technology; and integration of analysis programs into CAD/CAE and concurrent engineering systems. 24 ...... sheet metal, and the impact of a nose cone.
L
n
ructures ames
sion
-- THP, U-N92-25935
Uncl HI/39
Js
0088975
L
J,
_:2_
_
i
:, _:-
:_=--_
NASA
Conference
Publication
3142
Computational Structures Technology for Airframes and Propulsion Systems Compiled by Ahmed K. Noor Center
for Computational
Structures Technology University of Virginia Hampton, Virginia Jerrold
NASA
NASA
M. Housner
and James H. Statues, Jr. Langley Research Center Hampton, Virginia Dale A. Hopkins and Christos C. Chamis Lewis Research Center Cleveland,
Ohio
Proceedings of two workshops sponsored by the National Aeronautics and Space Administration, Washington, D.C., and the Center for Computational Structures Technology, University of Virginia, Hampton, Virginia, and held at Lewis Research Center, Cleveland,Ohio, June 26-27, 1991, and at Langley Research Center, Hampton, Virginia, September 4-5, 1991
National Aeronautics and Space Administration Office of Management Scientific and Technical Information Program 1992
PREFACE This document contains the proceedings of the two Workshops on Computational Structures Technology for Airframes and Propulsion Systems. The Workshops were jointly sponsored by the Center for Computational Structures Technology of the University of Virginia and NASA. The first workshop was held on June 26-27, 1991 at NASA Lewis Research Center and focused on computational technology for advanced propulsion systems. The second workshop was held on September 4-5, 1991 at NASA Langley Research Center and focused on computational technology for airframes. The attendees of the workshops came from government agencies, airframe and engine manufacturers and universities. The objectives of the workshops were to assess the status of CST in the aerospace industry, to identify the technical needs in the CST area, and to provide guidelines for focused future research leading to an enhanced capability for future national programs, such as the High-Speed Civil Transport and the National Aerospace Plane. Certain materials and products are identified in this publication in order to specify adequately the materials and products that were investigated in the research effort. In no case does such identification imply recommendations or endorsement of products by NASA nor does it imply that the materials and products are the only ones or the best ones available for the purpose. In many cases equivalent materials and products are available and would probably produce equivalent results.
Ahmed K. Noor Center for Computational University of Virginia
Structures
Technology
Jerrold M. Housner and James H. Starnes, NASA Langley Research Center Hampton, Virginia
Jr.
and Dale A. Hopkins and Christos NASA Lewis Research Center Cleveland, Ohio
C. Chamis
pF_C.EOK._G PAGE
m
BLANK
NOT
FILMED
CONTENTS °°.
PREFACE
....................................
ATTENDEES
111
..................................
vi_
INTRODUCTION
'
COMPUTATIONAL
STRUCTURES
Ahmed
K. Noor
COMPUTER
CODES
Christos
Pramote
CSM
DEVELOPED
IN INTEGRATED ....................................
AND
UNDER
ANALYSIS
WITH
AND
UVA
DEVELOPMENT
FOR
CST
AT NASA
....
LEWIS
5
43
ADAF17VE
UNSTRUCTURED 59
OF COMPUTATIONAL AT NASA LEWIS
STRUCTURES TECHNOLOGY RESEARCH CENTER ............
81
A. Hopkins
ACTIVITIES Jerrold
AT THE
NASA
LANGLEY
RESEARCH
CENTER
...........
91
M. Housner
OVERVIEW OF MECHANICS OF MATERIALS BRANCH COMPUTATIONAL STRUCTURES AREA ...................... C. C. Poe, ANALYSIS
ACTIVITIES
IN THE 121
Jr.
AND
James HIGH
CENTER
Dechaumphai
OVERVIEW ACTIVITIES
Dale
TECHNOLOGY
1
C. Chamis
PROGRESS MESHING
A BRIEF RELATED
" ..............
DESIGN
H. Stames,
SPEED
CML
Jr.
TECHNOLOGY and Charles
TRANSPORT
FOR
HIGH-SPEED
AIRCRAFT
STRUCTURES
. 137
J. Camarda .
.
.- ......................
173
R. L. McKnight STRUCTURES AEROSPACE
TECHNOLOGY PLANE
APPLICATIONS
FOR
THE
NATIONAL ...........
189
...........
209
T. E. Little m
i
LARGE
SCALE
Vipperla LIGHT
OPTIMIZATION
AN OVERVIEW
AND
MATERIALS
STRUCTURES
FOR
HIGH
SPEED
FLIGHT
.....
231
A. Thornton
MODELING
"BR_VVFLE"
K. T. Kedward, ALGORITHMIC Frank
ASTROS:
B. Venkayya
THERMAL
Earl
USING
HIGH-TEMPERATURE
R. M. McMeeking
DEVELOPMENT
COMPOSITE
STRUCTURES
......
253
and S. Janson
IN STRUCTURES
TECHNOLOGY
...........
269
Sagendorph V
PRE_ED_,r_
PAGE
BLA,"!K NOT
FILMLCD
STRUCTURAL Johnny
MECHANICS
SIMULATIONS
......................
H. Biffle
STRUCTURAL ANALYSIS FOR PRELIMINARY CIVIL TRANSPORT (HSCT) ............................. Kumar
303
DESIGN
OF HIGH
SPEED 321
G. Bhatia
APPLICATION OF INTEGRATED STRUCTURAL CIVIL TRANSPORT .................................
ANALYSIS
TO THE
HIGH
SPEED 335
C. R. Saff OVERVIEW MILITARY
OF COMPUTATIONAL STRUCTURAL AIRCRAFF ...............................
METHODS
FOR
MODERN 395
J. N. Kudva A PERSPECTIVE STRUCTURES Frank
ON TECHNICAL NEEDS IN COMPUTATIONAL TECHNOLOGY ............................
F. Abdi,
Gregory
COMPUTATIONAL PRACTICE]FUTURE Allan AIRFRAME
and Kenneth
STRUCTURES TECHNOLOGY NEEDS .........
B. Pifko LIFE
L. Savoni
and Harvey
375
J. Newell AT GRUMMAN-CURRENT ....................
395
Eidinoff
PREDICTION
............................
431
G. P. Sendeckyj PROBABILISTIC MAIN ENGINES Kevin
DESIGN APPLICATIONS .................................
FOR
THE
SPACE
TRANSPORTATION 447
O'Hara
COMPUTATIONAL STRUCTURAL ANALYSIS COMMERCIAL ENGINES ..............................
AND
ADVANCED 473
R. B. Wilson MILITARY Daniel
ENGINE
STRUCTURES
TECHNOLOGY
..........
491
E. Thomson
COMPUTATIONAL Bruce
COMPUTATIONAL
STRUCTURES
TECHNOLOGY
ENGINE/AIRFRAME
C. McClintick
vi
COUPLJ2qG
....
507
Specialty Technology
Workshops on for Airframes
Computational Structures and Propulsion Systems
Attendees Prof. Mr. Frank Abdi D/734-11-GB 13 Rockwell International 201 North Douglas Street E1 Segundo, CA 90245 (213) 922-0990; Fax (213)
Keith
Mechanical Engineering Department University of California Santa Barbara, CA 93106 (805) 893-3381; Fax (805) 893-8651 Dr. N. J. Kudva
414-0430
MS Dept. 3852-MF Northrop Aircraft Division Northrop Corporation One Northrop Avenue Hawthorne, CA 90250-3277 (213) 332-8300; Fax (213)
Dr. Kumar Bhatia Mail Stop 3T-AF Boeing Corporation P.O. Box 3707 Seattle, WA 98124 (206) 393-6993; Fax (206) 477-2345
Mr. Bruce Dr. Johnny H. Biffle Applied Mechanics Ill, Div. Sandia National Laboratory P.O. Box 5800 Albuquerque, NM 87185 (505) 844-5385; Fax (505) Dr. Chrisms
583-5170
846-9833 Mr. R. L. McKnight Mail Drop A333 General Electric Aircraft Engines P.O. Box 156301 Cincinnati, OH 45215 (513) 583-5068; Fax (513) 583-5170
Center
2,1000 Brookpark Road Cleveland, OH 44135 (216) 433-3252; Fax (216) 433-8011 Dr. Pramote Dechaumphai Mail Stop 395 NASA Langley Research Hampton, VA 23665 (804) 864-1357
332-5853
C. McClintick
Mail Drop A33A General Electric Company Highway 75 Cincinnati, OH 45215 (513) 583-5146; Fax (513)
1523
C. Chamis
Marl Stop 49-8 NASA Lewis Research
T. Kedward
Mr. James E. Newell Mail Code D/545-055-J'B 11 Rockwell International 6633 Canoga Park Canoga Park, CA 91303 (818) 773-5505; Fax (818) 773-5542
Center
Mr. Dale A. Hopkins Mail Stop 4%8 NASA Lewis Research Center 21000 Brookpark Road Cleveland, OH 44135 (216) 433-3260; Fax (216) 433-8011
Prof.
Ahmed
K. Noor
Dr. Jerrold M. Housner Mail Stop 240 NASA Langley Research Hampton, VA 23665 (804) 864-2906
Mr. Kevin T. O'Hara Mail Code D/611-055-RA02 Rockwell International 2227 Drake Avenue, Suite 45 Huntsville, AL 35805 (205) 880-4519; Fax (205) 880-4596
Center for Computational Structures Technology Mail Stop 210 NASA Langley Research Center Hampton, VA 23665 (804) 864-1978; Fax (804) 864-8089
Center
vii
Dr. Allan B. Pifko Mail StopA08-35 ResearchDepartment GrummanCorporation Bethpage,NY 11714 (516)575-1965;Fax (516)575-7716
Mr. Samuel Code RM
NASA Headquarters Washington, D.C. 20546 (202) 453-2760; Fax (202) 755-4068 Mr. Raymond B. Wilson Mail Stop 163-10 Engineering Division Pratt & Whimey 400 Main Street East Hartford, CT 06108 (203) 565-2901; Fax (203)
Mr. ClarenceC. Poe,Jr. Mail Stop 188E NASA LangleyResearchCenter Hampton,VA 23665 (804)864,3467 Dr. CharlesR. Saff Mail Code1021322 McDonnellDouglas P.O.Box 516 St. Louis,MO 63166 (314)233-8623;Fax(314)777-1171 Mr. FrankE. SagendorphIV Mail DropA333 GeneralElectricAircraftEngines P.O.Box 156301 Cincinnati,OH 45215 (513)583-5001;Fax (513)583-5170 Dr. GeorgeP. Sendeckyj WL/FIBEC Wright PattersonAir ForceBase,OH 45433 (513)255-6104;Fax (513)255-3717 Dr. JamesH. Starnes,Jr. Mail Stop190 NASA LangleyResearchCenter Hampton,VA 23665 (804)864-3168;Fax (804)864-7791 Mr. DanielE. Thomson WL/POTC Wright PattersonAir ForceBase,OH 45433 (513)255-2081;Fax (513)476-4531 Prof. Earl
A. Thornton
Light Thermal Structures Center School of Engineering and Applied Thornton Hall
L. Venneri
Science
University of Virginia Charlottesville, VA 22903 (804) 924-6291; Fax (804) 982-2037 Dr. V. B. Venkayya WI./FIBR Wright Patterson Air Force Base, OH 45433 (513) 255-7191; Fax (513) 255-3740 viii
565-9615
INTRODUCTION Performance requirements for future flight vehicles are rapidly increasing due to ambitious objectives of the U.S. civil and military aerospace programs. In aeronautics, future goals include higher cruising speeds, altitudes and thrust-to-weight ratios. The technology drivers for future aircraft include reduction in material, fabrication and maintenance costs; reduction in weight; extended life; higher operating temperature; and signature reduction. In space, future goals include lower transportation costs to space; long-duration space flights; planetary missions; and extraterrestrial bases. To successfully achieve the performance requirements space systems major advances are needed in: 1) computational and engineered materials such as high-temperature composites solution of coupled multidiscipline problems; 4) computational 5) accurate quantification of risk. The timely development essential to insure U.S. superiority in the aerospace field.
for planned and h, ture aeronautical and structures technology (CST); 2) advanced and advanced metallics; 3) formulation and simulation of concurrent engineering; and and deployment of these technologies is
Several national programs such as High-Speed Civil Transport (HSCT), National Aerospace Plane (NASP), National Launch System (NLS), and Integrated High-Performance Turbine Engine Technology (IHPTET) need major advances in a number of key areas of computational structures technology. To this end, there are a number of primary pacing items and related tasks that must be addressed by the research community. The joint NASA/University of Virginia Workshops held at NASA Lewis Research Center, June 26-27, 1991 and at NASA Langley Research Center, September 4-5, 1991 focused on the status of computational structures technology and the pacing items of this technology. The list of pacing items given in this introduction was compiled from a number of participants. It is anticipated that the items in the list can impact the design and operation of future flight vehicles in the following four ways: 1) by providing better understanding of the phenomena associated with response, failure and life, thereby identifying the desirable structural design attributes; 2) by improving the productivity of the design team, and reducing the response time to resolve operability problems; 3) by verifying and certifying designs, and making low-cost design modifications during the design process; and 4) by allowing major improvements and innovations in the design process so as to achieve a fully integrated design in a concurrent engineering environment. In such an environment, computer simulation is made of the entire life cycle of the flight vehicle including material selection and processing, multidisciplinary design, automated manufacturing and fabrications, quality assurance, certification, operation, health monitoring and control (e.g., maintenance and repairs), retirement and disposal. The ultimate aim of CST research is to impact the fully integrated design process. Primary.
Pacing
Items
The primary pacing items identified by the participants can be grouped into the following six headings: 1) computational material modeling; 2) fail.ure and life prediction methodologies; 3) hierarchical, integrated multiple methods and adaptive modeling techniques; 4) probabilistic analysis, stochastic modeling and risk assessment; 5) validation of numerical simulations; and 6) multidisciplinary analysis and design optimization. For each of the aforementioned items attempts should be made to exploit the major characteristics of high-performance computing technologies, as well as the future computing environment. The six primary pacing items are described subsequently. Note that some of the tasks within the pacing items are of generic nature, others are specific to either propulsion systems or airframes. 1. Computational Material Modeling. The reliability of the predictions of response, failure and life of structures is critically dependent on the accurate characterization and modeling of material behavior. The simple material models used to date are inadequate for many of the future applications, especially those involving severe environment (e.g., high temperatures). Needed work on material modeling'can be grouped in two general areas', a) modeling the response and damage of advanced material systems in the actual operating environment of future flight vehicles; and b) numerical simulation of manufacturing (fabrication) processes.
Advancedmaterialsystemsincludenewpolymercomposites,metalmatrix composites,ceramic composites,carbon/carbonandadvancedmetallics. The length scaleselectedin the model must be adequatefor capturing the responsephenomenaof interest (e.g., micromechanics,mesomechanics, macromechanics).For materialsusedin propulsionsystems,work is neededon themodelingof damage accumulationandpropagationto fracture;modelingof thermoviscoplasticresponse,thermal-mechanical cycling andratcheting;andpredictionof long-termmaterial behaviorfrom short-termdata, which are particularlyimportant. 2. Failure and Life Prediction Methodologies. Practical numerical techniques are needed for predicting the life, as well as the failure initiation and propagation in structural components made of new, high-performance materials in terms of measurable and controllable parameters. Examples of these materials are high-temperature materials for hypersonic vehicles; piezoelectric composites; and electronic, optical, and smart materials for space applications. For some of the materials, accurate constitutive descriptions, failure criteria, damage theories, and fatigue data are needed, along with more realistic characterization of interface phenomena (such as contact and friction). The constitutive descriptions may require investigations at the microstructure level or even the atomic level, as well as carefully designed and conducted experiments. Failure and life prediction of structures made of these materials is difficult and numerical models often constructed under restricting assumptions may not capture the dominant and underlying physical failure mechanisms. Moreover, material failure and structural response (such as instability) often couple in the failure mechanism. 3. Hierarchical. Inte_m'ated Multiple Methods and Adaptive Modeling Techniques. The effective use of numerical simulations for predicting the response, life, performance and failure of future flight vehicles requires strategies for treating phenomena occurring at disparate spatial and time scales, using reasonable computer resources. The strategies are based on using multiple mathematical models in different regions of the structure to take advantage of efficiencies gained by matching the model to the expected response in each region. To achieve the full potential of hierarchical modeling, there should be minimal reliance on a priori assumptions about the response. This is accomplished by adding adaptivity to the strategy. The key tasks of the research in this area are the following: 1) simple
2)
design-oriented
3) simulation
of local phenomena
4) automated
(or semiautomated)
6)
stages
of the design
process
and adaptive
through
global/local
methodologies
coupling
of different
mathematical/discrete
modeling
models
strategies
high fidelity modeling of details (such as damping, material nonlinearities, joints). For propulsion systems, this may require, among other things, efficient full three-dimensional multi-load analyses;
7) efficient 8) sensitivity neglected 4. developed conditions,
for use in the early
rational selection of a set of nested mathematical models for different regions, and discretization techniques for use in conjunction with the mathematical models. This, in turn, requires the availability of a capability for holistic modeling from micro to structural response with varying degrees of accuracy.
5) error estimation
2
models
methods
for engine
airframe
and rotor/engine-frame
analysis to assess the sensitivity of the response in the current mathematical model.
coupling. to each of the parameters
Probabilistic Analysis. Stochastic Modeling and Risk Assessment. The new methodology for treating general forms of uncertainties in geometry, material properties, boundary loading, and operational environment in the structural analysis formulation of structural
components to quantify
needs to be extended inherent uncertainties
to probabilistic design/risk assessment in the response of flight vehicles
of full flight vehicles. is obviously of great
The ability advantage.
However, the principal benefit of using any stochastic method consists of the insights into engineering, safety, and economics that are gained in the process of arriving at those quantitative results and carrying out reliability analyses. As future flight-vehicle structures become more complicated, failure mechanisms will be probabilistically modeled from the beginning of the design process, and potential design improvements will be evaluated to assess their effects on reducing overall risk. The results, combined with economic considerations, will be used in systematic cost-benefit analyses (perhaps also done on a probabilistic basis) to determine the structural design with the most acceptable balance of cost and risk. 5. Validation of Numerical Simulations. In addition to selecting a benchmark set of flight-vehicle structures for assessing new computational strategies and numerical algorithms, a high degree of interaction and communication is needed between computational modelers and experimentalists. This is done on four different levels, namely, 1) laboratory tests on small specimens to obtain material data; 2) component tests to validate computational models; 3) full-scale tests to validate the modeling of details; and 4) flight tests to validate the entire modeling process. 6. Multidisciplinary_ Analysis and Design Optimization. The realization of new complex aerospace vehicles requires integration between the structures discipline and other traditionally separate disciplines such as aerodynamics, propulsion and control. This is mandated by significant interdisciplinary interactions and couplings which need to be accounted for in predicting response, as well as in optimal design of these vehicles. Examples are the couplings between the aerodynamic flow field, structural heat transfer, and structural response of high-speed aircraft and propulsion systems; and the couplings between the control system and structural response in control-configured aircraft and spacecraft. This activity also includes design optimization with multi-objective functions (e.g., performance, durability, integrity, reliability and cost), and multi-scale structural tailoring (micro, local, and global levels). For propulsion systems it also includes design with damping for high-cycle fatigue, low-cycle-fatigue, and acoustic fatigue. Typically, in the design process questions arise regarding influence of design variable changes on system behavior. Answers to these questions, quantified by the derivatives of behavior with respect to the design variables or by parametric studies, guide design improvements toward a better overall system. In large applications this improvement process is executed by numerical optimization, combined with symbolic/AI techniques, and human judgement aided by data visualization. Efficiency of the computations that provide data for such a process, is decisive for the depth, breadth, and rate of progress achievable, and hence, ultimately, is critical for the final product quality. R¢li_*;cd Tasks For CST to impact the design process, the following three tasks need to be addressed by the research community: 1) development of automated or semi-automated model (and mesh) generation facilities; 2) pre- and postdata processing and use of advanced visualization technology; 3) adaptation of AI tools (knowledge-based/expert systems and neural networks) to CST.
3
N9e- 5912
Computational Structures Technology UVA Center for CST Ahmed University
and
K. Noor of Virginia
Ft_C-ED._t-J,G PAGE
BLA;_K
NOT
FILMED
5
OUTLINE Rapid
advances
disciplines,
including
computational modeling,
in computer
structural science,
outgrowth
of finite
is on some newly
the materials,
structures
computer
aspects
established
environment
technology
numerical
methods
Center
along
developed
for CST.
The outline
with the motivations Third,
and research
directions.
Fourth,
as an insightful
on the one hand, theory,
airframes
on the other
CST.
we look at the future
of the Center
the newly
established
is describe.d,
UVA
andfinally
systems,
modeling
,
• A look at the future • TechniCal needs • Computing environment • Research directions • UVA Center for CST • Description
of one research
• Summary Figure
l
project
such as
the background
CST is an
of technical Center
and goals
a brief discussion needs,
is made
computing
for CST is described.
a brief summary
goals for CST and motivations of CST material
material
as well as on the
is given.
• Computational
6
between
The focus of this presentation
Second,
in terms
blend
hand (see Ref. l).
in Fig. I. First,
for developing
and mechanics
and other disciplines
and propulsion
is shown
engineering
A new technology,
over the last three decades. future
modeling.
disciplines.
emerged
and approximation
• Background, development
_i __
and synthesis
can impact
projects
effect on various
and dynamics has recently
material
Fifth, one of the research presentation
analysis
of CST which
had a profound
(CST),
analysis
element
UVA
have structures
and dynamic
for CST are described on computational
hardware
for
of the
DIFFERENT
Current
CST work includes
• computational
material
• computational
methods
components (Refs.
thereof.
Some
ASPECTS
the following
facets
OF CST WORK
(Fig. 2):
modeling for predicting
the response,
of these activities
performance,
failure
have been labeled
computational
and optimization.
In addition,
and life of structures structural
mechanics
2 and 3).
• automated component
methods
of structural
in multidisciplinary
synthesis
analysis
and design
of many
engineering
CST is an important
fields.
• Computational material modeling • CSM - Computational methods for predicting
• •• •
Performance Failure, Responseand Life
}
structures a n_lfcomponents thereof
• Automated methods of structural synthesis and optimization. Also, CST is an important component in the multidisciplinary analysis and design of many engineering systems.
Figure
2
and - CSM
GOALS FOR CST Within
the aeronautics
First.
To predict,
structures
and space fields, in a reasonable
and their components
S_ond.
To complement
the following
amount
at actual
goals of CST can be identified
of time, the response,
operating
and supplement
four major
failure
(Fig. 3):
and life of flight-vehicle
conditions_
experiments
and flight testing,
and to help in the design
of
experiments. Third.
To explore
new phenomena,
which
are difficult
to understand
simulations. Fou_h.
To aid in the design
process
of flight vehicles.
First To predict response, perfor_nce, failure and life of flight-vehicle structures at actual operating conditions Second To design, complement and supplement experiments and other tests for flightvehicle structures Third To study ph_oomena which are difficult to understand by means other than numerical simulations (e,g,, damage and failure mechanisms of high-performance materials at high temperatures) Fourth To aid in the design of flight-vehicle structures Figure
8
3
by means
other than numerical
MOTIVATIONS There compelling
Some
the capacity
components
concepts
components
and expensive
mission-critical
and space
A third major emerging potential
testing,
and future
motivation
strategies
explored
is frequently
computer
modeling
awaits
experiment
Thefirst and/or
so large and complex
numerical
that they overtax
are the simulation
and the study of thermoviscoplastic
of
response
In other
structural
problems,
(e.g., damage
initiation
and propagation
simulation component-
CST relates
computing
systems
to reduce
or mission-oriented. replace
the limits
tests.
of
the in
in solving
that exploit
power
large-scale
only by developing the capabilities
the dependence
on
Moreover,
in some
This is because
of ground-test
to the anticipated
can be realized
algorithms
is needed
may, of necessity,
are likely to exceed
for developing
and mlmerical
CST (Fig. 4).
systems).
which
computers
CST
of these problems
systems.
is that numerical
high-performance
of these high-performance
computational
are still being
structures
simulations
forces,
propulsion
developing problems
Examples
crash impact
reason
areas in space,
large aerospacecrafts
numerical
made of new material
compelling
practical
large computers.
used in advanced
mechanics
for vigorously
of unsolved
involve
to multidirectional
A second extensive
is that a number
of even present-day
fundamental
motivations
of these problems
response
structural
structural
compelling
motivation
solutions.
aircraft
are three
FOR DEVELOPING
future
technology. and potential
structural
problems.
of The
new formulations,
of the new machines.
