International Journal of CAD/CAM Vol. 2, No. 1, pp.

29~37 (2002)

Virtual Disassembly
Jianzhong Mo1*, Qiong Zhang2 and Rajit Gadh3 1 eMedia Center, University of Wisconsin - Madison, 1500 Engineering Dr, Madison, WI 53706, USA 2 State Key Lab of CAD&CG, Zhejiang University, China 3 University of California - Los Angeles, Los Angeles, USA
Abstract − De-manufacturing is an entire process of collecting, disassembling, reusing, refurbishing, recycling, and/or disposing products that are obsolete or un-repairable. Designing the products for inexpensive and efficient disassembly enhances the ease of de-manufacturing. Virtual disassembly addresses the difficulty and the methods to disassemble a product in design stage rather than really disassemble a product at the end of its life cycle. Based on the virtual disassembly analysis results, design will be improved for better assembling/disassembling. This paper presents a systematic virtual disassembly methodology such as disassembly relation modeling, path/sequence automatic generation and evaluation. This paper also presents a new virtual disassembly interface paradigm via virtual reality technology for disassembly simulation in virtual environment. Keywords: Virtual Reality, Virtual Disassembly

1. Introduction
Currently, products are acquired with no clear plan of disposal of the product. Often, consumers are left with no other choice at the end of a product’s life but to dispose it off by throwing it in to the trash. The demanufacturing addresses the question of systematically designing for easier product dismantling and disposal. The two stages of de-manufacturing are (1) disassembling the assembly into its individual components (requiring disassembly analysis), and (2) recycling the individual materials that constitute each component (requiring component material recycling assessment). We developed a virtual-reality-based software tool- Motive3D - that supports collaborative de-manufacturing (disassembly, service, recycling and disposal) between manufacturer/ de-manufacturer/disposer and designer. This paper will address the methodology of virtual disassembly, a core part of Motive3D. Virtual disassembly addresses the difficulty and the methods to disassemble a product in design stage rather than really disassemble a product at the end of its life cycle. Our motivation of research on virtual disassembly is to complement traditional tools to bring de-manufacturing trade-off analysis into the design process, allowing engineers to adopt a “predict and prevent” approach. The rest of this paper is organized as following: Section 2 reviews the related research work; Section 3 presents virtual disassembly software environment Motive3D; Section 4 presents virtual disassembly
*Corresponding author: E-mail :

hardware environment; Section 5 is a conclusion.

2. Related Research
The disassembly evaluation addresses the problem of estimating the cost, time, or design effectiveness of disassembly plans. This evaluation can be served to determine the product design for disassembly and demanufacturing. Boothroyd and Alting [1], Jovane et al. [2], Gupta and McLean [3], Penev and Ron [4] have reviewed Design For Assembly (DFA) methods and discussed the research trends in Design For Disassembly (DFD). Subramani and Dewhurst [5] introduced time standard charts to make disassembly evaluation. Hrinyak et al. [6] examined the existing disassembly software. Bras and Emblemsvag [7] evaluated the cost incurrence of different design based on Activity-Based-Costing (ABC). Kroll [8] developed a rating method based on the difficulty index of every disassembly task. Suga et al. [9] introduced energy and entropy for disassembly. Meanwhile, several applications specific to recycling/ maintenance approaches have been developed. Kirby and Wadehra [10] suggested the Design For Environment (DFE) factor should be considered into design. Johnson and Wang [11] evaluated the recyclability and material recovery of a product. Zussamnn et al. [12], Ishii [13], Geiger and Zussmann [14] addressed end-of-life approach for DFD/demanufacturing. Mo et al. [15] investigated DFA-oriented assembly relation modeling. Virtual Reality (VR) is a new technology that creates a real-time visual/audio/haptic experience with computer systems including hardware and software. It provides a potential way for disassembly simulation. Siddique and

