MDO 0

1

• MULTIDISCIPLINARY DESIGN OPTIMIZATION
• A Paradigm Shift in Design Methodology
• for Complex Engineering Systems

K Sudhakar, PM Mujumdar, Amitay Isaacs Centre
for Aerospace Systems Design Engineering Dept.
of Aerospace Engineering, IIT Bombay
2
ENGINEERING DESIGN OPTIMIZATION

• Decision is objective and not subjective
• Forces a mathematical statement of the problem
• Forces modeling system performance goodness
  criteria
• Captures knowledge - What was the problem solved,
  how was it analyzed, how were the decisions
  taken, . . .

3
ISSUES IN POSING THE PROBLEM

• Of all variables that influence the design
• which to pick as design variables? x ? X
• How to confirm that all constraints are g
• specified (g, h)?
• Which one(s) f ? F to choose as objectives?
• How to evaluate f, g, h ?
• How to handle coupled multi-disciplinary
  (iterative)
• analysis

4
HIERARCHICAL STEPS IN TRADITIONAL DESIGN

• General arrangement and performance
• Representative Configurations
• General internal layout

Mission Requirements
Conceptual Design
Conceptual Baselines

• Optimization
• Parametric
• 1st level analysis

• Systems Specifications
• Detailed Subsystems
• Internal Arrangements
• Process Design

Preliminary Design
Selected Baseline

• Sophisticated analysis
• Problem Decomposition
• Disciplinary Optimization

Detailed Design
Production Baseline
Production and Support

• May lead to Sub-optimal designs

Source AIAA MDO White Paper, 1991
5
TRADITIONAL APPROACH TO PRODUCT DEVELOPMENT

• Short Conception phase unequal distribution of
  disciplines
• limited scope for optimization assessing
  impact of inter-disciplinary
• couplings.
• Correction of later problems
• Costly/ Lost Time/ Futile
• Solutions limited to specific discipline

Source AIAA MDO White Paper, 1991
6
LEVERAGE IN THE DEVELOPMENT PROCESS
Ballistic Missile System
Life cycle cost effectively rendered unchangeable
for a given design
Source AIAA MDO White Paper, 1991
7
THE DESIGN PROCESS PARADIGM SHIFT
Design process reorganized to gain information
earlier and to retain design freedom longer

• More up-front design
• More evenly distributed efforts of disciplines
  in early design
• Alleviate paradox
• Design decisions/trade-off reordered

Source AIAA MDO White Paper, 1991
8
CONCURRENT ENGINEERING Vs MDO
Ultimate objective - Balanced design by full and
formal multi-disciplinary
integration and optimization concurrently in
all disciplines
Source AIAA MDO White Paper, 1991
9
MDO ?

• Multi-disciplinary More than one discipline
  plays a role. Eg. In aerospace -aerodynamics,
  structures, controls, mission, . . .
• Design Process of translating requirements into
  detailed product specifications.
• Optimization Formal mathematical process of
  locating the best under constraints

10
What is MDO? Some popular definitions of
Multidisciplinary Design Optimisation

• A methodology for the optimal design of complex
  engineering systems and subsystems that
  coherently exploits the synergism of mutually
  interacting phenomena using high fidelity
  analysis with formal optimization
• MDO is a methodology that combines analysis and
  in individual disciplines into that for the
  entire system for optimization.
• "How to decide what to change, and to what extent
  to change it, when everything influences
  everything else."

11
(No Transcript)
12
HAS MDO BEEN AROUND?
Improvement in Aerodynamics siphoned off for
other system level benefits MDO in action
13
CONVENTIONAL DESIGN V/S MDO

• Conventional Aerospace Design Practice
• Heirarchichal
• Dependence on Parameter trends and trade-off
  studies
• Independent disciplinary design System level
  reviews
• Optimization limited to disciplines
• Resolution of interdisciplinary conflicts
  non-automated
• Relying heavily on previous experience
• Overall more heuristic, than formal
  mathematical optimization

14
CONVENTIONAL DESIGN V/S MDO

• Multi-disciplinary Design Optimization
• Low fidelity models Conceptual design
• Tightly coupled inter-disciplinary codes. (Close
  knit group) ??
• All or most of the following,
• Largely automated within a formal framework
• Formal mathematical methods and high fidelity
  computations indispensable
• Interdisciplinary couplings formally
  modeled/retained
• Design freedom to significantly affect system
  performance in multiple disciplines
  simultaneously
• Disciplinary authority. Parallel execution.