9
• Practicalproblems
awaithTg solutions
• Very large problems (e.g., due to high-degree and/or high-degree of sophistication needed
of integration in modeling)
• Problems for which fundamental mechanics concepts explored (e.g., failure mechanisms of structural components made from new _----_. material systems) • Reduce dependence on testing • Reliability of testing large space structures questionable • Exploit new and emerging
computing
required
are still being 7
in 1-G environment
technology
is
] •
It, tool
IMSOOO3
_=-
I
I
Transputers
Figure
10
4
COMPUTATIONAL MATERIAL MODELLING Objective and Status Considerable
attention has recently been devoted to computational
CST, the objective and status of the activity are summarized a hierarchy of material models (multilevel/multiscale) response, life and failure of structures. physics-based
material modeling.
in Fig. 5. The overall objective
to describe the different phenomena
This has resulted in a gradual evolution
models that predict the processing
In the context of
response
and properties.
is to develop
associated with
from empirical
However,
models to
many gaps still exist
in the hierarchy of models.
Objective: Development of a hierarchy (multilevel/ multiscale) of material models to describe the different phenomena associated with response, life and failure of structures Status: Gradual evolution from empirical models to physics-based models that predict processing response and properties. •
Many gaps of models
still exist
in the hierarchy
Figure
5
11
COMPUTATIONAL MATERIAL MODELING Hierarchy of models A hierarchy phenomena meter). quantum
of material
they describe
The disciplines mechanics;
used range
models
are shown
and the length involved
from atomistic
scale at which
include
computational
in Fig. 6. The models
computational
material
to single crystals
this phenomena
science;
chemistry,
which
and computational
to polycrystals
are arranged is studied covers
(from
10d°m
molecular
structural
to micromechanical
according
to the to one
dynamics
mechanics.
The models
and macromechanical
models.
Models Phenomenological/ macromechanical
Length scale, m
Discipline Computational structural mechanics
Polycrystals (homogenized models)
Computational material science
Single crystals Atomistic models
Molecular dynamics
Computational
Quantum mechanics
chemistry
Figure
12
Phenomena
10 0 10-2
• Structural response • Metal forming
10 4
• Plastic strain localization • Crack tip fields • Indentation fracture
N
Micromechanical
10 -6
• Void growth ° Polycrystalline slip • Microstructural effects
10 "8
• Dislocations • Particles and interfaces
10"10
• Creep diffusion • Cleavage • Discrete defects • Basic transport properties • Phase transformation
6
and
A LOOK AT THE FUTURE CST is likely
to play a significant
in the multidisciplinary advances number Ref.
design
and computational of primary
tools are needed pacing
three factors
1) characteristics
adapted
and certification
and secondary
1). The following
role in the future of future
computing
3) recent
and projected
flight vehicles.
in a number
are taken into account
environment;
of structures
in identifying
their technical
technology
For this to happen
of key areas of CST.
items that must be addressed
of future flight vehicles,
2) future
development
as well as
major
To this end, there are a
by the research the pacing
community items
needs and implications
(see
(Fig. 7):
for CST;
and
developments
in other fields of computational
technology
which
can be
to CST.
• Computational technology will play a significant role in the development of structures technology and in the multidisciplinary design and certification of future flight vehicles • Major and secondary
pacing
• Basis for determining
the pacing items
• Characteristics their technical for CST
items
of future flight vehicles, needs and the implications
• Future computing
environment
• Recent and projected developments in other fields of computational technology which can be adapted to CST Figure
7
13
TECHNICAL NEEDS FOR FUTURE FLIGHT-VEHICLE STRUCTURES AND THEIR COMPUTATIONAL IMPLICATIONS The technical
needs for future
flight vehicles
can be grouped
into three
major
areas
(Fig. 8),
namely: 1) new high-performance metallics
material
as well as intelligent/smart 2) novel structural
systems.
material
concepts
which
These
include
new composite
materials,
advanced
systems; include
structural
tailoring
and smart/adaptive
structural
concepts; 3) expanding complex
phenomena,
couplings
1) development stress/strain/temperature 2) highfidelity 3) development
14
of engineering
such as damage
problems
tolerance
(e.g., structure/fluid/thermal/control
The implications
V
the scope
of new material interaction
of the aforementioned of computational range
considered
models
technical
to include
systems;
investigation
of more
and study of interdisciplinary
problems). needs for CST include:
for new material
systems
over the entire
of interest;
representation of effective
of details computational
(e.g., material strategies
response, for large-scale
joints
and damping);
coupled
problems.
and
Needs • New high-performance material systems (including intelligent/smart materials) •
Novel structural concepts (e.g., structural tailoring and smart/adaptive structures)
• Investigation of complex phenomena interdisciplinary couplings Implications • Computational
material
• High fidelity
representation
• Strategies
for large-scale
and
models of details coupled
problems
Figure 8
15
TRENDS IN HIGH-PERFORMANCE COMPUTING
The trends in the 1950's
in high-performance
and early
and vectorization late 1970's. that before
1960's was a result
was introduced.
Recent
trend is moving
towards
teraflop
is shown
of advances
Development
the end of the century
point operations
computing
in device
of computers
distributed
computing
in Fig. 9. The increase
in speed of computers
technology.
In the late 1960's
with homogeneous
parallelism
heterogeneous
will be achieved
supercomputing. (speeds
reaching
began
trillion
floating-
Distributed heterogeneous supercomputing Homogeneous
Pipelining parallelism & vectorization Device tec
1950
1960
1970 Figure
16
1980 9,
1990
in the
It is anticipated
per second).
Speed
pipelining
2000
DISTRIBUTED HETEROGENEOUS MULTICOMPUTERS The basic The concept which
concept
refers
to an integrated
the network
is the computer.
numerical
simulations
accomplished networks.
New
CRAY
(Cedarlike); advanced
imposed
by combining buzzwords
The hardware (e.g.,
of distributed
3, C-90, massively
workstations
multicomputers
environment
The use of DHM
of disparate
like "META
Computer"
a plethora
minisuper systems RISC,
speeds
are currently
(e.g., Convex
SUN,
...); and HDTV
is highlighted
of networked
alleviate
used to refer
systems
in
This is
through
high-speed
to this concept.
large-grain clustered
application-specific hardware
in Fig. 9a.
on the size of
capacities.
platforms
C-380);
computer
the limitations
and memory
such as:
(e.g., Intel Touchstone);
SGI,
(DHM)
supercomputing
of architectures
computers
consisting
can greatly
supercomputer
the resources
parallel (IBM
computing
by current
can include SSI),
heterogeneous
vector
supercomputers
processors computer
and software
systems;
with video
facilities.
Description: An integrated computing environment consisting of networked computer systems - Garden (or Plethora) of new architectures - META Computer. Hardware
includes:
• Large-grain vector supercomputers (e.g., CRAY 3, C-90, SSI) • Minisuper computers (Convex C380) • Clustered processors (Cedarlike) • Massively-parallel systems (Touchstone - gamma, delta or sigma) • Application-specific computer systems • Advanced workstations (IBM RISC, SGI, SUN,...) • HDTV hardware and software with video facilities Figure
9a
17
DISTRIBUTED
HETEROGENEOUS
MULTICOMPUTERS (CONT'D.) The effectiveness area networks
parallel
is strongly
dependent
for data transfer
between
For local area networks
(LAN),
interface)
links are currently
net T-3 internet
be needed. the following
and UltraNet will be upgraded
Because
media,
of DHM
the different
computers.
FDDI
distributed
mass storage
key elements:
and a massive
systems
being
HiPPI
local and wide
(high-performance
For wide area networks
(WAN)
improvement
directly
connected
(IEEE through
in mass storage reference HiPPI
model),
channels;
facilities
mass
network.
LAN
FDDI HiPPi
100 Mb/sec. 1 Gb/sec.
WAN
T3-1nternet T5-1nternet
60 Mb/sec.
Mass Storage: • High bandwidth mass storage (IEEE reference model) • Large disk arrays directly connected through HiPPI (> 100 G Bytes) • Mass robotic media file transfer
network (> 200 M Bytes/sec.) Figure
z
18 i
!
!
10
will
and include
Networking:
• Massive
the NSF
(Fig. 10).
will have high bandwidth
file transfer
of high-speed
data interface),
used.
of data, significant
large disk arrays
high-speed
(Fiber
to the T-5 interact
of the very large volumes
Future
on the availability
robotic
WHY DISTRIBUTED HETEROGENEOUS MULTIPROCESSORS? The primary parallelism
motivations
than can be done on a single
superconcurrency
(Fig.
Experience be relatively
In distributed
algorithms
vector
(or ineffective) applications
heterogeneous
are to achieve
architecture.
These
higher
are referred
speeds
and higher
levels
to as hypercomputing
of and
11).
with different
effective
most of the practical
numerical
for using DHM
and parallel
on different
the sustained
multiprocessors,
used in each module,
architectures
sections
has shown
of the computational
speed is a fraction the physical
process.
of the peak performance
characteristics
will form the basis
that different
of the problem,
for identifying
architectures Therefore,
can for
of the machine. and the
the most suitable
platform
for that module.
Hypercomputing
and superconcurrency
• Optimal support for algorithmically diverse parts of an application program on architectually diverse machines. • Different architectures can be relatively effective or ineffective on different sections of the computational process (code profiling and machine matching), • Parameters and data lengths can affect choice of architecture. Figure
11
19
CROSS OVER POINTS As an example, a function vector
the speed
of the vector
lengths,
higher
of performing
length
the operation
is depicted
speed is achieved
FOR SAXPY
Z(I) = R*X(I)
in Fig. 12. As can be seen from by a different
this figure
t
I_
8K Conn Mach
_
•
mmi.w
,
,
wwl_w==w
/: -'C'i_AY X-MP
7,,I,4,
,/
MFLOPS
for each range
Y(I)
4K DAP 100
computers
machine.
Z(I) = R*X(I).
1000
+ Y(I) on four different
/
4,0
40 e o 4, ,i,° # @* @@
10
Convex 210
@@@ @@
@@@ ,@
I
110'
I
,,,,,,,I
llllJ
IO0
,,,,,,,I
1000 10000 Vector size Figure
2O
,
12
, ,,,,,,J
, ,,,,,,,!
100000 1000000
of
as
PROFILES The results analysis identified
of the previous
on a network
figure
OF PROCESSOR
are extrapolated
of n supercomputer
as the one which
results
to the entire computation
platforms.
in the highest
For each module,
speed for that module
•",
//\/_
: "
typel_::
spectrum
of a finite
the most suitable (Fig.
element
platform
is
13).
Processor " type 2 ",, -7
m
Processor
TYPES
_",
/
Speed
/
v
_--
Processor type n
Element Assembly form.
Constraint and
Solution of
boundary conditions
equations
Computation
Error analysis
Adaptive refinement
spectrum
Figure
13
21
FORMS OF DISTRIBUTED
HETEROGENEOUS
MULTICOMPUTERS (DHM) Three different second
different
forms
of DHM
workstations
are shown
are connected.
is the site DHM,
in which
DHM
are the NSF Centers
DHM,
in which
An example
different
at Illinois,
supercomputing
in Fig. 14. The first is micro
architectures
is executed
are connected
and San Diego,
at different
sites, the application
program
concept,
the four NSF Supercomputing
connecting
of this is the PASMproject
Pittsburgh
platforms
DHM,
platforms. Centers
and NOSC.
site.
Examples
National
of global
Metacenter
DHM.
Global DHM
Front-end Processor
J Complementary
(e.g., PASM
Purdue Project)
(e.g.,
Back-end
NSF-NCSA,
SDSC,
Processors
PSC
and NOSC)
Maximum databases interface
Figure
22
14
The of site
The third is the global
The proposed
is an example
of
University.
in each of the different
SITE DHM
Micro DHM
a number
at Purdue
at the same
sites are connected,
on different
in which
use of or display/ capabilities
DHM
SUPPORT Despite
its potential,
of adequate
DHM
transfer
the data format
rates
fraction,
computers
must cross
affect the overall
interfaces different
several
include:
of potential
pitfalls.
high computing
For DHM
effective
networks,
such mismatches
of support
facilities
Thus,
to be viable,
bandwidth. as packet
networks
must be capable
on the supercomputers.
to that of another.
data must be measured. into the overall
In particular,
speeds
is not likely to be identical
since it figures
transfer
In addition,
express);
to sustain
or translating
significant
Linda,
in order
of one computer
spent transforming
15). These
also has a number
FACILITIES
In addition,
the amount
this should
If transfers
between
size and transfer
of time
not be a two
rate differences
could
rate.
a number distributed
high-level
operating
programming
to aid in the partitioning
system
are needed
(e.g., MARK
abstractions
of the program
to realize
the full potential
or KRONOS);
of DHM,
network
and object-oriented
tools; expert
into tasks and scheduling
the tasks
(Fig.
language systems
(e.g.,
and user
on processors
of
types.
• Distributed
• Network
operating
language
• Object oriented
system
(e.g., MARK, KRONOS)
(e.g., Linda, Express)
tools and user interfaces Figure
15
23
FUTURE
The future directions
DIRECTIONS
FOR RESEARCH
for research in CST are listed in Fig. 16. For convenience,
they are divided into
major pacing items and related tasks. The major pacing items include the following 1) high fidelity modeling of material response, structural, geometrical well as the environmental
3) effective computational fluid/thermal/structural/control
details as
made of new materials;
strategies for large-scale problems which include:
analysis; sensitivity analysis; and multidisciplinary
integrated
analysis and
of large systems; and
4) validation and assessment
of the reliability of numerical
The related tasks include predata, postdata processing technology;
and topological
effects;
2) life prediction and analysis of failure of structural components
optimization
four:
and integration
simulations.
and effective use of visualization
of analysis programs into CAD/CAE
and concurrent
engineering
systems.
Major Pacing Items • High fidelity modeling of material response, structural, geometrical and topological details, environmental effects (e.g., boundary-layer transition, interference heating) Life prediction and analysis of failure for structural components and structures made of new materials Effective computational strategies for large-scale problems • Integrated fluidthermalstructural analysis • Sensitivity analysis • Multidisciplinary design and optimization Validation and assessment of reliability of aerodynamic/thermal/structural response predictions. Secondary • •
Pacing
Items
Predata, postdata processing and effective use of visualization technology Integration of analysis programs into CAD/CAE and concurrent engineering systems Figure 16
24
UVA CENTER The background was established Headquarters.
in July 1990. Research
The overall modeling,
and objectives
analysis,
four specific
goal
grants
high-risk
sensitivity
for CST are highlighted
Langley.
The primary
from NASA
funding
Langley
and use of AI methods.
source
17. The Center is NASA
and AFOSR.
is to serve as a focal point for the diverse optimization
in Fig.
CST activities
The Center
including
has the following
research
on advanced
topics
of CST;
by demonstrating
to the research
community
future
of research
in support
results
and in broadening
what can be done
(high-
research);
of the twenty-first
of the aeronautical
and space
and,
transfer
of the state-of-the-art CST (notably
directions
century;
4) to help in the rapid
can impact
at NASA
also been obtained
studies,
innovative
3) to help in identifying
and engineers
have
of the Center
2) to act as pathfinder,
missions
It is located
Center
objectives:
1) to conduct
potential,
of the UVA
FOR CST
of research
awareness
in CST as well as in other areas of computational
CFD and computational
among technology
researchers which
mathematics).
25
BACKGROUND: • Established in July 1990 • Located at NASA Langley in Hampton • Funded by NASA Headquarters, NASA Langley, AFOSR,... OVERALL
GOAL:
Serve as focal point for CST development
(including modeling,
analysis, sensitivity studies, optimization and use of AI methods) SPECIFIC OBJECTIVES: • Conduct innovative reseach on advanced topics of CST • Act as pathfinder, by demonstrating what can be done (highpotential, high-risk research) • Help in identifying future directions for research • Help in the rapid transfer of research results and broaden awareness of the state-of-the-art in CST as well as other areas of computational technology that can impact CST (serve as central clearing house for information) Figure
26
17
FUTURE
To accomplish 1) research.
This will be done in strong collaboration
a series of seminars,
4) publish quarterly newsletter research centers and universities,
•
•
•
•
•
with NASA (Langley and Lewis), UVA
workshops
and national symposia; and state-of-the-art
monographs
on timely topics; and
listing CST research activities at various government
as well as recent contributions
laboratories,
on selected topics.
Items
High fidelity modeling of material response, structural, geometrical and topological details, environmental effects (e.g., boundary-layer transition, interference heating) Life prediction and analysis of failure for structural components and structures made of new materials Effective computational strategies for large-scale problems • Integrated fluid/thermal/structural analysis • Sensitivity analysis • Multidisciplinary design and optimization Validation and assessment of reliability of aerodynamic/thermal/structural response predictions.
Related •
Pacing
four major activities (Fig. 18):
researchers;
3) write survey papers, special publications
Major
FOR RESEARCH
its mission the Center will carry out the following
faculty, industry, and university 2) organize
DIRECTIONS
Tasks
Predata, postdata processing and effective use of visualization technology Integration of analysis programs into CAD/CAE and concurrent engineering systems Figure
IIIIIII
18
27
The initial research • Design-oriented This activity techniques
include effective
UVA CENTER
FOR CST
projects
include
CST
technique
to high-speed
(see Fig. 19):
transport
and large
for evaluating
the sensitivity
for large-scale
and coupled
flexible
derivatives,
spacecrafts.
optimization
for large systems.
• Innovative
computational
hierarchical
adaptive
use of artificial
for the Center
with application
will focus on effective and AI methods
selected
strategies
modeling
neural
networks,
strategies,
hybrid
analysis
and development
problems.
techniques,
of intelligent/smart
novel
This activity partitioning
computational
will strategies,
modules.
INITIAL RESEARCH PROJECTS: • Design-oriented CST(with application to high-speed transport and large flexible spacecrafts) • Optimization, sensitivity analysis and Ai methods • Innovative computational and coupled problems. • • • •
strategies for large-scale structural
Hierarchical adaptive modeling Hybrid techniques Novel partitioning strategies Neural networks
• Intelligent computational modules Figure
28
19
UVA CENTER FOR CST (Cont'd.) The initial
research
• Computational development
also include
modeling
for heat transfer
thermal
buckling
(Refs.
(Refs.
buckling
modeling
of flight-vehicle
computational
and modeling
and damping.
of joints
• Failure
analysis
• Quality
assessment,
models,
and mechanisms adaptive
multilayered
4 and 5), analytic
analysis
and engine effective
of structural
and validation
lamination
models
for
for the accurate
9, 10 and 11).
structures.
coupling
thermal
thermoelastic
procedures
(Refs.
This includes
of composites,
three-dimensional
responses
of failure
control
composites.
predictor-corrector
and postbuckling
(mulfilevel/multiscale)
four (Fig. 20):
for thermoviscoplastic
6, 7 and 8), and effective
of the thermal
• High fidelity
models
analysis
the following
of high-temperature
of micromechanical
theories
prediction
projects
This includes:
of numerical
components
of numerical
hierarchical
simulations
and experiments,
made of new materials.
simulations.
INITIAL RESEARCH PROJECTS: • Computationalmodeling composites
of high-temperature
multilayered
• Micromechanical models • Thermal lamination models for heat transfer • Computational models for thermal buckling and postbuckling • High-fidelity modeling of flight-vehicle
and engine structures
• Hierarchical (multilevel/multiscale) computational material models • Effective coupling of numerical simulations and experiments • Modeling of joints and damping • Failure analysis and mechanisms of failure of structural components made of new materials • Quality assessment and adaptive controlof numerical solutions; and validation of numerical simulations Figure
20
29
CHARACTERISTICS OF AN EFFECTIVE COMPUTATIONAL STRATEGY FOR LARGE STRUCTURAL SYSTEMS The remainder Center,
of the presentation
viz., development
of effective
is devoted
computational
to a description strategies
of one of the research
for large-scale
projects
and complex
at the
structural
systems. The three major
characteristics
First, the strategy hierarchical
adaptive
sophistication,
modeling
as needed,
must be the simplest The second computational
should
computational
insight
strategy
about the response.
are listed
the actual
structure.
As was suggested
in Fig. 21.
This is accomplished
- in the sense of starting from a simpler model
and increasing
by Einstein,
by using
the level of
"The model
used
one, but not simpler."
characteristic
model.
give physical
to model
possible
of an effective
of the strategy
This is accomplished
is that it should
by obtaining
help in assessing
sensitivity
information
the adequacy about
of the
the modeling
details
as part of the analysis. The third characteristic by linking
the degrees
new computing
3O
systems
of freedom (vector,
of the strategy is that it should be highly used in the initial discretization, parallel
and AI capabilities).
efficient,
and by exploiting
which
is accomplished
the major
features
of
Characteristics
Accomplished
by
• Gives physical insight about response
=Hierarchical adaptive modeling - start from a simpler model and increase the level of sophistication, as needed, to model the actual structure 'The model used must be the simplest possible one, but not simpler."
• Helps in assessing adequacy of computational model
• Obtain sensitivity modeling details
• High computational efficiency
• Reduce number of degrees of freedom used in original discretization • Exploit major features of new computing systems (vector, parallel and AI capabilities)
Figure
information
about
21
31
BASIC IDEA OF PROPOSED COMPUTATIONAL STRATEGIES The basic idea of the strategies, response
of a complex
system
using
either the simpler
system
discrete
of the simpler
Sensitivity available
equations
derivatives
of the response
process
the response
to satisfy
model
are then embedded
of the system cost.
conjugate
is to generate model
of the original
to modeling
of the simpler gradient-PCG
model
the
associated
system
into those of the original
with respect
The response
the three criteria,
from that of a simpler
mathematical/discrete
(e.g., preconditioned of the original
appear
large perturbations
system
at very small computational
and an iterative generate
or a simpler
which
(Fig. 22).
complex
details
with
neglected,
system. are directly
is then used as a predictor
or multigrid-MG)
is applied
model.
• Response of a complex system is generated using large perturbations from that of a simpler model associated with either: • Simpler system • Simpler mathematical/discrete model of original system • Discrete equations of simpler system are embedded of original complex system
into those
• Sensitivity derivatives of the response of the system with respect to modeling details neglected, are directly available • Response of simpler model used as a predictor and an iterative process (PCG or MG) is appl,ed to generate response of original model Figure
32
22
_
_
The
to
APPROACHES FOR SELECTING SIMPLER MODEL Two general
approaches
The f'n'st is hierarchical multigrid
approach
dimensionality. mathematical
based
transformations.
on selecting
the simpler
which
a simpler
and restriction
model
includes model,
are outlined
the classical associated
operators
reflect
in Fig. 23 (Refi
multigrid
technique
with a mathematical the physical
12).
and the physical
model
assumptions
of lower
of that
model.
domain
or uncoupling
modeling
The interpolation
The second physical
for selecting
approach
is the decomposition
decomposition.
It is based
of the load-carrying
on either
mechanisms
The two approaches
or partitioning
are briefly
uncoupling
in structural discussed
strategy
which
of different
problems,
can be thought
fields
in coupled
of as problems,
or using symmetry
in the succeeding
figures.
• Hierarchical modeling (multimodel or multigrid) • Mathematical model of a lower dimensionality (multimodel or physical multigrid-PMG) • Coarse finite element grid (classical multigrid) • Decomposition or partitioning (physical domain decomposition-PDD) • Uncoupling of different fields in coupled problems (e.g., aerodynamics, thermal and mechanical fields) • Uncoupling of load-carrying mechanisms in structural problems (e.g., extensional and bending components) • Symmetry transformations Figure
23
33
HIERARCHICAL
The application The structure is referred
of the hierarchical
is modeled
by using
to as the actual
beam model.
Although
similar,
of the simpler
those
The degrees interpolation two-dimensional operation
the governing structure
model
structure
discrete
equations
are much
smaller
reflects
structure
to one-dimensional
are given
to a composite
corresponds
are related
beam model).
structure
and
{z} = [r](z}
[F]t = restriction [k] = [F]t[K][F] {q} = [F]t{Q}
assumptions
operator w °,
1
Figure
34
i1,112
24
As usual
and load vectors
Governing equations: Actual structure [K]{Z} = {Q} Simpler structure [k]{z} = {q}
Interpolation [F] reflects basic in dimensionafity reduction
and simpler
in Fig. 24.
discrete
model
thin-walled
structures
model
used in the dimensionality
the stiffness
actual
The resulting
to those of the simpler
in Fig. 24.
Relationship between simpler model:
is outlined
are
in number.
thin-walled
between
airframe
to a one-dimensional
for both the actual
the basic assumptions
is taken to be IF] t. The relations
and the simpler
strategy
STRATEGY
plate and shell elements.
The simpler
of the actual
IF], which
shell model
modeling
two-dimensional
structure.
of freedom
operator
MODELING
by the
reduction
(from
the restriction
of the actual
structure
GEOMETRIC INTERPRETATION OF SYMMETRY TRANSFORMATION APPROACH
The geometric unsymmetric
domain,
subvectors, statement
component. vector.