Mo et al. APOINT et al. COEDGE. Step (4) is implemented as Motive3D client. 1. integrating VR device for disassembly simulation. Automatic path/sequence generation and manual path/ sequence generation with VR device are complementary to each other. LOOP. that it. [19] investigated selective disassembly based on Conceptual Virtual Design System (COVIRDS). COVIRDS is a VR system coupled with 3D hand tracking.1. and/or users can edit them. [18] developed system called Virtual Environment for Design for Assembly (VEDA) for 2D models. [17] developed a prototype system called Virtual Assembly Design Environment (VADE). [20]. pp. This paper presents a methodology to fill in this gap by virtual disassembly based on 3D CAD models. Step (1). EDGE. The simplified B-rep of a component is shown in Fig. a component can be represented by BOOD. 29~37 Rosen [16] investigated disassembling in a virtual environment. LUMP. automatic generation of path and sequence. Virtual Disassembly Analyzer (VDA) VDA takes CAD models as inputs. Virtual Disassembly Software Environment . FACE. Its Virtual Environment (VE) will be discussed in Section 4. esp. we will focus on VDA. What the module of Product Preprocess will do is rebuilding assembly relation. Then VDA will generate disassembly sequence and path. VDA will generate assembly relation information based on the CAD models.Motive3D The three steps to achieving the goals of disassemblyoriented design are (1) investigate algorithms of assembly /disassembly path/sequence generation and evaluation. visualizing disassembly sequence and path. (2) and (3) are implemented as Motive3D server. 2. The architecture of Motive3D sever is shown in Fig. 3. and stereoscopic visualization which provides a high fidelity visualization and easy-to-use interface. Jayaram et al. Gupta et al. However current researches rarely address both problems together. SHELL. and SURFACE. and disassembly cost evaluation. Motive3D is developed according to above three steps. to deploy assembly/disassembly information on to the network and implement the software tool of Data Deployment Tool (DDT).1. VDA and VE. [21] investigated Internet-based virtual assembly /disassembly for e-manufacturing. and implement the software tool of Virtual Disassembly Analyzer (VDA). 3. In this section. 1. .1. Product Preprocess Motive3D accepts CAD models in B-rep. VDA will also complete disassembly evaluation. Srinivasan et al. DDT and Motive3D client will be discussed very briefly. Data Poster. 1. (2) investigate effective virtual environment to simulate disassembly process using virtual reality technology (VE). optimize them. and implement a 3D visualizer. investigate interactive 3D visualization. A very brief introduction of DDT and Motive3D Client will also be presented at the end of this section for system integrity. (3) investigate fast approaches Fig. International Journal of CAD/CAM Vol. and VERTEX as its topologic data structure. Motive3D server: VDA. VE and DDT. This paper will focus on Motive3D server. No. and Data Collector. CURVE. (4) Design data representation. as its geometric definitions. 2 (ACISTM type). voice command. calculating clearance 3. presented by Chu et al.

