A SYSTEM LIFE-CYCLE MODEL FOR DISASSEMBLY-ASSEMBLY LINE DESIGN

Copyright © 2002 IFAC 15th Triennial World Congress, Barcelona, Spain

www.elsevier.com/locate/ifac

A SYSTEM LIFE-CYCLE MODEL FOR DISASSEMBLY-ASSEMBLY LINE DESIGN Christian Mascle Section fabrication, Département de Génie Mécanique, École Polytechnique de Montréal, C.P. 6079, Succ. Centre-Ville, Montréal, Québec, H3C 3A7 Canada e-mail: [email protected] Abstract: In this paper, algorithms are presented in order to determine the best disassembly sequence for a sub-assembly and a system life-cycle model for disassembly-assembly line design. The main contribution consists in the determination of a sequence which has the minimal number of components to disassemble, taken into account several criteria: stability, the geometrical form, operations done in parallel, and the irreversible operations. The integration of assembly and disassembly is advantageous for the activation of higher value as per assembly, repair, maintenance and reuse of parts, components and products. Copyright © 2002 IFAC Keywords: Assembly, System Design, Computer-aided manufacturing, Adaptive systems, CAD/CAM Models, Reversibility, algorithms, Database Structures. 1.INTRODUCTION

the automatic manufacturing and assembly process, the processes during product use and de-production are mainly of a manual nature, although for repairs and maintenance. Tomiyama and Umeda (1997) introduced the expressions “intelligent systems of fabrication”, which will be considered probably as the new generation of fabrication systems. This article begins with a list of modern difficulties caused by technological advances, which inevitably, meet natural, social and human constraints.

In the future, de-production and repairs or maintenance could differ in two different ways. In contrast to maintenance, disassembly for the purposes of recycling could also take place with an industrial structure. Larger batch sizes and quantities could be achieved for the use of more highly automated disassembly systems. The determination of the best disassembly sequence for a sub-assembly has considerable implications for the points of view assembly, maintenance, recycling and re-use. Another idea is to suppose the respect of the same technological line for the fabrication of a product, and to use the disassembly methods already known and which respects the current standards. Constraints, which could appear in the disassembly process, are the existence of components with toxic and dangerous substances, liquid or with components with corrosion of deformation, etc. All these constraints can make heavy or in some cases, stop the disassembly realisation.

This study shows that our definition of assembly features (AF) see (Jabbour, et al., 1996 and 1998; Mascle, 1999), the starting point of any targeted objective, was consistent for maintenance and reuse of parts, components and products. The most interesting contribution is to use any data (geometric, technological, cost, etc.) in a feature-based model at each disassembly or assembly state. Any time it is possible to assemble or to disassemble a product and to know if the operation is reversible or not. The review examined various assembly and disassembly modelling and method studies, but also the technical potential given to industrialise the last phase of products life. Our purpose is to be able to work in a synthetic manufacturing environment to enhance all levels of decision and control. This process uses product, process and resource models to evaluate the productivity of new product concepts prior to

Westkämper, et al (1999) show objectives, requirements and first solutions for integrated assembly and disassembly. They underline that higher added value can be reached by industrialisation of disassembly for re-manufacturing, not only for recycling the materials. In contrast to

103

disassembly” which represents only the disassembly of a sub-assembly or a component, which will be repaired or replaced. So, they present a case of a defective aircraft and explain that in the case of maintenance of a sub-assembly of an aircraft, it is sufficient to disassemble only the respective subassembly. So the idea is the determination of the best disassembly for a sub-assembly. Authors introduce a new term of “wave propagation” which represents a disassembly analysis. It determinates the minimum number of components which are disassembled and an assurance of a necessary geometric accessibility for assuring the disassembly (for the space where the tool is introduced, the apparatus or the robot arm which help us to realise only the disassembly, only what is necessary). A good conception lies in the complexity reduction in geometric selective disassembly, using the wave propagation abstraction is important concerning the disassembly. This last one should be as easy as possible, not only concerning the recycling, but from the maintenance point of view too, when it is important to find the most efficient selective disassembly.