15
CONVENTIONAL DESIGN V/S MDO

• Multi-disciplinary Design Optimization (contd.)
• Special architectures for problem formulation/
  decomposition
• System level objectives, constraints, variables
• Disciplinary (local) constraints, variables
• Coupling variables and constraints
• System level and discipline level optimizations
• Human expertise judgement given due weightage

16
MDO CONCEPTUAL ELEMENTS
MD Optimization
Design-Oriented MD Analysis
Information Science Technology
Optimization Problem Formulation
Mathematical Modeling
Software Integration
S/W Engineering Practices
Decomposition
Cost v/s Accuracy Trade-off
Smart Reanalysis
Data Management, Storage Visualization
Interdisciplinary Feasibility
Data S/W Standards
Approximations
Design Space Search
Human Interface
Sensitivity Analysis
Source AIAA MDO White Paper, 1991
17
CHALLENGES IN MDO IMPLEMENTATION

• Information Science Technology
• Computational resources (CPU, memory, disk
  space)
• Distributed parallel processing
• Common parametric geometric model
• Software support
• s/w integration of proprietary, legacy,
  commercial, . .
• configuration control and data management
• collaborative work environment,
  person-person/machine
• Human expertise/experience capture

18
CHALLENGES IN MDO IMPLEMENTATION

• Multidisciplinary Analysis
• Well posed interfaces for disciplines
• Discipline and MD sensitivities
• Mathematical modeling of LC disciplines
• Automated grid generation for CFD, FEM
• Cost run-time of high fidelity analysis
• MD Optimization
• Problem definition
• MDO architectures
• Design Space Search

19
OPTIMIZATION ISSUES IN MDO

• Single level monolithic optimization
  (conventional)
• Decomposition
• Decomposed Analysis
• System level optimization
• Parallel disciplinary analysis
• Decomposed optimization
• Multi-level (system and subspace) optimization
• Parallel disciplinary analysis
• System sensitivity analysis
• Design oriented analysis, Surrogates
• Improved optimization algorithms
• Large number of design variables constraints
• Gradient free

20
OPTIMISATION ISSUES IN MDO

• Which optimisation algorithm to use?
• Gradient based? How to generate gradients?
• Evolutionary? How many function evaluations?
• Evaluation of gradients?
• Requirements on convergence more severe
• than that required for engineering analysis.
• Noisy functions?

21
MDO GROWTH OVER THE YEARS
1999
Second Special Issue of Journal of Aircraft
22
MDO STATUS AT A GLANCE

• Generalized MDO environment far from reach

• Most common applications
• One discipline (structures) with other
  discipline as constraints
• Simultaneous aerodynamic and structural
  optimization
• Wings
• Aircraft configurations (HSCT, BWB)
• Rotor blades
• Coupling of preliminary design with mission and
  performance
• optimization catching up
• Trajectory optimization in Space vehicles
• Disciplines - propulsion, trajectory analysis,
  weights, sizing
• Simultaneous structures, aerodynamics and
  control optimization

23
SUMMARY OF INDUSTRY MDO APPLICATIONS
Source AIAA MDO White Paper, 1998
24
An Example HSCT (1991-99)!

• HSCT-2
• 5 design variables, 6 constraints
• WINGDES, ELAPS, Range equation, engine deck
• Time for one cycle 10 minutes
• HSCT-3
• 7 design variables, 6 constraints
• ISAAC, COMET, Range equation. Engine deck
• Time for one cycle 3 hours
• HSCT-4
• 271 design variables, 31,868 constraints
• CFL3D, USSAERO, GENESIS, FLOPS, ENG10
• Time for one cycle 3 days (Analysis,
  sensitivity, optimizer step)
• 32 P Origin 2000, n x Sun Ultra-2, SGI R10000

CFL3D - CFD in Euler mode USSAERO Panel
code, GENESIS - FEM
25
HSCT - 4

• Detailed problem definition took more than 1 year
  to extract from people
• Requirements document touched 100 pages merely to
  define analysis process, tools used and data flow
• 90 of work went into preparing analysis codes
  for MDA and integrating them in a proper
  sequence. (Such experiences have prompted
  development of MDO Frameworks)

26
MDO_at_CASDE OVER THE YEARS

• Aug 1999 - CASDE asked to spur initiatives in
  MDO
• Aug 2000 - First meeting of SIG-MDO
• Jan 2001 - Professional Development Course on
  MDO
• Jun 2002 - Second meeting of SIG-MDO
• Mar 2003 - Third Meeting of SIG-MDO
• Sep 2003 - International Conference on MDO
• Feb 2004 - Fourth Meeting SIG-MDO
• Jan 2005 - Fifth Meeting of SIG-MDO

27
End of MDO Introduction
28
MDO STATUS AT A GLANCE

• MDO Elements
• Decomposition, Approximations, Sensitivity
  Analysis well researched but not matured to
  industry requirements
• Other issues ??

• Recent Trends - Generic MDO Technologies (NASA,
  EC )
• Distributed heterogeneous environment
• Problem specification and setup, decision support
  systems
• Generic software framework for MDO
• Multi-objective design optimization
• MADIC, FRONTIER, FIDO, iSIGHT, USMADE
• IPPD system for aircraft design (ASDL, GIT
  Atlanta)
• Knowledge Based Engineering (Boeing)
• MDO in preliminary design focused on
  aero-elasticity with FEM, CFD