Contacting constraints can be classified into two categories: mating constraints and fitting constraints. (2) Disassembly directionality Two other major tasks of Product Preprocess of VDA are disassembly directionality determination and interference checking among mating parts. Cj {Ca} contacting with C. If nv1 parallel to -nv2. vj). If DDi denotes the DD of component Ci. the centerline of sf2 is c2 STRAIGHT. Disassembly Directionality (DD) of a component C is a geometry entity containing direction vectors in which the component C can be disassembled from components {Ca}.Jianzhong Mo. between components and performing interference checking between components. f1 f2 φ . nv1 is the normal vector of sf1 . Fitting constraint: CONE type geometry sf1 of f1 FACE b1 BODY. nv1(ui. represents Fig. With respect to other types of surfaces. 2. m. and f1 f2 φ . n]. All of those tasks are finished based on B-rep CAD models. See Fig. j) is the DD based on mating constraints. et al. i [0. For an example. (1) Assembly relation establishment There is one assembly relation. should be established for automatic sequence generation. . nv1(ui. Mating constraint: PLANE type geometry sf1 of f1 FACE b1 BODY. vj ) is the normal vector of sf1 at knot (ui . vj). If c1 parallel to c2. DDik . 4. DDij = DDm(i. The assembly relation and contacting constraints should be established for automatic sequence generation. f2 ). n]. j ) DDf(i. nv2 is the normal vector of sf2 . k is the element number of {Ca}. DDm(i. the centerline of sf1 is c1 STRAIGHT. 3(a). f2 ). j [0. then b1 and b2 has mating constraints mating( f1 . 3. n are the dimension of rectangular arrays of control points of sf1 and sf2 . and PLANE type geometry sf2 of f2 FACE b2 BODY. the disassembly directionality of a component can be calculated. See Fig. special checking method should be used to establish contacting constraints. If for every i [0. and CONE type geometry sf2 of f2 FACE b2 BODY.. then DDi =DDi1 DDi2 . The paper represents the product to be disassembled with symbol A.                                             Fig. and DDij denotes DD of Ci with respect to Cj . Based on contacting constraints. the disassembly directionality of a component can be calculated. Contacting constraints (a) mating (b) fitting. DDf(i. 3(b). Virtual Disassembly  Fig. vj) . intersection. v) of f2 FACE b2 BODY.. nv2(ui. Based on contacting constraints. f1 f2 φ. m]. f2). represents intersection. m]. v) of f1 FACE b1 BODY. vj) parallel to -nv2(ui. then b1 and b2 has fitting constraints fitting( f1. represents intersection. then b1 and b2 has mating constraints mating( f1 . and SPLINE type geometry sf2 (u. Component’s CAD model in B-rep. contacting constraints. DD generation from contacting constraints. j [0. j) .j) is the DD based on fitting constraints. the mating definition of two B-spline surfaces will be: SPLINE type geometry sf1 (u. vj) is the normal vector of sf2 at knot (ui .

Interference checking. Fig. Two steps are conducted to improve performance: (1) interference checking of parts’ bounding boxes.1. (2) if parts’ bounding boxes collide each other. 5(d)] STRAIGHT [a radium. If the DD is PLANE. DD1 in Fig. 5(b)] LUMP [a sphere cone. the generation of DDij based on contacting constraints is shown in Fig. 4. 5(c)]  represents intersection. DD1 in Fig. 5(a)] PLANE [a fan. 3. 2. the component can not be disassembled. DD1 in Fig. If the DD is STRAIGHT. 6. 1. Ci Cj .2) if the intersection graph is not null. i j If the DD is LUMP. Step (2) has two sub-steps: (2. p2 is the other end of this straight line.  . 6. 29~37 Fig. DDi can be any of geometric entity on a unite Gaussian sphere : a) b) c) d) APOINT [a center point. do interference checking based on native geometry model (ACIS or Parasolid). C .2. C  A. p1 is the center of the unite Gaussian sphere. (3) Interference checking Interference checking is another important task executed in the module of Product Preprocess. indicates that component Ci is adjacent to The disassembly direction vectors dv is determined by following equation : dv=p2-p1. pp. p2 is any point on the arc of this plane. 5. (2. p2 : If the DD is APOINT. No. Disassembly directionality.1) firstly check the intersection graph between two BODYs (components). p2 is equal to p1. According to the definition. International Journal of CAD/CAM Vol. Propagation Waves and Automatically Sequence Generation Accessibility Graph (AG) is a directed graph representing product A’s assembly relation in which nodes correspond to the components of A. p2 is any point on the sphere surface of this LUMP. DD1 in Fig. and an arc. take intersection Boolean operation to determine the interfering volumes. An example of collision checking is shown in Fig.