mock-ups. Total companies commitment and to the solid savings others final models product ofplace 3% andto design. the on 6%use records ofAerospace of lifevirtual cycle the costs are expected (AGARD, 1998). Virtual Manufacturing has demonstrated tremendous benefits related to communication between product development teams and first time fit for quality. Products are designed by using solid models, with certain problems to cote and to tolerance parts or components. These same solid models are used to develop and validate manufacturing and assembly concepts prior to fabrication to hardware. Virtual manufacturing as simulation of part assembly or fabrication processes, cost relationships and virtual reality training require to use an adapted features model. Technology can be seen as the enabler in this case because the tools allow multi-user concurrent access to data. An extraordinary number of technical innovations have recently been made in the field of industrial process planning. Re-using requires the separation of parts, subsets or subassembly, but contrary to maintenance, reversible liaisons and their components attached can be broken. Designing parts, subassembly and liaisons in assembly for selective disassembly allows profitable recycling, thereby reducing the burden placed on the environment because of the ever-increasing number of obsolete products. Recycling requires the separation of parts of different materials from the product. Designing components of different materials in assembly for selective disassembly is less important. The proposed approach can only be applied with the following assumptions: ¾ initially the relative positions of parts or components are known and coincide with the assembled position of a normal working configuration, ¾ a part will be removable if and only if at least one of its finite translation mobility exists without any geometric interference (intersection with the remaining parts or components or its environment) and process is well-known. ¾ the geometric model of a functional sub-face representing the contacts is described using an exact Boundary Representation (B-Rep) model.

In another paper, see (Srinivasan, et al., 1999b), the separation importance of some components from an assembly is presented. This method applies with success for the maintenance and the recycling, as for an assembly. The method “selective disassembly” has tree steps: Identifying the components to be selectively disassembled with the help of a software program or a designer; Determining an optimal disassembly sequence possible for the selected components, for an easy realisation with a minimum number of components and with a minimal cost. So, after the study of both aspects, we must do an optimal re-identification of the components, which will be disassembled; Performing disassembly and making with accuracy; the disassembly order is observed. 2.1 Symbols and definitions Demand requirement – determinate automatically the optimal sequence (SO) for the studied component C

X

with the wave propagation (WP) in the way that it satisfies the functional objective. Wave propagation is applicable for the components where n >1.

The process of volume generation does not take into consideration the faces that are at the origin of the existence of a sub-face characterised by an assembly process that contributes to a deformation (such as crimping, clipsage, etc.). This same rule applies equally to each face declared to undergo a processing equivalent to an assembly process of the same type as those mentioned above.

Terms used: Ass - sub-assembly Ci, Cj - components of sub-assembly Ass ∆ - disassemblability IOCiCj - irreversible operation (i ≠ j) t - time τi - the wave - the number of components for disassembly nS for one of the disassembly sequence W - components on all the length of front of the wave

2.DISASSEMBLY USING THE CONCEPT OF THE WAVE PROPAGATION. Srinivasan H. and Gadh R. (1998 and 1999a) present the influence of the geometrical form on the disassembly. They use the same term “selective

The main attributes, which suggest the wave propagation for determining the optimal disassembly sequence, are:

104

For t ≥ 1 ; WP of Ci ∈τt-1 at Cj ∈ MACi . The wave propagation is possible if ∆Ci is false. Cj ∉ W and RIij = true then Cj ∈ τt. A component is dependent when a Ci component on the wave has only a relation with another component Cj on the wave τa.

1. The less number of components to change place. 2. We determinate the components Cj∈Ass to disassemble the element Cx chosen with the minimum movements. number of components and 3. The number of components which are to be analysed of Ass for the WP for the disassembly of an element C X. This one vary between 1 and n, depending of the CX position in Ass. The more CX is at an extremity of the sub-assembly, the more the analysed components number is smaller than n.

A component Ci can be defined on the wave τa-1. They are several components Cj, Ck, … dependent on the wave τa. So we have several directions of the wave propagation (if d = 1, it exists only one dependent component).

The graph junction of the sub-assembly corresponds to Ci components, which form the sub-assembly.

3. NEW METHOD FOR DISASSEMBLY USING A MODIFIED WAVE PROPAGATION CONCEPT

Disassemblability: It is the variable, which says if the component CX is removed (disassembled); therefore WP is determinate. If Cj = true the WP is determinate Cj = false we determinate the WP Multitude adjacent: MACi = multitude of components {Cj} which are in contact with Ci. Both components Ci and Cj (i ≠ j) are in contact one with the other, which means they have an or several communes surfaces. So their relation reduces the number of freedom grades between them. We define that as a multitude of the adjacent links MA, of the Cj components, which are in contact with Cj. The direction of disassembly: One direction of disassembly is noted di,j, and represents the direction of component Ci which is disassembled in report of Cj. Removal influence: RICiCj is a variable with a binary value which indicates if we can remove Ci in the absence of some components Cj ∈ MACi, i ≠ j. If RICiCj – true ⇒ it is necessary to disassemble Cj If RICiCj – false ⇒ it is not necessary to disassemble Cj. If we can disassemble one component by the movement of other component, we define that like the removal influence.

