You are on page 1of 40

Int J Flex Manuf Syst (2007) 19:140 DOI 10.

1007/s10696-007-9015-7

A review of some issues and identication of some barriers in the implementation of FMS
Tilak Raj Ravi Shankar Mohammed Suhaib

Published online: 9 March 2007 Springer Science+Business Media, LLC 2007

Abstract Global competition, advancements in technology and ever changing customers demand have made the manufacturing companies to realize the importance of exible manufacturing systems (FMS). These organizations are looking at FMS as a viable alternative to enhance their competitive edge. But, implementation of this universally accepted and challenging technology is not an easy task. A large number of articles have been reviewed and it is found that the existing literature lacks in providing a clear picture about the implementation of FMS. In this paper, work of various researchers has been studied and it is found that it is really a very difcult task for any organization to transform into FMS on the basis of existing research results. A wide gap exists between the proposed approaches/algorithms for the design of different components of FMS and the real-life complexities. Besides describing the gap in various issues related to FMS, some barriers, which inhibit the adaptation and implementation of FMS, have also been identied in this paper.
T. Raj (&) Department of Mechanical Engineering, YMCA Institute of Engineering, Faridabad 121006, India e-mail: tilakraj64@rediffmail.com T. Raj e-mail: Tilakraj45@yahoo.co.in R. Shankar Department of Management Studies, Vishwakarma Bhawan, Indian Institute of Technology Delhi, Shaheed Jeet Singh Marg, New Delhi 110 016, India e-mail: ravi1@dms.iitd.ernet.in M. Suhaib Department of Mechanical Engineering, Faculty of Engineering & Technology, Jamia Millia Islamia, New Delhi, India e-mail: suhaib_me@yahoo.co.in

123

T. Raj et al.

Keywords Flexible manufacturing systems Implementation Transition Adaptation Barriers 1 Introduction During recent years, the manufacturing organizations are facing many unpredictable market changes such as shortened product life cycles, technological advancement, intense pressure from competitors, and ever growing customers expectation for high quality products at a lower cost. The market conditions are becoming more dynamic and more customers driven. The manufacturing performance is no longer driven by the product price; instead other competitive factors such as exibility, quality, and delivery have become equally important (Chan and Swarnkar 2006). Hence, the manufacturers want such type of production technology in which changes can be made with minimum possible time and cost to produce medium to small batches of products. In todays competitive global market, for their survival, the manufacturing companies need to be exible, adaptive, responsive to changes and be able to produce a variety of products in a short time at a lower cost (Nagalingam and Lin 1999). So, manufacturing exibility is the most sought after property of the modern production systems and such type of exibility can be attained through the adaptation and implementation of FMS. A exible manufacturing system is an integrated, computer-controlled complex arrangement of automated material handling devices and numerically controlled (NC) machine tools that can simultaneously process mediumsized volumes of a variety of part types (Stecke 1983). This new production system has been designed in such a way that it has the efciency of a wellbalanced transfer line and ability to provide the exibility of a job shop. FMS is capable of producing a variety of part types and handling exible routing of parts instead of running parts in a straight line through machines (Chen and Ho 2005). Firms adopt the FMS as a means for meeting the mounting requirements of customized production (Kumar et al. 2006). It is also viewed as a technology, which can provide a high potential for productivity improvement in batch production of parts in discrete manufacturing systems. FMS can be very rapidly adjusted to part variety according to the changing market demands. It can respond quickly and smoothly to unexpected changes in the market and recently, it is setting new trends in the manufacturing world. It has been hailed as the solution to challenges facing manufacturing industries worldwide (Parker and Wirth 1999). To continue to meet the challenges of the marketplace, FMS has been acclaimed as the ultimate weapon in boosting productivity and competitiveness (Sarkis 1997). While the above arguments support the adoption of an FMS, its adoption is not an easy job. Diffusion of FMS, in specic and advance technology, is far from easy (Bellasi and Fadlalla 1998). It is generally found that industries, which claim to have adopted the FMS environment, use only a partial FMS. Industries face different types of problems while planning to adopt FMS.

123

Issues and barriers in the implementation of FMS

Many research papers have been published which reveal several types of methodologies for installation and control of FMS. But practically these techniques and methodologies have not been found to be suitable by the industries especially in developing countries. It has been noticed that industries in USA, Japan, Australia, UK, and Europe have responded to FMS to some extent but the response of industries in developing countries is very meager. Industries willing to adopt FMS have to struggle with a variety of problems. The purpose of this paper is to identify and study some of the involved barriers, which are critical during transition to FMS. The organization of the rest of the paper starts with a review of the general procedure for implementation of FMS and shortcomings in this area. Different issues related to implementation of FMS are then listed and discussed. Some important barriers, which inhibit the implementation of FMS in real-life scenario, are identied and listed. A summary of this work and conclusions are discussed in the last section.

2 Implementation procedure for FMS Several authors have presented articles, which describe various steps necessary for the implementation of FMS technology (e.g., Fry and Smith 1989; Primrose and Leonard 1991; Groover 2003; Rezaie and Ostadi 2007). Fry and Smith (1989) have described a detailed procedure for the implementation of FMS in Department of Defence of U.S. They have proposed the following eight distinct stages in the implementation of a exible manufacturing system, i.e., (i) identify the manufacturing requirements of the parts to be produced, (ii) identify and evaluate the alternative technologies, (iii) choose the appropriate technology, (iv) send out requests for proposals, (v) evaluate and select the vendor, (vi) installation of FMS, (vii) system conguration, and (viii) shake down the system. Primrose and Leonard (1991) have suggested that the implementation procedure for exible technology can be divided into three steps, i.e., (i) investment appraisal, (ii) technology selection and (iii) technical installation. Groover (2003) has also discussed FMS planning and implementation issue. He has proposed that this issue should be divided in two phases, i.e., (i) FMS planning and design phase and (ii) FMS operational phase. For the rst phase, he has proposed the consideration of part family, processing requirements, physical characteristics of the work parts, production volume, types of workstations, variations in process routings and FMS layout, material handling system, work-in-process inventory, cutting tools and pallet xtures. For handling the second phase, he has proposed that operational problems must be solved through proper scheduling and dispatching, machine loading, part routing, part grouping, tool management and pallet and xture allocation. Rezaie and Ostadi (2007) have introduced a dynamic programming model to analyze the optimal and phased implementation of exible technology in a manufacturing system.

123

T. Raj et al.

2.1 Discussion on research related to FMS implementation With the study of other articles on the same issue, it is found that no clear-cut procedure for the implementation of FMS has been proposed. Instead, many complex techniques like expert systems, articial intelligence and neural network, etc., have been experimented and suggested for the implementation and integration of different components of FMS, which are so complex that it is often infeasible to apply them in the real-life implementation of FMS. The existing research work is found to be handicapped in fully answering the following questions, which are generally asked by the manufacturing managers in the industry for a possible transition of their traditional manufacturing system to FMS: What types of technologies, standard industrial networks and protocols are available for FMS system integration and how are they managed? What types of software, sensors and other mechatronic components are available for the system integration? What are their related constraints and how are they managed in realistic situations? How will the unmanned operations of FMS be managed, especially in a third shift? How will the problems related to tool management and maintenance be looked after in the third shift? How quickly can one shift to FMS? If a typical traditional company wants to adopt FMS, then, how much time is needed for its transition? How much gain in prots (i.e., economical efciency), exibility, automation and productivity, etc., can be expected over the time horizon? If a decision for FMS installation is taken, then (i) how much time is needed to achieve the breakeven, (ii) when will the prots start pouring in, and (iii) how will the new production system be successful in a long run? How much exibility will be achieved? If a new random part is to be manufactured in FMS on an urgent basis and a sudden design change is required in the product conguration according to market demand, then, how quickly can it be handled in the FMS? How will pallets and xtures handle a variety of parts? Groover (2003) has suggested the use of a number of xtures for different parts. No doubt, different pallets and xtures can be utilized and procured for a number of parts, but (i) how much setting time will be required for different pallets and xtures, (ii) for how much time will the system be idle, (iii) how many parts can a single pallet or xture handle, (iv) if some how exible xtures are designed, then what about their cost and complexity, (v) if exible xtures are not available then, what will be the difference between an FMS and a simple automated production system, and (v) where is the exibility in the system in such a case? Generally, there is a scarcity of vendors supplying all the components of FMS, i.e., there are different manufacturers of computer numerical control (CNC) machines, robots, automatic guided vehicles (AGVs), co-ordinate measuring machines (CMMs), conveyors, automatic storage and retrieval

123

Issues and barriers in the implementation of FMS

system (AS/RS), etc. In developing countries, there is even scarcity of manufacturers of robots and AGVs. When these hardware components are purchased from different vendors, then what is the best way for the integration of these components? What about system maintenance and its up-gradation in a long run? What about training and upgrading the skills of engineers and concerned staff in the area of FMS in a country where these systems have not been introduced?

We believe that these questions are really irritating and these have not been properly addressed by the researchers. A dedicated research work is still pending for nding the solution to the real-life problems related to planning and implementation of FMS. Following sections illustrate more gap areas within different issues related to FMS. 3 Issues related to implementation of FMS Based on the published research papers regarding planning, design and implementation of FMS and discussion with the production managers in industries and with academicians, the following issues are discussed here: Issue Issue Issue Issue Issue Issue Issue regarding regarding regarding regarding regarding regarding regarding loading of parts in FMS scheduling techniques material handling exibility and its measurement machine tools operation, control and maintenance techniques human element and culture

3.1 Issue regarding loading of parts in FMS Loading involves decision about the assignment of work to different machine tools in the manufacturing system for the purpose of machining. Characteristics of different components of FMS and socio-technical environment, where the FMS operates, deeply affect the loading problem. Components of FMS, e.g., types of machine tools, control system of FMS, cutting tools and their handling and storage systems and tool magazine capacity, etc., put different types of constraints which must be considered while deciding and formulating a loading plan for FMS. A loading problem is strongly affected by the types of machine tools and their availability in the FMS. CNC machines are more versatile and precise in performing different operations in comparison to conventional machine tools. Besides this, they can be utilized in unmanned conditions also. So, before formulating a loading problem, one should consider the nature and capacity of all the machine tools involved in FMS. Normally, each machine of the FMS

123

T. Raj et al.

undergoes through a preventive maintenance plan as per the instructions of the machine tool manufacturer. Therefore, while following the schedule of the maintenance team of the plant, some machines or the whole FMS may be unavailable for particular periods of time. This introduces a constraint at the loading level on the capacity of machine. The behavior of an FMS is affected by the characteristics of its control system. At present, the control system of most of the existing FMSs is based on a supervisor, coordinating the behavior of various devices (machines, transport system, etc.) and on a set of computerized numerical controls (CNCs) and programmable logic controllers (PLCs), each of which controls a specic device. The supervisor is the proprietary software running on a central computer. A loading problem can be dened as ... given a set of parts to be produced, set of tools that are needed for processing the parts on a set of machines, and using a set of resources such as material handling systems, pallets and xtures, how should the parts be assigned and tools allocated so that some measure of productivity is optimized...(Kumar et al. 2006). Ever since, the rst article written by Stecke and Solberg (1981) on the productionplanning problem of FMS has been published, a lot of research has been devoted in this area by various researchers. Among the different levels of the production planning hierarchy proposed by Stecke (1983), the loading problem has received considerable attention. It is very important to organize the past research in this area before proceeding to identify the gaps between documented research and the actual real-life requirement of its industrial application. With a detailed-study of literature, it has been found that machine-loading problem is addressed by following approaches: (a) Analytical and Mathematical programming-based methods. The pioneering work in this area described the FMS planning problem into ve subproblems i.e. (i) part selection, (ii) resource allocation, (iii) machine grouping, (iv) production ratio determination and (v) loading; and formulated these as non-linear 01 mixed-integer programming (MIP) problem (Stecke and Solberg (1981); Stecke (1983)). Stecke (1986) improved her earlier work (Stecke 1983) by proposing a new non-linear MIP formulation. Sarin and Chen (1987) developed another mixed integer-programming model to determine routing of parts through machines and to allocate proper cutting tools for achieving overall minimum cost. Lashkari et al. (1987) added the aspects of rexturing and limitations on the number of available tools to the formulation of the operations allocation problem given by Stecke (1983). These formulations assumed that a product-mix problem was already solved and therefore limit the models to be suitable only for dedicated FMS (Mukhopadhaya et al. 1992). Lee and Jung (1989) proposed that the production phase of an FMS must take into account different objectives at the same time. Therefore, they introduced a multiple objective problem and tackled it by means of goal programming. Chen and Askin (1990) also discussed the multi-objective evaluation of FMS loading heuristics. Kumar et al. (1990) tried to nd out a joint solution to the machine grouping and loading problem. Sawik (1990) proposed a hierarchy for the production-planning task based on part type

