A Computable Fastener Representation to Support ...

Proceedings of DETC’97: 1997 ASME Design Engineering Technical Conferences and Computers in Engineering Conference September 14-17, 1997, Sacramento, California

DETC97/CIE-4310 A COMPUTABLE FASTENER REPRESENTATION TO SUPPORT COMPUTER-AIDED CONFIGURATION DESIGN FOR THE LIFE CYCLE

Brian Harper, Zahed Siddique, and David Rosen* Systems Realization Laboratory George W. Woodruff School of Mechanical Engineering Georgia Institute of Technology Atlanta, GA 30332-0405 ABSTRACT Most of a product’s life-cycle characteristics are determined during the configuration design stage, when the product’s components are selected and arranged spatially and logically. One set of choices that determines many life-cycle characteristics is fastener type selection. In this paper, an assembly modeling representation is presented that supports changes in fastener types and fastening mechanisms while maintaining consistent degrees-of-freedom among fastened components. A fastening mechanism template and a corresponding instantiation algorithm have been developed for tensile-compressive fasteners. The template consists of four main elements: an assembly representation template, CSG tree fragments (to allow geometry construction), geometric constraint templates, and a parametric relationship template to integrate analysis equations into fastener models. Each particular fastener type is modeled by a specific template that is developed manually using the general template as a guide. The instantiation algorithm maps a particular fastening template onto an existing assembly model (assemblies, components, geometry, and mating relationships) in order to add fasteners to a product. A similar fastener substitution algorithm enables the replacement of one fastener type with another. The use of the algorithms is illustrated in the configuration design of an automotive center console. The paper concludes with a brief demonstration of how fastener selection affects life-cycle product characteristics.

*

1.0

INTRODUCTION Life cycle concerns such as recycling and service are becoming increasingly important to engineering manufacturers. The discipline of Design For the Life Cycle (DFLC) is emerging from the study of engineering design, manufacturing, materials, economics, etc. At the center of this discipline is the concept of making informed design decisions with information from all relevant life cycle viewpoints. Appropriate design and analysis tools are needed to support this decision making. Recent research is providing the knowledge base for these tools, but tool development is proceeding slowly. One reason for the slow pace is the difficulty in exploring design alternatives and analyzing them using existing CAD tools. In this paper, we address an important aspect of these issues, namely the addition and replacement of fasteners into a product model. The choice of fasteners determines many characteristics of a product’s assemblability and disassemblability, which drives important recycling, servicing, and demanufacturing properties. Fasteners are one part of a product’s structure, the arrangement and connections of modules within a product. Many of a product’s life cycle characteristics are determined during the configuration design stage, when a product’s structure is developed (Dixon et al. 1988). In general, CAD system support for configuration design is limited to allowing a designer to assemble a product model from existing CAD models of components. Compare this with variational and parametric modeling technologies. With these technologies, a designer can create a parameterized model and explore many similar designs by changing parameter values and having the CAD system create updated models.

Corresponding author. Contact info: (404) 894-9668, fax: (404) 894-9342, [email protected]

1

Copyright © 1997 by ASME

There is no similar support for making configuration changes to a design, for example, swapping one fastener type with another (a bolted-joint with a snap-fit). Geometric primitives from one component must be deleted and replaced with other geometric primitives (replace hole with a snap feature), then new assembly relationships must be created among the new geometric elements. More comprehensive configuration changes are even more tedious to perform, often necessitating the replacement of entire component models. This is a point of frustration when the models are to perform similar or identical functions, as when fastener types are swapped. Solution of the configuration design problem will require a comprehensive, computable model of functionality, something that is many years away. Our approach to the problem of fastener selection and replacement is to develop a general, computable model of a class of fasteners (fastener template) and an instantiation algorithm that inserts or replaces fastener types within this class. A limited model of functionality is embedded within the fastener template. With this capability, though limited, we can quickly generate various design alternatives and explore their life-cycle characteristics since explicit representation of fasteners and assembly relationships enables easy access to relevant product information. Our presentation in this paper begins with a literature review on assembly representations and an introduction to our assembly modeling representation in Section 2. The fastener template and instantiation algorithm are described next, which build upon the assembly representation. Fastener selection and replacement are illustrated in the design example in Section 4. We also demonstrate how the resulting product models are used to support life-cycle assessments. A critique of these capabilities is included in the final section. 2.0