The disassembly process is studied for a good and easy separation of the components point of view, in the order to re-use the components (when they are still in a good shape) and the recycling of the others. The disassembly can be done in 2 manners: nondestructive and destructive. The non-destructive one is used when we want the component break invisibility. After the separation, components, which can be used again, can re-integrate the production process. So, the disassembly process must be clearly definite in order to know what will be re-used, what will be recycled, and eventually what will be eliminated. Engineers have to take into account the degradation process which appears, for example: the corrosion, the rust, the destruction of the geometrical form, parties which disappear and quality changes of each component of the product, during its utilisation period. We propose a topology of the disassembly process in the following way: the attachment dismantling process, the separation, the destruction of some components (geometric and physical way), the control of the quality of the components which can be re-used, their manipulation, their packaging and their stocking. The whole disassembly process is represented on the diagram basis, using the standard symbols in the disassembly process. So, we can recognise all operational times and the whole cycle of the disassembly process.

2.2. The wave propagation method

Leaving from the disassembly method of the wave propagation presented in the anterior paragraph, we will build a new method more adapted at the usual disassembly process and for assembly line design. So, we propose a simplification of the method already presented, concerning the attached components treatment and the other components. Moreover, we take into account the irreversible operation when we choose the disassembly sequence, by introducing the IOCiCj variable. Irreversible Operation: If IOCiCj - true ⇒ it is necessary to make the irreversible operation for disassembly Ci in rapport with Cj If IOCiCj - false ⇒ contrary.

In this section, we present the method of the wave propagation (Srinivasan and Gadh,. 1998 and 1999a,b). At t = 0, the disassembly wave contents only the CX component, τ0 = {Ci = CX} and W = {}. After each step, the wave of CX advances with a step (wave) at a time. So t = a, W=τ0 U τ1 U … U τa . So, if we leave from the CX component, we trace the wave. For each knot of the graph, we have a Ci component. We study the relation existing with components on other waves. In this manner, we can say that the disassembly graph is a study of adjacent relations, which are formed between the Ci components on the wavelength. The authors studied the wave propagation concerning only one component, and after that they will generalise the method for “d” dependent components. For t = 0, τ1 = τa ∈ {Cx}

105

After the disassembly sequence is determinate, we do an analysis. This one will help us to analyse if disassembly contains some operations which can be done in parallel, if we have any stability problems of the sub-assembly, and what is the influence on the accessibility and on the geometrical form. We present our improved wave propagation algorithm: Data: a component CX. Objective: Determination of the best disassembly sequence for the CX component by our wave propagation algorithm. 1. τ0 = {CX}; 2. It determines if CX can be disassembled: we study the disassemblability ∆; If ∆Cx = true ⇒ the sequence is determined ; If ∆Cx = false ⇒ the sequence is not as yet determined; 3. It determines the adjacent components MACx with whose CX is in contact; 4. Removal influence (RI); If RICxCj = true ⇒ it is necessary to disassemble Cj ; If RICxCj = false ⇒ it is not necessary to disassemble Cj. It chooses only the components of the MACx for which RICxCj = true ; 5. It calculates ∆Cj for the MACx components for which RICxCj = true; ∆Cj = true ⇒ the sequence is determined; ∆Cj = false ⇒ we determine the sequence; 6. For t = t + 1, the process is repeated and τt is determinate; 7. It determines for each component of a disassembly sequence, the IOCiCj. 8. From the determined sequence W, the sequence which has the minimal number of components ns is chosen and operator verifies if special process of disassembly, as the irreversible operations, is met. This one is the optimal sequence GO. 9. Analysis of the disassembly sequence if we have some operations which can be done in parallel, if we have any stability problems of the subassembly, and to know what is the influence of the accessibility and of the geometrical form.

