Division of Research School of Business Administration The University of Michigan November 1988 Revised, July 1990 A SYNTHESIS OF TOOL-MANAGEMENT ISSUES AND DECISION PROBLEMS IN AUTOMATED MANUFACTURING Ann E. Gray and Abraham Seidmann University of Rochester Kathryn E. Stecke The University of Michigan Working Paper No. 590-b

i I I

ABSTRACT Tool management issues have been receiving increasing attention as more complex automated manufacturing systems have come on-line. Plant tooling systems affect process design options, batching, machine loading and scheduling, and various real-time control decisions. The evidence is clear that a lack of attention to structured tool management has resulted in the poor performance of many automated systems. This paper critically evaluates major decision problems and recent research efforts related to the management of tools in automated manufacturing. These decisions range from selecting the optimal machining parameters and the most economic processing rate for a particular task, to the assignment of tools and parts to machines and the determination of the optimal tool stock levels for a particular production schedule. We outline and use an integrated conceptual framework for resource planning to examine how tool management issues, depending upon their scope, can be classified into tool level, machine level, and system level concerns. This framework identifies the critical control and information interfaces between the various manufacturing management tasks associated with tooling. Further, we evaluate the appropriateness of various tool management approaches in different manufacturing environments and point at several promising research directions.

-- --- ----

1. INTRODUCTION There are critical tool management issues that affect the productivity of many automated and flexible manufacturing systems (FMSs). Manufacturers and machine tool suppliers recognize that a lack of attention to such tool management issues is a primary reason for the poor performance of many facilities (Kiran and Krason [1988], Rhodes [1988], Martin [1989], and Gruver and Senninger [1990]). Conceptually, tool management can be classified into tool level, machine level, and system level issues. This classification allows us to portray how models of individual tools fit into machine level models, how these machine level models fit into system level models, and how technological constraints directly impact decisions at all levels. To ensure the quality performance of an automated system, a high level of integration is necessary between tooling capabilities and the other basic production functions, including process planning, scheduling, part design, and part programming. An effective, well managed information system is necessary to collect and distribute tooling data across these functions. Tool management is broad in scope, requiring: * a design strategy to coordinate tooling inventory, tool tracking, tool handling, and tool loading and unloading, * a planning strategy to ensure that the appropriate tools are available when needed and are provided in the right quantities, * a scheduling strategy to account for tool availability and tool changes, * a control strategy to coordinate either manual or automatic tool transfers between machines and tool cribs, and * a tool monitoring strategy to identify and react to unexpected tool wear and breakage. Besides being a critical issue in factory integration, tool management has direct cost implications. Industry data suggests that tooling accounts for 25% to 30% of both the fixed costs' and variable costs of production in an automated machining environment (Cumings [1986], Tomek [1986], and Ayres [1988]). In the metal cutting industry, as in other automated operations such as assembly and plastic molding, a large number of different tools must be managed. These tools require a substantial degree of sophistication to plan, schedule, control, monitor, and track them among the various machines in the plant. Thus the Integrated Resource Planning Hierarchy developed in Section 2 is applicable not only to the metal cutting industry, but also to other discrete part production systems. Section 3 addresses tool specific issues. Individual tool level decisions include the economic determination of tool types, feedrates, and cutting speed for any given part operation. Other decisions involve standardization of tool types, real-time data monitoring, and adaptive process planning. In Section 3 we group the models into subsections on Tool Life, Cutting Tool