zmin) s1 . Ci . …. Cj A. Then interference checking result Rp for path P of Ci can be re-calculated by Rp =Bsi k= 1. interference checking can be accomplished by collision checking of BODYs of component models. ymin . i j .. (xmax . and vector (xmin . zmax)s1 . q remains same for those two path states. zmax ) sj and (xmin . The interference checking result Rp for path P of Ci is calculated by Rp =BBei k=1. In VE. i. ymax . BBei is the extended bounding box of component Ci disassembled along path P=< s1. n is the number of paths. paths are generated through hand motions: translation. P2 .. twist. et al. that is. zmin) sj . and q is the orientation element of s represented in quaternion. βj jth wavefront of a β waves. If n<m. Px Py Pr with locally minimum component removals for C s ={Cx}. the path P is valid. zmin ) s2 . then S is called parallel sequence. . Virtual Disassembly  Fig.. j<=n. m is the component number of A.3. and free motion. where Pi is the disassembly path of Ci A.Jianzhong Mo. ymin . This intersection determines a sequence Sx ={Px. component Cj and stores an attribute DDij .. BBsij is also defined by two vectors: (xmax . BBk is the bounding box of component Ck . Disassembly directionality. and Pk denotes the shortest disassembly path for Ck. Pr}. There are two steps: (1) extended bounding boxes interference checking. p is the position element of s. 2. . For details about propagation waves. ymax .. (xmax . an intersections event occurs at Cy. 7(b) is an example. A disassembly wave is defined to topologically arrange components in A to denote the disassembly order. Pj is the disassembly path of Cj A. j <= n. ymax . zmax) s2 . In automatic path generation. Disassembly state s=[p. s2.. sn >.1.. In ACIS. and Pi and Pi+1 have the relation represents the relation of Pi Pi+1 or Pi ~Pi+1. This step involves low level interference checking based on native geometry in ACIS or Parasolid. ymin . refer to our previous paper [19]. Path and Sequence Evaluation Path evaluation involves the interference checking of one part with others along its disassembly path. Then (xmax .. If S has at least one relation of indifference (~). k  i Bk.   . {Cr } denotes removable component set of A. 3. If DDij is not APOINT. the arcs are corresponding to RG. zmax) sn ]. There are two types of disassembly waves:        a disassembly sequence with minimal components removal to remove selected component set Cs is determined. BBsij is the bounding box of component Ci at path state sj. zmin)i =min[(xmin . Removalability Graph (RG) is a directed graph where nodes correspond to the components of A. where of precedence. n is the number of path states. screw motion. ymin . where C s ={Cx }. Cj A. Let Bsi is the BODY of Ci sweeping along P=< s1 . i.        (1) τ waves propagating outwards from the component set C s. q]. ymin . Ci Cj. accessibility graph. where si is the disassembly state of component Ci A and n is the number of states. then S is called selective sequence. Otherwise do step (2) sweep volume interference checking. one path contains only two states and only translation movement is used for each state. Bk is the BODY of Ck . removalability. (xmin .. Py . BBei is defined by two vectors: vector (xmax . . zmin )i representing Bei’s minimum coordinates. indicates that Rij =TRUE. s2. …. and an arc. si sj . ….. Fig. Pj . . ymax . s2 .. rotation. Ci . sn >. ymax . zmax)i =max[(xmax . …. Disassembly waves (a) conception (b) sequence determination. Rij is determined by DDij .. else Rij = FALSE (un-removable). Cj A. n is the number of path states. zmax )i representing BBei’s maximum coordinates. Pi . Rij =TRUE (removable). Ci. Pn}. If Rp = φ . and removalability graph are the information on which propagation wave will be generated. where τi intersect βj (j i). zmin)sn ]. In this figure. 7. ymax . and represents the relation of indifference.. . sn >. Path (P) is an ordered list of disassembly states: P= < s1 . Removalability (Rij ) of a component Ci A is a Boolean value indicating if it can be removed from its adjacent components Cj . 2. otherwise S is called complete sequence. otherwise S is called sequential sequence. Let τi denotes the ith wavefront of a τ waves. k  i BBk. (xmin . m. m is the component number of A. ymin . …. (2) b waves propagating inwards from the boundary of A. Based on the intersection event between and β waves. m. Sequence (S) is a list of paths: S={P1. Fig. 7(a) shows the concept of disassembly waves. τi and βj are both represented by component set. and (xmin .

