SlideShare a Scribd company logo
1 of 119
&




We Provide You Confidence in Your Product ReliabilityTM
   Ops A La Carte / (408) 654-0499 / askops@opsalacarte.com / www.opsalacarte.com
MECHANICAL DESIGN
  for RELIABILITY
       (MDfR)
      SEMINAR
         April 9, 2008

  T. Kim Parnell, Ph.D.,P.E.
          408-203-9443
     kimp@OpsALaCarte.com
        Ops A La Carte LLC
      www.OpsALaCarte.com


          © 2006 Ops A La Carte
The following training and presentation
       materials are copyright protected by:
                Ops A La Carte LLC
                        and
      Parnell Engineering & Consulting (PEC)




April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability   2
                          © 2006 Ops A La Carte
Presenter’s Biographical Sketch
                                Kim Parnell
◈ Dr. Kim Parnell is a Senior Consultant with Ops A La Carte and Principal/Founder of
  PEC (www.parnell-eng.com), an engineering consulting firm that focuses on providing
  support for high-tech and medical device companies. He specializes in the mechanical
  engineering design and behavior of electronic and miniature components, MEMs,
  wireless sensors, telecommunications devices, biomedical devices, and shape memory
  metals. Dr. Parnell is an expert in the application of finite element analysis (FEA) to the
  solution of engineering problems, and consults actively in these areas as well as failure
  analysis and reliability. He served as a Visiting Associate Professor in the Mechanical
  Engineering Department at Stanford University and is a coach and mentor for the
  innovative Stanford Biodesign program. He is a member of the NanoBioConvergence
  (www.NanoBioConvergence.org) Board, and the CSIX Connect (www.csix.org) Board
  Dr. Parnell is a member of ASME, IEEE, ASM, and is currently Chair of the IEEE-CNSV
  (Consultants’ Network of Silicon Valley) (www.CaliforniaConsultants.org).
◈ He has worked with companies including Ops A La Carte, MSC Software, Rubicor
  Medical, Exponent Failure Analysis Associates, SST Systems, ATT Bell Laboratories,
◈
  Stanford University, and General Motors.
◈ Dr. Parnell holds Ph.D. and MSME degrees from Stanford University in Mechanical
  Engineering and a BES from Georgia Tech. He is a registered Professional Mechanical
◈
  Engineer in the State of California.

    April 9,9, 2008
    April 2008              MDfR-Mechanical Design for Reliability                      3
                                    © 2006 Ops A La Carte
Mechanical Design for Reliability-MDfR
                                T. Kim Parnell, PhD, PE

    8:15           8:30   Setup
    8:30           9:30   Introduction to Mechanical Reliability & Robust Design
    9:30          11:00   Finite Element Analysis (FEA)
  11:00           11:30   Random Vibration & Shock

  11:30           12:00   Tolerance & Worst Case Analysis

  12:00           1:00    Lunch
   1:00           1:30    Thermal Analysis & Electronic Cooling
   1:30           2:45    Probabilistic Design System (PDS) & Stochastics
   2:45           3:00    Forensic Methodology
   3:00           4:00    Failure Analysis & Prevention; Fatigue & Fracture
   4:00           4:30    Conclusions & Discussion

April 9,9, 2008
April 2008                   MDfR-Mechanical Design for Reliability                4
                                     © 2006 Ops A La Carte
Introduction to
                   Mechanical Reliability &
                      Robust Design




April 9,9, 2008
April 2008                    MDfR-Mechanical Design for Reliability                    5
                                      © 2006 Ops A La Carte




                     Six Sigma in Practice
                              What does it mean?
        99% Performance                                       6 Sigma Performance
                                                                 99.999663% Meet Spec
   202 Billion Pieces of Mail Delivered by USPS per year
2.02 Billion pieces Lost per Year                        680740 pieces lost per year

                      24/7 Power Delivery to Your Home
87 Hours without Power every Year                        Less than two minutes without power
                                                         per year
              510 Million Prescriptions Worldwide per year
5.1 Million Wrong Prescriptions per Year                 1719 Wrong Prescriptions per Year

                  27 Billion Credit Card Transactions per year
0.27 Billion wrong transactions per year                 90990 wrong transactions per Year

Ref: Scott Burr, 2005, “Design for Six Sigma” Hubenthal-Burr Associates
April 9,9, 2008
April 2008                    MDfR-Mechanical Design for Reliability                    6
                                      © 2006 Ops A La Carte
Two Types of Quality
  Two Types of Quality

  •     Type 1: Customer Quality - The features that customers want.
  •     Type 2: Engineered Quality - The problems customers do not want.

  •     Customer quality leads to the size of the market segment and
        includes items such as function, features, colors and designs. The
        better the customer quality, the bigger the market size becomes. In
        order to obtain the market size, the price must be reasonable.
  •     Engineered quality includes defects, failures, noise, vibration,
        unwanted phenomena, lowering the cost of manufacture, and
        minimizing manufacturing problems.


  Ref: “Design for Six Sigma Roadmap”, Shree Phadnis



April 9,9, 2008
April 2008               MDfR-Mechanical Design for Reliability        7
                                 © 2006 Ops A La Carte




                  Results & Meaning
      • Process provides snapshot of current system


      • No one tool makes an entire reliability program


      • Check step is critical before moving to
            recommendation around improvement plan




April 9,9, 2008
April 2008               MDfR-Mechanical Design for Reliability        8
                                 © 2006 Ops A La Carte
Robust Design
• Engineers dream of designing a product or
  process that exhibits the state of 'robustness.'
  However, few can actually make the claim that
  they know exactly what robustness means.
Formal definition by quality guru Dr. Taguchi:
• Robustness is the state where the technology,
  product, or process performance is minimally
  sensitive to factors causing variability (either in
  the manufacturing or user environment), and
  aging at the lowest manufacturing cost.
Ref: “Design for Six Sigma Roadmap”, Shree Phadnis

April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability   9
                                © 2006 Ops A La Carte




          How to Achieve Robust Design?
• “Brute Force” techniques
        – Intuitively select design parameters and
          tolerances and make system design trade-
          offs
        – Added design margins
        – Tighter tolerances
        – Does not work in this hyper competitive
          economy


April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability   10
                                © 2006 Ops A La Carte
Cost to Repair Design Problems




 •     As a rule, design engineering has lagged behind the shop
       floor in awareness of product and process quality.
 •     Very real costs are associated with inattention to design
       quality. If errors or omissions in design data are not
       addressed early, more costly changes are required later in
       the product development process.

 Ref: “Building Quality into Design Engineering”, Gavin Finn
April 9,9, 2008
April 2008                 MDfR-Mechanical Design for Reliability   11
                                   © 2006 Ops A La Carte




                  Robust Design Outline
 • 21st century challenges
 • Tools & techniques for Design For Six Sigma
   (DFSS)
 • Definition of robustness
    – What, how, significance
 • Business impact of robust design solutions
 • Systematic approach to robust design

 •     Ref: Ansys, 2004



April 9,9, 2008
April 2008                 MDfR-Mechanical Design for Reliability   12
                                   © 2006 Ops A La Carte
21st Century Challenges
• Product life cycles are expected to last for
  just a few months
• More customized products
      – 1700 car models as opposed to 900 ten years ago
      – More focus on “built-in-quality” or “built-in-reliability”
• To meet these demanding requirements
      – Products need to be developed in the shortest amount of time
        and
      – They have to be safe, reliable, and competitive
      – Robust and Affordable !
  “Product quality requires managerial,
   technological and statistical concepts
   throughout all the major functions of the
   organization …” Joseph M. Juran
April 9,9, 2008
April 2008                    MDfR-Mechanical Design for Reliability                   13
                                      © 2006 Ops A La Carte




                     Tools & Techniques
                   for Design for Six Sigma
    Planning        Concept              Optimization
                                Design                                 Tolerance
      VOC         Development                         Verification
                              Development Variation                     Analysis Manufacturing
      QFD          Screening


• Quality Function Deployment (QFD)
        – Customer requirements to technical specs
• Failure Modes and Effects Analysis (FMEA)
        – To study failures, their impact and frequency
• Other statistical tools for tolerance analysis,
  etc.
• Tools for concept screening, design,
  optimization, verification
• Probabilistic simulations for robust design
        – Design of Experiment (DOE), response surface
April 9,9, 2008
April 2008                    MDfR-Mechanical Design for Reliability                   14
                                      © 2006 Ops A La Carte
What is Robust Design?
• Works as intended regardless of variation in:
      –    Design
      –    Manufacturing process
      –    Material
      –    Resulting from Deterioration
      –    Operation (misuse)
• For Achieving robust design
      – Understand potential sources of variations
      – Quantify the effect of variations on product behavior and
        performance
      – Take steps to desensitize the design to these variations
      – Adjust the output to hit the target performance
April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability   15
                                © 2006 Ops A La Carte




                  What is Robust Design?
• De-sensitizing the design to variations




April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability   16
                                © 2006 Ops A La Carte
How to Achieve Robust Design?
• “Intelligent Design” by understanding
        – Product & Process design parameters that are
          critical to achieving a performance
          characteristic
        – What are the optimum values to both
                  • Achieving the desired performance
                  • Minimize the effect of naturally occurring and difficult to
                    control variations (noise)
        – Requires a systematic approach of deploying
          engineering simulations as part of the design
          process
April 9,9, 2008
April 2008                      MDfR-Mechanical Design for Reliability            17
                                        © 2006 Ops A La Carte




     Significance of Robust Design
• Robust design optimization helps
  – Identify parameter values that maximize
    performance and minimize the effect of “noise”
                  • More robust and affordable design
        – Identify parameters that have no significant
          effect on performance
                  • Relaxed tolerance, reduces product cost
        – Identify parameter values that reduce cost
          without affecting performance or variation
        – Cost-effective quality inspection
                  • Do not have to inspect for parameters that are not critical to
                    performance

April 9,9, 2008
April 2008                      MDfR-Mechanical Design for Reliability            18
                                        © 2006 Ops A La Carte
Business Impact
         of Robust Design Solutions

          Profitability = Market Size
                         x Market Share
                         x Margin on Sales




April 9,9, 2008
April 2008         MDfR-Mechanical Design for Reliability   19
                           © 2006 Ops A La Carte




                      Market Size
Increased market size by entering related
  markets. Some markets require product
  designs to be robust:
      – Defense, aerospace, jet engine, nuclear
        power, biomedical, oil industry, other mission
        critical solutions
      – Auto industry – 1st tier suppliers capturing
        more value



April 9,9, 2008
April 2008         MDfR-Mechanical Design for Reliability   20
                           © 2006 Ops A La Carte
Market Share
• Increased market share (in slow growth
  industry) through
        – Better quality products
                  • Less maintenance
                  • Higher resale value
        – Quick response to customer requirements
        – Affordable pricing
        – Attract business by implementing DFSS


April 9,9, 2008
April 2008                   MDfR-Mechanical Design for Reliability   21
                                     © 2006 Ops A La Carte




                         Margin on Sales
 • Product Cost Reduction Through
   – Relaxed tolerances
   – Material savings
   – Cost-effective quality inspection
   – Less rejects (lower scrap rate)
   – Shorter development time
   – Less warranty cost
   – Customer loyalty
 • Command Premium price
   – Better quality
   – Timely delivery
   – Satisfied customers
April 9,9, 2008
April 2008                   MDfR-Mechanical Design for Reliability   22
                                     © 2006 Ops A La Carte
Reliability Philosophies
       Two fundamental methods to achieving
       high product reliability:
                    – Build, Test, Fix

                    – Analytical Approach




April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability   23
                                  © 2006 Ops A La Carte




                         Build, Test, Fix
    • In any design there are a finite number of
      flaws.
    • If we find them, we can remove the flaw.

    •     Rapid prototyping
    •     HALT
    •     Large field trials or ‘beta’ testing
    •     Reliability growth modeling
April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability   24
                                  © 2006 Ops A La Carte
Analytical Approach
    •     Develop goals
    •     Model expected failure mechanisms
    •     Conduct accelerated life tests
    •     Conduct reliability demonstration tests
    •     Routinely update system level model

    • Balance of simulation/testing to increase
      ability of reliability model to predict field
      performance.
April 9,9, 2008
April 2008          MDfR-Mechanical Design for Reliability      25
                            © 2006 Ops A La Carte




              Issues with each approach
    Build, Test, Fix                       Analytical
    • Uncertain if design                  • Fix mostly known
      is good enough                         flaws
    • Limited prototypes                   • ALT’s take too long
      means limited flaws                  • RDT’s take even
      discovered                             longer
    • Unable to plan for                   • Models have large
      warranty or field                      uncertainty with new
      service                                technology and
                                             environments
April 9,9, 2008
April 2008          MDfR-Mechanical Design for Reliability      26
                            © 2006 Ops A La Carte
Drop Tests: What is the physics?
• Is drop deterministic or stochastic?
• Is drop predictable?
• Is drop optimizable?
• Does it make sense to speak of precision in drop
  simulations?
• Do we need to increase the number of elements
  in our drop models? What is the reasonable
  limit?
• What is the future of computer-based drop
  analysis?

• Is drop a chaotic phenomenon?

April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability   27
                          © 2006 Ops A La Carte




       Electronic Component Drops




April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability   28
                          © 2006 Ops A La Carte
Electronic Component Drops




April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability   29
                          © 2006 Ops A La Carte




       Electronic Component Drops




April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability   30
                          © 2006 Ops A La Carte
Example of Measured
                   Acceleration Signal
• A series of tests for chaos are performed with this signal.




April 9,9, 2008
April 2008            MDfR-Mechanical Design for Reliability   31
                              © 2006 Ops A La Carte




                  Log-linear Power Law
• Systems that exhibit a log-linear Power Spectrum are
  potentially chaotic.




April 9,9, 2008
April 2008            MDfR-Mechanical Design for Reliability   32
                              © 2006 Ops A La Carte
Typical Tests for Chaos
• Hausdorff (Capacity dimension). Signal has fractal
  dimension (1.8).
• Log-linear Power Spectrum (yes).
• Correlation dimension (5).
• Lyapunov Characteristic Exponents (+0.4).
• Poincare’ sections or Return Maps (check for structure).

• According to these tests, the measured drop/impact
  signal possesses a clear chaotic flavour. This explains
  why each drop is a unique event and cannot be
  optimised.
April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability    33
                                  © 2006 Ops A La Carte




                         Drop = Chaos
  •     Chaos can be described by closed-form deterministic
        equations. Chaos does NOT mean random.
  •     Chaos is characterized by extreme sensitivity to initial
        conditions.
  •     “Memory” of initial conditions is quickly lost in chaotic
        phenomena (“butterfly effect”).
  •     Examples of chaotic phenomena:
         – Tornados (weather in general)
         – Stock market evolution, economy
         – Crash, drops, impacts, etc.
         – Earthquakes
         – Avalanches
         – Combustion/turbulence
         – EEG (alpha-waves in brain)
         – Duffing, Van der Pol, Lorenz oscillators, etc.