-2 - Economics, Tool Standardization, and Information Requirements for Tool Planning and Monitoring. Section 4 identifies machine level issues related to tooling a single automated machine. We examine the technological capabilities of machine tools for storing tools, loading tools, and monitoring the condition of tools in operation. Typical machine level decisions include the simultaneous sequencing of parts and tools on a specific machine, the allocation of tools to magazine slots, and tool replacement strategies. Machine level performance measures include tool change times, machine throughput, tool replacement and tool regrinding costs. We cover these topics in subsections on Equipment Selection, Tool/Part Sequencing on a Flexible Machine, Tool Placement in a Magazine, and Tool Replacement. Section 5 reviews system management tooling issues. It deals with the impact of tool allocations among several machines, and the interactions among machining conditions and the overall system productivity. Additionally, in this section we address decision problems involving system setup, economic production rates, part routing and scheduling, tool requirements planning, tool sharing among machines, spares management, and tool inventory management. We evaluate the applicability of various methodologies to tool management and identify many open research problems. Section 5 contains subsections on Master Production Planning, Machine Sequencing and Process Monitoring, Process Planning for Economic Production Rates, Spares Management, and Tooling Inventory Management. Section 6 summarizes our conclusions. Conceptually, analyzing tool-specific, machine - level, and system-level issues allows us to evaluate how models of individual tools fit into machine level models, how these fit into system level models, and how technological and performance constraints directly couple decisions at all levels. From studying the literature, it is apparent that few models fully consider practical tooling issues. Models for equipment selection seldom consider the selection of tool types and requisite tool inventories. Most researchers have given little attention to the actual selection of tool storage, handling, loading, changing, and monitoring technologies. Planning models that include tooling constraints and the planning of tooling inventories are now being developed. Critical research areas identified include tool replacement strategies, spares management, and integrating the various tool management decision levels. Scheduling research sometimes considers the size of the tool magazine, and there have been some recent efforts to consider the number of magazine slots used by each tool. We noticed that issues of tool life, loading of duplicate tools, and tool change times are often overlooked. Several studies have proposed implementing adaptive control measures to optimize the machining parameters for each tool during actual production. This work has progressed from initially optimizing tool use on a single machine to optimizing system-wide tool use.

-3 - With further development, many existing design, planning, and scheduling models could realistically account for tooling issues. The lack of appropriate attention to these issues, however, invalidates many other models. 2. AN INTEGRATED TOOL RESOURCE PLANNING HIERARCHY In automated manufacturing, resource planning consists of many interconnected decisions, such as facility loading, tool allocation, and scheduling (see Figure 1). For example, critical process planning decisions and tool selection decisions depend upon the key attributes of the available machines (i.e., horsepower, precision, tool shank parameters), and upon the production volume of each type of part. Expensive, high-speed cutting tools with custom profiles are used for high volume manufacturing, whereas less expensive, general purpose tools are recommended for high mix, low volume operations. The availability of high power, high precision machines may permit both rough-cut and surface finishing operations at the same machine tool. Moreover, the availability of multiple tools of a given type, and the availability of space in the tool magazine, constrain the machine sequencing decisions. These decisions determine the allocation of part operations and tools to machines. Existing resource planning models like MRPII address the planning and control of material flows and machines (Nahmias [1989]), but are incapable of dealing simultaneously with the constraints imposed by tool requirements. The hierarchical structure of the conceptual framework in Figure 1 imposes a set of interrelated resource planning and tool management decisions that guarantee the appropriate coordination of key decision factors. The framework allows for the planning, scheduling, and control of tools, parts, and machines. In the resource planning framework, control flows top-down. The hierarchical structure acknowledges that the options available at a higher level of the structure (e.g., system level) are constrained by those available at lower levels (e.g., machine level and tool level). The Master Production Schedule triggers the Materials Resource Planning (MRP) and the Process Planning and Part Programming activities. The use of a Master Production Schedule enables the process planner to consider desired production volumes and lead times when setting up part routes and tool selections. Information from the MRP and Process Planning/Part Programming activities feeds into the Tooling Requirements Planning (TRP) decision module, which generates the profile of net tool requirements over time as a function of (a) the number of each part type to be produced, (b) the machining times for each operation using each tool, (c) the tool types required, (d) the expected tool lives, and (e) the probability of premature breakdowns. Several plants have already successfully set up TRP systems (Gayman [1986]). The allocations of tools, jigs, and fixtures to groups of machines facilitates machine grouping, or the generation of group technology cells. Machine grouping may reduce part setups and may reduce the traffic of both parts and tools. Scheduling can be particularly sensitive to the

Process Planning and Part Programming. —.. Feedback and Feasibility Information Flow * — Flow of Control Figure 1. The Integrated Resource Planning Hierarchy