It is important requires
governing
The transformation
vector
matrices
transformation
process
of the domain).
can be repeated The strategy
to an
into two equal
The equation
by applying
a matrix
length
in Fig. 25 is a
as the sum of a symmetric
and antisymmetric
model
as applied
and an transformation
to
in the figure.
for their determination. The simpler
in Fig. 25.
can be obtained
are shown
approach,
{Z} is partitioned
can be written
to note that each of the symmetric
only half the model
partitioning
vector
in the form shown
Each of the two vectors
{Z}s and {Z}as are coupled.
The symmetry
transformation
in Fig. 25. The response
of the fact that each unsymmetric
the original
(further
is given
for the symmetry
{Z}I and {Z}2, and can be written
antisymmetric
vector
interpretation
components
For an unsymmetric corresponds
to effect further
has several
advantages
domain
to the uncoupled reduction
of the response the equations set of equations.
in the size of the model
including
its suitability
for
parallelism.
35
Decomposition and Transformation matrices
t' tzl (Zl
+Z21 l+ z_ 1½ (z_ + Z2
o {Z}I nodes
{Z}1
1 (ZI.Z2) (Zl - Z 2 ))
= {Z}s + {Z}as
:'/---1
_____ :
tt
= [[T] s + [T]as]{Z}
/zl.--.--_
............. {Z} 2
13[']_[']
3
I:_[']_-['_
Comments on computational procedure • Each of {Z} s and {Z}a s can be determined by using a smaller (reduced-size) model or subdomain • Simpler model corresponds to uncoupled set of equations in {Z} s and {Z}a s
Figure
36
25
®
OPERATOR SPLITTING AND ITERATIVE SOLUTION PROCESS
!
The application partitioning
strategy,
The vector correspond
of operator is outlined
of fundamental
to the degrees
(uncoupled
equations).
unknowns
components
terms are identified
and iterative
solution
process,
in conjunction
with the
in Fig. 26.
of freedom
symmetric/antisymmetric The coupling
splitting
is partitioned
associated
with either different
of response.
by the parameter
The PCG iterative
into two subvectors
process
The discrete k. The simpler
{Z1} and IZ2}.
load-carrying
equations
are partitioned
structure
is used for generating
mechanisms,
corresponds the response
These or
accordingly. to the case k=0
of the actual
structure.
Partitioned
Equations
and Unknowns
{Z}I
O {Z}_1 nodes
. ...........
,A {Z} 2 nodes
([Kll ["21K:1 ){':}-{O.} Z1Q1 where L is a tracing parameter. = 1 --> original equations _, = 0 _ simpler-structure equations
[z}
Uncoupling of load-carrying mechanisms {Zl }' {Z2} are associated with
Iterative
Solution
different load-carrying mechanisms (e.g., membrane and bending)
Symmetry
transformation
symmetric/antisymmetric components of response vector
Process
• PCG used for solution • Left-hand-side matrix corresponding
to _,= 0 used as a preconditioner
Figure
26
37
APPLICATION TO NONLINEAR DYNAMIC ANALYSIS Cylindrical
The strategy center
circular
was applied cutout
to a nonlinear
subjected
i
\
/
dynamic
to uniform
,,,,1
j
panel with cutout
problem
pressure
of a composite
loading
cylindrical
panel
with an off-
(Fig. 26a).
x h_
u
w \
Boundary
3818 nonzero displacement degrees of freedom 6144 stress degrees of freedom
conditions
At x = 0, LI: u=v=w=
Loading
At y = 0, L2:
Uniform normal loading with intensity Po
W=_I=O
Figure
38
26a
_1= (_2 = 0
PERFORMANCE EVALUATION OF STRATEGY CRAY-Y MP4/432 (MENDOTA HEIGHTS)
The performance
of the proposed
27.
resulting
The speedup
of the study
are given
from
in Ref.
strategy
on the CRAY-YMP4/432
the use of four processors
at Mendota
was over an order
3818 displacements 6144 stress
971 displacements 1536 stresses
Semibandwidth of equations
700
315
Wall clock time
171
58.6 one processor 29.7 two processors 16.4 four processors
1.0
2.92 one processor 5.76 two processors 10.43 four processors
Speedup
in Fig.
The details
Partitioned Structure (Nearly Optimized Code) (246 MFLOPS)
Number of degrees of freedom
(first ten steps)
of magnitude.
is shown
12.
Full Structure (Optimized Code) (278 MFLOPS)
(see.)
Heights
ON
Figure
\
27
39
SUMMARY In summary
(Fig. 28), the goals
environment,
technical
needs,
aeronautical
and space
systems
UVA CST Center
structures.
(for example,
computational
contribute future flight
Its future
tools for structural
for research
development CFD,
advanced
complement
of planned
and activities
phenomena
interaction
and computational material
analysis
and synthesis
and future
to the development
of structures
technology,
computing
and projected
of the newly
established
projects.
associated
to experimental
strong
computational
high performance
of one of the research
of physical
requires
Future
in support
The mission
with the details
a valuable
mathematics,
of CST is bright;
significantly
along
our understanding
CST provides
flight-vehicle
computational
directions
have been identified.
enhanced
of structures.
The future
and future
have been described,
CST has greatly and failure
of CST have been described.
with the response
and analytical
with researchers
methods
for
in other fields
electromagnetics). models,
smart/intelligent
high performance
computers
as well as to improving
should
the design
of
vehicles.
• Goals of CST described • Future high-performance computing environment; technicaineeds and future directions for research, in support of aeronautical and space systems, identified • Mission
and activities
of UVA CST Center
described
• CST technology has greatly enhanced our understanding of physical phenomena associated with high temperatures • Future development of CST requires strong interaction with researchers in other fields (e.g., computational mathematics, CFD, and computational electromagnetics). Figure
4O
28
i
=.
REFERENCES 1. Noor, A. K. andVenneri, S.L. "AdvancesandTrendsin ComputationalStructuresTechnology," ComputingSystemsin Engineering,Vol. 1,No. 1, 1990,pp. 23-36. 2. Noor,A. K. andAtluri, S.N., "AdvancesandTrendsin ComputationalStructuralMechanics,"AIAA Journal,Vol. 25, 1987,pp. 977-995. 3. Grandhi,R. V., Stroud,W. J. andVenkayya,V. B. (eds),ComputationalStructuralMechanicsand MultidisciplinaryOptimization,AD Vol. 16,AmericanSocietyof MechanicalEngineers,NewYork, 1989. 4. Noor,A. K. andBurton,W. S., "Steady-State HeatConductionin MultilayeredCompositePlatesand Shells,"ComputersandStructures,Vol. 39,No. 1/2,March 1991,pp. 185-193. 5. Noor, A. K. andTenek,L.H., "Steady-State NonlinearHeatTransferin Multilayered Composite Panels,"Journalof EngineeringMechanics,ASCE 1992. 6. Noor, A. K. andBurton,W. S., "Three-DimensionalSolutionsfor ThermalBuckling of Multilayered AnisotropicPlates,"Journalof EngineeringMechanics,ASCE 1992. .
Noor, A. K. andBurton,W. S., "Three-DimensionalSolutionsfor theFreeVibrationsandBuckling of ThermallyStressedMultilayeredAngle-PlyCompositePlates,"Journalof AppliedMechanics 1992.
8. Noor, A. K. andBurton, W. S.,"Three-DimensionalSolutionsfor theThermalBuckling and SensitivityDerivativesof Temperature-Sensitive MultilayeredAngle-PlyPlates,"Journalof Applied Mechanics1992. .
Noor, A. K. andBurton,W. S.,"Predictor-CorrectorProceduresfor ThermalBuckling of Multilayered CompositePlates,"ComputersandStructures,Vol. 40, No. 5, 1991,pp. 1071-1084.
10.Noor,A. K. andPeters,J. M., "Postbucklingof MultilayeredCompositePlatesSubjectedto CombinedAxial andThermalLoads,"FiniteElementsin AnalysisandDesign1992. 11.Noor, A. K. andPeters,J. M., "ThermalPostbucklingof Thin-WalledCompositeStiffeners," ComputingSystemsin Engineering,Vol. 2, No. 1, 1991,pp. 1-16. 12. Noor,A. K. andPeters,J. M., "Strategiesfor Large-ScaleStructuralProblemson High-Performance Computers,"Communicationsin AppIied NumericalMethods,Vol. 7, 1991,pp. 465-478. 41
N92-25913
Computer
Codes Developed and Under Development at Lewis Christos C. Chamis NASA Lewis Research Center Cleveland, Ohio
. _,JC BLAt;K
NOT F!L ._t;D
43
Codes IPACS
The objective of this summary is to provide developed or under development at Lewis and with some typical early results.
a brief description (2) the development
of: status
(i) of
The computer codes that have been developed and/or are under development at Lewis Research Center are listed in the accompanying Charts 1 - 5. This list includes (i) the code acronym, (2) select physics descriptors, (3) current enhancements and (4) present (9/91) code status with respect to its availability and documentation. The computer codes llst is grouped by related functions such as: (i) Composite Mechanics, (2) Composite Structures, (3) Integrated and 3-D Analysis, (4) Structural Tailoring, and (5) Probabilistic Structural Analysis. These codes provide a broad computational simulation infrastructure (technology base-readlness) for assessing the structural integrity/durability/reliability of propulsion systems. These codes serve two other very They provide an effective means of technology transfer depository of corporate memory.
CODES
DEVELOPED
STRUCTURAL
BY AND ARE AVAILABLE
MECHANICS
Probabilistic Code
BRANCH,
Structural
Probabilistic
Structural
as the analysis
Analysis
functions: (i) they constitute a
FROM
THE
2 Oct 91
Analysis
Description/Current Enhancements
Name NESSUS
important and (2)
Status
includes
module/component
MHOST
risk and
Available
with
documentation
reliability CLS
IPACS
Probabilistic
Loads
Simulation
Shuttle
Available
with
Main Engine components/loads based on deterministic models
simulation
documentation
Probabilistic
of Composites/
Development
Couple
PCAN
Structural
Analysis
with NESSUS
Chart
44
for Space
1
CODES
DEVELOPED
STRUCTURAL
BY AND ARE AVAILABLE
MECHANICS Structural
Code
BRANCH,
FROM
2 Oct 91
Tailoring
Description/Current Enhancements
Name STAHYC
Structural
Tailoring
Computational
Status
of Hypersonic
efficiency,
THE
Structures/
Operational
documentation
STAEBL
Structural
STAEBL/ GENCOMP
Structural Tailoring Structures/Improved documentation
STAEBL/AERO
Structural Tailoring of Engine Blades for Aerodynamic Performance and Flutter
Completed and documented
STAEBL/TURBINE
Structural
Tailoring
of Turbine
Completed
& documented
STAT
Structural
Tailoring
of Swept Turboprops
Completed
& documented
CSTEM
Coupled
Tailoring
of Engine
Blades
Completed
of General Composites finite element and
Operational no documentation
Blades
Structural/Thermal/Electromagnetic
Tailoring/Gain
& documented
Available with documentation
familiarity
Chart 2
CODES
DEVELOPED
STRUCTURAL
MECHANICS
Integrated Code
and 3-D Inelastic
and large problem
FROM THE
2 Oct 91
Analyses Status
Lewis-owned Finite Element Analysis Computer Code - based on mixed iterative scheme/add mix-element
3.D INAN
BRANCH,
Description/Current Enhancements
Name MHOST
BY AND ARE AVAILABLE
Available with documentation
capabilities
A compilation of 9 different 3-D finite element codes for nonlinear structural and stress
Available with documentation
analysis with progressive level of sophistication, nonlinear simulation models/Dormant - needs code user familiarity BEST3D
Boundary Element with heat transfer
Structural
ESMOSS
Engine
Modeling
Structures
Dormant COSMO
Analysis
Software
code
Available with documentation
System/
Available
- needs code user familiarity
Component Specific Modular for Hot Engine Structures/Dormant - needs code user
with
documentation Available with documentation
familiarity
Chart 3
45
CODES DEVELOPED STRUCTURAL
BY AND ARE AVAILABLE
MECHANICS Composite
Code Name
BRANCH,
Composite
Blade Structural
MHOST, add updated CODSTRAN
HITCAN
Structures
Composite
Durabilily
Extension
to complete
Analysis/Add
CODES
DEVELOPED
STRUCTURAL
Structural
Analysis/
METCAN
BY AND ARE AVAILABLE BRANCH,
FROM
Composite
Analyzer
Incorporate
damping
capability
Probabilistic Probabilistic
Integrated capability
Ceramic
features;
Status
Available
- PMC/
Composite to ICAN
Metal Matrix Composite
Analyzer/
Analyzer/rime
and
documentation
Matrix Composite ply/fiber
Analyzer
- based
from COSMIC
Development
Operational sketchy documentation Development
substructuring
Inlegrate Composite Analysis for Structural Composite Sandwiches/Dormant - needs recovery
THE
2 Oct 91
Mechanics
Integrated
stress
Available
with
documentation
capability
Metal Matrix Laminate Tailoring/Capability to tailor for specific life
Chart
46
Operational
4
MECHANICS
on progressive ICAN/SCS
Analysis/
Description/Current Enhancements
unloading CEMCAN
Development
structures
Composite Code Name
Operational
ICAN
Chart
MMLT
Status
High Temperature Composite Structural Computational efficiency, documentation
PICAN
2 Oct 91
Description/Current Enhancements
COBSTRAN
ICAN
FROM THE
5
Development
The
schematic
embodied
in
progressive of
the
in an
fracture
local
through
computer
in
various
composite
constituent initiation
closes
the
loop
materials
state
local
STRUCTURE
the to
illustration
structural
The
code
the
global
material
capability
specific similar
PROGRESSIVE
to
the
corresponding
global
for
it local
simulation
structural are
VIA
evaluated.
CODSTtTAN
o GLOBAL
GLOBAL
0
ANALYSIS
ANALYSIS
f,';._'F:n _ :-,
/ I I
time
feedback
FRACTURE
STRUCTURAL
/
same
design concepts to CODSTRAN.
effects
response
the
computational
and
of
at
continuous the
physics
the
structural
response
Thus,
the
simulation
synthesizes
And
permits
growth.
of
computational
composites.
which
and
where component/system and HITCAN have structures
COMPOSITE
in
global
accumulation
excellent for
behavior
inherent the
between
an code
structures.
materials
scales
decomposes
damage
performance COBSTRAN
1 represents
constituent
the
progressively local
Figure
integrated
STRUCTURAL _
/
\
A ;:-;:lT_r.
\ LAMINATE
I,"_o..',.__ I
LAMINATE_ THEORY
r ICAN PLY-_t'."':!ifl;_il)i";ii__/; __
\ \ \ \
COMPOSITE MICROMECHANICS THEORY
UPWARD _ INTEGRATED _ OR "SYNTHESIS"
,,. "%
/"
, .. .,,_!!i;il;i;i)ii_i_ PLY
"_.
_ /--.
/ P '
.,-.,
C.O, _M..P..O_T.. ,E,,,,,,iCS / _L_)_c_,n,_mn,-un /
d t"
U,--"
/
CONSTITUENTS
/ MATERIAL PROPERTIES
"_
P (m T, M) _-- --_ -_
Simulation
of Composite
Damage
and Fracture
Figure
I I / /
LAMINATE, THEORY #
/ /
J
Propagation
/
/
TRACED OR "DECOMPOSITION"
/
via CODSTR
TOP DOWN
AN
I
47
Typical during slit
results progressive and
obtained by fracture
subjected
to
using CODSTRAN to in this composite
internal
pressure
is highly localized exemplifying this deformation evolution in monitoring. in
The
Figure
remains
3. so
damage
corresponding
The
the
tolerance,
surface
defect
(damage
tolerance)
defect
location exhibits
simplistic aspects
P
the
= 288
AFTER
a
to
internal is
PSI
(1.379
growth
the
usefulness
and
to
4.
PLY
BEFORE
DAMAGE
some
internal that
with these
I LONGITUDINAL (PSI)
Fracture
of Plies
PA)
TO 14
1 Longitudinal
Composite
Stresses
after
l,,itial
Shell T300/Epoxy[902/&lS/90,/+lS/9%/TIS/9%]
Figure 2
in with
1 and 9.
shown and the plan,
the
pressure the two
simulation
process.
_ 6,895
is case
inspection shell
computational
(I PSI
this
difference
Obviously, of
deformation
slit
in
The
higher
IN
I g 2
embedded
However,
development
MPA)
Ply
48
and
STRESSES
PLIES
PROGRESSION PLIES 13 _
an
The
evolution surface
sensors to capture in service health
The
behavior.
component/system
2.
localized
Figure
fracture.
DEFECT
IMPOSITION
Figure
for
more
in
deformation a longitudinal
locating tests or
pressure
shown
damage
structural
demonstrate
structural
in of
even
the with
catastrophically.
damage
limited to
is
"brittle-fracture-llke"
examples of
the
prior
exhibits
fractures
respect of
shown
pattern
evolution
shell
with
the
are
difficulty verification
deformation
deformation
until
between
the either
assess shell
embedded rather in
all
P
-
344
PSI
(2.375
MPA)
PLY I LONGITUDINAL STRESSES (PSI) (I
IHITIAL DEFECT PLIES
PSI
-
G,SgS
PA)
IN 9
&
220000.
18
19S000.
IMMEDIATELY BEFORE I
&
2
PLIES
16B000.
FRACTURE
158080.
IS000B. If
Ply
1 Longitudinal
Composite
Shell
Stresses
at 2.375
MPa;
before
Ply
1 Fracture_
T300/Epoxy[90_/=ElS/902/+lS]90_/_=lS/90_}
Figure 3
400
....
2.758
350 t
2.413
g
_ 250
_
A Defect in plies 1 & 2 (Case II)
1.724
tl/ ° °e_°_ 'n_''e__ _ _°_°°__v_ 200_
_
1.379
150 I 0.0
,
0.'2 DAMAGE
Damage
Propagation
Composite
Shell
wlth
,
, 0.4
1.034
(_)
Pressure
"r300/Epoxy[DO2/_-15/90_/=l=lS/90_/_FIS/90_]
Figure 4
49
Modules from the various codes in the llst can be stacked up (assembled) to develop computational simulation capability for specific structural response. The schematic in Figure 5 illustrates a combination to evaluate acoustic fatigue in composite panels In a hygrothermal environment. The combination of codes include: (I) CSTEM for acoustic source, (2) ICAN for composite properties synthesis, (3) MHOST for structural dynamic analysis due to acoustic excitations, (4) ICAN for ply stresses and strengths and (5) ICAN for cyclic load effects.
Computational
Simulation
of Acoustic
R.T. Constituent Properties
Fatigue
:-............. , /CAN _
_
EnvironmentDegraded Properties
Vibrating Panel
l
Acoustically Excited Panel
Acoustic Pressure
Dynamic Force Response MttOST_
CSTEM I
time
time
/CAN
Safety
I
_
No. of Cycles Figure
511
5
__1
Ply Stress/ Strength
Typical results from the computational simulation of acoustic fatigue on composites are shown in Figure 6 where the remaining strength in terms of margin of safety is plotted versus the number of cycles. Three different curves are shown: (i) The bottom curve is the base case. It is for a [(0/± 45/902)3] AS/E laminate at room temperature conditions. (2) The middle curve is for the base laminate at (200°F, I% moisture by weight) environmental conditions. The top curve is the base laminate with rearranged plies. Environmental effects enhance composite fatigue because they tend to soften the composite. Ply stacking sequences can be selected to significantly increase acoustic fatigue provided that other design requirements are not violated. The significant observation is that the infrastructure available in the computer codes listed in Charts 1 - 5 provide a base to tackle a variety of problems as they arise.
Demonstration:
acoustic
fatigue
0.60 .... : :_;,.
0.48
....
>.,
/
.............
([45/0/-45/9012)
;L,Q._ -- . ./-.
s
" ......................................................................................................
(]3 N-,-.
O9
,°
0.36
4--"
o ED
""_,_-(200
°g, 1%
Moisture)"Q...
0.24 *.
0.12
.
Base
Case
-
. *.'-%
',,
0
I
10
5
6
7
10
10
Number
I__LLIJJ
8 10
10
9
of Cycles
Figure 6
51
The major elements of the Advanced Composites Technology Program are depicted within the ellipse in Figure 7. These elements include activities in all major aspects of advanced composites technology: {i) Constituent materials development and characterization (top). {2) Demonstration of these constituent materials in simple structural elements {right). {3) Structural components made from these simple elements {bottom right). (4) Analysis methods development {bottom left). (5) Design and fabrication process {left). Another small element of this relatively inclusive program is the development of probabilistic methods to incorporate the uncertainties associated with the major technology elements. These uncertainties are depicted schematically around the ellipse in Figure 7 next to their respective program element.
ADVANCED (,4 Bold
New
COMPOSITE
Program
in Composites
TECHNOLOGY Research
& Technology)
I
Probability
J%
/ I
°%%'_'I_I' ] "'_//I/I I\\
IJ
S,re.. S,,.,,_.
Probab,,,,y L//._/__l A
1_, _
._-_,
/
.
_
Pr°b'b"tY I
lJ
Probability
/
/ Magnitude
Crileria
Figure7
52
]
oo=,_"a°"'e
/ __-_
"e'_°°'°I
_
I
/
The different
probabilistic computer codes.
Probabillstlc
Integrated
capability probabilistic
in
methods to The first Composite
for
two
uncertainties
are
integrated
for
using scale
uncertainties
of
40
the
process
The -
the
available
result
different
are is
computational
simulation
in
(bottom).
different
ICAN.
as
probabilistic
composite
properties
the
composite uncertainties,
integration
description which
accepts
These
inherent
Additional
introduced
the
code and
are
progresses
of
the
required
to
fully
composite.
Probabilistic
Simulation
Composite Material Level
of Composite
Deterministic Properlies f/:.__
Mechanics
with PCAN
Fabrication Variables
Probabilistic Propedies
-_ _ PI)F
le o o o o oL-_l-/_'-%-_q lo_o o.o.0 ? ooL2.__pt I= = = _ _ ___jpv-
Laminate
embodied into two as PICAN for
in Figure 8. The material properties
variables
through
descriptors
scale.
will be identified
The
illustrated constituent
upward
mechanics
composite
to
characterize
fabrication
composite
respective
from
Analyzer.
PICAN is schematically uncertainties for 29
uncertainties scales
be developed of these is
40
tI, OI Fabrication Variables
Properly
Prolierly
I 0
=3 (I
"'"
_(
4o
Ply
I
PI)F
tp, 0p
I
0........ 0 0 0 ]:".-1
Fabr |cation Variables
Property
PDF I o_o o o I_'--"
Subply
37
ts, 0 s Property
/\
l"rolleily
CDF
Fabrication
I
-
Properly
Variables
Constituents (fiber/matrix)
[ _'-_-["i_ ._._//f-J_
17 for tiber 12 for matrix
fvr, vvr Property
NOTAIION:
I = CDF
lhickness. =
Cllmlllalivo
0 =
nfisnliqnmeld, Dishihulion
fvr Ea]clion,
=
fiber
volume
Sulls(:ripls
ralio,vvr I =
Figure
"-
=
void
laminale,
p
volume =
ply,
ralin, s
=
PDF
=
probnbilily
(|e.sily
EJnc|iol}
_uhllly
8
53
Probabilistically "laminates and
for
values major
are two are
standard
for
ratio,
Figure
PICAN
in
9,
results
Figure
deviations
included
Poisson's
results in stiffness.
described
tabulated
9. on
though
are
preliminary,
lower
stiffnesses
side
All
line)
VERIFICATION
l_aminale
either
comparisons. (last
for
Probabilistic the
of the
verify
boulld
two
PICAN
mean.
111.5"111
different
given
for
Average values,
standard predictions
LAMINATE
(111c.,'111-2o)
[0I + 45_I0/±
the
three
are
experimental
within
FOR
of
values
experimenlal value
the
except
the
deviations. for
The
laminate
STIFFNESS
llpper houild (111_711 { 2o'}
45],
Long.
tllodilhlS
(MSI)
5.40
6.19
6.30
6.98
Trans.
niodiihls
(MSI)
2.46
3 .(17
3 O8
3.68
.l_lliZ;ll tlll)(hlhls
([tl']_[)
3.3.1
3.84
3.21
4.35
0.690
(}.8116
0 803
0.922
Majol
Poisson's
ralio
'
[0/_t:45/0/90/0)], l.ong,
lnl_duhls
(MSI)
11.41
13.30
131K)
15.09
'rranL
inodilhls
(M,_I)
3.69
4.30
4.20
4.96
(MSI)
1.40
1.59
1.50
1.78
ratio
0.276
0.313
0.325
0.350
[.ong. Illodulus
(MSI)
6.12
7.15
6.68
8.18
Trans.
