paradigm shift in complex system design

8/10/2019 Paradigm Shift in Complex System Design

1/95

Dr. Dimitri N. Mavris, Director ASDLDr. Dimitri N. Mavris, Director ASDLSchool of Aerospace EngineeringSchool of Aerospace EngineeringGeorgia Institute of TechnologyGeorgia Institute of TechnologyAtlanta, GA 30332Atlanta, GA [email protected]@ae.gatech.edu

A Paradigm Shift In Complex System DesignA Paradigm Shift In Complex System Design

Enabling Technologies for Strategic Decision Making ofAdvanced Design Concepts

By

Prof. Dimitri Mavris

Director

Aerospace Systems Design Laboratory

General Electric University Strategic Alliance

Boeing Professor in Advanced Aerospace Systems AnalysisSchool of Aerospace Engineering

Georgia Institute of Technology
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


2/95


Quality Issues to be AddressedQuality Issues to be Addressed

Successful Utilization of Concurrent Engineering (CE) ApproachesSuccessful Utilization of Concurrent Engineering (CE) Approaches

by the Japanese Automotive Manufacturersby the Japanese Automotive Manufacturers

N

umberofEnginee

ringProduct

ChangesProcessed

20-24

Months

14-17

Months

1-3

Months

Job#1

+3

Months

U.S. Company

Japanese Company

90%

Total Japanese

Changes Complete
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


3/95

Dr. Dimitri N. Mavris, Director ASDLDr. Dimitri N. Mavris, Director ASDL

School of Aerospace EngineeringSchool of Aerospace EngineeringGeorgia Institute of TechnologyGeorgia Institute of TechnologyAtlanta, GA 30332Atlanta, GA [email protected]@ae.gatech.edu

Motivation for PhysicsMotivation for Physics--based Conceptual Designbased Conceptual Design

Subsonic Transports

Supersonic Aircraft

Personal Air Vehicles

Uninhabited Air Vehicles

Rotorcraft

New Generation of Vehicles can

not be modeled accurately in the

absence of historical dataExtreme STOL


4/95



Traditional Development ProcessTraditional Development Process

Advanced Design Project Design

Requirements

Conceptual

Design

Conceptual

Baseline

Preliminary

Baseline

Allocated

Baseline

Detailed

Design

Production

Baseline

Production &

Support

Optimization Parametric

1stLevel Analysis

General Arrangement/Performance

Representative Configurations

General Internal Layout

System Specifications

Detailed Subsystems

Internal Arrangements

Process Design

Sophisticated Analysis

Problem Decomposition

Multidisciplinary Optimization

Problems with not foreseeing design flaws

Cannot rely on historical data

Communication between manufacturing and engineers is poor


5/95



Design StagesDesign Stages

Requirements Definitionunderstanding the

requirements posed by the customer/market

Conceptual Designinitial formulation,interpretation based on experience/background

knowledge

Preliminary Designtransforming the concept sothat the product will work and/or make money

Detailed Designtesting and fine-tuning


6/95



Uneven Distribution of Knowledge EffectsUneven Distribution of Knowledge Effects

100%

Conceptual DetailedPreliminary

100%100%

1 1 1 1. Aerodynamics

2 2 2 2. Propulsion

3 3 3 3. Structures

4 4 4 4. Controls

5 5 5 5. Manufacturing

6 6 6 6. Supportability

7 7 7 7. Cost

Time into the Design Process

DesignFreedom

Knowledge

about design


7/95



Traditional, PointTraditional, Point--Design PhilosophyDesign Philosophy

May be characterized as a manual, deterministic, data driven, serial or

parallel, disciplinary-centric, point design process

Design requirements, and technology assumptions are usually fixed

and a design space exploration is performed around one or a handful

of concepts (point solutions)

As organizations strive to decrease costs and reduce operational

overhead, the number of personnel available for given activities isdecreasing

At the same time, the demands on the organization for more in depth

analysis at the conceptual and preliminary stages is increasing

As a result, a paradigm shift is required to reduce design cycle time,allow for more iterations, and increase fidelity

Traditional organizations can be supported and enhanced by several

enabling technologies, to be presented here, that allow for this

transformation to take place in a practical fashion


8/95



Acquisition ProcessAcquisition Process

Short concept design phase with unequal distribution of

disciplines does not allow use of design freedom toimprove quality and integrate disciplines for

optimization

Uneven distribution of knowledgeand effort

Need better representation of all disciplines in earlier stages

(conceptual, preliminary)

If data is in the historical database, it is pointless to use

physics based analysis uses too many assumptions


9/95



Phases in Acquisition ProcessPhases in Acquisition Process

Pre-Milestone 0

Determination of Mission need and deficiencies

Phase 0

Concept exploration

Phase I Program definition and risk evaluation

Phase II

Engineering and manufacturing development

Phase III

Production, development, and operations support


10/95



AffordabilityAffordability -- Making the Right Decisions EarlyMaking the Right Decisions Early

