tasks with many design variables. If the user explicitly selects the BIGDOT algorithm using OPTCOD=BIGDOT on the DOPTPRM bulk data entry, the IPOTPT algorithm will be used in its place and a User Information Message will be printed that indicates this. It is expected that the results from the two algorithms will be similar, but not exactly the same.

MSC Nastran 2010 Release GuideMSC Nastran Error List

144

MSC Nastran Error ListThe current error lists for MSC Nastran can be obtained from the MSC Software Simcompanion website: http://simcompanion.mscsoftware.com/infocenter/index?page=content&id=KI8008006

http://simcompanion.mscsoftware.com/infocenter/index?page=content&id=KI8008006

Ap. A: Connectors MSC Nastran 2010 Release Guide

A Connectors

Connectors Output

MSC Nastran 2010 Release GuideConnectors Output

146

Connectors Output

Example A - Element Force and Element Stress

0 SUBCASE 1 FREQUENCY = 1.000000E+01 C O M P L E X F O R C E S I N W E L D E L E M E N T S ( C W E L D P ) (REAL/IMAGINARY) ELEMENT BEND-MOMENT END-A BEND-MOMENT END-B - SHEAR - AXIAL ID PLANE 1 (MZ) PLANE 2 (MY) PLANE 1 (MZ) PLANE 2 (MY) PLANE 1 (FY) PLANE 2 (FZ) FORCE FX TORQUE MX 30 -1.808122E-01 -1.966379E-02 -3.561784E-03 2.373087E-03 -3.979684E-01 -4.947790E-02 2.768694E-02 -2.765884E-02 2.170022E-02 2.359719E-03 4.290746E-04 -2.845342E-04 4.775866E-02 5.936962E-03 -3.321101E-03 3.320390E-030 SUBCASE 1 FREQUENCY = 2.000000E+01 C O M P L E X F O R C E S I N W E L D E L E M E N T S ( C W E L D P ) (REAL/IMAGINARY) ELEMENT BEND-MOMENT END-A BEND-MOMENT END-B - SHEAR - AXIAL ID PLANE 1 (MZ) PLANE 2 (MY) PLANE 1 (MZ) PLANE 2 (MY) PLANE 1 (FY) PLANE 2 (FZ) FORCE FX TORQUE MX 30 -1.808791E-01 -1.966534E-02 -3.602137E-03 2.367348E-03 -3.980280E-01 -4.946850E-02 2.765457E-02 2.769116E-02 2.171652E-02 2.360096E-03 4.389053E-04 -2.831364E-04 4.777317E-02 5.934670E-03 -3.313215E-03 3.328268E-030 SUBCASE 1 FREQUENCY = 1.000000E+01 C O M P L E X S T R E S S E S I N W E L D E L E M E N T S ( C W E L D P ) (REAL/IMAGINARY)

ELEMENT AXIAL MAX STRESS MIN STRESS MAX STRESS MIN STRESS MAXIMUM BEARING ID STRESS END-A END-A END-B END-B SHEAR STRESS STRESS 30 3.525211E+00 2.045557E+03 -2.038507E+03 6.397730E+01 -5.692688E+01 2.092356E+02 4.010323E+01 -4.228557E-01 2.446496E+02 -2.454954E+02 6.845903E+00 -7.691615E+00 2.511545E+01 4.812626E+000 SUBCASE 1 FREQUENCY = 2.000000E+01 C O M P L E X S T R E S S E S I N W E L D E L E M E N T S ( C W E L D P ) (REAL/IMAGINARY)

ELEMENT AXIAL MAX STRESS MIN STRESS MAX STRESS MIN STRESS MAXIMUM BEARING ID STRESS END-A END-A END-B END-B SHEAR STRESS STRESS 30 3.521090E+00 2.046250E+03 -2.039208E+03 6.432576E+01 -5.728358E+01 2.094103E+02 4.010902E+01 -4.218517E-01 2.448205E+02 -2.456642E+02 6.932804E+00 -7.776507E+00 2.515803E+