ASSEMBLY REPRESENTATION

2.1 Literature Review Before our assembly representation is discussed, a literature survey of previous work in the area is warranted. Wesley et al. (1980) and Lieberman and Wesley (1977) present an assembly modeling system called AUTOPASS. The system stores components and assemblies as nodes in a graph structure; links between the components store the type of relationship between the components. The structure is difficult to traverse due to the non-hierarchical nature of the graph. Lee and Gossard (1985) introduce a hierarchical assembly graph structure used to represent assemblies, subassemblies, mating relationships, and components. This information is stored in a set of arrays or data structures, which are in turn connected by ‘virtual links’. The virtual link is “the complete set of information required to describe the relationship and the mating features between the mating pair”. It is argued that this structure removes the need for explicit transformation matrices for each component, while storing realizable assemblies and assisting in tolerance analysis.

Rocheleau (1992) created an assembly representation based on the work of Lee and Gossard. A structure known as a connectivity set is used to store topological and mating relationship information. The components and connectivity sets make up the assembly tree for the product. Thus, the assembly is represented in a hierarchical manner. Xiaolin and Shensheng(1996) present an assembly relationship matrix in which pointers containing information about an assembly are stored. The diagonal entries of the matrix store information about a component (parameters, CSG construction, etc.), while the off-diagonal entries store information about the interaction between components, such as assembly and constraint relationships. The matrix can be used on the subassembly level (containing groups of components) as well as the assembly level (containing groups of subassemblies). The IKA system of Bachman et. al (1993) consists of assembly modeler built on top of a feature modeler, which is in turn built on top of a geometric modeler. Assembly features that define relationships between parts are utilized in creating the assembly model. These assembly features contain pointers to adjacent geometry, as well as information about the part and its geometry. In this manner, an effectoriented assembly that contains information useful to the designer can be created. Shah and Rogers (1993), in addition to a more extensive literature review, present a product modeling approach using structures that ‘define relationships between assemblies, parts, features, feature volume primitives, and evaluated boundaries’. These structures can then be used to infer information about the assembly, such as component locations. 2.2 Our Representation We use a product representation suitable for the top-down design of assemblies, one that emphasizes hierarchical structure and mating relationships. Our approach to modeling hierarchical structure is to allow designers to specify assembly modules (sub-assemblies), sub-modules, etc. in a sub-assembly hierarchy. Designers can break down modules into components, or can design components, then incorporate them into modules. While modules are being defined, designers can specify mating relationships between components. An underlying strategy in developing our assembly representation is to identify a small set of fundamental building blocks with which more complex structures can be built. In this case, the building blocks are the fits and against conditions (Popplestone et al. 1980) from which more complex mating relationships can be constructed.1 The fits and against conditions are called base mating relationships. Composite mating relationships can be defined as combinations of base relationships. For more complete product representations, additional base relationships will be needed, for example line contacts as in a

1 In the against condition, two planar surfaces are required to be coplanar and have normal vectors in opposite directions. The fits condition constrains the center-line of a cylindrical surface (peg) to be coincident with the center-line of a hole (concave cylindrical surface).