Pre-milestone0

Phase0

PhaseI

Determination ofMission Need and

IdentificationDeficiencies

Engineering &ManufacturingDevelopment

Production,Deployment, and

OperationalSupport

Approval to

Begin a New

Acquisition

Program

Approval to

Enter

Engineering

and

Manufacturing

Development

Production or

Deployment

Approval

Milestone 0 Milestone I Milestone II Milestone III

AoA I AoA II AoA III

LRIPApproval

Program Initiation

Acquisition Timeline

Cost Committed

Actual Cost

Expenditure

PhaseII

ConceptExploration

Program Definitionand Risk

Reduction

PhaseIII

Approval to

Conduct

Concept

Studies

DecisionDecision--Makers Need New MethodsMakers Need New Methods

to Make the Right Trades !!to Make the Right Trades !!

Emphasis of Affordability InitiativeEmphasis of Affordability Initiative

$$


11/95



Capability based/Affordability Paradigm ShiftCapability based/Affordability Paradigm Shift

A paradigm shift is underway that challenges the manner in

which complex systems are being designed Emphasis has shifted from design for performance to design

for affordability to design for overall capability

There is a need for a multidisciplinary approach to the problem

based on more sophisticated, higher fidelity tools There is a need for forecasting the economic viability of a

system with a high probability of success

Long-term goal: Creation of a virtual engineering environment

for simulation-based acquisition

Academia is reacting to this paradigm shift and is trying to change its

own culture in an attempt to meet future research needs and take

advantage of new funding opportunities


12/95



The Affordability

Balance

Definition of Affordability in Our ContextDefinition of Affordability in Our Context

O & S Cost

Survivability

Safety

Acquisition CostCapability

Availability

Maneuverability RDTE Cost

essEffectivenThisAchievetoInvestment

essEffectivenSystemWeaponROIT&S

Effectiveness = k1(Capability)+ k2(Survivability)+ k3(Readiness)+ k4(Dependability)

+ k5(Life Cycle Cost)

Affordability: The balance of benefits provided or gained from the system

to the cost of achieving those benefits. In a probabilistic, Modeling &

Simulation approach, Risk is inherent in these estimates.

Weapon System Effectiveness- Aircraft Example

Acquisition cost

Operation cost

Maintenance cost

Aircraft re placement

Crew replacement

training

RDT&E Cost

Operational Effectiveness

Performance

Maneuverability

Satisfying mission

requirements

Capability Dependability

Maintainability

Inherent availability

Reliability

Logistics support

Readiness

Susceptibility

Vulnerability

Survivability

defects

Cost

Reliability

Maintenance

Design defects

Operations

Safety

Lethality

CDF

CDF

. . .

. . .

Weapon System Effectiveness

Investment to Achieve this Effectiveness
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


13/95



PhysicsPhysics--based Conceptual Designbased Conceptual Design -- A Paradigm ShiftA Paradigm Shift

Design Freedom

0 %

100 %

RequirementsDefinition

Detail DesignPreliminaryDesign

ConceptualDesign

+ Manufacturing

Pre-milestone 0 Phase 0 Phase I

Determination ofMission Need and

Deficiencies

Engineering &Manufacturing

Development

Production,Deployment, and

Operation Support

Phase II

Concept

Exploration

Program Definitionand Risk

Reduction

Phase III

Knowledge

Acquisition TimelineAcquisition Timeline

Design TimelineDesign Timeline

Today

Future

Knowledge

becomes available

when time to make

decisionCost Committed

Mavris, D.N., DeLaurentis, D.A., Bandte, O., Hale, M.A., "A Stochastic Approach to Multi-disciplinary Aircraft Analysis and Design", AIAA 98-0912.

A B i f Whi h t B iA B i f Whi h t B i
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.gatech.edu/http://www.gatech.edu/http://www.gatech.edu/http://www.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


14/95



A Basis from Which to Begin:A Basis from Which to Begin:Generic IPPD DecisionGeneric IPPD Decision--Making ProcessMaking Process

COMPUTER-INTEGRATED ENVIRONMENT

PRODUCT

DESIGN

DRIVEN

PROCESSDESIGN

DRIVEN

REQUIREMENTS

& FUN CTIONAL

ANALYSIS

SYSTEM DECOMPOSITION

&

FUNCTIONAL ALLOCATION

SYSTEM SYNTHESISTHROUGH MDO

SYSTEM ANALYSIS&

CONTROL

ESTABLISH

THE NEED

DEFINE THE PROBLEM

ESTABLISHVALUE

GENERATE FEASIBLEALTERNATIVES

EVALUATE

ALTERNATIVE

7 M&P TOOLS AND

QUALITY FUNCTION

DEPLOYMENT (QFD)