147Appendix AConnectors

Example B - Element Stress Output

S T R E S S E S I N S E A M E L E M E N T S ( C S E A M )0 CORNER ------CENTER AND CORNER POINT STRESSES--------- DIR. COSINES MEANELEMENT-ID GRID-ID NORMAL SHEAR PRINCIPAL -A- -B- -C- PRESSURE VON MISES 10001 0GRID CS 8 GP CENTER X 7.747896E+05 XY -3.827610E+03 A 7.748116E+05 LX 1.00 0.00-0.01 -2.940678E+05 7.270885E+05 Y 1.074136E+05 YZ 1.673470E-10 B 2.502929E-09 LY-0.01 0.00-1.00 Z 2.328306E-09 ZX 1.746230E-09 C 1.073916E+05 LZ 0.00 1.00 0.00 90001 X 1.711155E+06 XY -7.739222E+03 A 1.711200E+06 LX 1.00 0.00-0.01 -8.162711E+05 1.342475E+06 Y 3.764584E+05 YZ -3.868169E+03 B 3.602726E+05 LY-0.01 0.23-0.97 Z 3.611997E+05 ZX -5.157554E+01 C 3.773406E+05 LZ 0.00 0.97 0.23 90002 X 1.707039E+06 XY -7.739222E+03 A 1.707084E+06 LX 1.00 0.00-0.01 -8.103257E+05 1.345162E+06 Y 3.646422E+05 YZ 3.868169E+03 B 3.572576E+05 LY-0.01-0.47-0.88 Z 3.592958E+05 ZX -5.157554E+01 C 3.666358E+05 LZ 0.00 0.88-0.47 90003 X -1.638654E+05 XY 8.400235E+01 A -1.563204E+05 LX 0.01 0.00-1.00 2.261826E+05 1.983753E+05 Y -1.563954E+05 YZ 3.868169E+03 B -3.583610E+05 LY 1.00-0.02 0.01 Z -3.582869E+05 ZX -5.157554E+01 C -1.638663E+05 LZ 0.02 1.00 0.00 90004 X -1.655750E+05 XY 8.400235E+01 A -1.580980E+05 LX 0.01 0.00-1.00 2.286519E+05 2.005495E+05 Y -1.581723E+05 YZ -3.868169E+03 B -3.622819E+05 LY 1.00 0.02 0.01 Z -3.622085E+05 ZX -5.157554E+01 C -1.655759E+05 LZ-0.02 1.00 0.00 90005 X 1.727548E+06 XY -7.739222E+03 A 1.727593E+06 LX 1.00 0.00-0.01 -8.258101E+05 1.352764E+06 Y 3.830623E+05 YZ -3.868169E+03 B 3.659432E+05 LY-0.01 0.22-0.98 Z 3.668194E+05 ZX 5.157554E+01 C 3.838939E+05 LZ 0.00 0.98 0.22 90006 X 1.722568E+06 XY -7.739222E+03 A 1.722612E+06 LX 1.00 0.00-0.01 -8.186157E+05 1.356026E+06 Y 3.703815E+05 YZ 3.868169E+03 B 3.612509E+05 LY-0.01-0.39-0.92 Z 3.628978E+05 ZX 5.157554E+01 C 3.719841E+05 LZ 0.00 0.92-0.39 90007 X -1.698541E+05 XY 8.400235E+01 A -1.598039E+05 LX 0.01 0.00-1.00 2.312129E+05 1.993407E+05 Y -1.598779E+05 YZ 3.868169E+03 B -3.639800E+05 LY 1.00-0.02 0.01 Z -3.639066E+05 ZX 5.157554E+01 C -1.698548E+05 LZ 0.02 1.00 0.00 90008 X -1.706989E+05 XY 8.400235E+01 A -1.607165E+05 LX 0.01 0.00-1.00 2.324332E+05 2.003621E+05 Y -1.607901E+05 YZ -3.868169E+03 B -3.658836E+05 LY 1.00 0.02 0.01 Z -3.658106E+05 ZX 5.157554E+01 C -1.706996E+05 LZ-0.02 1.00 0.00 10002 0GRID CS 8 GP CENTER X 7.406084E+05 XY 9.462977E+02 A 7.406120E+05 LX 1.00 0.00 0.00 -2.397683E+05 7.514921E+05 Y -2.130364E+04 YZ 4.292815E-10 B -2.130482E+04 LY 0.00 1.00 0.00 Z -1.583248E-08 ZX 1.333328E+03 C -2.400266E+00 LZ 0.00 0.00-1.00 90009 X 1.623490E+06 XY 1.970375E+03 A 1.623494E+06 LX 1.00 0.00 0.00 -6.671043E+05 1.446252E+06 Y 8.306726E+04 YZ 9.464031E+02 B 8.306052E+04 LY 0.00 1.00 0.00 Z 2.947562E+05 ZX 1.612613E+03 C 2.947584E+05 LZ 0.00 0.00-1.00 90010 X 1.623357E+06 XY 1.970375E+03 A 1.623361E+06 LX 1.00 0.00 0.00 -6.669126E+05 1.446268E+06 Y 8.317433E+04 YZ -9.464031E+02 B 8.316755E+04 LY 0.00 1.00 0.00 Z 2.942068E+05 ZX 1.612613E+03 C 2.942091E+05 LZ 0.00 0.00-1.00 90011 X -1.337316E+05 XY -7.777978E+01 A -1.233649E+05 LX-0.01-0.01-1.00 1.839953E+05 1.666055E+05 Y -1.233709E+05 YZ -9.464031E+02 B -2.949047E+05 LY 1.00 0.01-0.01 Z -2.948834E+05 ZX 1.612613E+03 C -1.337162E+05 LZ-0.01 1.00-0.01


