Osu 1133282089

download Osu 1133282089

of 182

Transcript of Osu 1133282089

  • 8/17/2019 Osu 1133282089

    1/182

    SELECTION AND OPTIMIZATION OF SNAP-FIT FEATURESVIA WEB-BASED SOFTWARE

    DISSERTATION

    Presented in Partial Fulfillment of the Requirement for

    The Degree Doctor of Philosophy in the Graduate

    School of The Ohio State University

    By

    Tieming Ruan, M.S.

    * * * *

    The Ohio State University

    2005

    Dissertation Committee:

    Approved by

    Professor Anthony F. Luscher, Adviser

    Professor Gary L. Kinzel

    Professor Krishnaswamy Srinivasan Adviser

    Professor Donald R. Houser Mechanical Engineering

    Graduate Program

  • 8/17/2019 Osu 1133282089

    2/182

      ii

    ABSTRACT

    When used in plastics, snap-fits can be a simple, quick and cost-effective method

    of assembling two parts. When designed properly, parts with snap-fits can be assembled

    and disassembled numerous times without any adverse effect on the assembly. Snap-fits

    also the aid in making products environmentally friendly because of their ease of

    disassembly, making components of different materials easy to recycle. Traditionally,

    snap-fit design methodology has been disorganized and anecdotal in nature, relying

    greatly on the skill and the experience of the individual designer. The most popular

    source of designing snap-fit is the design guides from resin suppliers. However the

    information disseminated by these guides is usually old, obsolete and inaccurate because

    they based on assumptions of small deformation and linear material property. To

    overcome these two disadvantages, the author developed a two-dimensional, plane stress,

    contact finite element model (FEM) considered the nonlinear material property of

     polymer. Using design of experiments (DOE) and response surface methodology (RSM),

    this research obtained a second order response surface equation to predict the retention

     performance of cantilever hook with high retention angles.

    With the development of Internet, there is also a demanding for a web-based

    snap-fit design tool that is independent to all operating systems, easily accessible and can

     be universally upgraded by simply updating the design tool at the server location. This

  • 8/17/2019 Osu 1133282089

    3/182

     iii

    thesis developed a web-based design tool for three different snap-fit features: cantilever

    hook, post & dome, and bayonet & finger. The response surface equations obtained by

    using the combination between FEM, DOE and RSM were applied. Constraint

    management (CM) was used to make several functions available such as sensitivity

    analysis, correction advisor.

    Optimization modules such as single objective optimization and multiple

    objectives optimization were applied to make the design tool more flexible and powerful.

    Single objective optimization was implemented in two steps. The first step is to use

    Golden Section Method to identify the search direction and the second step is to use

    Broydon-Fletcher-Goldfarb-Shanno (BFGS) method to find the minimum values on this

    search direction. To do multiple objectives optimization, the Weighted Sum strategy was

    choose. Each objective was assigned a weighted value based on their importance, then

    combine them into a single objective optimization problem.

  • 8/17/2019 Osu 1133282089

    4/182

      iv

     

    Dedicated to my parents, my wife and kids

  • 8/17/2019 Osu 1133282089

    5/182

      v

     

    ACKNOWLEDGMENTS

    I wish to thank my adviser, Anthony Luscher, for intellectual support,

    encouragement, and enthusiasm which made this dissertation possible, and for his

     patience in correcting both my stylistic and scientific errors.

    I thank Prof. Srinivasan and Prof. Houser for their patience and to be my

    committee members. Especially I thank Prof. Gary Kinzel to be my committee member

    and for providing the FORTRAN source code used in this web-based design tool.

    I am grateful to Gaurav Suri for discussing with me various aspects of this

    dissertation. I also want to thank my officemates who helped me to handle various

     problems, especially Jonathan Pillai and Leo Rusli.

    I want to thank my parents, my wife and kids. Without your support, I can’t finish

    this dissertation with a full-time job.

    This research was supported by a grant from CAPCE.

  • 8/17/2019 Osu 1133282089

    6/182

     vi

     

    VITA

    January 11, 1976 …………………………………….. Born – Zhejiang, China

    1996 ………..…………………………………………B.S. Mechanical Engineering,

    China Univ. of Mining & Tech.

    1999 ……………………………………… ………… M.S. Mechanical Engineering,

    China Univ. of Mining & Tech.

    1999 – 2002 …………………………………………..Graduate FellowThe Ohio State University

    2002 – present ………………………………………...Graduate Research AssistantThe Ohio State University

    2003 – present ……………………………………….. Sr. Associate R&D EngineerBayer HealthCare

    PUBLICATIONS

    Research Publication

    1. T. Ruan, Z. Wang and H. Chen, “A monitoring device for the fire of belt”, Journalof Coal Science & Engineering (China), 1998.

    2. Z. Luan, Z. Wang and T. Ruan, “Balance compound gear pump (motor)”,Transaction of Huainan Institute of Technology (Chinese), 1997.

  • 8/17/2019 Osu 1133282089

    7/182

     vii

     

    FIELDS OF STUDY

    Major field: Mechanical Engineering

  • 8/17/2019 Osu 1133282089

    8/182

     viii

     

    TABLE OF CONTENTS

    Abstract .............................................................................................................. ii

    Dedication ......................................................................................................... iv

    Acknowledgments.............................................................................................. v

    Vita ..................................................................................................................... vi

    List of Tables .................................................................................................... xi

    List of Figures ................................................................................................. xiii 

    Chapter 1 Introduction ....................................................................................... 1 1.1 Introduction to Snap-Fit Features ............................................................................. 21.2 Motivation for Research ........................................................................................... 4

    1.2.1 Snap-fit Design Approaches .............................................................................. 81.2.2 Web-based Snap-fit Design Tool..................................................................... 111.2.3 Constraint Management ................................................................................... 12

    1.2.4 Thesis Objectives ............................................................................................. 13

    Chapter 2 Literature Review ............................................................................ 15 2.1 Numerical Formulation for Snap-fit Features......................................................... 15

    2.2 Materials Issues in Snap-fit Design ........................................................................ 20

    2.3 Failure Models of Cantilever Hook ........................................................................ 222.4 Web-based Design Tool for Snap-fit....................................................................... 25

    2.5 Constraint Management .......................................................................................... 34

    Chapter 3 Finite Element Analysis of Cantilever Hook With a High RetentionAngle ................................................................................................................. 37 

    3.1 Combination of FEA with DOE.............................................................................. 37

    3.2 Nonlinear Structural Analysis of Snap-Fits ............................................................ 39

    3.2.1 Moving Contact ............................................................................................... 393.2.2 Geometric Nonlinearities................................................................................. 40

  • 8/17/2019 Osu 1133282089

    9/182

      ix

    3.2.3 Material Nonlinearities .................................................................................... 40

    3.3 Numerical Model of Cantilever Hook Feature ....................................................... 413.3.1 Modeling Techniques for Contact Problem..................................................... 42

    3.3.2 Improvement of Solution ................................................................................. 44

    3.3.3 Comparison between FEM Result and Experiment Result.............................. 46

    Chapter 4 The Empirical Model for Performance of Cantilever Hook With aHigh Retention Angle ....................................................................................... 49 

    4.1 Screening Experiment of Cantilever Hooks with High Retention Angles.............. 514.2 Central Composite Design ...................................................................................... 59

    4.3 Model Improvement................................................................................................ 64

    4.3.1 Transformation of Empirical Model ................................................................ 654.3.2 Remove the Influential Outlying Cases ........................................................... 66

    4.4 Verification Tests..................................................................................................... 74

    Chapter 5 Design Optimization and Constraint Management ...................... 78 

    5.1 Design Optimization ............................................................................................... 785.1.1 Single Objective Optimization......................................................................... 815.1.2 Multiple Objectives Optimization.................................................................... 82

    5.2 Constraint Management .......................................................................................... 83

    5.2.1 Constraint Representation................................................................................ 85

    5.2.2 Constraint Management Algorithms................................................................ 895.3 Mathematic Formulations for Snap-fits Features ................................................. 102

    Chapter 6 The Web-Based Application for Snap-Fits.................................. 108 6.1 Approaches for Web-based Application................................................................ 108

    6.1.1 Java ................................................................................................................ 108

    6.1.2 JavaScript....................................................................................................... 1106.1.3 VBScript ........................................................................................................ 110