ROBUST DESIGN

ASSESSMENT &

OPTIMIZATION

ON-LINE QUALITYENGINEERING &

STATISTICALPROCESS

MAKE DECISION

SYSTEMSENGINEERING METHODS

QUALITYENGINEERING METHODS

TOP-DOWN DESIGNDECISION SUPPORT PROCESS

Schrage, D.P., Mavris, D.N., "Technology for Affordability - How to Define, Measure, Evaluate, and Implement It?",

50th National Forum of the American Helicopter Society, Washington, D.C., May 11-13, 1994.
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.gatech.edu/mailto:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


15/95



What is needed for the Paradigm ShiftWhat is needed for the Paradigm Shiftto occurto occur??

Transition from single-discipline to multi-disciplinary analysis,design and optimization

Automation of the resultant integrated design process

Transition from a reliance on historical data to physics-basedformulations, especially true for unconventional concepts

Means to perform requirements exploration, technology infusion

trade-offs and concept down selections during the early designphases (conceptual design) using physics-based methods

Methods which will allow us to move from deterministic, serial,single-point designs to dynamic parametric trade environments

Incorporation of probabilistic methods to quantify, assess risk Transition from single-objective to multi-objective optimization

Need to speed up computation to allow for the inclusion ofvariable fidelity tools so as to improve accuracy


16/95



Elements needed to enable this Paradigm ShiftElements needed to enable this Paradigm Shift

Advances in MDA/MDO methods and techniques to encompass theholistic nature of the problem, emphasis on uncertainty associated with the

early design phases Creation of computational architecture frameworks to allow for easyintegration and automation of sometimes organizationally dispersed tools

Emergence of commercially available frameworks will further expeditethe usefulness of the proposed approaches

Creation of physics-based approximation models (surrogate or meta-

models) to replace the higher fidelity tools which are usually described astoo slow for use in the design process, cryptic in their use of inputs,interfaces and logic, and non-transparent (lack of proper documentation,legacy)

Use of probability theory in conjunction with these meta-models willenable us to quantify, assess risk and to explore huge combinatorial spaces

In fact it will enable us to uncover trends, solutions never before examinedin a very transparent, visual, interactive manner

Use of Multi-attribute decision making techniques, pareto optimality,genetic algorithms to account for multiple, conflicting objectives and fordiscrete settings


17/95



Varying Fidelity M&S InitiativeVarying Fidelity M&S Initiative

Aerodynamics Economics

Propulsion

Safety

Aerodynamics

Structures

Propulsion

Performance

Manufacturing

Economics

Safety

S & C

ManufacturingStructures

S & C Performance

Conceptual Design Tools(First-Order Methods)

Synthesis & Sizing

Preliminary Design Tools(Higher-Order Methods)

Geometry

Mission

IncreasingSophistication and

Complexity

Approximating FunctionsDirect Coupling of Analyses

Integrated Routines

Table Lookup
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


18/95



Key EnablerKey EnablerSurrogate ModelsSurrogate Models

Reliance on meta-models or surrogate models as a means to:

speed up processes,

protect proprietary nature of codes used,

overcome organizational barriers (protectionism of tools and data),

allow for the framework to be tool independent (no need for direct

integrations of codes; also enables our desire for variable tool fidelity

formulations), allow the designer to perform requirements exploration, technology

infusion trade-offs, and concept down selections during the early design

phases (conceptual design) using physics-based methods

Surrogate models can also be used at the integrated system level

to determine responses at that level. This will allow us to movefrom deterministic, serial, single-point designs to dynamic

parametric trade environments.


19/95



RSM is a multivariate regression technique developed to model the

response of a complex system using a simplified equation Regression data is obtained intelligently through the Design of

Experiments (DoE) techniques

RSM is based on the design of experiments methodology which gives

the maximum power for a given amount of experimental effort

Typically, the response is modeled using a second-order quadratic

equation of the form:

Where,biare regression coefficients for the first degree termsbiiare coefficients for the pure quadratic termsbijare the coefficients for the cross-product terms

ji

k

i

k

ijiji

k

iiii

k

iio xxbxbxbbR

1

1 1

2

11

R

Response Surface Methodology (RSM)Response Surface Methodology (RSM)
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.gatech.edu/http://www.gatech.edu/http://www.gatech.edu/mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


20/95



Design ofExperiments

For 7Variables

For 12Variables

Equation

Full Factorial 2,187 531,441 3n

CentralComposite

143 4,121 2n+2n+1

Box-Behnken 62 2,187 -D-Optimal

Design36 91 (n+1)(n+2)/2

Factors

Run X1 X2 X3 Response

1 -1 -1 -1 y12 +1 -1 -1 y23 -1 +1 -1 y3

4 +1 +1 -1 y45 -1 -1 +1 y56 +1 -1 +1 y67 -1 +1 +1 y78 +1 +1 +1 y8

Design of Experiments
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.gatech.edu/mailto:[email protected]://www.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


21/95



Enabling Tools and TechniquesEnabling Tools and Techniques

Established Techniques

Response Surface Method (Biology; Ops Research)

Design of Experiments (Agriculture, Manuf.)

