-
Strategic Multiscale Framework
A New Multiscale
Science - Engineering Environment
FFFooorrr
TTTeeeccchhhnnnooolllooogggyyy IIInnnnnnooovvvaaatttiiiooonnn,,,
EEEnnngggiiinnneeeeeerrriiinnnggg aaannnddd
MMMaaannnuuufffaaaccctttuuurrriiinnnggg
Alessandro Formica
March 2014
-
Alessandro Formica, March 2014 All rights reserved
2
TABLE OF CONTENTS
1. Multiscale and The Future of Technology Innovation,
Engineering and Manufacturing... pag. 3
2. Strategic Multiscale Framework Architecture.. pag. 7
3. Integrated Multiscale Science - Engineering Framework.. pag.
9
3.1 Architecture. pag. 9
3.2 Multiscale Data, Information and Knowledge Analysis and
Management Systempag. 10
3.3 Multiscale Science Engineering Information Space. pag.
18
3.4 The Information Driven Concept and Analysis Scheme... pag.
24
3.5 Multiscale Modeling & Simulation as Knowledge
Integrators and Multipliers. pag. 27
3.6 Multiscale Multiresolution Multiphysics Testing,
Experimentation and Sensing. pag. 31
3.7 Integrated Multiscale Science Engineering Analysis
Strategies.. pag. 34 3.7.1 Methodologically Integrated Multiscale
Science - Engineering Strategies pag. 34 3.7.2 Multiscale Science -
Engineering Analysis Schemes. pag. 41
3.8 Designing the R&D and Engineering/Manufacturing
Processes. pag. 44 3.8.1 R&D and Engineering/Manufacturing
Process Architecture pag. 44 3.8.2 R&D and
Engineering/Manufacturing Strategy Management System.. pag. 49
3.8.3 Integrated R&D and Engineering Analysis Strategies. pag.
51
4. Integrated Multiscale Science Engineering Technology, Product
and Process Development (IMSE-TPPD) Framework.. pag. 54
4.1 IMSE-TPPD Architecture and Overview pag. 54
4.2 Multiscale Multidisciplinary Science Engineering Cyber
Extended Enterprise Framework. pag. 55 4.3 Computer Aided R&D,
Engineering and Manufacturing /Processing (CARDE-MP) Framework....
pag. 56 4.3.1 Architecture.. pag. 56 4.3.2 Multiscale Manufacturing
and Processing.. pag. 57 4.3.3 Multiscale Environmental Monitoring
and Impact Analysis..... pag. 61
4.4 Multiscale Science Engineering Virtual Testing pag. 66
4.5 Virtual Multiscale Innovative Technology and Systems
Development Framework pag. 69 91
4.6 Virtual Multiscale Life Cycle Engineering Framework.. pag.
72
4.7 Multiscale Science Engineering Knowledge Integrator and
Multiplier (KIM) . Computing Information Communication
Infrastructural Framework....... pag. 74
About the Author.. pag. 77
-
Alessandro Formica, March 2014 All rights reserved
3
1. Multiscale and The Future of Technology Innovation,
Engineering and Manufacturing
Computational Multiscale has become a key asset in the R&D
and Engineering World and an important element for Technology,
Products and Processes Innovation. Multiscale methods helped to
establish a bridge between Science and Engineering and the related
domains of knowledge. Continuous advances in Computational Methods
(Virtual Environments) and High Performance Computing provided the
basis to define a new vision of Multiscale we refer to as
"Strategic Multiscale.
The term Strategic means that Multiscale Methodologies are
applied not only to improve Modeling and Simulation Methods, but,
also, to improve in a significant way R&D, Engineering and
Manufacturing Organization, Structure and Strategies.
Complexity of Products and Manufacturing Technologies and the
related R&D and Engineering Processes is continuously
increasing: researchers and engineers have to manage, integrate and
coordinate an ever widening spectrum of analytical, computational,
experimental, testing and sensing models, methods and
techniques.
A Fundamental Goal of the Strategic Multiscale Framework is to
address this Challenge outlining a set of new concepts, methods and
environment to Design the R&D and Engineering Process
A Distinguishing Element of the Strategic Multiscale Framework
is the new concept of Multiscale Modeling and Simulation as
Knowledge Integrators and Multipliers and Unifying Paradigm for
Scientific and Engineering Domains.
This new methodological Science - Engineering Framework allows
us to give a New Dimension and Meaning to the term Virtual as far
as Engineering and Manufacturing are concerned. and introduce a New
Field: Virtual Technology Innovation, which is the connection
element between Science and Engineering/Manufacturing Domains.
The Strategic Multiscale Framework defines a Comprehensive
Theoretical and Methodological Environment to design and implement
a New Generation of Virtual Science Based Technology Innovation,
Engineering and Manufacturing Strategies where Multiscale Modeling
and Simulation become Pivotal Elements of the R&D and
Engineering Manufacturing World overcopming classical divisions
between the Computational and the Experimental, Testing and Sensing
Areas.
A Fundamental Characteristics of the Strategic Multiscale
Framework is to allow for a smooth, continuous, efficient,
structured and timely transfer of scientific knowledge inside the
Technology Development, Engineering and Manufacturing Processes and
related Computational Frameworks.
The concept of Multiscale as Unifying Paradigm is not new. In
the mid of ninenties, several researchers in the Chemical
Engineering Field (Sapre and Katzer, Leou and Ng, and Villermaux)
and the author of this document (Alessandro Formica) highlighted
the need of a comprehensive Multiscale approach as a key Strategy
to establish a new Unifying Paradigm in order to enable a better
correlation between scientidfic and engineering advances and
related knowledge domains. Later on, Prof. Charpentier, past
European Federation of Chemical Engineering President highlighted
again the strategic relevance of this conceptual scheme.
-
Alessandro Formica, March 2014 All rights reserved
4
The definition of new Frameworks like the Integrated
Computational Materials Engineering (ICME) and the Integrated
Compuitational Materials Science Engineering (ICMSE) ones and the
launch of the US Presidential Materials Genome Initiative (MGI) put
the bases for a wide Industrial Application of Multiscale Science
Engineering Integration Strategies and Frameworks
The European Union FET FLAGSHIP Human Brain Preoject (HBP) is a
demonstration of the Strategic Value of Multiscale Science
Engineering Integration. To fully understand processes and related
relationships characterizing Brain Functions and Functionalities
over the whole range of scalea and Brain organization levels and
the fundamental relationships with diseases, new (Multiscale)
Computational, Experimental and Data Analysis Methodologies,
Techniques and Strategies will be developed and applied. New
(Multiscale) Methodologies will also be functional to develop a New
Generation of (Multilevel) Non Von Neumann Computing
Systems.Engineering and Manufacturing are quickly changing. Science
has already become a key issue and value for both the fields and
this trend will become increasingly important in the coming years .
Many Projects have clearly demonstrated that, today, is possibile
to use Multiscale (Science Engineering) Computational Methodologies
to Design and Manufacture new inherently Hierarchical Multiscale
Materials, Devicews, Components and Systems (Nano To Macro
Integration).
Fig. 1 Multiscale (Nano To Macro) System Design (MIT)
Multiscale as Unifying Paradigm for Chemical Engineering
Prof. Charpentier, past European Federation of Chemical
Engineering (EFCE) President, at the 6th World Congress of Chemical
Engineering - Melbourne 2001, described his Vision of Multiscale as
Strategic Paradigm for Chemical Engineering. We report his words:
One key to survival in globalization of trade and competition,
including needs and challenges, is the ability of chemical
engineering to cope with the society and economic problems
encountered in the chemical and related process industries. It
appears that the necessary progress will be achieved via a
multidisciplinary and time and length multiscale integrated
approach to satisfy both the market requirements for specific end
use properties and the environmental and society constraints of the
industrial processes and the associated services.
This concerns four main objectives for engineers and
researchers:
(a) total multiscale control of the process (or procedure) to
increase selectivity and productivity,
(b) design of novel equipment based on scientific principles and
new methods of production: process intensification,
(c) manufacturing end-use properties for product design: the
triplet processus-product-process engineering,
(d) implementation of multiscale application of computational
modeling and simulation to real-life situations: from the molecular
scale to the overall complex production scale.
-
Alessandro Formica, March 2014 All rights reserved
5
Multiscale Science Engineering Integration implies that we can
not only define a Science Driven Engineering, but, also a
Engineering Driven Science. Multiscale Computational Methodologies
(Virtual Engineering and Manufacturing) should consider the impact
of these global trends over their development, structure and
related implementation strategies in order to define their
Future.