    6.1.4 VB DHTML Application............................................................................... 1116.1.5 VB IIS Application ........................................................................................ 111

    6.2 Principles for VB IIS Application......................................................................... 113

    6.2.1 ASP Object Model ......................................................................................... 1146.2.2 Introduction to WebClasses ........................................................................... 116

    6.3 Snap-fit Web-based Application ........................................................................... 117

    6.3.1 New Specification Problem ........................................................................... 1216.3.2 Respecification Problem ................................................................................ 121

    6.3.3 Unspecification Problem................................................................................ 121

    6.3.4 Reverse Specification Problem...................................................................... 124

    6.3.5 Sensitivity Analysis ....................................................................................... 1256.3.6 Correction Advisor......................................................................................... 126

    6.3.7 Single Objective Optimization....................................................................... 127

    6.3.8 Multiple Objectives Optimization.................................................................. 127

  • 8/17/2019 Osu 1133282089

    10/182

      x

    Chapter 7 Case Studies ................................................................................. 131 7.1 Methods to Evaluate the Performance of Cantilever Hook .................................. 1317.2 Integrated Polymeric Lens Housing...................................................................... 132

    7.3 Cantilever Hook Application of Cabinet............................................................... 136

    7.4 Cantilever Hook Samples ..................................................................................... 139

    7.5 Case Study Conclusions........................................................................................ 145

    Chapter 8 Conclusions and Recommendations for Future Research ....... 147 8.1 Scientific and Engineering Contribution .............................................................. 1478.2 Recommendations for Future Research ................................................................ 150

    Appendix A “PROCEDURE” File for MSC.MARC and MSC.PATRAN ........ 151 

    Appendix B Visual Basic Subroutines for Cantilever Hook........................ 153 

    Bibliography.................................................................................................... 162 

  • 8/17/2019 Osu 1133282089

    11/182

     xi

     

    LIST OF TABLES

    Table 4.1: The upper limit and low limit of each parameter ............................................ 52

    Table 4.2: The screening array.......................................................................................... 55

    Table 4.3: The result of best subset regression................................................................. 58

    Table 4.4: The most significant factors for cantilever hook............................................. 61

    Table 4.5: The central composite design for cantilever hook........................................... 62

    Table 4.6: Regression analysis for retention force of cantilever hook ............................. 64

    Table 4.7: Regression analysis of logarithm transformation of cantilever hook .............. 67

    Table 4.8: Regression analysis of reciprocal transformation of cantilever hook.............. 68

    Table 4.9: The influential outlying cases.......................................................................... 71

    Table 4.10: Regression analysis for retention force after deleted influential outlying cases

    ........................................................................................................................................... 73

    Table 4.11: Verification test between FEA result and equation results............................ 76

    Table 4.12: The analysis of variance for Lack-of-Fit test................................................. 77

    Table 5.1: Parameters for strength of materials design equations .................................... 80

    Table 5.2: The occurrence matrix of cantilever hook ....................................................... 86

    Table 5.3: The occurrence matrix after specified x1, x2, x3, x4, and x6............................... 93

    Table 5.4: Block, input/output, level and order of cantilever hook example.................... 95

    Table 5.5: The occurrence matrix of respecification problem.......................................... 96

    Table 5.6: The occurrence matrix of unspecification problem......................................... 98

  • 8/17/2019 Osu 1133282089

    12/182

     xii

    Table 5.7: The response surface factors for the post & dome feature ............................ 103

    Table 5.8: Design variables for the post & dome feature ............................................... 104

    Table 5.9: The scaled variables for bayonet & finger feature......................................... 106

    Table 5.10: The design variables for bayonet & finger feature ...................................... 106

    Table 5.11: The response surface factors for bayonet & finger feature.......................... 107

    Table 6.1: The advantages and disadvantages of web-based languages......................... 113

    Table 7.1: The comparison of cantilever hook samples ................................................. 145

  • 8/17/2019 Osu 1133282089

    13/182

     xiii

     

    LIST OF FIGURES

    Figure 1.1: Typical experimental force curve for insertion of snap-fit features................. 4

    Figure 1.2: Typical experimental force curve for insertion of snap-fit features................. 5

    Figure 1.3: Common snap-fit feature topologies ................................................................ 6

    Figure 1.4: Equations for dimensioning cantilevers ........................................................... 7

    Figure 1.5: Force multiplier of insertion and retention....................................................... 7

    Figure 1.6: Cantilever hook and mating part ...................................................................... 9

    Figure 1.7: Suri’s idealized model of cantilever hook and mating part, shown in deformedconfiguration. .................................................................................................................... 10

    Figure 2.1: Percent engagement of cantilever hook.......................................................... 23

    Figure 2.2: The different failure modes with respect to the PE ........................................ 24

    Figure 2.3: The two different mechanical failure modes for cantilever hook .................. 24

    Figure 2.4: The IFP snap-fit design tool ........................................................................... 28

    Figure 2.5: AlledSignal Plastics’ snap-fit design guide.................................................... 29

    Figure 2.6: Cantilever snap-fit design tool from Eastman chemical company................. 30

    Figure 2.7: Jeff Raquest’s snap-fit calculator ................................................................... 31

    Figure 2.8: GE Plastics’ snap-fit wizard ........................................................................... 32

    Figure 2.9: Engineers Edge’s snap-fit straight beam calculator ....................................... 32

    Figure 2.10: Engineers Edge’s snap-fit tapered beam calculator ..................................... 33

    Figure 2.11: Brock & Wright’s design tool for snap fits.................................................. 33

  • 8/17/2019 Osu 1133282089

    14/182

     xiv

    Figure 3.1: The actual and bilinear stress-strain curve for plastic material...................... 42

    Figure 3.2: Typical cantilever hook meshed example ...................................................... 43

    Figure 3.3: The nonlinear stress-strain curve of Duraform Polyamide ............................ 46

    Figure 3.4: The comparison results for cantilever 1 ......................................................... 47

    Figure 3.5: The comparison results for cantilever 2 ......................................................... 47

    Figure 3.6: The comparison results for cantilever 3 ......................................................... 48

    Figure 4.1: The typical cantilever hook ............................................................................ 53

    Figure 4.2: The typical bayonet & finger.......................................................................... 54

    Figure 4.3: The relationship between beam thickness and retention force....................... 54

    Figure 4.4: The combination of central composite design................................................ 60

    Figure 4.5: Retention force vs. position............................................................................ 61

    Figure 4.6: The residual and normality plots for retention force...................................... 65

    Figure 4.7: Residual plot vs. fitted values for logarithm transformation.......................... 69

    Figure 4.8: Residual plot vs. fitted values for reciprocal transformation ......................... 70

    Figure 4.9: The residual plot after deleted outlying cases ................................................ 73

    Figure 4.10: Main effects for retention force.................................................................... 75

    Figure 5.1: A procedure for identifying disjoint blocks of equations............................... 87

    Figure 5.2: Design decomposition algorithm.................................................................... 91

    Figure 5.3: The new specification problem of cantilever hook ........................................ 94

    Figure 5.4: The respecification problem of cantilever hook............................................. 96

    Figure 5.5: Forward dependency algorithm...................................................................... 97

    Figure 5.6: The unspecification problem of cantilever hook............................................ 99

    Figure 5.7: Backward dependency algorithm................................................................. 101

  • 8/17/2019 Osu 1133282089

    15/182

     xv

    Figure 5.8: The reverse specification problem of cantilever hook ................................. 102

    Figure 5.9: A typical post & dome feature ..................................................................... 104

    Figure 6.1: A typical WebClass life cycle ...................................................................... 118

    Figure 6.2: The flow diagram of snap-fit web-based design tool ................................... 119

    Figure 6.3: The logon page ............................................................................................. 120

    Figure 6.4: The selection page ........................................................................................ 120

    Figure 6.5: The design interface of new specification.................................................... 122

    Figure 6.6: The design interface of respecification ........................................................ 122

    Figure 6.7: The final design interface of new specification............................................ 123

    Figure 6.8: The design interface of unspecification........................................................ 123

    Figure 6.9: The design interface of reverse specification ............................................... 124

    Figure 6.10: The typical sensitivity analysis page .......................................................... 125

    Figure 6.11: The typical correction advisor page ........................................................... 126

    Figure 6.12. Single objective optimization for retention force of cantilever hook......... 128