Quality Function Deployment, Pugh Diagram (Automotive)

Morphological Matrix (Forecasting)

MADM techniques (U.S Army, DoD)

Uncertainty/Risk Analysis (Control Theory; Finance)

Technology Readiness Levels (NASA)

ASDL Innovation

Feasibility/Viability Identification

Robust Design Simulation (RDS)

Technology Identification, Evaluation, Selection (TIES)

Joint Probabilistic Decision Making (JPDM)

Unified Trade-off Environment (UTE)

Virtual Integrated Stochastic System Technology

Assessment (VISSTA)


22/95



21420

Thrust

(lbs.)

14535O&S

TOGW

TOWOD

Vapp

Turn Radius

Turn Rate

Ps

AlternateRange

380 Area (ft^2) 520

Point Design forA notional Concept

Point Design Identifies a Single, Feasible DesignPoint Design Identifies a Single, Feasible Design

A point design is a single point on the thrust/wing area plot

This point will not satisfy evolvingmission requirements

A parametric design environment would allow movementaround this space

Constraints could also be changed in real time and theimpact on the design could be assessed


23/95



Integrated Design: Reduction in Cycle Time Through AutomationIntegrated Design: Reduction in Cycle Time Through Automation

Performing an integrated design involves linking conceptual andpreliminary design tools in a computational environment that

automatically passes information between design codes Enablers:

Computational environment such as ModelCenter or iSIGHT

Design codes with simple inputs/outputs without hard coding of designvariables or internal optimizations that may skew results

Integrated design provides tremendous advantages in designcycle time by eliminating the re-keying of information fromoutput files to input files.

The next slide shows a missile design environment. As an

integrated suite of codes, it takes 35 seconds to perform adesign. If the codes were not linked, it would take approximately45 minutesto pass the information back and forth and check forerrors!

Example: Integrated Missile Design Tool in theExample: Integrated Missile Design Tool in the ModelCenterModelCenter EnvironmentEnvironment


24/95



Example: Integrated Missile Design Tool in theExample: Integrated Missile Design Tool in the ModelCenterModelCenter EnvironmentEnvironment

Aero

Trajectory

Weights/Sizing

Propulsion

Plume

OPS

Cost

Reliab/

Safety

Design Variables

Linked Computer Codes


25/95



Varying Fidelity M&S InitiativeVarying Fidelity M&S Initiative

AerodynamicsEconomics

Propulsion

Safety

Aerodynamics

Structures

Propulsion

Performance

Manufacturing

Economics

Safety

S & C

ManufacturingStructures

S & C Performance

Conceptual Design Tools(First-Order Methods)

Synthesis & Sizing

Preliminary Design Tools(Higher-Order Methods)

Geometry

Mission

IncreasingSophistication and

Complexity

Approximating FunctionsDirect Coupling of Analyses

Integrated Routines

Table Lookup
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


26/95



PhysicsPhysics--Based Modeling and Simulation EnvironmentBased Modeling and Simulation Environment

Objectives: Attribute 1

(e.g. Cost)

Attribute 2(e.g. Performance)

Attribute 3

. . .

Customer

Satisfaction

Design & EnvironmentalConstraints

Synthesis

& Sizing

Technology

Infusion

Physics-

Based

Modeling

Activity and

Process-

Based

Modeling

Economic

Life-CycleAnalysis

Subject to

Economic &

DisciplineUncertainties

Impact of New

Technologies-Performance &

Schedule Risk

Decision Making

(MADM)