It is to be highlighted that the awarding of the Nobel Prize in
Chemistry to three scientists for the development of Multiscale
Models for Complex Chemical Systems has helped to create the
optimal intellectual and scientific context to introduce high level
Projects and Initiatives in the Multiscale Science Engineering
Integration field
Strategic Multiscale Framework Goals:
Defining a New Organization and Structure of the R&D and
Engineering World: Designing the R&D and Engineering Process
Architecture
Defining a New Frontier for Virtual Worlds and Application
Strategies: Virtual Multiscale Science Based Technology Innovation,
Engineering and Manufacturing
Easing knowledge transfer between the different stages of the
R&D and Engineering/Manufacturing process
Integrating and Strcuturine Data, Information and Knowledge from
the Scientific and Engineering Worlds
Defining new cooperation and partnering schemes among academy,
research, and industry. In the new science-engineering context,
engineering can become an important driver for science, overturning
historic relationships and dependencies and putting the bases for a
new way of doing science and engineering. Not only advances in
science can be stimulated and driven by technology progress and the
need to solve specific technological and engineering problems, but
research strategies will be more and more influenced by technology
roadmaps and vice versa.
Putting the bases to define a new structure and organisation for
the research and industrial world based, from an Infrastructural
point of view, on a new generation of Multiscale Multidisciplinary
Science Engineering Cyberinfrastructures and, from a methodological
point of view, on the here described Framework to bridge the gap
between disciplines and the different scientific and engineering
approaches.
Enabling new Technological Engineering Solutions (Multiscale
Engineering: From Multiscale Analysis to Multiscale Design). New
Frameworks enable the design of inherently Hierarchical Multiscale
Systems (materials, structures components, products and processes)
which is a fundamental condition to fully exploit in the industrial
environment the potentialities of Nano and Micro Technologies.
In such a context it is possible to realize a real fusion
between
science-driven engineering and engineering-driven science
which represents a key goal of the Strategic view of
multiscale.
The Royal Swedish Academy of Sciences has awarded the Nobel
Prize in Chemistry for 2013 to Martin Karplus of Universit de
Strasbourg, France and Harvard University, Cambridge, MA, USA;
Michael Levitt of Stanford University School of Medicine, Stanford,
CA, USA; and Arieh Warshel of the University of Southern
California, Los Angeles, CA, USA "for the development of multiscale
models for complex chemical systems
-
Alessandro Formica, March 2014 All rights reserved
6
General References
David L. McDowell, Jitesh H. Panchal, Hae-Jin Choi. Carolyn
Conner Seepersad, Janet K. Allen, Farrokh Mistree, 2010. Integrated
Design of Multiscale, Multifunctional Materials and Products -
Published by Elsevier .
Oden , J.T. , Belytschko , T. , Fish , J. , Hughes , T.J.R. ,
Johnson , C. , Keyes , D. , Laub , A. , Petzold , L. , Srolovitz ,
D. , Yip , S. , 2006 . Simulation-based engineering science:
Revolutionizing engineering science through simulation . In : A
Report of the National Science Foundation Blue Ribbon Panel on
Simulation-Based Engineering Science . National Science Foundation
: Arlington, VA .
Olson , G.B. 1997 . Computational design of hierarchically
structured materials . Science, 277 ( 5330 ) , 1237 1242 .
Alessandro Formica. Fundamental R&D Trends in Academia and
Research Centres and their Integration into Industrial Engineering
Report drafted on behalf of European Space Agency, July 2000
Alessandro Formica, Multiscale Science Engineering Integration A
New Frontier for Aeronautics, Space and Defense, Italian
Association of Aeronautics and Astronautics (AIDAA), March 2003
-
Alessandro Formica, March 2014 All rights reserved
7
2. Strategic Multiscale Framework Architecture
The theoretical and methodological basis of the Strategic
Multiscale Framework is constituted by the following key
elements:
The extension of the Model concept to the Experimentation,
Testing and Sensing Fields giving a new meaning to the Virtual
Engineering and Manufacturing concept and approach. In this
context, a new Vision of Multiscale Modeling & Simulation as
Knowledge Integrators and Multipliers and Unifying Paradigm for
Scientific and Engineering Methodologies and Knowledge Domains has
been defined. Multiscale Modeling and Simulation integrate the full
spectrum of science and engineering methodological approaches and
knowledge environments. This new Vision puts Computational
Frameworks and High Performance Computing at the center of the
R&D and Engineering/Manufacturing World even more than
classical Virtual concepts and approaches.
The Multiscale Science-Engineering Information Space concept to
integrate data, information and knowledge from computational models
and methods and experimental, testing and sensing models and
techniques to develop, validate and apply Computational Models.
Uncertainty Quantification (UQ) and Quantification of Margin of
Uncertainties (QMU) have become critical issues as the relevance of
Modeling and Simulation is continuously increasing.
The Information Driven Analysis concept and scheme which,
together with the Science Engineering Information Space concept is
a key element to shape Integrated R&D and
Engineering/Manufacturing Analysis and Design Strategies, following
the Multiscale Modeling and Simulation as Knowledge Integrators and
Multipliers concept and application environment.
New Multiscale Science Engineering Data, Information and
Knowledge Management Systems based upon the Multiscale Maps
concept
New Multiscale Methods to Model, Simulate and Design the
Technology Development and Engineering/Manufacturing Processes and
Products Life Cycle: Virtual Multiscale Innovative Technology and
Systems Development Framework and Virtual Multiscale Life - Cycle
Engineering Framework and Environmental Impact Analysis
Basic Conceptual, Theoretical and Methodological Framework
Integrated Multiscale Science Engineering Framework
Integrated Multiscale Science Engineering Technology,
Product
and Process Development
(IMSE-TPPD) Framework
Analysis And Design of a New Generation of Materials, Devices
Systems, and related Manufacturing Processes
-
Alessandro Formica, March 2014 All rights reserved
8
The Strategic Multiscale Framework embodies the following
Elements:
Integrated Multiscale Science Engineering Framework Described in
the Chapter 3 - which represents the theoretical, conceptual and
methodological core. Key Elements: Multiscale Data, Information and
Knowledge Analysis and Management System Multiscale Science
Engineering Information Space Multiscale Modeling & Simulation
as Knowledge Integrators and Multipliers Multiscale Multiresolution
Experimentation, Testing and Sensing Methodologically Integrated
Multiscale Science Engineering Strategies The Information Driven
Concept Multiscale Science - Engineering Analysis Schemes R&D
and Engineering/Manufacturing Process Architecture R&D and
Engineering Analysis and Design Strategy Management System
Integrated R&D and Engineering Analysis Strategies
Integrated Multiscale Science Engineering Technology, Product
and Process Development (IMSE-TPPD) Application Framework -
Described in the Chapter 4 - Multiscale Multidisciplinary Science
Engineering Enterprise Framework
Computer Aided R&D, Engineering and Manufacturing/Processing
(CARDE-MP) Framework which implements the Integrated Multiscale
Science Engineering Framework
Multiscale Manufacturing and Processing Multiscale Environmental
Monitoring and Impact Analysis
Multiscale Science Engineering Virtual Testing Virtual
Multiscale Innovative Technology and Systems Development
Framework
Virtual Multiscale Life Cycle Engineering Framework
The Multiscale Knowledge Integrator And Multiplier Computing,
Information and Communication (CIC) Infrastructural Framework
A distinguishing characteristics of the Strategic Multiscale
Framework is that it can incorporate and take advantage of a wide
range of existing Software Environments in many areas: from Data
Analysis, Workflow Management, Statistics, Graphics, Single and
Multiscale Computational Codes to name a few.
-
Alessandro Formica, March 2014 All rights reserved
9
3. Integrated Multiscale Science - Engineering Framework
3.1 Architecture
Main elements of the Conceptual and Methodological Framework
are:
Multiscale Science - Engineering Data, Information and Knowledge
Analysis and Management System
Multiscale Science Engineering Information Space
Information Driven Multiscale Science Engineering Analysis
Concept and Schemes
Multiscale Modeling & Simulation as Knowledge Integrators
and Multipliers and Unifying Paradigm for Scientific and
Engineering Methodologies and Knowledge Domains The role of
Multiscale as Unifying Paradigm and Language for Science and
Engineering was discussed by Alessandro Formica, some years ago in
the book - Computational Stochastic Mechanics In a Meta-Computing
Perspective December 1997 - Edited by J. Marczyk pag. 29 Article: A
Science Based Multiscale Approach to Engineering Stochastic
Simulations.