    Figure 6.13. Single objective optimization result for retention force of cantilever hook128

    Figure 6.14: Multiple objectives optimization for retention force of cantilever hook.... 129

    Figure 6.15: Multiple objectives optimization result for retention force of cantilever hook 

    ......................................................................................................................................... 129

    Figure 7.1: Integral polymeric lens housing ................................................................... 133

    Figure 7.2: Dimensions of snap-fits used in integral polymeric lens housing................ 133

    Figure 7.3: The curve of retention force of cantilever hook of lens housing.................. 135

    Figure 7.4: The web-based design tool for integral polymeric lens housing.................. 135

    Figure 7.5: Calculation of beam thickness at broken area for power supply’s snap-fit.. 136

  • 8/17/2019 Osu 1133282089

    16/182

     xvi

    Figure 7.6: Cantilever hook of power supply for Rockwell Allen Bradly PLC controller 

    ......................................................................................................................................... 137

    Figure 7.7: The detailed dimensions of the second case study ....................................... 137

    Figure 7.8: The retention force curve of cabinet snap-fit ............................................... 138

    Figure 7.9: The web-based design page for power supply’s cantilever hook................. 138

    Figure 7.10: The detailed dimensions of cantilever hook sample a................................ 139

    Figure 7.11: The detailed dimensions of cantilever hook sample b ............................... 140

    Figure 7.12: The detailed dimensions of cantilever hook sample c................................ 140

    Figure 7.13: The retention force curve of cantilever hook a........................................... 141

    Figure 7.14: The retention force curve of cantilever hook b .......................................... 142

    Figure 7.15: The retention force curve of cantilever hook c........................................... 142

    Figure 7.16: The web-based design page for cantilever hook sample a ......................... 144

    Figure 7.17: The web-based design page for cantilever hook sample b......................... 144

    Figure 7.18: The web-based design page for cantilever hook sample c ......................... 145

  • 8/17/2019 Osu 1133282089

    17/182

     

    1

    CHAPTER 1

    INTRODUCTION

    When used in plastics, snap-fits can be a simple, quick and cost-effective method

    of assembling two parts. When designed properly, parts with snap-fits can be assembled

    and disassembled numerous times without any adverse effect on the assembly. Snap-fits

    also aid in making products environmentally friendly because of their ease of

    disassembly, making components of different materials easy to recycle. Traditionally,

    snap-fit design methodology has been disorganized and anecdotal in nature, relying

    greatly on the skill and the experience of the individual design engineer. A popular

    source of design knowledge for snap-fit has been the design guides from resin suppliers

    such as Honeywell Plastics [1], General Electric Plastics [2] and Bayer Polymer [3].

    However the information disseminated by these guides is fragmentary and often

    inaccurate. There has been tremendous progress in the areas of more accurate retention

    equations, more accurate finite element models, identification of critical geometric

     parameters and high-performance topologies.

    In the current scenario most snap-fit designs are one-of-a-kind efforts with very

    little, if any, leveraging between different product families. Design knowledge generated

    in a particular project is leveraged for use in subsequent design only through the

  • 8/17/2019 Osu 1133282089

    18/182

      2

    experience of the designer, or through project reports describing the design process and

     performance data. It will be of great value if design information for snap-fit features

     became available to the designers in an abstracted as opposed to application-specific

    forum. What is needed are design equations, response surfaces, design heuristics,

    evaluation metrics and the like.

    It is often tedious to apply design equations for unique types of snap-fits to get

     parameters such as insertion and retention force. This is especially true if response

    surface modeling equations which can be algebraically long and tedious are used. For this

    reason an interface for the snap-fit design tool is advantageous. With the surge in growth

    and availability of the Internet and advanced programs such as Java Applet and Visual

    Basic’s Internet Information Service, it was decided that the easiest and most efficient

    way for distributing and maintaining a snap-fit design tool was to switch from a

    computer-based to a web-based application. With a web browser, a designer can access

    the snap-fit design tool independent of computer operating systems (i.e. versions of

    Windows OS, Mac OSX, Unix and Linux) from anywhere in the world. This chapter

     presents a brief introduction to snap-fit features, some commonly used terminologies, and

    then explains the motivation and goal of this research.

    1.1 Introduction to Snap-Fit Features

    The main performance attributes of a snap-fit which will be used in this thesis are:

    • Insertion Force ( i F  )  is the force that needs to be applied in the insertion

    direction of a snap-fit feature to engage it. Insertion force can be expressed as a

  • 8/17/2019 Osu 1133282089

    19/182

      3

    single maximum value or as a graph of the force versus position relative to the

    snap-fit feature (Figure 1.1) [4].

    • Retention Force ( r  F  )  is the force that needs to be applied in the separation

    direction of a snap-fit feature to disassemble it (Figure 1.2) [4]. If designed to be a

     permanent assembly, disengagement occurs due to fracture, permanent

    deformation or loss-of-engagement between the two mating parts.

    • Locking Ratio  is defined to be the ratio of the maximum retention force to the

    maximum insertion force of a snap-fit feature.

    Snap-fit are molded into plastic parts to provide attachment functionality. Unlike

    rivets, screws etc., which are discrete fasteners, snap-fit features are integral to the part.

    They thus help reduce part count and assembly time. The motions require for assembly

    are also usually along a single axis as other features on the parts should remove all

    degrees of freedom. Snap-fits have traditionally been used in lightly loaded application

    such as toys and other consumer products. However, with improvement in polymer

    technology and the rapid development of composite materials for structural applications,

    snap-fit features are now used in more demanding products. As an example, automotive

    under-hood applications that snap-fits have found use in includes air filter housings,

    throttle bodies, temperature and pressure sensors, and engine intake manifolds. These

    assemblies are expected to withstand harsh environmental conditions like high

    temperatures and pressures and contaminants like oil and corrosive gases. Lens housings

    in high-end projection television systems, pager housings, single-use cameras, and

    compact disc players are other examples of consumer products in which snap-fits have

  • 8/17/2019 Osu 1133282089

    20/182

      4

     been successfully applied. Quite often, snap-fits are used to provide secondary joining

    functionality to give the primary fastening (adhesive, ultrasonic welding) joints time to

    form. Commonly used snap-fit topologies are show in Figure 1.3 [4]. The primary cause

    restricting the use of snap-fit features in a larger number of polymeric products is the lack

    of confidence in their design process and performance attributes.

    0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-3

    -2.5

    -2

    -1.5

    -1

    -0.5

    0

    0.5

    1

    1.5

    Time

       I  n  s  e  r   t   i  o  n   f  o  r  c  e   (   N   )

     Figure 1.1: Typical experimental force curve for insertion of snap-fit features

    1.2 Motivation for Research

    The essential goal driving this research is the need for more accurate and versatile

    models of the performance of snap-fit design and a web-based snap-fit design tool that is

    independent of operating system platforms and can be distributed and upgraded

  • 8/17/2019 Osu 1133282089

    21/182

      5

    universally. As mentioned earlier, the models provided by the resin supplier are usually

    limited in scope and often inaccurate. An analytical model developed by Suri [4]

    overcame some limitations of current design guides. However it is unable to provide

    accurate prediction for high retention angles due to convergence issues. And with the

    development of the Internet, there is also a demanding for a web-based snap-fit design

    tool that is independent of all operating systems, easily accessible and can be universally

    upgraded by simply updating the design tool at the server location. The immediate need

    for an improvement in these areas forms the motivation behind the work done in this

    thesis.

    0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-3

    -2.5

    -2

    -1.5

    -1

    -0.5

    0

    0.5

    1

    1.5

    Time

       R  e   t  e  n   t   i  o  n   f  o  r  c  e   (   N   )

     Figure 1.2: Typical experimental force curve for insertion of snap-fit features

  • 8/17/2019 Osu 1133282089

    22/182

      6

     

    (a) Cantilever hook feature  (b) Post & dome snap-fit feature(Axisymmetric cross-section)

    (c) Bayonet & finger snap-fit feature(Two-dimensional cross-section)

    (d) Loop-hook snap-fit feature

    (e) Trap type snap-fit feature (f) Hollow-core hook snap-fit feature(Two-dimensional cross-section)

    Figure 1.3: Common snap-fit feature topologies

  • 8/17/2019 Osu 1133282089

    23/182

      7

     

    Figure 1.4: Equations for dimensioning cantilevers

    Figure 1.5: Force multiplier of insertion and retention

  • 8/17/2019 Osu 1133282089

    24/182

      8

    1.2.1 Snap-fit Design Approaches

    The development of equations for the detailed sizing of snap-fit features and for

     predicting their response can be accomplished using analytical, numerical or

    experimental method. The first two approaches, in particular, have been popular in the

     past. Because of cost and time issues, the experimental method is usually replaced by

    numerical method using finite element method to simulate experiments. In the following

    several sections, the analytical and numerical formulations will be discussed in detail.