April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability    34
                                  © 2006 Ops A La Carte
The Logistic Map
    • X(n+1) = k X(n)(1-X(n))
    Shows astonishingly complex
      behaviour:
            –     0 < k < 1, Extinction regime
            –     1 < k < 3, Convergence regime
            –     3 < k < 3.57, Bifurcation regime
            –     3.57 < k < 4, Chaotic regime
            –     4 < k, Second chaotic regime

    For more details see:
    Weisstein, Eric W. "Logistic Map." From MathWorld--A Wolfram
       Web Resource. http://mathworld.wolfram.com/LogisticMap.html

April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability   35
                                    © 2006 Ops A La Carte




                      Chaos & Predictability
• Phenomena that are chaotic, are unpredictable
  (nonrepeatable). The main reason is extreme sensitivity
  to initial conditions.
• Phenomena that are unpredictable, cannot be optimized.
  They must be treated statistically.
• All that can be done with chaotic phenomena is increase
  our understanding of their nature, properties, patterns,
  structure, main features, quantify the associated risks.
• Models for Risk Analysis must be realistic to be of any
  use.


April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability   36
                                    © 2006 Ops A La Carte
Understanding Risk
• Essentially, risk is associated with the
  existence of outliers           Outlier:
                                                                               - warranty
                                                                               - recall
                                                                               - lawsuit    }
                                                                               Most likely
                                                                               response
                                                                               (highest density)

    Note: DOE and Response Surface techniques
    cannot capture outliers
April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability                              37
                                    © 2006 Ops A La Carte




             What is Risk and Uncertainty
                    Management?
• Understand and remove outliers
• Shift entire distribution

                                is safe                               fails


                  Improved design
                                                            Initial design           Outliers:
                                                                                     unfortunate
                                                                                     combinations
                                                                                     of operating
                                                                     Outlier         conditions and
                                                                                     design variables
                                                                                     that lead to
                                                                                     unexpected
                                                                                     behaviour.

April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability                              38
                                    © 2006 Ops A La Carte
Example of Robust Design: MIR
              Space Station
• Robustness = survivability in the face of unexpected
  changes in environment (exo) or within the system (endo)




April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability   39
                                © 2006 Ops A La Carte




                  Example of Optimal Design
• M. Alboreto dies (Le Mans, April 2001) due to slight loss
  of pressure in left rear tire. The system was extremely
  sensitive to boundary conditions (was optimal, and
  therefore very very fragile!).




April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability   40
                                © 2006 Ops A La Carte
Optimization: a Dangerous Game

                                                                 Second order RS

First order RS



Optimum?




      Different theories can be shown to fit the same set of observed
      data. The more complex a theory, the more credible it appears!
April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability            41
                                © 2006 Ops A La Carte




                       Some Lessons

    • Boundary conditions are most important
    • Small effects can have macroscopic consequences
      (watch out for chaos, even in small doses!)
    • Precision is not everything!
    • Optimal components don’t give an optimal whole
    • Optimality = fragility
    • Robust is the opposite to optimal




April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability            42
                                © 2006 Ops A La Carte
Conclusions – Robustness, Chaos,
         and Predictability
     • Phenomena that possess a chaotic component
       cannot be optimized, but can be improved in
       statistical sense.
     • With such systems, it is possible to address:
        – Risk Analysis
        – Design for robustness
        – Increase understanding
     • Realistic models necessitate:
        – continuous 3D random fields (geometry)
        – discrete random field (spotwelds, joints)
        – randomization of ALL material properties
        – randomization of ALL thicknesses
        – variations of boundary/initial conditions
April 9,9, 2008
April 2008          MDfR-Mechanical Design for Reliability   43
                            © 2006 Ops A La Carte




     Finite Element Analysis (FEA)




April 9,9, 2008
April 2008          MDfR-Mechanical Design for Reliability   44
                            © 2006 Ops A La Carte
Finite Element Analysis (FEA)
• FEA is generally applicable for analyses such as
  stress, thermal, vibration, and dynamic cases
  1) Create a geometric model; subdivide into
     elements to create a finite element mesh
  2) Specify material properties for all components
  3) Apply loads and boundary conditions: thermal,
     pressure, deadweight, wave loads, etc.
• Linear analysis is typically used for design;
  nonlinear analysis is frequently required for
  failure investigation

    April 9,9, 2008
    April 2008        MDfR-Mechanical Design for Reliability   45
                              © 2006 Ops A La Carte




                      FEA & Testing

       • Finite element analysis (FEA) and physical
         testing are complementary
       • A comprehensive program needs to include
         both components
       • With judicious experimental validation, FEA
         can be used to reduce the amount of physical
         testing that is needed and shorten the design
         cycle


    April 9,9, 2008
    April 2008        MDfR-Mechanical Design for Reliability   46
                              © 2006 Ops A La Carte
FEA Simulation Tools
FEA simulation can
• Be a cost effective way to evaluate design choices.
• Provide insight into how varying parameters affects the
  design outcome.
• Simulate expensive processes where downtime is
  unacceptable.
• Help establish the critical relationship between design
  parameters and process parameters. THIS IS BIG IN
  SIX SIGMA!!!! And can translate into an enormous and
  unfair Competitive Advantage.
Ref: Scott Burr, 2005, “Design for Six Sigma”, Hubenthal-Burr Associates




 April 9,9, 2008
 April 2008                         MDfR-Mechanical Design for Reliability   47
                                             © 2006 Ops A La Carte




                             FEA Concepts
• Linear Analysis
   – Small Deflection & Small Strain
   – Elastic material
• Nonlinear Analysis
   – Large Deflection &/or Large Strain
   – Nonlinear material
              • Elastic/plastic
              • Rubber & Polymers
              • Temperature dependent properties
  – Contact
  – Shock & impact
• Multi-physics: thermal, fluid, electromagnetic, etc.
 April 9,9, 2008
 April 2008                         MDfR-Mechanical Design for Reliability   48
                                             © 2006 Ops A La Carte
Finite Element Software
   •     MSC.Marc
   •     MSC.Nastran, MSC.Adams
   •     MD-Nastran
   •     LS-Dyna
   •     Ansys
   •     Abaqus
   •     Comsol
   •     Cosmos
   •     Pro-Mechanica
   •     Algor
 April 9,9, 2008
 April 2008                MDfR-Mechanical Design for Reliability                                        49
                                   © 2006 Ops A La Carte




         The Real World is Nonlinear
• Simplified approximations
  compromise accuracy &                                                     F      Linear

  safety                                                                Load
                                                                                 Behavior


   – Cannot reliably predict geometric,                                 F                               Nonlinear
                                                                                                        Behavior
     material, & boundary nonlinearities
                                                                         u                                u
                                                                                   Displacement

• Nonlinear analysis is                                                                Linear
  essential for investigating:                                          Stress       Behavior

                                                                                        C
                                                                                                   PLASTIC
                                                                                                   Nonlinear
   – Large-strain effects, plasticity, friction,                                     TI
                                                                                   AS              Behavior
     dynamic loads, large deflections & motion                      F       F    EL
                                                                                            Yield Pt.
   – Metallic & non-metallic materials                                                 Strain
     exhibiting nonlinear behavior (e.g., shape
     memory alloys, plastics, elastomers,
     composites, etc.)
   – Effects of physical coupling with
     thermal, electrical, fluid, magnetic,
     acoustic, etc.




 April 9,9, 2008
 April 2008
 MSC.Marc                  MDfR-Mechanical Design for Reliability                                        50
                                   © 2006 Ops A La Carte
Simulate Reality With Nonlinear FEA
Capture fit, form, function, &
manufacturability through simulation
Examples:

                   Snap-Fit                               Automobile   Machining of
                   Buckle                                 Exhaust      Aircraft
                                                          Pipe         Fuselage
                                                          Hanger       Component




   - Make better design decisions earlier in your product’s life cycle
   - Explore & optimize product performance more quickly through
     less-expensive virtual testing
April 9,9, 2008
April 2008                    MDfR-Mechanical Design for Reliability     51
MSC.Marc
                                      © 2006 Ops A La Carte




                  Introducing MSC.Marc®
Nonlinear analysis for a wide range of
applications:




  MSC.Marc provides virtual insight into the behavior of components
  experiencing geometric, material, and/or boundary nonlinearities
April 9,9, 2008
April 2008                    MDfR-Mechanical Design for Reliability     52
                                      © 2006 Ops A La Carte
What is MSC.Marc?
Virtual Product Development (VPD) solution that helps
engineers simulate & optimize component performance
& manufacturing processes
– Long, proven history of innovation—established in
  1971 as the world’s first commercially available
  nonlinear FEA software
– Provides advanced, general-purpose, implicit
  nonlinear
  FEA technology with high-performance, linearly-
  scalable
  parallel processing capability
– Solves static & dynamic 2-D & 3-D structural &
  non-structural coupled multiphysics problems with
  highest accuracy & robustness
– Simulates all kinds of structural deformation &
  material nonlinearities:
      • Multibody contact & frictional effects
      • Buckling, cracking, bending, cutting, shaping, forming,
        welding, etc.
      • Highly-nonlinear materials, such as elastomers,
        composites, plastics, concrete, etc.
 April 9,9, 2008
 April 2008                          MDfR-Mechanical Design for Reliability                            53
                                             © 2006 Ops A La Carte




Elastic/Plastic Tensile Response
                                                                     True Stress                 E’
         Ultimate                                                                  D     Eng. Stress
         stress
                                                      C                                          E
         Yield                       B
         stress
                            A
         Proportional
         limit



                        0
                            Linear         Plastic               Strain            Significant
                                                                 hardening         necking
  Ref: Anderson, Biomaterials notes


 Typical stress/strain curve for steels. Strains become localized when
 necking occurs. Standard elongation highly dependent on gage
 length. Area reduction gives local strain.
 April 9,9, 2008
 April 2008                          MDfR-Mechanical Design for Reliability                            54
                                             © 2006 Ops A La Carte
Tensile Test for Steel
                        Pipeline Material




 April 9,9, 2008
 April 2008                  MDfR-Mechanical Design for Reliability                  55
                                     © 2006 Ops A La Carte




  Global Remeshing/Rezoning Capability
MSC.Marc has a unique automated capability overcomes
mesh distortions arising from extreme material deformation
 – Remeshing: New, better quality global mesh is automatically
   generated
 – Rezoning: Solution is transferred from the old mesh to the new
   mesh
 – Contact boundary conditions are also automatically reapplied
 – Triangular or quadrilateral elements for 2-D, tetrahedral or
   hexahedral for 3-D
 Compression




                                                                      3-D Example Showing
   Rubber Seal with       With Global
                                                      Animation of    Close-Up View of Mesh
    Distorted Mesh     Remeshing/Rezoning
                                                      Rubber Seal          Refinements


       Avoid simulation failure & loss in accuracy for materials
 April 9, 2008
 April 9, 2008
                 experiencing extreme deformation
                      MDfR-Mechanical Design for Reliability  56
                                     © 2006 Ops A La Carte
3-D Cyclic Symmetry with Remeshing
• Examples: Screw Twisting




MSC.Marc
April 9,9, 2008
April 2008                           MDfR-Mechanical Design for Reliability                  57
                                             © 2006 Ops A La Carte




 Advanced Friction Modeling Capability
MSC.Marc has simple and robust friction
capability. Simply turn it on & supply a
                                                                               Insertion &
coefficient of friction—or, manually                                            Extraction
provide necessary parameters                                                  Force on an
                                                                              Elastomeric
                                                                                      Seal




                Belt-Driven Pulley

  Simulate realistic contact conditions by including effects
  of friction
April 9, 2008
April 9, 2008       MDfR-Mechanical Design for Reliability 58
                                             © 2006 Ops A La Carte
Enabling You to Simulate Reality
 Use MSC.Marc’s
 innovative contact
 modeling capabilities
 to satisfy your virtual
 testing requirements
                                                Stresses in Lug,
                                                 Pin, & Clevis



                                                                          Gear Lever with Rubber Boot




   Staking of Intersecting                Friction & Heat
Hot Wires into a Plastic Block        Generation in Disc Brake                   Pin Insertion


   Validate product performance under a wide range of
   operating conditions & constraints
  April 9, 2008
  April 9, 2008     MDfR-Mechanical Design for Reliability                                         59
                                         © 2006 Ops A La Carte




                  MSC.Marc Tube Bending with
                    Contact & Remeshing




  April 9,9, 2008
  April 2008                     MDfR-Mechanical Design for Reliability                            60
                                         © 2006 Ops A La Carte
Coupled Thermal-Mechanical




April 9,9, 2008
April 2008                             MDfR-Mechanical Design for Reliability   61
                                               © 2006 Ops A La Carte




                    MSC.Marc - Machining
• Pocket Cutting
  – Definition of the Two Cutting Processes
Y                               Cutter and its axis definition 1
                                Surface for the first cut stage




Z

             Y


         X
Surface for the second cut stage

    Cutter and its axis definition 2
April 9,9, 2008
April 2008                             MDfR-Mechanical Design for Reliability   62
                                               © 2006 Ops A La Carte
MSC.Marc Machining - Verification
• Experimental verification
  of machining behavior for
  a 1.5 inch thick stock
  aluminum beam that is
  bent to a prescribed
  radius.
• 2.5 inch slot is cut through
  75% of its thickness,
  spring back is plotted
  along the ten gage points.



April 9,9, 2008
April 2008         MDfR-Mechanical Design for Reliability   63
                           © 2006 Ops A La Carte




Predicting Distortion from Machining

• Machining
  operations
  cause distortions
• MSC.Marc can
  be used to
  predict the
  distortion




April 9,9, 2008
April 2008         MDfR-Mechanical Design for Reliability   64
                           © 2006 Ops A La Carte
Coupled Electrostatic-Mechanical
                                                    • Capacitor plates close
                                                      by increasing charge
                                                      due to Coulomb’s Law
                                                    • Voltage peaks at 1/3
                                                      of original gap opening
                                                    • Popular for MEMS
                                                      applications:
                                                       – Optical-network
                                                         components
                                                       – Print Heads
                                                       – Projectors
                                                       – BioMEMS
April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability                65
                          © 2006 Ops A La Carte




   Electrostatic-Mechanical MEMs




April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability                66
                          © 2006 Ops A La Carte
Joule Heating-Mechanical MEMs

                                                           •Electrical current
                                                           flow causes heating
                                                           due to material
                                                           resistivity
                                                           •Mechanical
                                                           deformation due to
                                                           thermal expansion
                                                           of hot leg



April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability                    67
                          © 2006 Ops A La Carte




       Magnetostatic Improvements
• New tetrahedral
  elements for
  magnetostatics (181
  and 182)
• New Line (wire)                                                B


  element for
  magnetostatics (183)


                                                                 A

        A                       B
April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability                    68
                          © 2006 Ops A La Carte
Coupled Electromagnetic-Thermal
• Induction heating used for various industrial
  manufacturing processes, such as heat
  treatment for metals, soldering guns, and
  preheating for forging




April 9,9, 2008
April 2008                             MDfR-Mechanical Design for Reliability   69
                                               © 2006 Ops A La Carte




                  Example: RF Attenuator
                       Simulation
RF/microwave energy is attenuated through resistive losses in a
  Nichrome film attached to the microstripline waveguide. The
  energy is lost in the form of heat which is conducted both through
  the devices ceramic substrate and top insulating surface film.
                  Typical Packaged Device:




                  Image from KDI data sheet.

