ASSEMBLY PROCESS:ASSEMBLY TECHNOLOGIES AND SYSTEMS

ASSEMBLY TECHNOLOGIES AND SYSTEMS

Basic Structures of Assembly Systems

Assembly is the sum of all processes needed to join together geometrically determined bodies. A dimensionless substance (e.g., lubricants, adhesives) can be applied in addition (VDI 1982).

In addition to joining in the manufacturing process, handling of components is the primary func- tion of assembly. The assembly also contains secondary functions such as adjusting and inspecting as well as various special functions (see Figure 8).

Joining is defined by DIN 8593 as a part of manufacturing processes. In this case, the production of a compound consisting of several parts can be achieved by merging, pressing, pressing in, metal forming, primary shaping, filling, or by combining substances.

Handling is defined in VDI Guideline 2860 / 1 as the creation, defined varying, or temporary maintaining of a prescribed 3D arrangement of geometrical defined solids in a reference coordinate system. For this, procedures such as ordering, magazining, carrying on, positioning, and clamping are important. It is simple for a human being to bring parts into correct position or move them from one place to another. However, a considerably larger expenditure is necessary to automate this task. An extensive sensory mechanism often must be used.

Manufacturing of components is subject to a great number of influences. As a result, deviations cannot be avoided during or after the assembling of products. These influences must be compensated for, and thus adjusting is a process that guarantees the required operating ability of products (Spur and Sto¨ferle 1986).

Testing and measuring are contained in the inspection functions. Testing operations are necessary in all individual steps of assembly. Testing means the fulfillment of a given limiting condition. The result of the test operation is binary (true or false, good or bad). On the other hand, specifications are determined and controlled by given reference quantities while measuring. Secondary functions are activities, such as marking or cleaning operations, that can be assigned to none of the above functions but are nevertheless necessary for the assembly process.

Assembly systems are complex technical structures consisting of a great number of individual units and integrating different technologies.

There are different possibilities for the spatial lineup of assembly systems. One possibility is a line structure, which is characterized by:

• Clear flow of materials

• Simple accessibility of the subsystems (e.g., for maintenance and retrofitting)

• Simple lineup of main and secondary lines

• Use mainly for mass production (the same work routine for a long time)

Alternatively, an assembly system can be arranged in a rectangular structure, which is character- ized by:

• Very compact design

• High flexibility. The combination of opposing subsystems is easy to realize.

• Poor accessibility to the subsystems during maintenance and retrofitting

• Use mainly for small and middle lot sizes

In addition to different spatial lineups, other basic modifications of cell structure are possible for achieving the required efficiency (see Figure 10).

The number of work cycles that can be carried out on an assembly system depends on the size of the assembly system (number of cells) and the required productivity. The availability drops as the number of stations increases. Therefore, the distribution of work cycles onto several machines is necessary. In this case, the productivity increases with decreasing flexibility. The entire assembly system can be subdivided into different cells, which are connected by the flow of materials and information. The basic structure of a cell consists of a tabletop, a basic energy supply, the mounted handling devices, the internal conveyor system, and a safety device (protective covering, doors with electrical interlock). The individual cells should be built up modular and equipped with standardized interfaces (energy, pneumatic, and information technology). A strict module width for the spatial measurements is also useful. As a result, fast realization of the individual assembly cells, a high degree of reuse, and flexible use are guaranteed. The assembly cell itself consists of different com- ponents and units (Figure 9). The main components of an assembly cell are devices (integrated robots, modular system built of numerically controlled axes, pick-and-place devices, etc.), its mechanical construction, the design of the grippers, the actuating system, the control system, the method of programming, and the sensor technology used. Furthermore, the part supply is very important (ordered or disordered part supply; see also Section 2.3). Last but not least, the joining process (see also Section 2.2) must be adapted and optimized to meet all the necessary requirements.

The different alternatives for design of assembly systems are represented in Figure 10. The flexible cell is able to carry out all necessary assembly functions for a certain variant spectrum (three variants,

A, B, and C, are represented here). For this purpose, it has all the necessary main and secondary functions for the assembly process. High variant flexibility, sequential processing of the individual assembly tasks, as well as the resulting high nonproductive times, lead to the use of flexible cells, mostly for small and middle lot sizes.