123

Issues and barriers in the implementation of FMS

selection, part input sequencing, machine loading and operational scheduling. Chen and Chung (1991) discussed the effects of loading and routing decisions on the performance of FMS. Kouvelis (1991) described a two-level hierarchical scheme for determining optimal number of tools of each type required for efcient operation of manufacturing system over a planning horizon on minimum cost basis. At the rst level, a long-term operations assignment to machines is specied and at the second level, the optimal tooling decision is made. Akturk and Avci (1996) proposed a non-linear MIP solution methodology for simultaneously determining the optimum machining conditions and tool allocation to minimize the production cost of a multiple operation case where alternate tools can be used for each operation. Atmani and Lashkari (1998) presented a linear 01 integer-programming model for machine tool assignment and operation allocation in FMS. This model determines an optimal plan by minimizing the total costs of operations, material handling and setups. Gamila and Motavali (2003) discussed the problems of part loading, tool loading and part scheduling. They developed a mathematical model, using a 01 MIP technique, to select machines and assign operations and the required tools to the machines in order to minimize the summation of maximum completion time, material handling time and total processing time. (b) Heuristic-based methods. Stecke and Tallbot (1983) reported that the models proposed by Stecke (1983) may not be practical for real time applications because these models will require a great amount of computational efforts and therefore, large computer time in nding the solutions to real life problems. Hence, they proposed heuristic algorithms for minimizing the part movement and balancing the load on machines of equal size. Shanker and Sriniwasulu (1989) suggested three heuristic-based approaches for solving loading problems in case of random FMS. They proposed these heuristics for the problems having bi-criteria objectives. In the rst heuristic, the main objective is to minimize the workload unbalance and in the other two, the objective is to maximize the throughput. Mukhopadhyay et al. (1992) suggested a heuristic based approach to solve the loading problem in FMS by developing the concept of essentiality ratio for the objectives of maximization throughput time of the system and minimization of the system unbalance. Tiwari et al. (1997) used a pre-determined job-sequencing rule as an input to their proposed heuristic. They developed a Petri-net model for the proposed heuristic for delineating its graphical representation. Some researchers have utilized meta-heuristic approaches like tabu search and simulated annealing (SA) for operation allocation on different machines (Kim and Yano 1993; Tiwari et al. 1997; Tiwari and Vidyarthi 2000; Sarma et al. 2002; Swarnkar and Tiwari 2004). But these approaches sometimes give infeasible solutions due to different restrictions on processing times and non-availability of tool slots. (c) Genetic Algorithm (GA)-based methods. GA-based algorithms have a great potential to solve machine-loading problems of FMS. These are stochastic search techniques that rely on the process of natural solution (Goldberg et al. 1989). Although GA is a global search technique, its practical usefulness lies in its capability to handle the problem constraints. Kumar and

123

T. Raj et al.

Shanker (2000) solved the machine loading and part type selection problem by using GA methodology. Many other researchers have also developed GA-based heuristic for the different objectives such as minimization of throughput time, system unbalance, etc. (Mori and Tseng 1997; Guvenir and Erel 1998; Cormier et al. 1998; Eatson and Mansour 1999; Tiwari and Vidyarthi 2000; Liaw 2000; Cai and Li 2000; Borteldt and Gehring 2001; Kumar et al. 2006). (d) Simulation-based methods. When a system cannot be evaluated analytically to get a single solution then simulation may be a good option. Computer simulations provide almost complete and detailed behavior of the system, especially with respect to comparative analysis of solution variants, by which the best solution can be reached. In simulation-based methods, a symbolic representation of real world is coded into a computer program to evaluate the problems (Kovacs 1992). Stecke and Solberg (1981) carried out a simulation-based study of a dedicated FMS in which ve loading strategies were tested against sixteen dispatching rules. Their study indicated that the choice of the applicable loading strategy depends upon many variables particular to a system. As their study was based on a dedicated type of FMS, the loading strategies were simply the procedures to allocate the operations to a number of same types of machines. Kost and Zdanowicz (2005) have presented the problems involving identication, modeling and simulation of the operations of FMS in their study. They have suggested that in order to carry out the simulation in a correct way, one must know the precise duration time of particular operations, both involving machining and transportation. But at the modeling stage it is almost impossible to know the precise durations, mainly of transport operation. These may be dened only after the start-up of the system and after measuring them in real conditions. The works of both Kovacs (1992) and Kost and Zdanowicz (2005) contradict each other. Kovacs (1992) suggests that modeling is the best way of nding an optimal solution in the design of FMS but at the same time Kost and Zdanowicz (2005) propose that modeling cannot represent a real-life FMS problem. Few simulation-based methods have also been formulated by Mukhopadhyay et al. (1998) and Zolfaghari and Liang (1999). 3.1.1 Discussion on research related to FMS loading Though analytical and mathematical programming-based methods, e.g., integer programming, dynamic programming, branch and bound methods, etc. are robust in applications yet they tend to become impractical when problem size increases. A wide range of multi-objective problems has been addressed by different researchers. However, in such type of problems, some objectives turn out to be conicting in various situations. In real-life industrial applications, number of variables and constraints increases many times and it becomes even impossible to satisfy all the goals at a time. The MIP approach has been found to be computationally infeasible even for deterministic formulations. Moreover, even for moderate-sized FMSs, the

123

Issues and barriers in the implementation of FMS

computation time required for MIP approaches is considerably large. Heuristic approaches are generally based on rules and rely, basically, on empirical data. Hence, main limitation of heuristic approaches is their inability to estimate the results in a new or completely changed environment. Though simulation is an effective technique for dynamic analysis, but for FMS design, it lacks the ability to provide an optimal solution. In many cases, optimization of the loading problems in FMS is heavily inuenced by the spatial conguration of the system. The simulation techniques have another limitation, i.e., more number of iterations is needed to nd the best possible solution and there is always a possibility of returning to some recently found solutions which leads to non-optimized solutions. Most of the researchers have considered the loading problems with the constraints related to cutting tools (e.g., tool magazine capacity, limited number of tool copies, limited tool life, presence of all the tools on the magazine, elimination of infeasible couples like tool and machine or operation and machine) and availability of machining time. Very few researchers included the constraints related to pallets/xture allocation, surface roughness and operation completion time. The consideration of constraints related to cutting tools and tool magazine increases the complexity of problem formulation especially in case of non-linear terms of Stecke (1983). The problems including constraints of pallets or xture allocation have emphasized on the operations of a part while paying no attention to the types of machines involved in the operations. But the consideration of the type of machine is an important issue as it can inuence the major objectives in a big way; hence it needs proper attention of researchers. Another problem in some of the previous works related to part selection and machine loading is that machining speed for each job-tool, machine combination has been treated as a pre-determined constant. However, it is unrealistic to nd a machining speed for an unknown job-tool-machine combination. It is evident from the above discussion that all the four types of approaches are handicapped in producing an optimal solution to loading problems of FMS. Though good attention has been paid to machine loading problems, but, most of these studies focus on modeling the problems and emphasis has been put on the optimization of certain performance measures rather than on understanding the problem and nding the interaction between the different factors in the system. Researchers have proposed many methodologies for solving the loading problems of FMS but in the end, majority of them have shown the limitations of their own work in the real-life situations, e.g., Stecke(1983) and Atmani and Lashkari (1998). Stecke (1986) has used a MIP approach for solving the machine loading and grouping problems of FMS and presented the solution with the help of large computer (CDC 6600) by using a standard mixed integer code but she concluded her work with the statement: ...real-time FMS control requires very quick solutions to these planning problems. Many companies do not have a mixed integer code available. It is clear that the hierarchical

123

10

T. Raj et al.

approach was developed for a particular case-company and could not be applied to a general case. She has further suggested the development of efcient heuristic algorithms for providing good solutions to machine loading and grouping problems of FMS. Atmani and Lashkari (1998) conclude their work like this: The real-life, industrial application of this model will invariably result in large numbers of variables and constraints which may over-run the capabilities of the integer programming software currently available. Under these conditions the use of heuristic solutions procedures is recommended. Most of the researchers have used the objectives of minimizing the total machining time, machining cost, inventory cost, throughput time, minimization of makespan and frequency of tool movement for the loading and scheduling problems. But they have neglected the set-up cost, machining cost and the movement of parts between the machines. Use of these criteria, generally, creates imbalance in the machine workload and reduces machine utilization. Due to unbalanced workloads and unmatched production resources in FMS, slack time is inevitable and it is not negotiable. But this slack time has not been given due consideration in machine loading problems. Grieco et al. (2001) presented a wide review of different approaches to the FMS loading problems and in the end of their work, they commented like this: Analysis of the literature on loading shows that very few articles deal with the same loading problem. It results in the fragmentation of research effort. This situation is made even more critical by the fact that few articles present detailed real or realistic test cases, which can be adopted as test beds for subsequent research work. A common complaint of industrial managers is that theoretical approaches to FMS loading problems fall short of realism. The real life, industrial applications of above models will invariably result in a large number of variables and constraints, which may put a question mark on the real adaptation of these techniques, and it is found that following questions still remain unanswered: 1. 2. 3. 4. 5. What types of characteristics are desired in production planning hierarchy for deciding the loading module? What is the integration of loading module at different levels of production planning hierarchy? What is the best way to meet the due dates? What is the effect of characteristics of plant where the FMS operates on the loading model? How is the proposed loading model affected by (i) the number of shifts and their characteristics, (ii) ow of tools in the tool room, (iii) policies adopted for preventive maintenance, and (iv) arrival of new parts? What is the effect of tool life management, especially in case of unmanned shifts, for nding the full production capacity of FMS?

6.

From above discussion, it becomes clear that some attention is required to make the proposed loading models applicable to real-case situations. This important area needs further exploration and can be addressed in future research work.

123

Issues and barriers in the implementation of FMS

11

3.2 Issue regarding scheduling of parts in FMS Scheduling is the allocation of resources over time to perform tasks (Baker 1974). This is a well-known and very general denition of scheduling. The purpose of scheduling is to determine when to process which job and by which resource so that production constraints are satised and production objectives are met (Yu et al. 2003). In view of scheduling theory, a general FMS may be considered to be job shop with parallel machines and additional limited resources. There are, however, substantial differences between a conventional job shop and an FMS where one should additionally consider such FMS features as alternate routings, pallet and xture limitations and nite-in-process buffers (Sawik 1990). In classical scheduling environments, only one resource type is considered (the machine) and the basic scheduling problem reduces to deciding the allocation of this resource type to process the jobs. The FMS environment is quite different in many respects from the classical machine shop. The main difference of scheduling between a conventional production system and a exible manufacturing system is caused by exibility and automation in a modern manufacturing environment. In an FMS, one machine can process various different parts and one operation can be processed in various different machines. These exibilities increase signicantly the feasible solution space and complicate the FMS scheduling (Bing 1998). The increasing automation and complexity of FMS requires that an effective and improved scheduling technique should be developed for it. The successful development of realistic scheduling techniques for FMS could have a signicant effect on manufacturing industry. Liu and Maccarthy (1996) have identied and discussed ve major factors inuencing the FMS scheduling problems such as (i) system type such as a single exible machine (SFM), a exible manufacturing cell (FMC), a multi-machine exible manufacturing system (MMFMS) and a multi-cell exible manufacturing system (MCFMS), (ii) capacity constraints, (iii) job characteristics, (iv) production management environment and (v) scheduling criteria. 3.2.1 Scheduling techniques in FMS A review of literature indicates that there are hundreds of articles on scheduling problems proposed by different authors at different times. On the basis of study of some of these articles, the approaches used for scheduling problems can be divided in four categories, as follows: (a) Analytical and Mathematical programming-based methods. Stecke (1983) applied mathematical programming techniques to solve various FMS planning and machine loading problems and in the same article, she suggested that large problems couldnt be effectively handled with the proposed methodology. Chang et al. (1984) proposed a two-phase approximate methodology for quasi-real-time scheduling of FMS. In this methodology, phase-I utilizes the reduced enumeration to generate a number of schedules and phase-II is utilized in real-time to select the optimal schedule for a particular schedule.