-4 - availability of the right type and number of tools. Certain custom-made tools require long lead times for delivery. Their availability imposes tight constraints on scheduling. Some machines can share several tools during one planning period. This allocation policy creates tool scheduling and part synchronization problems. The Tool Inventory Control function deals with maintaining appropriate safety stocks of the various tools and mitigating the random effects of tool supply and demand. This function also includes monitoring the gradual wear of tools, forecasting replenishment orders for worn tools, and predicting tool regrinding requirements. Tool Allocation decisions are tightly coupled with Machine Sequencing and provide input for determining the actual Placement of Tools in the Magazine for each machine. As machines or tools break down, and as parts need reworking, there is an immediate effect on tool availability. Wear rates are accounted for in real-time control through the monitoring and Tool Replacement functions. The following sections discuss the impact of these decisions and the models developed to execute them. Additionally, many design issues are detailed. Since it is necessary to determine the tool handling technology and machine tool magazine capacity before actual startup, these design problems do not appear in the integrated resource planning hierarchy, but they clearly affect planning decisions. Figure 1 3. TOOL SPECIFIC ISSUES Tool specific issues include the number and types of tools, tool speed rates, tool feed rates, and the technology used to monitor and control machining and tooling conditions. With a given set of machine tools, these factors essentially determine the quality of the parts produced and the effective capacity of the machines. In automated manufacturing, these are more critical choices than in manual operations because of (a) the level of integration necessary between the various production functions and (b) the greater capital and time involved in developing hardware, software, and technical support for automated manufacturing. The flow of control described in Figure 1 shows how the data collected by individual tool monitoring activities affects tool replacement decisions due to wear and breakage. Individual tool monitoring and process monitoring data relate machining parameters, such as feed rate and depth of cut, to the rate at which tools must be replaced. This data is used to figure out the desired number and type of tools to be assigned during the tool allocation to machines. Classic tool life and tool economics models address the effects of limited tool life, tool breakage, and tool change times. Yet, many of these studies appear to be unknown to many researchers in the area of automated manufacturing management. As a result, certain published planning and scheduling models ignore the effects of these critical variables. The following

-5 - subsections address four key tool related issues that represent the major tool management concerns at the individual tool level: (1) tool life, (2) tool economics, (3) tool standardization, and (4) information requirements. 3.1 Tool Life The useful life of a tool depends primarily upon the machining environment, including the speed and feed rate, the material composition of the part and of the tool, and the depth of the cut. In pioneering research, Taylor [1907] developed the classical relationship between average tool life and cutting velocity through an empirical study of tool wear. The Taylor tool life equation VIt = k relates cutting speed V to expected tool life T. The empirical constants n and k depend on the cutting conditions and on the material composition of both the part and the tool. The constant value k is numerically equal to the cutting speed that gives a tool life of one minute. It is evident from the Taylor tool life equation that tool life decreases rapidly with an increase in cutting speed. Including the feed rate and the depth of the cut provides a better empirical description of tool life. Today many use the extended tool life equation (Cook [1973] and Jain and Gupta [1987]). This equation C has the form Vt =, where Vt is the equivalent cutting speed (or the cutting speed for a given dx fY tool life), d is the depth of the cut, f is the feed rate per revolution, and C, x, and y are empirical constants. These tool life equations provide expected values based on random tool life data. Empirical curve fitting of shop floor and laboratory failure data have justified the use of the normal, log normal, Weibull, exponential, and gamma distributions (as well as various combinations of them), to describe the life of a tool under various machining conditions (Wagner and Barash [1971], Ramalingam [1978], Ramalingam and Watson [1978a, 1978b], and Ramalingam et al. [1978]). Making decisions such as tool choice, cutting conditions, capacity planning, and tool replacements mandates characterizing and distinguishing failure patterns of various tools. In general, tool life distributions depend upon the nature of the failure mechanism (Cohen and Black [1977], Batra and Barash [1978], Yaohua et al. [1987], and Wayne and Buljan [1989]). As an alternative to selecting a single distribution, a failure rate function may be developed from the actual tool failure data. This is most appropriate when the tool is more likely to fail due to a single injury, such as a fracture, than due to gradual wear of the tool's surface, as is common (a) during high speed cutting, (b) when using impact sensitive tools, and (c) during rough machining (Pandit [1978]). 3.2 Cutting Tool Economics Tool wear increases the probability of a tool chipping or suffering some other type of catastrophic failure that may damage the part being machined (Yaohua et al. [1987]). To minimize the risk of damaging expensive parts, it is more economical to replace the tool early than to damage a part or have the machine shut down. The time interval between planned tool replacements is