Robust Design Simulation

Simulation

Operational

Environment

VIRTUAL INTEGRATED STOCHASTIC SYSTEM AND TECHNOLOGY ASSESSMENT (VISSTA) ENVIRONMENT


27/95



System Level Objectives

Systems Engineering Methods

Simplified Analysis

Historical-Based

Current

Module Integration

Proposed

Module Integration

Simulation

Environment

Sizing

Synthesis

Physics-based

Simulation

Variability Reduced Variability

Transparent,

Seamless

Integration

Stability &

Controls

Integration Methodology

Risk/Benefit

Analysis

Environmental,

Operational

Maintenance

Model

Technology

Readiness/

Risk Library

Probabilistic

Assessment

Uncertainty

Probability

Fuzzy Logic

Distributions

Constraints

Decision Support

Quality Engineering MethodsComputer Integrated Environment

ManufacturingRe- Manufacturing

Fluid Mechanics

Safety

Propulsion

Subsystems

Solid Mechanics

Economics

NeuralNetworks

Fuzzy

Logic

ResponseSurfaces

Knowledge-BasedSystems

Agents

ExpertSystems

Activity-Based Costing

Process-Based Models

Virtual Manufacturing

Probabilistic FEM

Virtual Wind Tunnel

Flight Simulation

Virtual Operation

Environment

VIPER-CAT

Integration Environment

Parametric DefinitionGeometry

Design Guidance

Knowledge

Based System

Decision Making

Processes

Constrained

Probabilistic

Optimization

Product Family

Design,

Enterprise Design

Comprehensive LifeCycle Customer

Requirements

Process

Product(Physics-Based)

FidelityUncertainty

Numerical

Optimization

(MDO)

VIRTUAL INTEGRATED STOCHASTIC SYSTEM AND TECHNOLOGY ASSESSMENT (VISSTA) ENVIRONMENT
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/mailto:[email protected]:[email protected]://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


28/95



Why Do We Need a CapabilityWhy Do We Need a Capability--Based Design Approach?Based Design Approach?

Noting that schedule slips have become ubiquitous in the acquisition of

complex systems, the Air Force is pursuing techniques which will facilitate

accelerated acquisition (also known as agile acquisition.) Theparadigm shiftin systems design advocates moving knowledge forward.

We now want to move the ability to examine capabilitiesto the conceptual

design phase

Assists future military planners

Identifies solutions which may be non-optimal in and of themselves, butmaximize a macro-level performance function

Improve interoperability of weapons systems and platforms through more

rigorous interoperability evaluation in a replicated battlefield environment

Identify technologies for systems and subsystems in the presence of changing

requirements and evolving threats

Facilitate Shift to Capability-Based Acquisition and Planning


29/95



Enabling CapabilityEnabling Capability--Based DesignBased Design

There is an overall desire to select systems and architectures based on theiroverall capability

Because these architectures rely on multiple, interoperable, heterogeneoussystems, an integrated design environmentis needed

Collaboration is required because an architecture is comprised of differentelements belonging to various entities

To perform trade studies between requirements, design criteria, and technologies,rapid parametric analysis capabilities are needed

Collaboration is hindered by competition and intellectual property issues

Surrogate Modelsmay be viewed as an enabler for capability-based design

If processes can be sped up to the point where they are not a computationalburden, the mapping of capabilities to candidate designs is trivial

An integrated, parametric modeling and simulation environment facilitates

bottom-up trade studies Probabilistics, coupled with surrogate models, enables large-scale design

studies where top-level capabilities can be mapped to systems and anyvariable can be treated as an independent variable


30/95



Collaborative Design Aided by Surrogate ModelsCollaborative Design Aided by Surrogate Models

IT issues and intellectual property concerns frequently limit collaborative activities

Surrogate models can be traded as a currency for exchanging information Generated using the tools specific to a collaborative partner

Proprietary concerns are mitigated since the surrogates are made for a specific problem(cannot be reverse engineered)

Brings the disciplinary experts into the conceptual design process as they generate thesurrogates

Equations are not operating system or platform-specific

Shields Intellectual Property

Provide intelligence to assets in an agent-based framework

www.phoenix-int.com

Surrogate Models

Integrated Suite of Tools Multi-Site Collaboration

www.engineous.com

i i i i f C i ii i i i f C i i ii


31/95



Assumptions

Establish heuristics, behaviors,

and actions for assets

Map heuristics, behaviors, and

actions to the environment

Specify tacticsExecution of actions to fulfill doctrine

DoctrineGuiding principles for actions

Scenario Modeling AssumptionsPhysical Assumptions

Political ClimateFriend or Foe, no-fly zones

GeographyRange where?

Basing Options

Deployment Status

Asset Allocation

Design Team

Scenarios/Missions/Threats

Simulation method

(force/force or one/one)

TechnologiesTechnologies

RequirementsRequirements

Asset FamiliesAsset Families

System-of-Systems-Level

Requirements Design Vars Technologies

Responses

Metrics/Objectives

Constraints

MoPs

System Level (Missile)


MoPs

Subsystem Level (Propulsion)MoPs

Subsystem Level (Sensors)MoPs

Subsystem Level (Avionics)

Requirements Design Vars Technologies Requirements Design Vars Technologies Requirements Design Vars Technologies

MoE

s

MoEs

MoPs ofvehicle become variables for next level

MoEsbecome MoPs

Campaign LevelCampaign LevelWarfighterViewWarfighterView

Mission LevelMission LevelEngagementModelEngagementModel

Responses

Metrics/Objectives

Constraints

MoPs

System Level (Platform)