Multiscale Multiresolution Multiphysics Testing, Experimentation
and Sensing
Methodologically Integrated Multiscale Science Engineering
Methodologies
New Methods, Tools and Strategies to Design the R&D and
Engineering Analysis Process
Integrated Multiscale R&D and Engineering Analysis
Strategies
-
Alessandro Formica, March 2014 All rights reserved
10
3.2 Multiscale Science Engineering Data, Information and
Knowledge Management System
A critical issue for a wide diffusion of the science based
engineering analysis and design approach in the industrial field is
the availability of Software Environments (CAD/CAE/CAM/CAP)
specifically conceived for multiscale science engineering
strategies and applications. Today, notwithstanding the growing
diffusion of multiscale inside university, research, and even
industry, software environments (CAD/CAE/CAM/CAP) specifically
conceived to implement multiscale science-engineering integration
visions and strategies are still in their starting phase. The lack
of software environments specifically conceived to implement a
multiscale science-engineering integration strategy represents a
fundamental hurdle to a large scale implementation of multiscale
inside innovative technology development and
engineering/manufacturing/processing fields. The new Data,
Information and Knowledge Management System proposed in this
Document rests on the concepts of:
Multiscale Multiphysics Multiresolution Maps
Multiscale Multi Abstraction Level Knowledge Domains
The Multiscale Multiresolution Maps here described is an
extension of the Map concept discussed by Alessandro Formica in the
Multiscale Science Engineering Integration: A new Frontier for
Aeronautics, Space and Defense White Book published on March 2003
by Italian Association of Aeronautics and Astronautics.
Multiscale Multiresolution Maps are Multiscale Multiresolution
Information and Knowledge Structures describing complex networks of
relationships and interdependencies between a large spectrum of
Information Variables characterizing Systems Structure and
Dynamics. Relationships and interdependencies between Information
Variables are worked out applying several mathematical techniques
such as multivariate analyses and neural networks [available inside
a specific Data Analysis Module. This Module correlates sets of
Data to Data Sources, Tasks and Integrated Strategy Maps] to raw
data coming from a wide range of Data Sources (analytical and
computational models, data bases, experimentation,
characterization, testing and sensing). covering the full spectrum
of scales (from atomistic to macro) and the full spectrum of
disciplines. Multiscale Maps structure Data and turn Data into
Information. Maps are organized in a hierarchical way: A Map can
incorporate a set of lower level Maps. For instance: a Multiscale
Physical Map linked to a specific Process (Hypervelocity Impact,
Combustion or Explosion, for instance) can be constructed by
assembling a range of Multiscale Physical Maps describing more
elementary physical (chemical and biochemical) phenomena (fracture,
fragmentation, phase change,..) related to a specific material or
component of a System. Accordingly any Element represented inside a
Map can be decomposed into more elementary Elements.
Representations can be static and dynamic. Multiscale Maps
incorporate error analyses and uncertainty quantification methods.
Multiscale Maps make an extensive use of Multiscale Multiresolution
Static and Dynamic Graphic Representations.
Multiscale Multi Abstraction Knowledge Domains are a further
organization level is represented by Knowledge Domain which are
Structures that can aggregate several Maps related to one or more
scales of the same typology or of different typologies related to
the same or different operational conditions, analysis and design
hypotheses and solutions. Maps can be set up and integrated inside
a Knowledge Domain applying several aggregation and clustering
schemes. Knowledge Domains can be organized in a Hierarchical Way.
Knowledge Domains can be related to specific R&D and
Engineering/Manufacturing Tasks and Phases. They can track
Knowledge structure and organization as we transition from a
R&D and Engineering Phase and Task to another one.
Multiscale Maps and Multiscale Knowledge Domains allow for an
effective insertion and management of the more fundamental
knowledge (basic and applied research) inside the sequence of
Technology Development and Engineering phases. At each step/phase
of the R&D and Engineering Process , Multiscale Maps and
Knowledge Domains are built taking full advantage of the knowledge
get in the previous step/phase.
-
Alessandro Formica, March 2014 All rights reserved
11
Several typologies of Maps are foreseen which describe
relationships among variables, structures and processes:
A) R&D And Engineering (for systems of any kind of
complexity)
Multiscale Analysis and Design Variable Maps tracking
relationships between Analysis and Design Variables . Multiscale
Analysis and Design Variable Maps are built applying available
knowledge and current working hypotheses at the drafting time;
statistical analysis schemes (multivariate, PCA) or other
techniques like neural networks to data coming from several
sources: data bases, computation, analytical theories,
experimentation, testing, sensing. Data integration and fusion
techniques are applied to reconcile and integrate data coming from
different sources characterized by a range of accuracy and
reliability degrees. Multiscale Analysis and Design Variable Maps
describe relationships between variables and parameters used to
characterize Systems Behaviour over a full range of space and time
scales and disciplines and over a range of operational
conditions..
Multiscale Physics Maps identifying the Physical, Chemical and
Biochemical Phenomena and Processes considered fundamental to carry
out a specific task and describe relationships and
interdependencies among them. The following table illustrates a
textual version of a simplified Physics Map:
Multiscale Architectural/Structural Maps describing
relationships between the hierarchy of Sub-Systems, Components,
Devices, Materials and Elementary Structures constituting an
Engineering System (or System of Systems) of arbitrary level of
complexity. This kind of Maps incorporates a special set of
Elements referred to as Interfaces which describe interconnections
among Architectural/Structural constituents inside a scale and
among different scales.
Engineering System Multiscale Monitoring and Control Maps
describing (Hierarchical) Networks of Sensors and Control Devices
and Systems and their relationships with Elements to be monitored
and controlled (described in the Multiresolution Multiscale
Architectural/Structural Maps). Transformation Processes induced by
control actions are described thanks to Multiresolution Multiscale
Physics Maps and Multiresolution Multiscale
Architectural/Structural Maps. This kind of Maps describes the
quantities monitored and controlled, time and space monitoring and
control resolution, sensing and control devices characteristics and
operational schemes
Multiscale Materials (Tantalum) Characterization (Livermore)
Physics Map
Atomistic length-scale modeling input: interatomic potentials
(calculated with quantum mechanics) output: dislocation generation,
motion and interaction with other defects scale physics: properties
of individual defects (dislocations, vacancies, interstitials,
dopants), defects mobility, diffusion, clusters, surface
reactions
Microscale length scale modeling input: dislocation generation,
motion and interaction with other defects output: yield and
hardening rules for single crystals scale physics : defect
interactions, precipitates, dislocation reactions, the early stages
of void growth, grain boundaries and the interactions between
dislocations and grain boundaries
Mesoscale Modeling input: yield and hardening rules for single
crystals output: mesoscale models of polycrystal aggregates (100s
of grains) scale physics : shear band, dislocation walls,
collective dynamics of microstructure, interface diffusion, grain
coarsening, recrystallization, crack growth, fracture
Mesoscale Homogenization / Continuum Model input: mesoscale
models of polycrystal aggregates (100s of grains) output: pressure
and strain path dependent yield surface for continuum code
hardening. scale physics : polycrystal plasticity, temperature
fields, hydrodynamic motion, textures, microstructures
homogenization, anisotropic hardening.
-
Alessandro Formica, March 2014 All rights reserved
12
Multiscale Functional Maps describing relationships between
Engineering System Architectural/Structural Elements and Functions
performed
Multiscale Requirements - Performance Property Structure Maps
describing relationships between Requirements, Performance,
Structural Elements and related Properties over the whole scales
and representation levels.
Multiscale Performance Property Structure
Manufacturing/Processing Maps describing the impact of Processing
techniques over the network of Performance - Structure - Property
relationships over the whole scales and representation levels.
Fig. 2 Physics Map Example (from Overview of the Fusion
Materials Sciences Program Presented by S.J. Zinkle, Oak Ridge
National Lab Fusion Energy Sciences Advisory Committee Meeting
February 27, 2001 Gaithersburg)
This figure depicts a Information Structure like the proposed
Multiscale Physics Maps. In this case the Multiscale Physics Map
describes relationships between physical phenomena and
chemical/physical structural transformations linked to Radiation
Damage Process for Metals
A cluster of Multiscale Physics Maps, linked to specific physics
phenomena or processes, architectural element and operational
conditions, can define what can be called a Physical (Chemical and
Biochemical) Phenomena and Processes Knowledge Domain. Knowledge
Domains are managed by the Multiscale Science Engineering Data,
Information and Knowledge Management System.
-
Alessandro Formica, March 2014 All rights reserved
13
Multiscale Multilevel Architectural and Structural Maps
Any System of arbitrary degree of complexity (an air
transportation system, an energy production system, an aerospace
vehicle, a chemical plant, a structure, a nanotechnology device, a
nanostructured material), can be recursively broken down in a set
of simpler (macro, meso, micro, nano and atomistic) Architectural
and Structural Elements and Interface/Interconnection Elements.