-6 - referred to as the "economic tool life." The economic tool life applies to (a) tools which are reground or to (b) disposable inserts. Furthermore, as machining speed increases, tools must be replaced more often and tooling expenses rise exponentially. With the increase in throughput rates, however, a part requires less machine and labor time and also provides the potential for higher revenues. This basic relationship illustrates that machining speed should be a variable when considering the economics of the facility operation. A simple relationship between the number of times a tool is reground and the total cost of regrinding operations is given by Cook [1973]. Cook computes the mean edge cost (Y) as: Y = (Y0 + nG)/(n + 1) where Yo is the initial cost of the tool, n is the number of times that a tool can be reground, and G is the cost of a single regrinding operation. The number of possible regrindings and the cost of each are inversely proportional to the relative time interval allowed between regrinds and to tool life. The economic tool life is found where the mean edge cost is minimized. This problem has been extensively studied since Gilbert [1950] presented a seminal work entitled "Economics of Machining." The interaction between machining conditions and the economic performance of a given machine is discussed by Hitomi [1971, 1976]. In practice, the task cycle time, which depends on the machining speed, feed rate, and spindle revolution rate, is the decision variable used to optimize production. In different environments, this may require maximizing production rate, minimizing variable cost, or maximizing profit rate. Hitomi [1989] derives the optimal machining speed for the case where tool replacement can be made within the setup time for a part. The extended tool life equation reproduced above is commonly used within the industry to find both the most efficient feed rate and the cutting speed for single-pass operations (Drozda and Wick [1983]). In multiple-pass operations, the number of passes and the cutting depth may be determined (Lambert and Walvekar [1978]). It may be worthwhile to determine the sequence of machine tools and part surfaces to be processed in order to meet the dimensional constraints of a finished part with maximal accuracy, as is done by Iwata and Sugimura [1987] using a simple branch-and-bound algorithm. Parameter optimization specific to milling, drilling, reaming, turning, single-pass, two-pass or multi-pass operations have also been treated in the literature (McCullough [1963], Crookall and Venkataramani [1971], Chang et al. [1982], Yellowley [1983], and Hough [1986]). Additionally, Trappey et al. [1987] find optimal machining conditions under a fuzzy set of constraints and Malakooti and Deviprasad [1989] specify a multiple-criteria approach aimed at simultaneously minimizing the production cost per part, the machine cycle time, and the surface roughness. Primrose and Leonard [1986] and Boucher [1987] stress the need to appropriately trade off material, labor, and tool costs by omitting irrelevant overhead allocations historically included in earlier studies. All of these studies deal with manufacturing a single part at a time and assume that tool life is deterministic.

-7 - In attempts to more realistically capture tool life economics, tool life is treated as a random variable whose distribution is determined by the cutting conditions (Ermer [1970], Hati and Rao [1976], Levy and Rossetto [1978a, 1978b], Zompi et al. [1979], and Sheikh et al. [1980]). Rossetto and Levy [1975] present a profit rate distribution function by superimposing random tool fractures on a continuous random tool wear process. Conard and McClamrock [1987] develop a stochastic control model that uses sensory feedback information to determine economic drilling conditions. One study suggests, however, that deterministic models are fairly close to their stochastic counterparts in prescribing the optimal cutting parameters, explained by the convex structure of their machining cost functions (Fenton and Joseph [1979]). The widespread industrial practice of using the same cutting tool for processing a mixture of part types minimizes the number of tool changes and the number of tools required and increases part routing flexibility. However, existing tool life models are unable to provide reliable predictions of tool life under these conditions. The single machine studies discussed here are important precursors to the system level decision models of setting cutting speed and optimizing overall system performance. 3.3 Tool Standardization Metal cutting facilities commonly require hundreds of tool types and maintain thousands of tools in inventory. Standardizing tools can be done either through part or process redesign, or simply by comparing the capabilities of similar tool types and assigning more operations to the same tool type. Standardization results in substantial savings in tool inventories and data management and may improve system reliability by reducing the need for custom tools with long lead times for delivery (Hartley [1984]). Group technology methodologies have been proposed to aid in process planning efforts (e.g., Burbidge [1975, 1990] and Chang and Wysk [1985]). Most of these are limited to the generation of tool commonality subsets. Daskin et al. [1990] is one of few recent studies aimed at the practical issues of tool standardization. It describes a punch-and-die facility which was converted to laser punch-press technology capable of cutting 900 round holes unique in diameter, depth-of-cut, and tolerance. The study details an algorithm for selecting the smallest set of tools that can punch the holes, subject to tool magazine capacity and tool change constraints. Industrial implementation of tool standardization on a large scale, however, will only be possible once general classification, coding, and pattern clustering schemes are developed for automating the standardization process. 3.4 Information Requirements for Tool Planning and Monitoring Interactions among different levels of the integrated tool resource planning hierarchy (Figure 1) is better facilitated when a common tool management data base is used. The data record for a tool type, for instance, should be linked to vendors, part types, machines, and specific operations for each part/machine combination. Each of the numerous tools in the plant must be