Responses

Metrics/Objectives

Constraints

Responses

Metrics/Objectives

Constraints

MoPs

System Level (Missile)System Level (Missile)


MoPs

MoPs


MoPs


MoPs



MoE

s

MoEs

MoPs ofvehicle become variables for next level

MoEsbecome MoPs

Campaign LevelCampaign LevelWarfighterViewWarfighterView

Mission LevelMission LevelEngagementModelEngagementModel

Responses

Metrics/Objectives

Constraints

MoPs

System Level (Platform)Responses

Metrics/Objectives

Constraints

Responses

Metrics/Objectives

Constraints

MoPs


Capability Options

Strategic Challenges

System Level

Weapons and Platforms

Many Heterogeneous Assets Interoperating

Subsystem Level

Propulsion, Avionics, Structures

Technologies and Design Variables

System-of-Systems Level

Campaign/Theater/

Engagement Analysis

Realizing the Vision for CapabilityRealizing the Vision for Capability--Based DesignBased Design

Capability Based DesignCapability Based Design System of Systems AffordabilitySystem of Systems Affordability
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.gatech.edu/http://www.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/http://www.gatech.edu/http://www.gatech.edu/http://www.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/mailto:[email protected]://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


32/95



Capability Based DesignCapability Based Design -- System of Systems AffordabilitySystem of Systems Affordability

EconomicEconomicSecuritySecurity

NationalNationalSecuritySecurity

L / D SFC IR/RCS DOC/SortieEW

National

Level

Campaign

Level

Asset

Level

Attributes maneuverability speed payload $ RDTE $ O&S range susceptibility

Discipline

Level

Technologies

System

EffectivenessDependability Survivability Capability Lethality Total Own. Cost

Requirements

Doctrine

Missions

Needs

Probabilistic

Matching

Systems

Capabilities

S&T $

S i iS i i S iS i


33/95



Surrogate Modeling Enables MultiSurrogate Modeling Enables Multi--Level Trade StudiesLevel Trade Studies



Responses

Metrics/Objectives

Constraints

MoPs

System Level (Missile)


MoPs





MoEs

M

oEs

MoPs of vehicle become variables for next level

MoEs become MoPs

Campaign LevelCampaign LevelWarfighter ViewWarfighter View

Mission LevelMission LevelEngagement ModelEngagement Model

Responses

Metrics/Objectives

Constraints

MoPs


Environment

allows flow-upand flow-downEnabler to performtrades between

dissimilar systems

(eg: satellites vs.

stealth UAVs) with

MoEs at multiple

levels

P t i D i U i I t t d D i LP t i D i U i I t t d D i L S lS l


34/95



Parametric Design: Using an Integrated Design on a LargeParametric Design: Using an Integrated Design on a Large--ScaleScale

The integrated design environment is an enabler for a parametric

design study

Instead of passing in a series of input variables, a parametric

design can take a distributionof inputs.

In this manner, an entire design space can be explored, rather

than small perturbations around a single point design

Large design spaces may take too long to explore by traditional

means

The integrated design environment above can be used to generate

metamodels of the design process

These metamodels, custom made for a given range of inputs, can beevaluated in a spreadsheet hundreds of times per second

Metamodels represent another order of magnitude in reduction for design

cycle time.

Problem Definition:Problem Definition:1 2 3 4 5 6 7 8


35/95



Problem Definition:Problem Definition:HSCT conceptHSCT concept

50,000 ft.

1. Taxi & T.O.F.L.=11,000 ft.

3. CruiseM=0.9

8. Abort3000 ft.

10. LandF.L.= 11,000 ft.

7. LoiterM=0.6

9. ReserveM=0.6

2. Climb

67,000 ft.

35,000 ft.

4. Climb

5. CruiseM=2.4

6. Descent

200 nm100 nm750 nm50 nm

5,000 nm

Societal Need:

Next generation supersonic aircraft

Increased commercial traffic growth

Increased comfort, safety, and affordability

Potential concept:High Speed Civil Transport*

Definethe

Problem

DefineConceptSpace

Modelingand

Simulation

InvestigateDesignSpace

Feasibleor

Viable?

IdentifyTechnologies

EvaluateTechnologies

SelectTechnologies

* Potential concept is actuallyestablished in the following step

Define Concept Space:Define Concept Space:1 2 3 4 5 6 7 8


36/95



p pp pMorphological MatrixMorphological Matrix

Definethe

Problem

DefineConceptSpace

Modelingand

Simulation


Feasibleor

Viable?