123

12

T. Raj et al.

Although authors have proved that this approach performs better than the traditional job shop scheduling heuristics yet their work does not comment on the computational efforts required for the formulation of integer programming model for second phase. Chan (1999) has studied the effectiveness of operational control rules in scheduling an FMS with the use of computer simulation. He has proposed three control rules: (1) dynamic alternative routing (DAR), (2) planned alternative routing (PAR) and (3) no alternative routing (NAR) for the selection of alternative routing for each part and applied these rules to FMS with either local buffers with innite capacity or without local buffers. He also examined the inuences of two types of loading stations, i.e., universal loading station and dedicated loading station on the performance of FMS. A hypothetical model of FMS has been considered with some basic assumptions like (i) demand for each part is known and (ii) the traveling time of AGVs between stations is insignicant, when compared to processing times. The results of this study are insignicant in real life situations where part demand is variable and traveling time of AGVs is highly signicant. Sawik (1990) has formulated the production scheduling of an FMS as a multi-level integer program. He has proposed a hierarchical decision structure that includes the problems of (1) part type selection, (2) machine loading, (3) part-input sequencing, and (4) operation scheduling. The author has presented the integer programming formulation for all these problems as well as algorithm for part selection and operation scheduling. Though the proposed algorithm provides a period-by-period production schedule, which can be used on-line as well as off-line, yet some important characteristics of FMS like transportation time and capacity, buffer limitations, etc. have not been included in the proposed algorithm, which makes it non-realistic for practical purposes. Some other analytical approaches have also been formulated by researchers (Nasir and Elsayad 1990; Ghosh and Gaimon 1992; Sodhi et al. 1994; Dolinska and Besant 1995; Jinyan et al. 1995; Rajamani and Adil 1996; Das and Nagendra 1997; Guerrero et al. 1999; Chan 1999; Saygin and Kilic 1999). (b) Heuristic-oriented methods. A large number of mathematical models have been developed for solving the scheduling problems in FMS. But, main limitation of these models is that large amount of calculations are needed for nding the feasible solution. This makes the mathematical approaches infeasible in realistic conditions. Hence, in most of the cases, these formulations have been used as a basis for developing the heuristics for FMS scheduling. Abdin (1986) developed a heuristic for the scheduling problem in FMS with alternate machines selection for an operation. Lee and DiCesare (1992) developed a heuristic based Petri-Net model, which could handle complexities and uncertainties in FMS environment such as rescheduling and routing exibilities. Zhao and Wu (2001) have suggested a genetic algorithm (GA) based heuristic to solve the job-sequencing problem of FMS with multiple routes. This means that parts, of all types, can be processed through

123

Issues and barriers in the implementation of FMS

13

alternative routings and there can be several machines for each machine type. For solving these general scheduling problems, they have proposed a genetic algorithm approach and have introduced the concepts of virtual and real operations. Concepts of exible-routing scheduling problem, which involve routing selection, machine selection and processing sequence selection, are introduced. Chan et al. (2003) have supported this fact that FMSs are more sensitive to system disturbances and they require an immediate response to changes in the system states. This can be achieved by real time dynamic scheduling. For this purpose, they developed an algorithm, which is based on pre-emptive approach. They proved that systems performance when using dynamic scheduling could be improved by changing the corresponding dispatching rule at a correct frequency. But, main drawback of their proposed algorithm is that they have considered only machine dispatching rules and other important operation rules such as AGV dispatching, machine selection rules, tool selection rules, etc. have not been taken into account. These rules are very important as far as the dynamic nature of FMS is concerned. Many other researchers have also developed heuristic-based approaches for FMS scheduling such as Widmer (1991), Fanti et al.(1992), Van Laarhoven et al.(1992), Barness and Chambers (1995), Sakawa et al. (1996), He et al. (1996), Bryne and Chutima (1997), Hertz et al. (1998), Ecker and Gupta (2005) and Low et al. (2006). (c) Expert system-based methods. Shaw and Whinston (1989) have described a scheduling approach, which employs a knowledge-based system to carry out the non-linear planning method developed in articial intelligence. A prototype of this scheduling system has been applied to solve the scheduling problem in FMS. This scheduling method is characterized by its knowledgebased organization, symbolic representation, state-space inference, and its ability for dynamic scheduling and plan revision. It provides a foundation for integrating intelligent planning, scheduling and machine learning in FMSs. Chandra and Talvage (1991) and Naidu and Vishwanandham (1992) have developed and proposed intelligent systems consisting of a rule-base for alternate machine selection. Dadone et al. (1997) have proposed a fuzzy logic system (FLS) and fuzzy multiple attribute decision making (MADM) techniques for scheduling of FMS. It is the authors opinion that the importance of common sense and human expertise in scheduling, together with fuzzy logic ability to mimic human reasoning, along with the ease of dealing linguistic variables makes it a very suitable and powerful tool for scheduling in an FMS. But in the proposed model, tool management and failure of workstations and/ or transport systems have not been considered which affect the industrial feasibility of fuzzy scheduler. Sharafali et al. (2004) have considered the problem of production scheduling in an FMS with stochastic demand. They have proposed a model similar to polling model with the objective of minimizing the total average cost. A polling model is a multiple channel queuing system in which queues are served in a cyclic or some other predetermined order by a single server. Based on their ndings with this model, they have proposed that in FMS a part family

123

14

T. Raj et al.

can be mixed with other part family and wastage of production time can be reduced. Sallez et al. (2004) have proposed a product-based heterarchical (PBH) approach for the dynamic scheduling of FMS. Authors have shown that specic advantage is its robustness and adaptability to its environment. However, a serious drawback of the proposed approach is that it becomes infeasible in case all the entities of the system are independent. (d) Simulation-based methods. Chang et al. (1984) examined various dispatching rules for dynamic scheduling in a exible manufacturing system with the use of simulation programs. The main problem with the simulation-based methods is that it does not provide any solutions but it only tests the resulting performance so that the best heuristic rule for a given system can be selected. Montazeri and Van Wassenhove (1990) also reviewed the performance of different dispatching rules for simulation model of an FMS. They concluded that dispatching rules have a good impact on the different performance criteria, e.g., makespan, machine utilization and buffer utilization. But, general results are not available as the performance of scheduling rules not only depends upon the selected criteria but also on the conguration of the production system. Harmonosky and Robohn (1995) identied some factors, which can inuence the implementation time of simulation as a real-time control tool. Weintraub et al. (1999) proposed simulation-based scheduling decisions in minimizing the maximum lateness of large-scale system. Jain and Foley (2002) conducted a simulation study aimed at understanding the impact of an interruption on a schedule in order to build a knowledge base for intelligent selection of a response from a set of alternatives and identied signicant major factors and their interactions. They have reported that the impact of interruption is more disruptive in typical FMS than in conventional ow shop environments and the impact of an interruption at a bottleneck machine is much higher than an interruption at a non-bottleneck machine. It is found that most of the work related to scheduling problem of FMS has been dedicated to static conditions which does not take into account the unexpected problems such as arrival of rush orders, machine breakdowns, non-availability of tools, AGV problems, etc. These problems are very much a part of the practical operation of FMS and render the existing schedule infeasible. Shi-jin et al. (2006) have also supported this argument and have proposed a ltered-beam-search (FBS)-based heuristic algorithm to solve the dynamic scheduling problem in a large job shop FMS environment with realistic disturbances. They have evaluated the performance of the proposed algorithm with only computational experiments. At the end of their proposed study, they are doubtful about the effectiveness and robustness of the scheduling system with this algorithm and its implementation in real life cases. 3.2.2 Discussion on research related to FMS scheduling The mathematical programming models provide an understanding of the problem structure and the factors contributing to the complexity and growth in problem size. They may be used in testing and comparing new solution

123

Issues and barriers in the implementation of FMS

15

methods. But, the main limitation of these models is due to heavy computations requirement for nding a solution for the target problems. This computational complexity makes these methodologies impractical for real-time control in most of the applications. Another problem with the existing methodologies based on analytical mathematical programming is that they are generally off-line techniques and their ability is limited in handling the reallife manufacturing environment. It is because of their failure to handle the realistic and dynamic behavior of FMS and lack of providing the solutions within the reasonable time period. In addition to above cited limitations, another shortcoming of analytical approaches is that, the constraints on the storage of work-in-process inventory and on the material handling devices such as AGVs, robots, etc., have not been included in most of these models. Both of these constraints are very critical for the real-time implementation and design of FMS scheduling techniques. Several heuristic methods based on branch and bound, genetic algorithms, simulated annealing, tabu search and Langrangean relaxation approaches have been developed for FMS scheduling problems. But main drawback of such methods is that they require an initial solution and the quality of nal solution depends upon the initial solution. Smith et al. (1993) have stated, unfortunately, increased importance of scheduling FMS does not impact the complexity of the scheduling problem. Also, limited inventory buffers, faster machine setup times, more alternate routings and the existence of tooling constraints all serve to complicate the FMS scheduling further. Scheduling in an FMS environment is more complex and difcult than in conventional manufacturing environment. This is primarily due to the usage of versatile machines, which are capable of performing many different operations resulting in many alternative routes for part types. Another problem which has been noticed in the existing literature is that, in most of the scheduling problem formulations, set-up time has been assumed to be negligible or a very small portion of processing time. But, in realistic conditions, set-up time reduction is the main expectation from the exible manufacturing system and cannot be neglected as such. There is an utmost need that research communities must, now, focus on developing some practical methods, which quickly solve the real-world scheduling problems because no perfect solution has yet been found for these problems due to the complexity of FMS environment. There are some important factors, which make the scheduling problem very complex such as: There are varieties of part families to be produced in FMS. At any time of production, a mix-part family may be loaded for production in FMS. A part may have many attributes. Different part types share common tools (Sharafali et al. 2004). An optimal solution requires simultaneously partitioning the production order into the minimal number of batches.

123

16

T. Raj et al.

Basic purpose of FMS is to produce a variety of products, which require congurable or similar type of tool changing devices as well as transportation and storing devices. But in real life cases, production managers may try to avoid this situation.

Form the above discussion it can be concluded that though a large number of research papers have appeared in the eld of FMS regarding its scheduling problems but most of them are based on either static conditions or their solutions are referred to particular cases only. Researchers have not adequately considered the real-life scenario, which is dynamic in nature and requires such type of scheduling methodologies, which are robust and effective. Following gaps, still, exist between the reported research and the real-life scheduling of FMS: 1. 2. In most of the studies, the impact of product variety has not been adequately studied for real-life applications. The impact of uncertainties such as machine breakdown, shortage of tools, shortage of materials, sudden rush of orders and quality failures have not been given due consideration in the preparation of schedulers. Constraints related to material handling devices like AGVs, i.e., their load carrying capacity, exibility in handling a variety of parts and routing exibility, etc., have not been paid proper attention in most of the studies. In most of the cases, there is no integration between process planning and production planning which leads to suboptimal solutions of the scheduling problems.

3.

4.

3.3 Issue regarding material handling in FMS An FMS consists of a set of machine tools and a material handling system (MHS) linked by a network of computers controlling and interfacing with them. Unlike the traditional MHS, where a human element is involved in the transportation of materials between various locations, human intervention is almost non-existent in FMS (Mahadevan and Narendran 1990). This has been made possible by developments in guided-vehicle technology and computercontrolled MHSs. Among the various equipments employed for material handling in FMS, automated guided vehicles (AGVs) have been the most popular choice. An automated guided vehicle system (AGVS) features battery powered, driverless vehicle moving on a guided path layout. No human intervention is needed for the guidance, steering or control of such vehicles. An off-board controller is used to send dispatcher commands for the identication of the load, destination of the load and other instructions related to loading and unloading of the load. It has programming capabilities for path selection and can be recongured easily to accommodate changes in production volume, product mix, product routing and equipment interfacing requirements. These exibilities are essentially required in an FMS environment (Rajotia et al. 1998).

123

Issues and barriers in the implementation of FMS

17

3.3.1 Design issues for AGVs The power and smaller size of microprocessors have given guided vehicles the capability of operating autonomously, even in complex transport applications. The design of a material handling system using AGVs must address the following issues (Mahadevan and Narendran 1990; Vis 2006; Shankar and Vrat 1999): 1. 2. 3. 4. 5. 6. 7. 8. The number of vehicles required. The layout of the AGV tracks in the shop. Decision regarding provision of control zones and type, number and capacity of buffer for the vehicles. Vehicle dispatching rules. Trafc management: prediction and avoidance of collisions and deadlocks. Positioning of idle vehicles. Battery management. Failure management.