148

S T R E S S E S I N S E A M E L E M E N T S ( C S E A M )0 CORNER ------CENTER AND CORNER POINT STRESSES--------- DIR. COSINES MEANELEMENT-ID GRID-ID NORMAL SHEAR PRINCIPAL -A- -B- -C- PRESSURE VON MISES 90012 X -1.333445E+05 XY -7.777978E+01 A -1.228786E+05 LX-0.01-0.01-1.00 1.834361E+05 1.662438E+05 Y -1.228843E+05 YZ 9.464031E+02 B -2.941010E+05 LY 1.00-0.01-0.01 Z -2.940796E+05 ZX 1.612613E+03 C -1.333288E+05 LZ 0.01 1.00-0.01 90013 X 1.608190E+06 XY 1.970375E+03 A 1.608193E+06 LX 1.00 0.00 0.00 -6.591450E+05 1.435601E+06 Y 7.755697E+04 YZ 9.464031E+02 B 7.755026E+04 LY 0.00 1.00 0.00 Z 2.916883E+05 ZX 1.054044E+03 C 2.916916E+05 LZ 0.00 0.00-1.00 90014 X 1.608294E+06 XY 1.970375E+03 A 1.608298E+06 LX 1.00 0.00 0.00 -6.592958E+05 1.435494E+06 Y 7.790112E+04 YZ -9.464031E+02 B 7.789438E+04 LY 0.00 1.00 0.00 Z 2.916921E+05 ZX 1.054044E+03 C 2.916954E+05 LZ 0.00 0.00-1.00 90015 X -1.357683E+05 XY -7.777978E+01 A -1.230557E+05 LX-0.01-0.01-1.00 1.835484E+05 1.627915E+05 Y -1.230615E+05 YZ -9.464031E+02 B -2.918279E+05 LY 1.00 0.01-0.01 Z -2.918155E+05 ZX 1.054044E+03 C -1.357617E+05 LZ-0.01 1.00-0.01 90016 X -1.356183E+05 XY -7.777978E+01 A -1.228063E+05 LX-0.01-0.01-1.00 1.833317E+05 1.627466E+05 Y -1.228120E+05 YZ 9.464031E+02 B -2.915773E+05 LY 1.00-0.01-0.01 Z -2.915649E+05 ZX 1.054044E+03 C -1.356115E+05 LZ 0.01 1.00-0.01 10003 0GRID CS 8 GP CENTER X 1.284104E+03 XY -4.129671E+01 A 1.285049E+03 LX 1.00 0.02 0.00 -2.544763E+02 1.610510E+03 Y -5.206752E+02 YZ 6.457412E-10 B -5.216197E+02 LY-0.02 1.00 0.00 Z -2.142042E-08 ZX 8.847564E-08 C -2.142039E-08 LZ 0.00 0.00-1.00 90017 X 2.416318E+03 XY -5.225373E+01 A 2.417096E+03 LX 1.00 0.01 0.00 -4.974160E+02 3.080820E+03 Y -1.092519E+03 YZ -4.401031E+01 B -1.094830E+03 LY-0.01 1.00 0.03 Z 1.684483E+02 ZX -5.868064E-01 C 1.699818E+02 LZ 0.00 0.03-1.00 90018 X 2.495232E+03 XY -5.225373E+01 A 2.496019E+03 LX 1.00 0.02 0.00 -6.114019E+02 3.040780E+03 Y -9.753662E+02 YZ 4.401031E+01 B -9.776510E+02 LY-0.02 1.00-0.03 Z 3.143401E+02 ZX -5.868064E-01 C 3.158376E+02 LZ 0.00-0.03-1.00 90019 X -8.759534E+01 XY -3.033968E+01 A -2.349008E+01 LX-0.42-0.08-0.91 1.068964E+02 1.504556E+02 Y -5.015059E+01 YZ 4.401031E+01 B -1.968506E+02 LY 0.88-0.30-0.38 Z -1.829433E+02 ZX -5.868064E-01 C -1.003486E+02 LZ 0.24 0.95-0.19 90020 X -1.375186E+02 XY -3.033968E+01 A -7.612272E+01 LX-0.44 0.04-0.90 1.790078E+02 2.061600E+02 Y -9.965966E+01 YZ -4.401031E+01 B -3.093700E+02 LY 0.88 0.21-0.42 Z -2.998451E+02 ZX -5.868064E-01 C -1.515307E+02 LZ-0.17 0.98 0.13 90021 X 2.982642E+03 XY -5.225373E+01 A 2.983355E+03 LX 1.00 0.01 0.00 -8.526937E+02 3.380599E+03 Y -8.465572E+02 YZ -4.401031E+01 B -8.487935E+02 LY-0.01 1.00 0.03 Z 4.219966E+02 ZX 5.868065E-01 C 4.235193E+02 LZ 0.00 0.03-1.00 90022 X 2.948929E+03 XY -5.225373E+01 A 2.949649E+03 LX 1.00 0.01 0.00 -8.039978E+02 3.369357E+03 Y -8.420306E+02 YZ 4.401031E+01 B -8.444359E+02 LY-0.01 1.00-0.04 Z 3.050948E+02 ZX 5.868065E-01 C 3.067800E+02 LZ 0.00-0.04-1.00 90023 X -2.039378E+02 XY -3.033968E+01 A -1.402171E+02 LX-0.43-0.02-0.90 2.671824E+02 2.728479E+02 Y -1.611178E+02 YZ 4.401031E+01 B -4.434710E+02 LY 0.90-0.16-0.42 Z -4.364915E+02 ZX 5.868065E-01 C -2.178590E+02 LZ 0.13 0.99-0.09 90024 X -1.412352E+02 XY -3.033968E+01 A -7.500764E+01 LX-0.41 0.04-0.91 1.766120E+02 1.980148E+02 Y -9.800105E+01 YZ -4.401031E+01 B -3.004030E+02 LY 0.89 0.22-0.39 Z -2.905998E+02 ZX 5.868065E-01 C -1.544254E+02 LZ-0.18 0.98 0.12