April 9,9, 2008
April 2008                             MDfR-Mechanical Design for Reliability   70
                                               © 2006 Ops A La Carte
Multi-field Solver- RF Attenuator
• Analysis results:
                    E-field




                  H-field

                                                              Resultant
                                                            temperature


April 9,9, 2008
April 2008         MDfR-Mechanical Design for Reliability                 71
                           © 2006 Ops A La Carte




           Random Vibration & Shock




April 9,9, 2008
April 2008         MDfR-Mechanical Design for Reliability                 72
                           © 2006 Ops A La Carte
Random Vibration Capabilities
 • MSC.Patran and MSC.Nastran contain
   very easy to use random vibration
   analysis
 • The overview will have some specific
   references to the capabilities in these
   codes



April 9,9, 2008
April 2008                 MDfR-Mechanical Design for Reliability               73
                                   © 2006 Ops A La Carte




            What is random analysis?
        • Statistically based frequency-dependent
              dynamic loading




                          Time                                      Frequency

                  – Don’t know exactly what the load is at any
                    instant in time
                  – Do know the loading envelope

April 9,9, 2008
April 2008                 MDfR-Mechanical Design for Reliability               74
                                   © 2006 Ops A La Carte
What calls for random analyses?
             • Random/white noise events
               – Acoustic pressure loading
               – Rocket/jet engine exhaust
               – Rocket/jet engine thrust vibration
               – Road-loads/transportation
               – Earthquakes
             • Base vibration or Acoustic excitation

                  Loading defined in Frequency Domain
April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability               75
                                    © 2006 Ops A La Carte




     Frequency Response Analysis
             • Frequency Domain
                    – Steady-state harmonic response
                    – Load/Response varies w/ Frequency
                    – Magnitude and phase are important


                                                                         Tip
                     Base
1.



        Frequency                                                    Frequency



April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability               76
                                    © 2006 Ops A La Carte
How random analyses typically happen
    • Typically based on a PSDF spectrum


    ( ~ )2/Hz


                                           Frequency



            – Almost looks like Frequency Response
            – Loading is in terms of (something)2/Hz
April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability   77
                                    © 2006 Ops A La Carte




                    So what do you do???
                  • Frequency response analysis
                     – Apply 1 unit of what is specified in
                       (~)2/Hz across all frequencies
                  • Define input spectrum
                     – table of (~)2/Hz vs. Frequency
                  • Do the math…
                     – This is where it all comes together
                  • Data recovery
                     – Know what you are looking for/at

April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability   78
                                    © 2006 Ops A La Carte
Frequency Response Post-Processing
                                        Frequency Response




              1.                                                 10.



                       Frequency                                         Frequency
 Input spectrum                                                                           Output
scales like this…                                                                     spectrum scales
                                                                                        similarly…
            5.8
                                                              580.




                          Frequency                                     Frequency

                                         Random Response
April 9,9, 2008
April 2008                           MDfR-Mechanical Design for Reliability                             79
                                             © 2006 Ops A La Carte




  Frequency Response Post-Processing
                                    Frequency Response




                  1.



                        Frequency                                         Frequency




                            Frequency                                     Frequency


                                    Random Response
April 9,9, 2008
April 2008                           MDfR-Mechanical Design for Reliability                             80
                                             © 2006 Ops A La Carte
Random Results: What to look
               for
  • RMS values
          – Area under curve
  • Number of positive crossings                                    Frequency


          – Apparent frequency for fatigue
  • Cumulative RMS
          – Which frequencies contribute the most



April 9,9, 2008
April 2008                 MDfR-Mechanical Design for Reliability               81
                                   © 2006 Ops A La Carte




             So what is MSC.Random?
        • User-developed MSC.Patran application
                  – Customer-requested features and
                    functions
        • Officially available since MSC.Patran v9
                  – Available under Utilities/Analysis menu
        • Improved User Interface for:
                  – Frequency Response Analysis
                  – Interactive Random Analysis
                  – Post-processing
April 9,9, 2008
April 2008                 MDfR-Mechanical Design for Reliability               82
                                   © 2006 Ops A La Carte
So what is MSC.Random?
        • Improved frequency response user
          interface
                  – Sets up frequency response analysis
                  – Unit loading(s)
                  – Large mass, force or acoustic pressure
                  – Support for FREQi, I=1,5 entries
        • .xdb access of frequency response
              results
                  – Fast & efficient

April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability   83
                                    © 2006 Ops A La Carte




             So what is MSC.Random?
        • Interactive Random Analysis
                  – Calculations done “on the fly” (w/o having
                    to re-run MSC.Nastran)
                  – RMS fringe plots
                  – Specialized XY-Plotting
        • Interactive v. batch environment
                  – Change your random analysis inputs w/out
                    re-running MSC.Nastran


April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability   84
                                    © 2006 Ops A La Carte
MSC.Random Benefits
        • Simplified analysis setup
                  –   GUI interface
                  –   Engineer-driven specialized forms
                  –   No XYPLOT entries required
                  –   Full on-line help
        • Improved productivity
                  – .xdb access of results
                  – Re-analyze without restarting
        • Quality results
                  – Specialized interactive plots
                  – Log-log integration
                  – Cumulative RMS

April 9,9, 2008
April 2008                      MDfR-Mechanical Design for Reliability             85
                                        © 2006 Ops A La Carte




                               New Patran                                Old GUI
     New GUI
                                Random
                              • Similar over all operation.
                              • New interface has more
                                options and controls.
                              • New interface has easier and
                                more flexible ways to input
                                data.
                              • More result types supported
                              • New GUI allows more room to
                                add additional features.
                              • Incorporates many
                                enhancements requests.


April 9,9, 2008
April 2008                      MDfR-Mechanical Design for Reliability             86
                                        © 2006 Ops A La Carte
Patran Random Enhancements
                                     XY plot                                    Num. Pos.
                                                                 Node Ids       Crossings
                                                                                (NEW)




User Custom Title      Reflector Clearance



                                                                        RMS Value (NEW)



                                                                        RMS Scale (NEW)




 April 9,9, 2008
 April 2008             MDfR-Mechanical Design for Reliability                              87
                                © 2006 Ops A La Carte




                   Tolerance & Worst Case
                          Analysis




 April 9,9, 2008
 April 2008             MDfR-Mechanical Design for Reliability                              88
                                © 2006 Ops A La Carte
Worst-Case vs. Statistical Tolerancing
• When solving tolerancing problems, one must
  choose between worst-case tolerancing and
  statistical tolerancing. Both of these methods
  have their pros and cons.
• If worst-case tolerancing is used, all tolerances
  must be specified as worst-case tolerances. If
  statistical tolerancing is used, all tolerances must
  be specified as statistical tolerances.
• In reality, the behavior of certain inputs is best
  described using worst-case tolerances while the
  behavior of other inputs is best described by
  statistical tolerances. Still other inputs are not
  adequately represented by either type of
  tolerance.
    Ref: “Process Tolerancing: A Solution to the Dilemma of Worst-Case vs. Statistical Tolerancing”, Wayne A. Taylor

April 9,9, 2008
April 2008                         MDfR-Mechanical Design for Reliability                                    89
                                            © 2006 Ops A La Carte




    Tolerance & Worst Case Analysis
 Traditional Design
• Tolerances specify design requirements.
• Tolerances control design parameters by
  defining how much variation is acceptable.
• Tolerances give insight into the designer’s intent.


Ref: Scott Burr, 2005, “Design for Six Sigma” Hubenthal-Burr Associates




April 9,9, 2008
April 2008                         MDfR-Mechanical Design for Reliability                                    90
                                            © 2006 Ops A La Carte
Variation
Sources of Variation
• Processes
      – Variation occurs when a part is processed.
      – Different processes often have different inherent capabilities.
• Raw Materials
      – Inhomogeniety
      – Property variation
• Environment
      – Temperature and Pressure Variation
      – Noise
      – Vibration
• Product Degradation
      – Due to Heat
      – Component variation through time
      – Stresses and failure mechanisms

    Traditional Design is unconcerned with Process Variation as long as
    you meet SPEC!!

April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability             91
                                  © 2006 Ops A La Carte




Variation in Mechanical Assemblies
Three sources of variation occur in
 mechanical assemblies
• Dimensional variations (lengths & angles)
• Geometric form and feature variations (flatness,
  roundness, angularity, etc.)
• Kinematic variations (small adjustments between
  mating parts)

Ref: “A Comprehensive System for Computer-Aided Tolerance Analysis of 2-D and 3-D
   Mechanical Assemblies” , Kenneth Chase, et.al.


April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability             92
                                  © 2006 Ops A La Carte
There is Good News
                      and Bad News
• Tolerancing has NO influence on Variation
• You CANNOT count on Inspection to assure your
  design parameters and tolerances.
• Proper Design Choice is a highly effective
  countermeasure against Variation
• FEA Simulation is a highly effective method for
  evaluating Design Choices, Variation and linking
  process parameters to design performance.

Ref: Scott Burr, 2005, “Design for Six Sigma” Hubenthal-Burr Associates

 April 9,9, 2008
 April 2008                MDfR-Mechanical Design for Reliability         93
                                   © 2006 Ops A La Carte




                                    Lunch!!




 April 9,9, 2008
 April 2008                MDfR-Mechanical Design for Reliability         94
                                   © 2006 Ops A La Carte
Thermal Analysis
                  & Electronic Cooling
      Wireless Packages




April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability   95
                                  © 2006 Ops A La Carte




  Thermal Airflow Comparison
Laminar vs. Turbulent flow regimes
• Laminar airflow
     – Possible overheating.
     – Insufficient temperature gradient


• Turbulent airflow
     – Heat transfer greatly enhanced.
     – Boundary layer dissipated.



April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability   96
                                  © 2006 Ops A La Carte
Thermal Airflow Comparison

                      Laminar Flow                               Turbulent Flow
           Ta                                           Ta
                                Tb                                             Tb


           Ta                                           Ta

                          Tb>Ta                                        Ta=Tb
         Heat
         Transfer         104 BTU/Hr                                    106 BTU/Hr
    April 9,9, 2008
    April 2008                MDfR-Mechanical Design for Reliability                      97
                                      © 2006 Ops A La Carte




            Why does Turbulence Enhance
                  Heat Transfer?




•      Turbulent boundary layer is thin compared to laminar
•      Turbulence has unsteady (time varying) velocity & pressure
       components
•      Turbulent flows include eddies & vortices; “eddy transport” is a major
       heat transfer factor
•      Turbulent flow has shorter entry length for flow to become fully
       developed
•      These features enhance the heat and mass transfer in the turbulent
       regime
Ref: “Component or PCB Oriented Thermal Design” http://www.coolingzone.com/library.php?read=530
    April 9,9, 2008
    April 2008                MDfR-Mechanical Design for Reliability                      98
                                      © 2006 Ops A La Carte
How do you Achieve Turbulent Flow?
• Typically easy in an electronics enclosure
• “Roughness” associated with components
  (leads, packages, capacitors, etc.) trips the
  boundary layer
• Add ribs, helical ridging or similar
• Related to the characteristic Re (Reynolds
  number)


 April 9,9, 2008
 April 2008                 MDfR-Mechanical Design for Reliability                  99
                                    © 2006 Ops A La Carte




                   PCB Heat Transfer
• Thermal vias in the package and board serve as
  significant heat flow path from the top plane of the
  package to the top plane of the board
• Entry flow effects are important
Ref: “So Many Chips, So Little Time, Device Temperature Prediction in Multi-Chip
   Packages”, http://electronics-cooling.com/articles/2006/2006_aug_cc.php




 April 9,9, 2008
 April 2008                 MDfR-Mechanical Design for Reliability                 100
                                    © 2006 Ops A La Carte
Cell Phone PCB Thermal Analysis




April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability      101
                                  © 2006 Ops A La Carte




                      Thermal Design
• Both heat and cold adversely affect circuits
• Low temperatures cause different
  problems
• Thermal cycling can lead to fatigue
• Thermal shock     severe

Ref: “Hot, Cold, and Broken: Thermal Design Techniques”, 2007, EDN,
   http://www.edn.com/article/CA6426879.html



April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability      102
                                  © 2006 Ops A La Carte
System Thermal Design
• Airflow (forced convection) can reduce
  thermal resistance up to 5X
• Fans add noise; need to be sized
  based on volume flow in the system
• Ambient temperature effect
• Turbulence and natural convection
  Ref: “Turbulent Heat Transfer in
  Buoyancy-Driven Natural Convection
  in Vertical Enclosures”


Ref:
www.ceere.org/beep/docs/FY2002/Turbulent_Flow_in_Enclosure.pdf




April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability           103
                                    © 2006 Ops A La Carte




        Thermal / Structural / Drop Test
          Analysis of a Circuit Board
• Determine Temperature distribution and stresses due
  to CTE Mismatch.
• Determine dynamic response due to frequency
  response and random excitation
• Determine acceleration response and stresses due to
  Drop Testing


                  IC                                                 Leads

                                                                     PWB
     Connector

April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability           104
                                    © 2006 Ops A La Carte
Circuit Board Demonstration

                                Thermal
                                Analysis                        Structural
                                                                 Warping
                                                                                                   Modal
                         Determine Temperature              Determine Stresses                    Analysis
                         Distribution in board              due to CTE mismatch

                                                                                       Evaluate stress
                                                                                       and deformation



                                                                       Determine Response
                                                                       Levels due to input PSD
                                                                                                          Frequency
                               Determine stresses                                                         Response
                               due to Drop Testing
                                                                                                          Analysis

                                             Drop                  Random
                                             Test                  Vibration


    April 9,9, 2008
    April 2008                     MDfR-Mechanical Design for Reliability                                      105
                                               © 2006 Ops A La Carte




        Summary - MD Nastran Workflow
•    MD Nastran eliminates two software applications from the process.
                                                               Step 1                            Ref: MSC Software

                                                         Thermal Analysis


                                                              Step 2
                                MD Nastran




                                                        Structural Warping


                                                               Step 3
                                                          Modal Analysis


                                                               Step 4
                                                    Frequency Response Analysis


                                                               Step 5
                                                         Random Vibration


                                                               Step 6
                                                             Drop Test

    April 9,9, 2008
    April 2008                     MDfR-Mechanical Design for Reliability                                      106
                                               © 2006 Ops A La Carte
References – Thermal & Cooling
 • Electronic Cooling Solutions
       http://www.ecooling.com/references.html
 • Electronics Cooling
       http://www.electronics-cooling.com
 • Coolingzone
       http://www.coolingzone.com/library.php




April 9,9, 2008
April 2008           MDfR-Mechanical Design for Reliability   107
                             © 2006 Ops A La Carte




 Probabilistic Design System (PDS)
• Essentially a DOE using finite element analysis
• Key variables are specified with some sort of
  distribution (e.g., Gaussian distribution, etc.)
• Series of simulations are analyzed varying the
  parameters within the specified distributions
• Terms such as “stochastic analysis” are also
  used



