Robotics & Computer-lnteorated Manufacturing, Vol. 10, No. 1/2, pp. 41-48, 1993 0736-5845/93 $5.00 + 0.00 Printed in Great Britain (~C~ 1992 Pergamon Press Ltd

• Paper

CAD SYSTEM WITH PRODUCT ASSEMBLY/DISASSEMBLY PLANNING FUNCTION

EIJI ARAI* a n d KAZUAKI IWATA t

*Department of Precision Engineering, Shizuoka University, Johoku, Hamamatsu 432, Japan and TDepartment of Computer Controlled Machinery, Osaka University, Osaka, Japan

Assembly design of mechanical products is one of the most important subjects in the CAD/CAM systems of the near future. The product model (geometric model) is the input in many CAD/CAM systems in order to determine the assembly/disassembly sequence. Disassembly is executed with cooperation between the computer system and the designer. The kinematic simulation should support the designer; however, most simulation systems do not have enough functions to do this. There are many problems left, mainly caused by the lack of functionality of the kinematic simulation of mechanical products. When product designers are requested to consider product assembly and extract the fundamental assembly plan from their design solutions, the kinematic simulation and evaluation functions, to obtain the product assembly plan, are most important in CAD systems.

This paper deals with a kinematic simulation system used in assembly planning that simulates physical phenomena including the effect of gravity, and a prototype simulator is developed. An evaluation method for the ease of assembly/ disassembly based on the kinematic simulation system is also proposed, which gives a decision support tool. Here, the assembly sequence is considered as the reverse of the disassembly sequence, which is justified through taking the kinematic simulation into consideration.

The part to be removed is decided by comparing the evaluation standard values of candidate parts when several parts can be removed at the same time. With the use of the product model presented by 3-D solid geometry, possible movements of each part in the product are calculated, and the removal possibilities from the product can be searched. When there are several ways to remove the part from the product, the best disassembly operation for the part is selected on the basis of shortest distance of part movement.

In the evaluation, the effect of gravity is considered, which may cause the next assembly/disassembly operation to be difficult. The developed simulation system is also effective in detecting the necessity for jigs or fixtures in the assembly sequence.

Through a case study, it is shown that the developed system is valid and effective in enabling product designers to obtain product assembly plans.

1. INTRODUCTION Manufacturing software has played a very important role, especially in mechanical products industries. The core is the CAD/CAM system. The target in a CAM system has shifted from NC programming in machin- ing to process planning in machining and assembly planning of products.

A practical level of support for designers and plan- ners in assembly planning of mechanical products has not been realized. An assembly planning system is one of the most important in the CAD/CAM field to realize the flexible assembly system, x

Research contributing to product assembly plan- ning has mostly taken the standpoint that the assem- bly sequence is simply the reverse of the disassembly sequence. 2 The product model (geometric model) is the input in order to determine the disassembly se- quence, and disassembly is executed with cooperation between the computer system and the designer.

However, the actual assembly sequence is not al-

41

ways the reverse of the disassembly sequence. For example, the dimensions may be different in an assem- bled product from those in the parts. Elastic forces may be utilized in the assembly.

There are two methods to develop a product assem- bly planning system: one is to generate the disassem- bly sequence taking the actual conditions into consideration first, and rearrange it into the assembly sequence; the other is to generate the assembly se- quence directly with the use of know-how used in product assembly.

Taking the importance of disassembly planning, especially in maintenance engineering, into consider- ation, we adopt the first method in the development of a product assembly planning system. The objective of this paper is to propose a new and valuable methodol- ogy and to introduce the fundamental tools in the product assembly area, where assembly planners make decisions with the use of computer support.

When product designers are requested to consider

42 Robotics & Computer-Integrated Manufacturing • Volume 10, Numbers 1/2, 1993

product assembly and extract the fundamental assem- bly plan of their design solutions, the kinematic simu- lation in assembly and evaluation functions for the ease of product assembly are most important in CAD systems.