If the duplex cell is used, the assembly tasks will be distributed over two handling units. These always have the full variant flexibility. With regard to function, the cells need not be completely flexible. They must be able to carry out only a part of the assembly tasks. The function size of the multiple grippers (in the case of unchangeable variation of the functions) can be increased by parallel assembly with two independent handling devices. Even shorter cycle times can be achieved than with the flexible cell.

Serial cells are flexible only with respect to the assembly of variants. However, the function size of an individual cell is highly limited. The spatial lineup of the cells is responsible for the fact that the cell with the longest cycle time determines the total cycle time of the assembly process. Consid- ering the spatial extension of serial structures, it is obvious that the integration level of the system is smaller than that of flexible cells or duplex cells.

Parallel cells will be used if only one assembly cell is responsible for the complete assembly of a variant. Each individual cell has a high degree of functional flexibility, which unfortunately can only be used for one variant. Therefore, the potential of the resources cannot be exploited because identical assembly conditions must be used during the assembly of different variants. If the individual variants show only a small number of similar parts, which on top of that often have different gripping conditions, splitting up variants will be advantageous. As a result, a smaller variant flexibility will be necessary.

The serial structure will react extremely sensitively to fluctuation in the number of variants. At worst, restructuring or reconstruction of all cells will be required. In contrast, the rate of utilization of the system will remain almost stable if full variant flexible cells are used. Because the handling devices of parallel cells have full functional flexibility, it is not necessary to carry out an adaptation of the number of flexible systems. Due to changes in lot size, parallel cells designed for special variants will even have too much or too little capacity, which cannot be removed without adequate measures.

Joining Technologies

Classification and Comparison of Joining Technologies Industrialized manufactured products predominantly consist of several parts that are usually manu- factured at different times in different places. Assembly functions thus result from the demand for joining subsystems together into a product of higher complexity with given functions. According to German standard DIN 8580, joining is defined as bringing together two or more workpieces of geometrically defined form or such workpieces with amorphous material. The selection of a suitable joining technique by the technical designer is a complex function. In DIN 8593, manufacturing methods for joining (see Figure 11) are standardized.

In addition to the demands on the properties of the product, economic criteria have to be consid- ered (e.g., mechanical strength, optics, repair possibilities). The ability to automate processes and design products with the manufacturing process view are important points to focus on. Therefore, examples of the joining techniques in automation presented are given here.

The joining processes can be divided into various classes on the basis of several criteria:

• Constructional criteria: strength, form, and material closure connections or combinations of these possibilities

• Disassembly criteria: in removable and nonremovable connections

• Use of auxiliary joining part, e.g., screws

• Influence of temperature on the parts

Some important joining techniques and applications are presented below. Pros and cons of the joining techniques will be discussed in regard to the features of function, application, and assembly.

Bolting

Screwing is one of the most frequently used joining processes in assembly. Screwdriving belongs in the category of removable connections (ICS 1993). Joinings are removable if the assembly and separation process can be repeated several times without impairing the performance of the connection or modifying the components. The specific advantages of detachable connections are easy mainte- nance and repair, recycling, and allowing a broad spectrum of materials to be combined.

The most important component of the bolt connection is the screw as the carrier of substantial connecting functions and the link between the components. The screw’s function in the product determines its geometrical shape and material. Important characteristics of a screw are the form of thread, head, shaft end, tensile strength, surface coating, tolerances, and quality of the screw lots.

Screwing can be manual, partly automated, or fully automated. The bolting tools used can be classified according to structural shape, control principle, or drive. Substantial differences occur in the drive assigned, which can be electric, pneumatic, or hydraulic (only for applying huge torques). Figure 12 shows the basic structure of a pneumatic screwdriving tool.

A basic approach is to increase the economy of automated screwing systems and reduce deadlock times. Process stabilization can be achieved by using fast control loops in process control and di- agnostic support of the operators for error detection and recovery (see Figure 13) (Steber 1997). Further, position errors of the parts can be compensated automatically by adaptation of the coordinate

system in flexible handling systems, such as robots and image-processing systems. The automated screwing technique is becoming more and more important in mass or serial production. Apart from reduction unit costs, longer production times, and higher output, it offers the advantage of continu- ously high quality that can be documented.