SelectTechnologies

Config

Mission

P

ropulsion

Aero

Struct

Alternatives

Characteristics 1 2 3 4

Vehicle Wing & Tail Wing & Canard Wing, Tail &

Canard Wing

Fuselage Cylindrical Area Ruled Oval

Pilot Visibility Synthetic Vision Conventional

Conventional &

Nose DroopRange (nmi) 5000 6000 6500

Passengers 250 300 320

Mach Number 2 2.2 2.4 2.7

Type MFTF Turbine Bypass Mid Tandem

Fan Flade

Materials Conventional High T CompCombustor Conventional RQL LPP

Nozzle Conventional Internal

Flow Alteration Mixed Ejector Mixer Ejector &

Acoustic Liner

Low Speed Conventional

FlapsConventionalFlaps & Slots

C C

High Speed Conventional NLFC Active Control HLFC

Materials Aluminum Titanium High Temp.

Composite

Process IntegrallyStiffened

SpanwiseStiffened

Monocoque Hybrid

Purpose: Establish the concept space that may fulfill the customer requirements and establish a

datum point for the feasibility investigation

Performed with the aid of the Morphological Matrix technique

Procedure:Define Alternatives Space

Functionally decompose the existing

system into contributing

characteristics

For each characteristic, list all the

possible ways in which it might be

satisfied

Select a datum point; permutations

are concept alternatives

Define Design Space Further decompose the system from

the Alternatives Space to elementary

attributes, such as geometric and

propulsive characteristics

E l f P t i D i E i f S i B i J tE l f P t i D i E i f S i B i J t
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


37/95



Example of a Parametric Design Exercise for a Supersonic Business JetExample of a Parametric Design Exercise for a Supersonic Business Jet

E l f P t i D i E i f S i B i J tE l f P t i D i E i f S i B i J t


38/95



Example of a Parametric Design Exercise for a Supersonic Business JetExample of a Parametric Design Exercise for a Supersonic Business Jet

Each aircraft to the left is anexample of a complete design.

Parametric design provides the

user with the power to testhundreds or thousands of designs,where previously, time permitteda single design point only.

Each aircraft to the left has A complete analysis of the

propulsion system

An aerodynamic analysis tocalculate accurate drag polars

They have all been sized for themission requirements, which areALSO parametrically scalable. Achange in desired range will re-generate this matrix of designs.

The creation of a single one ofthese aircraft designs can take lessthan a minuteor up to a day,depending on the desired fidelityof the design tools.

Man in the loop Genetic Algorithm


39/95



Man in the loop Genetic Algorithm

Sonic Boom Profiles for Various SBJ Configurations

Conventional Baseline Swept Configuration Highly Swept Configuration

w/ Long VTail

Unconventional Joined

Wing Design

Define Concept Space:Define Concept Space:Define Define Modeling Investigate Feasible Identify Evaluate Select1 2 3 4 5 6 7 8


40/95



C p SpC p SpDefine Design SpaceDefine Design Space

Variable Minimum Maximum Units Description

SW 7500 9000 ft2 Wing area

TWR 0.29 0.33 ~ Thrust-to-weight ratio

TIT 3000 3400 o

R Turbine Inlet Temperature

FPR 3.5 4.5 ~ Fan Pressure Ratio

OPR 18 21 ~ Overall Pressure Ratio

CLdes 0.08 0.12 ~ Design lift coefficient

X2 1.54 1.69 ~ LE kink x-location*X3 2.1 2.36 ~ LE tip x-location*

X4 2.4 2.58 ~ TE tip x-location*

X5 2.19 2.37 ~ TE kink x-location*

X6 2.18 2.5 ~ TE root x-location*

Y2 0.44 0.58 ~ LE kink y-location*

t/c_root 3 5 % Wing root t/c ratio

t/c_tip 2 4 % Wing tip t/c ratio

SHref 400 700 ft2

Horizontal Tail area

SVref 350 550 ft2

Vertical Tail area

* Variables Nondimensionalized by wing semi-span

X2,Y2

X3

X4

X5

X6

Definethe

Problem

DefineConceptSpace

Modelingand

Simulation


Feasibleor

Viable?



SelectTechnologies

Note: The geometric and

propulsive parameters may

vary in the ranges definedwith the same likelihood

since at the outset, there

should be no preference of

values. Hence, uniform

distributions are assigned

to each parameter.