April 9,9, 2008
April 2008           MDfR-Mechanical Design for Reliability   108
                             © 2006 Ops A La Carte
Probabilistic Design System (PDS)
 • Tools like Engineous iSight can be used to drive
   other applications and finite element codes to
   give a PDS
 • ANSYS includes PDS capabilities in the
   Workbench
 • MSC Adams Insight provides excellent DOE
   and Response Surface capabilities
 • MSC.Software MD-Nastran is adding stochastic
   and DOE capability for multi-disciplinary
   applications

April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability   109
                                  © 2006 Ops A La Carte




    Sources of Uncertainty in Engineering
Physical uncertainty
      Material properties
                                      F                                  F
      Boundary and initial conditions
      Assembly and manufacturing imperfections
      Environmental loads
      Geometry


Non-physical uncertainty
      Modelling errors (material model, element type, etc.)
      Choice of method (linear, non-linear, etc.)
                                 non-
      Solver
      Computer
      Engineer


April 9,9, 2008
April 2008                MDfR-Mechanical Design for Reliability   110
                                  © 2006 Ops A La Carte
Definition of a Stochastic Problem
                                    Model




                                                               Random response




Multiple executions
                                                                    Single computer run
are a Simulation                                                        is an Analysis




 April 9,9, 2008
 April 2008           MDfR-Mechanical Design for Reliability                     111
                              © 2006 Ops A La Carte




              Introduction to Stochastics




 April 9,9, 2008
 April 2008           MDfR-Mechanical Design for Reliability                     112
                              © 2006 Ops A La Carte
Variation in Design
• Traditional Simulation                            • Reality
      – “Deterministic”                                    – No two examples of a
      – All parameters are known                             product are exactly alike
        precisely                                          – No two tests will produce
      – One solution – uniquely                              the same results
        determined                                         – It’s not always obvious why
      – Evaluation of whether a                              some samples pass
        design is “good” is usually                          acceptance tests while
        based on nominal                                     others fail
        conditions                                         – Simulations from two
                                                             different engineers will give
                                                             slightly different results


April 9,9, 2008
April 2008                 MDfR-Mechanical Design for Reliability                    113
                                   © 2006 Ops A La Carte




              How Do Designers Deal with this
                       Variation?
• By trying to eliminate it.
• Make reality behave like a deterministic simulation rather
  than making the simulation a better reflection of reality
        –    Tight tolerances
        –    Elaborate process specifications
        –    Statistical quality control
        –    Conservatism factors
        –    Scrap parts that show “too much” variation
        –    Expensive and time consuming tests to determine “real world”
             performance
• The result?
        – Reduces but doesn’t eliminate variation
        – Can be very expensive
April 9,9, 2008
April 2008                 MDfR-Mechanical Design for Reliability                    114
                                   © 2006 Ops A La Carte
The Alternative
• Make simulations vary like reality
• Perform multiple simulations, each with slightly
  different parameter values that reflect the
  variations that occur in actual manufacturing and
  use
• Multiple simulations provide a distribution of
  outcomes, rather than a single outcome
• Evaluate which variations have a large impact
  on performance and reliability and should be
  controlled and which have a small impact and
  can be allowed to vary over a wider range
• This is the idea behind stochastic simulation
April 9,9, 2008
April 2008         MDfR-Mechanical Design for Reliability   115
                           © 2006 Ops A La Carte




                       Stochastic
• sto·chas·tic
  1 : involving a random variable <a
  stochastic process>
  2 : involving chance or probability <a
  stochastic model of radiation-induced
  mutation>
• Source: Merriam-Webster's Medical
  Dictionary, © 2002 Merriam-Webster, Inc.

April 9,9, 2008
April 2008         MDfR-Mechanical Design for Reliability   116
                           © 2006 Ops A La Carte
Why Stochastics Makes Sense for
          Electronics Applications
       (and many other applications!)




April 9,9, 2008
April 2008                    MDfR-Mechanical Design for Reliability   117
                                      © 2006 Ops A La Carte




                   Electronics Applications
  • Examples:
          –       Computer equipment and components
          –       Cell phones and communications equipment
          –       Biomedical devices
          –       Appliances
          –       Automotive and aerospace components
          –       Manufacturing equipment
  •     Typically have large production volume
  •     Emphasis on miniaturization
  •     High cost of failure (liability, warranty)
  •     Operate in a variety of frequently harsh and
        uncontrolled environments

April 9,9, 2008
April 2008                    MDfR-Mechanical Design for Reliability   118
                                      © 2006 Ops A La Carte
Design Objective for Electronics
                  Devices
• The design objective for electronics
  devices is typically a probabilistic one
• For example, mean time between failure
  (MTBF) or probability of failure (POF)
• Translating deterministic results to
  probabilistic terms is difficult at best
• Probabilistic design objectives require
  probabilistic design techniques
April 9,9, 2008
April 2008                 MDfR-Mechanical Design for Reliability       119
                                   © 2006 Ops A La Carte




                        Business Value
• Stochastics provides:
      – Time to Market
              • Efficient design space surveys
              • Shorten development time by enabling engineers to analyze
                results, not run tests
      – Reduce warranty/risk
              • Greater understanding of key parameters in the design
              • Design for manufacture
      – Improve perceived quality/robustness
              • Fewer returns
              • Fewer complaints
              • Less warranty costs

April 9,9, 2008
April 2008                 MDfR-Mechanical Design for Reliability       120
                                   © 2006 Ops A La Carte
Stochastic Crash Analysis




April 9,9, 2008
April 2008           MDfR-Mechanical Design for Reliability   121
                             © 2006 Ops A La Carte




                  MSC.ADAMS/Insight
• Insight is a framework for performing stochastic
  simulation
• Insight was designed to be flexible enough in its
  implementation that it can work with virtually any
  simulation or analysis tool
• When coupled to MSC.SimManager, a web-
  based portal for simulation management, it can
  provide efficient stochastic simulations to an
  entire enterprise

April 9,9, 2008
April 2008           MDfR-Mechanical Design for Reliability   122
                             © 2006 Ops A La Carte
How Does It Work?
• The user defines parameters (called “factors”) which vary
  and the way they vary (a list of potential discrete values,
  a probability distribution, etc.)
• The user defines responses, i.e. – dependent variables
  that vary as a result of the variation in the input factors
• The user defines how the input factors and output
  responses are related to the simulation model
• Insight then performs multiple simulations.
      – Each simulation has a set of input factors whose values are
        automatically selected based on a variety of methods
      – The responses of each simulation are collected
      – Essentially “virtual tests” of the design
• The result is a statistical sampling of inputs and the
  resulting outputs

April 9,9, 2008
April 2008               MDfR-Mechanical Design for Reliability       123
                                 © 2006 Ops A La Carte




         What Does Insight Provide?
• The selection of factors for each stochastic simulation
  can be based on Latin Hypercube or Monte Carlo
  methods
        – Other non-stochastic virtual experimental design methods,
          such as Full Factorial, Central Composite, Box-Behnken,
          Placket Burman, Sweep Study, D-Optimal, and Design of
          Experiments (DOE) are also available
• Statistical analysis of the variation in responses to the
  variation in the inputs
• Response surfaces that represent the relationship
  between factors and responses of the system as curve-
  fit polynomials
• Statistical data that can be exported to tools such as
  Excel, Visual Basic, and MATLAB
• Export results to a web server

April 9,9, 2008
April 2008               MDfR-Mechanical Design for Reliability       124
                                 © 2006 Ops A La Carte
Typical Output Using Insight’s Web Interface

                                                                 Includes
                                                                 response
                                                                 surface data.
                                                                 “+” and “-”
                                                                 buttons enable
                                                                 the user to
                                                                 interactively
                                                                 change values
                                                                 of factors see
                                                                 their effect on
                                                                 the responses.




April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability          125
                                © 2006 Ops A La Carte




                  What are the Benefits?
• Response surfaces allow a user to select values for
  factors needed to achieve a desired response
• Statistical analysis identifies which factors have the
  greatest affect on responses and can provide the
  greatest benefit
      – Relax tolerances on design parameters that have little affect
      – Tighten tolerances on those that have a greater affect
      – Assess the probability of a specific outcome (e.g. – “stress will be
        greater than allowable”)
• Physical tests can be more rapidly correlated by
  comparison to the results of “virtual” tests
• Designs can be automatically improved using “stochastic
  design improvement” (SDI) techniques


April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability          126
                                © 2006 Ops A La Carte
What are the Applications?
• Drop Testing
      – Vary impact energy and orientations, material
        properties at high strain rates
• Ball Grid Array Durability
      – Vary thermal input, ball size and shape, load on
        individual balls in the array, solder material properties
        at temperature
• Mechanical Systems Reliability
      – Vary manufacturing tolerances, actuator forces,
        freeplay
• Controller Algorithm Development
      – Vary gains, phase lags, sensor accuracy

April 9,9, 2008
April 2008            MDfR-Mechanical Design for Reliability   127
                              © 2006 Ops A La Carte




                    An Example:
                  BGA-Ball Grid Array




April 9,9, 2008
April 2008            MDfR-Mechanical Design for Reliability   128
                              © 2006 Ops A La Carte
The Problem
• Fatigue analysis of a single solder ball
• Loads come from a less detailed model of a ball
  grid array. This is the most severely loaded ball
  in the array.
• Shape and size of solder balls varies. For very
  small solder balls, even small variations can
  have significant impact on strain and hence
  fatigue life
• Fatigue methodology for solder under plastic
  deformation is semi-empirical and subject to
  significant scatter

April 9,9, 2008
April 2008             MDfR-Mechanical Design for Reliability   129
                               © 2006 Ops A La Carte




                     The Approach
• Create a parametric model of the solder ball
  using MSC.Patran software
        – Height, diameter, the size of small fillets at the top
          and bottom of the solder ball, the load on the solder
          ball, the yield strength of the solder
• Use Insight to define normal distributions for
  each of these parameters
• Conduct 128 finite element simulations with
  MSC.Marc using Monte Carlo methodology to
  sample the design space
• Determine the statistical distribution of plastic
  strain, and hence, fatigue life

April 9,9, 2008
April 2008             MDfR-Mechanical Design for Reliability   130
                               © 2006 Ops A La Carte
The Process
                                Define
       Define Parametric                              Define Responses     Monte Carlo
                           Distributions for
             Model                                        of Interest       Sampling
                             Parameters




                                                                         MSC.Patran FEA
                                            Response
                                            Surfaces


                                                                          MSC.Marc FEA

                               Factor/
                              Response
                             Correlations
                                                                           Database of
                                                                            Factor and
                                                                         Response Values

                                          Outcome
                                         Probabilities


April 9,9, 2008
April 2008                   MDfR-Mechanical Design for Reliability                         131
                                     © 2006 Ops A La Carte




  Solder Ball – Parameterized Geometry



     Fillet Height
                                                                                           Height

                                               Diameter




April 9,9, 2008
April 2008                   MDfR-Mechanical Design for Reliability                         132
                                     © 2006 Ops A La Carte
Solder Ball
                   Finite Element Model




April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability   133
                                © 2006 Ops A La Carte




                  Solder Ball – Total Strain




April 9,9, 2008
April 2008              MDfR-Mechanical Design for Reliability   134
                                © 2006 Ops A La Carte
Solder Ball – Plastic Strain




April 9,9, 2008
April 2008         MDfR-Mechanical Design for Reliability   135
                           © 2006 Ops A La Carte




          Work Space – Inputs and
        Responses (only 36 out of 128
                displayed)




April 9,9, 2008
April 2008         MDfR-Mechanical Design for Reliability   136
                           © 2006 Ops A La Carte
Work Space Review




April 9,9, 2008
April 2008           MDfR-Mechanical Design for Reliability   137
                             © 2006 Ops A La Carte




         Influence of Input Variables on
                Max. Plastic Strain




April 9,9, 2008
April 2008           MDfR-Mechanical Design for Reliability   138
                             © 2006 Ops A La Carte
Work Space Correlations
                   (Pearson Correlation)




April 9,9, 2008
April 2008             MDfR-Mechanical Design for Reliability   139
                               © 2006 Ops A La Carte




                  Work Space Correlations
                   Input vs. Responses




April 9,9, 2008
April 2008             MDfR-Mechanical Design for Reliability   140
                               © 2006 Ops A La Carte
Work Space Scatter Plot
     Diameter vs. Max. Displacement




April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability   141
                          © 2006 Ops A La Carte




     Work Space Scatter Plot
 Mesh Factor vs. Max. Plastic Strain




April 9,9, 2008
April 2008        MDfR-Mechanical Design for Reliability   142
                          © 2006 Ops A La Carte
Probabilistic Design System (PDS)
          with ANSYS FEA




April 9,9, 2008
April 2008          MDfR-Mechanical Design for Reliability            143
                            © 2006 Ops A La Carte




                    Stress Results
      Max.Principal Stress                              Hoop Stress




April 9,9, 2008
April 2008          MDfR-Mechanical Design for Reliability            144
                            © 2006 Ops A La Carte
Rank Order Correlation
  Sensitivities & Histogram of Stress
                                                            – Wrist Pin ID has the greatest
                                                              influence on the fatigue stress




April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability                     145
                                    © 2006 Ops A La Carte




          Probabilistic Simulations using
                ANSYS to DFSS
              •   Introduction
              •   An Example
              •   Motivation
              •   Features
              •   Benefits
              •   Probabilistic Methods
              •   Probabilistic Results/Interpretation
              •   Summary
              Ref: 2004, ANSYS DFSS Workshop

April 9,9, 2008
April 2008                  MDfR-Mechanical Design for Reliability                     146
                                    © 2006 Ops A La Carte
Introduction
Purpose of a Probabilistic Design System (PDS)

         Input
          Input                  ANSYS
                                 ANSYS                             Output
                                                                   Output
Material Properties                                          Deformation
Geometry                                                     Stresses / Strains
Boundary Conditions                                          Fatigue, Creep,...

                     It’s a reality that input
                  parameters are subjected to
                  scatter => automatically the
                     output parameters are
                        uncertain as well!!
April 9,9, 2008
April 2008          MDfR-Mechanical Design for Reliability                147
                            © 2006 Ops A La Carte




                     Introduction
  Purpose of Probabilistic Design System (PDS)


                      ANSYS PDS
                      ANSYS PDS

Questions answered with probabilistic design:

• How large is the scatter of the output parameters?
• What is the probability that output parameters do not
  fulfill design criteria (failure probability)?
• How much does the scatter of the input parameters
  contribute to the scatter of the output (sensitivities)?


April 9,9, 2008
April 2008          MDfR-Mechanical Design for Reliability                148
                            © 2006 Ops A La Carte
An Example
                          Example: Lifetime of Components !!!
Random input           Finite-Element             Random output
  variables                 Model                   parameters
   Material                                       • LCF lifetime
 • Strength                                       • Creep lifetime
 • Material
                                                  • Corrosion lifetime
   Properties
                                                  • Fracture mechanical lifetime
    Loads                                         •…
 • Thermal
 • Structural                                              Evaluate reliability of products !
  Geometry/
  Tolerances                                               Evaluate quality of products !