Riveting / Clinching

Riveting is one of the classical joining processes. It is in particularly wide use in the aircraft industry. What all rivet methods have in common is that an auxiliary joining part, the rivet, must be provided. Blind rivets are often used in device assembly because accessibility from one side is sufficient.

Recently the punching rivet method has become increasingly important in the automobile industry. It is one of the joining processes with less heat influence on the material than with spot welding (Lappe 1997). The advantage of this technique is that no additional process step for punching the hole into the parts is necessary. The punching rivet itself punches the upper sheet metal, cuts through, and spreads in the lowest metal sheet. Figure 14 shows the profile of a punching rivet with typical characteristics for quality control. It is characterized by the fact that different materials, such as metal and aluminum, can be combined. The use of new lightweight design in automotive manufacturing means that materials such as aluminum, magnesium, plastics, and composites are becoming increas- ingly important. Audi, with the introduction of the aluminum space frame car body concept, is a pioneer in the application of punching rivets.

In clinching, the connection is realized through a specially designed press-driven set of dies, which deforms the parts at the join to provide a friction-locked connection. No auxiliary jointing part is necessary, which helps to save costs in supply and refill of jointing parts.

Sticking

Figure 15 shows a comparison of various destructive detachable joining technologies. The progress in plastics engineering has had positive effects on the engineering of new adhesives. Sticking is often

used in combination with other joining processes, such as spot welding and punching rivets. The connection is made by adhesion and cohesion (Spur and Sto¨ferle 1986).

Adhesives can be divided into two groups. Physically hardening adhesives achieve adherence by two different mechanisms. The first is by cooling of the melted adhesive, and the second is by the evaporation of solvent or water (as the carrier) out of the adhesive. Because the adhesive does not interlace, it is less resistant to influences such as heating up, endurance stress, or interaction of solvent. Chemically hardening adhesives solidify themselves by a chemical reaction into a partially interlaced macromolecular substance characterized by high firmness and chemical stability. Adhesives can also be differentiated into aerobic and anaerobic adhesives.

The quality and firmness of an adhesive depends on the conditions at the part surface. The wettability and surface roughness of the parts, as well as contamination (e.g., by oil), play a substantial role. To ensure quality of sticking, therefore, often a special surface treatment of the jointing parts is necessary. This represents an additional process step, which can be automated too. Typically, car windows are automatically assembled into the car body in this way.

Welding

Welding methods can be subdivided into melt welding and press welding methods. In melt welding, such as arc welding, metal gas-shielded welding (e.g., MIG, MAG) or gas fusion welding, the con- nection is made by locally limited heating to just above the liquidus temperature of the materials. The parts that should be connected and the usually used additional welding materials flow together and solidify. In pressure welding, such as spot welding, the connection is realized by locally limited heating followed by pressing or hammers. Welding methods differ according to their capacity for automation. In the building of car bodies, fully automated production lines with robots are usually already in use. Fusion welding, such as gas-shielded arc welding, makes higher demands on auto- mation. Therefore, robots are suitable, which should be also equipped with special sensors for seam tracking. Therefore, both tactile and contactless sensors (e.g., optical, capacitive, inductive) are used. With the increasing power density of diode lasers, laser beam welding with robots is becoming much studied.

Peripheral Functions

Handling Devices

For complex operations, industrial robots are normally used. If the handling task consists of only a simple pick-and-place operation, specially built devices with one or more linear axes are probably the better choice. These handling devices are classified by their degrees of freedom (DOF), which indicate the number of possible translational and rotational movements of a part. Therefore, six DOF—three translational and three rotational—are required to arrange an object in a defined way in a 3D room. Opening and closing of grippers is not counted as a degree of freedom, as this movement is used only for clamping and does not, strictly speaking, move the part.