    Simplified Strength of Materials Analytical Formulation

    Design guides from major resin suppliers use a strength of materials to develop

    simple equations (Figure 1.4 and 1.5) for the detailed sizing of snap-fits. For example the

    cantilever hook is one of the most common snap-fits and its basic shape is shown in

    Figure 1.6. It is typically idealized as a cantilever beam with a rigid catch at the end. The

    transverse force required to deflect the end of the beam ( P ) by an amount equal to the

    offset of the catch is determined using Euler-Bernoulli beam theory. This is related to the

    forces acting on the catch in the engagement direction ( i F ) to determine the insertion

    force for the feature (Figure 1.6). A similar approach is used to determine the retention

    force. Strength of materials formulations however suffer from a number of limitations.

    Consideration of the equilibrium of the catch will show that the beam is subject to both

    an axial load and a moment, in addition to the transverse load  P . The deflections due to

    these forces and moments cannot be found by simple superposition. The bending moment

    in the beam due to the axial component depends on the deflection of the end, rather than

     being constant. This effect has not been considered in previous snap-fit analysis [5]. The

  • 8/17/2019 Osu 1133282089

    25/182

      9

    assumption that the maximum deflection of the end of the beam during insertion equals

    the offset of the catch also introduces some inaccuracy, because the end of the beam

    rotates, in addition to the transverse deflection. The Euler-Bernoulli beam theory is based

    on small deformation and long-beam assumptions, which are commonly violated in real

    life.

    Figure 1.6: Cantilever hook and mating part

    Applied Mechanics Equations

    Suri [4] has worked on an analytical formulation that includes a full applied

    mechanics formulation and solution. It involved idealizing the catch as a rigid body

    supported on a flexible structure (Figure 1.7). A set of equations that comprehensively

    described the system in its deformed configuration is formulated. The equation system

    was iteratively solved for several such configurations to obtain a model of insertion and

    retention processes for snap-fits. The model showed excellent agreement with

    experimental results for most snap-fit geometries. However, as the retention angle

    approaches 90° it becomes impossible to find a solution to the set of equations.

    Insertion Direction

    Retention Direction

    Fi

    y

    P

    CatchCantilever Hook 

    Mating

    Part 

  • 8/17/2019 Osu 1133282089

    26/182

      10

     

    Figure 1.7: Suri’s idealized model of cantilever hook and mating part, shown in deformedconfiguration.

    The Numerical Formulation

    The numerical formulation approach involves the development of response

    surface equations and main effect plot using Design of Experiment (DOE) and statistical

    methods. The expected significant variables to the snap-fit’s performance and response

    variables are first identified. The portion of the design space that is of interest is then

    determined. An array of experimental designs spanning the region of interest is then

    created using standard statistical software such as SAS®

      and Minitab®

    . Physical or

    computational experiments at each of the points in the DOE array are conducted. Because

    of the cost related to manufacturing physical test samples and the test time, the use of

    computational experiments has been more popular in the literature. Typically, finite

    elements models of the snap-fit are created and queried for performance attributes.

    Response surface equations are fit to the values of the response variables at the

    experimental point. These equations can then be used to predict the performance of the

    snap-fit at points in the design space other than those in the experimental array. The

     possibility of satisfactorily extrapolating the equations to other regions of the design

  • 8/17/2019 Osu 1133282089

    27/182

      11

    space also exists. Main effect plots depicting the sensitivity of the response variables to

    the factors are also created for quick determination of the most important design

    variables. Such an approach has been used for the cantilever hook [5], the bayonet &

    finger [6, 7], the compressive hook [8] and the post & dome snap-fit features [9].

    While the numerical design formulation is an effective and efficient method for

    generating design equations for snap-fit, it has some disadvantages. Some limitations

    arise from it being a purely statistical method rather than being based on a fundamental

    understanding of snap-fit’s performance. The response surface equations are simply

    mathematical equations fit to the response variable values. They are very useful and

    effective when the analytical formulation is very difficult to model the response surface

    equation and the design space has just a single active physical phenomena.

    1.2.2 Web-based Snap-fit Design Tool

    Using the web as a development environment is a relatively new phenomenon

    called web-based application. The CAD/CAM area is also facing a transformation from

    computer-based to web-based applications. Computer-based application means that the

    design software packages need to be installed and operated on a standalone computer.

    Famous examples in CAD/CAM are AutoCAD®

      and ProEngineer ®. Eudora

    ®  and

    Outlook ®

     are also examples of computer-based application. Compared to computer-based

    applications, web-based applications refers to applications or services that are resident on

    a server that is accessible using a web browser such as Internet Explorer ® or Netscape

    ® 

    and is therefore accessible from anywhere in the world via the Internet and control using

    the Web interface that browser provides. A popular example of a Web-based application

  • 8/17/2019 Osu 1133282089

    28/182

      12

    would be Microsoft’s Hotmail®

      that is simply a large-scale email program that runs on

    Microsoft’s server computers. They’ve allowed anyone with Internet access and a

     browser to connect to their server and check email. So instead of using Eudora®

      or

    Outlook ®

     to check email, the client can check email by using Hotmail® anywhere. This

    approach has the following advantages:

    • It allows users to log onto the system from anywhere in the world as long asthey have a computer, an Internet connection and a web browser.

    • Applications are resident on the server instead of client’s computer so thatusers don’t need to worry about software distribution and updates.

    • User accessibility to data is definable.

    • System administration can be performed remotely.

    On the other hand, there are some limitations for web-based applications. The

    major concern with web-based applications is that of control and security. Although most

    Web-based application servers protect against hackers with some security systems, no

    computer on the Internet is truly hacker-proof even though some web-based applications

    further protect their customer’s data by encrypting it in some way. Another limitation is

    speed. Web-based applications usually execute quickly, but their response is noticeably

    slower. In other words, clicking on a button or link doesn’t result in an instantaneous

    reaction by the program. This is simply because of the time lag that it takes for data to

    travel via Internet connection from client’s computer to the application server and back.

    1.2.3 Constraint Management

    Constraint management is a way of planning, organizing, evaluating and

    controlling complex systems and it has been used in a number of fields such as

  • 8/17/2019 Osu 1133282089

    29/182

      13

    variational geometry based Computer-Aided Drafting systems, economical management,

    mechanical design, chemical process synthesis, and artificial intelligent based constraint

    satisfaction procedures. Most of work done in the field of constraint management is

    closely related to sparse matrix research since the underlying matrix representations of

    the designs tend to be quite sparse. Kinzel and his students [57-62] developed an

    interactive engineering design framework consisting of a constraint manager coupled

    with a friendly graphical user interface. Constraint management was implemented into

    this design tool and several algorithms – new specification, respecification,

    unspecification and reverse specification – were developed.

    1.2.4 Thesis Objectives

    This thesis’s objective is to make significant contributions to advancing the

    accuracy and effectiveness of snap-fit modeling techniques in these areas. There are two

    objectives for the current research.

    • Development of an improved numerical formulation for modeling the

    performance of a cantilever hook feature with a high retention angle.   As

    mentioned earlier, current snap-fits design guides and Suri’s analytical model are

    unable to predict the performance of high-retention cantilever hooks. Nonlinear

    and contact finite element analysis are able to overcome this limitation. Design of

    experiment and response surface methodology will also allow us to generate an

    empirical model to predict the performance of high-retention cantilever hooks.

    • Creating a web-based design tool for snap-fit features. Usually empirical

    models developed by statistical methods have many polynomial terms and are

  • 8/17/2019 Osu 1133282089

    30/182

      14

    tedious to use. A web-based, platform-independent design tool of snap-fits

    implemented those empirical models will be very convenient to use. Three

    different snap-fits, Cantilever Hook, Post & Dome and Bayonet & Finger, will be

     built in the design tool and constraint management will be applied. Modules such

    as sensitivity analysis and correction advisor will also being applied. Two

    different optimizations, single objective optimization and multiple objectives

    optimization, will be applied in the design tool.