manufacturing and other operations specifications, historical assembly-disassembly sequences for assembly, maintenance and re-using. It notes A a binary matrix definite by: A(i,j) = 1 only if Cj is an immediate predecessor of Ci. Waves and sequences are deduced with the help of a nonexhaustive algorithm of branch and bound type. 4.RESULTS AND DISCUSSIONS In this section, we present some results on which we will apply the wave propagation algorithm with our proposed improvements. We choose a tap (fig. 1). We want to disassemble the waterproofs cone C3. We determine the best sequence of disassembly for that component. So CX = C3, t = 0, τ0 = {C3} ; ∆C3 = false, so we continue the wave propagation. We determine MAC3 = { C2, C4, C6 }. We will determine the wave τ1. We determine the RI of C3 in function of components of MAC3 = { C2, C4, C6 }. Therefore RIC3C2 = true, RIC3C4 = true, RIC3C6 = true, and IOC3C2 = false, IOC3C4 = false, IOC3C6 = false. The wave τ1 = { C2, C4, C6 }. As RIC6C4 = true, RIC2C4 = true, and IOC6C4 = false, IOC2C4 = false, we disassemble together on the same wave C4 and C6 or C2 and C6 (C4 is an attachment component). For the components for this wave, we determine the disassemblability ∆. Therefore ∆C2 = false and ∆C6 = false, and ∆C4 = true (component type leaf), therefore we continue the wave propagation. We have MAC6 = {C7, C8 } and MAC2 = {C1}, RIC6C7 = true and RIC6C8 = true, and IOC6C7 = false and IOC6C8 = false, and ∆C7= false and ∆C8 = false, and RIC6C7 = true, RIC6C8 = true, and RIC7C8 = true, IOC6C7 = false, IOC6C8 = false, and IOC7C8 = false. For the component C2 we have RIC2C1 = true and IOC2C1 = false. The wave, at t= 2 is τ2 = { C1, C7, C8 }. We continue the process which applies the same method, and we find the sequence W1 = { C15, C14, C1, (C2, C4), C3 } with nS1= 6, and W2 = { C13, C12, C11, C10, C9, (C7 ,C8), (C4, C6), C3} with nS2 = 10. We choose the sequence W1 which has nS1 = 6. Therefore, the optimal sequence SO in this case is SO = W1 = { C15, C14, C1, (C2, C4), C3 }. The wave propagation for the waterproofs cone C3 is presented in figure 5.

We can generate a graph for recoverable components (re-used or recycled) for maintenance and reassembly or for re-using and assembly. In this graph, we sort the components, which can be recuperated of the eliminated components. The problem is similar when we deal with multi-variant products, we have to add a level of complexity to products and assembly systems design. Obtaining a very good line design for products with a lot of re-used components can become very complex. The best way to help the line designer is to propose a set of assistance tools and a methodology. Four object-oriented databases will help the user to design both product and assembly line: Product and parts database, operating techniques database, equipment database and historical sequences database. The input data needed by the software are following: Production specifications, company parameters, parts specifications, product specifications, assembly,

Fig.1 Tap

106

minimal (nS = 5, in this case). The great advantage of our algorithm are put in evidence in the case when we have several disassembly sequences and we want quickly to determine the optimal sequence with the minimal number of components to disassemble.

C1 – lower valve body; C9 – spacer; C10–fastener component; C2 – seal component; C11 – open - close wheel; C3 – waterproofs cone; C12 – slice Grower; C4 – waterproofs nut; C5 – waterproofs corps ring;C13 – nut M8; C 14 – blocking axe; C6 – axe tap; C15 – nut; C7 – top corps tap; C8 – waterproofs axe ring; C16 – spacer. Remark 1:On long term, some wear and tear can appear for some components of the sub-assembly. For example, it can be a wear and tear of the nut C13, on the top axe of C6. In this case, the disassembly is realisable, by a partial or total destruction of the components; therefore we have an irreversible operation. Remark 2: We can have assembly case by a welding between the wheel C11 and the axe C6. In this case the IOC6C11 = true. The disassembly is possible by a destruction of the C11 and C6 components. Therefore, it is preferable, in this case, to choose the disassembly sequence W1, because by achieving some irreversible operations, dismantling of components appears. So, we prefer to choose a disassembly sequence, which has a larger number of components, without dismantling the components.

Matrix No. 1

Matrix No. 2

Fig.3 The matrix of the immediate predecessors nS = 8 Wave No. 1 Wave No. 2 Wave No. 3 Wave No. 4 Wave No. 5 Wave No. 6

At the Figure 2, we present a pump with sprockets wheels, which have the function of bringing the oil under pressure by the de-pressure created between the sprocket of the sprockets. So, we propose ourselves to disassembly the ruling axle 11. We will determine all disassembly sequences possible with the algorithm presented in that article. In this way, we will construct the matrix of the immediate predecessors in Figure 3. After that, the program determines automatically the disassembly sequences in Figure 4, respectively for each matrix of the predecessor.