(MSI)
6.12
7.15
6.62
8.18
2.37
2.72
2.34
3.07
0.290
0.317
0.350
0.344
Shear modulus Major I'oisson's I(0/+45/90),1,
,qlicar
Majoi
inoduhis lllodulus
(M_I)
PoiSSOli'S ialio
Figure i
54
9
mean
experimental
The other computer code under development as a part of the ACT program IPACS, Integrated Probabillstic Assessment of Composite Structures. This consists of PICAN and NESSUS with modules from COBSTRAN for automatic finite element model generation and composite configuration description. A schematic the computational simulation capability in IPACS is depicted in Figure i0. Preliminary probabilistic results obtained from IPACS of a composite panel, loaded in compression (Figure II), are tabulated in Figure 12. Inclusion of probabilistic boundary conditions for two panels appear to verify IPACS' predictions for buckling loads. A few summary remarks are listed in Chart 6.
IPACS
: Inlegraled
Probabilistic
Assessmont
of Composito
is
of
Slrucluros
COMPONENTS
ELEMENT
_
_
FINITE
ELEMENT
_
FINITE
GLOBAL
GLOBAL
.a j LAMINA
TE
(
N_SSUS
_)
LAMINA
LA
_
°OMPOS,TE %
MICRO-MECHANICS THEORY
I I I
TE
LAMINATE
'_'
....OM OS.E MICRO-MECHANICS THEORY
NONUNEAR MULTIFACTOR MODEL CONSTITUENTS
MATERIAL
PROPERTIES
P - F(I+ T, M)
Figure I0
55
GEOMETRY
-:/
OF
THE
PLATE
-/ Figure il
II'ACS
VERIFICATION
I ;lnlhl;llc
FOl+t BUCKLING
lowt'r
ht_tllld
(nlcm,
lilt'till
2,0
I+,OAI)S
CXl+elin+cnlal
Ulq+Ct llmmd
vahlc
...................................................
I 2,,)
' (m(':m
2o(o) lluckting
?,I!
hind
I
_t)t9t)) "hut I,linq, Io:ld IO( J: 30),
It)( -t-30)
" bucklh_g hind It)(-±,lS),
I0(:I
513
561
662
6RP,
555
(,tit;
5 _
crack
o---o-
.o
o
,o-
o-
J length
Test B I I
o ="
11
o -o- .o .o '_J' '-"='_ "0
-
,0, -o- o- ,o- =-'*"" _ "" *"
Test A
I crack I length (ram)
o Test B
0
Average Half Crack Length (ram)
"
0
/
[]
."
9
o
,:'
o 7
..'
o O
[]
.." ,'''
[]
5 AIBE prediction
_,J'-j 3
..... [] o
.J..|,JJ
0
J|.=..1,
10
20
average crack predicti0n measurements
...
30
|,
40
Cycles (thousands) Figure
136
15
J..J...,
i
50
60
were
crack
as
cycles
whereas,
the
the
cracks
growth
plotted
cycles;
exceeded
with
lives. aluminum
interaction
are
predictions
measured
proximity rates
holes
of
the and
growth
sheets
crack
and
AIBE
the
open
Fatigue
evaluating
developed
crack
thick
lengths
lengths
average growth
The of
for
was
fatigue
co-linear
lengths.
cycles.
method
cracks
2.3-mm
hole.
crack
DAMAGE
(AIBE)
wide,
each
crack
load
crack
code
of
within
than
interacting
equal
unequal
on
closer
element
contained
of
MULTI-SITE
multiple
sides
method
the
OF
boundary
on
sheets
basis
were
The
computer
both
versus
100%.
for
conducted
The out
crack
a
were
2024-T3.
propagating
MECHANICS
unequal
predictions as caused
much
as
-25918
Analysis and Design Technology for High-Speed Aircraft Structures James
.=
F
H. Starnes, Jr. and Chades J. Camarda NASA Langley Research Center HamptOn, Virginia
137
INTRODUCTION
Viable supersonic and hypersonic aircraft structures must be structurally efficient and designed to operate reliably with combined mechanical and thermal loads. To provide such structures requires the development of verified structural analysis and design technology that is necessary to predict the response and failure characteristics of wing and fuselage structures made from advanced metallic and composite materials. Research is being conducted at NASA Langley Research Center to understand the response and failure characteristics of high-speed aircraft structures and to develop the necessary structural analysis and design technology for future high-speed aircraft. The present paper describes selected recent high-speed aircraft structures research activities at NASA Langley Research Center. Selected topics include: the development of analytical and numerical solutions to global and local thermal and structural problems, experimental verification of analysis methods and identification of failure mechanisms, and the incorporation of analysis methods into design and optimization strategies. The paper describes recent NASA Langley advances in analysis and design methods, structural and thermal concepts, and test methods.
138
AIRCRAFT
STRUCTURES
RESEARCH
High-speed aircraft structures research at NASA Langley Research Center is focused on the development of structural mechanics technology for supersonic and hypersonic aircraft primary structures. A goal of the research is to understand the thermal and structural behavior of complex structures made from advanced metallic and non-metallic materials using advanced fabrication techniques. Another goal is to develop structurally efficient and cost-effective structural concepts which exploit the beneficial characteristics of advanced metallic and non-metallic composite materials.
Provide the scientific basis and structural mechanics technology
for
aircraft primary structures
Develop structurally
efficient, cost-effective
structural concepts that
exploit the benefits of advanced composite and advanced metallic materials
139
STRUCTURAL
MECHANICS
RESEARCH
APPROACH
The key to understanding the physics of a structural mechanics problem is the ability to conduct precise structural and thermal experiments. Optimally designed experiments, which insure satisfaction of prescribed boundary and loading conditions and the accurate measurement of response quantities, are essential. Experiments are designed so parameters can be varied systematically and representative failure mechanisms can be identified and understood. The development of verified analytical methods, whether classical or numerical or combinations of both, is closely coupled to the experiments. The ability of the analysis methods to predict accurately the actual response and failure mechanisms verifies the methods. Anomalies in the correlation between analysis and experiment are resolved by careful studies of the observed behavior and, if necessary, additional experiments are conducted and improved analytical methods and models are developed. Once the analytical methods are verified and an understanding of a given problem is assured, the next logical step is the appropriate simplification of the analysis to increase computational efficiency and enable its incorporation into a formal optimization or structural sizing procedure. During the optimization process, many analysis and response sensitivity calculations must be performed and, hence, it is often necessary to simplify the analyses as much as possible to make the optimization process tractable. Iteration between the development of analysis and optimization procedures is necessary to insure accuracy of the analysis and proper representation of constraint boundaries. Iteration between experiments and optimization methods development is needed to assure that optimally designed structures represent actual physical behavior and to insure an accurate physical description of the response. The overall goal of structural mechanics research is to provide a better understanding of the physics of the problems of interest, including the true limits of performance, which leads to less conservatism in the design and, hence, a more structurally efficient design for wing and fuselage primary structures.
!
f_
140
Y
Experiments
_"
"_
" (Des,gn)
"_
J
_iMathematics)
• Systematically * Understand , Identify STRUCTURAL
Computational
) _
vary parameters true limits
failure
MECHANICS
Methods
of performance
mechanisms RESEARCH
APPROACH
141
STRUCTURAL
ANALYSIS
AND
DESIGN
TECHNOLOGY
One contribution that will decrease the time needed to design a new aircraft structure is a structural modeling tool that will decrease the time needed to develop a discrete structural model of a complex aerospace vehicle. Work is underway at NASA Langley to enhance the Solid Modeling Aerospace Research Tool (SMART) to enable rapid structural modeling of external and internal structure from a given aerodynamic shape. In addition, it is intended that the enhancements to SMART will allow the rearrangement and resizing of internal and external structural elements in a rapid manner. Both global and local analysis methods are being developed to predict the structural and thermal response for static and transient, linear and nonlinear problems. Some of the analytical methods currently being investigated are: advanced reduced basis methods, operator splitting techniques, flux- or stressbased finite element algorithms, Ritz-based methods, and classical solution methods. Several methods for optimizing large structural systems, such as a future supersonic transport wing which has thousands of degrees of freedom and is subjected to hundreds of mechanical and thermal constraints, are currently being explored. Equilibrium programming and other structural sizing methods are being developed to address the structural optimization of large nonlinear systems. Efficient methods to calculate structural and thermal sensitivity derivatives are being implemented into existing general-purpose finite element codes to facilitate formal optimization and to explore other uses for sensitivity derivative information such as parameter estimation. These structural sizing tools are being used to tailor structural designs to exploit the beneficial properties of advanced materials. Structural
modeling
aerodynamic
tools for internal
external
Detailed and local analysis thermal loads optimization constraints
Sensitivity parameters
142
analyses
configurations
from
shape
Global analysis methods and thermal loads
Design, thermal
structural
for combined
methods
and tailoring
to identify
that affect
structural
mechanical,
for combined
methods
important
pressure
mechanical
for mechanical
geometric
performance
internal
and
and material
and
THERMAL-STRUCTURAL
CONCEPTS
The degree of coupling between the structural, thermal, and fluids disciplines increases proportionally as the speed of the vehicle increases. For a hypersonic vehicle such as the National Aero-Space Plane, which is designed to cruise at Mach numbers exceeding 16 and to fly single stage to orbit, the selection of a hot structures concept, an insulated structural concept, and a cooled structural concept is not straight forward and often requires re-evaluation of the concepts at the vehicle design level. A good understanding of fluid flow, heat transfer, and structural mechanics is often necessary early in the design cycle to insure proper synergism in the design. Examples of solutions to coupled problems include the design of a refractory-metal/refractory-composite heat-pipe-cooled wing leading edge and a liquid-metal-cooled engine cowl leading edge. Both of these concepts will be discussed in detail later in the paper. Supersonic vehicles such as a future supersonic transport may not experience the same degree of coupling between the thermal and structural disciplines as a hypersonic vehicle; however, important structural design options will be governed by heat transfer. For example, the selection of wing structural concepts which contain fuel may be governed by the inherent insulative properties of a sandwich structure that may be lighter in weight than a stiffened structure with insulation. Organic composite and metallic structures will be compared in the preliminary design of a supersonic transport. Cost is an important consideration in the competitiveness of a commercial supersonic transport in addition to weight. Also, damage tolerance, durability and thermal stability may become critical design constraints in addition to traditional strength and buckling constraints. Details of preliminary studies for a wing structure will be presented later in the paper. Some hypersonic structural concepts currently being investigated include: a carbon-carbon elevon (hot structure), a refractory-composite/heat-pipe-cooled wing leading edge (passively cooled structure), and a liquid-metal-cooled engine cowl leading edge (actively cooled structure). Structural materials for these concepts include advanced refractory-composite materials and advanced metal-matrix composites.
143
Speed, weight, cost, and supportability - Uncooled or hot structures - Cooled structures
influence concept selection
- Thermal protection systems and insulated structures Current HSCT wing and fuselage concept candidates - Sandwich structure - Stiffened-skin structure with or without fuel tanks and fuselage insulation - Organic composites and metals Some current HSCT structural issues Some -
Weight and cost Damage tolerance and damage containment Durability for 60,000 flight hours at Mach 2.4 cruise Thermal stresses and thermal stability at global and local levels current hypersonic structural concepts Actively cooled structure C/C, C/SiC, advanced metal matrix Heat pipes THERMAL-STRUCTURAL f
144
CONCEPTS
TEMPERATURE
i
DISTRIBUTIONS FOR CIVIL TRANSPORT
HIGH-SPEED WINGS
SUPERSONIC
Temperatures of a future high-speed supersonic civil transport (HSCT) wing were calculated for two different flight Cruise Mach numbers. Temperatures were calculated assuming radiation equilibrium on the heated surfaces and neglecting conduction. As shown in the upper left figure, a Mach 3 cruise flight condition results in temperatures close to 500 °F, 50 ft. aft of the leading edge. Corresponding maximum temperatures for a Mach 2.4 cruise flight condition are approximately 300 °F. The large difference in wing temperatures between a Mach 2.4 and a Mach 3 cruise condition has a significant impact on the selection of materials to satisfy the life and durability requirements of the vehicle. Lower-surface skin temperature distributions for both the Mach 3 and Mach 2.4 cruise flight conditions are shown in the figures at the right. For a Mach 3 flight trajectory, temperatures over the acreage areas on the lower surface of the vehicle are about 460 °F and temperatures along the stagnation lines of the wing leading edges are approximately 525 °F. Temperatures along lower surface acreage areas for the Mach 2.4 flight case are 275 °F and are 315 °F near the leading edges. Temperature gradients through the depth of a threedimensional model of a section of the wing bounded by ribs and spars is shown in the figure at the lower right for a Mach 2.4 flight condition. The model accounts for flow of heat by conduction near massive sections of the wing such as ribs and spars. The model accounts for unsteady flow of heat through the honeycomb wing skins to the fuel contained within the wing structure. Detailed three-dimensional thermal models are necessary to predict accurately the temperature gradients which are necessary for accurate determination of thermal stresses.
145
WEIGHTS AND STRESS SUPERSONIC CIVIL
RESULTANTS TRANSPORT
FOR WING
HIGH-SPEED COVER
Stress resultants in a future high-speed supersonic civil transport wing cover are shown in the figure. Thermal and mechanical loads for a 2.5g pullup maneuver were used to size the vehicle and to determine stresses and unit weights. Two different materials were used for the wing structure, a conventional titanium alloy, Ti-6AI-4V, and a quasi-isotropic graphite reinforced organic matrix composite material. The normal stresses in the wing upper cover panels are shown in the figures at the right. As shown in the figure, maximum normal stresses occur in the section of the wing where the leading edge crank occurs. In addition, thermal stresses were considerably lower for the composite structural design than for the titanium structural design due to the lower coefficient of thermal expansion for the composite material. Unit weight distribution in the upper wing cover is shown in the figure at the left. As expected, the weight of the composite wing cover is much less than that of the titanium wing cover.
Unit wei
146
Iht, Ibs/ft 2
INTERACTIVE INTERNAL
MODEL GENERATION CAPABILITY STRUCTURAL CONFIGURATION
FOR AIRFRAME RESEARCH
The time required to generate a three-dimensional structural finite element model from an external aerodynamic shape can be very long, on the order of months. Work is underway at NASA Langley to enhance the Solid Modeling Aerospace Research Tool (SMART) to enable rapid structural modeling of external and internal structure from a given aerodynamic shape. Currently, the computer time required for linear structural analysis is much less than the actual time to create and modify a structural finite element model. As shown in the figure, the purpose of the present research is to reduce the time to model internal and external structure from months to days. This reduction in modeling time will enable a more rapid assessment of internal structural dimensions and arrangements and speed up the optimum placement and sizing of internal structure. Planned enhancements will enable the generation of internal structural configurations such as those shown and enable the rapid rearrangement of internal structures as illustrated by the different structural arrangements in the models shown. Planned enhancements to SMART include a means for: 1) creating and editing structural elements for the wing and fuselage; 2) integrating wing and fuselage structural components; 3) integrating tail and fuselage components; 4) remapping aerodynamic loads data to the structural model; 5) applying point and distributed loads to the structural model; and 6) preparing loads data for visual presentation. Examples of structural elements and components to be modeled include wing spars, ribs, shear webs and cover panels and fuselage skin, frames, bulkheads, Iongerons, and keel beams.
147
DESIGN CONSTRAINT CRITICALITY SPEED SUPERSONIC CIVIL
FOR MINIMUM WEIGHT TRANSPORT WING DESIGN
HIGH-
A future high-speed supersonic civil transport (HSCT) model was optimized using formal optimization procedures to satisfy element stress, local buckling, and displacement constraints. The vehicle structure was sized for a 2.5g pullup maneuver and a 12-foot tip deflection constraint. Sine-wave rib and spar elements were used to minimize thermal stresses in the wing and honeycomb sandwich panels were used for the upper and lower cover panels. Various regions of the structure were governed by different design constraint conditions, either element stress, local buckling or minimum gage constraints. The degree of criticality of each element can be easily monitored by using a simple color coding scheme to identify regions where constraints are critical or satisfied. Upon investigation of a critical region, it can be determined which constraint is approaching
criticality.
Element
Stress
and
Local
Buckling
Constraints
12 ft. Tip Deflection Constraint ...... 2.5g Symmetric Pull-up Maneuver
CONSTRAINT_ CRITICAL
m
Honeycomb
148
Sandwich
Upper
Cover
Panels
SATISFIED
i
I
I
SENSITIVITY OF CIVIL TRANSPORT
MINIMUM-WEIGHT WING DESIGNS
HIGH-SPEED TO MATERIAL
SUPERSONIC PROPERTIES
The sensitivity of a future high-speed supersonic civil transport (HSCT) to the wing-tip deflection constraint is illustrated in the figure. The minimum-weight wing design of a candidate supersonic transport wing is plotted as a function of wing-tip deflection limit for three different structural material choices: a titanium alloy(Ti-6AI-4V), an advanced aluminum alloy (FVS0812), and a quasi-isotropic graphite-bismalimide (Gr/BMI) material (IM7/5260). When the tip displacement is relaxed (e.g., for a tip displacement limit of 15 feet), the titanium and advanced aluminum designs are similar inweight and considerably heavier than the Gr/BMI composite design. As the dip deflection limit is reduced and approaches 5 feet, the advanced aluminum and Gr/BMI designs become similar in weight and considerably less than the titanium design. The advanced aluminum chosen is similar in stiffness to a quasi-isotropic Gr/BMI composite structure. The benefits of the composite material in producing a lighter weight design can be realized if tailoring the structural design to exploit the benefits of the directional properties of the material is permitted. Future work will address the potential benefits of structural tailoring on minimum weight design.
2.5 G Supersonic Strength,
Buckling,
Pull-up
and Tip Displacement
Constraints
70000 __,,,,__Ti-6AI-4V
60000
50000
Minimum Wing Weight, Ibm
_AI
(FVS0812) Gr/BMI (quasi-isotropic)
4000£
30000 20000
10000
0 5
10 Tip Displacement
15 Limit, ft
149
STIFFENED
PANEL
DESIGN
CODE
m
PASCO
The Panel Analysis and Sizing Code (PASCO) (refs. 1 and 2) is a computer design code which combines a rigorous buckling analysis with a nonlinear mathematical optimization algorithm to perform structural analysis and minimum-weight optimization of longitudinally-stiffened composite panels. PASCO is restricted to prismatic structures having an arbitrary cross section. The PASCO program can accommodate the design of fuselage and wing structural panels which can be loaded by any combination of in-plane loads, lateral pressure and thermal loads. Initial "bow-type" imperfections in the panel geometry can also be analyzed using PASCO. PASCO uses a linked-plate representation of geometry in which individual plates are assembled to construct a structural panel cross section such as those shown in the figure. Two or more individual plates are assembled to form a substructure, which is repeated to create the entire panel cross section. Plates are constructed as a balanced symmetric laminate of a prescribed number of plies with orthotropic material properties. PASCO can perform a local or global buckling analysis of a stiffened panel for various combinations of free or supported boundary conditions along the panel edges. Local buckling loads and mode shapes, such as those shown in the figure, are routinely calculated by PASCO. In addition, PASCO can perform a structural analysis of the loaded panel and minimize panel weight subjected to various stress and buckling constraints. A Macintosh version of PASCO has been developed (ref. 3) and an interactive graphical interface to the Macintosh version of PASCO, called MacPASCO, has also been developed. The graphical interface was created to simplify user input and model checkout (ref. 4). PASCO, the Macintosh version of PASCO, and the graphical interface, MacPASCO, are available through COSMIC.
150
STRUCTURALPANEL
-----
LOADING
BUCKLING MODE UNDEFORMED
BLA i i
e
COMPLEX BUCKLING MODES OF ARB ITRARY PANEL CONFIGURATIONS
STIFFENED
,
PANEL
BOW-TYPE IMPERFECTION
DESIGN
CODE
-- PASCO
151
STRUCTURAL EFFICIENCY DETERMINED GRAPHITE/THERMOPLASTIC
FOR OPTIMIZED PANELS
One application of PASCO to size structurally efficient stiffened composite panels is shown in the figure. Minimum weight designs for compression-loaded graphite-thermoplastic panel concepts were developed for a range of loading intensities and the results are compared with current aluminum designs in the figure. The weight, normalized by the planform area, A, and the panel length, L, is shown for different values of applied load, Nx, normalized by the panel length. Two graphite-thermoplastic concepts are compared, one with a corrugated core and two face sheets and one with a hat-stiffened configuration. The concepts are based on a cost-effective fabrication process that allows the corrugated core and hat-stiffened section to be thermoformed and subsequently attached to the face sheets.
-4
.............................................................
20 x 10 j- Commercial aircraft | aluminum wing
.,/
| compression panels _. [. _ 101 __..--r"TC/__j'_
Graphite-thermoplasti. . c panel_ _Corrugated-core
ibiin 3
at-sti
I*.... 5O0
100 NxtL,Ibtin
i
i : [:7' ; b
152
2
L---_ 1000 --_
STRUCTURAL
EFFICIENCY
PANELS
IS
OF
INSENSITIVE
OPTIMIZED TO
CORE
SANDWICH DENSITY,
COVER
Pcore
A structural optimization study of a sandwich panel concept with composite face sheets has been conducted which includes damage tolerance constraints as well as strength and bucking constraints (ref. 5). The results of the study indicate that imposing a maximum strain constraint of 0.0045 in./in, will provide designs with thick enough face sheets to tolerate reasonable low-speed impact damage.
The
significantly compressive weights, minimum
results
of the study
indicate
that core
density
Pcore does
not
affect the weight, W, of the designs over a range of applied loads, Nx. Since core density did not strongly affect these
a heavier more damage-tolerant weight increase.
core
can
be used
for the design
design with
8 6
PCORE = 9.5 Ib/ft 3
W, 4 Ib/ft 2
__
• Orthotropic
I b/ft 3
facesheets
• Response mechanisms: global buckling, facesheet wrinkling, material failure
2
0
I
I
I
10
20
3O
• Damage tolerance _x < 0.0045
constraint:
in./in.
N x, kips/in.
153
EFFICIENT RITZ-BASED ELEMENTS DEVELOPED STRUCTURAL ANALYSIS OF BEAM AND PLATE
FOR THERMALSTRUCTURES
The structural design of supersonic and hypersonic aircraft requires the efficient and accurate calculation of structural temperatures, the transfer of these temperatures to a discrete structural model, and the efficient and accurate calculation of the structural response. Discrete thermal and structural models are often dissimilar and require some form of mapping to transfer temperatures from the thermal model to the structural model. In addition, discrete structural elements like beam and plate elements have no analogous thermal elements which can be used to calculate the temperature distribution through the element thickness. Thermal models require significantly more detail to predict the through-the-thickness variation of temperature. The objective of the present research is to develop thermal elements which are compatible with structural beam and plate elements. Compatible thermal and structural elements can reduce modeling time, problem size, and the computation time required to obtain accurate thermal stresses. Ritz-based thermal and structural elements were developed as a means to alleviate some of the problems associated with thermal-structural analysis. Structural and thermal energy functionals are developed which include parallel formulations for internal energy and the energy associated with boundary conditions. The Ritz method is used to develop the governing equations for the thermal and structural elements. Temperature and thermal stress results for a Ritz-based analysis of a heated beam are compared to linear finite element results as shown in the figure. The structure has mixed thermal boundary conditions and is fixed against translations and rotations at each end as shown in the upper sketch. Results from finite element thermal and structural analyses are shown for comparative purposes. Both the Ritz and finite element results were chosen from convergence studies which examined changes in the temperature and stress as a function of degrees of freedom. Convergence studies for the Ritz-based element required only an increase in the interpolation function order while finite element studies necessitated mesh refinement and associated changes in loads and boundary conditions defined at nodes. The Ritz-based analysis requires only 12 degrees-of-freedom for accurate prediction of temperatures and 34 degrees-of-freedom for accurate prediction of thermal stresses. The conventional finite element analysis requires 99 degrees-offreedom for accurate temperature calculation and 198 degrees-of-freedom for accurate thermal stress calculation. The Ritz-based elements are capable of representing mixed boundary COnditions including convection for a steady-state thermal analysis and prescribed displacements for structural analysis. Orthotropic and layered media can also be modeled with the Ritz-based elements. More detailed information on Ritz-based elements can be found in reference 6.
154
4O 100
225
Z
ends restrained against translation and rotation --o
prescribed
temperature °F
_/_G_'[_> heat load
Btu / in. Temperature
45
2O
375.
°F
324.