2


cam-and-follower and more constraints for more general curved surfaces. Components are represented using form features and typical geometric construction approaches including constructive solid geometry (CSG). Specifically, we use form features in which a feature is defined by a set of geometric primitives and operations arranged as a CSG tree (Shah 1991). Geometric primitives and features are defined with size parameters (e.g., a cylindrical solid has radius and length parameters), while components and assemblies may have size parameters defined by the designer. Handles, such as center-lines, center-planes, and certain surfaces, are introduced into features as useful geometric abstractions. In specifying mating relationships, it is useful to reference handles instead of the geometric modeler-specific surfaces. Parametric modeling capabilities are included that enable positions and sizes of features, components, and modules to be controlled and updated during design. See Shah (1991), Bronsvoort and Jansen (1993), and Rosen (1993) for more complete descriptions of feature-based design. With the information stored in this representation, we hypothesize that life cycle assessments are enabled and configuration design is better supported than in current CAD systems. To summarize, the proposed assembly representation consists of three structures: • Sub-assembly hierarchy that records the decomposition of the product into modules; • Component mating graph (CMG) that records mating pairs of components; • Handle mating graph (HMG) identify particular mating handles on two components; Additionally, the product representation contains two more structures: • Constraint graphs of parametric relationships; • Feature and geometric construction definitions for components. As an example, a simple cup-lid assembly is shown in Figure 1 with hinged-joint and snap-fit mating relationships between the cup and lid. In this example, these relationships are modeled using base against and fits relationships. CMG is given by the components, the four against and fits mating relationships, and the solid directed arcs, while HMG consists of feature handles (triangle nodes), the mating relationships, and the dashed directed arcs. For simplicity, only the snap, slot, and hinge features are modeled, although additional features could be used. Also, the sub-assembly hierarchy and HMG are not included. To provide the designer with better design primitives, composite mating relationships can be defined, used, and represented as combinations of against and fits base relationships. For example, a hinged-joint relationship can be defined as a combination of one against relationship and one fits relationship as shown in Figure 2. Similarly, a snap-fit mating relationship can be defined as two, three, or four against relationships (ours consists of two). Using these composite mating relationships, the cup-lid assembly can be represented as shown in Figure 3. Provided that only against and fits relationships are needed, it is very easy to add

additional composite mating relationships by defining them as aggregations of base relationships. In fact, our snap-fit and hinged-joint relationships were defined in this way in a few minutes.

a)

screendump of cup-lid

CUP

LID Against

Hinge Bottom

Fits

Hinge Top

Against

b) Figure 1

Slot

Fits

Snap

assembly graph

Assembly representation of cup-lid assembly.

AGAINST FITS

Figure 2 Hinged-joint model illustrating against and fits mating relationships.

3


3.0

Hinged-Joint CUP

LID Against

Hinge Bottom

Fits

Hinge Top

Snap-Fit

Against

Snap

Slot

Fits

Figure 3 Assembly representation of cup-lid assembly with composite mating relationships. Tensile E lements

... n-1

n B ot tom Component

a)

n-2

3

1

2

Top Component

Compressive elements

General form for fastener. hj

h 2i-1

h2i

b) Compressive Element Handles Figure 4 General fastener elements and handles

FASTENER REPRESENTATION Our fastener representation is an extension of the assembly representation described above. Fasteners are implemented as composite mating relationships that carry additional information such as fastener geometry important to instantiation in assemblies. The class of fasteners explored is the set of fasteners that utilize pure tension and compression, rather than shear or friction. Screws, bolts, and snap-fits are examples of tensile-compressive fasteners. Figure 4a shows an example of a fastener of this type. Note that the elements being fastened are in compression while the fastening agent(s) are in tension. A typical compressive element is shown in Figure 4b, with two planar handles and a centerline handle identified. In this representation, we identify a top component, a bottom component, and a set of zero or more middle components. 3.1 Fastener Template As shown in Table 1, the general fastener template consists of four elements: the assembly representation, CSG tree, geometric parameters and geometric constraints. The assembly representation contains information on the fastening and compressive agents, as well as mating relationships between the two. Information on the orientation of the fastener is contained in the assembly representation. The CSG tree gives information on the construction of the fastener, as well as information on what operations must be performed on the compressive elements for instantiation of the fastener. The model constructed is parameterized according to the information in the geometric parameters field. Finally, the geometric constraints detail the necessary conditions for the fastener to be instantiated correctly - such as collinear centerlines of the compressive elements. In the assembly representation section of Table 1, the expression for MA denotes against mating relationships among all components in the fastening relationship. The expression implies a sequence of components and mating relationships, arranged linearly as in Figure 4. That is, starting from the right, the sequence of components is The expressions for MHA and MHF are formulated similarly for handles. This fastener template is then used as the guide to construct templates for specific fastener types, as presented in the next sub-section.