   Boundary                                                Evaluate warranty costs !
  Conditions
 • Gaps                                            To evaluate is the first step
 • Fixity
                                                   to improvement !

 April 9,9, 2008
 April 2008                    MDfR-Mechanical Design for Reliability                       149
                                        © 2006 Ops A La Carte




                                   Motivation
 Influence of Young’s Modulus and Thermal Expansion
 Coefficient on thermal stresses:
     thermal = E · · T

 Deterministic Approach:
  Emean and mean => evaluate expected value:                                       expect


 Probabilistic Approach:

                                   P(     thermal >
                                                1.05 expect) P( thermal > 1.10 expect)
   ‘E’ scatters           ±5%            16% (~1 out of 6)      2.3% (~1 out of 40)
   ‘E’ and ‘       ‘ scatter ±5%         22% (~1 out of 5)                 8% (~1 out of 12)

 April 9,9, 2008
 April 2008                    MDfR-Mechanical Design for Reliability                       150
                                        © 2006 Ops A La Carte
Scatter in Material Properties
                     and Loads
            Property                                        SD/Mean %
            Metallic materiales, yield                      15                 Source: Klein, Schueller
                                                                               et.al. Probabilistic Approach
            Carbon fiber composites, rupture                17                 to Structural Factors of
                                                                               Safety in Aerospace.
            Metallic shells, buckling strength              14                 Proc. CNES Spacecraft
                                                                               Structures and Mechanical
            Junction by screws, rivet, welding              8                  Testing Conf., Paris 1994

            Bond insert, axial load                         12
            Honeycomb, tension                              16
            Honeycomb, shear, compression                   10
            Honeycomb, face wrinkling                       8
            Launch vehicle , thrust                         5
            Transient loads                                 50
            Thermal loads                                   7.5
            Deployment shock                                10
            Acoustic loads                                  40
            Vibration loads                                 20
April 9,9, 2008
April 2008                       MDfR-Mechanical Design for Reliability                                151
                                         © 2006 Ops A La Carte




                                      Motivation
                                      ±5-100%
                                                     Materials,
                                                      Bound.-
                                                      Cond.,
                                                     Loads, ...
                                      Thermal
                                      Analysis
CAD                                                  FEM
Geometry
                       CFD                                                LCF                  ±??%
± 0.1-10%
                                      Structural
                                                     FEM
                       Materials,
                                      Analysis                            Materials
                        Bound.-
                       Cond., ...             Materials,                  ±30-60%
                                               Bound.-
                        ±5-50%
                                               Cond.,
                                      ±5-100% Loads, ...

April 9,9, 2008
April 2008                       MDfR-Mechanical Design for Reliability                                152
                                         © 2006 Ops A La Carte
PDS Benefits
Deterministic Analysis:                         Probabilistic Analysis:
• Only provides a YES/NO answer                   • Provides a probability and reliability
                                                    (design for reliability)
• Safety margins are piled up                     • Takes uncertainties into account in a
  “blindly” (worst material, maximum                realistic fashion => This is closer to
  load, … worst case)                               reality => Over-design is avoided
  1 worst case assumption        =10-2
  2 worst case assumptions =10-4
  3 worst case assumptions =10-6
  4 worst case assumptions =10-8
  ...
   => Leads to costly over-design                 • “Tolerance stack-up” is included
• Only “as planned“, “as is” or the                 (design for manufacturability)
  worst design


 April 9,9, 2008
 April 2008               MDfR-Mechanical Design for Reliability                       153
                                  © 2006 Ops A La Carte




                         PDS Benefits
Deterministic Analysis:                            Probabilistic Analysis:
• Sensitivities do not take range of                 • Range/width of scatter is “built-in” into
  scatter or possibilities into account                probabilistic sensitivities
• Sensitivities do not take
  interactions between input                         • Interactions between input variables
  variables into account (second                       are inherently taken care of
  order cross terms)




 April 9,9, 2008
 April 2008               MDfR-Mechanical Design for Reliability                       154
                                  © 2006 Ops A La Carte
PDS Benefits
                    Illustration of the Benefits of
Probabilistic Analysis over Deterministic Analysis
                                                                     Probabilistic Analysis




                                                       Deterministic Analysis

 April 9,9, 2008
 April 2008                 MDfR-Mechanical Design for Reliability                            155
                                    © 2006 Ops A La Carte




                   ANSYS PDS Features
  • Works with any ANSYS analysis model
            • Static, dynamic, linear, non-linear, thermal, structural, electro-
              magnetic, CFD ..
  • Allows large number random input and output
    parameters
  • 10 statistical distributions for random input parameters
  • Random input parameters can be correlated
  • Probabilistic methods:
            • Monte Carlo - Direct & Latin Hypercube Sampling
            • Response Surface - Central Composite & Box-Behnken
              Designs
 April 9,9, 2008
 April 2008                 MDfR-Mechanical Design for Reliability                            156
                                    © 2006 Ops A La Carte
ANSYS PDS Features
• Use of distributed, parallel computing techniques for
  drastically reduced wall clock time
• Comprehensive probabilistic results
          • Convergence plots, histogram, probabilities, scatter
            plots, sensitivities, ...
• State-of-the art statistical procedures to address the
  accuracy of the output data
          • Confidence intervals




April 9,9, 2008
April 2008             MDfR-Mechanical Design for Reliability            157
                               © 2006 Ops A La Carte




                  Probabilistic Methods
Monte Carlo Simulation:                                                        Fully Parallel
    Perform numerous analysis runs based
    on sets of random samples, and then
    evaluate statistics of derived responses.

      • Direct (Crude) Sampling Monte Carlo                      (DIR)
      • Latin Hypercube Sampling Monte Carlo                    (LHS)
      • User defined                                            (USR)



April 9,9, 2008
April 2008             MDfR-Mechanical Design for Reliability            158
                               © 2006 Ops A La Carte
Probabilistic Methods
Monte Carlo Simulation Method Scheme:
Simulation of input                                                     Statistical analysis of
   parameters at                                                         output parameters
 random locations




  X1              X2      X3


                               Repetitions = Simulations



                                         ANSYS
                                         ANSYS

April 9,9, 2008
April 2008                     MDfR-Mechanical Design for Reliability                        159
                                       © 2006 Ops A La Carte




                  Probabilistic Methods
  Finite Element Runs for Monte Carlo
  For Monte Carlo Simulation the number of simulations does not
  depend on the number of random input variables, but on the
  probabilistic result you are looking for:

  • For assessment of the statistics of output parameters (Mean, sigma)
                 Nsim 30 … 100
  • For histogram and cumulative distribution function
                 Nsim 50 … 200
  • For assessment of low probabilities P (tails of the distribution)
                 Nsim 30/P … 100/P




April 9,9, 2008
April 2008                     MDfR-Mechanical Design for Reliability                        160
                                       © 2006 Ops A La Carte
Probabilistic Methods
 – Response Surface Methods:




                                                                                     Fully Parallel
     Select specific observation points for
     each random variable, run analyses,
     establish response surface for each
     response parameter, perform Monte
     Carlo Analysis on Response Surface.
       • Central Composite Design                                  (CCD)
       • Box-Behnken Matrix                                        (BBM)
       • User defined                                              (USR)


  April 9,9, 2008
  April 2008             MDfR-Mechanical Design for Reliability                       161
                                  © 2006 Ops A La Carte




                     Probabilistic Methods
   Response Surface Methods Scheme:
Simulation of input parameters                                       Statistical analysis of
    at specific locations                                             output parameters



                             X1               X2           X3



   Evaluate input
    Evaluate input                                                Monte Carlo Simulations
                                                                  Monte Carlo Simulations
 parameter values
  parameter values                                                 on Response Surface
                                                                    on Response Surface
            DOE         Repetitions = Simulations                  Response Surface Fit
                                                                   Response Surface Fit


                                   ANSYS
                                   ANSYS

  April 9,9, 2008
  April 2008             MDfR-Mechanical Design for Reliability                       162
                                  © 2006 Ops A La Carte
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar
Ops a la carte mechanical design for reliability (mdfr) seminar

More Related Content

Featured

AI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfAI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfmarketingartwork
 
PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024Neil Kimberley
 
Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)contently
 
How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024Albert Qian
 
Social Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsSocial Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsKurio // The Social Media Age(ncy)
 
Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Search Engine Journal
 
5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summarySpeakerHub
 
ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd Clark Boyd
 
Getting into the tech field. what next
Getting into the tech field. what next Getting into the tech field. what next
Getting into the tech field. what next Tessa Mero
 
Google's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentGoogle's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentLily Ray
 
Time Management & Productivity - Best Practices
Time Management & Productivity -  Best PracticesTime Management & Productivity -  Best Practices
Time Management & Productivity - Best PracticesVit Horky
 
The six step guide to practical project management
The six step guide to practical project managementThe six step guide to practical project management
The six step guide to practical project managementMindGenius
 
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...RachelPearson36
 
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...Applitools
 
12 Ways to Increase Your Influence at Work
12 Ways to Increase Your Influence at Work12 Ways to Increase Your Influence at Work
12 Ways to Increase Your Influence at WorkGetSmarter
 

Featured (20)

AI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfAI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
 
Skeleton Culture Code
Skeleton Culture CodeSkeleton Culture Code
Skeleton Culture Code
 
PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024
 
Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)
 
How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024
 
Social Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsSocial Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie Insights
 
Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024
 
5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary
 
ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd
 
Getting into the tech field. what next
Getting into the tech field. what next Getting into the tech field. what next
Getting into the tech field. what next
 
Google's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentGoogle's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search Intent
 
How to have difficult conversations
How to have difficult conversations How to have difficult conversations
How to have difficult conversations
 
Introduction to Data Science
Introduction to Data ScienceIntroduction to Data Science
Introduction to Data Science
 
Time Management & Productivity - Best Practices
Time Management & Productivity -  Best PracticesTime Management & Productivity -  Best Practices
Time Management & Productivity - Best Practices
 
The six step guide to practical project management
The six step guide to practical project managementThe six step guide to practical project management
The six step guide to practical project management
 
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
 
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
 
12 Ways to Increase Your Influence at Work
12 Ways to Increase Your Influence at Work12 Ways to Increase Your Influence at Work
12 Ways to Increase Your Influence at Work
 
ChatGPT webinar slides
ChatGPT webinar slidesChatGPT webinar slides
ChatGPT webinar slides
 
More than Just Lines on a Map: Best Practices for U.S Bike Routes
More than Just Lines on a Map: Best Practices for U.S Bike RoutesMore than Just Lines on a Map: Best Practices for U.S Bike Routes
More than Just Lines on a Map: Best Practices for U.S Bike Routes
 