Thermal
Analysis
- Temperature
273
I
223 172 i_¸¸
_121
Ritz-based element
Finite element - 99 degrees of freedom
12 degrees of freedom
70.7 20.0
Stress -52.5
StrUctural
Analysis
- Thermal
Stress
-64.9
o x
ksi
-77.3
F
-89,6 -102
F
-114
Ritz-based element 34 degrees of freedom
Finite element - 198 degrees of freedom
EFFICIENT RITZ-BASED ELEMENTS DEVELOPED STRUCTURAL ANALYSIS OF BEAM AND PLATE
-127 -139
FOR THERMALSTRUCTURES
155
NASA/BOEING STUDY THERMOPLASTIC
OF POSTBUCKLING BEHAVIOR SHEAR WEBS WITH HOLES
FOR
Graphite-thermoplastic panels are being considered for supersonic aircraft applications. The results of an experimental study of graphite-thermoplastic shear webs with circular cutouts is shown in the figure (ref. 7). The cutout size was varied in the study from a diameter of 0 to 3 inches and the specimens were loaded to failure in a picture-frame shear test fixture. All panels buckled before failure. The results in the figure show the out-of-plane deflection as a "function of applied shear load for difference cutout sizes. All panels had out-ofplane deflections ranging from approximately 4 to 6 times the 0.080 inch-thick shear-web thickness.
5000 4000 Applied shear flow, Ib/in.
16-ply
quasi-isotropic d=0 /
Q
Failure
3000
/rd /?
= .75 in. d=l.5in.
.30
.40
2000 1000
.10 .20 0 Out-of-plane ................
156
_:_::
-.
deflection,
.50
e_
|_
.............................
in. ...........
............
_-=;:_;_;
..........
POSTBUCKLING
The
results
RESPONSE
of a typical
AND SHEAR
postbuckling
FAILURE WEBS
analysis
OF
THERMOPLASTIC
of a graphite-thermoplastic
shear
web with a 0.75-inch-diameter cutout is shown in the figure. Large shear stress gradients are shown in the right figure that correspond to the locations of failure in the panel shown in the lower figure. The analytical response prediction compares well with the test data as shown in the left figure.
_ 4000
Experiment I
3000F
Nonlinear analys_s | Thickness , *
| lb q'in.
........... __-_--stress
shear resultants
Ib Qmax = 396 in. _:i
Failure ,l_ r
q
200
W
.q
* |
q
d
q
looo_
0
1
2 W't
3
4
Qmin = -165 specimen
Ib In.
157
CARBON-CARBON
CONTROL SURFACE (UNASSEMBLED)
TEST
COMPONENT
Based on results of previous conceptual studies, it is believed that a refractorycomposite material such as Advanced Carbon-Carbon (ACC) offers significant advantages which warrant its selection for control surfaces on a hypersonic vehicle. A generic elevon configuration was selected which would carry significant mechanical loads and reach maximum temperatures as high as 3000 °F. At the present time the elevon system has been designed, fabricated and assembled. The elevon was fabricated by LTV Corporation under the direction of NASA Langley. The major carbon-carbon (C/C) and refractorymetal (Rene' 41) parts which compose the elevon assembly are shown in the figure. The C/C parts consist of a 3-foot by 5-foot built-up structure with rib and skin panels, a torque tube, ten rib-to-tube attachment fittings, a closure piece, and many C/C fasteners and nuts. Rene' 41 pieces include: attachment rings, lugs, fasteners and cleats used to join the C/C torque tube to the wing support structure. Sub-component testing has begun on the C/C torque tube at NASA Langley as will be shown later in the paper. Testing of the full-scale elevon component will be performed in the structures laboratory at NASA Dryden. Significant advancements in C/C design, analysis, and fabrication technology were necessary for fabrication of the test component. The large size of the test article and the need for close tolerances were a significant challenge that required advancement in the state of the art. Fabrication of the rib and skin panel built-up structure, the torque tube, and the rib-to-tube attachment fittings required significant expertise to develop. The high design torque loads and the large difference in coefficients of thermal expansion between the C/C torque tube and the Rene' 41 actuator lug posed a major challenge. The high design temperatures (up to 3000 °F) required the use ofc/c fasteners for the assembly and is the first application of C/C fasteners to a structure that will be subjected to cycllc thermal and mechanical loads. Further details of the C/C elevon can be found in reference 8.
158
I !
!
CARBON-CARBON
CONTROL
SURFACE
TEST
COMPONENT
(UNASSEMBLED)
ORtCtNAL BLACK
AND
WHITE
F;',_ E I>HOTOC, RApH
159
i
CARBON-CARBON
CONTROL
SURFACE
TEST
COMPONENT
(ASSEMBLED)
The figure illustrates a completely assembled Carbon-Carbon (C/C) elevon for a hypersonic vehicle complete with C/C fasteners and attachment rings and Rene' 41 fasteners and attachment pieces. The accurate fit-up and adherence to close dimensional tolerances is a testament to the significant advances in C/C
manufacturing
and
rigorous
fabrication
procedures
program.
OR_GIN,A,L P;,GE BLACK
-
160
AND
WHITE
PHOTOGRAPr_
made
during
this
ELEVON
A three-dimensional
NASTRAN
NASTRAN
MODEL AND DISTRIBUTION
thermal
finite
ITS
element
TEMPERATURE
model
of the carbon-
carbon elevon was developed to calculate detailed temperatures and temperature distributions. The detailed model was necessary to accurately predict temperatures and thermal loads necessary for accurate thermal stress calculation. Temperatures at critical regions where the refractory-metal Rene' 41 fasteners and fittings are located also necessitated accurate temperature prediction. The finite element model uses over 2000 elements and has over 11,000 degrees of freedom. Temperature contours in the elevon and torque tube for a generic flight profile which simulated ascent, cruise, and descent are shown at the right of the figure. Maximum temperatures in the ACC carboncarbon elevon are 3000 °F and maximum temperatures of the Rene' 41 fasteners are 1600 °F.
TEMPERATURE DISTRIBUTION 3100 _ NASTRAN MODEL
2800
2200
161
DETAILED
STRESS
ANALYSIS
OF TUBE
CARBON-CARBON
TORQUE
The carbon-carbon short torque tube is being tested to obtain preliminary performance data for a 9.5-in.-diameter torque tube constructed of 42plies of woven carbon-carbon material. All loads and attachments are similar to those of the carbon-carbon elevon torque tube which will be tested later. The short torque tube model, shown in the figure, was taken from a larger model which consisted of the torque tube, attachment rings, load arms, and support collar. The 0.5-inch-walls of the torque tube were modeled with plate bending elements having quasi-isotropic material properties. Cleat holes and bolt holes were included in the model since the proximity of these openings has an influence on the stress field. Mechanical fasteners were modeled with rigid beam elements which transmit only compressive loads to bearing surfaces. Perfect fit-up was assumed at all bolt holes and cleat holes. The torque loads are transferred to the tube at the bolt holes resulting in a clockwise rotation. These loads are reacted by the cleat holes of the support collar. The compressive bearing strains at the bolt holes occur on the opposite face from the compressive strains at the cleat holes. The largest compressive strains in the torque tube occur where the couple forces from the load arms are applied to the tube. The compressive strain limit of .0013 in./in, for a carboncarbon lamina is about one-third of the tensile strain limit. There is some local bending at the cleat holes and bolt holes since single-lap joints are used to transfer loads. A local three-dimensional finite element model was used to investigate bending effects. Displacements from the planar model analysis were specified as boundary conditions for the local model analysis. The outer fiber compressive strains from the solid model analysis were about 2.5 times greater than the membrane strains obtained from the planar model analysis. This maximum Value iS a.ppr0ximately equal to the allowable compressive strain specified by the LTV Corporation which fabricated the torque tube.
162
I I-
F-
Ee .onng14
.non?g4
CLEAT BEARING .ono494
iiiiiili_ i;i!!ii!i_N. .nn0283
.nonf1728
-.
nno
138
-,
n00340
-,
0n0550
r BEARING BOLT Z DETAILED
STRESS
ANALYSIS
OF
CARBON-CARBON
/ TORQUE
TUBE
163
REFRACTORY-METAL/REFRACTORY-COMPOSITE COOLED WING LEADING
HEAT-PIPEEDGE
A refractory-metal/refractory-composite heat-pipe-cooled wing leading edge concept, shown schematically in the left of the figure, is currently being considered for use on a hypersonic aircraft. The heat-pipe concept has the potential to reduce leading edge weight by 50 percent over an all actively cooled leading edge and to result in a more reliable and redundant design. In addition, the heat-pipe-cooled leading edge concept eliminates the need for active cooling during descent and even the more severe ascent portions of the trajectory. Elimination of active cooling from the leading edge design greatly reduces systems complexities and weight. The concept uses thin refractory-metal "D-shaped" heat pipes embedded within a refractory-composite structure. The heat pipes are spaced close to one another and arranged normal to the leading edge. The heat pipes cool the stagnation region by efficiently transporting heat aft to the upper and lower surfaces of the wing where the heat is rejected by external radiation to space. The heat pipes effectively isothermalize the leading edge because the mechanism for transporting the heat is very efficient and relies on the evaporation and condensation of a high temperature working fluid; lithium in this particular design. The heat pipes are self contained and require no external pumping for the lithium working fluid. The lithium evaporates in the stagnation region and condenses in the aft sections of the heat pipe. The liquid condensate is pumped back to the stagnation region by the capillary pumping action of an internal wick structure. The heat pipes are sized to be redundant in the event of an individual heat pipe failure and the refractory-composite structure surrounding the heat pipes is also designed to offer ablative protection in the event of a massive heat pipe malfunction. If active cooling is necessary during ascent, it can be accommodated by internal radiation to an actively cooled heat exchanger as shown in the figure. A thermal parametric trade study was performed (ref. 9) using a threedimensional finite element model of a portion of a single heat pipe, shown by the shaded region in the figure on the left. Maximum temperatures for a 0.5-inch-radius design are shown in the figure on the right for the case of an uncooled leading edge and an internally cooled leading edge design. Parameters varied in the study were the wetted length of the heat pipes and the heat-pipe spacing. As shown in the figure, if the spacing between heat pipes is less than 0.1 in., a 24-in.-Iong heat pipe cools the stagnation region sufficiently to reduce maximum temperature below a 3000 °F temperature limit for the refractory-composite structure.
164
REFRACTORY METAL/REFRACTORY COMPOSITE HEAT-PIPE-COOLED WING LEADING EDGE
-o-
uncooled
--o-
cooled
/----
L = 12 in.
3600
thp = 0"005 in'_
3400
tsic=0.01in.--__.__ ___.,,.
_
3200
H2 coolant
3000 Trnax, °F
__-D
layup
2800
L=36in._
__-
2600 2400
'- _/
\ -"_/
trc = 0.04 in. J
"
/
3-D weave _Reqion
\
modeled in FEA
2200 200( 0
t_ rh p = 0.25 in.
HEAT PIPES EMBEDDED IN REFRACTORY COMPOSITE WITH OPTIONAL INTERNAL RADIATIVE COOLING
0.04 Distance
0.08
0.12
0.16
between
heat pipes,
0.20 x, in.
THERMAL PARAMETRIC STUDY FROM 3-D FINITE ELEMENT ANALYSIS FOR R = 0.5 in.
165
CARBON-CARBON/HEAT-PIPE-COOLED STRESSES AT
A detailed thermal ACC4 carbon-carbon
THE
LEADING STAGNATION
EDGE LINE
THERMAL
stress analysis of a molybdenum heat pipe embedded structure, results in very high compressive stresses
within in the
molybdenum tube as shown in the figure (ref. 9). At elevated temperatures, the refractory-metal molybdenum "D" tube expands into the lower coefficient of thermal expansion refractory-composite structure. The linear structural analysis does not account for the relief in thermal stress by yielding of the metallic tube. In addition, the structural analysis assumes a perfect bond between the refractory-metal and refractory-composite materials. Methods for alleviating the high thermal stresses are currently being investigated. The use of a soft carbon strain isolator material placed between the metal "D" tube and the refractorycomposite differential stresses.
structure thermal
is currently being expansion without
investigated the creation
as a means of excessive
of allowing thermal
the
ay, psi _
coolant
'x t¢..¢
/"
Region modeled
__ in FEA
qs = 900 Btu/ft2-s R = 0.5 in. s =36 tc_c = 0.06 in.
y
166
in.
x = 0.5 in.
3-D Carbon-carbon
structure
THERMAL-STRUCTURAL
PARAMETER
ESTIMATION
Parameter estimation methods are currently being developed to determine thermal as well as structural parameters and to help facilitate correlation between experimental and analytical results. Parameter estimation techniques are currently being used to determine the thermal contact resistance of refractory-metal heat pipes embedded within a refractory-composite structure. The estimation procedure is implemented within a commercially available thermal-structural finite element program called the Engineering Analysis Language (EAL) System (ref. 10). The flow chart illustrated in the figure depicts the procedural flow which begins with the development of a finite element model which represents a physical experiment. Certain parameters are considered known constants in the finite element model and certain parameters are considered to be variable. Analytical results are compared with actual or test results and a measure of the error or lack of correlation between results is calculated. If the correlation is poor, the sensitivity of the response to parameter variations is calculated and a least-squares procedure is used to minimize the error and predict a new value for the parameters. The procedure is repeated until the error or lack of correlation between actual and calculated responses is reduced below some prescribed tolerance. A numerical experiment is illustrated in the figure to highlight the convergence of the parameter estimation procedure implemented into EAL. The transient thermal history of a point located on the outer surface of a tube heated internally is predicted numerically assuming a known value for thermal conductivity (kactual=0.0013875 Btu/in.-s-R). The estimation procedure begins with an initial estimate for kactual which is kcalculated=0.001 Btu/in.-s-R. The procedure automatically converges to the actual value for thermal conductivity in only three iterations. General methods for estimating multiple parameters of complex problems have been incorporated into the program. Future plans include extension of the methods to structural problems and the use of sensitivity information to optimally design experiments to determine accurately the variable parameter values.
167
THERMAL-STRUCTURAL
Initial Model
PARAMETER
ESTIMATION
[ !
Select Parameters _, Model with Compare Actual
I
re= 0.30
I
"";
ri = 0.25
k = '_
]
•
,
•
i¢
J
¢
Convergenceof Thermal Conductivity 1.5e-3
kactual
1.4e-3
= 0.0013875 •
k
1.3e-3
(Btu/s-ln-R)
I e
Update
Minimize Error
I Parametersl-
I (Least Squares)
lated
1.2e-3
1.1e-3 1.0e-3
i
0
I
I
i
I
2 Iteration
168
Btu/s-in-R O
i
I
3
i
!
4
ANALYSIS
OF
LIQUID-METAL-COOLED LEADING EDGE
NASP
ENGINE
COWL
Hypersonic vehicles experience extremely high aerodynamic heating because of the high Mach numbers that the vehicle attains while traveling within the earth's atmosphere. A particularly severe condition exists when the vehicle accelerates and the bow shock from the nose of the vehicle intersects the bow shock of the engine cowl lip. The shock-shock interference produces extremely high and local heating, 50,000 to 100,000 Btu/ft2-s acting over a region 0.01 to 0.02 inches wide. To predict accurately the thermal performance of a convectively cooled cowl leading edge subject to high local heating requires an accurate description of the internal fluid flow and heat transfer. Computational fluid dynamics (CFD) analysis of the coupled convective flow field and solid conduction to the cowl skin was used to predict accurately the maximum temperatures and temperature gradients within the thin copper skin of an engine cowl leading edge. Previous results, using engineering approximations to determine heat transfer characteristics, did not adequately represent the growth of a local thermal boundary layer and were overly conservative. A section of the curved cowl lip (see figure) was analyzed as a planar section. The schematic diagram in the figure is not to scale; the true horizontal scale is on the order of inches, the vertical scale is on the order of hundredths of inches, and the heat pulse width is on the order of thousandths of inches. The coolant enters the cowl lip with a uniform temperature and velocity profile. Velocity boundary layers develop and cause the velocity profile to change to one having a nearly uniform velocity core with regions of high shear near the walls. Near the region of the high heat pulse, a thermal boundary layer begins to form in the coolant. The coolant outside this thermal boundary layer is at, or very close to, the inlet temperature while the temperature within the thermal boundary layer is hotter than the inlet temperature. Results for a local 0.015-in.-wide heat pulse which has a magnitude of 50,000 Btu/ft2-s is shown at the right of the figure. The cowl lip walls are 0.02-in.-thick copper with a temperature limit of 1200 °F. Results of maximum surface temperature as a function of coolant velocity are shown in the figure for three different coolants: hydrogen, water, and liquid sodium. The high conductivity and thermophysical properties of sodium result in high heat transfer coefficients and lower maximum temperatures. Results of the study indicate that, depending on coolant velocity constraints, several liquids could potentially accommodate high local heating representative of the shock-shock heating condition for a hypersonic vehicle.
169
ANALYSIS OF CONVECTIVELY COOLED ....... ENGINE COWL L_DING EDGE
"
"_ _Coolant
Velocity boundary layers Thermal boundary layer :::¢
Metal skins
170
NASP
AIRCRAFT
STRUCTURES
RESEARCH
TOPICS
The development of verified structural analysis and design technology for future supersonic and hypersonic vehicles requires research on a number of topics as indicated on the figure. Advanced structural concepts are needed for costeffective structurally efficient designs. Structural tailoring can be used to exploit the beneficial properties of advanced materials. Failure often initiates at stress gradients in a structure, so gradient producing discontinuities and eccentricities must be understood. Local two-dimensional and three-dimensional analyses of these local gradients are needed that are consistent with global twodimensional plate and shell models. Nonlinear effects associated with postbuckling design philosophies and pressure and thermal loads must be accurately predicted analytically and minimum-weight designs developed for nonlinear structural response. Failure mechanisms must be understood and failure analyses developed to predict accurately the onset of these failure mechanisms. Damage tolerance requirements must be understood for future high speed vehicles and designs must be developed that safely tolerate local damage. The interaction between subcomponents and elements in built-up structures needs to be understood and minimum-weight joint technology needed to connect the elements and subcomponents is needed. Scaling laws for composite and metallic structures are needed to minimize vehicle development costs. Thermal effects and heat transfer into the structure must be predicted addition,
to determine thermal stresses light-weight thermal protection
concepts withstand
are needed. High-speed combined mechanical,
and thermal buckling systems and actively
aircraft pressure
of a structure. In cooled structural
structures must be designed and thermal loading during
to their
flight profiles so the interaction of these loads on structural performance must understood. All of the analysis and design methodology developed for highspeed vehicles must be verified in the laboratory with the appropriate experiments with panels and wing-box and fuselage-shell models. •
Structural
efficiency
studies
and advanced
•
Structural
tailoring
•
Gradients,
•
2-D global analysis with 2-D and 3-D local analysis
•
Postbuckling
•
Nonlinear
•
Failure mechanisms
•
Damage tolerance
•
Subcomponent
•
Scaling
•
Thermal
•
Thermalstresses
•
Combined
•
Panels and subscale
and anisotropic
discontinuities,
cutouts,
and geometric
analysis
be
concepts
effects and eccentricities
nonlinear
effects
and sizing procedures and failure analysis and low-speed
interaction
and optimum
laws for composite
andthermal
effects
joints
structures
effects and heat transfer
mechanical,
impact damage
into interior structure
buckling
pressure, fuselage-shell
and thermal
loads
and wing-box
models
171
REFERENCES
Anderson, M. S.; Stroud, W. J.; Durling, B. J.; and Hennessy, K. W.: PASCO: Structural Panel Analysis and Sizing Code, User's Manual. NASA TM 80182, November 1981.
.
Stroud, W. J.; and Anderson, and Sizing Code, Capability 80801, November 1981.
.
Structural Panel Analysis Foundations. NASA TM
Lucas, S. H.; and Davis, Randall C.: User's Manual for the Macintosh Version of PASCO. NASA TM 104115, September 1991.
,
Lucas, S. H.; and Davis, NASA TM 104122, 1991.
,
Randall
C.:
User's
Manual
for MacPASCO.
Cruz, Juan R." Optimization of Composite Sandwich Cover Panels Subjected to Compressive Loading. NASA CP 3104, Part 2, 1990.
.
Vause, R. F.: Compatible Thermal and Structural Elements for Thermal Analysis of Beam and Plate Structures. Presented at AIAA/ASME/ASCE/AHS/ASC 32nd Structures, Structural Dynamics, Materials Conference, Baltimore, MD, April 8-10, 1991. AIAA Paper 91-1154.
.
the the and No.
Rouse, M.: Effect of Cutouts and Low-Speed Impact Damage on the Postbuckling Behavior of Composite Plates Loaded in Shear. Part 2, A Collection of Technical Papers presented at the AIAA/ASME/ASCE/AHS/ASC 31st Structures, Structural Dynamics, and Materials Conference, Long Beach, CA, April 2-4, 1990, pp. 877-892. AIAA Paper No. 90-0966.
.
.
.
10.
172
M. S." PASCO: and Analytical
Ho, T.; Matza, E. C.; Medford, J.; and Watabe, S.: Design Concept for NASP Control Surface. NASA CR 18173, October, 1988.
Study
Glass, David D.; and Camarda, Charles J.: Preliminary Thermal/Structural Analysis of a Carbon-Carbon/Refractory-Metal HeatPipe-Cooled Wing leading Edge. Presented at the First Thermal Structures Conference, University of Virginia, Charlottesville, Virginia, November 13-15, 1990. Whetsone, W. D.: EISI-EAL Engineering Analysis Language Reference Manual- EISI-EAL System Level 2091. Engineering Information Systems, Inc., July 1983.
_92-25919
High Speed Civil Transport General
R. L. McKnight Electric Aircraft Engines Cincinnati, Ohio
173
CONCORDE
HSCT
3,000
RANGE
100
PAYLOAD
400,000 EXEMPT
COHHUNITY FARE
APPROX20 of
5,000
(PASSENGERS)
WEIGHT
PREHIUH
Comparison
(NML)
250
STANDARD
LEVELS
HSCT
design
requirements
TO
300
STANDARD
N0x EMISSIONS (GM/KG) the
6,000
750,000 FAR 36 STAGE III
(LB)
NOISE
TO
3
with
the
TO
8
present
CONCORDE.
HIGH SPEED CIVIL TRANSPORT ENVIRONMENTAL
IMPACT MAJOR DRIVER
-
EMISSIONS
-
NOISE
CRITICAL COMPONENTS -
HIGH TEMPERATURE
COMBUSTOR
-
LIGHT WEIGHT EXHAUST NOZZLE
NEW AND ADVANCED MATERIALS
174
-
HIGH TEMPERATURE CMC'S
-
HIGH TEMPERATURE IMC'S/MMC'S
day
KEYM_R
MATERIAL REO_UIREMENTS
HIGH OPERATING TEMPERATURE HIGH THERMAL STRESS RESISTANCE ACOUSTIC/VIBRATORY DURABILITY ENVIRONMENTAL DURABILITY DAMAGETOLERANCE SHAPE-FORMING CAPABILITY REASONABLE COST Design combustor
projections material
indicate will
need
that to
a
successful
possess
the
HSCT noted
characteristics.
175
KEY NOZZLE MATERIAL REOUIREMENTS
HIGH SPECIFIC STRENGTH THERMAL STABILITY ENVIRONMENTAL RESISTANCE THERMAL MECHANICAL/ACOUSTIC FATIGUE RESISTANCE THERMAL SHOCK/STRESS CAPABILITY DAMAGE TOLERANCE GOOD FABRICABILITY AFFORDABLE COST Design
projections
nozzle
material
characteristics.
i
176
indicate will
need
that to
a
possess
successful the
noted
HSCT
HSCT PROGRAM WILL PROVIDE HIGH-TEMPERATURE ADVANCED COMPOSITE MATERIALS, INCLUDING FIBERS AND MATRICES IMPROVED PROCESSES FOR FABRICATING ADVANCED COMPOSITE COMPONENTS ANALYTICAL TOOLS FOR COMPOSITE AND COMPONENT DESIGN, FABRICATION, AND LIFE PREDICTION IMPROVED PROCEDURES FOR TESTING COMPOSITE SUBCOMPONENTS
COMPOSITE SUBCOMPONENTS FOR ENGINE TEST ANALYTICAL TOOLS DEPEND ON MATERIAL CHARACTERISTICS MANUFACTURINGBEHAVIOR FUNCTIONAL BEHAVIOR FAILURE BEHAVIOR FAILURE MODES DAMAGEACCUMULATION CONSTITUTIVE MODELS THERMAL i
STRUCTURAL
i
MACRO, MESO, MICRO
LIFE MODELS -
MACRO, MESO, HICRO
CLOSE COUPLING BETWEEN LIFE MODELS AND CONSTITUTIVE MODELS 177
.4.
Tailodng of Composite Structures I
Thermal. and GasDynm_ic
_r
_; ........
..,
..
, .;_._._._:_; Probablislic
(Analysis OpSmlzer I_lhods) Moisture - Temperature - Stress Mat_al ProperSes Space
_-_J
SensiSvity Methods
CSTEMOVERVIEW _CSTEM"
is
contract, of
Graded
gram
the
were
to
produce
tailoring.
ulation
specialized
The
enabling
element
effective, CSTEM of
nate.