Several authors have contributed in the area of design issues of AGVs. Some of the previous researches in this area are discussed here. Maxwell and Muckstadt (1982) studied the problem of the design of an AGV system. They did pioneering work in analytical modeling of operational features of an AGV system. In an environment comprising primarily of assembly operations for nished products, they proposed a time-independent model to estimate the minimum number of vehicles required to support the material handling needs. A standard transportation problem was formulated which assigned empty vehicle trips between various stations to minimize the total empty vehicle travel time. Egbelu and Tanchoco (1984) examined the effects of some heuristic rules for dispatching AGVs in a job shop environment. Their study considered the case where the volume of material ow was large. Later on, Egbelu and Tanchoco (1986) analyzed the possibility of bi-directional ow in guided paths. Based on a simulation analysis, they concluded that bi-directional ow increases the throughput, especially in systems that require only a small number of vehicles. Bozer and Srinivasan (1989) suggested a tandem conguration for reducing software and control complexity of AGV-based MHS. Ozden (1988) conducted a simulation study of a small FMS to consider simultaneously design parameters such as trafc pattern, number of AGVs, the load carrying capacity of each AGV and queue capacity of the machines. Malmborg (1990) suggested a scheme for computing travel time of empty vehicle, which is opposite to the approach taken by Maxwell and Muckstadt (1982). Important differences arise on two countsnumber of empty trips and travel time of each such trip. The frequency of empty trip was based on total number of loads delivered at or picked up form each station rather than net ows. Further, instead of minimizing empty travel time, Malmborgs scheme maximized it. It meant that each vehicle, after being unloaded at a delivery

123

18

T. Raj et al.

station, was routed to the farthest station. The total empty vehicle travel time thus obtained was considered by Malmborg (1990) as an upper bound solution as opposed to the lower bound in the Maxwells model. Further, Malmborg (1990) argued that actual empty travel would be a weighted average of these two bounds. The weighing factor is a function of vehicle dispatching strategy. 3.3.2 Scheduling and routing algorithms for AGVs Scheduling of AGVs aims to dispatch a set of AGVs to pick up the jobs from a centralized load/unload station and dispatch them to different workstations according to some priority. Certain goals are kept in mind while deciding the scheduling of AGVs. These goals are related to resources such as minimizing the number of vehicles while maintaining the system throughput or minimizing the total travel time of all vehicles (Akturk and Yilmaz 1996). Routing of AGVs involves nding a suitable route, e.g., shortest distance path, shortest time path or minimal energy path for every vehicle from its origin to its destination based on current situation (Daniel 1988). For solving scheduling and routing problems of AGVs, many algorithms have been suggested in the literature, e.g., Egbelu and Tanchoco (1986), Kolen et al. (1987), Bartholdi and Platzman (1989), Kaspi and Tanchoco (1990), Kim and Tanchoco (1991), Tanchoco and Sinreich (1992), Lin and Dgen (1994), Langevin et al. (1994, 1996), Taghaboni and Tanchoco (1995), Akturk and Yilmaz (1996), Klien and Kim (1996), Lee et al. (1996), Rajotia et al. (1998), Barad and Sinriech (1998) and Qiu and Hsu (2001). Qiu et al. (2002) conducted a detailed survey of such algorithms and suggested that the existing work can be classied into four general categories: Algorithms for general path topology Path layout optimization Algorithms for specic path topologies Dedicated scheduling algorithms.

Works in the rst category usually treat the problem as a graph theory problem and use approaches such as shortest path algorithm to get an optimal route. Optimization techniques such as integer programming, have been used for path network design in which the routing control is generally simple. Algorithms for path networks are restricted to specic topologies such as single-loop, multi-loops, meshes, etc., and these are developed to decide the route and control of AGVs. Contributions made by different authors in the development of AGV scheduling algorithms have been discussed by Qiu et al. (2002). 3.3.3 Discussion on research related to FMS material handling Several approaches have been applied to solve the design, routing and scheduling problems of AGVs. These approaches range from simple logic to

123

Issues and barriers in the implementation of FMS

19

typical mathematical techniques such as integer programming formulations, branch and bound techniques, heuristic and simulation. But the approaches discussed in the literature are quite diverse in nature and it is very difcult to nd out the optimal solution in real-life cases. Hoff and Sarker (1998) also noted this difculty and commented, the AGV system is still a topic for research, as guide path, pickup and delivery station location, idle position and dispatching rules for the vehicles are being developed and improved. The proposed algorithms and heuristics are all rened and some are more computationally difcult than others. Besides these difculties in the use of proper techniques in preparation of different algorithms, following shortcomings may be noted in the reported research work: 1. Most of the analytical models, which have been developed for AGV system, deal with small number of AGVs and workstations. As the number of AGVs increase, they present non-optimal results. Most papers concerning the design of AGV system ignore or simplify the empty vehicle trafc. Very few authors consider the multiple design problems of AGV system. Mutual relationship between AGVs and other material handling devices like AS/RS has not been adequately explored. Generally, the layout problems and control problems of AGV system have been treated separately despite the fact that these issues are highly inter-related.

2. 3. 4. 5.

A large number of techniques have been suggested and developed for designing and control of AGVs in different situations by various researchers. However, if these techniques are carefully examined then it is observed that most of these methodologies have been proposed by academic institutions or research laboratories and their actual feasibility in an industrial setting is quite limited. Further research is still needed to nd the industrial feasibility of such techniques and following questions are to be answered: What is the cost of AGV system in comparison to human labor in the context of FMS? What is the exibility of AGV system in comparison to human controlled equipments? What is the feasibility of implementation of AGV system in different work environments? What is the exibility range of AGV system and how can it handle different varieties of parts? What type of industry-based optimization techniques is available for AGV system path layout? What type of industry-based models is available for calculating the number of vehicles for different work situations, dispatching rules, etc.?

123

20

T. Raj et al.

Countries like India, where labor is very cheap, hesitate to adopt these AGVs because of their costly and complex controlling techniques. Production managers generally argue that when human labor is cheaply and conveniently available then, why to go for such complicated material handling systems. Human labor is highly exible and does not require special control algorithms. In such an environment, automated material handling systems, present a big barrier in the implementation of a complete FMS. 3.4 Issue regarding exibility and its measurement Flexibility is one of the critical dimensions in enhancing the competitiveness of organizations. Manufacturers of discrete parts face increasing demands for small-to-medium sized lots of customized products, requiring a production process, which can provide exibility as well as economy (Kouvelis 1991). Technological developments and market demands are forcing organizations to become more exible. As a result, process exibility is fast becoming a major priority for many organizations. Flexibility is one of the most sought-after properties in modern manufacturing systems (Shewchuk and Moodie 1998). A great deal of research has already taken place in dening various types of exibilities in manufacturing environment. Different authors have dened exibility in their own ways. According to Chen and Chung (1996), exibility refers to the ability of the manufacturing system to respond quickly to changes in part demand and part mix. Das (1996) has dened it as the ability of a system or facility to adjust to changes in its internal or external environment. Tsubone and Horikawa (1999) have suggested that exibility can be dened as the ability of a system to adapt quickly to any changes in relevant areas such as product, process and workload or machine failure. Several researchers have classied exibility under different categories. Buzacott (1982) has classied it in two categories, i.e., job exibility and machine exibility. Park and Son (1988) and Son and Park (1990) have identied four types of exibilityprocess, product, demand and equipment exibility. Browne et al. (1984) have proposed eight types of exibilities including machine exibility, routing and expansion, etc., Azzone and Berteles (1989) have suggested six types of exibility: process, product, production, routing, expansion and volume exibility. Sethi and Sethi (1990) have identied eleven types of exibility: product, process, program, production, volume, routing, expansion, operation, machine, material handling and market exibility. The identied types of exibility typically refer to different elements and attributes of a production facility such as machine, product, processing, operation, routing, capacity, expansion, design and system (Das 1996). Other researchers have studied the relationship between exibility and productivity (Gustavsson 1984; Buzacott and Madelbaum 1985). While it is helpful to understand the fundamental concept of exibility, it is also essential to develop measurement systems for exibility so that rms can better benet from utilizing these concepts of exibility. A exibility measure is a formula, algorithm, methodology or the like, for generating a value for a

123

Issues and barriers in the implementation of FMS

21

given exibility type under given condition (Shewchuk and Moodie 1998). Chen and Chung (1996) have suggested that without proper development of exibility measurements and consequently, a better understanding of the relationship among different exibility levels and system performance, it can be quite difcult for rms to position manufacturing as one of their competitive priorities. They have discussed alternative measures for the assessment of machine exibility and routing exibility and have illustrated the capability and applicability of these measures with examples. Dixon (1992) has applied factor analysis technique for measuring three dimensions of exibility, i.e., mix, new product and modication. Das (1996) has provided a theoretical basis for measuring the exibility of manufacturing systems. He has introduced a new concept of multiple levels of measures for each type of exibility. Tsubone and Horikawa (1999) have compared the machine exibility and routing exibility in terms of manufacturing performance in different shop environments. They conducted a simulation-based investigation to analyze the impact of these types of exibility on the average ow time of parts under various conditions of job ow pattern. Hutchinson and Sinha (1989) have used a decision theoretic approach in developing a model to assist a decision maker in evaluating the cost-exibility tradeoff. They have also examined the sensitivity of the model with respect to several parameters of interest. They have tried to quantify the exibility in monetary terms. Similar attempts have been made by other researchers also like Son and Park (1987), Son (1991), Suresh (1990), Azzone and Bertele (1989), Troxle and Blank (1989), Venk (1990), Zahir (1991), Stam and Kuula (1991), Suresh (1991) and Demmel and Askin (1992). 3.4.1 Discussion on research related to FMS exibility From the above discussion, it is clear that a lot of efforts have been put by different authors in dening and measuring exibility. But in spite of this, following problems are generally faced in industries. First problem is with the denition of exibility. Though many denitions have been used in the literature, but there is no agreement on one general denition of exibility. This creates a lot of confusion in the minds of production managers in real-life situations. As Shewchuk and Moodie (1998) have pointed out, there is no general agreement on how to dene exibility. This is due to multidimensional nature of exibility and various views of exibility that result: exibility has been viewed and studied as a physical property, an attribute of decision making, an economic indicator and a strategic tool. Their viewpoint is strongly supported by Beskese et al. (2004), despite the wide interest, exibility remains to be poorly understood in theory and poorly utilized in practice. One of the reasons for this is the lack of general agreement on how to dene exibility. The second problem in the domain of exibility issue is its measurement and quantication. No clear-cut methodology is available to practicing managers for measuring the exibility of a system. Classication of exibility into various types has further complicated this situation. For example, product

123

22

T. Raj et al.

exibility, operation exibility, process exibility, machine exibility, production exibility, manufacturing exibility are terms, which look alike and are very difcult to be differentiated in practical industrial environment. These terms have only helped in enhancing the confusion in the minds of manufacturing managers and no other big objectives have been fullled with these confusing terms. Owing to its nature, any measure of exibility has to be user or situation specic. Several approaches, e.g., multi-dimensional approach, decision-theoretic approach, Petri-nets approach and information theoretic approaches, etc., have been suggested in the literature. But, their use in real-life cases is yet to be explored. Most of the literature only presents a list of proposed measures but very rarely mentions the objectives and criteria against which a particular measure can be judged. This explains the existence of many measurement schemes according to different types of exibilities. Researchers have not been able to develop and present the universally accepted techniques over which the manufacturing people can rely. This poor understanding about manufacturing exibility is, thus, inhibiting the progress towards the utilization of exibility concepts in industry and impeding manufacturing managers from evaluating and changing the exibility of their operations. Other shortcomings observed in relation to exibility and its measurements are as follows: It is difcult to understand the classication or dimensions of exibility from the literature. It has been realized that there is still a large gap between the benets promised by nancial evaluation tools and the benets realized in real life cases. It is difcult to measure the manufacturing exibility. Practically, managers in manufacturing industries face difculty in calculation of machine and routing exibilities in the absence of a clear-cut model for the same. Another problem in designing of FMS is related to how to reach to a compromise between exibility and throughput of the system. Whenever discussion about transition to FMS takes place, the industrial managers put-up such type of blunt questions as (i) what is exibility of the manufacturing system, (ii) is there any exact denition and nature of it, (iii) how exactly, can it be measured, (iv) is there any universally accepted formula for its quantication, and (v) how can it be implemented? There are certain constraints regarding different resources in FMS which inhibit to achieve the desired level of exibility in the system, e.g., (i) machine tools have limited range to support a limited category of parts, (ii) robots are designed or selected for a denite range of parts, (iii) AGVS have well dened load carrying capacity, and they can be dispatched and routed through a xed path and (iv) xtures and other work-holding devices are highly rigid.