Mechanically controlled inserting devices offer one to two DOF. They are suitable for simple handling tasks, such as inserting. Due to their strict mechanical setup, they are compact, robust, and very economical. Their kinematics allows them to reach different points with a predefined, mechan- ically determined motion sequence. This motion can be adapted to the different handling tasks by the use of different radial cams (control curves). Due to the sensor-less mechanical setup, it is an inherent disadvantage of the system that the precision of the movement is not very high. In an open control loop, only the end positions are detected by sensors.

If the handling task demands higher accuracy or flexibility, numerically controlled (NC) axes or industrial robots are recommended. Using two or three linear axes allows more complex and precise handling tasks to be performed than with mechanical handling devices.

Industrial robots are built in different setups. The most common are the SCARA (selective com- pliance assembly robot arm) robot and the six-DOF robot. The SCARA robot usually has four DOF, three translational and one rotational, whereas the z-stroke represents the actual working direction. The other axes are used for positioning. SCARA robots are often used for assembling small parts automatically with very short cycle times. Six-DOF robots are used for more complex tasks because they are more flexible in their movements so they can grip parts in any orientation.

Grippers

Grippers have to perform different tasks. Not only do they have to lock up the part to be handled (static force), but they also have to resist the dynamic forces resulting from the movement of the handling device. They also have to be made so that the parts cannot move within them. Additional characteristics such as low weight, fast reaction, and being fail-safe are required. There are different techniques for gripping parts. Grippers are typically classified into four groups by performing prin- ciple (Figure 17): mechanical, pneumatic, (electro-) magnetic, and other. With a market share of 66%,

mechanical grippers are the most commonly used. Mechanical grippers can be subdivided by their kind of closing motion and their number of fingers. Three-finger-centric grippers are typically used for gripping round and spherical parts. Two-finger-parallel grippers perform the gripping movement with a parallel motion of their fingers, guaranteeing secure gripping because only forces parallel to the gripping motion occur.

Because these gripper models can cause harm to the component surface, vacuum grippers are used for handling damageable parts. Two-dimensional parts such as sheet metal parts are also handled by vacuum grippers. Using the principle of the Venturi nozzle, an air jet builds up a vacuum in the suction cup that holds the parts. When the air jet is turned off, the parts are automatically released.

Heavy parts such as shafts are lifted not by mechanical grippers but with electromagnetic grippers. However, secure handling, not exact positioning, is needed when using these grippers.

To handle small and light parts, grippers using alternative physical principles, such as electrostatic and adherent grippers, are used because they do not exert any pressure that could cause damage to the part. Fields of application include microassembly and electronics production.

Security of the gripped parts and the workers is an important consideration. Toggle lever grippers, for example, ensure that the part cannot get lost even if a power failure occurs. In contrast, parts handled by vacuum grippers fall off the gripper if the air supply fails.

Feeding Principles

To be gripped automatically, parts have to be presented to the handling device in a defined position and orientation irrespective of the above-mentioned gripping principles. For small parts, which are assembled in high volumes, the vibratory bowl feeder is commonly used. This feeder integrates three different tasks: singulation, orientation, and presentation.

The parts are stored in the bowl feeder as bulk material in the bunker, which is connected via a slanted feeding track to the pick-up point. Using suitable vibrations generated by lamellar springs through an electric motor, the parts move serially toward the pick-up point. The movement is a result of the superposition of micro-throws and a backwards slide.

A disadvantage of the bowl feeder is that the parts may be damaged by the micro-throws and the relative movement and contact between the parts and between the single part and the surface of the bowl feeder. Additionally, this feeding principle leads to the annoyance of noise. Another disadvan- tage is that the bowl feeder is extremely inflexible with regard to different parts because it is strictly constructed for the use of one special part. Also, it is not yet possible to automate the process of constructing a bowl feeder. Rather, whether the bowl feeder will meet the requirements is up to the experience and skill of the person constructing it. Nevertheless, this is and will continue to be one of the most important feeding technologies for small parts.

Larger and more expensive parts, or parts that must not be damaged on their surface can be presented to the handling device palletized by using deep-draw work trays.

Generally speaking, a degree of orientation, once reached, should never be allowed to be lost again, or higher costs will result.