Interconnections and Integration develop along two lines:
Horizontal (same scale) and Vertical (different scales). We
distinguish two kinds of Systems: Technological Systems and Natural
Systems where the Technological System (or System of Systems)
operates.
Fig. 3 Two dimensional multilevel multiscale view of an
aircraft. (from the Validation Pyramid and the failure of the A-380
wing Presentation given by I. Babuska (ICES, The University of
Texas at Austin), F. Nobile (MOX, Politecnico di Milano, Italy), R.
Tempone (SCS and Dep. of Mathematics, Florida State University,
Tallahassee) in the context of the Workshop Mathematical Methods
for V&V SANDIA , Albuquerque, August 14-16, 2007
Three new features distinguish this kind of Maps and related
Multiscale Multilevel Science Engineering CAD Systems:
Multiscale Multilevel Architectural/Structural Element Networks
Analysis and Description. New CAD Systems should describe the full
set of multiscale multilevel (inside a single scale) Architectural
and Structural Elements of a System (or System of Systems) -
including the Operational Environment - and related
interconnections. Interconnection Elements describe two way
interactions between Elements. This feature is of particular
importance for System Engineering analyses and if we like to assess
the impact of the System upon the environment where it operates and
the effects of the Environment on the System for the whole Life
Cycle and the whole spectrum of operational conditions including
extreme ones and accidents.
Zooming and Selected Multilevel Multiscale view capabilities.
Users should have the possibility to select a full spectrum of
views at different levels of resolution, scales and abstraction
ways. Multiple views should be visualized in order not to lose
connections among different levels of abstraction, resolution and
scales. The zooming function should allow users to transition from
a levels of abstraction, levels and scales in an interactive
way.
Multi Abstraction Levels: we can select groups (clusters) of
architectural/structural elements of different typologies over a
spectrum of scales and resolution levels as needed to carry out
specific analyses and design tasks.
-
Alessandro Formica, March 2014 All rights reserved
14
This kind of Maps gives a comprehensive picture of the:
Architectural and Structural and Interface/Interconnection
Elements (from macro to atomic levels as needed) which constitute a
system and its related Horizontal and Vertical organization: from
the System (or System of Systems) down to elementary structures
(atoms/molecules, groups of atoms and molecules)
Materials, Energy, Chemical and Biochemical Substances Flow
(pollutants emitted toward the Natural System for instance,) among
the Elements constituting the System or System of Systems
Analysis and Design Variables their relationships and
interdependencies and links between Analysis and Design Variables
and Architectural and Structural Elements
Properties of the full set of Architectural and Structural
Elements
Performance and Requirements for the full set of Architectural
and Structural Elements. Performance are calculated and/or measured
during the R&D and Engineering Process while, Requirements are
defined and refined by designers.
Architectural and Structural Maps evolve along the Technology
Development and Engineering Analysis and Design Process thanks to
Analysis and Design Modules and Strategy Modules. Maps are built
using the available knowledge; as analysis and design activities
proceed, they are interactively modified. Different Maps can be
linked to different Architectural Hypotheses and Decisions for
different purposes and tasks during the R&D and Engineering
Process. Maps are recorded, organized and managed in specific
Architectural and Structural Map Data Bases. Architectural and
Structural Elements Maps are related to: Functional Maps
Monitoring and Control Maps
Physics Process Maps
Multiscale Monitoring and Control Maps
This kind of Maps gives a comprehensive picture of the
Multiscale Multilevel Networks of Monitoring and Control Devices
and Systems their interconnection schemes and their functionalities
and operational modes. Multiscale Monitoring and Control Maps are
related to: Architectural and Structural Maps
Physics Maps (Physical and Bio-Chemical Phenomena and Processes
Monitored and effects of Control actions)
Multiscale Functional Maps
We define two types of Functional Maps. The first one, which can
be called Direct Functional Map, describes Functions carried out by
the
System and the full hierarchy of its Elements. Direct Functional
Maps link Architectural/Structural Elements to Functions and they
describe what functions are performed by Architectural/Structural
Elements.
The second one, which can be called Inverse Functional Map
relates Functions to Architectural/Structural Elements over the
full spectrum of hierarchy levels
Functional Maps are linked to:
Architectural and Structural Maps
Physics and Processes Maps
Functional Maps defined during the Technology Development and
Engineering Process are recorded, organized and managed by specific
Functional Maps Data Bases. Maps are indexed in such a way as to
relate them to specific R&D and Engineering Phases and
Tasks.
-
Alessandro Formica, March 2014 All rights reserved
15
Multiscale Multiphysics Maps
We use the term Physics to indicate a more or less complex
cluster of elementary physical and biochemical phenomena/processes
occurring inside a scale or developing over a spectrum of scales.
Phenomena/Processes are, for instance, failure, stress corrosion
cracking erosion, phase transformation, A Process can be broken
down in a full hierarchy of more elementary Processes and
Phenomena. The distinction between processes and phenomena is, to
some extent, arbitrary. It is a matter of opportunity. Phenomena
and Processes can concern more Architectural/Structural
Elements.
Physics Maps are linked to:
Architectural/ Structural and Functional Maps.
Monitoring and Control Maps
Requirements - Performance Property Structure Maps
Performance Property Structure - Processing Maps
Multiscale Manufacturing Systems any level of the hierarchy -
Operational Modes - Environmental Emission Maps
Physics Maps are software environments which describe :
the full set of physical (biological and chemical, as needed)
phenomena and processes which rule the dynamics of
Architectural/Structural Elements (Interconnection Elements
included) of a System under analysis/design for a specific Task and
their interactions inside a scale and over different scales.
The full hierarchy of (geometrical, physical and bio- chemical)
Architectural/Structural transformations related to a specific set
of Phenomena/Processes linked to a specific R&D and Engineering
Task .
Relationships between the full hierarchy of processes, phenomena
and Architectural/Structural transformations for a specific
Task
Maps are indexed in such a way as to relate them to specific
R&D and Engineering Phases and Tasks.
Physics Maps are linked to Multiscale Methodologically
Integrated Strategy Maps described in the Paragraph 3.8.1.
Multiscale Methodologically Integrated Strategy Maps describe what
Computational Models, Experimentation, Testing and Sensing
Techniques/Procedures are applied to analyze specific physical
phenomena/processes and their interconnection networks, sequence of
execution and data. Physics Maps are built using the available
knowledge, as R&D and Engineering proceed, they are
interactively modified.
Physics Maps defined during the R&D and Engineering Process
are recorded, organized and managed by specific Physics Maps Data
Base.
Integration of the previously defined Multiscale Maps allow to
correlate:
functions to physical phenomena and processes (linking
Multiscale Functional Maps with Multiscale Physics Maps
Properties (Multiscale Architectural/Structural Maps) to Physics
(Multiscale Physics Maps)
-
Alessandro Formica, March 2014 All rights reserved
16
Multiscale Performance Properties Structure Processing Maps
The definition of the Performance Properties Structure
Processing relationships has become a cornerstone of the modern
Materials Science and Engineering and R&D and Engineering at
all. Prof. Gregory Olson, Northwestern University has been one of
the pioneers of this strategy. Prof. Olson described this approach
in a Science Magazine article: Vol. 277 (29 August 1997) pp.
1237-1242.
Fig. 4 (from Questek) illustrates the application of a
Performance Properties - Structure Processing Map to the design of
new alloys.
Performance Properties Structure - Processing Maps are indexed
in such a way as to relate them to specific R&D and
Engineering, Phases and Tasks. Performance Properties - Processing
Structure Maps defined during the R&D and Engineering Process
for different purposes and tasks are organized and recorded in the
Performance Properties - Structure - Processing Map Data Bases The
Multi Abstraction Level feature of the Maps can be seen in the
figure: each box is a specific abstraction level. Each Box refer to
a cluster of processes occurring over u spectrum of scales and
resolution levels.