In addition to these, constraints regarding control methodology also act as the main barrier in making the production system a really exible one. Production managers often feel handicapped in differentiating process

123

Issues and barriers in the implementation of FMS

23

exibility and product exibility. In such an environment, there is an utmost need to develop models for the measurement and quantication of exibility based on simple inputs such as variety of parts and availability of different resources. 3.5 Issue regarding machine tools in FMS Machine tools are the main component of exible manufacturing systems. Various types of machine tools like CNC machines and machining heads in SPMs, etc., are used in FMS. It has been reported in the literature that CNC machines are the basic component of FMS because of their softwired nature. But, Overmars and Toncich (1994) have pointed out some limitations of CNCs (e.g., longer cycle time; longer chip-to-chip tool change time; closed architecture of CNC machines, etc.) in the integrated FMS environment. Besides these drawbacks, associated with the usage of CNC machines, major factor is their high cost. Another type of machine tools used in FMS is special purpose machines (SPMs). These machines are designed and selected according to some specic product requirements, but whenever some design changes are congured in the products, then, such machines are not found to be practically enough exible to cope with the big changes in product design. This situation puts up a big barrier in the implementation of FMS. Hence, there is a need to make these SPMs sufciently exible. One way to do so is to introduce recongurability in the machining heads and use of CNC machining heads instead of conventional machining heads. In this way, conventional SPMs would be called exi-SPMs offering the desired exibility in the manufacturing system. Cutting tools and various issues related to them also affect the performance of an FMS. Cutting tools of different materials behave in different ways and they have different tool lives. For example, cemented carbide tools can be utilized at higher cutting speeds than high-speed steels where as ceramics and CBN tools can be utilized at still higher cutting speeds than cemented carbides. Tool life of these tools is also different. When these tools are worn out, they are sent to grinding section for resharpening. Grinding section may or may not be free at that time to serve them. These cutting tools may have to be in queue for getting resharpened. So, this factor puts a great constraint on the decision of loading and scheduling problems. Gray et al. (1993) and Veeramaani et al. (1992) give extensive surveys on the tool management issues of automated manufacturing systems and emphasize that the lack of tooling consideration has resulted in the poor performance of these systems. Kouvelis (1991) has identied cutting tool utilization as an important parameter for the overall system performance. One should also consider the nature and number of pallets and xtures for mounting the work-pieces on them. Whenever a new product is added, the production facility experiences a cost for setting up new cutting tools, xtures,

123

24

T. Raj et al.

reprogramming of CNC machines, reconguring the machining heads in RMS (recongurable manufacturing systems) and down time cost of the production facility for new setup (Das 1996). Minimum set-up cost and time are the main expectations from FMS. But, practically following limitations or constraints are generally experienced on shop oor conditions: New cutting tools (for new parts) may not be readily available. FMS can produce parts that are large in variety and small batches. However, with the increase in part variety, the number of tool types and tool cost will also increase. Tool cost contributes to about 25%30% of the total xed cost and variable cost in a FMS (Keung et al. 2003). Fixtures represent the interface between the FMS and the parts to be machined. But, generally, parts differ in shape, size, material and nature of required operations and the xtures may not accommodate this variety of parts. There may be limits to the reconguration of machining heads for handling the new products in special purpose machines (SPMs). Set-up time for the new part may be high because of various constraints related to cutting tools, machine tools and material handling devices. This puts a question mark on the exibility to be achieved through FMS in real sense. There may be difculty in reprogramming of the system. Another problem is to change the layout of machines within the manufacturing system. Atmani and Lashkari (1998) have rightly pointed out that though in recent years, the concept of FMS has emerged as a viable answer to the problems of exibility and efciency, but proper planning is a key condition to FMS success in dealing with problems of poor utilization of equipments and tools, and costly and frequent setups. Thus, the difcult task facing the production planner is the optimal selection of machines and tools. Tool magazine or tool turret capacity also inuences the total number of tools to be used in the machining process. This puts constraints on the set of operations that can be assigned to a machine during a given period. Machine tool builders normally try to equip their machining centers with large tool magazines (e.g., 5060 slots) in order to reduce the capacity constraint. However, large tool magazines result in high seek times and sometimes greater than the time required to perform an operation (Grieco et al. 2001).

3.6 Issue regarding operation, control and maintenance techniques for FMS Although exible manufacturing systems have increased the exibility and provided system users with many other advantages but many companies have

123

Issues and barriers in the implementation of FMS

25

failed to take full benets of these advantages because of increasing complexity in operation and control techniques. An FMS consists of a group of machines or other automated workstations, which form into modular subsystems, such as CNC machines, robots, vision system and a process station. These are interconnected by a material handling system and usually driven by a computer (Maleki 1991). Individual controller unit controls each modular system and these controllers perform their intended tasks under the supervision of a higher-level controller. Though many new techniques and technologies such as neural networks, fuzzy logic, expert systems and articial intelligence, etc., have been proposed in the literature for controlling and integrating various components of FMS but majority of these controls and simulation methodologies are highly complex. They are found to be untenable in practical FMS implementations and in a large context; it is difcult to determine the industrial feasibility of such concepts and methodologies. Overmars and Toncich (1994) have pointed out that the interfacing between FMS controller and the individual group controllers (e.g., coordination of gantry robot with CNC machines and other robots in order to avoid collision or interference) is often complex, weak and unreliable where multi component manufacturing is concerned. At a technological level, the main questions are concerned in getting different hardware and software systems to work together. Most of the rms feel that software represents the major technical problem. Hardware problems mainly revolve around linking machine tools with handling systems and since the technological developments have been evolving incrementally over a long period, they pose few insurmountable difculties. On the software side the picture is more uid due to rapid proliferation of operating systems, communications options and control programs. The key issue is one of software integrationof providing structures and standards within which communication can take place on and between different levels in a control hierarchy. The key aspect of FMS is its ability to adapt to changes in the control tasks. This exibility includes the quantities and varieties of part types which it can produce, the orders in which the operations may be performed and the ability to re-route parts back into ow paths. In short, the control platform should have the capability to automate the ow of information (Mcdermott and Yao 1997). Grieco et al. (2001) have found following drawbacks in the traditional supervisory control system: 1. 2. 3. It is difcult to integrate hardware devices supplied by different companies into the same system. It is difcult to integrate the FMS with other systems operating in the same company. It is difcult to integrate new sensors in existing control architecture.

These are not the only operation and control techniques, which create problems in running of FMS, but maintenance of FMS is also of very complex nature. Vineyard et al. (1999) have reported that during an FMSs extended

123

26

T. Raj et al.

useful life it will experience a different wear and tear history than a traditional machine tool operating during the same time period. According to them, an FMS will operate upto 90% utilization rate whereas a traditional machine tool would probably be utilized at only 20% and it will result in the FMS incurring four times the wear during any given time period. Hence, it will signicantly increase the importance of maintenance. A typical FMS will not only have mechanical, but also electronic, hydraulic, electro-mechanical, software and human element, each having different failure rate distribution. In this respect, maintenance policies for FMS must focus on the understanding of the entire system instead of individual units or components because of integrated nature of units (Cho and Parlar 1991). Pintelon et al. (1995) have also noted that complexity and stochastic nature of maintenance requirements in such environments make deciding on an appropriate maintenance policy a very difcult task. 3.7 Issue regarding human element and culture in FMS Even if FMSs have a high degree of automation, people are normally present to supervise the system, load/unload parts, introduce new xtures and remove unnecessary ones and remove worn tools and replace them with new ones. The behavior of the system is, therefore, affected by the availability of humans (Grieco et al. 2001). Replacing a conventional manufacturing system with FMS alters the physical work environment. No doubt, FMS results in improved work environment, however, the human impact of this transition has yet to be thoroughly evaluated. Although the available literature includes a number of ethnographic studies and experimental work still there is a general lack of rigorous methodology. Contradictory ndings also indicate a real need for additional research work in this area. Investigations focusing on the behavioral impact of FMS systems with respect to operators should be carried out. Since, implementation of full FMS means unmanned operations in the manufacturing area, so, there is very little need of human labor. Hence, existing work force will not agree for such a transition and may put up a great resistance against the implementation of FMS. In particular, worthwhile research is required for analyzing the impact of FMS on workers and the whole society and should be able to answer the following questions: How many workers will be retrenched with the transition to FMS? Cant these workers be adjusted in other work areas? What will be the training and relocation cost for such workers? Whether the workers/employees support the idea of transition to FMS or not? If not, then, how much resistance is expected from them? Whether the organization is ready for any untoward consequences in this age of competition? Whether the organization can bear the loss of short-term or long-term shut down?

123

Issues and barriers in the implementation of FMS

27

What will be long-term human consequences of FMS implementation? What types of additional skills and qualities are essential for FMS supervisor?

Investigation in above mention areas can help clarify the relationship between FMS performance and human element. All these studies may help determine whether the FMS has positive, negative or very minute human consequences. In-depth case studies and experimental investigations are needed to answer these important questions.

4 Identication of barriers On the basis of the literature review and experiences of interaction with manufacturing managers and academicians, it has been found that despite of a lot of research in the area of FMS, there are certain barriers, which inhibit the production managers during transition to FMS. The purpose of this section is to identify these barriers. Several authors have identied the high cost of FMS as a major reason for its low level of acceptance (e.g., Rao and Deshmukh 1994; Koren et al. 1999; Kumar et al. 2006). Rao and Deshmukh (1994) have noted that the adoption of FMS involves a huge investment and a high degree of uncertainty. Kumar et al. (2006) have reported that as FMSs are very expensive, it is essential to manage them effectively to achieve desired goals with less investment-risks. According to Koren et al. (1999), FMSs consist of expensive, general-purpose computer numerically controlled (CNC) machines and other programmable automation. It is not only the use of CNC machines, which adds cost to the FMS but other components, like robots, AGVS, automated storage and retrieval system, conveyor system and their control methodologies also increase the cost of system extensively. Increased level of automation in FMS is also a matter of concern. No doubt, automation inuences the efciency of system but at the same time, it should be cost effective. Efciency of conventional machine tools is between 15% and 30%, with NC machine tools it is about 50% while with machining centers it is about 75% and with FMS it is up to 90% (Balic and Pahole 2003). The efciency of FMS is very high, but if we take the economical aspects and cost effectiveness into account, the total efciency is low (Balic et al. 1995). Another barrier cited by Nelson (1986) is uncertainty. The adaptation of relatively new technology, not surprisingly, increases the risk involved, owing to uncertainties such as those of implementation cost, implementation schedule and product performance. The adoption of high-tech and expensive production system such as FMS adds nancial, technical and organizational risks to the rm (Gupta 1988). Ching and Loh (2003) have raised the issue of good management in successful implementation of FMS. Rao and Deshmukh (1994) have also concluded that implementing the high cost sophisticated

123

28

T. Raj et al.

automated system is a general challenge for a manufacturing rm, as it requires clear-cut goals and policy direction from top management. Another barrier in transition to FMS is that of exibility and its measurement. Craven and Slatter (1988) have discussed that most of the existing FMSs are not exible in true sense of word. Although many systems do have volume exibility yet they lack in product exibility, i.e., they can handle only a limited variety of different parts types. Many exibility measurement schemes also exist but there is a lack of universal acceptance of any one scheme. Sharafali et al. (2004) have discussed that operating an FMS is more complicated than its conventional counterpart. Buyurgan et al. (2004) have also noted that problems related to FMS are relatively complex as compared to traditional manufacturing systems in which lead times are longer, inventory levels are higher and utilization rates are lower. It becomes very difcult to solve real-life FMS planning, scheduling and operational problems because each machine in an FMS is quite versatile and capable of performing many different operations, the system can machine several part types simultaneously and each part may have alternate routes through the system. Hutchison (1991) has reported that the design and operation of FMS has proven to be difcult in practice due to their exibility, complexity and need for system-wide coordination, which makes strategic and tactical decisions, complicated. Spur et al. (1986) have reported that the complexity of these systems has impaired their reliability. Mak et al. (1999) have discussed that the design of FMS is a formidable task and requires the expertise in various disciplines due to its high degree of integration and complexity. One of the problems encountered in the design and implementation of FMS is the layout of machines within the manufacturing cells (Solimanpur et al. 2005). Kouvelis (1991) and Gamila and Motavalli (2003) have discussed the problem of tool management in FMS. Since a large number of tools are used in FMS to machine a variety of parts, it requires a great degree of sophistication to plan, control and monitor tools in the plant. A major bottleneck in a true FMS is the inability of hard-dedicated xtures to be setup and changed quickly and automatically (Markus et al. 1990). Set-up time for new parts is too high. Set-up cost and maintenance cost are very high in an FMS (Buzzacott and Yao 1986). Chan et al. (2000) raised the difculty in design of FMS and commented that FMS design is a very complex task due to two important characteristics(i) a wide variety of alternate system control strategies and congurations available to designers, and (ii) a task in which a variety of selection criteria are involved which are difcult to quantify. They, further, pointed out that the current modeling techniques are not able to solve the FMS design problem as a whole. According to Chan and Chan (2004), there are many operational problems that arise in FMSs domain, for example, scheduling, routing and dispatching problems. McCutcheon (1993) has also reported in his work that many FMS projects are prone to severe implementation problems because of their technical risks. He has listed the following factors, which contribute to technical risks of FMS:

123

Issues and barriers in the implementation of FMS

29

1. 2. 3. 4.