Linkage

Normally a product is not assembled and produced by only one machine. Therefore, multiple handling devices, machines, and processes have to be arranged in a special way. The different ways of doing this are shown in Figure 19. The easiest, which has been widely used since the industrial revolution, is the loose linkage. This is characterized by the use of discrete ingoing and outgoing buffers for each machine or process. It is thus possible to achieve independence for the different cycle times. Disadvantages are that very high supplies are built up and the flow time is very high. An advantage is that the production will keep running for some time even if one machine fails, because the other machines have supplies left.

The stiff linkage works the opposite way. The workpieces are transported from station to station without a buffer in between. The corresponding cycle time is equal to the process time of the slowest machine. As a consequence, if one machine is out of order, the whole production line has to be

stopped. A well-known application of this principle can be found in automobile production, where, because there are no bypasses, the whole assembly line has to be stopped if problems occur at any working cell.

The elastic linkage can be seen as a combination of the other two. The stations are connected by different transports (belts, live roller conveyors, etc.). Because the single transports are not connected with each other, they can be used as additional buffers. The failure of one machine does not neces- sarily cause the whole production line to be stopped.

Manual Assembly Systems

Description of Manual Assembly Systems and Their Components

Manual assembly systems are often used within the area of fine mechanics and electrical engineering. They are suitable for the assembly of products with a large number of versions or products with high complexity. Human workers are located at the focal point of manual assembly systems. They execute assembly operations by using their manual dexterity, senses, and their intelligence. They are supported by many tools and devices (Lotter 1992). The choice of tools depends on the assembly problem and the specific assembly organization form. The most frequently used forms of organization are, on the one hand, assembly at one separate workstation, and, on the other, the flow assembly with chained assembly stations. Assembly at only one workstation is also an occasional form of assembly orga- nization. The choice of the form of organization depends on the size of the product, the complexity of the product, the difficulty of assembly, and the number of units.

Workstations are used for small products or modules with limited complexity and a small number of units. High version and quantity flexibility are the most important advantages. Also, disturbances affect other workstations to only a small extent.

The components for the basic parts and the assembly parts are the substantial constituents of manual assembly systems. The assembly parts are often supplied in grab containers. The distances to be covered by the workers arms should be short and in the same direction. The intention is to shorten the cycle time and reduce the physical strain on the workers. This can be realized by arranging the grab containers in paternoster or on rotation plates. Further important criteria are glare-free lighting and adapted devices such as footrests or work chairs (Lotter and Schilling 1994).

When assembly at a workstation is impossible for technological or economical reasons, the as- sembly can be carried out with several chained manual assembly stations (Lotter 1992). Manual assembly systems consist of a multiplicity of components, as shown in Figure 20. The stations are chained by double-belt conveyors or transport rollers. The modules rest on carriers with adapted devices for fixing the modules. The carriers form a defined interface between the module and the

superordinate flow of material. For identifying the different versions, the carriers can be characterized. Identification systems separate the different versions and help to transport them to the correct assem- bly stations.

Criteria for the Design of Manual Assembly Systems

There are many guidelines, tables, and computer-aided tools for the ergonomic design of workstations. The methodology of planning and realizing a manual workstation is shown in Figure 21.

The following criteria have to be considered in designing manual assembly systems.

• Design of the workstation (height of the workbench and work chair, dimensions of the footrest, etc.)

• Arrangement of the assembly parts, tools, and devices

• Physical strain on workers (forces, torque, noise)

Ergonomic design requires the adjustment of the work height, chair height, grab distance, and so on to the human dimensions. It should be possible for workers to do the work by sitting or standing (Konold and Reger 1997). To ensure that most of the possible workers work under ergonomic con- ditions, the human dimensions between the 5th and 95th percentiles are considered in designing manual workstations.

One of the most important criteria is the organization of the assembly sequences. The assembly operation should be easy and require little force to be exerted by the worker. Devices like a screw- driver can reduce physical strain on the worker. An optimized arrangement of the assembly parts and components can rationalize the movements. Therefore, the grab room is partitioned into four areas (Lotter 1992):

1. The grab center

2. The extended grab center

3. The one-hand area

4. The extended one-hand area

Assembly should take place in the grab center because both hands can be seen. Part supply should be in the one-hand area.