standard time for accomplishing a particular task). DDT will generate an HTML page file. m is the component number of A. Multisensory input device are coupled with special nodes of scene graph. DDT and VE together are also called A3D.4. The generated SD sequences are then simulated in A3D system. its disassembling path (such as distance. The first process of sequence evaluation is finding out a sequence from a sequence set which has maximum NDB. 3. and NDBi denotes the net de-manufacturing benefit of Ci. then NDBi =DBi-DCi . 9 shows an aero engine assembly without 3D menus and toolbars. de-manufacturing benefits. NDBi NDBj . and (3) sequences S are determined using the WP abstraction. The Net De-manufacturing Benefit (NDB) of the component is the difference of its de-manufacturing benefit and disassembly cost. position. 3. CAD models in ACIS/Parasolid format are triangulated and displayed within WroldToolKit rendering environment. (Assembly/disassembly in 3D). After that. and Data Collector make up Motive3D’s client side. DBi denotes the de-manufacturing benefit of Ci . Let S={P0 . and Data Collector They are the client side applications. Data Deployment Tool(DDT) DDT takes disassembly sequence. 8. Technology Used in Implementation VDA and DDT make up Motive3D’s server side. disassembly tools are not considered as a constraint. If DBi <DBj or DCi >DCj. ACIS / Parasolid. pp. If there is no applicable sequence available. Data Poster. 1.3. fixture cost. 3D Visualzier will visually display assembly/disassembly path and sequence interactively. state numbers) and standard variables (such as labor rate. Disassembly sequence for . OpenGL. Otherwise P is not valid. 29~37 Poster when the page is viewed by users via web browser. Disassembly costs vs. All the output of DDT will be saved at HTTP server side. Currently. IGES to ACIS translator and UG to Parasolid translator are also embedded into A3D. P1 . rotated angles. No. It’s our future research work. as shown in Fig.  3. De-manufacturing requires that maximum NDB will be gained at the earliest stage of disassembling process. Data Poster. Product assembly structure is translated into scene graph for 3D operations. a design recommendation is thrown out. In an A3D system: (1) user inputs CS via a menu-interface. If Rp = φ . 3D Visualizer. Then DDT organizes the sequence and path data into a special file together with meshed CAD geometric information. that is. Fig. Data Collector is used to collect the feedback from the users. Java3D. Disassembly Cost (DC) of one component can be determined based on attributes of the component (such as its weight. i<j<n. et al). a design recommendation is thrown out. Server side is implemented in C/C++. orientation.2. De-manufacturing Benefit (DB) of one component can be determined based on its reuse value (shape reuse and part reuse) and recycling value (material reuse). (2) Disassemblability and Removal Influence are determined from AG. and WorldToolKit. 2. the path P is valid.. VDA. The second process is the evaluation of paths of this sequence. i<j<n. Fig. 3D Visualizer. volume. DCi denotes the disassembly cost of component Ci A. which will launch 3D Visualizer and Data Fig. 8. This requirement will be satisfied if DBi DBj and DCi DCj . path and evaluation as inputs.. and HTML. Data Poster will post disassembly instructions and de-manufacturing cost/benefits.. International Journal of CAD/CAM Vol. . VDA and DDT are supported by Virtual Environment (VE). Pn} denotes the disassembly sequence of an assembly. If there is invalid path in a path. total NDB turns into minus. a design recommendation will be thrown out to CAD. Fig. 9. Client side is implemented in Java. 8 also illustrates that total NDB will reach the maximum at the disassembly stage of Pk for Ck .