Parametric Description of a Wing PlanformParametric Description of a Wing Planform
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/http://www.asdl.gatech.edu/http://www.gatech.edu/mailto:[email protected]://www.asdl.gatech.edu/


41/95



Other Design Variables

for the Aerodynamic Screening

xwing

t/c at root

t/c at tipNacelle Scaling

Horizontal Tail Area

CL Design

Root Airfoil (loc. max. thickness)

Tip Airfoil (loc. max. thickness)

Nacelle X-location

Wing Reference Area

X5, 0 Y-axis

Planform Variables

(Normalized by Span)

(X1, Y1)

X2

X3 (X4, Y1)

Xwing

naY1naY2

X-axis

0, 0

Parametric Description of a Wing PlanformParametric Description of a Wing Planform

Possible Wing PlanformsPossible Wing Planforms


42/95



Possible Wing PlanformsPossible Wing Planforms

Parametric Technology Space:Parametric Technology Space:


43/95


gy pgy pFamily of DesignsFamily of Designs

Modeling and Simulation:Modeling and Simulation:

Define Define Modeling Investigate Feasible Identify Evaluate Select

1 2 3 4 5 6 7 8


44/95


ggVehicle ModelingVehicle Modeling

M&S environment:

Relates responses to inputs via a physics-based

M&S environmentMetamodels are employed to facilitate the use of

higher-fidelity analysis for unconventional

configurations

InputVariables

OutputResponses

Response Data

FLOPS/ALC

CA

DoE

FLOPS/

ALCCA

Aero RSEs=f(design)

C

LCD

FLOPS (Flight Optimization System): A NASA-Langley

vehicle synthesis and sizing code, well-suited for the conceptualand preliminary design of subsonic transport aircraft.

ALCCA (Aircraft Life-Cycle Cost Analysis):Developed by

NASA-Ames and enhanced by ASDL; calculates life-cycle costs

and airline economics for transport aircraft.

Design

Variables & Distributions

Tech. (k)or

Response =f (design variables), or

=f (technology k factors)

theProblem

ConceptSpace

andSimulation

DesignSpace

orViable?

Technologies Technologies Technologies

Creation of Modeling and Simulation EnvironmentCreation of Modeling and Simulation Environment


45/95


Creation of Modeling and Simulation EnvironmentCreation of Modeling and Simulation Environment

WATEWeight Analysis of

Turbine Engines Code

FLOPSFlight Optimization

Code

NEPPEngine Performance

Program

ALCCAAircraft Life Cycle

Cost Analysis Code

Multi-Disciplinary DOEMissionRequirements

MarketRequirements

TechnologySetting

FidelityMultipliersEconomic

Assumptions

Vehicle Size

VehiclePerformance

VehicleEconomics

NOx CO2 NOISE

EmissionsModules

Airframe Fixed Given Engine Architecture

ThrustRequired

ThrustAvailable

AB

EngineEngineArchitecturesArchitectures

Aircraft NeedsAircraft Needs

ASDL Probabilistic Methods ProcessASDL Probabilistic Methods Process


46/95


ASDL Probabilistic Methods ProcessASDL Probabilistic Methods Process

Aero

Structures

Weights

Etc.

DISCIPLINARY RSEs

SYNTHESIS & SIZING

FPI

RSEs

Objective0%

100%

Probability

Respons

es

Respons

es

Metrics/Objectives

Metrics/Objectives

Constraints

Constraints

Responses

Responses

Metrics/Objectives

Metrics/Objectives

Constraints

Constraints

Responses

Responses

Metrics/Objectives

Metrics/Objectives

Constraints

Constraints

Concept Space TechnologySpace

RequirementsSpace

%$/RPM

TOFLmodSLNmod

CDF

DynamicContour

Plots

Competitive Assessment

Strategic Decision Making

x

x

Engine Weight

Thrust

x

x

x

x

Design Point

ArchitectureA

ArchitectureB

Growth Spurs

Aspiration Space

PhysicsDrivenGrowth

CustomerDriven Requirements

(Concept &Technology Set Specific)

TechnologyInsertionImpact

x

x

Engine Weight

Thrust

x

x

x

x

Design Point

ArchitectureA

ArchitectureB

Growth Spurs

Aspiration Space

PhysicsDrivenGrowth

CustomerDriven Requirements

(Concept &Technology Set Specific)

TechnologyInsertionImpact

Viewing RSEsViewing RSEs-- Prediction ProfilesPrediction Profiles


47/95


Viewing RSEsViewing RSEs-- Prediction ProfilesPrediction Profiles

Uses of Prediction Profile

1) Debugging: Review each sensitivity,

checking for those that dont make intuitivesense: investigate

2) Fidelity: Adjusting the regressor variables

to investigate the strength of their impact on

responses

3) Life/Technology: Model the impact of

new technologies (or the degradation of

current systems) by using metric-factors as

regressors.

Prediction Profile: This displaysprediction tracesfor each X variable. A prediction trace is the

predicted response as one variable is changed while the others are held constant at the current

values. The Prediction Profile can recompute the traces as you vary the value of an X variable.*

Regressor Variables

Responses

Variable Limits

Prediction Trace

Hairlines

Calculated Value

Input Value

AREA

Dynamic Interactive Design Space TradeDynamic Interactive Design Space Trade--off Environment for an SSToff Environment for an SST
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:dim

paradigm shift in complex system design

Documents