  • 8/17/2019 Osu 1133282089

    31/182

      15

    CHAPTER 2

    LITERATURE REVIEW

    2.1 Numerical Formulation for Snap-fit Features

    The cantilever hook is the most widely utilized snap-fit feature (Figure 1.3(a)).

    Trantina [10] created a model of cantilever and imposed a displacement load on the dwell

    surface equal to the offset of the hook. This type of model allows the geometric and

    material nonlinearity of the hook to be modeled, and, therefore, produces a reasonably

    accurate prediction of the transverse stiffness of the hook. A serious limitation to this

    type of model, however, is that it cannot model hooks in retention. This is especially true

    for non-removable hooks with retention face angles that are close to or equal to ninety

    degrees.

    Hotra et al.  [11] modeled retention of a compressive finger by use of contact

    elements but did not model the insertion process. The modeling approach taken was to

    have the cantilever hook modeled as an elastic structure which is incrementally pushed

    into a retention block which is modeled as a rigid surface.

    Luscher [5, 12] applied a combination of finite element analysis and design of

    experiments to the cantilever hook snap-fit feature. A four-factor, two-level, orthogonal

    array was used to study insertion and a five-factor, two-level array was used to study

  • 8/17/2019 Osu 1133282089

    32/182

      16

    retention. The amount of engagement was also defined as a factor to incorporate the

    effect of warpage and shrinkage on the performance of the feature. The finite element

    results were compared to experimental data. Discrepancies between the two were

    explained. The sensitivity of the feature to geometric variables was discussed for

    insertion and retention separately. An optimal hook design formulation was also

     presented. The design equations developed by Luscher were first order and the retention

    face angles investigated were 65° and 90° degrees. With such a large spread of retention

    angles, it was not possible to accurately capture the subtle variations in retention force

     between angles from 85° to 90° degrees.

    The post & dome feature is a high performance snap-fit that is self-datuming and

    can take some shear loading in addition to retention (Figure 1.3(b)). It provides a higher

    locking ratio than traditional cantilever hooks, and its retention strength is less dependent

    on friction. Nichols [9] created two design arrays to study the post & dome snap-fit

    feature. The first is called the catch array; a focused array constructed to determine the

    optimal preload and molded-in undercut at the interface between the post & dome

    segments. Optimal performance is judged with respect to both the snap-fits maximum

    retention force and maximum locking ratio. The second array, called the macro array, is a

    comprehensive design of experiments array. The objective of the macro array is to

    generate design data for a variety of dome geometries. Design equations are provided by

    subsequent analysis of the finite element data using Response Surface Methods (RSM).

    The designed equations provide estimates for the maximum insertion force, insertion

    strain, and retention force applicable for an idealized post & dome snap-fit feature. To

    accommodate the snap-fit’s geometric nonlinearity, the post & dome are modeled with

  • 8/17/2019 Osu 1133282089

    33/182

      17

    three dimension 8-node brick elements via the MARC®

      finite element package to

    generate design data. This three-dimensional contact modeling utilizing an updated

    Lagrangian [13] approach provides more accurate displacement predictions than small

    displacement beam theory.

    A bayonet & finger snap-fit feature contains a bayonet, a retention finger and a

    support finger (Figure 1.3(c)). Wang et al. [14] analyzed a bayonet & finger feature using

    a contact type finite element model in ABAQUS. High-impact polystyrene (HIPS) was

    modeled as an elastic-perfectly plastic material. Finite element analysis predicted a snap-

    through failure model. The simulation was compared to results from experimental tests.

    Wang and Gabriele [15] created a finite-element model of the bayonet & finger

    snap-fit feature. The model is used to simulate the insertion and retention processes of the

    feature as well as snap-through and buckling phenomena. A “master-slave” approach is

    used for contact modeling in ABAQUS. An elastic-perfect plastic material model is used

    to describe the plastic behavior of high-impact polystyrene (HIPS). Finite-element

    analysis predicts a snap-through failure mode. The simulation is compared to results from

    experimental tests. Reasonably accurate correlation between analysis and experiment is

    observed for insertion. Results for retention are not presented.

    Lewis et al. [16]  expanded upon the previous work. A design of experiments

    approach with two and three dimensional finite-element methods was used to generate

    approximate linear response surfaces based on feature geometry which could calculate

    insertion and retention forces for the bayonet & finger feature. Five dimensions were

    chosen as the significant design factors. Two levels were chosen for each factor. A

    fractional factorial experiment design was used to reduce the number of experiments

  • 8/17/2019 Osu 1133282089

    34/182

      18

    (analyses). Sixteen different trials were performed for the study. Material properties were

    not considered as a factor in the experiment. Material properties for HIPS were used, with

    an elastic-perfect plastic material model. Insertion was modeled using a 2-D model and

    retention using a 3-D model. Sensitivity information was generated using a level average

    analysis technique. Design equations for predicting the insertion force and retention

    force based on part dimensions are presented in the paper. An attempt at confirmation of

    the validity of the equations using a sample feature design fails. The FEA analysis results

    for the feature show large discrepancies when compared to the design equation

     predictions. No experimental verification is presented.

    Shen [7] first found an optimum catch geometry of bayonet using an L-9

    orthogonal experimental array. Then another three-lever, five-factor experimental array

    was used to obtain the response surfaces. These response surfaces approximated the

    amount of over-engagement as well as the insertion and retention forces for values of the

    design factors. The calculations were done through a finite element package MARC®

    . A

    spring was connected between the root of the bayonet and an external node to simulate

    the base-part stiffness of the bayonet. The sensitivity of the values of this spring constant

    to the feature performance was investigated.

    The compressive hook (Figure 1.3(e)) is a common snap-fit used in thermoplastic

    automotive electrical connectors. It derives its name form its retention mechanism which

    uses compression to provide the locking force. Hotra et al. [17] use an approach similar

    to Wang et. al. [13] to model performance of compressive hook. The analysis is

     performed for 15% glass reinforced poly-butylene terephehalate (PBT). The material data

    is experimentally determined at 0.08%/sec strain rate. Bending and shearing failure

  • 8/17/2019 Osu 1133282089

    35/182

      19

    modes are investigated separately. The simulation results are compared to experimental

    data and are shown to be in good agreement. The paper emphasized the need for the use

    of advanced analysis techniques to model the performance of complex snap-fit features.

    Hotra et. al. [11] extended the above work. Poly-phenylene oxide (PPO) was also

    considered, beside PBT. Bending and shear were again considered in separate analyses

    and the lower of the two taken as the actual failure mode. Tests were performed with a

    hard gage as a mating part in one case, and the actual terminal as the mating part in the

    other case. This was done to study the influence of mating part flexibility on the

     performance of the feature.

    Roy [18] adopted an analytical approach for predicting the performance of the

    compressive hook feature. The hook was modeled as a cantilever beam with variable

    cross-section. In particular, it was divided into three difference beam sections and

    Castigliano’s theorem used to determine the load-deflection relationships. A graphical

    solution method was also presented. Stiffness matrices for the feature were derived using

    an approach similar to the finite element method. Expressions for the insertion

    (assembly) force were derived based on equilibrium of the feature. For retention,

     buckling was recognized as the primary model of failure. The Rayleigh-Ritz method was

    used to derive approximate expression for column buckling loads under different end

    conditions. The same method was also used to estimate eigen-frequencies for longitudinal

    and transverse vibration models of the feature. All the results were compared to results

    from a finite element analysis conducted using ANSYS.

    Other snap-fit features such as cylindrical snap-fit hinges and ball snap-fits, have

     been studied using numerical simulation (FEM) and design of experiment.

  • 8/17/2019 Osu 1133282089

    36/182

      20

    Matuschek and Michaeli [19] used a three-level, five-factor design of experiment

    to model the performance of cylindrical snap-fit hinge. Each experiment is a finite

    element model of the cylindrical snap-fit feature with the appropriate dimensions. A

    quadratic response surface is generated using the results of these analyses. The sensitivity

    of the maximum stress in the feature and the releasing force to the geometric variables is

     presented graphically. An equidistant grid strategy is used to partition the design space of

    the feature snap-fit identify the optimal set of dimensions for a set of given design

    criteria.