The system complexity. The amount of new hardware and software, which the system incorporates. The number of new kills and routines required for its operations, and The builders experience with the technologies involved.

Based on the literature review and discussions with the experts both from industry and academia, 23 barriers were identied. These barriers are enlisted in Table 1. These barriers are listed in a generalized manner. It is not necessary that all the organizations, which are opting for transition to FMS or implementation of it, will face all of these barriers. But the type of barriers and their impact depends upon the type of organization, type of its end products and their number of models, etc. Hence, any organization opting for FMS must foresee the types of barriers and their impact. They must evaluate the feasibility of transition to FMS before its adaptation.

5 Summary and conclusions The confusion starts from the reported denition of FMS itself in the literature. There is a lack of general agreement over just what an FMS is. Denitions of FMS, provided by different researchers, vary in nature. Some authors dene FMS as a system consisting of CNC machines linked by automatic material handling systems while others include inspection equipments like coordinate measuring machines or machine vision and/or even washing stations in their denition of FMS. Some specify AGVs for parts loading and unloading while others specify robots for the same purpose. This creates a lot of confusion in the minds of industry people as to how many and what type of components should be included in an FMS. Not only its denition, but also other issues related to FMS design and planning such as loading, scheduling, material handling, etc., have not been addressed in a practical way in the literature. Most of the researches have been reported by academia as theory-based approaches, which are difcult to be digested by industry experts. Many technical difculties associated with FMS implementation are felt in real-life scenario and these technical problems have slowed down the growth rate of FMS adaptation. Some surveyors have also reported failure results of FMS. This is the main reason that in the initial phases of FMS adoption, for example during 1980s, though adoption of FMS became the central objective in the push for progress for many manufacturers but slowly its lustre started to fade and situation became more uid because of many technical problems. Practicing managers have found it difcult to exploit these multi-machine systems to their full potential. Today, the situation is so worst that some manufacturing experts are not interested, even, to discuss this high technology system, which was, once, acclaimed as the ultimate weapon in boosting productivity and competitiveness for the industries.

123

Table 1 Barriers and their references/sources References/source Sarin and Chen (1987), Atmani and Lashkari (1998), Rao and Deshmukh (1994), Kovacs (1992)

30

123
Belassi and Fadlalla (1998), Rao and Deshmukh (1994) Rao and Deshmukh (1994) Belassi and Fadlalla (1998), Rao and Deshmukh (1994) Rao and Deshmukh (1994), Koren et al. (1999), Kumar et al. (2006), Park (2005) Belassi and Fadlalla (1998), Rao and Deshmukh (1994) Browne et al. (1984), Chen and Chung (1996), Sethi and Sethi (1990), Das (1996), Tsubone and Horikawa (1999), Beskese et al. (2004), Giachetti et al. (2003), Azzone and Bertele (1991), Rao and Deshmukh (1994) Cordero (1997) Chan et al. (2006) Rao and Deshmukh (1994) Shankar and Vrat (1999), Maxwell and Muckstadt (1982), Egbelu and Tanchoco (1984, 1986), Bozer and Srinivasan (1989), Mahadeven and Narendran (1990), Kim and Tanchoco (1991), Langevin et al. (1994), Kolen et al. (1987), Rajotia et al. (1998), Barad and Sinriech (1998), Gaskins and Tanchoco (1987), Bartholdi and Platzman (1989), Tanchoco and Sinreich (1992), Berman and Edan (2002), Vis (2006), Ho and Liu (2006) Maffei and Meredith (1994), Rao and Deshmukh (1994) Robertson and Gatignon (1986), Rao and Deshmukh (1994), McCutcheon (1993) Solimanpur et al. (2005) Expert opinion Akturk and Avci (1996), Kouvelis (1991), Sharafali et al. (2004), Keung et al. (2003), Gamilla and Motavalli (2003) Cho and Parlar (1991), Pintelon et al. (1995 ), Vineyard et al. (1999), Buzzacott (1986), Ching and Loh (2003), Kenne and Boukas (2003)

SN

Description of barriers

2 3 4 5 6 7

Complex operational and control techniques of FMS Fear of failure Employees resistance in transition to FMS Lack of clear vision High cost of FMS Non-commitment of top management Difculty in measuring the manufacturing exibility

8 9

10 11

Poor rate of return over investment Non-availability of any support from Government Cheap labor Difculty in use of automated guided vehicles

12

13

14

15

16

Non-availability of trained personnel for handling the complex environment of FMS Vendor selection problem in supply of hi-tech equipments Difculty in changing the current layout of machines, i.e., layout problems in FMS Big losses, e.g., loss of market share during transition period Tool management problems

T. Raj et al.

17

FMS maintenance problems

Table 1 continued References/source Stecke (1983), Sarin and Chen (1987), Chen and Askin (1990), Gamilla and Motavalli (2003) Expert opinion

SN

Description of barriers

18

19

20

21

Difculty in handling the loading problems of FMS Social implications due to retrenchment of employees Difculty in handling the scheduling problems of FMS FMS planning problems

Issues and barriers in the implementation of FMS

22 23

Non-availability of good vendors Throughput time

Shaw and Whinson (1989), Dadone et al. (1997), Jain and Foley (2002), Zhao and Wu (2001), Yu et al. (2003), Smith et al. (1993) Stecke (1983), Sarin and Chen (1987), Atmani and Lashkari (1998), Sabuncuoglu and Lahmar (2005), Gamilla and Motavalli (2003) Robertson and Gatignon (1986), Rao and Deshmukh (1994), McCutcheon (1993) Kost and Zdanowicz (2005)

31

123

32

T. Raj et al.

It is very much true that FMS has many advantages and a number of success stories about its utilization have been reported in the literature but, frankly speaking, it is not easy to ride this wild horse. Every thing regarding its implementation as well as its use is not as smooth as has been reported in many research articles. But, behind the scene, the story is quite different and interesting to explore. The warning issued by Green (1986), today also, carries sufcient weight: Dont be misled by what appear to be successful installation stories...Remember that for the benets gained each user has most likely gone through an extensive shakedown. Most FMS stories dont tell you about the birth pains, they just brag about the baby. (Source: OGrady 1989). It is agreed that though good research work has been done in the domain of FMS but, until unless, it is not handy for its easy and smooth implementation in real-life cases, then, what is its use? But, every thing is not lost and there is, still, a ray of hope. Only need is to pay proper attention towards the development of such techniques, which make the FMS a successful one. In order to have a true FMS, all its components, e.g., work-holding and work-transferring devices like xtures and pallets, AGVs and robots, machining heads, etc., are to be exible. If any industry decides to shift to FMS, then, it becomes important to know the feasibility of transition to FMS. Not only this, one should also know the situations in which it is preferable to other manufacturing systems. The company should know the potential benets from FMS before a detailed analysis is carried out. Adaptation of FMS is not only a technical issue but also dependent upon other features such as type of products and their volume, readiness on part of the management, etc. It is generally found that the best way of judging the feasibility of FMS is to gather knowledge about its industrial feasibility. More knowledge about its system components means that one can better understand its design and operational concepts and have deeper insight into crucial variables during its implementation. For this purpose, answers to the following questions can be searched: 1. 2. 3. 4. 5. 6. What types of concerns do motivate the company to adopt or transit to FMS? How much increase in exibility, productivity, automation and economic advantage is anticipated with the transition to FMS? How many FMS applications are there in the industry and at what rate they are being installed? What is their success/failure rate? To what extent, the transition to FMS is applicable? Is a total plant conversion to FMS a feasible and desirable option? What will be the cost of this changeover? What will be the payback period?

For nding the answers to these questions, literature reviews, mail, and telephonic surveys, case studies may be conducted. In case studies, organizations that are currently using FMS, can be studied in detail and the ndings

123

Issues and barriers in the implementation of FMS

33

may be documented in the form of comparisons of rms and their system characteristics. It is true that questions and gaps, identied in relation to different issues, are really irritating and deserve proper research attention. Both academia and practicing manufacturing managers should discuss all these issues on a common platform and future research should be oriented towards addressing core issues in the feasible implementation of FMS in real-life situations. At the moment, we conclude our work with the following comment in response to that of Green (1986): No doubt, babys birth is associated with different kinds of normal pains, even caesarean may be necessary in some cases; if the baby is hale and hearty, then, all such kinds of pains are forgotten. Only need of the hour is to design and develop different kinds of painkillers in the form of hardware and software technologies which make the FMS implementation easier.

References
Abdin MF (1986) Solution of scheduling problems of job shop type FMS with alternative machine tools. Comput Ind Eng 11(1):1921 Akturk MS, Avci S (1996) Tool allocation and machining conditions optimization for CNC machines. Euro J Operat Res 94:335348 Akturk MS, Yilmaz H (1996) Scheduling of automated guided vehicles in a decision making heirarchy. Intl J Product Res 34(2):577591 Atmani A, Lashkari RS (1998) A model of machine-tool selection and operation in FMS. Intl J Product Res 36(5):13391349 Azzone G, Bertele U (1989) Measuring the economic effectiveness of exible automation: a new approach. Intl J Product Res 27:735746 Azzone G, Bertele U (1991) Techniques for measuring the effectiveness of automation. Manuf Syst 48:145 Baker KR (1974) Introduction to sequencing and scheduling. John Wiley & Sons, New York Balic J, Zivec Z, Cus F (1995) Model of a universal manufacturing interface in CIM for small and medium sized companies. J Mater Process Technol 52(1):102114 Balic J, Pahole I (2003) Optimization of intelligent FMS using data ow matrix method. J Mater Process Technol 133(12):1320 Barad M, Sinriech D (1998) A Petri Net model for the operation design and analysis of segmented ow topology (SFT) AGV system. Intl J Product Res 36(5):14011426 Barness JW, Chambers JB (1995) Solving the job shop scheduling problem with Tabu search. IIE Trans 27(2):257263 Bartholdi JJ, Platzman KK (1989) Decentralized control of automated guided vehicles on a simple loop. IIE Trans 21(1):7681 Belassi W, Fadlalla A (1998) An integrating framework for FMS diffusion. OMEGA: Intl J Manag Sci 26(6):699713 Berman S, Edan Y (2002) Decentralized autonomous AGV system for material handling. Intl J Product Res 40(15):39954006 Beskese A, Kahraman C, Irani Z (2004) Quantication of exibility in advanced manufacturing systems using fuzzy concept. Intl J Product Econ 89(1):4556 Bing WX (1998) Analytic process of resource: a practical approach for scheduling. Comput Ind Eng 35(12):97100 Borteld A, Gehring H (2001) A hybrid genetic algorithm for the container loading problem. Euro J Operat Res 131(1):143161 Browne J, Dubois D, Rathmill K, Sethi SP, Steck KE (1984) Classication of exible manufacturing systems. FMS Mag 2(2):114117