4


Assembly Representation: T = set of component(s) in tension (fasteners) C = set of components in compression X =T∪ C M = set of mating relations; MA = against mating relations; MF = fits mating relations H = set of handles; HP = planar handles; HC = center-line handles

M M M M

= {( xi , mi , xi +1 ) | x1 , x p+1 ∈ T , x 2 K x p ∈ C, m ∈ M A , i = 1K p} F = {( x1 , m j , x j +1 ) | x1 ∈ T , x n ∈ X , x 2 K x p ∈ C, m ∈ M F , j = 1K n − 1} HA = {( h2 i −1 , mi , h2 i ) | parent ( h2 i ) = parent ( h2 i +1 ), h ∈ H P , m ∈ M A , i = 1K p} HF = ( h1 , m j , h j +1 ) | h1 ∈ T , hn ∈ X , h ∈ HC , m ∈ M F , j = 1...n − 1 A

{

}

n = | X |; p = | M A | CSG Tree:

(fastener specific) Geometric Constraints: ∀(hi , m, hj ) ∈M ∀(hi , m, h j ) ∈M

, collinear(hi , hj ) AH , coplanar ( hi , h j )

FH

Parametric Constraints: (fastener specific) Table 1

Generic Fastener Template.

3.2

Example Fastener Models In order for instantiation of a particular fastener to occur, a specific fastener template must be created based upon the general template described above. This section describes such templates for two types of fasteners - the bolted joint and the snap-fit. Shown in Table 2 is the snap-fit template, without parametric constraints. In this case, only the component that carries the snap feature is a tensile component. The expressions for mating relations are similar to those of the generic fastener template. The CSG tree for the tensile component is shown in Figure 5a, while the CSG trees for each compressive component are shown in Figure 5b. Figure 6 shows the assembly representation for the snapfit fastener without explicitly showing the compressive elements. Here, it is assumed that the base of the snap has been joined to the top surface of the bottom component (this is included in the CSG representation). Therefore, only two mating relationships are required. The bottom surface of the snap head must be against the top surface of the top component. In addition, the centerline of the compressive element(s) and the centerline of the snap-fit share a fits relationship.

U

a)

b)

5

Snap feature.

Hole feature to be added to compressive elements. Figure 5 Snap feature CSG trees


Assembly Representation: T = {component with snap = S} C = set of components in compression X =T∪ C M = set of mating relations; MA = against mating relations; MF = fits mating relations H = set of handles; HP = planar handles; HC = center-line handles

M M M M

= {( x i , m i , x i +1 ) | x1 = x n +1 = S, x 2 K x m ∈ C, m ∈ M A , i = 1K n} F = {( x1 , m j , x j +1 ) | x1 = S, x 2 K x p ∈ C, m ∈ M F , j = 1K n − 1} HA = {( h2 i −1 , m i , h2 i ) | parent ( h2 i ) = parent ( h2 i +1 ), h ∈ H P , m ∈ M A , i = 1K m} HF = ( h1 , m j , h j +1 ) | h1 ∈ T , hn ∈ X , h ∈ HC , m ∈ M F , j = 1...n − 1

A

{

}

n = | X |; p = | M A | CSG Tree:

For S, shown in Figure 5a. For each c ∈ C, shown in Figure 5b.

Geometric Constraints: ∀(hi , m, h j ) ∈M ∀(hi , m, h j ) ∈M

, collinear(hi , h j ) AH , coplanar ( hi , h j )

FH

Parametric Constraints: (not yet determined) Table 2

Snap-Fit Fastener Template.

The bolted-joint fastener contains two tensile elements the bolt and the nut. For brevity, only the assembly representation part of this template is included. Figure 7 shows the assembly representation for the bolted joint. The bottom portion of the graph shows these two components, as well as the mating relationships between them. There is one mating relationship between the bolt and the nut; the centerline handles of each have a fits relationship. Also included in the graph are mating relationships with compressive elements. The bolted joint joins at least two components. The bottom surface of the bolt head and the top surface of the top compressive element share an against relationship; in addition, the bolt centerline and the centerline of the top component must be collinear; thus, a fits relationship is defined. Similar relationships are defined between the bottom component and the nut. The bottom face

of the bottom component shares an against relationship with the top face of the nut; another fits relationship is defined between the centerlines of the two components. For completeness, relationships between the compressive elements are shown as well. The final part of the template is the fastener-specific geometric constraints. For both the snap-fit and the boltedjoint to be instantiated correctly, the compressive elements must have collinear centerlines. For both types of fasteners, the top face of the top compressive element must be flat and parallel to the bottom face of the fastener in order to mate correctly. In addition, the bolted joint requires that the bottom face of the bottom compressive element be flat and parallel to the top face of the nut. Finally, for the bolted joint, the bottom face of the bolt head and the top face of the nut head must be parallel.