-8 - located, tracked for use limits, checked for repairability, and followed through regrind and offset processes (Gruver and Senninger [1990]). Information requirements for both planning and monitoring tooling are extensive. Data on the behavior of tools under different cutting conditions is required for tool selection, in process planning, and in coding and classifying tools for standardization. Tools must be monitored for wear to permit planning for replacement and regrinding. If wear is monitored continuously, adaptive control can be implemented to adjust machine speed and feed rates appropriately. Moreover, inspecting tool conditions off-line increases the non-productive machine times and may result in workpiece damage when the tool fails between the scheduled inspections (Tarn and Tomizuka [1989]). When tool breakage is detected the system can react by arranging for a replacement tool, terminating the processing of the part if it is already damaged, and/or possibly rerouting subsequent parts to other machines (Kendall and Bayoum [1988]). Several companies have developed sophisticated information systems to (a) coordinate delivery of the proper tools to specific machines in time, (b) provide location information, (c) correlate the number of tools needed for the quantity of parts to be produced, and (d) offer acceptable substitutes when needed (Gaymon [1986] and Wick [1987]). As shown in Figure 1, these tool delivery systems interface with machine loading and sequencing functions. Bar-code labeling of tools or tool cabinets, or memory chips embedded in the shanks of toolholders is used to track tool and collect real-time data (Cumings [1986] and Ryan [1987]). In lieu of these developments, it is possible to bypass many of the static-deterministic models of tool life and move directly to adaptive control schemes, where tool performance is directly controlled during a machining task. To do so it is necessary to understand the main issues associated with operating individual machines with multiple tool types as discussed in the next section. 4. TOOL MANAGEMENT ISSUES AT THE INDIVIDUAL MACHINE LEVEL Machine level decisions are influenced by both higher system level decisions and the technology constraints and capabilities of the individual tools discussed in Section 3. Thus, as can be interpreted from Figure 1, individual tools can be allocated to the magazines of the various machines after capacity requirements planning decisions are finalized and machine grouping is determined. There are three key tool management issues at the single machine level: (a) loading, and sometimes placement, of a set of tools in the machine's magazine, (b) determining the part input sequences to meet certain magazine constraints, and (c) establishing tool replacement strategies. To ensure a smooth operation, rules for exception handling and the proper methods of continuously monitoring the system must also be determined. 4.1 Equipment Selection Several of the information and control features available on machine tools support tool management. These options include tool holding and changing capabilities and tool breakage and

-9 - wear monitoring functions. Specifications of a tool magazine and an automatic tool changer include (a) the tool storage capacity, (b) the type of accessing system, (c) whether cutter loading is manual or automatic, (d) the tool shank system used, and (e) the maximum tool diameter, length, and weight. Tool magazines with 30 to 60 tool slots are common, and 70 to 100 tool slots are sometimes available. Some vendors offer machines equipped with several interchangeable tool magazines and others provide a carrier that shuttles the tools between the individual magazines and a centralized tool storage which can contain several hundred tools. This capability is particularly useful for lathes because of the relatively short economic tool lives of many turning tools. Tool magazine speed and capacity are among the most significant parameters for the determination of expected system throughput (Arbel and Seidmann [1984]). Yet, little work has been done to evaluate the relative cost imposed on the system by the size of the tool magazine, by interchangeable tool magazines versus changing tools at the machine, or by manual versus automated tool delivery and loading. Most current research on equipment selection does not consider tooling costs, tool change technologies, magazine size, tool commonalties and tool lives. Alberti et al. [1989] separate tooling and fixture costs from the equipment investment decision, claiming that they do not affect system performance. Believing otherwise, Graves and Redfield [1988] consider tool costs, tool commonalties, and tool change times in equipment selection. They assume, however, that when several tasks using the same tool are assigned to the same machine only one tool is required. This accounts for tool commonalities and saves space in the tool magazine, but it may not always be appropriate for tools with short lives relative to machining time, where duplicates may be necessary, nor does it account for systems in which spare tools are used to ensure system reliability. 4.2 Tool/Part Sequencing on a Flexible Machine The total number of tools required to process a set of parts on a flexible machine is usually larger than the available magazine storage capacity. As a result, a required tool may be absent from the magazine and a tool change must occur before that operation can begin. Tang and Denardo [1988a, 1988b] explore this issue for a single machine with a limited tool magazine, assuming that production requirements are known in advance. The decisions are: (1) how should the parts be sequenced, and (2) which tools to change on the machine prior to processing each part. Their objectives are to minimize the number of group tool change instances or to minimize the number of individual tools changed. The former is appropriate only when the changing time is roughly constant regardless of the number of tools changed. These studies assume that there is a deterministic change time and that all changes are due to part mix, ignoring tool changes due to wear. Bard and Feo [1989] address the problem of minimizing the total setup, tool replacement and machining times for individual batches subject to tool magazine and metal volume removal

-10 - constraints. This approach requires that all feasible tool paths be generated manually before being considered by the optimization algorithm. Silver [1990] studies the possibility of slowing down the processing rate in order to reduce the inventory holding cost in a single machine economic lot sequencing problem. Mittal and Lewis [1989] present an MIP formulation to minimize the sum of the machining time, the tool change times, and the tool travel times. They use a special set of constraints to handle tool life economics and tool changes due to accumulated wear. Their model considers various tooling aspects, but it does not include the option of loading duplicate tools in the magazine. 4,3 Tool Placement in a Magazine The selection and placement of tools in a tool magazine involves many important issues. The machining of a typical part can require a sequence of operations using many cutters of various sizes. Usually a tool covers one or three slots in a tool magazine. Two three-slot tools may only take five slots when placed side by side. This means that the number of magazine slots required for each operation depends on the actual placement of the tools in the magazine. Another potential consideration is weight balancing of a tool magazine (Stecke [1983] and Rajagopalan [1986]). Tool magazine weight balancing and tool overlaps of magazine slots are formulated as a mixture of integer and nonlinear capacity constraints. Tool search time (Stecke [1988]) is important in some environments. This is not the time to find the tool in the magazine (the computer tracks each tool location), but the time required for the magazine to rotate into position for the next tool interchange. Tool search time can take eight to ten seconds; however, if aluminum parts, for example, are being cut, some cutters might only be used for two to three seconds at a time. For a high volume production, the resulting idle search time can be high; however, the correct placement of tools in the tool magazine can reduce such idle time. Some operations may have several cutters in common. If these operations are assigned to the same machine, only one copy of each cutter may need to be loaded, saving magazine capacity. On the other hand, multiple copies may be beneficial, or even necessary, if they are used often or have short lives. It then becomes desirable to load duplicate (sister) copies of these tools into the magazine. This can reduce the number of times that a machine is stopped to change tools but reduces the effective magazine capacity and the machine flexibility. An important, unanswered research problem is the optimal number of sisters of each tool type to load into a magazine. Walas and Askin [1984] address the problem of sequencing operations within part programs and assigning tools to slots for punch presses to minimize the part cycle time, including both table move times and tool change times. Their formulation is a combination of the TSP and the quadratic assignment problem. Comparing three part programs generated by TSP-based commercial software used by a specific company, the Walas and Askin algorithm generates cycle times 8.2% to 24.5% shorter. This study uses two latent assumptions: (a) that it is possible to

-11 - permute the sequence of operations, and (b) that weight balancing of the tool magazines can be ignored. 4.4 Tool Replacement A complete tool replacement strategy specifies a tool change schedule based upon the economic service lives of tools, and a control policy regarding unscheduled tool changes following breakage. Tool replacement strategy is two-fold, consisting of, first, a decision on when to replace a particular tool due to wear or failure, and second, a decision on which additional tools to change early, given that a tool change must take place. The most realistic replacement strategies consider the distributed nature of tool lives under actual machining parameters, as well as the option to change several tools once one fails (Bao [1980] and LaCommare et al. [1983]), rather than considering only expected lives and single tool replacement (McCullough [1963], Cook [1966], and Armarego and Brown [1969]). All of these tool replacement studies consider one machine in isolation. If the machine does not have the potential to create a bottleneck, then a tool change may not result in lost system throughput. On a bottleneck machine, one would be more likely to change several tools when one fails. Sharit and Elhence [1989] go beyond the single machine model to examine tool replacement strategy at the system level. Rather than proposing an automated, optimizing strategy, the study emphasizes the limitations of both human and computer at making the tradeoff between economic tool replacement costs and system throughput in a real-time, dynamic environment. They suggest determining an appropriate mix of human and computer input into the decision process. Currently, many tool replacement models are deficient because they (a) ignore the relationship between the processing rates and the tool replacement policy, and (b) tend to overlook the impact of sharing tools on setup times and on resulting production lot-sizing decisions. 5. SYSTEM MANAGEMENT ISSUES At the factory management level, tooling issues arise in production planning, scheduling, spare tool management and tool inventory management. Production planning involves machine grouping and tool allocations to machines. Once scheduling is complete, facility loading takes place, involving machine sequencing and tool placement in the magazine. The integrated resource planning hierarchy presented in Figure 1 illustrates the necessary interface between the machine level decisions presented in the previous section and the system level decisions discussed here. 5.1 Master Production Planning Each time production requirements for a facility change, the system set-up problems must be resolved. The set-up problems for an automated facility are more difficult than for production lines and job shops because additional part mix and routing flexibilities greatly increase the number of decision variables to be addressed simultaneously. Effective planning models must take into