The forces and torque also have to be considered. Overstressing the worker with too great or too extended physical strain is not permitted. The maximum limit values for static and dynamic load depend on the frequency of the operation, the hold time, and the body attitude. Further correction factors are sex, age, and constitution of the worker.

Today many possible forms of computer-aided tools have been developed for optimization of devices, minimization of forces, time registration, and simplification of movements. They also enable shorter planning time, minimized planning costs, and extended possibilities for optimization of move- ments (Konold and Reger 1997).

Automated Assembly Systems

Automated assembly systems are used mainly for the production of series and mass-produced articles. In the field of indexing machines, a distinction is made between rotary indexing turntables and rectilinear transfer machines. The essential difference between the two systems is the spatial arrange- ment of the individual workstations.

Rotary indexing turntables are characterized by short transport distances. Therefore, high clock speeds are possible. The disadvantage is the restricted number of assembly stations because of the limited place. Rectilinear transfer machines can be equipped with as many assembly stations as needed. However, the realizable cycle time deteriorates through the longer transport distances between the individual stations.

Indexing machines are characterized by a rigid chain of stations. The construction design depends mostly on the complexity of the product to be mounted. The main movements (drives for transfer systems) can be effected from an electrical motor via an adapted ratchet mechanism or cam and lever gears or can be implemented pneumatically and / or hydraulically. Secondary movements (clamping

of parts, etc.) can be carried out mechanically (e.g., via cam and lever gears), electromechanically, or pneumatically. The handling and assembly stations are often driven synchronously over cam disks. If small products are assembled under optimal conditions, an output of up to 90 pieces / min will be possible. However, this presupposes a products design suitable for the automation, vertical joining direction, easy-to-handle components, processes in control, and a small number of product variants. The total availability of the assembly system is influenced by the availability of the individual feeding devices.

The number of stations needed depends on the extent of the single working cycles that have to be carried out (e.g., feeding, joining, processing, testing, adjusting). Therefore, the decision between rotary indexing and rectilinear transfer machines depends not only on the necessary number of work- stations and the required space but also on the entire assembly task. On the one hand, short cycle times and high accuracy during the assembly of small products can be achieved by using typical indexing machines. On the other hand, only the assembly of mass-produced articles and small product variants is possible. Normally the reusability of the individual components is also very small. Con- sequently, these product-specific special-purpose machines can only be amortized heavily.

In order to balance this disadvantage, modern rotary indexing turntables are available on the market. These have higher flexibility, greater reusability, easier reequipping, and higher modularity. The product-independent components (basic unit with the drive, operating panel, protective device, control box, control system, transport unit) should be strictly separated from the product-dependent components (seat, handling and assembly modules). Defined and standardized interfaces are very useful for integrating the product-specific components into the basic system very quickly and with little effort. With modular construction, it is possible to combine the high capability of the indexing machines with economical use of the resources.

Flexible Assembly Systems

Different objectives can be pursued in order to increase the flexibility of assembly processes. One the one hand it may be necessary to handle different workpieces, on the other hand different versions of one product may have to be produced in changing amounts. Another contribution to flexibility is to use the advantages of off-line programming in order to optimize cycle time when the product manufactured is changed.

Assembly of Different Versions of a Product in Changing Amounts

For producing changing amounts of different versions of a product, an arrangement as shown in Figure 24, with a main assembly line to which individual modules from self-sufficient manufacturing

cells are supplied, is very suitable. In order to decouple the module manufacturing from the main assembly line with respect to availability of the different modules, a storage unit is placed between module assembly and main assembly line to keep necessary numbers of presently needed versions of the modules permanently ready. Thus, even lot size 1 can be manufactured and at the same time the system remains ready for use even when one module of a module line fails. The individual manufacturing cells are connected by a uniform workpiece carrier-transport system, whereby the workpiece carriers—which are equipped with memory capacity—function as means of transport, assembly fixtures, and for product identification simultaneously. As the workpiece carriers are equipped with a mobile memory, all necessary manufacturing and test data are affiliated with the product through the whole process. Thus, the product data are preserved, even if the system controller breaks down.