    Bader and Koch [20] were the first to use a viscoelastic material model in the

    finite element analysis of a ball type snap-fit feature. A “standard linear solid”

    viscoelastic model consisting of springs and dashpots was used. Plasticity was modeled

    through the introduction of an additional mechanical element that remains inactive below

    the yield point for the material. The dependency of the spring and dashpot parameters on

    strain rate was modeled using a logarithmic curve fit. Plastic damage, recovery and stress

    relaxation behavior were experimentally observed and phenomenological models were fit

    to the data. Finite element analysis of the insertion and retention processes was performed

    using the commercial software MARC and MENTAT. The results obtained using

    viscoelastic and elastic-plastic material models were compared to experimental test data.

    2.2 Materials Issues in Snap-fit Design

    There have been some efforts to promote the use of advanced material models for

     polymers in the field of snap-fit design. Most of these have, however, been limited to

    relatively simplistic elastic-plastic material formulations, in order to model residual strain

  • 8/17/2019 Osu 1133282089

    37/182

      21

    in the polymer. This thesis will attempt to extend the state-of-the-art in snap-fit analysis

     by demonstrating improvement in the accuracy of finite element analysis predictions. A

     brief review of past work dealing with improved material models for snap-fit design is

     presented below, to emphasize the benefit of the proposed approach.

    In order to evaluate the applicability and accuracy of experimental test data,

    Knapp et al. [21] compare three tensile test methods. It is well known that polymer

     behavior is dependent upon strain rate and that conventional tensile tests subject the

    specimen to variable strain during the test. In this work, tests were conducted at 0.002/s

    and 0.02/s strain rate. The experimental stress-strain data fit to an analytical curve and the

    value of tangent modulus (Et) at each strain value was determined. This material model

    was used in a finite element analysis and its results compared to experimental data. The

    authors propose that designers should use true stress-strain data gathered at strain rate

    values appropriate for their application. Trantina and Minichelli [22] describe a software

    developed for automating snap-fit finite-element analysis. Deflection limited models are

    analyzed. Some elementary results describing the effect of dimensions on stress and

    strain on snap-fits are presented. The authors use a 0.1/s strain rate for determining the

    material properties. The work is also summarized, in more detail, in Trantina and

    Minnichelli [10, 23].

    Sawyer et al. [24] investigate the applicability of commonly reported coefficient

    of friction (µ) values for snap-fit design. The analysis of surface friction between

     polymers is extremely complex [25-27]. Widely different values of µ have been reported

    in literature. For example, coefficients of friction (µ) ranging from 0.28 to 0.62 have been

    reported for unfilled nylon 6/6. In this study, friction measurements were made directly

  • 8/17/2019 Osu 1133282089

    38/182

      22

    from polycarbonate cantilever hook lock pair contacts, and their dependence on load,

    sliding speed and contact geometry was investigated. An experimental setup for

    accomplishing this is detailed. A power law dependence of µ on normal load is suggested

    in the paper. Changes in the value of µ with sliding speed were found to be minimal. The

    contributing mechanisms to friction, viscoelastic deformation and adhesion are discussed

     briefly. The authors contend that adhesive contributions occur at a scale smaller than that

    modeled by finite elements and as such should be included in the value of µ. On the other

    hand, the effect of viscoelastic deformation should be captured by the finite element

    model of the snap-fit and as such need not be included in the value of µ.

    2.3 Failure Models of Cantilever Hook

    In the past cantilever hook has always been selected primary in terms of retention

    force and the molding tooling costs. Failure modes of cantilever hook are usually ignored

    even although it was already recognized and understood that it will be an important factor

    to the performance of cantilever hook.

    Luscher [5] noted a snap-fit mode of failure called loss-of-engagement at small

    Percent Engagement (PE) values (Figure 2.1). In loss-of-engagement failure the snap-fit

    slips past the mating part without any mechanical damage to either half of the feature. At

    high  PE   values, the snap-fit’s failure mode switched to either a shear and/or tensile

    material failure.

    Luscher et al. [28] extended the previous work. They determined the exact

    dividing lines between these different modes of failure with respect to the  PE   (Figure

    2.2). Well-defined transitions between modes of failure were observed with the cantilever

  • 8/17/2019 Osu 1133282089

    39/182

      23

    hook. The different failure modes also observed at different levels of  PE . Three distinct

    failure modes were observed: shear failure, tensile failure and loss of engagement. Loss

    of engagement was the exclusive failure mode from 20 to 66  PE  while shear failure was

    the only mode of a failure from 70 to 100  PE . Finally, tensile failure occurred at all

    values over 100 PE . There was only one exception, a single shear failure point at the 40

     PE  level. It was considered a statistical anomaly, most likely due to a material, molding

    or alignment condition. Figure 2.3 shows an example from each of the two mechanical

    failure modes in the cantilever hook. As shown in the Figure 2.3, the direct shear failure

    seems to occur progressively over a certain distance. Note in this figure that the failure

    seems to have a “stair-step” pattern showing that the mating part is tearing out the catch

    starting at the retention face. It is not able to tear out the catch along a line parallel to the

    removal direction. The tensile failure case is different. The failure surface looks like it

    occurs at a single point in time.

    Figure 2.1: Percent engagement of cantilever hook

  • 8/17/2019 Osu 1133282089

    40/182

      24

     

    Figure 2.2: The different failure modes with respect to the PE

    (a) The tensile failure mode (b) The shear failure mode

    Figure 2.3: The two different mechanical failure modes for cantilever hook

    It is very important to identify the transition from one failure mode to another in

    order to accurately calculate the retention force provided by the cantilever hook feature.

  • 8/17/2019 Osu 1133282089

    41/182

      25

    2.4 Web-based Design Tool for Snap-fit

    The Internet is changing not only our daily lives, but also the professional field of

     product design and new product development.

    Wang et al. [29] developed a fluid simulation package integrated with the

    Internet, which allow the user to perform fluid simulation on the World Wide Web

    (WWW), to simulate the traditional Computational Fluid Dynamics (CFD) problem -

    Transient Natural Convection in a Cavity. This simulation package was developed using

    two different tools: one is Microsoft®

    /COM ActiveX and VBScript and the other is Java

    and Java Applet.

    Chung and Wright [30] proposed a web-based engineering framework for an

    infrastructure, which integers various advanced design/manufacturing systems such as

    CAD, FEM and CAM, provides relationships among product data, and coordinates with

    manufacturing system. Designers access this framework through web browsers and start

    the design with some initial parameters.

    Colton and Dascanio [31] described a vision and current development in a

    distributed design and manufacturing environment, and emphasized how current CAD

    tools will evolve to facilitate the distributed design and fabrication process. They also

     presented the development of a set of web-based design tools for fabricating parts using a

    machining process via the Internet and experiments on machining 2-1/2 D and freedom

     parts through their Java-based design tool showed the feasibility for a networked

    machining service via the Internet.

    Kim et al. [32] developed a web-accessible CAD tool that assists a designer in

     producing designs that are manufacturable on a 3-axis milling machine, which simplify

  • 8/17/2019 Osu 1133282089

    42/182

      26

    the previous complex testing for manufacturability by focusing on 3-axis milling only

    and by making use of new fixturing techniques using a plastic compound or a metal alloy

    for reference-free part encapsulation (RFPE). They also build a Java-based front end of

    this design tool that can be downloaded as an applet over the Internet.

    Velάsquez et al. [33] created a system named Tool Trial System (TTS) which is

    capable of collating and disseminating information relating to tool trials amongst a

    variety of user groups. TTS provided a platform from which is possible to submit and

    retrieve highly specific technical tooling data on the WWW can be downloaded by

    remote users in the form of Java applets, through any computer with Internet connection

    and using conventional Java enabled browsers without the requirements of using middle

    tiers software or hardware between clients and server sides.

    Rajagopalan et al. [34] built an Internet-based infrastructure to provide designers

    with access to multiple layered-manufacturing services. This system contains three

     primary operatives: Design Clients, Manufacturing Services and Process Brokers. The

    Design Clients allows designers to submit completed designs for algorithmic

    decomposition, or alternately, to compose a design from primitives and library

    components that have been primed with some process-related information.