which enables large-scale motion. VR enabled A3D provides high fidelity visualization and an easy-to-use interface for sequence/path generation and visualization. Left hand wears 5th Data glove. 11(a). shown in Fig. virtual desktop. 11. this section will present the method to record paths by hand gestures.1. element of (θ. y. and voice device. Those three computers connected by gigabit switch to support Distributed Single Logical Screen (SLS/D). and its gesture is coupled to the selected target. To select the target component for path recording. and if left hand rotates. 10.2. Virtual Reality Devices Virtual Environment is a platform that provides 2D. Virtual Disassembly Environment 4. The input interfaces are for 3D mouse. The following section will discuss this issue. tracking device.O. When left hand grasps. 11(b). If left hand translates. it means standby. element of (x. The display system is called grv-120 provided by Panoram Technologies. Integrator version 3 is used for active matrix Stereographic glasses and emitters control. Three projectors display 3D models from Visualize Center on to the curved screen. φ. . such as devices for 3D Hand tracking. z. and note taking. voice command and stereoscopic visual display. When left hand grasps. et al. The integrated system is called HP Visualize Center. a one large. So the target will move as the left hand moves. 10. φ. The outputs are 3D sound. high-resolution image can be displayed with lights on for work group sessions. The system supports a 6 channels surround sound audio system. The layout of Visualization Laboratory is shown in Fig. grabbing device. Gestures and locator (a) left hand (b) right hand. Rendering and display system. 3D and voice interface. stereo display. an Anir Mouse is used (right hand).0 and supports OpenGL application transparently. z) of HG will be changed. three Ascension FOB 6 D. It fits into a 16' 6" X 16' 6" X 10' space. as shown in Fig. y.Jianzhong Mo. it is ready to record HG. project visualization system for Group VR applications. The rendering system is running on hp-ux 11. θ. information sharing 4. Fig. As an example of using Virtual Reality in Virtual Disassembly. theater-style computing center. The bright. Virtual Disassembly  A3D also supports Virtual-Reality (VR) devices. motion trackers and a 5th Data glove are used (left hand). Computing and Display Hardware eMedia Center Visualization Laboratory is a unique. Both hands are needed to record paths.F. To track the position and orientation of hand gesture {HG: (x. ϕ ) of HG will Fig. The GVR-120 is a curved screen immersive. ϕ )}. 4. Three HP UNIX J5600 computers with dual PA-RISC 8600 CPUs and fx10 graphics card per display makes up the 3D rendering system. which enables users to configure their workstations to utilize multiple add-on graphics cards.

Journal of Design and Manufacturing. 117. One recording method is manually clicking a mouse button by right hand to record current HG. improve ergonomics and carry out disassembling the products virtually. 238-250. be changed. there are 3D toolbar and 3D menu in Motive3D’s Virtual Environment. [7] B. “The use of activity based costing. Hrinyak. [10] J. “Planning product disassembly for material recovery”. Feldmann. Kirby and I. CIRP Annals. Gupta. In VR environment of stereo. Subramani and P. Connacher. CIRP Annals. 42(2). earlier in the development cycle. 33(11). Int. “A key issue in product life cycle: Disassembly”. Kriwet. 19-24. and W. et al. M. K. Ron (1996). 49-52. Zussmann. 2. [3] S. a virtual hand is modeled. 1. Hoffmann (1996). [13] K. [16] Z. as well as part selection. 1. [2] F. F. Dewhurst (1994). Fig. to visualize their full-system design ideas on the computer. 42(2). CIRP Annals. H. Geiger and E. pp. Disassembling an engine (stereo interface). Siddique and D. References [1] G. [6] M. The future work will integrate disassembly tools into system as constraints for path/sequence generation/evaluation. I. Computer-Aided Design. uncertainty and disassembly action charts in demanufacture cost assemblies”. “Enhancing design for disassembly: a benchmark of DFD software tools”. It allows engineers. Computers and Industrial Engineering. [14] D. “Repair time estimation for early stages of product development”. R. Bras and J. ASME Design Engineering Technical Conference. IEEE International Symposium on Electronics and the Environment. J. 32-36. 285292. An alternative way is checking interference after recording a path. G. Saneshige. Lyons (1997). 4(2). 29(12). and design feedback. [5] A. such recording gestures every one frame. “Determination of a disassembly strategy”. 43(1). Jovane. “Probablistic reactive disassembly planning”. ASME Journal of Mechanical Design. Mo. C. R. Fig. realistic and efficient interface interaction that enables disassembly simulation in 3D. 225-228. and make these models disassembled. The virtual hand is attached to 3D mouse and moves as user’s hand moves. voice and pinch gloves. 12(3). Kroll (1996). J. Corresponding to regular window applications’ 2D toolbars and 2D menus. 29~37 to any commercial CAD systems. 96-DETC/DFM-1271. CIRP Annals. 96-DETC/DTM-1509. 495-506. 651-658. R. 12 shows Motive3D’s 3D interface in Virtual Environment. K. Zussmann (1996). 45(1). McLean (1996). “Quantitative disassembly evaluation”. [8] E. as shown in Fig. “DFA-oriented assembly relation modeling”. Another way is recording gestures by definite frame. and Z. Ishii (1995). J. Roth (1993). Jayaram. This paper also presents a new interface paradigm with multi-sensory input/ output providing an effective. just as they would in the real world. J. Wang (1995). Boothroyd and L. Wadehra (1993). Emblemsvag (1995). 5. L. Johnson and M. 42-47. DE-vol 82. and J. [17] S. with realistic 3D models. Seliger (1994). Lu (2000). [12] E. [9] T. Proceedings of the ASME DET and CIE Conference. 625-636. International Journal of Production Research. that is. two frames. A. 11(a). path/ sequence automatic generation and evaluation independent . and W. “Design for assembly and disassembly”. real time collision detection is introduced. “Development of a disassembly evaluation tool”. which will prevent moving the target further at particular direction. 129-137. 9-14. 3119-3142. and C. 847-860. thus to improve product design for easy disassembly. No. Cai. 1993 IEEE International Symposium on Electronics and the Environment. users can manipulate the parts. These 3D toolbars and menus ease the command selection and activation in stereo mode. D. Bras. 31(1-2). Zhang. 34(2). Conclusion Virtual disassembly provides a good solution to simulate the disassembly process bottlenecks and evaluate operational sequences. Alting. Usually real time collision detection is time-consuming. Rosen (1997). and G. Armillotta. W. editing and visualization. Z. Suga. A. [4] K. “Disassembly of products”. To visualize 3D toolbar and menu selection. This paper presents a systematic methodology for disassembly relation modeling. International Journal of Production Research. 12. Eversheim. VR enabled A3D provides high fidelity visualization and an easy-to-use interface for sequence/path generation. Computer Integrated Manufacturing. “Designing business machines for disassembly and recycling”. W. “Automated Product Disassembly Reasoning”. To prevent generating interfering paths. B. Fujimoto (1996). “Life-Cycle engineering design”. “Disassembly Oriented Assessment Methodology to Support Design for Recycling”. generate/edit/simulate disassembly path/ sequence manually within a virtual environment. Penev and A. Seliger and N. Alting (1992). [11] M. Proceedings of the ASME DET and CIE Conference. [15] J. International Journal of CAD/CAM Vol.