6


External Mating relationships among components

SNAP-FIT TEMPLATE Fits

Against Top Surface of Top Compressive Element

SNAPFIT

Feature Mate

Bottom Surface of Snap head

Figure 6

Centerline of Compressive Element

Feature Mate

Centerline

Partial assembly representation for Snap-Fit.

Top Surface

Bottom Surface Centerline

Top Surface

Centerline

..........

MID COMP 1

Bottom Surface

MID COMPN

Bottom Surface

Top Surface Feature Mate

Feature Mate

Feature Mate

Feature Mate

Centerline TOP COMP

Fits

Against

Against

Fits

BOTTOM COMP

Centerline

Bolted joint template

Top Surface

Bottom Surface

NUT

BOLT

Fits

Fits

Fits Against

Against Feature Mate

Feature Mate

Feature Mate

Centerline

Centerline Feature Mate

Feature Mate Surface

Surface

Template

Mating relationships between features

Tensile elements

Compression elements

Feature Handle

Mating Relationships between components

Figure 7

Assembly Representation for a Bolted-Joint.

7


3.3

Fastener Instantiation Algorithm The instantiation algorithm utilizes the information stored in the fastener template in order to instantiate a specific fastener into an existing assembly model. The input to the algorithm is the set of against mating relationships (component_mates) amongst the compressive elements (components to be fastened) and a set of hole features that the fastener’s tensile element will pass through for each compressive element. In addition, fastener orientation (i.e., which compressive element mates with which end of the fastener) may be specified. Variables are defined in Tables 1 and 2. The algorithm is as follows: Input component_mates = list of against mating relationships among compressive elements. hole_features = list of holes that tensile element will pass through. Step 1. Preparation operations a. From component_mates, generate C by reference to mating components. i. For each feature referenced by an element in component_mates, find the parent component of that feature ii. If the component is not already in C, add it. b. Sort components in C based on against mating relationships. i. For each component in C, check the number of mating relationships from component_mates that reference that component ii. If the number of mates = 1, label component as top or bottom component iii. After top has been identified, add other component referenced by the mating relationship; continue until bottom component is reached. Step 2. CSG Tree Instantiation a. Instantiate CSG tree for tensile component(s). i. Read CSG portion of template; create appropriate geometry ii. Perform necessary boolean / positioning operations on fastener geometry b. Instantiate CSG tree for compressive elements i. For each c ∈ C, read appropriate CSG tree information from template ii. Check instantiation conditions; i.e., check if geometric elements and handles already exist in the assembly model iii. Instantiate geometric elements and handles not already present. Step 3. Assembly Representation Instantiation a. Instantiate | T | against mates and add to MA. b.

Instantiate MA mating relationships by completing MA and MHA. Select one tensile component (planar) handle for use with the first against mate, then the other tensile component handle is used with the pth against mate.

c. Instantiate p - 1 fits mates for the compressive components and any additional fits mates for fastener (tensile) components. d. Instantiate MF mating relationships by completing MF and MHF. Step 4. Geometric Constraints Instantiation Check collinearity and coplanarity constraints. If any fail, return to 3b and reverse the selection of handles. Step 5. Geometric Parameter Instantiation Set relevant parameter values. As an example, consider the assembly consisting of three components attached by a snap-fit as shown in Figure 8a. Two against relationships exist between the three components; one more is created during snap-fit instantiation. Upon fastener instantiation, component sorting begins. Assume that the components are ordered as . Here, a choice must be made whether to perform the CSG union with the snap feature base and component A or C. Assume the former. Then, MA = {(A, ma1, B), (B, m a2, C), (C, m a3, A)} MHA can be expressed similarly. Two fits relations are now created, one for each compressive component. Upon instantiation into the assembly, the fits mating relations are expressed as: MF = {(A, mf1, B), (A, m f2, C)}Again, for brevity, MHF will not be included here. 3.4 Fastener Substitution Algorithm In order to substitute one type of fastener with another, the fastener instantiation algorithm can be modified to reflect the different input information. Input to the fastener substitution algorithm consists of the fastener to be replaced and the template for the replacing fastener. In this algorithm, the subscripted ‘R’ denotes “to-be-replaced.” The algorithm is as follows: Input fast_replaced = fastener to be replaced, F R. fastener_new = template of fastener to be added. Step 1. Preparation operations a. From FR, generate CR , TR, and X R by following successive fits relations. b. Sort components in XR. c. Determine mating relations MAR, MFR, MAR, MHAR, MFR, MHFR. d. Form C, T, and X from C R , T R , and X R and fastener_new. Step 2 Reverse CSG Operations From FR’s template, get CSG operations For each x ∈ XR, Remove appropriate CSG tree fragments from x. Step 3. CSG Tree Instantiation Execute Step 2 of Fastener Instantiation algorithm.