This paper deals with a kinematic simulation sys- tem used in the assembly planning that simulates the physical phenomena including the effect of gravity, and a prototype simulator is developed. At the same time, a method of evaluating ease of assembly based on the kinematic simulation system is proposed. These are the first steps in developing the phenomena simu- lator and show the technical possibility of realization of product assembly CAD systems in the near future.

2. P R O D U C T DISASSEMBLY P L A N N I N G PROCESS

The product assembly planning process is the phase in which the assembly sequence and the assembly opera- tions of the product are decided. From the standpoint that the assembly sequence is the reverse of the disassembly sequence, a CAD system to support the disassembly planning should be developed first. The system developed is based on the kinematic simula- tion of mechanical products, which improves the shortcomings of the assumption that assembly is the reverse of disassembly. For instance, the necessity for jigs and fixtures can be extracted through the simula- tion.

Figure 1 shows the total flowchart for the disassem- bly planning system. First, the product model based on a 3-D solid model is input, and information on contacts among parts is automatically detected from the product model. Next, parts to be removed are detected in the product model, and the best part among them is decided by an evaluation based on the kinematic simulation which is mentioned in the fol- lowing sections. The disassembly sequence is com- puted by removing parts one by one, and re-building the product model through the kinematic simulation. Last, information on the assembly sequence and the necessity for jibs and fixtures is computed from the disassembly sequence with reference to the simulated physical phenomena.

3. DETECTION OF DISASSEMBLY OF PARTS The possibility of disassembly of one part depends on the existence of movement without interference from other parts. Figure 2 shows the flow chart to detect removable parts. The movements considered here are limited to linear movements and simple rotations, and contacts among parts are also limited to face to face contacts and line contacts. Part shapes are limited to the combination of cuboids, cylinders and cones.

3.1. Calculation oj'possible movements It is essential to detect the disassembly of parts from the product in order to calculate the possible move- ments for each part in the product. Possible move- ments are limited by the geometric conditions caused

Product model

l" Search for contacts

among parts

Calculation of possible movements of each part

Search for removable parts

Calculation of evaluation standards for each part

Decision for removal of part

Disassembly

Disassembly sequence

Fig. 1. planning system.

Re-building product model

No

Total flowchart for the disassembly/assembly

by contacts between parts. With use of 3-D solid models, since in many models geometric shapes are approximated by planes, the detection of contacts requires removal of errors typically shown in Fig. 3. Figure 4 shows the typical contacts seen in mechanical products which can be detected when executed based on the exact values of radii of cylinders and cones. In the developed system, the limit of the kinds of contacts as shown in Fig. 4 introduces an exact detection of contacts between parts.

Possible movements for each part while keeping contact are calculated based on the types of contact detected) Possible movements without keeping con- tact are detected by calculating the inner products between the direction vectors in space and the normal vector of the contact face as shown in Fig. 5.

3.2. Possibility of disassembly of parts The possibility of disassembly of the parts is detected by moving the part along each of the calculated

CAD system with product assembly/disassembly planning function • E. ARM and K.,|WATA 43

Product model 9 =1 For each part

I Detection of contacts I

I Calculation of possible movements

Yes Is there at least one movement that removes the part ? .

Search for other movements each of which removes

the part

Calculate the shortest distance

to remove the part

1 I Disassembly

possible

No

No T For each movement iP

Is there a change in contacts between the part and others ?

Yes

No

Kinematic simulation I

Change the product model ]

Yes

All possible movements finished ?

Yes

All parts searched ?

I Disassembly impossible

Fig. 2. Flowchart to detect removable parts.

Fig. 3. Typical errors caused by planar approximation.

possible movement sequences with t h e use of the kinematic simulation system.

The search for the possibility of disassembly of parts from the product is executed for each part. When a part has a possible movement along which the part can be removed, other possible movements are searched through which the part can be also removed. The distances necessary to remove the part from the product are calculated by the search movements and the movement which gives the shortest distance is stored in the database.

44 Robotics & Computer-Integrated Manufacturing t l Volume 10, Numbers 1/2, 1993

Cy l inde r -cy l i nde r

Cy l i nde r - cone

C o n e - c o n e

Cy l i nde r -p l ane

C o n e - p l a n e

Face con tac t Line contac t

Fig. 4. Typical contacts in mechanical products including curved surfaces.