S T R E S S E S I N S E A M E L E M E N T S ( C S E A M )0 CORNER ------CENTER AND CORNER POINT STRESSES--------- DIR. COSINES MEANELEMENT-ID GRID-ID NORMAL SHEAR PRINCIPAL -A- -B- -C- PRESSURE VON MISES 10004 0GRID CS 8 GP CENTER X 2.857427E+00 XY -5.724032E-02 A 7.595094E+00 LX-0.01 0.00-1.00 -3.483943E+00 6.644488E+00 Y 7.594403E+00 YZ -5.420588E-10 B 1.257286E-08 LY 1.00 0.00-0.01 Z 1.257285E-08 ZX -6.053597E-09 C 2.856735E+00 LZ 0.00 1.00 0.00 90025 X 1.577084E+00 XY 5.002377E+00 A 6.410882E+00 LX 0.72-0.69 0.05 -2.286139E+00 9.076537E+00 Y 1.218060E+00 YZ 1.910919E-01 B -3.610427E+00 LY 0.69 0.72 0.03 Z 4.063272E+00 ZX 2.547832E-03 C 4.057961E+00 LZ 0.06-0.02-1.00 90026 X -5.694098E+00 XY 5.002377E+00 A -8.618130E-01 LX 0.72-0.04 0.69 8.216681E+00 1.117548E+01 Y -6.043215E+00 YZ -1.910919E-01 B -1.292361E+01 LY 0.69 0.06-0.72 Z -1.291273E+01 ZX 2.547832E-03 C -1.086462E+01 LZ-0.01 1.00 0.07 90027 X 3.802032E+00 XY -5.116858E+00 A 1.589757E+01 LX-0.39 0.01-0.92 -4.467181E+00 1.785942E+01 Y 1.373110E+01 YZ -1.910919E-01 B -4.134058E+00 LY 0.92 0.01-0.39 Z -4.131593E+00 ZX 2.547832E-03 C 1.638033E+00 LZ-0.01 1.00 0.01 90028 X 1.120986E+01 XY -5.116858E+00 A 2.345315E+01 LX-0.39 0.92-0.02 -1.516737E+01 1.288174E+01 Y 2.131121E+01 YZ 1.910919E-01 B 9.069369E+00 LY 0.92 0.39 0.01 Z 1.298105E+01 ZX 2.547832E-03 C 1.297959E+01 LZ 0.02-0.02-1.00 90029 X -5.801891E+00 XY 5.002377E+00 A -9.705124E-01 LX 0.72 0.04 0.69 8.360745E+00 1.124266E+01 Y -6.152861E+00 YZ 1.910919E-01 B -1.313789E+01 LY 0.69-0.05-0.72 Z -1.312748E+01 ZX -2.547842E-03 C -1.097383E+01 LZ 0.01 1.00-0.06 90030 X 1.536345E+00 XY 5.002377E+00 A 6.378676E+00 LX 0.72-0.69-0.05 -2.238929E+00 9.062510E+00 Y 1.195284E+00 YZ -1.910919E-01 B -3.641881E+00 LY 0.69 0.72-0.03 Z 3.985157E+00 ZX -2.547842E-03 C 3.979991E+00 LZ-0.06 0.02-1.00 90031 X 1.171584E+01 XY -5.116858E+00 A 2.348970E+01 LX-0.40 0.92 0.03 -1.534583E+01 1.260059E+01 Y 2.126247E+01 YZ -1.910919E-01 B 9.489802E+00 LY 0.92 0.40-0.01 Z 1.305916E+01 ZX -2.547842E-03 C 1.305798E+01 LZ-0.02 0.02-1.00 90032 X 4.514247E+00 XY -5.116858E+00 A 1.643190E+01 LX-0.39-0.01-0.92 -4.943524E+00 1.805928E+01 Y 1.423316E+01 YZ 1.910919E-01 B -3.919224E+00 LY 0.92-0.01-0.39 Z -3.916836E+00 ZX -2.547842E-03 C 2.317893E+00 LZ 0.01 1.00-0.01

149Appendix AConnectors


150

Nastran 2010 Doc Release

Documents

equivalent

normalized

begin superseid

comma separated

upper hessenberg

equivalent

high performance

bulk data