8


Step 4. Assembly Representation Instantiation Update mating relations MA, MF, MA, MHA, MF, MHF. from MAR, MFR, MAR, MHAR, MFR, MHFR. respectively, to reflect the new handles and any new fasteners. Step 5. Geometric Constraints Instantiation Check collinearity and coplanarity constraints. If any fail, return to Step 4 and reverse the selection of handles. Step 6. Geometric Parameter Instantiation Set relevant parameter values.

C

B

A

a)

Snap Fit Example. NUT

C

B

A

BOLT

b) Bolted Joint Example. Figure 8 Fastener examples Continuing the example from the previous subsection, consider the substitution of the snap-fit fastener with a boltedjoint (bolt-nut combination), as shown in Figure 8b. One choice for the sorted sequence of components is . After reversing the CSG operations from the snap-fit template, the CSG operations from the bolted-joint are executed. The mating relations are similar as compared with the previous example. Starting with Step 4, one against relationship must be created between Bolt and A. The against relationship between C and A is updated to relate C and Nut. Two new fits relationships must be created between Bolt and A and between Bolt and Nut; the existing fits relations can be updated to reference Bolt rather than A. From these operations, the following mating relations result: M A = {(A, m a1, B), (B, m a2, C), (C, m a3, Nut), (Bolt, ma4, A)} M F = {(Bolt, m f1, B), (Bolt, m f2, C), (Bolt, m f3, A), (Bolt, mf4, Nut)} For brevity, MHA and MHF will not be included here.

4.0

APPLICATION TO LIFE CYCLE DESIGN After a product has been modeled in our prototype CAD system, CODA (COnfiguration Design of Assemblies) (Rosen, 1994) using the fastener template, information can be transferred over to a virtual environment which allows the designer to assess the ease of disassembly of a product. A disassembly process is interactively generated; this involves the environment suggesting feasible removal directions for components and the designer specifying the disassembly sequence. Virtual disassembly takes place by using appropriate tools to remove fasteners; once the necessary fasteners have been removed, components can be separated from the assembly. This process continues until the product has been completely disassembled (Siddique and Rosen, 1996). Information from the fastener template described above can be utilized in this process, automating the sequence generation to a large extent. In addition, the instantiation and substitution algorithms allow the designer to immediately see the impact of a change in fastener selection on the disassemblability of a product, thus allowing efficient design for the life cycle. To illustrate the usefulness of the template, consider a simplified model of a Chrysler LHS center console with four components (bin, endcap, armrest and hinge), shown in Figure 9. Two fasteners (fastener 1 and fastener 2) fasten the hinge, endcap and the bin together. Two other fasteners (fastener 3 and fastener 4) attach the hinge and the armrest. The fits and against relationships for the fasteners relative to the other components can be seen in Figure 10; screws have been instantiated for all four fasteners. The centerlines of the screws have a fits condition with the centerlines of the corresponding holes of the components. The bottom surfaces of the screws have against relationships with the top surfaces of the corresponding top components (for example, the bottom surface of screw 4 has an against mating relationship with the top surface of the hinge). A precedence diagram for disassembly of the center console can be seen in Figure 11. This diagram indicates a partial ordering for the disassembly sequence; for instance, screw 3 and screw 4 must be removed prior to the removal of the armrest. By querying the CAD model and utilizing the information contained in the fastener template, this graph can be automatically generated. For this model, the template would include the information that screw 3 attaches the armrest to the hinge. The removal axis for the screw is along the fits centerline that defines the mate; directionality is toward the first component in the sorted list of components from the fastener template. From this information, the environment can determine the appropriate type of tool and check for clearance and removal feasibility. This information can then be used to (1) generate acyclic graphs representing the disassembly process and (2) generate disassembly tables used for disassemblability evaluation of the product. This method of disassembly process generation, facilitated by the fastener template, provides a useful tool for fastener selection during the configuration design stage. The fastener template and the instantiation and substitution algorithms presented in this paper provide a means for quickly modifying the product model and the process model of the design.