: : : : : :

3 7 8 5 9 10 13 11

Disassembly sequence 1

nS = 5 Wave No. 1 Wave No. 2 Wave No. 3 Wave No. 4

: : : :

3 1 12 14 11

Disassembly sequence 2

Fig.4 The disassembly waves for the pump with sprockets wheels 5. CONCLUSION In conclusion of this research, the disassembly must be seen in the point of view of each assembly or component can be disassembled as easy as possible and with a minimum number of operations. In this way, it is not necessary to disassemble the whole product and this disassembly is the least destructive. So, it is important to choose geometrical forms and surfaces for the disassembly that can assure an easy access to all other components. We propose a new algorithm, which determines the disassembly sequence depending on the minimal number of components to be disassembled and other various criteria. In this article, we insist on the fact that the best sequence is not always the one that has the minimal number of components. In fact, various situations in some irreversible operations have to be done, could occur and some dismantling of some components should appear. Our algorithm takes into consideration the irreversible operations, and determines the disassembly sequence that avoids the realization of these irreversible situations, therefore, the dismantling of components. The advantage in using our algorithm consists in fact that we avoid the dismantling of components for the disassembly. Moreover, in the determination of the disassembly sequence, we take into account the parallel operation to be effectuated.

Fig.2 The pump with sprockets wheels C8 – simmering; C1 – housing; C9 – fittings; C2 – conducted axle; C 10– body1 ruling axle; C3 – screw; C11 – ruling axle; C4 – washer; C 12 – body2 ruling axle; C5 – cover; C 13 – body1 conduct axle; C6 – o'ring; C14 – body2 conduct axle; C7 – security ring; For the chosen example, we find two disassembly games. The first has the component number to disassembly nS = 8, and the second one has nS = 5. Finally, the program do an optimisation, choosing automatically the disassembly sequence with nS

107

Once the immediate predecessors were determined, it took only seconds or even a fraction of a second to deduce the waves and the sequences. The best disassembly sequence is determined with a minimal number of components to disassemble, which also avoids irreversible operations. Finally we propose a

way to help the line designer with a set of assistance tools and a methodology. We need four objectoriented databases to help the user to design both product and assembly line.

τ4 τ3

C15

C15 τ1

τ2 C8

C14

C14

τ1 τ0

C C2

C6

C3 C4

C4

τ5

C7

τ6 C11

C9 C10

τ7 C12 C13

Fig.5 The wave propagation for the waterproofs cone the wave propagation abstraction. ComputerAided Design, 30 (8), 603-613. Srinivasan, H., Figueroa, R. and Gadh, R. (1999b). Selective disassembly for virtual prototyping as applied to de-manufacturing. Robotics and Computer Integrated Manufacturing, 15, 231245. Tomiyama, T. and Umeda, Y. (1997). Life Cycle Design for the Post Mass Production Paradigm. Proc. of the CIRP 1997 International Design Seminar on Multimedia Technologies for Collaborative Design and Manufacturing. University of Southern California, Los Angeles, USA, 117-125. Westkämper et al (1999). Integrated Development of Assembly and Disassembly. CIRP STC A, 1-9.

ACKNOWLEDGEMENT This project is sponsored by the Natural Sciences and Engineering Research Council of Canada. REFERENCES AGARD (1998). Virtual manufacturing. Report 821. Hull, Canada. Jabbour, T., C. Mascle, and R. Maranzana (1996). Représentation des caractéristiques d’assemblage d’un produit mécanique. Revue Internationale de la CFAO et d’Informatique Graphique, 5 (11), 545-566. Jabbour, T., C. Mascle and R. Maranzana (1998). A data base for the representation of assembly features in mechanical products. International Journal of Comptutional Geometry & Applications, 8 (5&6), 483-508. Mascle, C. (1999). Feature-based assembly Model and Multi-agents system structure for ComputerAided-Assembly. Proceedings of the IEEE International Symposium on Assembly and Task Planning (ISATP’99). Porto, Portugal, 8-13. Srinivasan, H. and Gadh, R. (1999a). Selective Disassembly : Representation and comparative Analysis of Wave Propagation Abstraction in Sequence Planning. IEEE International Symposium on Assembly and Task Planning (ISATP ’99). Porto, Portugal, 129-134. Srinivasan, H. and Gadh, R. (1998). A geometric algorithm for single selective disassembly using

108