The
element
tural
analysis
CSTEM
178
and
shown
are
analysis
tailor
are
structures
having
disciplines large acoustics,
prohigh
involved
deformation
of
and
are
struccoupled
dis-
the of
sim-
aeropropulsion
implemented
simultaneously
carried
with to
model out.
mesomechanics
system.
the
obtain
micromechanics the
this
a computerizedmultidiscipline
problems
complexities
begins
finite
design
is
NASA
in
structures.
a
special
geometrical,
composite
3D
finite
material,
structures
for
cost
performance.
models
Also
the
for
multiple
analysis,
system
the
objectives
anisotropic,
thermal
environment
analysis
The
capabilities
composite
elements
under
Analysis/Tailoring
technical
cost.
include
CSTEM
for
to
optimum
ety
developed
heterogeneous
low
They
technical
and
The
and
The
formulated
loading,
program
The
anisotropic
cipline
computer
anisotropic,
nonlinear.
analysis,
the
Structures."
performance
highly
tural
for
Structural/Thermal/Electromagnetic
Composite
structural all
acronym
"Coupled
is The
constituent material then
analysis
models
to models
properties
and
characteristics constructed is obtain and
and
then
how
the
carried
local they
uses
for
are
global down
stresses
a vari-
the
lamistruc-
through and
related
the
strains. to
the
In
each
enabling
element
code
iterative
technical
has
been
solution
participating
discipline,
developed.
An
techniques
analysis
performs
modules.
Each
passing only the required input the modules as well as returning an
analysis
The
by
structural
similar
The
on portions
advanced
heterogeneous ative to the tion
point
coordinate emental global
system. coordinate
of
the
or
coupling
module
is
finite-
among
self
the
contained,
information be needed
20-nodedisoparametric
finite
set
of
of
the
structural
desired
Material
version
element nodal
the
as
between input for
bricks
codes
in many
displacement,
equations
with
the The
of
module
orientation may
of also
orientation structural
nodal
computer
is
It
force,
used is amultiproblems since
include
the be
can point
between
the material properties of the based on the volume ratios of
anisotropic,
be input relby integrawith
referenced
module
program
its
material
to
the the
elemental
can
it needs for composite make up the composite. the
and
ways.
separately.
Material properties skewed integration
properties
properties properties that
data bank containing erties are calculated
3D
controlling
loadings. The solution technique allows the solution of very large
internally.
adapted
nonlinear
analysis
acceleration,
system,
calculated
internal
16-,
material capability. material axes and then
anisotropicmaterial the constituent an
8-,
features
to obtain
with
geometry and control any results that may
isoparametric
and pressure solver, which
work
more
uses
other
the
stand-alone,
program
module.
centrifugal,
temperature, block column can
module
to many
accommodates
it
a following
a decoupled, executive
also
el-
and
generate
materials, This is done INHYD,
global
which
constituents. the constituents.
the
using using
accesses The
a
prop-
179
COUPLED STRUCTURAL/THERHAL/? COHPUTERHARDWARE/ALGORITHHS DESIGN OF EXPERIHENTS (TAGUCHI) TOO HUCH DATA-NEED EVALUATION STRATEGY/DATA BASE STRATEGY ANALYSIS-EXPERIHENT-ANALYSIS
FEEDBACKLOOP
HARDWARE/CODE/USERGOODNESS-OF-FIT PROBABILISTIC
DESIGN
-
HATERIAL SYSTEM
-
STRUCTURES
RISK ANALYSIS -
DETERMINISTIC
-
PROBABILISTIC
STRUCTURALTAILORING AND OPTIHIZATION
180
-
HULTIDISCIPLINE
-
HULTIOBJECTIVE
HIGH SPEED CIVIL TRANSPORT A NEW LEVEL OF COOPERATION AMONG t
GOVERNMENT/INDUSTRY/UNIVERSITIES
181
tNI_UT GEOMETRY
/
LOAD
NO
BUCKLING ANALYSIS?
ANALYSIS?
YES
NO
STRUCTURAL RESULTS PRINT
MATERIAL
NO
_
TRANSFORM STRESS/STRAIN UPDATE
STIFFNESS
NO
EIOENVALUE ANALYSIS7
DISPLACEMENT
UPDATE STRUCTURAL STIFFNESS
HGURE 1. CSTEM FLOWCHART
Flowchart CSTEM.
executive
to
182
the
These
analysis
the
of
modules
package
modules
process
as -
at
modules
used
entry
or
is
analysis are
with
routine,
tailoring these
major
the
a
through
the in
as
load
stand-alone the
analysis
which
of
case case
main
portion the level.
entry
of
F.SIGN VARIAB_ 0NSTRAINTS
1
DESIGN VARIABLE SET INITIAL VALUES
co NSTRAINT
FIGURE
CSTEM
tailoring
computer
program
consists
of
the
actual
to
abstracted
from
structural,
2, TAILORING
obtained major
tailoring,
parameters
VALUES
capability
two
be
and
been
built
from
NASA
Lewis.
and
CONMIN,
The
and
FLOWCHART
has
ANALIZ,
tailored.
thermal,
PROCESS
modules:
STAEBL
]
coupled
acoustic
the This
which
which CONMIN
on
with analysis
the
program
performs
supplies module
STAEBL
the was
CSTEM modules.
183
I I Compression
Tension
Tension q
I i
A
I I
I I Compression
ILT/ILS Ten/Camp Static/Fatigue Creep/Stress Oxidation
© Tension/Compression Load e/d, w/d (Ten) ab (Camp 8 = 4%)
Tension/Compression Load Static/Fatigue Creel:#Stross Oxidation
Impacl After Impact
t
Static/Fatigue compresslon/Tenslon Creep/Stress Oxidation
Joined Panels Biaxial Loading
Validation is
an
of
important
required
for
Configuration
part the
for
interactions
]84
of
the
models
HSCT
stress
laboratory
oxidation
in
will
shapes.
difficulties
and simpler
design
methodology
Validation
specimens
component
represented
and
process.
creep,
generic
fabrication not
of
analytical
static,
shaped
representative account
the
and be
When stress
laboratory
will fatigue
designed tested
or
be arenas. to these
failure specimens.
mode
be will
CRITICAL
SCREENING
THERMAL
CYCLING
THERMAL
MECHANICAL
FATIGUE DETAILED
ACOUSTIC
TENSILE
1 PRELIMINARY m
PRELIMINARY
CHARACTERIZATION
LOW
CYCLE
BURNER
:
CREEP
LCF
FATIGUE
FATIGUE
RIG EVALUATION
AND
HCF
CRACK
GROWTH
TOUGHNESS THERMAL
MECHANICAL
EFFECT
OF
EFFECT
OF
ORIENTATION
PHYSICAL
PROPERTIES
SCREENING
ORIENTATION
FATIGUE
OXIDATION
EFFECTS
EFFECT
ON
OF
STRENGTH
DOWN SELECT 4 MATL'S.
CREEP OXIDATION THERMAL
AND
STRENGTH
SHOCK
STATISTICAL
ON
STRENGTH BURNER
CONFIDENCE
DETERMINED THERMAL
THERMAL
ON
CYCLIC
EXPOSURE
TENSILE
PHYSICAL
CHARACTERIZATION
FATIGUE
PROPERTIES
SHOCK DESIGN
DATA
PACKAGE
MATERIAL
PROPERTIES
A
detailed
HSCT
evaluation
materials
will
test be
developed.
mechanical
and
physical
techniques
for
environmental
and
candidate
NDE
plan
testing, and
for
the It
candidate
will
include
characterization thermal
behavior,
techniques.
185
CONSTITUTIVE
I
HEAT
MODELS
TRANSFER
ATTAC HM E NTS ACOUSTICS
INTEGRATED
DESIGN
CODE
METHODOLOGY
DAMAGE
PROBABILISTIC
METHODS
THEORIES
o STRENGTH
(WEIBULL)
o DESIGN
(MONTE
1
I COUPON
BENCHMARK
TESTS
TESTS
The
theoretical
the
rISCT
design base
186
program
models
and
will
methodology established
CARLO)
be
and by
the
analytical integrated
life
prediction
coupon
and
tools to
provide system
benchmark
developed an with tests.
in
overall a
firm
MMC
model-
Creep
.35
vf,
1 IOOF,
Redistribution 120ksi fiber-stress
300
250 n
13--x--
matrix-stress
2OO ,u
J_ V
(n 150 (n (I) i,. (,n
I O0
5O
I
0
200
I
I
I
4OO
6OO
8O0
time
A
creep
model has
simulation in
not
a
of
nonlinear
occurred
(hours)
an
MMC
finite after
I
1000
50
unit element
cell
utilizing
code.
an
overlay
Stabilization
hours.
187
MMC
model-
Creep ..35 vf,
Redistribution
1 IOOF,
120ksi
250"
--B--
fiber-stress
---x--
motrix-stress
...-----El
200
'_ 150 (/I (/)
.,-, 100 ¢)
5O
0
0
!
_
10
20
w
time
A
creep
model not
188
simulation in
a nonlinear
totally
complete
of
Ir
30
!
40
50
(hours)
an
MMC
finite after
unit
cell
element I000
utilizing
code.
hours.
an
overlay
Stabilization
is
N92-25920
Structures Technology Applications National Aerospace Plane
for the
T. E. Little Pratt & Whitney West Palm Beach, Florida
189
NASP
Design
Requirements Challenge Structure Technologies
Current
Achieving the National Aerospace Plane (NASP) operational objectives of Mach 25 and single stage to orbit (SSTO) will subject the vehicle to extreme loading conditions and will require large, actively cooled structures while meeting very stringent weight goals.
• Mach 25 • Single stage to orbit (SSTO)
Lo_oad g
Configuration
• High temperatures
• Actively
• High acoustic
• Large panels
loads
• Minimum
• High pressures • Shock interactions • Aerodynamic
190
loads
cooled structure
weight
Trajectories
For
Hypersonic
Vehicles
The NASP is an air breathing, single stage to orbit vehicle (A/B SSTO) which is to take off, achieve orbit and land under its own power. Typical trajectories for the X15 experimental aircraft, the Space Shuttle and NASP show that NASP will achieve much higher velocities at lower altitudes.
400 [-
_
//
x.15 7
.
/
,, "_SHUTTLE
A/B SSTO
100 / 0
_ ___-- -
._//
(ASCENT) Z
_ 10 VELOCITY,
j
I 20
I
1
3O
k ft/s
191
Heat
The
combination
of
very
Flux/Life
high
speed
Comparison
at
relatively
low
altitudes
causes
the NASP airframe to experience heat fluxes greater than current gas turbine blades. The Cowl Leading Edge (CLE) is subjected to extremely high heat fluxes from the convergence of different shock waves on a very small position of the leading edge. Fortunately, the NASP design life requirement is "only" 150 cycles compared to 4000 cycles for a gas turbine engine. However, the cyclic application of such extreme thermal loads presents a very challenging structural problem.
GAS
TURBINE BLADE 4,0OO
CLE 4,000_oo.00o-
g5.000 3,500-
.b
_
3,000-
_
2,500-
_2,000X 1,500_
40.O00NOZZLE
SSME
__I.000(.o 09
TFi:IOAT i 4,400 NASP
ol___J
192
ENGINE
2, IO0
NASP 200
A/F
GAS
TLII_INE
I----]
BLADE
_J_J
IO0
500-
ENGINE t 50
SSME 50
Engine
Predicted
acoustic
engine.
levels
These
the
NASP
the
percentage
acoustic
design.
permissible
for
NASP
static
Gee
Engine
1 77
Load
much
greater
are
a
pressure
very levels
pressure
continued
.196
are loads
Dynamic of
Acoustic
Comparison
than
for
a
have
fluctuation
gas
been in
estimated
the
f
,
in based
on
combustor
operation.
Turbine
Shuttle Payload 159
Bay_
100
NASP 200
turbine
consideration
signiffcant
¢
Engine
18.3 /
ffl
1SO
_
-_Vctiv dCL > C' _h -
flutter
speed
divergence lift-curve
frequency
speed slope
217
Having
identified
the
performance
measure
sign variables and the constraints, mization problem is given in this F(x) _
Performance
Function
_(x) _
Constraint
Functions
xl,x2,...,Xn--*
.Minimize
Design
(such
a formal slide.
as
weight),
statement
of
Variables
or maximize
a function
F(x)
F(_) = r(x,, _,_,..._,,) subject
to the
inequality
constraints
_,(xl,z,...x.}
and
eq.ality
< _,
constraints
_,(z,, z2...x,,)= _,
218
j = 1,2...k
j = k-I-
l,...s
the the
de-
opti-
A
two
ble)
variable and
lines.
The
constraint gradients
design
space
unacceptable gradients
of
functions provide
is
illustrated
(infeasible) the
are
objective
designated
search
information
with
regions
acceptable
separated
function by
VF
for
and locating
by
(feasithe
constraint
(performance) VG
and
respectively. the
the These
optimum
(best)
design.
i ×2 COHSII1AINr
z'
kx_
!
FEASIBLE
_oNsTRAINF
V _ INFEASIBLE
REGION
BOUND
AI1Y
VG2
REGION _
Xl
--
219
The key elements of the trated in this schematic tified i.
by
dotted
Loads
boxes
of an aircraft The three main
are illusareas iden-
are:
determination
• Inertia • Steady 2.
line
preliminary design (block) diagram.
loads
(mass
properties)
Airloads
Structural
model-optimization
• Analysis • Sensitivity • Optimization 3.
Aeroservoelasticity • Aeroelastic instabilities • Controls
Two key transformations (load and often contribute to singularities polations
(spline
or
other
_nertia
Loads
displacement transformations) due to improperly defined inter-
interpolations).
I
Maneuver
_
Loads
I
F_exible Air Lqads
f
t Steady Aero Model
__
i
Loads RigidAir
t_
Forces Transformation
Structural Modes
_
....
AIC
"
_t
Corrections Aeroelastic
_
Displacements
Eigenvalue Analysis Stress Analysis
] :_
,
.i I
Sensitivity
Optimization
|
Search
Aer°elastic
l
I,.1Unsteady ModelAer° '
/
Analysis
I__.
/ °b'e_"ve Gr_a'e"'_ [_
Fig 1, Interdisciplinary
Directon
Step Size Optimality
Condition
Performance II
220
I
/
Model
SIC
_
Transform
_ I
Design
Eval
Interest
in
interdisciplinary
A
of
structural
number
tested.
Some
grams
seventies.
ing
various
tural
for
ment
(RAE)
in
in
'_ASTROS'' the
maining
Force
At
It
program
It
is -
is
being
loads
-
the
is
being
-
well
of
Wright
a
struc-
Establishto
system
include
developed
by
aeroelasticity (DASSAULT)
as
pro-
during
basically
developed as
are
integrat-
Aircraft enhanced
structures
US
ASTROS
Review of Structural
.
it
of
is
world. and
TSO
Laboratory
Royal
Optimization
Directorate
describe
the
system the
-
feasibility
by
the
developed
FASTOP
optimization
similar in
being -
_'STARS''
an
STRuctural
Dynamics
viewgraphs
a
marketed
automated
Flight
is
around
Dynamics the
present,
includes
are ASOP
Flight
developed
_LAGRANGE''
ELFINI
here.
established The
England.
Germany.
France.
Air
have
system
optimization.
for
the
widespread
systems
listed
disciplines.
aeroelasticity. MBB
are
They
optimization
is
optimization
examples
developed
the
systems
other
System
and in
countries.
was
Laboratory.
developed The
re-
system.
Optimization
Systems
ASOP--FASTOP • TSO
I
•
i
STARS •
LAGRANGE •
ELFINI •
I I
ASTROS
221
ASTROS
is
a
computer tem
which
(Matrix tem The
is
and
ported
by
a
namic
and
of
and
on
for
aeroelastic The
optimization
enhancements
control in
is
flutter
analytical
static
ipate
and
used
for
aeroelasticity
response
the
module
calculation.
gradients
of
constraints
are
module include module
is a is
included at
method the
calculating unsteady
sensitivity
present based
weakest,
in
on and
ADS
variety
package,
does
of anal-
and
criteria. not
optimization.
AUTOMATEDSTRUCTURALOPTIMIZATIONSYSTEM
222
is
objective a
sensitivity
optimality it
maneuaerody-
the and
the
the
steady
analysis
and
flutter
dy-
mechan-
on
on
constraints
frequency,
of
based
of
based
The
the
displacement,
is
purpose is
model
method
constitute
variety
module
sup-
Eigenvalue
a
ysis.
The
optimization.
are
dynamics
element
includes
Stress,
planned
finite
analysis
The
function.
in
airloads
sys-
extremely
database
and
the
static
theory
based
structures on
The
The
an
and
analysis
loadings.
MAPOL
executive
provide
response
aerodynamic
theory
and
transient
ver namic
applicable
sys-
database called CADDB. Both the executive
system
The based
executive called
This
and
thermal
loads.
elements
is
an
frequency
analysis.
ical
flexible executive
sophisticated
language
scientific called ICE.
modules. and
highly through
computer
highly The
a is
_anguage).
designed CADDB is
are
_ASTROS''
with
ASTROS
Oriented
engineering
library
of
high-level
well of
database
to
analysis
a
environment.
six
central
with
in
Program by a version
the
friendly
system
Control
written
is supported interactive
user
optimization
abstraction
system
is
design
architecture.
partic-
OBJECTIVES AND PAYOFFS OBJECTIVES •
t
AN AUTOMATED PRELIMINARY
•
•
DESIGN
INTERDISCIPLINARY
DESIGN
PROVIDE
FOR
STRUCTURAL
EMPHASIZE OF THE
TOOL
FEATURES
TASK
A NATIONAL
RESOURCE
t
PAYOFFS •
IMPROVED
COMMUNICATION
•
IMPROVED
DESIGN
•
REDUCED
DESIGN
AMONG
DESIGN
TEAM
MEMBERS
TIME
ASTROS ARCHITECTURE
EXECUTIVE SYSTEM
I FUNCTIONAL MODULE
1
"
l
IFUNCTIONA MODULE
k]
MODULE
I
J
L SOL!TION
223
Architectural Highlights •
•
Executive
System
-
Provides
High
-
Enables
Multidisciplinary
•
Design
for Engineering
Necessitated
•
Control
Database Customized
•
Level
Dynamic
Recoding
and Design
of Software
Resources
Memory
-
Enables
-
Provides
Utility
Major
Analysis
Unrestricted
Problem
Programmer
Size
with Precise,
Explicit
Control
Required
By Modules
Library
-
Special
Purpose
-
Emphasis Algorithms
-
Machine
Routines
Placed
Dependent
on High
Quality,
Robust,
Self
(Sort,
Documented
Functions
Isolated
Functional
and Utilily Modules
Modules Distinction -
Each
Between
Module
Blurred
:
Establishes Base Address in Memory Opens Required Data Base Entities Closes All Dala Base Enlities Prior to Exit Frees All Memory Blocks Prior to Exit
-
224
Search,
Inlermodular
Communication
is Through
lhe
Data
Base
etc.)
ENGINEERING DISCIPLINES
_
AERODYNAMIC
VREO
LOADS
/
r_
K ---=----
"_AEROELASTIC
STRUCTURAL
ANALYSIS
/
STABILITY
_S
U__---_----"-_
_
9
_.\ _ / \'
OPTIMIZATION
CONTROL
STRUCTURAL
E_._...,.
._
/
,
/
TECHNIQUES
RESPONSE
ANALYSES
o
THERMAL STATIC
ANALYSIS
LOADS
-5o
_'_,.,.,,
_-'°r" _ _,._a_r EIGENVALUE ANALYSIS
_ _,_oo
FREQUENCY DYNAMIC
(Hz)
ANALYSIS
225
BASIC ASTROS SEGMENTS
['"'_'_'i z_'°"[ DESIGN
[ F,NA, ANA,¥S,S [
---_ R__2ES, GN i
l
Software
Resources
Structural
Optimization
226
for ASTROS NASTRAN
Analysis
Static Aerodynamic
Unsteady
FINAL ANALYSIS
Loads
Aerodynamic
Algorithms
Loads
--
USSAERO
--
Doublet CPM
--
MDOT
Lattice
Ten Software Contributions of ASTROS •
Framework
For Multidisciplinary
•
Engineering
Data
•
High
•
Obsolescence
•
Unlimited
•
Exploitation
•
Built
•
Improved
Special
•
Balanced
Approach
•
hltegration
Level
Analysis
and
Design
Base
Executive
System
of Rigid
Problem
Formats
Size
of Microcomputers
In Maintenance
Features Purpose
Utilities
to Software
of Dispersed
Design
Development
Team
Ten Engineering Contributions of ASTROS •
Multidisciplinary
°
Analytical
•
Approximation
•
QUAD4
•
Improved
•
Innovative
•
Nuclear
•
Advanced
•
Design
•
Aerodynamic
Analysis
Sensitivity
and
Analysis
Concepts
Element
Supersonic
Blast
Public
Unsteady
Design
Analysis
Methods Variable
in a Production
in the
Flutter
Design
Code
Domain Aerodynamics
Technique
with
Finite
of Dynamic
Elements
and
Advanced
Aerodynamics
Reduction
Linking
Influence
Coefficients
For
Static
Aeroelasticity
227
AN ARCHETYPICAL
ASTROS APPLICATION
GIVEN: STRUCTURAL MATERIAL
CONFIGURATION PROPERTIES
DESIGN
FLIGHT
CONDITIONS
DESIGN
ALLOWABLES
DETERMINE THICKNESSES
OF DESIGNED
OPTIONALLY POSSIBLE
705 276 1167
NODES
DESIGNED
FIXED
-
MASS
DESIGN
BALANCE
CONSIDERATIONS
BOUNDARY CONDITIONS FLIGHT CONDITIONS
MULTIPLE
STORE
LOADINGS
ELEMENTS
ELEMENTS
VARIABLES
I • ROD AREAS • SHEAR ELEMENT • MEMBRANE • BARS
THICKNESSES
ELEMENT
• CONCENTRATED
THICKNESSES
MASSES
CONSTRAINTS • STRESS-STRAIN • DISPLACEMENT • MODAL FREQUENCY •
AEROELASTIC EFFECTS - LIFT EFFECTIVENESS -
•
22.8
AILERON EFFECTIVENESS DIVERGENCE SPEED
FLUTTER
VALUES
MULTIPLE MULTIPLE
DESIGN PARAMETERS DESIGN
ELEMENTS
RESPONSE
User Input Data Stream ASTROS
MAPOL
Solution
Solution
Control
Bulk
NASTRAN
DMAP
Algorithm
Commands
Case
Data Entries
•
Control
Bulk
ASTROS Theoretical
Sequence
Data
Commands
Entries
Documentation
Manual
Describes ASTROS Methods Emphasizes Innovative Features •
User's
Manual
Input and Output Description Techniques to Obtain Additional Output Creation and Modification of MAPOL Sequences •
Application
Manual
Documentation Resources Modeling Guidelines Sample Cases •
Programmer's
Manual
Code Installation Module Description Data Base Calls Utility Calls 229
Conclusions
•
Potential
for Realization
in Multidisciplinary •
Widespread
Interest
•
Need Extensive
•
Validation
of CAD
Environment Around
the World
Applications
to Assure
Reliability
i I i
• SHORTER SCHEDULES • EXTENSIVE PARAMETRIC STUDIES • OPTIMALDESIGNS - BESTPERFORMANCE • LIGHTWEIGHTSTRUCTURES • TECHNOLOGY TRANSFER • RESEARCH TOAPPLICATION • SYSTEMTO SYSTEM(LESSONS LEARNED)
230
N92-25922
Light Thermal Structures and Materials for High Speed Flight Earl A. Thornton Light
Thermal
Structures
University Charlottesville,
:of V|rg]n
Center ia
Virginia
231
INTRODUCTION
As a hypersonic interacting
with local
local pressures experience
intense,
not possible
a means
shocks
highly
years
because
viscoplastic
rate dependent
localized
flow, creep
edges
expose
of integrated
sweep
across
structural engine
surfaces
structures
have evolved
at elevated material
temperatures
to severe
which
behavior.
to meet this need.
from the elastic
and stress relaxation.
due to dynamic
loads
Over the last twenty
years,
These
constitutive
through
the plastic
Rate-dependent
plasticity
range
models
including
effects
was
provide strain-
are known
is to describe
Structures
Center
computational
focused
on the investigation
In the first part of the paper,
In the second
part of the paper,
and experimental
research
of the response
finite element
the thermal-structures
programs of structures
thermoviscoplastic
to be
experimental
INTERACTION
FOR
AEROSPACE
PLANE
_
COWL
program
Vehicle
LEADING
underway and
analysis
I
FLOW
it
loads.
inelastic
response
interactions
shocks
temperatures.
to local heating.
highlighted.
These
response
to model
models
in the atmosphere,
is leading
aerothermal
a material's
of this paper
at the Light Thermal
layers.