Ops a la carte mechanical design for reliability (mdfr) seminar

  • 1. & We Provide You Confidence in Your Product ReliabilityTM Ops A La Carte / (408) 654-0499 / askops@opsalacarte.com / www.opsalacarte.com
  • 2. MECHANICAL DESIGN for RELIABILITY (MDfR) SEMINAR April 9, 2008 T. Kim Parnell, Ph.D.,P.E. 408-203-9443 kimp@OpsALaCarte.com Ops A La Carte LLC www.OpsALaCarte.com © 2006 Ops A La Carte
  • 3. The following training and presentation materials are copyright protected by: Ops A La Carte LLC and Parnell Engineering & Consulting (PEC) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 2 © 2006 Ops A La Carte
  • 4. Presenter’s Biographical Sketch Kim Parnell ◈ Dr. Kim Parnell is a Senior Consultant with Ops A La Carte and Principal/Founder of PEC (www.parnell-eng.com), an engineering consulting firm that focuses on providing support for high-tech and medical device companies. He specializes in the mechanical engineering design and behavior of electronic and miniature components, MEMs, wireless sensors, telecommunications devices, biomedical devices, and shape memory metals. Dr. Parnell is an expert in the application of finite element analysis (FEA) to the solution of engineering problems, and consults actively in these areas as well as failure analysis and reliability. He served as a Visiting Associate Professor in the Mechanical Engineering Department at Stanford University and is a coach and mentor for the innovative Stanford Biodesign program. He is a member of the NanoBioConvergence (www.NanoBioConvergence.org) Board, and the CSIX Connect (www.csix.org) Board Dr. Parnell is a member of ASME, IEEE, ASM, and is currently Chair of the IEEE-CNSV (Consultants’ Network of Silicon Valley) (www.CaliforniaConsultants.org). ◈ He has worked with companies including Ops A La Carte, MSC Software, Rubicor Medical, Exponent Failure Analysis Associates, SST Systems, ATT Bell Laboratories, ◈ Stanford University, and General Motors. ◈ Dr. Parnell holds Ph.D. and MSME degrees from Stanford University in Mechanical Engineering and a BES from Georgia Tech. He is a registered Professional Mechanical ◈ Engineer in the State of California. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 3 © 2006 Ops A La Carte
  • 5. Mechanical Design for Reliability-MDfR T. Kim Parnell, PhD, PE 8:15 8:30 Setup 8:30 9:30 Introduction to Mechanical Reliability & Robust Design 9:30 11:00 Finite Element Analysis (FEA) 11:00 11:30 Random Vibration & Shock 11:30 12:00 Tolerance & Worst Case Analysis 12:00 1:00 Lunch 1:00 1:30 Thermal Analysis & Electronic Cooling 1:30 2:45 Probabilistic Design System (PDS) & Stochastics 2:45 3:00 Forensic Methodology 3:00 4:00 Failure Analysis & Prevention; Fatigue & Fracture 4:00 4:30 Conclusions & Discussion April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 4 © 2006 Ops A La Carte
  • 6. Introduction to Mechanical Reliability & Robust Design April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 5 © 2006 Ops A La Carte Six Sigma in Practice What does it mean? 99% Performance 6 Sigma Performance 99.999663% Meet Spec 202 Billion Pieces of Mail Delivered by USPS per year 2.02 Billion pieces Lost per Year 680740 pieces lost per year 24/7 Power Delivery to Your Home 87 Hours without Power every Year Less than two minutes without power per year 510 Million Prescriptions Worldwide per year 5.1 Million Wrong Prescriptions per Year 1719 Wrong Prescriptions per Year 27 Billion Credit Card Transactions per year 0.27 Billion wrong transactions per year 90990 wrong transactions per Year Ref: Scott Burr, 2005, “Design for Six Sigma” Hubenthal-Burr Associates April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 6 © 2006 Ops A La Carte
  • 7. Two Types of Quality Two Types of Quality • Type 1: Customer Quality - The features that customers want. • Type 2: Engineered Quality - The problems customers do not want. • Customer quality leads to the size of the market segment and includes items such as function, features, colors and designs. The better the customer quality, the bigger the market size becomes. In order to obtain the market size, the price must be reasonable. • Engineered quality includes defects, failures, noise, vibration, unwanted phenomena, lowering the cost of manufacture, and minimizing manufacturing problems. Ref: “Design for Six Sigma Roadmap”, Shree Phadnis April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 7 © 2006 Ops A La Carte Results & Meaning • Process provides snapshot of current system • No one tool makes an entire reliability program • Check step is critical before moving to recommendation around improvement plan April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 8 © 2006 Ops A La Carte
  • 8. Robust Design • Engineers dream of designing a product or process that exhibits the state of 'robustness.' However, few can actually make the claim that they know exactly what robustness means. Formal definition by quality guru Dr. Taguchi: • Robustness is the state where the technology, product, or process performance is minimally sensitive to factors causing variability (either in the manufacturing or user environment), and aging at the lowest manufacturing cost. Ref: “Design for Six Sigma Roadmap”, Shree Phadnis April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 9 © 2006 Ops A La Carte How to Achieve Robust Design? • “Brute Force” techniques – Intuitively select design parameters and tolerances and make system design trade- offs – Added design margins – Tighter tolerances – Does not work in this hyper competitive economy April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 10 © 2006 Ops A La Carte
  • 9. Cost to Repair Design Problems • As a rule, design engineering has lagged behind the shop floor in awareness of product and process quality. • Very real costs are associated with inattention to design quality. If errors or omissions in design data are not addressed early, more costly changes are required later in the product development process. Ref: “Building Quality into Design Engineering”, Gavin Finn April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 11 © 2006 Ops A La Carte Robust Design Outline • 21st century challenges • Tools & techniques for Design For Six Sigma (DFSS) • Definition of robustness – What, how, significance • Business impact of robust design solutions • Systematic approach to robust design • Ref: Ansys, 2004 April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 12 © 2006 Ops A La Carte
  • 10. 21st Century Challenges • Product life cycles are expected to last for just a few months • More customized products – 1700 car models as opposed to 900 ten years ago – More focus on “built-in-quality” or “built-in-reliability” • To meet these demanding requirements – Products need to be developed in the shortest amount of time and – They have to be safe, reliable, and competitive – Robust and Affordable ! “Product quality requires managerial, technological and statistical concepts throughout all the major functions of the organization …” Joseph M. Juran April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 13 © 2006 Ops A La Carte Tools & Techniques for Design for Six Sigma Planning Concept Optimization Design Tolerance VOC Development Verification Development Variation Analysis Manufacturing QFD Screening • Quality Function Deployment (QFD) – Customer requirements to technical specs • Failure Modes and Effects Analysis (FMEA) – To study failures, their impact and frequency • Other statistical tools for tolerance analysis, etc. • Tools for concept screening, design, optimization, verification • Probabilistic simulations for robust design – Design of Experiment (DOE), response surface April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 14 © 2006 Ops A La Carte
  • 11. What is Robust Design? • Works as intended regardless of variation in: – Design – Manufacturing process – Material – Resulting from Deterioration – Operation (misuse) • For Achieving robust design – Understand potential sources of variations – Quantify the effect of variations on product behavior and performance – Take steps to desensitize the design to these variations – Adjust the output to hit the target performance April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 15 © 2006 Ops A La Carte What is Robust Design? • De-sensitizing the design to variations April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 16 © 2006 Ops A La Carte
  • 12. How to Achieve Robust Design? • “Intelligent Design” by understanding – Product & Process design parameters that are critical to achieving a performance characteristic – What are the optimum values to both • Achieving the desired performance • Minimize the effect of naturally occurring and difficult to control variations (noise) – Requires a systematic approach of deploying engineering simulations as part of the design process April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 17 © 2006 Ops A La Carte Significance of Robust Design • Robust design optimization helps – Identify parameter values that maximize performance and minimize the effect of “noise” • More robust and affordable design – Identify parameters that have no significant effect on performance • Relaxed tolerance, reduces product cost – Identify parameter values that reduce cost without affecting performance or variation – Cost-effective quality inspection • Do not have to inspect for parameters that are not critical to performance April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 18 © 2006 Ops A La Carte
  • 13. Business Impact of Robust Design Solutions Profitability = Market Size x Market Share x Margin on Sales April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 19 © 2006 Ops A La Carte Market Size Increased market size by entering related markets. Some markets require product designs to be robust: – Defense, aerospace, jet engine, nuclear power, biomedical, oil industry, other mission critical solutions – Auto industry – 1st tier suppliers capturing more value April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 20 © 2006 Ops A La Carte
  • 14. Market Share • Increased market share (in slow growth industry) through – Better quality products • Less maintenance • Higher resale value – Quick response to customer requirements – Affordable pricing – Attract business by implementing DFSS April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 21 © 2006 Ops A La Carte Margin on Sales • Product Cost Reduction Through – Relaxed tolerances – Material savings – Cost-effective quality inspection – Less rejects (lower scrap rate) – Shorter development time – Less warranty cost – Customer loyalty • Command Premium price – Better quality – Timely delivery – Satisfied customers April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 22 © 2006 Ops A La Carte
  • 15. Reliability Philosophies Two fundamental methods to achieving high product reliability: – Build, Test, Fix – Analytical Approach April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 23 © 2006 Ops A La Carte Build, Test, Fix • In any design there are a finite number of flaws. • If we find them, we can remove the flaw. • Rapid prototyping • HALT • Large field trials or ‘beta’ testing • Reliability growth modeling April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 24 © 2006 Ops A La Carte
  • 16. Analytical Approach • Develop goals • Model expected failure mechanisms • Conduct accelerated life tests • Conduct reliability demonstration tests • Routinely update system level model • Balance of simulation/testing to increase ability of reliability model to predict field performance. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 25 © 2006 Ops A La Carte Issues with each approach Build, Test, Fix Analytical • Uncertain if design • Fix mostly known is good enough flaws • Limited prototypes • ALT’s take too long means limited flaws • RDT’s take even discovered longer • Unable to plan for • Models have large warranty or field uncertainty with new service technology and environments April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 26 © 2006 Ops A La Carte
  • 17. Drop Tests: What is the physics? • Is drop deterministic or stochastic? • Is drop predictable? • Is drop optimizable? • Does it make sense to speak of precision in drop simulations? • Do we need to increase the number of elements in our drop models? What is the reasonable limit? • What is the future of computer-based drop analysis? • Is drop a chaotic phenomenon? April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 27 © 2006 Ops A La Carte Electronic Component Drops April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 28 © 2006 Ops A La Carte
  • 18. Electronic Component Drops April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 29 © 2006 Ops A La Carte Electronic Component Drops April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 30 © 2006 Ops A La Carte
  • 19. Example of Measured Acceleration Signal • A series of tests for chaos are performed with this signal. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 31 © 2006 Ops A La Carte Log-linear Power Law • Systems that exhibit a log-linear Power Spectrum are potentially chaotic. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 32 © 2006 Ops A La Carte
  • 20. Typical Tests for Chaos • Hausdorff (Capacity dimension). Signal has fractal dimension (1.8). • Log-linear Power Spectrum (yes). • Correlation dimension (5). • Lyapunov Characteristic Exponents (+0.4). • Poincare’ sections or Return Maps (check for structure). • According to these tests, the measured drop/impact signal possesses a clear chaotic flavour. This explains why each drop is a unique event and cannot be optimised. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 33 © 2006 Ops A La Carte Drop = Chaos • Chaos can be described by closed-form deterministic equations. Chaos does NOT mean random. • Chaos is characterized by extreme sensitivity to initial conditions. • “Memory” of initial conditions is quickly lost in chaotic phenomena (“butterfly effect”). • Examples of chaotic phenomena: – Tornados (weather in general) – Stock market evolution, economy – Crash, drops, impacts, etc. – Earthquakes – Avalanches – Combustion/turbulence – EEG (alpha-waves in brain) – Duffing, Van der Pol, Lorenz oscillators, etc. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 34 © 2006 Ops A La Carte
  • 21. The Logistic Map • X(n+1) = k X(n)(1-X(n)) Shows astonishingly complex behaviour: – 0 < k < 1, Extinction regime – 1 < k < 3, Convergence regime – 3 < k < 3.57, Bifurcation regime – 3.57 < k < 4, Chaotic regime – 4 < k, Second chaotic regime For more details see: Weisstein, Eric W. "Logistic Map." From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/LogisticMap.html April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 35 © 2006 Ops A La Carte Chaos & Predictability • Phenomena that are chaotic, are unpredictable (nonrepeatable). The main reason is extreme sensitivity to initial conditions. • Phenomena that are unpredictable, cannot be optimized. They must be treated statistically. • All that can be done with chaotic phenomena is increase our understanding of their nature, properties, patterns, structure, main features, quantify the associated risks. • Models for Risk Analysis must be realistic to be of any use. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 36 © 2006 Ops A La Carte
  • 22. Understanding Risk • Essentially, risk is associated with the existence of outliers Outlier: - warranty - recall - lawsuit } Most likely response (highest density) Note: DOE and Response Surface techniques cannot capture outliers April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 37 © 2006 Ops A La Carte What is Risk and Uncertainty Management? • Understand and remove outliers • Shift entire distribution is safe fails Improved design Initial design Outliers: unfortunate combinations of operating Outlier conditions and design variables that lead to unexpected behaviour. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 38 © 2006 Ops A La Carte
  • 23. Example of Robust Design: MIR Space Station • Robustness = survivability in the face of unexpected changes in environment (exo) or within the system (endo) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 39 © 2006 Ops A La Carte Example of Optimal Design • M. Alboreto dies (Le Mans, April 2001) due to slight loss of pressure in left rear tire. The system was extremely sensitive to boundary conditions (was optimal, and therefore very very fragile!). April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 40 © 2006 Ops A La Carte
  • 24. Optimization: a Dangerous Game Second order RS First order RS Optimum? Different theories can be shown to fit the same set of observed data. The more complex a theory, the more credible it appears! April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 41 © 2006 Ops A La Carte Some Lessons • Boundary conditions are most important • Small effects can have macroscopic consequences (watch out for chaos, even in small doses!) • Precision is not everything! • Optimal components don’t give an optimal whole • Optimality = fragility • Robust is the opposite to optimal April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 42 © 2006 Ops A La Carte
  • 25. Conclusions – Robustness, Chaos, and Predictability • Phenomena that possess a chaotic component cannot be optimized, but can be improved in statistical sense. • With such systems, it is possible to address: – Risk Analysis – Design for robustness – Increase understanding • Realistic models necessitate: – continuous 3D random fields (geometry) – discrete random field (spotwelds, joints) – randomization of ALL material properties – randomization of ALL thicknesses – variations of boundary/initial conditions April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 43 © 2006 Ops A La Carte Finite Element Analysis (FEA) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 44 © 2006 Ops A La Carte
  • 26. Finite Element Analysis (FEA) • FEA is generally applicable for analyses such as stress, thermal, vibration, and dynamic cases 1) Create a geometric model; subdivide into elements to create a finite element mesh 2) Specify material properties for all components 3) Apply loads and boundary conditions: thermal, pressure, deadweight, wave loads, etc. • Linear analysis is typically used for design; nonlinear analysis is frequently required for failure investigation April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 45 © 2006 Ops A La Carte FEA & Testing • Finite element analysis (FEA) and physical testing are complementary • A comprehensive program needs to include both components • With judicious experimental validation, FEA can be used to reduce the amount of physical testing that is needed and shorten the design cycle April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 46 © 2006 Ops A La Carte
  • 27. FEA Simulation Tools FEA simulation can • Be a cost effective way to evaluate design choices. • Provide insight into how varying parameters affects the design outcome. • Simulate expensive processes where downtime is unacceptable. • Help establish the critical relationship between design parameters and process parameters. THIS IS BIG IN SIX SIGMA!!!! And can translate into an enormous and unfair Competitive Advantage. Ref: Scott Burr, 2005, “Design for Six Sigma”, Hubenthal-Burr Associates April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 47 © 2006 Ops A La Carte FEA Concepts • Linear Analysis – Small Deflection & Small Strain – Elastic material • Nonlinear Analysis – Large Deflection &/or Large Strain – Nonlinear material • Elastic/plastic • Rubber & Polymers • Temperature dependent properties – Contact – Shock & impact • Multi-physics: thermal, fluid, electromagnetic, etc. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 48 © 2006 Ops A La Carte