[19] H. [18] S. 15. 225-228. China. Figueroa. 231-245. and R. T. M. R. 29(8). Robotics and Computer Integrated Virtual Disassembly  Manufacturing. Q. “Virtual assembly using virtual reality techniques”. Computer-Aided Design. “Selective disassembly for virtual prototyping as applied to demanufacturing”. in 1994. Computers and Industrial Engineering. Gupta and C. Gadh (1999). “Multisensory Interface for a Virtual Reality Based CAD System”. et al. 493-499. August 22-24. and R.   received his BS. He received his PhD from Shanghai Jiao Tong University in 1998. 31(12). McLean (1996). [21] J. “Internet-based collaborative virtual assembly/disassembly for e-manufacturing”. 709-725. 575-584. .-C. [20] C. He has been a Research Associate at University of Wisconsin-Madison. USA and a Researcher at Philips Research East Asia. 29(10). China. USA. Computer Aided Design. MS and PhD all in Computer Science from Zhejiang University.   has been a Research Associate at eMedia Center of University of Wisconsin-Madison. R. Gadh (1998). and Data-Centered Visualization.Jianzhong Mo. Virtual Reality in Engineering. His research interests include CAD/CAM. Mo. Dani. Kunming. Srinivasan. H. CAD/Graphics’2001. Zhang. and R. 2001. pp. “Disassembly of products”. He is a member of ASME and IEEE. Chu. Gadh. 1996 and 1999 respectively.

He received his PhD from Carnegie Mellon University in 1991. USA. .   is a Professor at the Henry Samueli School of Engineering and Applied Science at University of CaliforniaLos Angeles. Mobile Field Service. Virtual Reality for Design. CAD. His research interests include Web-based Product Design. and Wireless Internet for Collaboration.