One example
constitutive
at elevated
The purpose
materials
and boundary
of an inability
plastic
at high speeds
the study of structural
for representing
important
accelerates
and heat fluxes.
Until recent
unified
vehicle
is
is outlined. IIII
EDGE
nose
b°whock°?sCh \ , -\ __
inlet -_ Cowl _ uw,-
_ _
_\
"_'_
_ _ " _
_t
(_ _.
_'__...
X Cowl bow / shock -/ Figure
232
_(_
1
Internal
\_oo,ing
__"_
_//
RESEARCH
Finite element years
analysis
and provides
an important
thermoviscoplastic convectively
method
presented
structures
to localized
of structures
problems
at elevated quasistatic
Reference
method
illustrate
has been under 1 describes
for hypersonic
the effectiveness temperatures.
finite element
development past efforts
structures.
15
and presents
Applications
of the approach
References
for about
a
to
and provide
insight
2-3 use the computational
thermoviscoplastic
analysis
of aerospace
heating. studies
for intense
thermoviscoplastic
therrnoviscoplastic
O
behavior
is a need for further
to represent
design
structures
models
capability.
computational
in ref. 1 to perform
subjected
There response
hypersonic
viscoplastic
constitutive
simulation
finite element
cooled
into the transient
with unified
OBJECTIVES
computational such as leading
to:
(1) understand
local heating, behavior,
edges
(2) investigate
(3) perform
methods,
the phenomena
finite element
thermal-structural
and (4) understand
of hypersonic
that occur
in the viscoplastic
modeling
experiments
the high-temperature
techniques
required
to validate behavior
of difficult
vehicles.
Investigate Thermoviscoplastic (TVP) response of thin panels subject to intense local heating. Evaluate finite element Thermal-Structural analyses with unified TVP constitutive models by comparison with experimental data.
..,H
ii
......
i iiiiiiii!iiiiiiiiiii::i_i::iiiiii_i_ '....
.....
Figure
2
233
THERMOVISCOPLASTICSTRUCTURALANALYSIS
The behaviorof a thermoviscoplastic structuresubjectedto intenseheatingis analyzedassumingthat: (1) thermo-mechanical couplingin theconservationof energyequationcanbeneglected,(2) thestructural responseis quasistatic,and(3) deformationsareinfinitesimal. With theseassumptions,anunsteady thermalanalysismay beperformedfirst to determinethetemperatures.Then,usingthesetemperatures, thestructure'sviscoplasticresponseis determined.The structuralresponseis obtainedby solvingthe equilibriumequationswritten in rateform. The solutionis obtainedby solvingtheinitial value-boundary valueproblemby time marching(ref. 1).
[KII_}
= {Fp} +/FT]
+ {i_} + {Fs}
where: r [K]
=
_Jf2e [ B ]T [ E(T)
{FP}
=
f_e [ B
]T
{iTT}
=
f_2e
] T [ E(T)
[ B
Stiffness
] [B] dC2
Plastic
[ E(T) ] {_P}d_
A'i _ dE2
] {a(T)}
Matrix
Strain
Temperature
m
{[_c}
=
f
{FB}
=
f_e
Surface
oa_e [NI T {_} ds •
[NIT
113} dE2
Body
Figure
234
3
Tractions
Forces
CONSTITUTIVE
The Bodner-Partom phenomenological model
constitutive
observations
has gone through
The current
model
The strategy temperature
integration
employed
rates.
Then
is of the internal
modifications
and was extended
thermal
in the viscoplastic variables
the constitutive
time, the equilibrium
equations
are solved
rates, then advancing
the constitutive
value boundary-value
problem
concepts
related
to dislocation
for anisotropic
is as follows:
the equilibrium
equations
again.
with
equations
are integrated
of the stresses,
equations
type that is based
on dynamics.
work hardening
The
materials.
effects.
algorithm
specified,
ALGORITHM
state variable
by physical
values
At time
AND VISCOPLASTIC
and supported
With updated
1.
points.
several
model
(ref. 4) also includes
and internal
displacement
MODEL
This sequence
are solved
forward
temperatures
the initial distribution
the nodal
in time at the element
and internal
of determining
in time is continued
to obtain
variables the nodal
until the desired
of stress,
Gauss
at the new displacement
history
of the initial-
has been obtained.
= t, initialize
(3ij
and
Z
P
1
2.
Calculate
3.
Assemble
4.
Calculate
1_} = [B]{fl}
5.
Calculate
{6} = [El{ _-
6.
Calculate
Z = ml(
7.
Integrate
(3"ij and
8.
If
_ij
-- (Sij
and
( t + At )
I
component Quantifiable reliability
I --
"
Figure 10 SIMPLIFIED
PROBABILISTIC
ANALYSIS
Fatigue Strength
Stress Probability of Occurrence
9a,, ,,_
Factor of Safety on Endurance /
/
/
Factor of Safety
_end F'S(end)
=
Ftu m- nOp.nd Sm- nOalt
n = 2 - 95% Reliability in = 3 - 99.7% Reliabilityg
Fa,,o;o-/ Region
Deslcln Herdwpre to Demaqe over
to be Insensitive Time
i • •
Thicker Walls (lower o) Low Stress Concentrations Smooth surface finishes
•
Alternating
Stresses
below
O'al t (Ksi) (_end on Endurance
Cycle to Failure
Figure 11 460
Operatlng_'_ Stress
[
DAMAGE
TOLERANT
DESIGN
APPROACH
Fracture Mechanics
Cht_r_¢terlze Material Flaw Sizes • Establish a Flaw Size/Distribution
_a
Probability of Occurrence
I •
3,_
,
-J
• • • •
Design Flaw Size
!
Database
Collect historical databases - typical castings Fully characterize material properties Conduct 1st article - section castings Section production hardware - random selection
-4.
Flaw Size, a -4 10
De$1an Hardware Flaw Tolerant • • • •
to be Dama0e/ d_dn
Thicker walls (lower stress) Minimize stress concentrations Established maximum flaw size Stress intensity below AKth
:/
-5
f
10
-6
:
10
AKIh
a t0
I 100
AK
Figure
DETAILED
12
PROBABILISTIC Most Sophisticated
Engine
Component
Component
J Component
Loads
ANALYSIS
Approach
Reliabilities
Responses
• Can be assessed
• Develop model of physical process
using closed form or FEM solution
• Anchor model to prior engine data
• Define analysis approach
• Define primitive engine variables
• Characterize variables
• Calculate
• Calculate response
component loads
input
Component
Failure
Rate
• Define failure mode algorithms • Define material property characteristics • Characterize other input variables • Conduct analyses
component
Enalne Reliability •
Sum of component failure rates • Probabilistic analysis component reliabilities, and • Deterministic component reliabilities for remainder of components
Figure 13 461
COMPONENT LOADS ANALYSIS Vibralion Measurements
]_,_,,A,A I _ rV W"1 time
# /
/ Pressure Measuremenls
Influence Coefficients • Relateengineenvironment Io component environment • Relalecomponenlenvironment Iocomponent loads
Temperalure Measuremenls lime
Turbine BladeLoading
'
I Com_nent Load# • V'bratory
P _lory
• Pressure • Thermal
P
Terr_ralure Pressuret,
|_
• TotalLoa_ng
P
Figure 14 COMPONENT
RESPONSE
ANALYSIS NessusTurbineBlade Course Model
TurbineBladeLoading
P I _,_lory
Input Variables i p
pI I
P e, ure, /'_
Geometric
P _._ances P
Structural / Response
Occurrence ProbabililYol
I p
Figure
462
OperalingStress
15
COMPONENT Generic
FAILURE
RATE
Model Process
Failure Model Input Variables ¢.,I
Failure Model
I i
• Dimensions/Geometry • Environment/Loads
_.
• Stress Concentrations • Model Accuracy
Finite Element
J
"""
• • • •
Predicted Failure Risk
Strength Fatigue (HCF, LCF) Crack Growth Displacement
t
Model
s
Material Characterization Model
• Stress Response • Displacement Response
• Material
Capability /_
Operating
\
'/
'
Failure Region Computed
Data
: Design Curve Statistical Variance • Surface Finish • Environment
Figure 16
BENEFITS
Quantitative
OF PROBABILISTIC
measure
of integrity
Uncertainty of each variable the analysm process
APPROACH
of design and reliability
is explicitly
considered
Most significant design variables (drivers) the order of their effect on reliability Design trade studies
can be assessed
in
are ranked in
via reliability
Gaps in the design data are surfaced - program resources can be effectively used to obtain necessary data
Figure 17
463
SPACE
TRANSPORTATION
MAIN
ENGINE
_Operational Parameters • • • • • • • • •
Thrust, klb (vac) Specific impulse, s (vac) Chamber pressure, psia Engine mixture ratio, o/f Area ratio, E Mission life Weight, Ib Gimbal capability, deg Throttling
583 430 2,250 6.0 45:1 10 7,900 _+10 70%
Design Features • Gas generator cycle • Liquid hydrogen/liquid oxygen • Milled channel chamber/tubular gas cooled nozzle • 3 stage H2T/P-1 stage O2T/P • Tank headstart • Open loop control • No bleeds • 152 in. long x 87 in. dia
Figure 18
MAIN
COMBUSTION
CHAMBER
Figure 19
464 _=
..... J_
COMPONENT
RELIABILITY
STME Initial Reliability
ALLOCATION Allocation
Requirement 0.99 at 90% Confidence
Y
0.99982
Reliability Design Goal Rsy s = 0.999
Combustion Devices Components
Design to Allocation
Injector Combustion Chamber Nozzle Gas Generator Total
0.99997 0.99991 0.99986 0.99993 0.99967
Controls 0.99997
Figure 20 STME
GAS
GENERATOR
Initial Reliability
Engine
ENGINE
Allocation
Level 0999
Turbomschtnery
Controls
i
Mo_llorln
9
Fuel
0¸99997
PCA IG
& SLt! Co,nt
CC CC
Con
M.dl
Ignile_
A
I_ter
8
Fuel
TurboPump
990998
0
999090
0
999990
Inducer
0 999g94
Pump
0
999999
Btg
0 099097
[nduce_
Pump
Assy
.k
0
gggggg
Til
_gniter
B
0
999999
Pep
0
099998
MQin
0
99991)2
1st
0
999994
2nd
0
999999
Inlerstg
Mcl
HG
(15)
Temp
C_o
"rm_
Flow VI_
O 999990
9oli Inlet
Wear
099990
TurT_n4
999996
Tufb
999999
2rid
0
999991
MelLn
(2)
0
999990
Rtg
Ring
Module
0
999999
Tud)
Whl
0
9999g9
Torqmlr
& Cond
Hams/Connect
Mdl
0
ts!
999998
2rid
5h=llt
Ass
0 999739
Pneum
Cnkl
0
999969
PC),
Inie¢lo[
0
099999
Elect
90g993
Ignlt_
AHF
0
909902
0
99999g
Spark
P_ugs
0
99999ii
0
999990
0
099990
9
Spdng
Chamt:_
In|e¢lor
0¸9999?0
Combustn
C, hrr_
Camp_|s
Controt
Ylv/Ach:
Valves
0999910
Main Main
Liner
0
0¸999990
Manifold
0 ggggTo
Ht
0¸009993
Body
0
99999O
tgnlf
Seal
0
Seel9
0¸999980
0
999860
GG
0
999930
(,I)
Ring
Stg
Brgs
999994
0 990975 0
Mild Noz]_te
SI9
0¸999997
999903
Housing
TUrt_ne "ru_blne
0¸990994
Tmb
0
Tu,b
999594
.41ty
Manilo_(I Whl
0¸999004
tst
Seal
0¸990958
2nd
Assy
0 999932
B_ds _lds
(2) (2)
0
099908
0
99997!
0
999994
Stg
999943
0
099060
0¸999900
/_ay
SIg
0
Gene_'_lo¢
Injector Combusloq
Body Bodl,
O 999975
Ht
B_gs
)4)
Ox;_zr
Contd
0
vlv
lntrcnc!
9g99g0
Heal
O=ld;zr
9909,1f
Contd
V_
E:changer
Bea;Ing
0
Attaching
900099
0
999999
0 909990
0 099997
O=_zr
Ocis/t.ns
0¸999940
0
0 990997
Ex(:tmngr
Glmbal
900g50
999906
0
0 999967 0 099997
5cisaorl
ggggg0
999999
0
0 9ggggg
Oxi_zr
GG
0
ggg9g4
0
0 999999
Exc_
Igntr
Bids
[4)
Fuel
Moln
0999955
Bids
0
0 999974
Actuators Mean Fuel
Inle!
Beadnge
H_drosl¢
GaS
Assy
9999S0
0¸999993
Disk B;all
Noztte
^ssy
Body
Impeller
Boadngs
Hydros(c
Malo
Thmt
Main
0¸999914
Comp 0 99g02
Volule
Oisk Gall
Br 9
0
999537
Syteme
0¸90999!
Assy
9
0¸999991 090097
0
Assy
GO
0¸990901
I.Is 0
In_t SIg
Tuf_epump
Pump
9
O gggG7
In_r
R_
0
Oxidizer
TCA
TIP 09984
imf_r
StO
0
(4)
Hsg
Pep stg
0
(t0)
(0|
De,ect
FTP/GG
(_)
SenSor Pos
Accel
TCA
(2)
0 909930
Sprmg
Igniler
Plesm
0.099722
0
0
GG
Senso¢s
el
0¸99972
(3(3
OTPhJEX
099959
TIP
Ac_u_
0
099997
0
99g980
0
999997
0
999997
0
999900
0
9gggg2
9
ggggg8
0
99999,_
Figure 21
465
MAIN COMBUSTION FUEL OUTLET MANIFOLD
TO COMPONENT FUEL DRAIN MANIFOLD
_ ',,\'_|
CHAMBER
! __ OVER.OARD ORA,. UNE tT/f_'/.J b_lln
BURST
PROGRAM
,_#_E_T
PLUG-AFT
CAVITY
[ "%/-_1_._-'_
c.osEouT_,._,.r/#C. , SLOTTED NARLOY (INNER WALL)
Z LINER
STRUCTURAL
__
_AC_. LINER/JACKET
-'_\I
CAVITY
\\
\
\\
""_,b__
_\
¢...,,...,_/_ _ z. _ __
_
\.
y
\\
(-/--7--7--T_-J_
COOLANT
.
\)\|
.
,,,C_EL
CLOSEOUT
coPpER
_ e__z ._ ...L_.:,,_
SECTION
BARRIER
A-A
Figure 22 PROBABILISTIC
Primitive
Varlabte_
• Characteristic exhaust velocity --.il,,. • Inlet nozzle area of the LOX turbine • Fuel pump efficiency
ANALYSIS
OF THE MCC LINER
Engine Model • Numerical solution • Influence coefficients • Statistics of primitive variables
Variables
that affect the MCC
• Inlet coolant temperature • Chamber pressure • Coolant flow rate
Loca! Variables • Curvature enhancement •
MCC Liner Thermal Model • Closed form solution
Hot spots
• Thermal conductivity • Hot gas wall thickness • Channel
•
Heat load
• Scaling • FE/FD models
depth and width Correlated Pressures
Liner Structural
Local Variables • Hot-gas wall thickness • Channel width • Land width • Material
properties
--Ii,,-
• Channel bending stress • Land tensile stress • Low-cycle fatigue • Ratcheting
Figure 23 466
Model
Temperatures & on the HOt Ga_ Wall
PROBABILISTIC Channel .0016
,0014
-
I
STATISrf¢
1,000,000
Stress '
''
'
..............
Mean ................... Slondard Deviation Coef of Varialion Minimum Maximum
,0012
z
Bending
I
--
ANALYSIS
I
STREH,;rH
3950.46 367.30 0 09
16032 65 2112 5t 0 13
2579.76 6255 08
4456.73 22628 16
..... .....
................ ................
I
SrRESS --
Simulations
0010 Probabi|ily < .000 00¥
._
at
Reaching
Y eld
Moment
.0008
._% .0006 o_
:::::JIEND:STR ,5o0.
t2500, STRESSI
17.500 PSI
Figure 24 PROBABILISTIC Land ,0025
.0020
-
:'UOI5
-_
0010
-
• 0005
-
Tensile
Stress
I I -STATISTIC .............. Mean ' Slandard Deviation ..... Coef. of Variation .....
I STRESS 5747.47 253.65 0,04
Minimum Maximum
4759.17 7011.44
1,000,000
I¢
ANALYSIS
................ ................
--
I STRENGTH 18057.10 1279.84 0.07 7287.27 23354.15
Simulations
Probability
_f
Reach+i_
_ield
Stress:
_%
I _OQDi 4500.
.... 8500.
1-2500.
' 16500
20500:
STRESS. PSI
Figure 25
467
PROBABILISTIC
ANALYSIS
Low Cycle Fatigue
OOO5
Mean ................... S|ond_rd Oeviotion ..... Coe[. o( V_¢ialion ..... Minimum ................ Maximum ................ Number of Simulalions ..
3098.31 1309.05 0,42 476.t7 27216.11 I000000
ooo z w
000.
0O02
O0Ol
oooo 40o
5400.
10400. 15400. 20400. NUMSE_ OF CYCLES TO FAILURE
25400.
Figure 26 PROBABILISTIC
ANALYSIS
Ratcheting
0.14
LINER
Failure Rate at a Hot Spot
Mean life Standard deviation Coefficient of variation Minimum life in simulation Maximum lile in simulation Number of simulations
0.16
OF MCC
9.69 4.62 0.48 0.59 59.96 1,000,000
0.12 _" 0.10
A I| . |rJ 6 I i_ iIv "ll_l|l] _IV r_l/lll iI " 11 if J _,[tJ
--
o
0.08 A- 0.06
Random V_rlables Channel width Thermal conductivity Wall thickness Characteristic efficiency, C* Land width Channel depth Curvature enhancement
Sensitivity 0.977 0.176 0.102 0.056 0.018 0.017 0.002 0.002
0.O4 0.02
/"
"'
I 0
10
_1_
_
Fuel pump efficiency
r-'-------J 20 Number
3O
I 5O
of Cycles to Failure
Figure 27 468
I 4O
6O
PROBABILITY DENSITY HIGH CYCLE Main Combustion
FUNCTIONS FATIGUE
Chamber
Failure FirsI-Order
Mode:
AFT Manifold
High-Cycle
Rellablllly
FOR
Casting
Fatigue
Method
(Level
III)
0006 Pressure ............. Thickness ............ Stress Cone .......... Ro_dom Vibration Force Ullim0te Strength .... Mo4elin 9 Error ....... [nduronce Limit ......
.0005
-0.39 0,29 -0,64 -0,25 0.iT -0.32 0.40
.0004 z EHDURAHCE .0003
=
[OUIVALENT
AL(ERKAT]HG
LIMIT
[HORIAAL
STRESS
.0002
0001
0000 0
5000.
10000
15000.
20000
STRESS,
Figure
PRELIMINARY RELIABILITY USING PROBABILISTIC Combustion
Chamber
25000,
30000
,]5000
PSI
28
ASSESSMENT ANALYSIS Aft Manifold
Manifold DETERMINISTIC
SAFETY
FACTORS (OeJlgn Reqm'l)
Primary High Shell
Stress:
Cycle
Fatigue:
Fracture Mechanics: (&K threshold/AK dynamic)
Probabilistic
3.9
(1.5)
2.2
(1,25)
1.0
(1,0)
Analysis
Random
Variables
Manifold
Pressure
Mean
STD Dev
Reliability Shell
Thickness
Sldeload
psi
360 psi
0.400
in.
.007
184,000lbs
Vibration Stress
3,600
Load
Concert,
Ultimate Endurance
87,000 Factor
27,600
Ibs
4,350
1.7
Burst HCF Crack Initiation
In. Ibs
R > 0.9to0 R = 0.998
Ibs
Chamber
0.10
Strength
96,700
psi
4,833
Umlt
27,800
psi
1,390 psi
Reliability
Allocation:
psi
R -- 0.99992
Figure
29
469
COMBUSTION CHAMBER SAFETY AND RELIABILITY SUMMARY Aft Msnlfold Inlet
t.
Jackel Forwerd
t = 0.75
itl
in,
J
FS. y = 1.7 F.S. en¢l >4.0
] J
F.S. FS
u y
= 39 =2_
AN Flange
r--
6Kih/_K
I
_l
F.S
end
- 2.2
(Nozzle}
I_ _
_ t 1
6Klh/AK
= 3.5
_
I
Fbnge
'_-I
- 0.94
Rbult _
>
RHC F
>.*0
gt00
Weld
,.2Sin.
FS.
u
• 4.0
FS_ F.S
_d
>4.0 = 2.4
I
-20 -15
F.S. _d
.2_6
Att CIoRI
= 1.2
M:
l llll_,_
I
lilll .... IIIII__ ":. "
I I l---J
I i
I
I
I
Combustion
Figure
Allocation
HCF
LCF
_/
Impeller
_/
q
q ",,/
Devices
Injector LOX post Injector close-out welds
_
_/ ",,/
Interpropellant
_
q
q
_/
V
_/
inlet
manifold
Fuel
inlet splitter
Manifold to Liner
_/
",,/ _/
q, -y
close-out
MCC liner
= 0.99992
Fract. Mech. Wear
_/
Fuel
II
APPLICATIONS*
_/ _/
plate
• 9100 • .9100
30
Turbine blades Turbine disk Hydrostatic bearings Inducer blades Combustion
Rb_l R HCF
.,,ooII
Turbomachinery
vanes
-39 >4.0
II
""_ ""°°
Reliability
FUTURE
>40
FS.y F.S end
F.S.o,_ ,4.o II ,,KmI_,K ._.4 II
I
Chamber
Hardware/Process
F.S,u
"_"'--I-.'/_"
llbl_"..... I
V
* Not complete list Figure31 470
,=
kl.
FS, FSy
&K_/t,K
Slats
0.40
FSu ._9
Splitter Jacket
Shell
Shell
Buckling/ Displ. Instabilit_
FUTURE
Hardware/Process
HCF
Systems Hardware Interactions Fluid Coupling in LOX system
Fract. Mech.
Wear
Displ.
Buckling/ Instability
_/
Process Control Casting flaw size Material surface finish
"J
joints
* Not complete
LCF
q
Stability issues
Welds/braze
APPLICATIONS*
_/
_
q
_/
-_
list
Figure 32
RECOMMENDED
TECHNOLOGY
• Simplified probabilistic analysis tools (rapid analysis) • Required early in design process (sizing, trend analysis) • Closed-form solutions for estimating variable distributions • Simplified probabilistic analyses or estimates from FS solutions • Simplified reference guide for conducting PDA • Many engineers are not familiar with process • Elimination result misinterpretation (due to lack of knowledge) • General purpose probabilistic tool for empirically derived equations • Nonlinear solutions - define by empirical relationships • Process control problems - fabrication issues • Meaningful graphics display of results • Graphic overlays - solution comparisons as a result of trade studies • Carpet plots showing reliability versus selected variables • Risk
management
tools
for
evaluation
of technical
risk
Figure 33
471
N92- 59zz
Computational Advanced
Structural Analysis and Commercial Engines R. B. Wilson Pratt & Whitney
PRECEDING
PAC_
BLANK
NOT
FILMED
473
ENGINE
ENVIRONMENT
- PRESENT
•
Materials - Monolithic TI, NI alloys, limited composites (not primarily structure)
•
Engine configurations
•
Limiting Issues - Compressor - Turbine
primarily turbofans
exit temperature
inlet temperature
(1150 to 1300 ° F)
(2000 to 3000 ° F)
The history of computational structural analysis at Pratt & Whitney will be reviewed and anticipated requirements for the design, development and support of advanced commercial engines will be discussed throughout the following paper. The present commercial engine environment titanium and nickel alloys.
474
is comprised primarily
of turbofan engines containing
monolithic
ENGINE
ENVIRONMENT
- FUTURE
• Materials - Monolithics
with modestly higher capability
- Composite
primary structure
- Woven and braided composites - Metal matrix and ceramic matrix composites temperature areas •
in high
Engine Configurations - Derivatives
of existing turbofans
"- Ducted or unducted propfans (efficiency noise requirements)
and 1997
- HSCT - "Super" Turbojet •
Key Issues - Engine/airframe
integration
- Gear systems - Fan integrity - Reliance on load sharing
Future commercial engines will make extensive use of composite materials to meet demanding high temperature requirements and aggressive weight goals, since only modestly higher capabilities can be anticipated in monolithic materials. Engine configurations will include ducted and/or unducted propfans and "super" turbojets (HSCT) in addition to turbofans.
475
BUSINESS
ENVIRONMENT
Intense competition - Shorter design cycle - Reduced design/development Increased
costs
emphasis on:
- Safety - Reliability - Durability - Cost - Efficiency - Environmental
impact - emissions
and noise
Competitive pressures will require a shorter design/development cycle and continuing reduction in costs. same time a variety of economic, durability and environmental issues will assume increasing importance.