This kind of software environments contribute to characterize
and manage relationships between processing and manufacturing
activities and the resulting architecture/structures
These Maps identify :
defects (typology, physical and chemical characteristics,
density and distribution : statistical and deterministic analysis)
linked to specific processes and manufacturing activities and
steps
bio chemical and structural features and transformations linked
to specific processes and manufacturing conditions, procedures and
technologies
This kind of Maps are related to Multiscale Physics Maps
-
Alessandro Formica, March 2014 All rights reserved
17
B) Manufacturing and Processing
Multiscale Manufacturing and Processing Systems Architectural,
Functional and Monitoring & Control Maps describing: the
(multiscale/multilevel) architecture (hierarchical networks of
units [from Plants to Cell,
Robots and Machines/Tools] at different scales and complexity
levels of any kind of Manufacturing/Processing Systems and related
interconnections and interdependencies (material flow). At the
highest abstraction level, a Unit can represent a whole
Manufacturing/Processing Systems incorporating several Plants and
other Elements. The representation scheme is recursive: A Unit can
be decomposed into a network of simpler Units, a simpler Units can
be, in turn, be decomposed into other networks of even more
elementary Units over the whole hierarchy of scales and
representation levels, as needed.
the full spectrum of functions carried out by the units
constituting Manufacturing/Processing Systems and their
relationships and interdependencies. Multiscale Physics Maps and
Multiscale Structural Maps are applied to describe physical,
chemical and bio-chemical transformations and processes occurring
at and over the full spectrum of Units.
the (Hierarchical) networks of multiscale monitoring and control
(M&C) devices and systems over the full spectrum of scales and
levels This kind of Maps describes the quantities monitored and
controlled, time and space resolution, sensing and control devices
characteristics and operational schemes
the (Hierarchical Network) of Inspection Systems, their
Functions and Operational Modes
Multiscale Multilevel Manufacturing Processes Execution Flow:
this kind of Maps describes, for each specific Manufacturing
Process of any level of complexity the execution flow
(manufacturing/process sequence of steps) throughout the full set
of Plants. Process Units, Cells, Machines/Robots,., the work
performed at each step, the characteristics of the Unit, the
structural/chemical/physical transformations worked out (also using
Architectural/Structural Maps and Physics Maps), the inspections
performed, the materials flow..
Multiscale Manufacturing Systems any level of the hierarchy -
Operational Modes - Environmental Emission Maps Theses Maps
represent a new Generation of Maps specifically conceived to
evaluate the impact on the Environment of Manufacturing/Processing
Systems for a wide range of operational conditions and design
solutions. Maps describe relationships among Manufacturing System
(any level), its Operational Modes and related Emissions (any
kind).
Multiscale Maps represents a key element of a new Multiscale
Computer Aided Research, Development, Engineering (CARDE-MP)
Software Systems. Main objectives:
Developing new schemes allowing for a more in-depth analysis and
structuring of data, information and knowledge and related
correlations and interdependencies
Integrating the full spectrum of Data Sources (Data Bases,
Analytical Theories, Computational Models, Experimentation ,
Testing and Sensing). The Information Space and the Modeling and
Simulation as Knowledge Integrators and Multipliers concepts and
methods can ease this kind of Integration
Developing new CAD/CAE/CAM/CAP Environments specifically
conceived to Design and Produce new Hierarchical Multiscale Nano To
Macro Multifunctional Systems in the context of an Integrated
Science Engineering Approach
Multiscale Maps are indexed and related to specific R&D and
Engineering Tasks and Phases, Design Hypotheses and Design
Decisions and Operational Conditions.
The Multiscale Science Engineering Data, Information and
Knowledge Management System records, organizes and manages all the
previously defined Maps and Knowledge Domains (Item A and Item B).
Each Map and Knowledge Domain is characterized by a set of Tags
which link it to a specific task, phase and operational conditions
and analysis and design hypothesis inside the R&D and
Engineering/Manufacturing Analysis and Design Process.
-
Alessandro Formica, March 2014 All rights reserved
18
3.3 Multiscale Science - Engineering Information Space
This concept was presented by Alessandro Formica in the Report
Fundamental R&D Trends in Academia and Research Centres and
Their Integration into Industrial Engineering (September 2000),
drafted for European Space Agency (ESA). The Multiscale
Science-Engineering Information Space is associated to any
analytical, computational model/method, and experimental, testing
and sensing procedure and technique applied to a specific task. The
Multiscale Science-Engineering Information Space defines:
what spectrum of information about physical/biological/chemical
phenomena and processes at what level of accuracy and reliability
(Uncertainty Quantification (UQ) and Quantification of Margin
of Uncertainty (QMU))
can be get by a computational model or
experimental/testing/sensing technique/procedure applied in a
specific context for a specific task.
A set of model variables characterize analytical and
computational models. A set of method variables characterize the
specific method applied to perform simulations. A set of system
variables characterizes the system to be modeled and simulated or
subjected to experimental, testing and sensing analyses. A set of
experimental, testing and sensing variables characterizes
experimental, testing and sensing techniques and procedures.
The Science Engineering Information Space also applies to
cluster of computational models and experimental/testing/sensing
techniques/procedures linked through multiscale multiphysics
coupling schemes. In this case we can define coupling scheme
parameters which describe the method used to couple models and/or
experimental/testing/sensing techniques/procedures.
With the term system we refer to the system (materials, device,
component,.) under analysis.. A set of variables describe the
geometrical, biological, chemical and physical structure of the
system.
With the term Operational Environment, we refer to External
Fields and Loading Conditions
With the term model we refer to the mathematical/computational
representation of the system under investigation. A set of
variables characterize and describe the models (boundary
conditions, external fields/loading conditions, space and time
dimensions, discretization techniques, particles number and
typology,.). In the proposed framework we extend the concept of
Model to the Experimental/Testing/Sensing world as explained in the
Paragraph 2.4
With the term method we refer to the specific deterministic and
statistical analytical and computational method (Monte Carlo.
Classical Molecular Dynamics, Quantum Molecular Dynamics, Density
Functional Theory, Dislocations Dynamics, Cellular Automata,).
With the term experimental/testing/sensing technique and
procedure variables we refer to the variables which describe
technical characteristics of the experimental and testing apparatus
and the specific operational modes and conditions (globally
referred to as procedure)
Information Space Construction To build the Information Space of
a specific (single scale or multiscale) computational model with
reference to a specific system and analysis task (fracture,
delamination, oxidation,), we perform a set of simulations, varying
in a systematic way parameters/variables which characterize the
physical (chemical and biochemical) phenomena/processes of interest
in the context of a specific task including external forces. Then,
we validate computational models using a set of experiments, tests
and sensing measures to track the boundaries of the Information
Space and evaluate accuracy and reliability (Uncertainty
Quantification UQ). Information Spaces can be built also for
experimental, testing and sensing techniques and procedures. In
this case a Cross Validation strategy is applied which foresee the
comparison of a spectrum of experimentation, testing and sensing
techniques.
The next page Box synthetically describes the key role and
significance played by new Verification & Validation Strategies
(Uncertainty Quantification and Quantification of margin of
Uncertainty) for the Computational field.
-
Alessandro Formica, March 2014 All rights reserved
19
The Predictivity and Validation Issues
The National Nuclear Security Program (NNSA), in the context of
the Advanced Simulation and Computing (ASC) Initiative, established
the Predictive Science Academic Alliance Program (PSAAP) focusing
on the emerging field of predictive sciencethe application of
verified and validated computational simulations to predict the
behavior of complex systems where routine experiments are not
feasible. The goal of these emerging disciplines is to enable
scientists to make precise statements about the degree of
confidence they have in their simulation-based predictions. Five
PSAAP Centers have been created: California Institute of
Technology: Center for the Predictive Modeling and Simulation of
High-Energy Density Dynamic Response of Materials; Purdue
University: Center for Prediction of Reliability, Integrity and
Survivability of Microsystems (PRISM); Stanford University: Center
for Predictive Simulations of Multi-Physics Flow Phenomena with
Application to Integrated Hypersonic Systems; University of
Michigan: Center for Radiative Shock Hydrodynamics (CRASH);
University of Texas at Austin: Center for Predictive Engineering
and Computational Sciences (PECOS)
The following text, drawn from the Presentation Can Complex
Material Behavior be Predicted? Given by Prof. Michael Ortiz,
Caltech PSAAP Center Director, at the DoE NNSA Stockpile
Stewardship Graduate Fellowship Program Meeting Washington DC, July
14, 2009, illustrates objectives and approach underlying the
general PSAAP Strategy and Methodology concerning Validation and
Predictivity challenges:
PSAAP Caltech High-Energy-Density Dynamic Response of Materials
(Hypervelocity Impact Application Field) Center objective: rigorous
certification of complex systems operating under extreme
conditions. l
Overarching Center objectives: Develop a multidisciplinary
Predictive Science methodology focusing on high-energy-density
dynamic response of materials Demonstrate Predictive Science by
means of a concerted and highly integrated experimental,
computational, and analytical effort that focuses on an
overarching ASC-class problem: Hypervelocity normal and oblique
impact at velocities up to 10km/s
Overarching approach: A rigorous and novel Quantification of
Margin of Uncertainty (QMU) methodology will drive
and closely coordinate the experimental, computational,
modeling, software development, verification and validation efforts
within a Yearly Assessment format
Two issues deserve to be highlighted:
The central role of the Uncertainty Quantification and
Quantification of Margin of Uncertainty issues in the context of
the Computational Models Validation effort to shape R&D and
Engineering activities. This vision can be, to some extent, related
to the previously illustrated concepts: Multiscale Science
Engineering Information Space, Range of Validity and Information
Driven R&D and Engineering Strategy
The key role of Computational, Analytical and Experimental
Efforts Integration. New (multiscale) experimental techniques and
analytical (theoretical) developments are fundamental to develop
and apply new and more powerful (predictive) computational models
and strategies. The Vision is in line with our Methodologically
Integrated R&D and Engineering approach
-
Alessandro Formica, March 2014 All rights reserved
20
The Information Space, should also include Multiscale Analysis
and Design Variable and Multiscale Physics Maps worked out during
the previously described construction process.