Moving d i rect ions Normal v e c t o r . / . /

Moving d i rect ions

rmal

(x, y, z) . . . . the moving direction

(a, b, c) . . . . the normal vector of contact face

1 a ~ x + b ~ y + c ~ z > O

vec tor

(x, y, z) . . . . the moving direction

(al , b l , c l ) . . . . the normal vector of contact face (a2, b2, c2) . . . . the normal vector of contact face

1 a l e x + b l ~ y + cl~z_-> 0

a 2 $ x + b2$y + c 2 ~ z > O

Fig. 5. Calculation of the possible movements without keeping contact.

When a part cannot be disassembled in one move- ment, the system detects this possibility by the com- bination of several possible movements. The system supports the designers in searching for the position where the contact status of the part changes (decrease/ increase in the number of contacts) as shown in Fig. 6 and rebuilds the product model through kinematic simulation. By using the re-built product model, the possibility of disassembling the part from the product is searched similarly.

Possible movements and re-built product models are stored in the database where the movements are described as nodes in the tree structure shown in Fig. 7.

Detection of the change of contact status is executed by using ray tracing when the contacts increase and by calculating the length or angle of the contact geomet- ric elements (faces and lines) along the moving direc- tion when the contacts decrease.

3.3. Kinematic simulation When a part is disassembled from the product, other parts may move by gravity if the disassembled part plays the role of a fastener of other parts, which it is essential to know especially in obtaining information

Fig. 6. Detection of change in contact status through several movements.

CAD system with product assembly/disassembly planning function • E. ARM and K. [WATA 45

Fig. 7.

Initial status

I 1

g Initial status

g j ~ Initial status

t Finish 1 ( ' ~

Tree structure to store the possible movements.

on the necessity for jigs or fixtures in assembly. The kinematic simulation system calculates the progress of disassembly and the effect of gravity to rebuild the new product model according to the simulation result. Figure 8 shows the flowchart of the simulation system.

The developed system takes only the effect of gravity into consideration among the actual physical phe- nomena as a first step.

Comparing the possible movements with the direc- tion of gravity, whether the part moves by gravity is detected. Gravity may move the part when the part is allowed straight movements vertically downward, or allowed rotation with a perpendicular line from the centre of gravity of the part to a non-vertical rotation axis. Although, in practice, movements caused by gravity are affected by friction, the evaluation of the effect friction is not considered in the developed system.

In the kinematic simulation, two different structures of the products are considered: open loop and closed loop. Forces and torques on joints can be calculated automatically in the open loop structure; however, torques on joints cannot be determined in the closed loop structure. On the other hand, the degree of freedom is reduced in the closed loop structure which can be calculated automatically. In the developed system when the structure includes closed loops, torques on joints are assumed to be zero or infinite which enables kinematic simulation of 3-D products including closed loop structures such as that shown in Fig. 9.

The kinematic simulation is executed until the

Product model

Search for contacts ] among parts

Calculation of possible ] movements for each part

Evaluation of effect ] caused by gravity etc.

No effect 1,9 Effect exists

Calculation of the position data after parts moved

Conflict~interference evaluation [

Generation of next step [ new product model I

Clock

t No conflict / interference Conflict / interference

Calculation of conflict time and position data

I

i Display of the movements

Fig. 8. Kinematic simulation flowchart.

46 Robotics & Computer-Integrated Manufacturing • Volume 10, Numbers 1/2, 1993

Fig. 9. Kinematic simulation of the closed loop structure.

contact status among parts is changed, such as occur- rence of a conflict as in Fig. 10 or a part is completely removed from the product. When parts move by gravity, the simulation system reports to the designers the existence of the effect caused by gravity. The designers can input the system to hold the moving parts at the original position (killing the gravity effect) which means the use of jigs or fixtures to hold the parts in the actual assembly/disassembly process, in- formation which is also stored in the database and utilized as jigs/fixtures information.