  • 28. Finite Element Software • MSC.Marc • MSC.Nastran, MSC.Adams • MD-Nastran • LS-Dyna • Ansys • Abaqus • Comsol • Cosmos • Pro-Mechanica • Algor April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 49 © 2006 Ops A La Carte The Real World is Nonlinear • Simplified approximations compromise accuracy & F Linear safety Load Behavior – Cannot reliably predict geometric, F Nonlinear Behavior material, & boundary nonlinearities u u Displacement • Nonlinear analysis is Linear essential for investigating: Stress Behavior C PLASTIC Nonlinear – Large-strain effects, plasticity, friction, TI AS Behavior dynamic loads, large deflections & motion F F EL Yield Pt. – Metallic & non-metallic materials Strain exhibiting nonlinear behavior (e.g., shape memory alloys, plastics, elastomers, composites, etc.) – Effects of physical coupling with thermal, electrical, fluid, magnetic, acoustic, etc. April 9,9, 2008 April 2008 MSC.Marc MDfR-Mechanical Design for Reliability 50 © 2006 Ops A La Carte
  • 29. Simulate Reality With Nonlinear FEA Capture fit, form, function, & manufacturability through simulation Examples: Snap-Fit Automobile Machining of Buckle Exhaust Aircraft Pipe Fuselage Hanger Component - Make better design decisions earlier in your product’s life cycle - Explore & optimize product performance more quickly through less-expensive virtual testing April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 51 MSC.Marc © 2006 Ops A La Carte Introducing MSC.Marc® Nonlinear analysis for a wide range of applications: MSC.Marc provides virtual insight into the behavior of components experiencing geometric, material, and/or boundary nonlinearities April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 52 © 2006 Ops A La Carte
  • 30. What is MSC.Marc? Virtual Product Development (VPD) solution that helps engineers simulate & optimize component performance & manufacturing processes – Long, proven history of innovation—established in 1971 as the world’s first commercially available nonlinear FEA software – Provides advanced, general-purpose, implicit nonlinear FEA technology with high-performance, linearly- scalable parallel processing capability – Solves static & dynamic 2-D & 3-D structural & non-structural coupled multiphysics problems with highest accuracy & robustness – Simulates all kinds of structural deformation & material nonlinearities: • Multibody contact & frictional effects • Buckling, cracking, bending, cutting, shaping, forming, welding, etc. • Highly-nonlinear materials, such as elastomers, composites, plastics, concrete, etc. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 53 © 2006 Ops A La Carte Elastic/Plastic Tensile Response True Stress E’ Ultimate D Eng. Stress stress C E Yield B stress A Proportional limit 0 Linear Plastic Strain Significant hardening necking Ref: Anderson, Biomaterials notes Typical stress/strain curve for steels. Strains become localized when necking occurs. Standard elongation highly dependent on gage length. Area reduction gives local strain. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 54 © 2006 Ops A La Carte
  • 31. Tensile Test for Steel Pipeline Material April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 55 © 2006 Ops A La Carte Global Remeshing/Rezoning Capability MSC.Marc has a unique automated capability overcomes mesh distortions arising from extreme material deformation – Remeshing: New, better quality global mesh is automatically generated – Rezoning: Solution is transferred from the old mesh to the new mesh – Contact boundary conditions are also automatically reapplied – Triangular or quadrilateral elements for 2-D, tetrahedral or hexahedral for 3-D Compression 3-D Example Showing Rubber Seal with With Global Animation of Close-Up View of Mesh Distorted Mesh Remeshing/Rezoning Rubber Seal Refinements Avoid simulation failure & loss in accuracy for materials April 9, 2008 April 9, 2008 experiencing extreme deformation MDfR-Mechanical Design for Reliability 56 © 2006 Ops A La Carte
  • 32. 3-D Cyclic Symmetry with Remeshing • Examples: Screw Twisting MSC.Marc April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 57 © 2006 Ops A La Carte Advanced Friction Modeling Capability MSC.Marc has simple and robust friction capability. Simply turn it on & supply a Insertion & coefficient of friction—or, manually Extraction provide necessary parameters Force on an Elastomeric Seal Belt-Driven Pulley Simulate realistic contact conditions by including effects of friction April 9, 2008 April 9, 2008 MDfR-Mechanical Design for Reliability 58 © 2006 Ops A La Carte
  • 33. Enabling You to Simulate Reality Use MSC.Marc’s innovative contact modeling capabilities to satisfy your virtual testing requirements Stresses in Lug, Pin, & Clevis Gear Lever with Rubber Boot Staking of Intersecting Friction & Heat Hot Wires into a Plastic Block Generation in Disc Brake Pin Insertion Validate product performance under a wide range of operating conditions & constraints April 9, 2008 April 9, 2008 MDfR-Mechanical Design for Reliability 59 © 2006 Ops A La Carte MSC.Marc Tube Bending with Contact & Remeshing April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 60 © 2006 Ops A La Carte
  • 34. Coupled Thermal-Mechanical April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 61 © 2006 Ops A La Carte MSC.Marc - Machining • Pocket Cutting – Definition of the Two Cutting Processes Y Cutter and its axis definition 1 Surface for the first cut stage Z Y X Surface for the second cut stage Cutter and its axis definition 2 April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 62 © 2006 Ops A La Carte
  • 35. MSC.Marc Machining - Verification • Experimental verification of machining behavior for a 1.5 inch thick stock aluminum beam that is bent to a prescribed radius. • 2.5 inch slot is cut through 75% of its thickness, spring back is plotted along the ten gage points. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 63 © 2006 Ops A La Carte Predicting Distortion from Machining • Machining operations cause distortions • MSC.Marc can be used to predict the distortion April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 64 © 2006 Ops A La Carte
  • 36. Coupled Electrostatic-Mechanical • Capacitor plates close by increasing charge due to Coulomb’s Law • Voltage peaks at 1/3 of original gap opening • Popular for MEMS applications: – Optical-network components – Print Heads – Projectors – BioMEMS April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 65 © 2006 Ops A La Carte Electrostatic-Mechanical MEMs April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 66 © 2006 Ops A La Carte
  • 37. Joule Heating-Mechanical MEMs •Electrical current flow causes heating due to material resistivity •Mechanical deformation due to thermal expansion of hot leg April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 67 © 2006 Ops A La Carte Magnetostatic Improvements • New tetrahedral elements for magnetostatics (181 and 182) • New Line (wire) B element for magnetostatics (183) A A B April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 68 © 2006 Ops A La Carte
  • 38. Coupled Electromagnetic-Thermal • Induction heating used for various industrial manufacturing processes, such as heat treatment for metals, soldering guns, and preheating for forging April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 69 © 2006 Ops A La Carte Example: RF Attenuator Simulation RF/microwave energy is attenuated through resistive losses in a Nichrome film attached to the microstripline waveguide. The energy is lost in the form of heat which is conducted both through the devices ceramic substrate and top insulating surface film. Typical Packaged Device: Image from KDI data sheet. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 70 © 2006 Ops A La Carte
  • 39. Multi-field Solver- RF Attenuator • Analysis results: E-field H-field Resultant temperature April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 71 © 2006 Ops A La Carte Random Vibration & Shock April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 72 © 2006 Ops A La Carte
  • 40. Random Vibration Capabilities • MSC.Patran and MSC.Nastran contain very easy to use random vibration analysis • The overview will have some specific references to the capabilities in these codes April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 73 © 2006 Ops A La Carte What is random analysis? • Statistically based frequency-dependent dynamic loading Time Frequency – Don’t know exactly what the load is at any instant in time – Do know the loading envelope April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 74 © 2006 Ops A La Carte
  • 41. What calls for random analyses? • Random/white noise events – Acoustic pressure loading – Rocket/jet engine exhaust – Rocket/jet engine thrust vibration – Road-loads/transportation – Earthquakes • Base vibration or Acoustic excitation Loading defined in Frequency Domain April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 75 © 2006 Ops A La Carte Frequency Response Analysis • Frequency Domain – Steady-state harmonic response – Load/Response varies w/ Frequency – Magnitude and phase are important Tip Base 1. Frequency Frequency April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 76 © 2006 Ops A La Carte
  • 42. How random analyses typically happen • Typically based on a PSDF spectrum ( ~ )2/Hz Frequency – Almost looks like Frequency Response – Loading is in terms of (something)2/Hz April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 77 © 2006 Ops A La Carte So what do you do??? • Frequency response analysis – Apply 1 unit of what is specified in (~)2/Hz across all frequencies • Define input spectrum – table of (~)2/Hz vs. Frequency • Do the math… – This is where it all comes together • Data recovery – Know what you are looking for/at April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 78 © 2006 Ops A La Carte
  • 43. Frequency Response Post-Processing Frequency Response 1. 10. Frequency Frequency Input spectrum Output scales like this… spectrum scales similarly… 5.8 580. Frequency Frequency Random Response April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 79 © 2006 Ops A La Carte Frequency Response Post-Processing Frequency Response 1. Frequency Frequency Frequency Frequency Random Response April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 80 © 2006 Ops A La Carte
  • 44. Random Results: What to look for • RMS values – Area under curve • Number of positive crossings Frequency – Apparent frequency for fatigue • Cumulative RMS – Which frequencies contribute the most April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 81 © 2006 Ops A La Carte So what is MSC.Random? • User-developed MSC.Patran application – Customer-requested features and functions • Officially available since MSC.Patran v9 – Available under Utilities/Analysis menu • Improved User Interface for: – Frequency Response Analysis – Interactive Random Analysis – Post-processing April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 82 © 2006 Ops A La Carte
  • 45. So what is MSC.Random? • Improved frequency response user interface – Sets up frequency response analysis – Unit loading(s) – Large mass, force or acoustic pressure – Support for FREQi, I=1,5 entries • .xdb access of frequency response results – Fast & efficient April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 83 © 2006 Ops A La Carte So what is MSC.Random? • Interactive Random Analysis – Calculations done “on the fly” (w/o having to re-run MSC.Nastran) – RMS fringe plots – Specialized XY-Plotting • Interactive v. batch environment – Change your random analysis inputs w/out re-running MSC.Nastran April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 84 © 2006 Ops A La Carte
  • 46. MSC.Random Benefits • Simplified analysis setup – GUI interface – Engineer-driven specialized forms – No XYPLOT entries required – Full on-line help • Improved productivity – .xdb access of results – Re-analyze without restarting • Quality results – Specialized interactive plots – Log-log integration – Cumulative RMS April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 85 © 2006 Ops A La Carte New Patran Old GUI New GUI Random • Similar over all operation. • New interface has more options and controls. • New interface has easier and more flexible ways to input data. • More result types supported • New GUI allows more room to add additional features. • Incorporates many enhancements requests. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 86 © 2006 Ops A La Carte
  • 47. Patran Random Enhancements XY plot Num. Pos. Node Ids Crossings (NEW) User Custom Title Reflector Clearance RMS Value (NEW) RMS Scale (NEW) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 87 © 2006 Ops A La Carte Tolerance & Worst Case Analysis April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 88 © 2006 Ops A La Carte
  • 48. Worst-Case vs. Statistical Tolerancing • When solving tolerancing problems, one must choose between worst-case tolerancing and statistical tolerancing. Both of these methods have their pros and cons. • If worst-case tolerancing is used, all tolerances must be specified as worst-case tolerances. If statistical tolerancing is used, all tolerances must be specified as statistical tolerances. • In reality, the behavior of certain inputs is best described using worst-case tolerances while the behavior of other inputs is best described by statistical tolerances. Still other inputs are not adequately represented by either type of tolerance. Ref: “Process Tolerancing: A Solution to the Dilemma of Worst-Case vs. Statistical Tolerancing”, Wayne A. Taylor April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 89 © 2006 Ops A La Carte Tolerance & Worst Case Analysis Traditional Design • Tolerances specify design requirements. • Tolerances control design parameters by defining how much variation is acceptable. • Tolerances give insight into the designer’s intent. Ref: Scott Burr, 2005, “Design for Six Sigma” Hubenthal-Burr Associates April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 90 © 2006 Ops A La Carte
  • 49. Variation Sources of Variation • Processes – Variation occurs when a part is processed. – Different processes often have different inherent capabilities. • Raw Materials – Inhomogeniety – Property variation • Environment – Temperature and Pressure Variation – Noise – Vibration • Product Degradation – Due to Heat – Component variation through time – Stresses and failure mechanisms Traditional Design is unconcerned with Process Variation as long as you meet SPEC!! April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 91 © 2006 Ops A La Carte Variation in Mechanical Assemblies Three sources of variation occur in mechanical assemblies • Dimensional variations (lengths & angles) • Geometric form and feature variations (flatness, roundness, angularity, etc.) • Kinematic variations (small adjustments between mating parts) Ref: “A Comprehensive System for Computer-Aided Tolerance Analysis of 2-D and 3-D Mechanical Assemblies” , Kenneth Chase, et.al. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 92 © 2006 Ops A La Carte
  • 50. There is Good News and Bad News • Tolerancing has NO influence on Variation • You CANNOT count on Inspection to assure your design parameters and tolerances. • Proper Design Choice is a highly effective countermeasure against Variation • FEA Simulation is a highly effective method for evaluating Design Choices, Variation and linking process parameters to design performance. Ref: Scott Burr, 2005, “Design for Six Sigma” Hubenthal-Burr Associates April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 93 © 2006 Ops A La Carte Lunch!! April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 94 © 2006 Ops A La Carte
  • 51. Thermal Analysis & Electronic Cooling Wireless Packages April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 95 © 2006 Ops A La Carte Thermal Airflow Comparison Laminar vs. Turbulent flow regimes • Laminar airflow – Possible overheating. – Insufficient temperature gradient • Turbulent airflow – Heat transfer greatly enhanced. – Boundary layer dissipated. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 96 © 2006 Ops A La Carte
  • 52. Thermal Airflow Comparison Laminar Flow Turbulent Flow Ta Ta Tb Tb Ta Ta Tb>Ta Ta=Tb Heat Transfer 104 BTU/Hr 106 BTU/Hr April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 97 © 2006 Ops A La Carte Why does Turbulence Enhance Heat Transfer? • Turbulent boundary layer is thin compared to laminar • Turbulence has unsteady (time varying) velocity & pressure components • Turbulent flows include eddies & vortices; “eddy transport” is a major heat transfer factor • Turbulent flow has shorter entry length for flow to become fully developed • These features enhance the heat and mass transfer in the turbulent regime Ref: “Component or PCB Oriented Thermal Design” http://www.coolingzone.com/library.php?read=530 April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 98 © 2006 Ops A La Carte
  • 53. How do you Achieve Turbulent Flow? • Typically easy in an electronics enclosure • “Roughness” associated with components (leads, packages, capacitors, etc.) trips the boundary layer • Add ribs, helical ridging or similar • Related to the characteristic Re (Reynolds number) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 99 © 2006 Ops A La Carte PCB Heat Transfer • Thermal vias in the package and board serve as significant heat flow path from the top plane of the package to the top plane of the board • Entry flow effects are important Ref: “So Many Chips, So Little Time, Device Temperature Prediction in Multi-Chip Packages”, http://electronics-cooling.com/articles/2006/2006_aug_cc.php April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 100 © 2006 Ops A La Carte
  • 54. Cell Phone PCB Thermal Analysis April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 101 © 2006 Ops A La Carte Thermal Design • Both heat and cold adversely affect circuits • Low temperatures cause different problems • Thermal cycling can lead to fatigue • Thermal shock severe Ref: “Hot, Cold, and Broken: Thermal Design Techniques”, 2007, EDN, http://www.edn.com/article/CA6426879.html April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 102 © 2006 Ops A La Carte
  • 55. System Thermal Design • Airflow (forced convection) can reduce thermal resistance up to 5X • Fans add noise; need to be sized based on volume flow in the system • Ambient temperature effect • Turbulence and natural convection Ref: “Turbulent Heat Transfer in Buoyancy-Driven Natural Convection in Vertical Enclosures” Ref: www.ceere.org/beep/docs/FY2002/Turbulent_Flow_in_Enclosure.pdf April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 103 © 2006 Ops A La Carte Thermal / Structural / Drop Test Analysis of a Circuit Board • Determine Temperature distribution and stresses due to CTE Mismatch. • Determine dynamic response due to frequency response and random excitation • Determine acceleration response and stresses due to Drop Testing IC Leads PWB Connector April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 104 © 2006 Ops A La Carte
  • 56. Circuit Board Demonstration Thermal Analysis Structural Warping Modal Determine Temperature Determine Stresses Analysis Distribution in board due to CTE mismatch Evaluate stress and deformation Determine Response Levels due to input PSD Frequency Determine stresses Response due to Drop Testing Analysis Drop Random Test Vibration April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 105 © 2006 Ops A La Carte Summary - MD Nastran Workflow • MD Nastran eliminates two software applications from the process. Step 1 Ref: MSC Software Thermal Analysis Step 2 MD Nastran Structural Warping Step 3 Modal Analysis Step 4 Frequency Response Analysis Step 5 Random Vibration Step 6 Drop Test April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 106 © 2006 Ops A La Carte