It is possible to apply different schemes to build the
Information Space for a specific task. For instance:
fixing model and methodology variables and varying external
conditions and/or system variables (typology and architecture of a
material or device)
fixing external conditions and system variables and varying
model and/or methodology variables (for a molecular dynamics model:
simulation time, force fields typology, number of particles,).
any other possible combinations
The Information Space, for each specific computational
model/method (or cluster of models: multiscale multiphysics)
applied to a specific task includes information about the computing
resources needed to perform simulations and the experimental,
testing and sensing techniques used to validate it
Information Space Relevance
Three considerations underlie the definition of the Multiscale
Science Engineering Information Space concept and method:
rationally correlating advances for models/methods and
multiscale multiphysics coupling schemes with the capability of
getting information thought to be important to carry out specific
R&D and Engineering/Manufacturing tasks.
rationally defining the role of models/methods and related
multiscale multiphysics coupling schemes inside a more general
R&D and Engineering analysis and design process and the
interdependencies among different models, methods, techniques and
coupling schemes.
formally tracking and planning the development path (roadmap)
for models, methods, techniques and related coupling schemes as
linked to specific R&D and Engineering analysis and design
tasks, and assessing the relative importance of the different
models/methods and related coupling schemes to get some Information
at a specific level of accuracy and reliability.
We can consider an aerodynamic design task, for instance. The
ability to run a 30/50-million grid points Navier Stokes simulation
in the same lapse of time, or less, as a 1-million grid points
simulation, is surely an important result from an engineering
analysis and design point of view. But, what is the relative weight
between model dimension and physics (turbulence) modeling as
function of a particular task (calculation of aerodynamic
coefficients, for instance) at a certain level of accuracy and
reliability?
In this way, can we get more reliable and accurate information
instrumental to reducing cost and development time and introduce
innovative technological solutions? The answer is not so
straightforward. Turbulence plays a key role in flow dynamics
phenomena of critical importance for the design of a wide range of
systems. Suppose the biggest simulation model used the same
turbulence model (or a slight modification) as the one employed in
the smallest one, what is the relationship among the number of grid
points, turbulence modeling (model variables) and the capacity of
getting the needed engineering information at the right level of
accuracy (for instance : CP - CL or vortex dynamics look at the
V-22 vortex ring state story) ? Is the number of grid points or the
turbulence modeling the dominant knowledge factor from a designer
point of view?
The situation becomes even more critical when the physics and
chemistry to be taken into account are highly complex
(aerothermodynamics and combustion, for example). It is sufficient
to think at a combustion chamber or an hypersonic vehicle. Several
variables such as complex thermo chemical phenomena, the
interaction between turbulence and chemistry, multiphase and phase
change phenomena, condition the information space linked to a
model.
We introduce, now, the Range of Validity concept for the
Multiscale Science-Engineering Information Spaces associated to
models/methods and experimental, testing and sensing techniques and
procedures.
-
Alessandro Formica, March 2014 All rights reserved
21
Range of Validity is the range of the Multiscale
Science-Engineering Information Space inside which we can get a set
of information from specific models/methods and experimental,
testing and sensing procedures/techniques and possible coupling
schemes at a certain level of accuracy and reliability (uncertainty
quantification). It is of fundamental relevance to determine how
the Range of Validity changes as model, method, experimental &
testing and coupling scheme variables change. The range of validity
is a key element to determine (for a specific task) :
how good computational models and experimental, testing and
sensing techniques and coupling schemes should be to get
Information we think to be needed to carry out a task at a
predefined error and uncertainty level.
how to define the right mix of computational models/methods and
experimental & testing procedures/techniques and coupling
schemes to get what we think to be the right information at the
right level of accuracy and uncertainty to perform a specific
R&D and Engineering analysis and design task..
Fig. 5 (Center for Computational Materials Design NSF) describes
a framework to define in a formal way the Range of Validity (or
Applicability Domain) of a model
The Multiscale Science-Engineering Information Space formalizes
what, today, is being performed in an empirical and semi-empirical
way. Such a formal procedure allows us to rigorously evaluate the
relative weight of the several model/method/technique variables as
function of the Information Space and the best research/development
paths for computational models/methods and experimental &
testing techniques to address specific challenges.
The Multiscale Science-Engineering Information Space concept and
method enables researchers and designers to jointly define
development roadmaps for computational models and experimental,
testing and sensing techniques.
-
Alessandro Formica, March 2014 All rights reserved
22
The need of defining the Information Space associated to
computational method and experimental techniques, in the context of
the Verification & Validation process, has been analyzed, for
instance, by Tim Trucano in Uncertainty in Verification and
Validation: Recent Perspective Optimization and Uncertainty
Estimation, Sandia National Laboratories Albuquerque, NM 87185-0370
SIAM Conference on Computational Science and Engineering, February
12-15, 2005, Orlando, Florida - SAND2005-0945C.
Fig. 6 The figure (from the previously quoted document)
illustrates the Information Space concept
Thanks to the Multiscale Science Engineering Information Space
concept and method, it is possible to define Costs/Benefits
Function for models/methods and related coupling schemes as
referred to different Technology Development and Engineering tasks.
Benefits are referred to the Information get and Costs to the
resources needed to develop, validate and apply
models/methods/techniques/coupling schemes. This kind of Function
could be useful to Technology Development and Engineering Project
Managers to better manage and allocate human, organizational and
financial resources.
The Multiscale Science Engineering Information Space and the
Range of Validity concepts can be related with new Verification and
Validation (V&V) strategies and methods. Uncertainty
Quantification (UQ) is a key challenge for Computational Science
and Engineering. UQ and Quantification of Margin of Uncertainty
(QMU) [performance (measured) vs. requirements (set)] , are
becoming (have already become) one of the new driver and objective
for the Computational World. The Predictive Science Academic
Alliance Program (PSAAP) managed by US National Nuclear Security
Agency (NNSA) is a clear example of application of these
statements.
The Multiscale Science-Engineering Space is of fundamental
importance to define and implement Methodologically Integrated
Multiscale Science-Engineering Strategies which foresee the
coherent use of several different single and multiscale
computational models and methods, and several different single and
multiscale experimental, testing and sensing techniques working
over a full range of scales.
The Multiscale Science Engineering Information Space is becoming
of increasingly importance for Science and Engineering because for
a specific tasks is common using a spectrum of computational models
and a spectrum of experimental techniques and methods. Integration
calls for rigorous methodologies to determine what kind of
Information can be get from computations and what from
experimentation, testing and sensing.
-
Alessandro Formica, March 2014 All rights reserved
23
According to the previous analysis, the Multiscale
Science-Engineering Information Space concept and method is
instrumental to identify:
shortcomings and limitations of computational models/methods and
related multiscale multiphysics coupling schemes for specific
R&D and Engineering tasks
development lines (roadmaps) for computational models and
methods and multiscale coupling schemes to achieve specific R&D
and Engineering objectives
shortcomings and limitations and development lines (roadmaps)
for experimental, testing and sensing techniques and procedures and
related multiscale multiphysics coupling schemes
integrated roadmaps for jointly developing multiscale
multiphysics analytical, computational and (multiscale)
experimental, testing and sensing techniques to deal with specific
R&D and Engineering Tasks
integrated strategies for jointly applying multiphysics
multiscale analytical, computational and (multiscale) experimental,
testing and sensing techniques/procedures to deal with specific
R&D and Engineering Tasks
-
Alessandro Formica, March 2014 All rights reserved
24
3.4 The Information Driven Concept and Analysis Scheme
The relevance of Information, as a key element to shape R&D
and Engineering Strategies, is winning an increasing attention.
Several studies have been performed, for instance, by Jitesh H.