    Manufacturing Service consists of a highly automated machine that can be used to build

    ceramic parts, and the associated software components for design decomposition, process

     planning and machine control. The Process Broker implements a number of supporting

    services including process selection and optimal part orientation.

    Ebbesmeyer et al. [35] described a web-based tool Virtual Web Plant (VWP), a

    tool to integrate 3D models from various CAD plant design tools and to display them

  • 8/17/2019 Osu 1133282089

    43/182

      27

    interactively through Internet. Using the application of object-oriented database, it is

     possible to define various views of the logical plant structure so that the user is able to

    navigate easily through both the plant structure and the project documentation. The

    special advantages of an objected-oriental database for the storage of the graphical data

    are also shown.

    Huang et al. [36] described a web site, WAPIP, which has been developed

    specifically to support new product introduction activities. It provides databases for

    software vendors and researchers to register their web applications with the “wapip”

    search engine. It also provides facilities to support practitioners in product design and

    manufacture to search rapidly for the right web applications suitable for solving their

     problems.

    Many response surface equations for snap-fit feature’s performance [5, 9, 15]

    have already been studied using DOE and FEM. But it is often tedious to look for design

    equations for unique types of snap-fits to calculate the insertion force, the retention force

    and the locking ratio. If found, these equations are usually long, complex, and difficult to

    use. For this reason, an easy using front end will be very helpful during the designing

     process of snap-fit features.

    Oh et al. [37] created a calculator (Figure 2.4) in which the design equations of

    seven snap-fit features (annular snap, bayonet & finger, post & dome, cantilever hook,

    cantilever-hole, compressive hook, L-shaped hook and U-shaped hook) were

    implemented. This calculator aids in designing snap-fits to meet specific loading

    requirements by allowing the designer to size the feature to obtain desired estimates for

    maximum insertion and retention forces. This calculator was developed in JAVA®

     

  • 8/17/2019 Osu 1133282089

    44/182

      28

    language that is independent of operating system platforms and can be distributed at a

    company site-wide over an intranet or worldwide over the Internet. This makes it easily

    accessible to a user, and universal upgrades can be achieved by simply updating the

    software at the server location.

    Figure 2.4: The IFP snap-fit design tool

    Honeywell Plastics [1] developed an online Snap-Fit Design Workspace (Figure

    2.5). It is a web-based application that serves as an engineering tool in snap-fit design.

    The program provides a workspace for the designer to test different scenarios by

    adjusting various input parameters and selecting different engineering material for the

    snap-fit design. The following snap-types were included: five different uniform beams,

  • 8/17/2019 Osu 1133282089

    45/182

      29

    two different tapered beams, two different U  shaped beam cases, and an L shaped beam.

    The cantilever beam formulas used in conventional snap-fit design poorly estimates the

    amount of strain at the beam/wall interface because they do not include the deformation

    in the wall itself. To obtain a more accurate prediction of total allowable deflection and

    strain for short beams, a deflection magnification factor was applied in this design

    workspace.

    Figure 2.5: AlledSignal Plastics’ snap-fit design guide

    Eastman chemical company [38] also developed an interactive online support tool

    (Figure 2.6) that receives inputs from the user to generate a technical recommendation for

    cantilever snap-fit design. This design tool calculates the theoretical strain that occurs

    when a cantilever is deflected. And the maximum strain occurs on the outer layer of the

  • 8/17/2019 Osu 1133282089

    46/182

      30

    thickness, usually at the base of the latch. Strains can also be calculated for a latch,

    which varies linearly in both thickness and width from the base of the tip to latch. After

    the users specify the material, its brand and geometry dimensions for cantilever, this

    design tool will predict the deflection force and outer fiber strain.

    Figure 2.6: Cantilever snap-fit design tool from Eastman chemical company

    Jeff Raquet [39] used Java Applet to create a snap calculator (Figure 2.7) to assist

    in the development of correct parameters in the design of a plastic snap-fit. Designers

    need to select a material and give values to several specified inputs. Then based on these

  • 8/17/2019 Osu 1133282089

    47/182

      31

    inputs, the snap-in force, engagement force and stress at root are computed. It also

     provide an “optimize” option to minimize the snap-fit’s volume.

    Figure 2.7: Jeff Raquest’s snap-fit calculator

    GE Plastics [2] (Figure 2.8), Engineers Edge [40, 41] (Figure 2.9 and Figure 2.10)

    and Berkeley Manufacturing Institute of University of California - Berkeley [42] (Figure

    2.11) also developed similar online snap-fit calculators. All of them used conventional

    design equations based on small deformation and long beam assumptions.

  • 8/17/2019 Osu 1133282089

    48/182

      32

     

    Figure 2.8: GE Plastics’ snap-fit wizard

    Figure 2.9: Engineers Edge’s snap-fit straight beam calculator

  • 8/17/2019 Osu 1133282089

    49/182

      33

     

    Figure 2.10: Engineers Edge’s snap-fit tapered beam calculator

    Figure 2.11: Brock & Wright’s design tool for snap fits

  • 8/17/2019 Osu 1133282089

    50/182

      34

    2.5 Constraint Management

    For the current research, constraint management will be an important section. So

    it’s necessary to review the literature of Constraint Management (CM). Constraint

    Management techniques have been used in a number of fields, such as variational

    geometry based Computer-Aided Drafting systems, mechanical design, chemical process

    synthesis, and artificial intelligent (AI) based constraint satisfaction procedures. Most of

    the work done in the field of constraint management is closely related to sparse matrix

    research since the underlying matrix representations of the designs tend to be quite

    sparse.

    Steward [43, 44] was one of the earliest investigators who examined the structure

    of simultaneous equations. Algorithms to partition and reorder directed graphs were

    shown by Christensen and Rudd [45].

    Duff [46] presented an excellent survey paper on sparse matrix research that

    included a survey of algorithms for matching and decomposing sparse matrices. Later,

    Duff and Reid [47] developed an algorithm (Harwell MC13A) to permute a square

    occurrence matrix to its block triangular form in order to determine simultaneous

    equations sets. Their algorithm was based on Tarjan’s algorithm for determining the

    strong components of a directed graph. In order to accomplish its purpose, MC13A

    required the original matrix to be converted to a zero-free diagonal (maximum

    transversal) form. They  compared various algorithms to achieve this configuration and

    developed algorithm MC21A to permute a sparse, square matrix to the maximum

    transversal form.

  • 8/17/2019 Osu 1133282089

    51/182

      35

    Shacham [48] presented a new method to partition simultaneous nonlinear

    algebraic equations into smaller irreducible subsets. Prasad and Kinzel [49] presented an

    algorithm to perform decomposition of incidence matrices as the number of degrees of

    freedoms is reduced in a mathematical model. Akin [50] refers to the occurrence matrix

    as a Boolean Inference Array and presents an algorithm for the detection of solvable sets

    of nonlinear equations that is quite similar to the work of Prasad and Kinzel. The work of

    Pothen and Fan [51] is aimed at converting a sparse rectangular matrix to its block

    triangular form. Their two stage decomposition techniques is based on the Dulmange-

    Mendelsohn decomposition algorithm.

    A number of researchers have studied constraint management in the field of

    variational geometry. Sutherland [52] was amongst the first researchers to treat 2-

    dimensional drafting as a constraint satisfaction problem. Later works of Lin and Light

    [53], Gossard [54], Serrano [55] added more to the mathematical basis behind CM.

    Serrano used directed acyclic graphs (DAGs) to determine dependencies among variables

    and to identify sets of simultaneous equations. A constraint modeler was also shown that

    allows for interactive addition/deletion of constraints. Chung and Schussel [56] compared

     parametric and variational geometry environments and described a commercial package.

    Srinivasan [57, 58] presented the Design Shell: an interactive engineering design

    framework consisting of a constraint manager coupled with a friendly graphical user

    interface. Sridhar [59, 60]  formulated strategies to interactively handle inequality

    constraints in the design shell framework. Agrawal et al. [61] identified and developed

    various constraint management, numerical solution, and numerical optimization

    methodologies necessary for the performance of the shell as an interactive and effective

  • 8/17/2019 Osu 1133282089

    52/182

      36

    design tool. Thomas [62] extended the design shell constraint management strategies

    towards the development of a framework for performing term designs.