  • 57. References – Thermal & Cooling • Electronic Cooling Solutions http://www.ecooling.com/references.html • Electronics Cooling http://www.electronics-cooling.com • Coolingzone http://www.coolingzone.com/library.php April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 107 © 2006 Ops A La Carte Probabilistic Design System (PDS) • Essentially a DOE using finite element analysis • Key variables are specified with some sort of distribution (e.g., Gaussian distribution, etc.) • Series of simulations are analyzed varying the parameters within the specified distributions • Terms such as “stochastic analysis” are also used April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 108 © 2006 Ops A La Carte
  • 58. Probabilistic Design System (PDS) • Tools like Engineous iSight can be used to drive other applications and finite element codes to give a PDS • ANSYS includes PDS capabilities in the Workbench • MSC Adams Insight provides excellent DOE and Response Surface capabilities • MSC.Software MD-Nastran is adding stochastic and DOE capability for multi-disciplinary applications April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 109 © 2006 Ops A La Carte Sources of Uncertainty in Engineering Physical uncertainty Material properties F F Boundary and initial conditions Assembly and manufacturing imperfections Environmental loads Geometry Non-physical uncertainty Modelling errors (material model, element type, etc.) Choice of method (linear, non-linear, etc.) non- Solver Computer Engineer April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 110 © 2006 Ops A La Carte
  • 59. Definition of a Stochastic Problem Model Random response Multiple executions Single computer run are a Simulation is an Analysis April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 111 © 2006 Ops A La Carte Introduction to Stochastics April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 112 © 2006 Ops A La Carte
  • 60. Variation in Design • Traditional Simulation • Reality – “Deterministic” – No two examples of a – All parameters are known product are exactly alike precisely – No two tests will produce – One solution – uniquely the same results determined – It’s not always obvious why – Evaluation of whether a some samples pass design is “good” is usually acceptance tests while based on nominal others fail conditions – Simulations from two different engineers will give slightly different results April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 113 © 2006 Ops A La Carte How Do Designers Deal with this Variation? • By trying to eliminate it. • Make reality behave like a deterministic simulation rather than making the simulation a better reflection of reality – Tight tolerances – Elaborate process specifications – Statistical quality control – Conservatism factors – Scrap parts that show “too much” variation – Expensive and time consuming tests to determine “real world” performance • The result? – Reduces but doesn’t eliminate variation – Can be very expensive April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 114 © 2006 Ops A La Carte
  • 61. The Alternative • Make simulations vary like reality • Perform multiple simulations, each with slightly different parameter values that reflect the variations that occur in actual manufacturing and use • Multiple simulations provide a distribution of outcomes, rather than a single outcome • Evaluate which variations have a large impact on performance and reliability and should be controlled and which have a small impact and can be allowed to vary over a wider range • This is the idea behind stochastic simulation April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 115 © 2006 Ops A La Carte Stochastic • sto·chas·tic 1 : involving a random variable <a stochastic process> 2 : involving chance or probability <a stochastic model of radiation-induced mutation> • Source: Merriam-Webster's Medical Dictionary, © 2002 Merriam-Webster, Inc. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 116 © 2006 Ops A La Carte
  • 62. Why Stochastics Makes Sense for Electronics Applications (and many other applications!) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 117 © 2006 Ops A La Carte Electronics Applications • Examples: – Computer equipment and components – Cell phones and communications equipment – Biomedical devices – Appliances – Automotive and aerospace components – Manufacturing equipment • Typically have large production volume • Emphasis on miniaturization • High cost of failure (liability, warranty) • Operate in a variety of frequently harsh and uncontrolled environments April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 118 © 2006 Ops A La Carte
  • 63. Design Objective for Electronics Devices • The design objective for electronics devices is typically a probabilistic one • For example, mean time between failure (MTBF) or probability of failure (POF) • Translating deterministic results to probabilistic terms is difficult at best • Probabilistic design objectives require probabilistic design techniques April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 119 © 2006 Ops A La Carte Business Value • Stochastics provides: – Time to Market • Efficient design space surveys • Shorten development time by enabling engineers to analyze results, not run tests – Reduce warranty/risk • Greater understanding of key parameters in the design • Design for manufacture – Improve perceived quality/robustness • Fewer returns • Fewer complaints • Less warranty costs April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 120 © 2006 Ops A La Carte
  • 64. Stochastic Crash Analysis April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 121 © 2006 Ops A La Carte MSC.ADAMS/Insight • Insight is a framework for performing stochastic simulation • Insight was designed to be flexible enough in its implementation that it can work with virtually any simulation or analysis tool • When coupled to MSC.SimManager, a web- based portal for simulation management, it can provide efficient stochastic simulations to an entire enterprise April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 122 © 2006 Ops A La Carte
  • 65. How Does It Work? • The user defines parameters (called “factors”) which vary and the way they vary (a list of potential discrete values, a probability distribution, etc.) • The user defines responses, i.e. – dependent variables that vary as a result of the variation in the input factors • The user defines how the input factors and output responses are related to the simulation model • Insight then performs multiple simulations. – Each simulation has a set of input factors whose values are automatically selected based on a variety of methods – The responses of each simulation are collected – Essentially “virtual tests” of the design • The result is a statistical sampling of inputs and the resulting outputs April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 123 © 2006 Ops A La Carte What Does Insight Provide? • The selection of factors for each stochastic simulation can be based on Latin Hypercube or Monte Carlo methods – Other non-stochastic virtual experimental design methods, such as Full Factorial, Central Composite, Box-Behnken, Placket Burman, Sweep Study, D-Optimal, and Design of Experiments (DOE) are also available • Statistical analysis of the variation in responses to the variation in the inputs • Response surfaces that represent the relationship between factors and responses of the system as curve- fit polynomials • Statistical data that can be exported to tools such as Excel, Visual Basic, and MATLAB • Export results to a web server April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 124 © 2006 Ops A La Carte
  • 66. Typical Output Using Insight’s Web Interface Includes response surface data. “+” and “-” buttons enable the user to interactively change values of factors see their effect on the responses. April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 125 © 2006 Ops A La Carte What are the Benefits? • Response surfaces allow a user to select values for factors needed to achieve a desired response • Statistical analysis identifies which factors have the greatest affect on responses and can provide the greatest benefit – Relax tolerances on design parameters that have little affect – Tighten tolerances on those that have a greater affect – Assess the probability of a specific outcome (e.g. – “stress will be greater than allowable”) • Physical tests can be more rapidly correlated by comparison to the results of “virtual” tests • Designs can be automatically improved using “stochastic design improvement” (SDI) techniques April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 126 © 2006 Ops A La Carte
  • 67. What are the Applications? • Drop Testing – Vary impact energy and orientations, material properties at high strain rates • Ball Grid Array Durability – Vary thermal input, ball size and shape, load on individual balls in the array, solder material properties at temperature • Mechanical Systems Reliability – Vary manufacturing tolerances, actuator forces, freeplay • Controller Algorithm Development – Vary gains, phase lags, sensor accuracy April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 127 © 2006 Ops A La Carte An Example: BGA-Ball Grid Array April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 128 © 2006 Ops A La Carte
  • 68. The Problem • Fatigue analysis of a single solder ball • Loads come from a less detailed model of a ball grid array. This is the most severely loaded ball in the array. • Shape and size of solder balls varies. For very small solder balls, even small variations can have significant impact on strain and hence fatigue life • Fatigue methodology for solder under plastic deformation is semi-empirical and subject to significant scatter April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 129 © 2006 Ops A La Carte The Approach • Create a parametric model of the solder ball using MSC.Patran software – Height, diameter, the size of small fillets at the top and bottom of the solder ball, the load on the solder ball, the yield strength of the solder • Use Insight to define normal distributions for each of these parameters • Conduct 128 finite element simulations with MSC.Marc using Monte Carlo methodology to sample the design space • Determine the statistical distribution of plastic strain, and hence, fatigue life April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 130 © 2006 Ops A La Carte
  • 69. The Process Define Define Parametric Define Responses Monte Carlo Distributions for Model of Interest Sampling Parameters MSC.Patran FEA Response Surfaces MSC.Marc FEA Factor/ Response Correlations Database of Factor and Response Values Outcome Probabilities April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 131 © 2006 Ops A La Carte Solder Ball – Parameterized Geometry Fillet Height Height Diameter April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 132 © 2006 Ops A La Carte
  • 70. Solder Ball Finite Element Model April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 133 © 2006 Ops A La Carte Solder Ball – Total Strain April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 134 © 2006 Ops A La Carte
  • 71. Solder Ball – Plastic Strain April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 135 © 2006 Ops A La Carte Work Space – Inputs and Responses (only 36 out of 128 displayed) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 136 © 2006 Ops A La Carte
  • 72. Work Space Review April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 137 © 2006 Ops A La Carte Influence of Input Variables on Max. Plastic Strain April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 138 © 2006 Ops A La Carte
  • 73. Work Space Correlations (Pearson Correlation) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 139 © 2006 Ops A La Carte Work Space Correlations Input vs. Responses April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 140 © 2006 Ops A La Carte
  • 74. Work Space Scatter Plot Diameter vs. Max. Displacement April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 141 © 2006 Ops A La Carte Work Space Scatter Plot Mesh Factor vs. Max. Plastic Strain April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 142 © 2006 Ops A La Carte
  • 75. Probabilistic Design System (PDS) with ANSYS FEA April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 143 © 2006 Ops A La Carte Stress Results Max.Principal Stress Hoop Stress April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 144 © 2006 Ops A La Carte
  • 76. Rank Order Correlation Sensitivities & Histogram of Stress – Wrist Pin ID has the greatest influence on the fatigue stress April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 145 © 2006 Ops A La Carte Probabilistic Simulations using ANSYS to DFSS • Introduction • An Example • Motivation • Features • Benefits • Probabilistic Methods • Probabilistic Results/Interpretation • Summary Ref: 2004, ANSYS DFSS Workshop April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 146 © 2006 Ops A La Carte
  • 77. Introduction Purpose of a Probabilistic Design System (PDS) Input Input ANSYS ANSYS Output Output Material Properties Deformation Geometry Stresses / Strains Boundary Conditions Fatigue, Creep,... It’s a reality that input parameters are subjected to scatter => automatically the output parameters are uncertain as well!! April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 147 © 2006 Ops A La Carte Introduction Purpose of Probabilistic Design System (PDS) ANSYS PDS ANSYS PDS Questions answered with probabilistic design: • How large is the scatter of the output parameters? • What is the probability that output parameters do not fulfill design criteria (failure probability)? • How much does the scatter of the input parameters contribute to the scatter of the output (sensitivities)? April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 148 © 2006 Ops A La Carte
  • 78. An Example Example: Lifetime of Components !!! Random input Finite-Element Random output variables Model parameters Material • LCF lifetime • Strength • Creep lifetime • Material • Corrosion lifetime Properties • Fracture mechanical lifetime Loads •… • Thermal • Structural Evaluate reliability of products ! Geometry/ Tolerances Evaluate quality of products ! Boundary Evaluate warranty costs ! Conditions • Gaps To evaluate is the first step • Fixity to improvement ! April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 149 © 2006 Ops A La Carte Motivation Influence of Young’s Modulus and Thermal Expansion Coefficient on thermal stresses: thermal = E · · T Deterministic Approach: Emean and mean => evaluate expected value: expect Probabilistic Approach: P( thermal > 1.05 expect) P( thermal > 1.10 expect) ‘E’ scatters ±5% 16% (~1 out of 6) 2.3% (~1 out of 40) ‘E’ and ‘ ‘ scatter ±5% 22% (~1 out of 5) 8% (~1 out of 12) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 150 © 2006 Ops A La Carte
  • 79. Scatter in Material Properties and Loads Property SD/Mean % Metallic materiales, yield 15 Source: Klein, Schueller et.al. Probabilistic Approach Carbon fiber composites, rupture 17 to Structural Factors of Safety in Aerospace. Metallic shells, buckling strength 14 Proc. CNES Spacecraft Structures and Mechanical Junction by screws, rivet, welding 8 Testing Conf., Paris 1994 Bond insert, axial load 12 Honeycomb, tension 16 Honeycomb, shear, compression 10 Honeycomb, face wrinkling 8 Launch vehicle , thrust 5 Transient loads 50 Thermal loads 7.5 Deployment shock 10 Acoustic loads 40 Vibration loads 20 April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 151 © 2006 Ops A La Carte Motivation ±5-100% Materials, Bound.- Cond., Loads, ... Thermal Analysis CAD FEM Geometry CFD LCF ±??% ± 0.1-10% Structural FEM Materials, Analysis Materials Bound.- Cond., ... Materials, ±30-60% Bound.- ±5-50% Cond., ±5-100% Loads, ... April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 152 © 2006 Ops A La Carte
  • 80. PDS Benefits Deterministic Analysis: Probabilistic Analysis: • Only provides a YES/NO answer • Provides a probability and reliability (design for reliability) • Safety margins are piled up • Takes uncertainties into account in a “blindly” (worst material, maximum realistic fashion => This is closer to load, … worst case) reality => Over-design is avoided 1 worst case assumption =10-2 2 worst case assumptions =10-4 3 worst case assumptions =10-6 4 worst case assumptions =10-8 ... => Leads to costly over-design • “Tolerance stack-up” is included • Only “as planned“, “as is” or the (design for manufacturability) worst design April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 153 © 2006 Ops A La Carte PDS Benefits Deterministic Analysis: Probabilistic Analysis: • Sensitivities do not take range of • Range/width of scatter is “built-in” into scatter or possibilities into account probabilistic sensitivities • Sensitivities do not take interactions between input • Interactions between input variables variables into account (second are inherently taken care of order cross terms) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 154 © 2006 Ops A La Carte
  • 81. PDS Benefits Illustration of the Benefits of Probabilistic Analysis over Deterministic Analysis Probabilistic Analysis Deterministic Analysis April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 155 © 2006 Ops A La Carte ANSYS PDS Features • Works with any ANSYS analysis model • Static, dynamic, linear, non-linear, thermal, structural, electro- magnetic, CFD .. • Allows large number random input and output parameters • 10 statistical distributions for random input parameters • Random input parameters can be correlated • Probabilistic methods: • Monte Carlo - Direct & Latin Hypercube Sampling • Response Surface - Central Composite & Box-Behnken Designs April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 156 © 2006 Ops A La Carte
  • 82. ANSYS PDS Features • Use of distributed, parallel computing techniques for drastically reduced wall clock time • Comprehensive probabilistic results • Convergence plots, histogram, probabilities, scatter plots, sensitivities, ... • State-of-the art statistical procedures to address the accuracy of the output data • Confidence intervals April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 157 © 2006 Ops A La Carte Probabilistic Methods Monte Carlo Simulation: Fully Parallel Perform numerous analysis runs based on sets of random samples, and then evaluate statistics of derived responses. • Direct (Crude) Sampling Monte Carlo (DIR) • Latin Hypercube Sampling Monte Carlo (LHS) • User defined (USR) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 158 © 2006 Ops A La Carte
  • 83. Probabilistic Methods Monte Carlo Simulation Method Scheme: Simulation of input Statistical analysis of parameters at output parameters random locations X1 X2 X3 Repetitions = Simulations ANSYS ANSYS April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 159 © 2006 Ops A La Carte Probabilistic Methods Finite Element Runs for Monte Carlo For Monte Carlo Simulation the number of simulations does not depend on the number of random input variables, but on the probabilistic result you are looking for: • For assessment of the statistics of output parameters (Mean, sigma) Nsim 30 … 100 • For histogram and cumulative distribution function Nsim 50 … 200 • For assessment of low probabilities P (tails of the distribution) Nsim 30/P … 100/P April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 160 © 2006 Ops A La Carte
  • 84. Probabilistic Methods – Response Surface Methods: Fully Parallel Select specific observation points for each random variable, run analyses, establish response surface for each response parameter, perform Monte Carlo Analysis on Response Surface. • Central Composite Design (CCD) • Box-Behnken Matrix (BBM) • User defined (USR) April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 161 © 2006 Ops A La Carte Probabilistic Methods Response Surface Methods Scheme: Simulation of input parameters Statistical analysis of at specific locations output parameters X1 X2 X3 Evaluate input Evaluate input Monte Carlo Simulations Monte Carlo Simulations parameter values parameter values on Response Surface on Response Surface DOE Repetitions = Simulations Response Surface Fit Response Surface Fit ANSYS ANSYS April 9,9, 2008 April 2008 MDfR-Mechanical Design for Reliability 162 © 2006 Ops A La Carte