-12 - account tool magazine sizes, tool commonalities, tool changing times, and tool lives. Choosing to simultaneously produce a set of part types with common tooling requirements will reduce the need for tool changes when magazine sizes are active constraints. Mazzola et al. [1989] propose an MRP framework for automated machining which provides for tool magazine constraints and tool commonalities. Their framework ignores tool changes due to tool wear and is generally appropriate when tool wear is not a significant cause of tool replacements. Because the entire system set-up problem is too large to be solved directly, it is often decomposed into subproblems to be solved independently and iteratively. Each problem employs a surrogate objective for some criteria, i.e., maximizing expected production or minimizing part movements among machines. Stecke [1983] introduces various FMS production planning problems (including part-type selection, machine grouping, machine loading, production ratio, and resource allocation), which the FMS manager has to address to set up a system before production begins. Tool management issues are particularly visible in the part-type selection, machine grouping, and loading problems. Tool/part scheduling for a particular single machine problem was discussed previously, and in Section 5.1.1 we discuss how tooling affects part type selection. In Section 5.1.2, approaches to jointly solve machine grouping and loading problems are discussed, as are various approaches to solve the loading problem independently. In Section 5.1.3 we focus on the choice of tool handling systems. 5.1.1. Part-Type Selection There are two basic approaches to addressing the part-type selection problem. A batching approach partitions the part types into distinct and separate batches and batches are machined oneat-a-time (Whitney and Gaul [1985], Hwang [1986], Rajagopalan [1986], and Afentakis et al. [1989]). When a batch is finished, all tools are taken out of the tool magazines and other tools are loaded for the next batch. A flexible approach selects the part types to be produced next, and machines them according to ratios that balance workloads until all requirements for some part type are met. Then the tools for this part type are taken out of the tool magazines and new tools are loaded for another part type (Stecke and Kim [1989, 1991]). Although tools are changed more frequently with a flexible approach, the time to change tools is much less. The flexible approach results in a more uniform utilization of machines and setup personnel. It leads to better system performance than batching in terms of decreased order leadtimes and increased productivity (Stecke and Kim [1988]). But the flexible approach is likely to require more duplicate tooling and may require a more sophisticated tool transport system. 5.1.2 Machine Grouping and Loading The machine grouping problem is to partition the machines into groups in such a way that each machine in a particular group is able to perform the same set of operations. The objective of the loading problem is to allocate the operations and required tools of the selected part types among