Flexible Handling Equipment

Figure 25 shows an industrial robot with six DOF as an example of handling equipment with great flexibility of possible movement. Especially in the automobile industry, flexible robots are commonly used, and a further increase in investment in automation and process control is predicted. Robots are introduced in automobile assembly to reduce strenuous tasks for human workers, handle liquids, and precisely position complicated shapes. Typical applications include windshield-installation, battery mounting, gasoline pouring, rear seat setting, and tire mounting. Another important application where the flexibility of a robot is necessary is manufacturing automobile bodies with stop welding. In order to reach each point in the workspace with any desired orientation of the welding tongues in the tool center point, at least three main axes and three secondary axes are necessary. Because now each point of the workspace with each orientation can be reached, all kinds of workpieces can be handled and assembled.

CAD-CAM Process Chain

In traditional production processes such as lathing and milling, numerical control is widespread and process chains for CAD-based programming of the systems are commonly used. In contrast, winding

systems are even now often programmed using the traditional teach-in method. Figure 26 shows a winding system for which a CAD-CAM process chain has been developed by the Institute for Man- ufacturing Automation and Production Systems in order to increase flexibility with respect to product changes. The disadvantages of on-line programming are obvious: because complex movements of the wire-guiding nozzle are necessary while the wire is fixed at the solder pins and guided from the pin to the winding space, considerable production downtimes during programming result. Furthermore collisions are always a danger when programming the machine with the traditional teach-in method. The tendency for lower production batches and shorter production cycles combined with high in- vestment necessitates a flexible off-line programming system for assembly cells such as winding systems. Therefore a general CAD-CAM process chain has been developed. The process chain allows the CAD model of the bobbin to be created conveniently by adding or subtracting geometric prim- itives if necessary. Afterwards the path of the wire-guiding nozzle is planned at the CAD workstation, where the user is supported by algorithms in order to avoid collisions with the bobbin or the winding system. Simulating the wire nozzle movements at the CAD system allows visual and automated collision detection to be carried out. If a collision occurs, an automatic algorithm tries to find an alternative route. If no appropriate path can be calculated automatically, the wire nozzle movements can be edited by the user. To this point, the result of this CAD-based planning system is a data file that is independent from the winding machine used. A postprocessor translates this file into a machine-specific data file, which is necessary for the numerical control of the winding system. This fault-free NC data file can be sent from the CAD workstation to the machine control unit via network, and the manufacturing of the new product can start (Feldmann and Wolf, 1996, 1997; Wolf 1997).

2.7. Control and Diagnosis

Despite major improvements in equipment, advanced expertise on assembly processes, and computer- aided planning methods, assembly remains susceptible to failure because of the large variety of parts and shapes being assembled and their corresponding tolerances. Computer-aided diagnosis makes a substantial contribution to increasing the availability of assembly systems. Assuring fail-safe pro- cesses will minimize throughput times and enable quality targets to be achieved.

Diagnosis includes the entire process chain of failure detection, determination of the cause of failure, and proposal and execution of measures for fault recovery. Realizing and operating a high- performance diagnosis system requires comprehensive acquisition of the assembly data. For this purpose, three different sources of data entry in assembly systems can be used. First, machine data from controls can be automatically recorded. Second, for extended information, additional sensors based on various principles can be integrated into the assembly system. Third, the data can be input manually into the system by the operator.

Diagnosis systems can be divided into signal-based, model-based, and knowledge-based systems. A crucial advantage of knowledge-based systems is the simple extendability of the database, for example concerning new failures. Therefore, they are frequently used with assembly systems.

Dependent on the efficiency of the diagnosis system, different hierarchically graded control strat- egies are possible. In the simplest case, the diagnosis system only supplies information about the occurrence and number of a disturbance. The user must determine the cause of failure and execute

the fault recovery on his own. Efficient diagnosis systems offer information to the user about the number and location of the failure. Further, appropriate diagnostic strategies are available that allow computer-aided detection of the cause of failure. The systems also suggest appropriate measures for fault recovery. In addition, diagnosis systems can independently initiate reaction strategies after the automatic determination of the cause of failure. These strategies directly affect the control of the assembly system or the assembly process. Due to the particularly high complexity of this procedure, it is applicable only to simple and frequently occurring errors.