9


Decreasing precedence

Disassemble Screw1

y

Armrest x

z

Fastener 4

Fastener 3

Disassemble Center Console

Disassemble Subassembly

Fastener 2

Hinge

Endcap

Bin

Figure 9

Disassemble Screw2

Center Console assembly with screw fasteners

Fits Fits

Screw4 Screw3

Fits

Screw2 Fits Against Screw1

Against

Against

Figure 10

Fastener mating relationships

Disassemble Screw3

Disassemble Endcap Disassemble Armrest

Disassemble Bin

Figure 11

Fastener1

Disassemble Hinge

Disassemble Screw4

Center console disassembly precedence diagram

One possible disassembly sequence for the center console assembly shown in Figures 9 and 10 can be summarized as: 1. Grab screwdriver and unfasten screw 1 and screw 2. 2. Release screwdriver and remove the two unfastened screws. 3. Separate the subassembly consisting of the hinge and the armrest. 4. Remove the Endcap. 5. Remove the Bin. 6. Grab the screwdriver and unfasten screw 3 and screw 4. 7. Release screwdriver and remove the two unfastened screws. 8. Remove the armrest. 9. Remove the hinge. Once such a sequence has been generated, an assessment of the ease of disassembly of the product can take place, using either tabular or automated assessments (Kroll, et. al., 1994). Once the disassembly process has been specified and evaluated, the effect of changing fastener types can be evaluated. For instance, all the fasteners can be changed to bolted joints; the impact on the disassembly process can then be seen. In this case, changing screwi in the above assembly to bolted jointi can have a significant impact. When an attempt is made to follow the same disassembly sequence as above, a problem is encountered. When removing bolt 1 and bolt 2, the head of the bolt must be grasped with a wrench; the nut must be secured as well. The latter step presents an accessibility problem; the endcap and bin are situated so as to prevent grasping the nut with any sort of tool. Clearly, changing fastener 1 and 2 to bolted joints adversely impacts the ease of disassembly (as well as assembly) and is likely not the best choice for this situation. Similarly, the effect of changing the fasteners to snap fits can be easily explored. Here, the disassembly steps differ somewhat for the screw-fastened assembly. Depending on the integrity of the snap fit, disassembly without any sort of tool could be possible; otherwise, some sort of prying tool would be appropriate for disassembly. In either case, since the fasteners are now not separate components, steps involving their explicit removal can be eliminated. A factor to take into account is that snap fits may break during disassembly, rendering the component they are attached to ill-suited for reuse; this fact is accounted for during the subsequent disassembly process evaluation. One possible solution would be to attach the base

10


of all four snap fits to the hinge, which allows the rest of the components to be reused in their present form. With this configuration, it is possible that the advantages of snap-fits during assembly could warrant their selection by the designer in this case. Again, tabular or automated assessment tools can be used in order to evaluate the design. From this case study it can be seen that the fastener template and the instantiation and replacement algorithms enable disassembly process generation. Information that is available from the template can be used to: • Partially order the disassembly sequence • Determine the required tools for unfastening the fasteners • Determine feasible removal directions for fasteners and, subsequently, the components they mate with • Determine requirements for visibility, accessibility etc. Finally, by facilitating the change of fasteners in a product model, the fastener instantiation and substitution algorithms can allow designers to immediately see the impact of fastener selection on a product. 5.0