  • 8/17/2019 Osu 1133282089

    53/182

      37

    CHAPTER 3

    FINITE ELEMENT ANALYSIS OF CANTILEVER HOOK WITH A HIGH

    RETENTION ANGLE

    This chapter describes the finite element model of cantilever hook with a high

    retention angle. Firstly three different nonlinearities of finite element analysis were

    described: Moving Contact, Geometric Nonlinearities and Material Nonlinearities. Then

    finite element analysis techniques applied in this thesis were explained because in a

    typical FEA simulation usually 80%-90% time is spent to find the best parameter settings.

    Finally a two dimensional, contact FEA model was created in MSC.MARC, and the

    results were compared to the sample testing results with the same geometry to verify the

    accuracy of the FEA model.

    After an accurate FEA model was obtained in this chapter, in the next chapter it

    was been used for the simulation of each design point and an empirical model was

    generated by using Design of Experiment (DOE) and Response Surface Methodology

    (RSM).

    3.1 Combination of FEA with DOE

    Modern finite element analysis and design of experiments techniques can be

    combined into a powerful quality-engineering tool [63]. FEA has been widely used to

  • 8/17/2019 Osu 1133282089

    54/182

      38

    simulate the snap-fit behaviors in the recent years and proved to be a powerful tool to get

     better understanding of snap-fit performance. Usually FEA users worked in the way of

    “one problem-one model” interactively. That is, for a given problem (say, a cantilever

    hook with certain geometry and under certain working condition such as loading and

     boundary condition), designers usually build a fixed model, solve it and get the required

    results (the insertion force, the retention force, etc). This has proved to be useful as

    shown in many applications.

    However, when using FEA to do DOE and design optimizations, the snap-fits

    have similar geometry but the parameters of components (dimensions and material

     properties) are changeable. Generally, even a simple DOE often involves more than ten

    runs. Working in one problem-one model way, we need to build many models

    interactively or manually which is time consuming.

    Fortunately, in MARC FEA package, besides working in interactive mode, it also

    allowed to use “PROCEDURE” file (Appendix A). “PROCEDURE” file allows users to

     build the model in terms of parameters instead of fixed value, which in turn make it very

    easy to change the design. Compared to the traditional one problem-one model method,

    the combination between DOE and FEM has the following advantages:

    • By expressing all the parameters in terms of variable the model becomes very

    flexible. In order to rebuild other models, designers just need to change the

     parameters’ value. This feature is especially suitable for DOE and design

    optimizations.

    • Store the model in text file (generally in size of a few kb) instead of in db file (in

    size of a few Mb) to save space.

  • 8/17/2019 Osu 1133282089

    55/182

      39

    • Run batch of models overnight to make full use of FEA system.

    3.2 Nonlinear Structural Analysis of Snap-Fits

    The reasons that the snap-fits design guides are not accurate for cantilever hook

    with a high retention angle are material and structural nonlinearities. Suri’s analytical

    mathematical model has difficulties to converge for high-retention cantilever hooks.

     Nonlinear structural behavior arises from a number of causes, which can be grouped into

    these principal categories:

    • Moving Contact

    • Geometric nonlinearities

    • Material nonlinearities

    For cantilever hook with a high retention angle, all these three nonlinearities exist

    in the retention stage. The following paragraphs explained these three nonlinearities in

    details.

    3.2.1 Moving Contact

    Many common structural features exhibit nonlinear behavior that is status-

    dependent. Status changes might be directly related to load, or they might be determined

     by some external cause.

    Situations in which contact occurs are common to many different snap-fit

    applications. Contact problems are highly nonlinear and require significant computer

    resource to solve. Contact problems present two significant difficulties. First, users

    generally do not know the regions of contact until you’ve run the problem. Depending on

    the loads, material, boundary conditions, and other factors, surfaces can come into and go

  • 8/17/2019 Osu 1133282089

    56/182

      40

    out of contact with each other in a largely unpredictable and abrupt manner. For example,

    during the retention of cantilever hook there actually exist several contact pairs

    depending on contact locations. Since the deformation of local contact areas it’s very

    difficult to identify the point when these contact pairs change. The second reason to cause

    contact problem highly nonlinear is that most contact problems need to account for

    friction. There are several friction laws and models to choose from, and all are nonlinear.

    Frictional response can be chaotic, making solution convergence difficult.

    3.2.2 Geometric Nonlinearities

    Small deflection and small strain analyses assume that displacements are small

    enough that the resulting stiffness changes are insignificant. In contrast, large strain

    analyses account for the stiffness changes that result from changes in an element’s shape

    and orientation. All design guides of snap-fits and Suri’s analytical model were based on

    small deformation theory. However cantilever hook with a high retention angle

    experiences large deformations, its changing geometric configuration can cause the

    structure to respond nonlinearly.

    3.2.3 Material Nonlinearities

    A number of material-related factors can cause your structure’s stiffness to

    change during the course of an analysis. Nonlinear stress-strain relationships of plastic,

    multilinear elastic, and hyperelastic materials will cause a structure’s stiffness to change

    at different load levels (and, typically, at different temperatures). Creep, viscoplasticity,

    and viscoelasticity will give rise to nonlinearities that can be time-, rate-, temperature-,

    and stress-related. Most polymeric materials used for cantilever hooks have a time-

  • 8/17/2019 Osu 1133282089

    57/182

      41

    dependent behavior, in which a certain amount of deformation is recoverable over time.

    For the purpose of numerical modeling, it was assumed that the material was time-

    independent.

    For all design guides of snap-fits none of these material nonlinearities were

    included in their design guides. All of them assumed that the maximum stress always

    under the yield stress that means no plastic deformation occurred during the retention. So

    that in order to overcome the inaccuracy caused by the ignoring of plastic deformation,

    it’s necessary to include the material nonlinearity in the FEA model. Several plastic

    material options are available in most of FEA packages such as MSC. MARC and

    ANSYS to describe material plasticity behavior. These options are the Bilinear

    Hardening, Multilinear Kinematic Hardening, Nonlinear Kinematic Hardening, Bilinear

    Isotropic Hardening, Multilinear Isotropic Hardening and Nonlinear Isotropic Hardening.

    Since in this thesis only Bilinear Hardening was used to describe the material plasticity

     behavior of plastics, the Bilinear Hardening option was shown in Figure 3.1. Although

    this is an approximation of the actual material’s curve it is an improvement over ignoring

     plasticity. For more details of other options, please refer to [64].

    3.3 Numerical Model of Cantilever Hook Feature

    In order to compute the insertion force, insertion strain, retention force and

    locking ratio of cantilever hook, the author created a two-dimensional, contact and

    nonlinear finite element model on Marc for each experiment. A mesh was generated

    using plane stress quadrilateral elements with full integration, which can prevent

    instability called hourglass mode [13] in a bending dominant problem. An example of the

  • 8/17/2019 Osu 1133282089

    58/182

      42

    meshed cantilever hook model on Marc is illustrated in Figure 3.2. A fine mesh is

    required on the insertion face, dwell face and retention face of cantilever catch where are

    the crucial regions at engagement and disengagement.

    Figure 3.1: The actual and bilinear stress-strain curve for plastic material

    3.3.1 Modeling Techniques for Contact Problem

    The modeling of the contact conditions is an advanced technique in the finite

    element method. To simulate the contact condition of the finite element models, the

    cantilever hook was specified as a deformable contact body and the fixed displacement

     boundary conditions were applied to the end of the cantilever hook (Figure 3.2). The

    mating part was defined a rigid body and the nodes on it were tied together as a group

    and fixed in the vertical direction.

    The analysis of surface friction between polymers is extremely complex [24-26].

    The classical "laws" of friction in which friction is proportional to applied load, and is

    independent of area of contact, velocity, and time of contact are poorly held for most

     polymers. In addition the conditions of a sliding snap-fit in terms of its speed, area of

    Inelastic

    Region

    Elastic

    Region

    Strain

    Stress

    Yield

    Stress 

    Inelastic

    Region

    Elastic

    Region

    Strain

    Stress

    Yield

    Stress

    Workhardening

    Slope

    (a) An actual stress-stain curve (b) Bilinear hardening option

  • 8/17/2019 Osu 1133282089

    59/182

      43

    contact, and residence time are not usually replicated in resin manufacturers tests and

    reported friction values. Rather than delve into this area, which is not a key focus of this

    research, it will be assumed that background research will be done to determine a value

    of coefficient of friction which is appropriate for the snap-fit in question. Therefore in

    these