Through the kinematic simulation, all possible parts to be disassembled are detected and the move- ments for each removal are stored.

4. DECISION OF DISASSEMBLED PART When only one part can be disassembled from the product, the part is removed and the new product is re-built with the use of the kinematic simulation system. If there are several candidates to be disassem- bled, the designers have to decide the best part. The developed system supports decision making by offer- ing an evaluation function and a branch-and-bound calculation system.

4.1. Evaluation function Three standards to evaluate the ease of disassembly are proposed as follows:

I. Function to evaluate the effect of physical phe- nomena after removing the part,

2. Times of movements in order to remove the part from the product,

3. Moving distance between the original position and the final position of the disassembled part.

According to the above three standards, the objec- tive function is described as follows to show the degree of difficulty to disassemble a part:

f = a*w + b*x/X + c*y/Y+ d*z/Z

where Wis 1 if other parts move by gravity when a part is removed, and 0 otherwise,

x is the time of movements necessary to disassemble a part,

X is the maximum time to remove one of the candi- date parts from the product,

y is the total distance to disassemble a part by using straight movements,

Y is the maximum distance for all candidates, z is the total angle to disassemble a part by using

rotations, and Z is the maximum angle for all candidates.

The function is linearly connected by four coeffi- cients a, b, c, and d which show the importance of the four factors, and are determined with the use of AHP (Analytic Hierarchy Process) by designers.

4.2. Decision of the disassembled part The part to be disassembled is decided by comparing the objective function value of candidate parts when several parts can be disassembled at the same time. The objective function is also used when there are several ways to disassemble one part.

In the tree such as Fig. 7 to show the possible disassembly process, the evaluation function valuef is also calculated at each node. The branch-and-bound method is used to detect the optimum disassembly sequence to minimize the value o f f in the tree.

Fig. 10. Kinematic simulation with effect of gravity.

5. CASE STUDY Figure l l shows a typical simulation example to determine the disassembly sequence executed by the developed system. Figure 1 l(a) shows the initial prod- uct model which is described by a 3-D solid model.

The product consists of five parts. The calculated disassembly sequence is shown in (a) to (g). Instead of the arm part in (f), another part can be disassembled, which causes the rotational movement of the arm by gravity as shown in (h). According to the objective function, the system adopts the arm.

6. CONCLUSION A CAD system to support designers in making a product assembly/disassembly plan is developed. The system is based on kinematic simulation which is

CAD system with product assembly/disassembly planning function • E. ARAI and K. IWATA 47

(a) (b)

(c) (d)

,>

(e) (f)

(g) (h)

Fig. 1 !. Disassembly sequence of a mechanism (a)-(g) and influence of gravity (h).

48 Robotics & Computer-Integrated Manufacturing • Volume 10, Numbers 1/2, 1993

essential to construct the product assembly planning system.

With use of the product model presented by 3-D solid geometry, possible movements of each part in the product are calculated. For each part, the possi- bilities of disassembly can be searched by using the possible movements. When the designers want to designate the movement of a part, the search must be more effective. If there are several ways to remove the part from the product, the best disassembly operation of the part is selected.

There may be several parts which can be disassem- bled at the same time from the product. The selection of the best requires evaluation of the ease of disassem- bly. One approach for such evaluation is proposed. In the evaluation, the kinematic simulation is essential to take the physical phenomena into consideration. As a

first step, the effect of gravity is considered. The developed simulation system is built in the evaluation system, and is also used to point out the necessity of jigs or fixtures in the assembly sequence.

Through a case study, the developed system is shown to be valid and effective in supporting product designers to determine a product assembly/disassem- bly plan through the kinematic simulation functions.

REFERENCES 1. Sata, T.: Computer integrated manufacturing: present

state and future. Preprints of International Seminar on Factory Automation, pp. 64 77, 1986.

2. Sekiguchi, H. et al.: Method of developing part specifica- tions from assembly drawing of machine unit. J. J S P E 51 (2): 359 365, 1985 (in Japanese).

3. Arai, E.: Kinematic simulation system of mechanical products. IMS Activity Report 1986, pp. 82-94, 1987.