123

34

T. Raj et al.

Bryne MD, Chutima P (1997) Real-time operational control of an FMS with full routing exibility. Intl J Product Econ 51(12):109113 Bozer YA, Srinivasan MM (1989) Tandem conguration for AGV systems offers exibility and simplicity. Ind Eng 21(2):2327 Buyurgan N, Soygin C, Engin Kilic SES (2004) Tool allocation in exible manufacturing system with tool alternatives. Robotics Comput-Integrat Manuf 20(4):341349 Buzzacott JA (1982) The fundamental principles of exibility in manufacturing systems. In: Proceedings of rst international conference on exible manufacturing systems, Brighton, UK, pp 1322 Buzzacott J, Mandelbaum M (1985) Flexibility and productivity in manufacturing systems. In: Proceedings of IIE conference, Chicago, pp 404413 Buzzacott J, Yao D (1986) Flexible manufacturing systems: a review of analytical models. Manag Sci 32(7):890905 Cai X, Li KN (2000) A genetic algorithm for scheduling staff of mixed skills under multi-criteria. Euro J Operat Res 125(2):359369 Chan FTS (1999) Evaluation of operational control rules in scheduling a exible manufacturing system. Robot Comput Integrat Manuf 15(2):121132 Chan FTS, Chan HK (2004) Analysis of dynamic control strategies of an FMS under different scenarios. Robot Comput Integrat Manuf 20(5):423437 Chan FTS, Swarnkar R (2006) Ant colony optimization approach to a fuzzy goal programming model for a machine tool selection and operation allocation problem in an FMS. Robot Comput Integrat Manuf 22(4):353362 Chan FTS, Jiang B, Tang NKH (2000) The development of intelligent decision support tools to aid the design of exible manufacturing systems. Intl J Product Econ 65:7384 Chan FTS, Chan HK, Lau HCW, Ip RWL (2003) Analysis of dynamic dispatching rules for a exible manufacturing system. J Mater Process Technol 138(13):325331 Chan FTS, Chung SH, Chan LY, Finke G, Tiwari MK (2006) Solving distributed FMS scheduling problems subject to maintenance: genetic algorithms approach. Robot Comput Integrat Manuf 22(56):493504 Chandra J, Talavage J (1991) Intelligent dispatching for exible manufacturing. Intl J Product Res 29(11):22592278 Chang YL, Sullivan RS, Bagchi U (1984) Experimental investigation of quasi-real-time scheduling in exible manufacturing systems. In: Proceedings of the rst ORSA/TIMS conference on exible manufacturing systems, Stecke KE, Rajan Suri (eds) Ann arbor, MI, pp 307312 Chen IJ, Chung CH (1991) Effects of loading and routing decisions on performance of exible manufacturing systems. Intl J Product Res 29(11):22092225 Chen IJ, Chung CH (1996) An examination of exibility measurements and performance of exible manufacturing systems. Intl J Product Res 34(2):379394 Chen J-H, Ho S-Y (2005) A novel approach to production planning of exible manufacturing systems using an efcient multi-objective genetic algorithm. Intl J of Mach Tools Manuf 45:949957 Chen Y-J, Askin RG (1990) A multiobjective evaluation of exible manufacturing system loading heuristics. Intl J Product Res 28(5):895911 Ching WK, Loh AW (2003) Iterative methods of exible manufacturing system. Intl J Appl Math Comput 141:553564 Cho DI, Parlar M (1991) A survey of maintenance models for multi-unit systems. Euro J Operat Res 51(1):123 Cordero R (1997) Changing human resources to make exible manufacturing systems (FMSs) successful. J High Technol Manag Res 8(2):263275 Cormier D, OGrady P, Saini E (1998) A constraint based genetic algorithm for concurrent engineering. Intl J Product Res 36(6):16791697 Craven FW, Slatter RR (1988) An overview of advance manufacturing technology. Appl Ergonom 19(1):916 Dadone P, Van Landingham HF, Maino B (May 1997) Fuzzy techniques for controlling exible manufacturing systems. In: Proceedings of the IASTED international conference control97, Cancum, Mexico

123

Issues and barriers in the implementation of FMS

35

Daniel SC (1988) Real-time conict resolution in automated guided vehicle scheduling. PhD Thesis, Department of Industrial Engineering, Pennsylvania State University, USA Das SK (1996) The measurement of exibility in manufacturing systems. Intl J Flexible Manuf Syst 8(1):6793 Das SK, Nagendra P (1997) Selection of routes in a exible manufacturing facility. Intl J Product Econ 48(3):237247 Demmel JG, Askin RG (1992) A multi objective decision for the evaluation of advanced manufacturing system technologies. J Manuf Syst 11:179194 Dixon JR (1992) Measuring manufacturing exibility: an empirical investigation. Euro J Operat Res 60:131143 Dolinska M, Bessant CB (1995) Dynamic control of exible manufacturing system. Intl J Adv Manuf Technol 10:131138 Eatson FF, Mansour N (1999) A distributed genetic algorithm for deterministic and stochastic labour scheduling problems. Euro J Operat Res 118(3):505523 Ecker KH, Gupta JND (2005) Scheduling tasks on a exible manufacturing machine to minimize tool change delays. Euro J Operat Res 164(3):627638 Egbelu PJ, Tanchoco JMA (1984) Characterization of automatic guided vehicle dispatching rules. Intl J Product Res 22(3):359374 Egbelu PJ, Tanchoco JMA (1986) Potentials for bi-directional guide-path for automatic guided vehicle based Systems. Intl J Product Res 24(5):10751097 Fanti MP, Maione B, Piscitelli G, Turchiano B (1992) Two methods for real-time routing selection in exible manufacturing system. In Proceedings of IEEE international conference on robotics and automation, pp 11581166 Fry TD, Smith AE (1989) FMS implementation procedure: a case study. IIE Trans 21(3):288293 Gamila MA, Motavalli S (2003) A modeling technique for loading and scheduling problems in FMS. Robot Comput Integrat Manuf 19(12):4554 Gaskins RJ, Tanchoco JMA (1987) Flow path design for automated guided vehicle systems. Intl J Product Res 25(5):667676 Giachetti RE, Martinez LD, Saenz OA, Chen C (2003) Analysis of the structural measures of exibility and agility using a measurement theoretical framework. Intl J Product Econ 86(1):4762 Ghosh S, Gaimon C (1992) Routing exibility and production scheduling in a exible manufacturing system. Euro J Operat Res 60(3):344364 Goldberg DE, Korb B, Deb K (1989) Messy genetic algorithm: motivation analysis and rst results. Complex Syst 3(5):493530 Gray AE, Siedmann AS, Stecke KE (1993) A synthesis of decision models for tool management in automated manufacturing. Manag Sci 39(5):549567 Green RG (1986) Flexible manufacturing systems: are they in your future? Tool Product, July-1986, pp 3538 Grieco A, Semeraro Q, Tolio T (2001) A review of different approaches to the FMS loading problem. Intl J Flex Manuf Syst 13(4):361384 Groover MP (2003) Automation, production systems and computer integrated manufacturing. Prentice-Hall, Inc, New Delhi Guerrero F, Lozano S, Koltai T, Larraneta J (1999) Machine loading and part type selection in exible manufacturing systems. Intl J Product Res 37(6):13031317 Gupta YP (1988) Organizational issues of exible manufacturing systems. Technovation 8:255 269 Gustavsson S (1984) Flexibility and productivity in complex production processes. Intl J Product Res 22:801808 Guvenir HA, Erel E (1998) Multicriteria inventory classication using a genetic algorithm. Euro J Operat Res 105(1):2937 Harmonosky CM, Robohn SF (1995) Investigating the application potential of simulation to realtime control decision. Intl J Comput Integrat Manuf 8(2):126132 He Z, Yang T, Tiger A (1996) An exchange heuristic embedded with simulated annealing for due-dates job-shop scheduling. Euro J Operat Res 91(1):99117 Hertz A, Laporte G, Mittaz M, Stecke K (1998) Heuristics for minimizing tool switches when scheduling part types on a exible machine. IIE Trans 30:689694

123

36

T. Raj et al.

Ho YC, Liu HC (2006) A simulation study on the performance of pick-up dispatching rules for multi-load AGVs. Comput Industr Eng 51(3):445463 Hoff EB, Sarker BR (1998) An overview of path decision and dispatching methods for automated guided vehicles. Integrat Manuf Syst 9(5):296307 Hutchison J (1991) Current and future issues concerning FMS scheduling. OMEGA Intl J Manag Sci 19(6):529537 Hutchison GK, Sinha D (1989) A quantication of the value of exibility. J Manuf Syst 8(1):4757 Jain S, Foley WJ (2002) Impact of interruption on schedule execution in exible manufacturing systems. Intl J Flex Manuf Syst 14(4):319344 Jinyan M, Chai SH, Youvi W (1995) FMS job shop scheduling using Lagarangian relaxation method. In: Proceedings of IEEE international conference on robotics and automation, pp 490495 Kaspi M, Tanchoco JMA (1990) Optimal ow path design of uni-directional AGV system. Intl J Product Res 28(6):10231030 Kenne JP, Boukas EK (2003) Hierarchical control of production and maintenance rates in manufacturing. J Qual Mainten Eng 9(1):6682 Keung KW, Ip WH, Yuen D (2003) An intelligent hierarchical workstation control model for FMS. J Mater Process Technol 139(13):134139 Kim YD, Yano CA (1993) Heuristic approach for loading problems in exible manufacturing systems. IIE Trans 25(1):2639 Kim CW, Tanchoco JMA (1991) Conict-free shortest-time bi-directional AGV routing. Intl J Product Res 29(12):23772391 Klien CM, Kim J (1996) AGV dispatching. Intl J Product Res 34(1):95110 Kolen AWJ, Rinnooy-Kan AHG, Trienekens HWJM (1987) Vehicle routing with time windows. Operat Res 35(2):266274 Koren Y, Heisel U, Jovane F, Moriwaki T, Pritschow G, Ulsoy G, Van Brussel H (1999) Recongurable manufacturing systems. Annal CIRP 48(2):527540 Kost GG, Zdanowicz R (2005) Modeling of manufacturing systems and robot motions. J Mater Process Technol 164165:13691378 Kouvelis P (1991) An optimal tool selection procedure for the initial design phase of a exible manufacturing systems. Euro J Operat Res 55(2):201210 Kovacs GL (1992) Pilot plants to assist FMS/CIM Evaluation. Desig Ind Appl Comput Integrat Manuf Syst 5(1):5460 Kumar N, Shankar K (2000) A genetic algorithm for FMS part type selection and machine loading. Intl J Product Res 38(16):38613887 Kumar P, Tewari NK, Singh N (1990) Joint consideration of grouping and loading problems in a exible manufacturing system. Intl J Product Res 28(7):13451356 Kumar A, Prakash, Tiwari MK, Shankar R, Baveja A (2006) Solving machine-loading problem of a exible manufacturing system with a constraint-based genetic algorithm. Euro J Operat Res 175(2):10431069 Langevin A, Lauzon D, Riopel D (1996) Dispatching, routing and scheduling of two automated guided vehicles in a exible manufacturing systems. Intl J Flex Manuf Syst 8(3):247262 Langevin A, Montreuil B, Riopel D (1994) Spine layout design. Intl J Product Res 32:429442 Lashkari RS, Dutta SP, Padhye AM (1987) A new formulation of operation allocation problem in exible manufacturing systems: mathematical modeling and computational experience. Intl J Product Res 25(9):12671283 Lee DY, DiCesare F (1992) FMS scheduling using petri nets and heuristic search. In: Proceedings of the IEEE international conference on robotics and automation, Nice, France, vol 2, pp 10571062 Lee J, Tangjarukij M, Zhu Z (1996) Load selection of automated guided vehicles in exible manufacturing systems. Intl J Product Res 34(12):33883400 Lee SM, Jung HJ (1989) A multi-objective production planning model in a exible manufacturing environment. Intl J Product Res 27(11):19811992 Liaw C (2000) A hybrid genetic algorithm for the open shop scheduling problem. Euro J Operat Res 124(1):2842 Lin JT, Dgen PK (1994) An algorithm for routing control of a tandem guided vehicle system. Intl J Product Res 32:27352750