Panchal, Janet K. Allen, David L. McDowell and colleagues at
Georgia Institute of Technology. Alessandro Formica highlighted the
role of Information to drive modeling and simulation strategies in
the White Paper HPC and the Progress of Technology : Hopes, Hype
and Reality published in US by RCI Ltd on February, 1995. In this
document he discussed the concept of Engineering Information
Analysis. The issue was also dealt with in the context of the
Accelerated Insertion of Materials (AIM) Program (1999) managed by
US DARPA. The following text is drawn from DARPA Proposer
Information Pamphlet BAA 00-22 clearly describes the theme and
related challenges:
The need for an Information-Driven strategy . .There are many
interrelated technical challenges and issues that will need to be
addressed in order to successfully develop new approaches for
accelerated insertion. These include, but are not limited to, the
following: The construction of the designers knowledge base: What
information does the designer need and to what fidelity? How does
one coordinate models, simulations, and experiments to maximize
information content? What strategies does one use for design and
use of models, computations, and experiments to yield useful
information? How can redundancies in the data be used to assess
fidelity ? The development/use of models and simulation: What
models are required to be used and/or developed in the context of
the designer knowledge base? How can models of different time and
length scales be linked to each other and to experiments? How can
the errors associated with model assumptions and calculations be
quantified? How can models be used synergistically with
experimental data ?
The use of experiments: Are there new, more efficient
experimental approaches that can be used to accelerate the taking
of data? How can experiments be used synergistically with models?
How can legacy data and other existing data base sources be used
?
The mathematical representation of materials: How can one
develop a standardized mathematical language to: describe
fundamental materials phenomena and properties; formulate reliable,
robust models and computational strategies; bridge interfaces; and
identify gaps between models, theory and experimental materials
science and engineering? How can this representation be used to
develop hierarchical principles for averaging the results of models
or experiments while still capturing extremes ?
In the context of the Integrated Multiscale Science Engineering
Framework, Information is a key element which, to a large extent,
drives and shapes R&D and Engineering/Manufacturing
Strategies.
The term Information Driven means that R&D and
Engineering/Manufacturing strategies for specific Tasks have to
address what can be called The Information Challenge for R&D
and Engineering :
What Information at what level of accuracy and reliability
(uncertainty quantification) is needed to accomplish a task
What Relationships and Interdependencies between analysis and
design variables should be tracked over a full range (as needed) of
space and time scales to accomplish a task
What kind of information sources (analytical, computational,
experimental, testing and sensing models/techniques) are needed and
how they can be combined to get the previously identified
information
-
Alessandro Formica, March 2014 All rights reserved
25
Accordingly, the following key issues define the The Information
- Driven Analysis Scheme for R&D and
Engineering/Manufacturing:
Select a set of scales and resolution levels (the choice is not
unique and it is related to a specific Phase and Task)
identify physical phenomena, geometry and variables at the
different space and time scales which influence the dynamics of a
system at the reference scale at a certain level of accuracy and
fidelity (different scenarios for accuracy and fidelity can be
taken into account).
identify at a qualitative and quantitative level relationships
and interdependencies among phenomena, geometry, equations and
variables at the different scales
assess how and to what extent (qualitative and quantitative
evaluation) the capability of getting information thought to be
needed to describe the dynamics of the system at the reference
scale at a certain level of accuracy and fidelity is affected by
the spectrum of phenomena at the other scales.
assess how requirements defined at a scale determine and affect
requirements to the other scales The definition of how information
and requirements propagate in a qualitative and quantitative way
(in a deterministic and/or probabilistic fashion and taking into
account uncertainties) from a scale to another scale, from a
resolution level to another resolution level, is a key step to
:effectively deal with physics as well as with system and process
complexity
Assess what Information at what level of accuracy and
reliability is thought to be needed to accomplish a R&D and
Engineering task . Thought to be needed means that the process is
iterative, we start with some hypotheses and just Multiscale
Science Engineering Strategies and related Data, information and
Knowledge Analysis schemes and tools give us the possibility to
improve evaluation about the Information needed to execute the
task. Example : What Information (what physical and chemical
phenomena and processes related to materials, structures and
chemically reacting flows and their interactions) at what level of
accuracy and uncertainty should we know to analyze the dynamics of
a Thermal Protection Systems of an Hypersonic Vehicle for a
specific operational environment?
Evaluate what physical length scales and related physical and
biochemical phenomena rule the dynamics of the system under
analysis for a specific Tasks, what is the relative weight, what
are relationships and interdependencies between phenomena and
processes inside a scale and between different scales (to be
described thanks to Multiscale Maps).
Evaluate what Information at what level of accuracy and
reliability can existing analytical, computational models,
experimental, testing and sensing techniques and related coupling
scheme give us (to be described using the Multiscale Science
Engineering Information Space).
Assess what characteristics (Information Spaces) should new
models/techniques and related coupling schemes have
Assess what combination of old and new computational,
analytical, and experimental/testing methodologies at different
levels of scale and resolution do we need to get the right
information at the right level of accuracy and completeness for the
different tasks in the different R&D and design stages. A
critical step for the rational design of the R&D and
engineering processes is a proper selection, integration, and
sequencing of computational and analytical models and
experimental/testing methodologies with varying degrees of
complexity and resolution. To do that we have to define the
Science-Engineering Information Space associated to each
methodology.
Assess how good analytical and computational models,
experimental, testing and sensing techniques and related coupling
schemes should be to get the previously identified information
thought to be needed to accomplish a task. How good means
evaluating how much physical realism should be incorporated into
the models and what scales hierarchy has to be taken into account.
Not in all the cases, of course, we really need complex multiscale
methodologies going down to the Schrdinger equations: simple single
scale models can be accurate and reliable enough.
-
Alessandro Formica, March 2014 All rights reserved
26
Note: This kind of Information is critical to evaluate what new
analytical and computational models and what new experimental,
testing and sensing procedures/techniques should be developed and
integrated to deal with a specific analysis task. It is important
to identify not only what we know, but, in particular, what we do
not know, what we should know, how we should know it (what
combination of scientific and engineering methodologies and
technologies should be needed). In this context, the lack of
Knowledge becomes and important element to guide Strategies.
Furthermore, another very critical issue is that we need a
rational approach to link advances in the different methods at the
different scales with the new information we need to meet
challenges in the different tasks in the different stages of the
R&D and engineering process. How do we effectively and timely
evaluate the impact of scientific methodological and information
advances at an atomic, molecular, and grain (for materials) level
on new technological and engineering solutions if we do not have
conceptual and methodological (multiscale) frameworks to link
methods and information at the different scales: from atomic to
continuum? The Multiscale Science-Engineering Information Space can
represent a first step to deal with these critical issues. If we
like to shape new cooperative schemes between industry, from one
side, and academia and research, from the other side, we have to
define specific methodologies to evaluate the industrial and
technological value of new scientific methodological advances.
It is to be highlighted that this Analysis Scheme is adaptive
and iterative. It should be carried out at the starting time of any
R&D and Engineering/Manufacturing Phase and Task using
available data, information and knowledge and formulating
hypotheses: Results get during the execution of a Phase and related
Tasks will provide data, information and knowledge that allow to
update and improve the Analysis Scheme and initial Hypotheses Phase
after Phase, Task after Task.
The Information-driven approach is a fundamental element to
assess if, where, when and to what extent we have to go down along
the hierarchy of scales. Not in all the cases, of course, we should
go down until Schrodinger equations from the continuum. Dont Model
Bulldozers with quarks (Goldenfeld and Kadanoff, 1999)
-
Alessandro Formica, March 2014 All rights reserved
27
3.5 Multiscale Modeling and Simulation as Knowledge Integrators
and Multipliers and Unifying Paradigm for Scientific and
Engineering Methodologies and Knowledge Domains
The Vision of Multiscale Modeling & Simulation as Knowledge
Integrators and Multipliers (KIM) and Unifying Paradigm for
Scientific and Engineering Knowledge Domains and (Experimentation,
Testing and Sensing) Methodologies characterizes the Integrated
Multiscale Science-Engineering Framework and it represents the
conceptual context inside which the Framework is applied to R&D
and Engineering Processes. The KIM notion was presented by
Alessandro Formica in the: HPC and the Progress of Technology :
Hopes, Hype, and Reality RCI. Ltd Management White Paper February
1995
Multiscale Multiphysics Modeling and Simulation can be regarded
as Knowledge Integrators and Multipliers (KIM) and Unifying
Paradigm for Scientific and Engineering Knowledge Domains and
Methodologies because Multiscale Models are able to integrate and
synthesize, in a coherent framework, Data, information, and
Knowledge from:
a number of disciplines,
a wide range of scientific and engineering time and space
domains,
multiple scientific and engineering models (science-engineering
integration) linked by a spectrum of coupling schemes.