476
At the
DESIGN/ANALYSIS
ENVIRONMENT
•
Fewer S/Task
•
Fewer, less experienced
•
Less time
•
Vast increase in computing
•
Increasingly
people
resources
complex analysis
- Three-dimensional - Increasing
models required
nonlinear requirements
- Detailed analysis of assemblies - Off-design
analysis to assure robust designs
- Closer fluids/structures
coupling
•
Sharing of analytical work among joint venture partners
•
Increasing
reliance on vendors for:
- Software - Design - Analysis - Testing
Structural design and analysis will need to be accomplished more quickly, with fewer personnel resources. Increasingly complex analysis will be required. The process will benefit from vastly increased computing resources.
477
ROLE OF ANALYSIS ENGINE
LIFE
IN
CYCLE
A brief historical overview of the role of structural analysis during the commercial engine life cycle is given in the next few figures.
478
PRE-HISTORY
Mechanical - Handbook
(1968?)
Design calculations
- One-dimensional
- sometimes
analysis
- Modest use of two-dimensional
•
Engine development/support through testing
"THE ONLY ENGINE!"
computerized
SIMULATION
analysis
addressed
primarily
FOR AN ENGINE
IS AN
Twenty to twenty-five years ago design issues were addressed primarily through handbook calculations, modest use of two-dimensional analysis. Major structural issues arising during engine development support were addressed primarily through testing, frequently of full engines.
with very and field
479
PRESENT
Mechanical
Design
- Supported by extensive 2D and 2.5D analysis, and selective 3D analysis, during design cycle - Additional
3D analysis for design verification
Significant
investment
Engine Development/Field
of highly trained personnel
Support
- Analysis drives design changes - Testing to provide analysis input and verify changes
"IT'S
GOING
TO TAKE
HOW
LONG?"
Presently, mechanical design is supported by extensive analysis, primarily two-dimensional. Development and field support are usually driven by analysis, with testing to provide analysis input and verify changes.
480
FUTURE
•
Mechanical
(1993-2005)
Design
- Use of "artificial intelligence" will require hands-off automation of much 2D and 3D analysis - Vastly increased
3D requirements
- Routine use of nonlinear capabilities - Requirement for concurrent organizations
(2D and 3D)
work by diverse
- Increased problem sizes - driven by automation closer coupling in engines
and
- Vast increase in analysis throughput • Design automation • Concurrent engineering • Evaluation of off-design • Engine Development/Field
conditions Support
Mostly 3D analysis needed Frequently nonlinear - Rapid turnaround necessary
"YOU WANT ME TO BUY HOW COMPUTERS?"
MANY
Future designs will require much more complex analysis types to be used on a routine basis. cycles will require increasing automation of portions of the design and analysis process.
Shortened design
481
ADVANCES
1. Analysis
- What problems can be solved?
2. Computing
3. Analysis
- What resources
482
are available?
Process - What does the user have to do?
4. Data/Information
Advances in computational requirements.
REQUIRED
structural
Handling.
analysis will be required in several
areas in order to support
future
TRANSIENT/DYNAMIC
ANALYSIS
•
Fan and containment design more exotic - design concepts and materials
•
Durability standards
•
Pressure on weight and manufacturing
•
Testing very expensive,
•
Analysis must be:
more demanding
with long lead times
- Timely Reliable - Usable by non-specialist •
Improvements
needed in:
- Element technology - Time integration - Material modeling •
HSCT will introduce new problems - i.e., thermal shock
Transient and dynamic analysis will become more important due to economic pressures, the use of exotic materials and the requirements of new engine types.
483
COUPLED
ANALYSES
•
Typically CFD, thermal and stress analysis done separately
•
Inefficient,
•
When coupled analysis is done, level of individual analyses considerably simplified
error prone process
Advanced engines will require some coupled analyses in full detail - for example, thrust reverse transition for a propfan will require full Navier-Stokes sophisticated structural analysis.
Many problems in future engines will require coupling of structural analysis with thermal and/or computational fluid dynamics analysis.
484
MATERIAL
MODELS
Further refinement of state variable models for monolithic materials - cover entire range of thermal/mechanical loading
•
Increased complexity of"traditional composites"woven and braided fiber configurations
°
Metal and ceramic matrix composites - material model must recognize environmental effects
•
Stress analysis and damage accumulation coupled problem
become a
Material modeling requirements, and the interfacing of these models with stress analysis programs will become vastly more complex. High temperature requirements will require further development of state variable models and computationally effective material models will be required for a variety of traditional and exotic composite materials.
485
OTHER
AREAS
•
Contact analysis
•
Improved triangular
•
Quality assurance
•
Rezoning/remeshing/substructuring
•
Hybrid (FEM/BEM)
and tetrahedral
elements
for analysis results
methods
Sensitivity analysis Coupled
stress/fracture
mechanics
Advances in many other areas will also be required, especially to support timely, complex analysis by relatively inexperienced personnel.
486
COMPUTING
Improvements in available analysis demands.
CAPABILITY
computing
capability
•
All predictions are too conservative.
•
All available capability will be utilized.
•
Networked workstations of increasing pre-/post-processing and analysis Compute
Massively
power for
supercomputers parallel systems
- Use of workstation
Computer
increased
Servers
"Traditional"
•
will support
networks as a parallel system
development
is ahead of analytical software
Computational structural analysis will depend on networks including supercomputers functioning as compute servers. In general, structural hardware capabilities.
capability from workstations through analysis software lags behind current
487
ANALYSIS
PROCESS
The analysis process is driven by design/development
requirements.
Only three time scales matter Impact the design process - Verify design while low cost changes possible
•
- Respond quickly to development/field
problems
Processes must be usable by "ordinary" technicians
engineers
and
Analytical tools must be usable by line technicians and engineers. Analysis results must be available in a timely manner in order to have impact on design and development.
488
REQUIREMENTS
•
Close coupling of stress pre-processors
with CAD tools
•
(Close to) automated
•
Order of magnitude improvement boundary condition definition
•
Simple geometry changes should be simple to evaluate!
meshing
in load and
Major improvements are needed to allow the more complex analyses required for future engines to impact the design process.
489
CONCLUSIONS
•
Vastly more analysis will be required.
•
More complex analysis will be required.
Analytical
development
required
- Material models - Dynamics Nonlinear
•
Analysis
problems
process improvements
equally important.
Real challenges exist for all communities involved in the development government, software developers and industrial users.
490
and use of structural
analysis tools -
Na2- 5934
Military Engine Computational Structures Technology Daniel E. Thomson WL/POTC Wright
Patterson
Air Force
Base,
Ohio
491
Agenda
IHPTET
overview
Codes we use now Codes we are developing The Future Summary
Integrated High Performance Turbine Engine Technology Initiative (IHPTET)
Goal: Double turbine engine propulsion capability by the year 2003
50% of goal will come from advanced materials and structures, the other 50% will come from increasing performance
492
IHPTET's _Effect on Computational Advanced
Structures Materials
Ceramics Intermetallics Single crystal castings Composites Organic
Matrix Composites
Metal Matrix Composites Ceramic
(MMC)
Matrix Composites
Higher Temperature
IHPTET's
(OMC)
(CMC)
Conditions
Effect on
Computational (continued)
Structures
Advanced Tactical Requirements 4000 hours/8000 section 2000 hours/4000 section Advanced
Structural
Fighter
Service
Life
TAC cycles
- cold
TAC cycles
- hot
Designs
Integrated design
structures/materials
Integrated
structures/aero
design 493
i_ Codes We Use Now MSC NASTRAN NOSAP-M X3D UDHeat CRACKS
90
PATRAN
NOSAP-M
Soft body (birds, ice) impact analysis Developed by General Air Force Contract Scaled-down
version
Electric
under
of ADINA
20 node brick element Uses water jet to model impact Implicit time integration Orthotropic
494
material
capability
X3D
Nonlinear
dynamics
and impact
analysis
Developed by University of Dayton under Flight Dynamics Directorate Explicit Model
Research contract
Institute
time integration both the structure
Includes
and the impactor
solid and shell elements
Fiber failure Currently
and delamination
performing
analysis
code verification
Z F-16 BIRD DISPLACEMENTS
IMPACT FOR
05-JUN-91 F16BE
AT F.S, TIME
112, STEP
=
SYMMETRIC, 173I
/
350 TIME
KNOTS 6.0000001E-03
04:27:38 -
0.5
INCH
-
POLYCARBONATE:
SY=7140
!
SU=16000
495
UDHeat
Explicit in time) transfer
code (finite element in space, for steady state and transient
Boundary
finite difference conductive heat
Conditions
Convective Surface
or volume
Heat sources
heat flux
and sinks
(applied
to nodes)
2D elements Infinite
in one direction
Axisymmetric 3D elements Orthotropic
CRACKS
material
capability
90
Preprocessing Material
database
Spectrum
profile graphs
Stress intensity and plotting Generation Residual
496
and plotting
factor generation
of analysis
input file
stress table generation
CRACKS
90
Analysis Residual
strength
Crack growth Residual
life
stress
Load interaction Crack growth
rates
Comprehensive
tabular
Brief summary
of results
output
Postprocessing Single or multi-curve plots
growth
Single or multi-curve
life plots
Residual
strength
rate
plots
497
P" Codes We Are Developing Blade Life Analysis (BLADE) Engine Structural (ENSAC) CRACKS
and Design Evaluation Analysis
Consultant
xx
Probabilistics
Background
(BLADE-ST)
Stress Technology Inc (STI) has worked under the sponsorship of the Electric Power Research Institute (EPRI) since 1980 Developed menu
BLADE-ST driven
contains blade preprocessing generic blade and root types contains
materials
library
of over 40
data base
computes steady stresses, natural frequencies, mode a life analysis
dynamic stresses, shapes, and performs
Recent Accomplishments
Integrally
bladed rotor analysis
F110 high pressure analysis 498
turbine
blade
Blade Life Analysis ahead Design Evaluation for _ Gas Turbine Engines (BLADE-GT) OBJECTIVE: "start-to-finish"
To develop a user-friendly, finite element solver
FEATURES: User friendly
geometry
input
Automated
mesh generation
Automated
boundary
Static Forced
and dynamic vibration
Heat transfer
condition finite
generation
element
solver
solver
and thermal
stress
solver
Life analysis
Funding
for BLADE-GT
Small
Business
Jointly
funded
Innovative
(SBIR)
from four AF organizations
Aero Propulsion Engine
Research
System
and Power Program
Directorate
Office
(SPO)
B2 SPO Arnold
Engineering
Development
Center
499
Engine Structural _ Analysis Consultant (ENSAC)
i_
Application Characterization GOAL: Build an expert system that advises engineers in the use of general purpose structural analysis codes USERS: Engineers with some background in structural analysis, limited practical experience
500
but
Application
Overview
Preparation of input data files for general purpose finite element structural analysis programs Find applicable Complex hardware
and reliable
combination capabilities,
material
properties
of software capabilities, and CPU costs
Analysis model must simulate correct physical behavior, preserve desired accuracy, and minimize CPU costs A similar capability currently exists in an expert YAstem called Structural Analysis Consultant CON)
Status
Selected ART-IM language
as programming
pull down menus dialog boxes windowing Preliminary
demo due I July 1991
prototype partial capability
501
CRACKS xx
OBJECTIVE: Extend the capabilitiy include crack growth in composite
of CRACKS90 systems
to
Completed: An influence function-based method was developed to calculate the stress intensity factor as a function of crack length for either an edge crack or a surface crack in bimaterial construction In Progress: Models of several additional damage accumulation modes are being developed Interface
crack
Delamination
growth
Crack-to-delamination
5O2
transition
Probabilistic Design System Development
Probabilistic
Methodology
Conventional, deterministic design uses single values for material properties, manufacturing tolerances, mission usage, and other parameters Safety factors variability Method
are applied
is inherently
to account
for
conservative
Probabilistic design makes full use of the known statistical distribution of all parameters, and combines them to produce a statistical prediction the result Conservatism level
is reduced
to known,
Design can be optimized for weight, life, or any selected criterion
of
acceptable performance,
503
Probabilistic Design System Development Probabilistic with General Started
Rotor Design Electric (GE)
September
System
(PRODS)
1990
Similar program was also started terminated due to lack of funding,
with P&W but was in February 1991
Both GE and P&W have very large probabilistic method development Anticipate one or more proposals design later this summer Other engine probabilistics
GE PRODS
Program
companies
IR&D efforts
in
on probabilistic
also show
interest
in
Program
OBJECTIVE: probabilistic
Develop, validate and apply a rotor design system methodology
APPROACH:
Six phase
program
Data acquisition Method
Development
Validation Application Application Method PAYOFF: materials; FUNDING:
504
Test
Extension Improved predicted
design safety;
$1.8 million,
capability in advanced and weight savings
over 4 years
Probabilistics
Summary
Probabilistic Improved
design
is real
computational
power
makes
it feasible
More knowledge of the variability of material properties, behavior, usage, and manufacturing effects is necessary Efficient essential
usage
But acceptance
of composite
of the methodology
must be demonstrated, must
materials
it
will take time
validated
be sold as a design
makes
and applied
tool
The Future
Improved
composite
AI Programming/Expert Composite
analysis
methods
Systems
Life Prediction
User Friendly
Systems
Animation/Simulation
of Dynamic
Phenomena
Crude animation of X3D results has revealed phenomena unseen in "snap-shots" Structural
optimization
for composite
components
5O5
Summary
IHPTET Effective
goals
require
analysis
a strong
of composite
analytical
base
materials
is critical
Life Analysis Structural Accurate critical
Optimization
life prediction
User friendly
systems
Post processing
506
for all material
systems
are desirable
of results
is very important
is
N92-25935
Computational Structures Technology Engine/Airframe Coupling Bruce
C. McClintick
General Electric Cincinnati,
Company Ohio
507
COMPUTATIONAL STRUCTURES TECHNOLOGY ENGINE / AIRFRAME COUPLING BC McCLINTICK
There experience,
are many
happening.
of engine
for the past 14 years,
3-dimensional engine static load conditions. utilization
aspects
structures However,
of the 3-dimensional Since
/ airframe
coupling
has been with the generation
to structures.
and analysis
My personal
of full
with regard to the design and operation of the engine under several problems arose about 6 years ago which required the models
then, full 3-dimensional
under
dynamic
engine
basis to help understand dynamics related problems. addresses, is engine related aircraft vibration.
508
related
loads
structural
to more models
fully understand have
One such problem,
been which
what
was
used on a regular this paper
Engine related aircraft vibration is noise within the fuselage of the aircraft which can be both felt and heard. This noise is generally caused by rotor imbalance and is transmitted from the engine
through
from the lower aircraft
system.
the structural
speed Many
only during certain unloaded, variations major
roles
vibration
rotor,
engine/airframe
PERCEIVED CAUSED •
encountered
must
Also, nearly
the primary matches
cause
frequencies
have also been non-linear
To effectively be treated
RELATED
NOISE
BY ROTOR
predict
engine
of noise
is
of the
and observed
related
becoming used play aircraft
as a system.
AIRCRAFT
VIBRATION
IS BOTH FELT AND HEARD IMBALANCE
PRIMARILY FAN ROTOR - LARGEST ROTOR - LOWEST SPEED ROTOR
CAN BE NON-LINEAR
IN NATURE
•
BEARING
TYPE
•
BEARING
STIFFNESS
•
MOUNT
ROTORS
STIFFNES_
=_ FRAMES
FUSELAGE •
more
Engine mounts becoming unloaded, bearings and bearing stiffness, and the types of bearings
of vibration.
ENGINE
of the aircraft.
as this rotor
of the problems
flight envelopes. in both mount
in the transmission
the entire
,
components
or fan rotor,
MUST
:=_ MOUNTS
:=_ CABIN/COCKPIT BE TREATED
=_ STRUT
=_ WING
=_
=_ PASSENGERS/CREW
AS A SYSTEM
509
Historically the engine has always been treated as a necessary evil to 'make the airplane go' and little emphasis was given to the engine other than making sure it stayed on the wing. The engine manufacturers sized the engine for thrust, designed for ultimate loads, minimized the affects of rotor modes and provided a reasonable balance at factory acceptance. The main point is not to over simplify the aircraft and engine design, which is indeed quite complex, but to point out that little was done in the design of the system to reduce or understand noise within the fuselage.
HISTORICAL #
#
ENGINE
/ STRUT
SYSTEM
•
ENGINE
A NECESSARY
•
FLU'ITER
STABILITY
•
ULTIMATE
LOADS
•
LIVE
RESULTS
WITH
EVIL
STRENGTH
ENGINE •
SIZED -
FOR LIKE
•
ULTIMATE
•
ENGINE
•
510
/ WING
THRUST ENTROPY LOADS
MODES
-
NOT
-
SIMPLE
-
RELIANCE
REASONABLE
- REQUIREMENT
ALWAYS
INCREASES
STRENGTH OUT
OF ROTOR
AT NORMAL DYNAMIC
OPERATING TEST
RANGE
SPEEDS
ANALYSES
ON ENGINE BALANCE
OPERATING UTILIZED DATA
AT FACTORY
ACCEPTANCE
As new
aircraft
were developed
it became
were less tolerant to noise and to maintain cockpit and cabin. Thus, tighter vibration on-wing some
trim balance
of the more
previously beginning
procedures
troublesome
used only for static of aircraft/engine
were
the norm.
vibration
problems,
analyses,
began
design
PASSENGERS/CREW
apparent
that both the passengers
for structural
Also,
as an aid to determining
3-dimensional
to be utilized borne
finite
for dynamics.
of
models,
This marked
the
80'S
LESS TOLERANT
TO PERCEIVED
SMOOTH TIGHTER
• •
ON-WING TRIM BALANCE PROCEDURES AFFER MARKET STRUCTURAL CHANGES
VIBRATION
AND QUIET REQUIRED TO SELL AIRPLANES PRODUCTION VIBRATION LIMITS
ELEMENT / STRUT
TO 'SYSTEM'
MODELS
•
ENGINE
• •
LARGE, COMPLEX MODELS (100,000 - 200,000 USED ONLY FOR SPECIAL PROBLEMS
•
SPECIAL
BEGINNING
the cause
element
noise.
• •
3-D FINITE
and crew
sales it was imperative to have a quiet, vibration free limits were imposed on engine manufacturers and
EARLY
•
more
PRIMARILY
ROUTINES
OF DESIGN
WRITTEN
FOR STATIC LOADS
FOR ROTOR
FOR STRUCTURAL
BORNE
DOF)
DYNAMICS NOISE
511
Large, complex engine structural models used for static models used for studying engine related aircraft vibration.
TYPICAL
512
ENGINE
STRUCTURAL
loads
MODEL
and deflections
were the first
FOR STATICS
New aircraft development in the later 1980's saw the integration of the engine and airframe as a system in attempts to predict the response of the aircraft cabin and cockpit to rotor imbalance. Aircraft / engine teams were formed to develop the necessary methodology to analyze, test and modify the system design to decrease structurally transmitted noise from the engine.
LATE *
GE UNDUCTED •
•
•
ENGINE
FAN - UDF _ RELATED
-
SIGNIFICANT
-
ENGINE
ENGINE
MOUNT
-
BLADE
-
AND
'ACTIVE'
80'S
AIRCRAF'I" DESIGN
/ AIRCRAFt
VIBRATION PARAMETER TREATED
AS
A SYSTEM
DESIGN OUTLOADS
- VIBRATION
VIBRATION
TRANSMISSION
REDUCTION
CONTROL
513
GE's Unducted Fan program vibration related data throughout analytically
monitored
GE UNDUCTED
514
for various
utilized a full airplane/engine structural model the airframe. Thus, cabin and cockpit vibration design
to generate was
changes.
FAN - UDF ® DUAL
ENGINE
DYNAMIC
ANALYSIS
Engine vibration related noise is now included in engine specifications. Thus, it is now necessary to include engine vibration transmission as a design parameter and develop the necessary methodology to predict noise within the aircraft before designs become 'fixed' and expensive to modify and change. Part of this methodology development is the correlation of the 3-dimensional finite element dynamic models with both engine testing and aircraft flight testing. Modelling techniques also need to be developed which address the specific needs for dynamic analyses and vibration transmission.
THE ENGINE
SPECIFICATION
3-D
- ENGINE
•
DESIGN
PARAMETER
•
ENGINE
MOUNTS
VIBRATION ENGINE
•
HARDWARE
• • •
- ENGINE DESIGNED
/ AIRCRAFT
RELATED
VIBRATION
NOISE
TRANSMISSION
TO REDUCE
TREATED
DESIGNED
ELEMENT TESTS
ENGINE
MODELS
FLIGHT
TESTS MODEL
AND
AS A SYSTEM BUILT
TO REDUCE
NOISE
MODELS
ENGINE
SYSTEM
VIBRATION
TRANSMISSION
•
FINITE
90'8
- PLANNED WITH
- PLANNED
AND
ACTUAL AND
PREDICTIONS
MODELLING TECHNIQUES USED FOR DYNAMICS AND VIBRATION
DESIGNED
TO CORRELATE
HARDWARE DESIGNED TO TEST
TO CORRELATE DATA
SPECIFICALLY TRANSMISSION
515
The complex, many DOF static structural model is still utilized for dynamic analyses; however, each of the structural components are reduced to a more manageable number of DOF's using dynamic reduction. The dynamic reduction reduces the components to its boundary DOF plus a limited number of DOF used to describe its vibratory modes. Also, many routines have been written to include such items as rotor gyroscopic stiffening, structural and viscous damping, non-linear stiffness and damping versus load and/or frequency and non-linear rub springs.
CURRENT 3-DIMENSIONAL
516
ANALYSIS
STRUCTURAL MODEL
TECHNIQUES
SYSTEM
- ALSO
•
COMPLEX
•
REDUCTION OF COMPONENTS TO DYNAMIC EQUIVALENCY - RETAIN OVERALL COMPONENT FLEXIBILITIES -
RETAIN
LOCAL
-
RETAIN
ALL
-
SIGNIFICANT
•
ROTOR
•
BEARINGS
USED
MODEL
INTERNAL
INTERNAL
STATICS
FLEXIBILITIES IMPORTANT
REDUCTION
GYROSCOPIC
FOR
MODES
IN DOF
STIFFENING
TERMS
- NON-LINEAR
-
STIFFNESS
VS ROTOR
-
STIFFNESS
/ DAMPING
VS FREQUENCY
-
STIFFNESS
/ DAMPING
VS BEARING
•
STRUCTURAL
•
ENGINE
MOUNT
LOADING LOAD
DAMPING STIFFNESS
VS LOAD
VS FREQUENCY
NASTRAN is currently used as a common analysis base between the aircraft and engine manufacturers and agreements are obtained between companies as to modelling number ranges. This permits ease of transmitting models to each other and similar DMAP can be utilized. Unique post-processing routines are utilized to review results from the steady state frequency response analyses.
CURRENT COMMON
ANALYSIS
•
DIRECT
•
LARGE
UNIQUE
ANALYSIS
TECHNIQUES
BASE
- NASTRAN
SOLUTION, AMOUNT
STEADY OF 'DMAP'
-
MATRIX
-
GYRO
-
NON-LINEAR
BEARINGS
-
NON-LINEAR
MOUNTS
-
SPECIAL
COMPONENT
•
RESPONSE
•
ANIMATION
•
BEARING
STATE
FREQUENCY
RESPONSE
WRITI'EN
INPUT STIFFENING
OUTPUT
POST-PROCESSING
•
(CONTINUED)
ROUTINES
ENERGIES (LOADS,
REQUIREMENTS
(POTENTIAL
DEFLECTIONS)
AND
KINETIC)
VS FREQUENCY
PLOTS / MOUNT
RESPONSE
517
This current static
and dynamic
on approximately than 5000.
new generation analyses. 15 major
strut/nacelle/engine
For dynamic components
__MODEL USED TO GENERATE
518
analysis to reduce
structural the model the number
DYNAMICAIJLY
model
is being
will undergo of DOF
SIM!!,AR
utilized
dynamic
from over
for both reduction
150,000
COM_PONENTS
to less
Component dynamic matrices are generated using fixed boundary component modes. The resulting mass and stiffness matrices are of the form shown. Each of the N component modes can be obtained from Ki]Mi and are coupled to the M boundary DOF by only the mass matrix. Also, full structures
can be generated
from
matrices generated from both a symmetric with the aircraft structure.
a symmetric
and anti-symmetric
COMPONENT N: COMPONENT MASS
DYNAMIC
structure
by combining analysis.
DYNAMIC
This is commonly
done
MATRICES
M: COMPONENT
DOF
the dynamic
BOUNDARY
STIFFNESS
MATRIX
DOF
MATRIX |
M1
|
M2