123

Issues and barriers in the implementation of FMS

37

Liu J, Maccarthy BL (1996) The classication of FMS scheduling problems. Intl J Product Res 34(3):647656 Low C, Yip Y, Wu T (2006) Modelling and heuristics of FMS scheduling with multiple objectives. Comput Operat Res 33:674694 Maffei MJ, Meredith J (1994) The organizational side of exible manufacturing technology. Intl J Operat Product Manag 14(8):1734 Maleki RA (1991) Flexible manufacturing systems. Prentice-Hall, Englewood cliffs, NJ, pp 90100 Malmborg CJ (1990) A model for the design of zone control automated guided vehicle systems. Intl J Product Res 28(10):17411758 Mahadevan B, Narendran TT (1990) Design of an automated guided vehicle-based material handling system for a exible manufacturing system. Intl J Product Res 28(9):16111622 Mak KL, Wong STW, Lau HYK (1999) An object-oriented rule-based framework for the specication of exible manufacturing systems. Comput Industry 39(2):127146 Markus A, Ruutkay Z, Vancza J (1990) Automating xture designfrom imitating practice to understanding principles. Comput Industry 14(13):99108 McCutcheon DM (1993) Impacts of vendor project management methods on exible manufacturing systems implementation: a eld study. J Eng Technol Manag 10(4):339362 McDermott KJ, Yao WA (1997) Developing a hybrid programmable logic controller platform for a exible manufacturing system. Intl J Flex Manuf Syst 9(4):367374 Maxwell WL, Muckstadt JA (1982) Design of automatic guided vehicle systems. IIE Trans 14(2):114124 Montazeri M, Van Wassenhove LN (1990) Analysis of scheduling rules for an FMS. Intl J Product Res 28(4):785802 Mori M, Tseng CC (1997) A genetic algorithm for multi-mode resource constrained project scheduling problem. Euro J Operat Res 100(1):134141 Mukhopadhyay SK, Midha S, Krishna (1992) A heuristic procedure for loading problems in exible manufacturing systems. Intl J Product Res 10(9):22132228 Mukhopadhyay SK, Singh MK, Srivastava R (1998) FMS loading: a simulated annealing approach. Intl J Product Res 36(6):15291547 Nagalingam SV, Lin GCI (1999) Latest development in CIM. Robot Comput Integrat Manuf 15(6):423430 Naidu EB, Vishwanadham N (1992) An expert system for real-time scheduling in exible manufacturing system. Inform Decision Technol 18:151170 Nasir N, Elsayad EA (1990) Job shop scheduling with alternative machines. Intl J Product Res 28(9):15951609 Nelson CA (1986) A scoring model for exible manufacturing system project selection. Euro J Operat Res 24(3):346359 OGrady P (1989) Flexible manufacturing systems: present developments and trends. Comput Industry 12(3):241251 Overmars AH, Toncich DJ (1994) A new FMS architecture based upon networked DSP servo technology. Intl J Flex Manuf Syst 6(4):311331 Ozden M (1988) A simulation study of multiple-load-carrying automated guided vehicles in a exible manufacturing system. Intl J Product Res 26(8):13531366 Park CS, Son YK (1988) An Economic evaluation model for advanced manufacturing systems. Eng Econ 34:126 Park SC (2005) A methodology for creating a virtual model for a exible manufacturing system. Comput Industry 56(7):734746 Parker RP, Wirth A (1999) Manufacturing exibility: measures and relationships. Euro J Operat Res 118(3):429449 Pintelon LMA, Van Puyvelde FLB, Gelders LF (1995) An age-based replacement policy with non-zero repair times for a continuous production process. Intl J Product Res 33(8):21112123 Primrose PU, Leonard R (1991) Selecting Technology for Investment in Flexible Manufacturing. Intl J Flex Manuf Syst 4(1):5177 Qiu L, Hsu WJ (2001) A bi-directional path layout for conict-free routing of AGVs. Intl J Product Res 39(10):21772195 Qiu L, Hsu WJ, Huang SY, Wang H (2002) Scheduling and routing algorithms for AGVs: a survey. Intl J Product Res 40(3):745760

123

38

T. Raj et al.

Rajamani D, Adil GK (1996) Machine loading in exible manufacturing systems considering routing exibility. Intl J Adv Manuf Technol 11(5):372380 Rajotia S, Shankar K, Batra JL (1998) Determination of optimal AGV eet size for an FMS. Intl J Product Res 36(5):11771198 Rao KVS, Deshmukh SG (1994) Strategic framework for implementing exible manufacturing systems in India. Intl J Operat Product Manag 14(4):5063 Rezaie K, Ostadi BA (2007) Mathematical model for optimal and phased implementation of exible manufacturing systems. Appl Math Comput 184(2):729736 Robertson TS, Gatignon H (1986) Competitive effect on technology diffusion. J Market 50:112 Sabuncuoglu I, Lahmar M (2005) An evaluative study of operation grouping policies in an FMS. Intl J Flex Manuf Syst 15(3):217239 Sallez Y, Trentesaux D, Berger T, Tahon C (2004) Product-based and resource-based heterarchical approaches for dynamic FMS scheduling. Comput Ind Eng 46(4):611623 Sakawa M, Kato K, Mori T (1996) Flexible scheduling in a machining center through genetic algorithms. Comput Ind Eng 30(4):931940 Sarma UMBS, Kant S, Rai R, Tiwari MK (2002) Modeling the machine loading problem of FMS and its solution using a Tabu-search based heuristic. Intl J Comput Integr Manuf 15(4):285 295 Sarin SC, Chen CS (1987) The machine loading and tool allocation problem in a exible manufacturing system. Intl J Product Res 25(7):10811094 Sarkis J (1997) An empirical analysis of productivity and complexity for exible manufacturing system. Intl J Product Econ 48:3948 Saygin C, Kilic SE (1999) Integrating exible process plans with scheduling in exible manufacturing systems. Intl J Adv Manuf Technol 15(4):268280 Sawik T (1990) Modelling and scheduling of exible manufacturing systems. Euro J Operat Res 45(7):177190 Sethi AK, Sethi SP (1990) Flexibility in manufacturing: a survey. Intl J Flex Manuf Syst 2(4): 289328 Shankar K, Sriniwasulu A (1989) Some selection methodologies for the loading problem in exible manufacturing system. Intl J Product Res 27(6):10191034 Sharafali M, Co HC, Goh M (2004) Production scheduling in a exible manufacturing system under random demand. Euro J Operat Res 158(1):89102 Shaw MJ, Whinston AB (1989) An articial intelligence approach to the scheduling of exible manufaturing systems. IIE Trans 21:170183 Shankar R, Vrat P (1999) Automated guided vehicle: an overview. In: Deshmukh SG, Rao PVM (eds) Proceedings of DST sponsored SERC School on AMT, pp 8293 Shewchuk JP, Moodie CL (1998) Denition and classication of manufacturing exibility types and measures. Intl J Flex Manuf Syst 10(4):325349 Shi-jin W, Li-feng X, Bing-hai Z (2006) Filtered-beam-search-based algorithm for dynamic scheduling in FMS. Robot Comput Integrat Manufact (in press) Smith AE, Fry TD, Philipoom PR, Sweigart JE (1993) A comparison of two intelligent scheduling systems for exible manufacturing systems. Exp Syst Appl 6:299308 Sodhi MS, Askin RG, Sen S (1994) Multiperiod tool and production assignment in exible manufacturing systems. Intl J Product Res 32(6):12811294 Solimanpur M, Vrat P, Shankar R (2005) An ant algorithm for the single row layout problem. Intl J Comput Operat Res 32:583598 Spur G, Seligar G, Vichweger B (1986) Cell concepts for exible automated manufacturing. J Manuf Syst 5(3):171179 Stam A, Kuula M (1991) Selecting a exible manufacturing system using a multi criteria analysis. Intl J Product Res 29:803820 Stecke KE, Solberg James J (1981) Loading and tool control policies for a exible manufacturing system. Intl J Product Res 19(5):491490 Stecke KE, Tallbot FB (1983) Heuristic algorithms for exible manufacturing systems. In: Proceedings of 7th international conference on production research, Windsor, Ontario, pp 570576 Stecke KE (1983) Formulation and solution of nonlinear integer production planning problems for exible manufacturing systems. Manag Sci 29(3):273287

123

Issues and barriers in the implementation of FMS

39

Stecke KE (1986) A hierarchical approach to solve machine grouping and loading of exible manufacturing systems. Euro J Operat Res 24(3):369378 Suresh NC (1990) Towards an integrated evaluation of exible automation investments. Intl J Product Res 28:16571672 Suresh NC (1991) An extended multi-objective replacement model for exible automation investments. Intl J Product Res 28:18231844 Son YK, Park CS (1987) Economic measure of productivity. Quality and Flexibility in Advanced Manufacturing Systems. J Manuf Syst 6(3):193206 Son YK, Park CS (1990) Quantifying opportunity costs associated with adding manufacturing exibility. Intl J Product Res 28:11831194 Son YK (1991) A cost estimation model for advanced manufacturing systems. Intl J Product Res 29:441452 Swarnkar R, Tiwari MK (2004) Modeling machine loading problems of FMSs and its solution methodology using a Tabu search and simulated annealing-based heuristic approach. Robot Comput Integrat Manuf 20(3):199209 Taghaboni F, Tanchoco JMA (1995) Comparison of dynamic routing techniques for automatic guided vehicle systems. Intl J Product Res 33(10):26532669 Tanchoco JMA, Sinriech D (1992) OSLoptimal single loop guided paths for AGVs. Intl J Product Res 30(3):665681 Tiwari MK, Hazarika B, Vidyarthi NK, Jaggi P, Mukhopadhyay SK (1997) A heuristic solution approach to machine loading problem of FMS and its Petri-Net model. Intl J Product Res 35(8):22692284 Tiwari MK, Vidyarthi NK (2000) Solving machine loading problem in exible manufacturing system using genetic algorithm based heuristic approach. Intl J Product Res 38(14):33573384 Troxle JN, Blank L (1989) A comprehensive methodology for manufacturing system evaluation and comparison. J Manuf Syst 8:175183 Tsubone H, Horikawa M (1999) A comparison between machine exibility and routing exibility. Intl J Flex Manuf Syst 11(1):83101 Van Laarhoven PJM, Aarts EHL, Lenstra JK (1992) Job shop scheduling by simulated annealing. Operat Res 40(1):113125 Venk S (1990) Strategic optimization cycle as a competitive tool for economic justication of advanced manufacturing systems. J Manuf Syst 9:194205 Veeramani D, Uptom DM, Barash MM (1992) Cutting tool management in computer integrated manufacturing. Intl J Flex Manuf Syst 3(34):237265 Vineyard M, Gyampah KA, Meredith JP (1999) Failure rate distribution for exible manufacturing systems: an empirical study. Euro J Operat Res 116(1):139155 Vis IFA (2006) Survey of research in the design and control of automated guided vehicle systems. Euro J Operat Res 170(3):677709 Yu H, Reyes A, Lang S, Lloyd S (2003) Combined Petri net modeling and AI based heuristic hybrid search for exible manufacturing systempart I Petri net modeling and heuristic search. Intl J Comput Ind Eng 44:527543 Weintraub A, Cormier D, Hodgson T, King R, Wilson J, Zozom A (1999) Scheduling with alternatives: A link between process planning and scheduling. IIE Trans 31(11):10931102 Widmer M (1991) Job shop scheduling with tool constraints: a Tabu search approach. J Operat Res Soc 42(1):7582 Zahir MS (1991) Incorporating the uncertainty of decision judgments in analytical hierarchy process. Euro J Operat Res 53(2):206216 Zhao C, Wu Z (2001) A genetic algorithm approach to the scheduling of FMS with multiple routes. Intl J Flex Manuf Syst 13(1):7188 Zofaghari S, Liang M (1999) Jointly solving the group scheduling and machine speed selection problem: a hybrid Tabu simulated and simulated annealing approach. Intl J Product Res 37(10):23772397

123

40

T. Raj et al. Mr. Tilak Raj is the Assistant Professor in Mechanical Engineering Department in YMCA Institute of Engineering, Faridabad, India. He is B.Sc. (Engg.) in Mechanical Engg., M.E. in Production Engg. and pursuing his Ph.D. research. He has consulted Indian industry on improvement in the manufacturing techniques. His area of expertise is manufacturing technology.

Dr. Ravi Shankar is the Associate Professor in the Department of Management Studies, Indian Institute of Technology, Delhi, India. He is the Group Chair of Sectoral Management and Programme Coordinator of MBA (Telecom) at IIT Delhi. His areas of interest are Operations Management, Supply Chain Management, Knowledge Management, Operations Research, and Telecom Systems Management. He is Ph.D. from IIT Delhi. His research papers have appeared in IEEE Transaction on Systems Man and Cybernetics Part A: Systems and Humans, European Journal of Operational Research, International Journal of Computer Integrated Manufacturing, OMEGA: The International Journal of Management Science, Decision Support Systems, International Journal of Production Research, International Journal of Production Economics, International Journal of Supply Chain Management, Computer and Operations Research, Computer and Ind. Engg., International Journal of Quality & Reliability Management, etc. Dr. Mohammed Suhaib is the Associate Professor in the Department of Mechanical Engineering, Faculty of Engineering & Technology, Jamia Millia Islamia (A Central University), New Delhi, India. He is B.Tech., M.Tech. and Ph.D in Mechanical Engineering. His areas of research interest are Robotics, Automation & Manufacturing, and Flexible Manufacturing Systems.

123

You might also like