a wide spectrum of Computational, Experimentation, Testing and
Sensing Multiscale Science Engineering Data and Information Spaces
built during the development, validation, application and
improvement phases of the same Multiscale Models
several Maps generated by a wide range of methodologies
(analytical theories, computation, experimentation, testing and
sensing) during the development, validation, application and
improvement phases of the same Multiscale Models
In this context, we propose to extend the concept of Model to
include not only its mathematical formulation, but, also,
Information Spaces and Maps linked to it for specific tasks. We
also extend the concept of Model from the Computational to the
Experimental, Testing and Sensing World
This Vision give a New Dimension to the Virtual Engineering and
Manufacturing Concept and Strategy and Science Engineering
Integration Methodologies and Environments open the way to define a
New Field: Virtual Technology Innovation
Multiscale Information Spaces and Multiscale Maps embody and
organize Data, Information and Knowledge get by the full spectrum
of analytical theories, a set models at different scales and the
related experiments, tests and sensing measures used to develop,
validate and improve them. It is to be highlighted that all the
existing Modeling and Simulation concepts, application strategies
and methodologies, such as Virtual Prototyping , Simulation - Based
Design, Simulation - Based Acquisition, Simulation Based
Engineering Science (SBES) and Virtual Engineering, can be
considered as particular cases of this more general concept and
strategy.
We would like to emphasize that the KIM concept puts Multiscale
Modeling and Simulation and, accordingly, HPC, at the centre of the
R&D and Engineering/Manufacturing Processes much more than the
classical Virtual Engineering and Manufacturing and Simulation
Based Engineering Science concepts. Multiscale Modeling and
Simulation become a key element to shape complex (multi and single
scale) Experimental, Testing and Sensing Strategies.
The concept of Model as Knowledge Integrator is certainly not
new. This view, in the mid of nineties, was clearly described in
the chemical engineering field by James H. Krieger, in the article
Process Simulation Seen As Pivotal In Corporate Information Flow -
Chemical & Engineering News, March 27, 1995. The text reported
the following statement of Irving G. Snyder Jr., director of
process technology development, Dow Chemical : "The model
integrates the organization. It is the vehicle that conveys
knowledge from research all the way up to the business team, and it
becomes a tool for the business to explore different opportunities
and to convey the resulting needs to manufacturing, engineering,
and research." . In the same article other companies such as BNFL
and Du Pont expressed similar points of view.
-
Alessandro Formica, March 2014 All rights reserved
28
Note: Continuous advances in computational modeling and
computing power makes it possible to build computational models
which simulate the experimental or testing apparatus, the system to
be probed and related interactions. This kind of modeling is an
interesting asset to plan experimentation, testing and sensing and
analyze results.
Key element of the KIM Vision is the extension of the concept of
Model to the Experimental, Testing and Sensing World as detailed in
the following:
The Concept of Experimental, Testing and Sensing Model
In the proposed theoretical and methodological framework it is
necessary to extend the concept of Model from the Computational to
the Experimental, Testing and Sensing World. In the context of the
Experimental, Testing and Sensing World, for Model, as referred to
a specific Experimental, Testing, Sensing activity carried out with
specific techniques, working in a specific operational mode and
probing a specific system for a specific task, we mean an
Information and Knowledge Structure that define:
Characteristics (structure, composition, initial dynamics state,
boundary conditions, external loadings) of the System to be
probed
Characteristics of the equipment in terms of resolution, scale,
physical and biochemical phenomena which can be probed
Characteristics of the specific Experimental, Testing and
Sensing operational conditions and modes applied for specific
R&D and Engineering Tasks
The Multiscale Science Engineering Information Space related to
it
Multiscale Physics Maps .
As in the Computational World, it is easy to define the concept
of Multiscale Experimental, Testing and Sensing Model. In this case
the Information/Knowledge Structure refers to a cluster of
different equipments and it embodies information about:
Interaction schemes among the different equipments
Data and Information Flow among the different equipments
Multiscale Computational Modeling and Multiscale Experimentation
Integration Materials Research Society Bulletin
An important recognition of the key strategic relevance of the
development of multiscale experimental techniques and their
integration with multiscale computational modeling comes from the
article Three-Dimensional Materials Science: An Intersection of
Three-Dimensional Reconstructions and Simulations (Katsuyo Thornton
and Henning Friis Poulsen, Guest Editors), published in the
Materials Research Society (MRS) Bulletin June 2008.
..For example, by combining a nondestructive experimental
technique such as 3D x-ray imaging on a coarse scale, FIB-based 3D
reconstruction on a finer scale, and 3D atom probe microscopy at an
even finer scale, one has an opportunity to capture materials
phenomena over six orders of magnitude in length scale. This will
bring materials researchers closer to the ultimate dream of a
direct validation of multiscale models, both component by component
and ultimately as an integrated simulation tool. In conjunction
with the advances on the modeling side, such comprehensive
experimental information is seen as very promising for establishing
a new generation of models in materials science based on first
principles..
-
Alessandro Formica, March 2014 All rights reserved
29
Even if attention to the integration issue is positively
increasing, particularly for models development and verification
and validation phases, there are still conceptual and
methodological relationships not thoroughly examined between
challenges and advances in modeling and simulation, and progress
and challenges in experimental, testing and sensing techniques.
Experience is showing us that ever more complex and large scale
computations call for increasingly sophisticated and expensive
experimental/testing/sensing techniques both in the model
development, validation and improvement phases. Advances in
modeling and simulation are intimately linked to progress in
experimental, testing and sensing methods and techniques and vice
versa. A direct correlation and strong mutual dependencies, in the
model development, validation and improvement phases, exist between
the two fields sometimes regarded as antithetic. It is important to
take into account that, if computational methods and computing
technologies are continuously progressing, also experimental,
testing and sensing techniques are making continuous significant
progress.
It is sufficient to think at the impact on materials research
that the Scanning Tunneling Microscopy (STM) and Atomic Force
Microcopy (AFM) techniques have had.
It is advisable to consider a joined development of new
Computational Methods and Strategies with new Experimentation,
Testing and Sensing Development Techniques and Strategies and vice
versa.
Furthermore more and more complex and powerful 3D and 4D
experimental, testing and sensing techniques increasingly call for
complex computational models to interpret, analyse and organize
data and define integrated measurement and characterization
strategies. A priority target is to develop a unified conceptual
context to synergistically take advantage of advances in both the
fields and not only for the computational models development and
validation phases, as it occurs today, but, also, in the
application phase. All of that in the context of Integrated
Frameworks and Strategies An effective R&D and Engineering
Strategy should find the way to synergistically take advantage of
advances in both the fields. In several cases, today, advanced
HPC/Modeling/Simulation and experimental/testing/sensing programs
are conceived and managed as separated realities. This situation
can lead to costs increase and hamper and limit the effectiveness
of both the programs. The new Vision reconcile development streams
and roadmaps in the two fields.
In the R&D and Engineering Process, today and, more and
more, in the future, we have to integrate a full spectrum of
(interdependent and interlinked) scientific and engineering models
and codes with a wide spectrum of experimental, testing and sensing
(scientific and engineering) data with a full spectrum of
scientific and engineering analytical formulations. Data get from
experimentation, testing and sensing covers several physical and
biochemical disciplines and domains and several different space and
time scales. It is clear that, increasingly, we have to deal with
very complex interaction patterns intra the experimentation,
testing and sensing world, intra the computational modeling world
and inter the experimentation, testing, sensing and computational
modeling worlds. Multiscale Science Engineering Information Spaces,
Multiscale Maps and the Kim vision can be a first step to realize
this integration. The KIM concept is a fundamental theoretical and
methodological basis. Methodologically Integrated Multiscale
Science - Engineering Strategies are built upon it. Classical
Modeling & Simulation Application Strategies in the innovative
technology development field are significantly hampered and limited
the following fundamental contradiction: when we develop innovative
technologies and innovative engineering solutions, we often enter a
territory where theories are not well developed and reliable, and
the availability of experimental and testing data is fragmented or
lacking at all. Accordingly, we face a fundamental and intrinsic
problem: Modeling & Simulation is the reference strategy to
limit risks, costs, and development times by heavily reducing the
resort to complex and expensive experimental and testing
activities. However, contrary to what happens in the mature or
evolutionary technology environment, we cannot adopt this strategy
because we still need very significant experimental and testing
activities to develop and validate the needed computational models.
That is what is called a classical Catch 22 situation: (i.e.) a
situation which involves intrinsic contradictions.
-
Alessandro Formica, March 2014 All rights reserved
30
This contradiction is certainly not ignored. In the presentation
Modeling and Simulation in the F-22 Progr