CONCLUSIONS The configuration design stage often has profound effects on a product’s life cycle characteristics. A CAD system that supports configuration design would allow designers to assess the impact a product makes on the environment while modification of the relevant portions of the design is still feasible. A CAD implementation of one portion of configuration design, fastener selection, has been explored in this paper. The general fastener template presented allows the explicit integration of fastener data into a CAD model. By using the instantiation algorithm to map a specific fastener template to an assembly, fastener selection is facilitated. In addition, the fastener substitution algorithm allows the impact of a fastener change to be easily assessed. An example using a portion of a Chrysler LHS center console showed the benefits of the implementation to life cycle design activities, in particular, automated disassembly planning. While results from the implementation of the template and associated algorithms are not yet complete, several intermediate conclusions can be drawn: • The fastener template and instantiation algorithms allow the development of several fastener alternatives for a given product model. The template stores information that facilitates automatic assessments such as disassembly planning; thus, the template and algorithm allow the rapid generation of both product and process alternatives. • The fastener template is extensible and allows the inclusion of any fastener within the class of tensile / compressive fasteners. Developing additional fastener templates is straightforward. In addition, the instantiation algorithm will work with any of these specific fastener templates. • The explicit representation of fasteners in a CAD model makes the model more complete and thus open to several different types of life cycle analyses.

REFERENCES Ambler, A. P. and R. J. Popplestone (1975). “Inferring the Positions of Bodies from Specified Spatial Relationships.” Artificial Intelligence Vol. 6, pp. 157 - 174. Bronsvoort, W. F. and Jansen, F. W., 1993, “Feature Modeling and Conversion - Key Concepts to Concurrent Engineering.” Computers in Industry, vol. 12, pp. 61-86. Dixon, J. R., Duffey, M. R., Irani, R., Meunier, K., and Orelup, M., 1988, “A Proposed Taxonomy of Mechanical Design Problems”. ASME Computers in Engineering Conference, 41-46. Hassenzahl, S. L., 1994, Extensions of CAD Representations to Support Configuration Design of Assemblies. Master’s Thesis, Georgia Institute of Technology. Kroll, E., B. Beardsley, A. Parulian and D. Berners (1994). “Evaluating Ease-of-Disassembly for Product Recycling”, ASME Materials and Design Technology 94, New Orleans, Louisiana, (January 1994), vol. 62, pp. 165-172. Lee, K., and Gossard, D.C. 1985. “A Hierarchical Data Structure for Representing Assemblies: Part 1,” ComputerAided Design, Vol. 17, No. 1, pp. 15-19 Lieberman, L.I., and Wesley, M.A. 1977. “AUTOPASS: An Automatic Programming System for Computer Controlled Mechanical Assembly,” IBM Journal of Research and Development, Vol. 21, No. 4, pp. 321-333 Popplestone, R. J., Ambler, A. P., and Bellos, I. M., 1980, “An Interpreter for a Language for Describing Assemblies.” Artificial Intelligence, Vol. 14, pp. 79-107. Rocheleau, D.N. 1992. Assembly Modeling of Spatial Mechanisms Using Geometric Constraints. Ph.D. Dissertation, University of Florida. Rosen, D. W.(1994). CODA - COnfiguration Design of Assemblies: Prototypical CAD System. Internal Report, Georgia Institute of Technology. Rosen, D. W., 1993, “Feature-Based Design: Four Hypotheses for Future CAD Systems.” Research in Engineering Design, 5(3/4), pp. 125-138. Shah, J. J., 1991, Conceptual Development of Form Features and Feature Modelers. Research in Engineering Design, 2(2), pp. 93-108. Shah, J. and Rogers, M. (1993). “Assembly Modeling as an Extension of Feature-Based Design.” Research in Engineering Design 5(3/4), pp. 218-237 Siddique, Z. and Rosen. D. W. (1996). “An Approach to Virtual Prototyping for Product Disassembly.” ASME 1996 Design Engineering Technical Conferences and Computers in Engineering Conference, Irvine, CA, August 18-22, 96DETC/CIE-1345. Wesley, J.P., Lozano-Perez, T., Lieberman, L.I., Lavin, M.A., and Grossman, D.D., 1980. “A Geometric Modeling System for Automated Mechanical Assembly,”, IBM Journal of Research and Development, Vol. 24, No. 1, pp. 64-74 Xiaolin, H. and Shensheng, Z. 1996. “Assembly Model for Concurrent Engineering,” Proceedings of SPIE, Vol. 2644 pp. 665-670

ACKNOWLEDGMENTS We acknowledge the support of the National Science Foundation through grant DMI-9420405 and DMI-9414715. We also thank the School of Mechanical Engineering at Georgia Tech for their partial support of Zahed Siddique.

11