-13 - the machine groups subject to technological and capacity constraints. These two problems can be considered jointly or separately and iteratively.. Balancing the aggregate expected workload across machines has been suggested as a potential surrogate for maximizing expected production in a flexible machining environment. Balancing is an objective of Kusiak [1983], Stecke [1983, 1985a, 1985b, 1988], Ammons et al. [1985], Stecke and Morin [1985], Whitney and Gaul [1985], Berrada and Stecke [1986], and Stecke and Kim [1989]. Stecke and Solberg [1981] show that the loading and control policies which may work towards maximizing production in a conventional environment may not be suitable for application in a more flexible environment because they do not take advantage of the potential system flexibility. Stecke [1983] investigates machine grouping and loading decisions under five different loading objectives, including balancing machine processing times, maximizing the number of consecutive operations on a machine, balancing the workload per machine for a system containing groups of pooled machines of equal sizes, and unbalancing the workload per machine for a system contaning groups of pooled machines of unequal sizes. The major problem constraints are tooling requirements and tool magazine capacities. The common solution has been to solve the MIP after linearization of the nonlinear terms. The loading problem for a nonstationary part mix and machine dependent processing times is considered by Shanker and Srinivasulu [1989]. They use a bicriterion objective of minimizing the workload imbalance and maximizing the throughput rate while considering critical resources such as the number of tools available and the number of magazine slots. Tomek [1986] suggests several approaches to allocate operations and cutters to machines based upon his experience in planning several Czechoslovakian FMSs. These systems have difficult tooling problems (many tools required for each part being machined), identical machines, and a tool delivery system that can deliver up to five cutters at a time. The loading approaches suggested include (1) assigning part types (all operations) to specific machines subject to throughput requirements, current tool magazine content, and technological (process) constraints, (2) assigning a set of tools to a group of parts considering common tooling requirements, and (3) assigning tools to machines and allowing parts to travel between machines. The appropriate approach is a function of the time and complexity of changing tools and of moving parts between machines versus moving tools between machines and a tool crib or spare tool magazine. Machine grouping and loading has also been studied for some more restrictive system configurations. Chakravarty and Shtub [1984] consider these problems for a flexible flow line, where similar part types must follow the same route. Parts of different types are first grouped together by similarities among tool requirements. Na et al. [1987] present a nonlinear integer programming formulation for facility loading with workload balance constraints aimed at

-14 - minimizing the amount of tool traffic among machines and between machines and a central tool crib. The studies by Kusiak [1983], Ammons et al. [1985], and Na et al. [1987] do not consider tool commonalities. They ignore the fact that when operations requiring the same tool type are assigned to the same machine, (a) space is saved in the tool magazine, (b) fewer tools may be needed, and (c) tool changeovers may be avoided. Models that do consider tool commonalities often assume that only one tool of each type is needed to process several parts types on the same machine. This may not be the case if tool lives are short relative to processing requirements. Overall, we found a lack of consideration for tool lives and tool reliability to be one of the major limitations of some planning and scheduling papers in this area. Most approaches center on a single period planning horizon. Many culminate in a solution found by partial enumeration which limits the size of the solvable problems. Reports of their applicability to real production problems are rare. 5.1.3 Manual versus Automatic Tool Handling and Loading Some vendors (i.e., Cincinnati Milicron) offer automatic tool delivery and loading. In the Mazak FMS (Florence, Kentucky), automated guided vehicles are used to transport magazines with a capacity of forty tools to the presetting rooms and back to the required machine tools. The parent company of Mazak, Yamazaki, in Nagoya, Japan also has interchangeable tool magazines but never uses this capability. The foreman prefers to change only the few tools that are worn. Using tool transporters requires a very large additional investment in the tools, magazines, and the delivery system. Some setup time on the tool magazines is reduced but at the expense of requiring a parts batching approach, which can result in system idle time. Sometimes, special tools are delivered to the machine on the same pallet as the part itself. Also, all tools are changed although they are worn to varying degrees. This option necessitates another level of coordination and causes additional scheduling problems. Even if automated, interchanging tools takes some time. Typically, tools are changed in the magazines manually. A study of a COMAU-Torino FMS notes that the major operational problems are tooling and loading (Stecke [1989]). The scheduling problems in this industrial facility turned out to be relatively easy. We noted that currently there is no formal characterization of the operational tradeoffs between automated and manual tool handling (and loading) systems. This is one of the many open design problems in automated manufacturing. 5.2 Machine Sequencing and Process Monitoring Scheduling and control issues arise upon completion of the capacity requirements planning stage (Figure 1). The complexity of scheduling and control generally increases with machine, operation, and routing flexibilities. Few scheduling models fully consider the implications of tooling constraints. Although tool changeovers due to part variety and tool magazine constraints may be included, seldom are tool life and tool changeover times due to tool wear included. Tools

-15 - are resources which must be scheduled and controlled along with parts. When a machine breaks down, workpieces must be rescheduled and delivered along with tools to alternate machines (Carrie and Bitici [1989] and Veeramani et al. [1989]). In a scheduling and control model, Chakravarty and Shtub [1986] include tool magazine capacity constraints and tool changeover times for part-mix changes and allow for periodic review of schedules to correct for problems such as bottlenecks, machine breakdowns, and urgent orders. An order release policy may take into account the time necessary to interchange entire tool magazines instead of individual tools (Chakravarty and Lin [1989]). The rate of tool exchange can be a basic measure of the workload on the tool management system (Rhodes [1988]). Several heuristic scheduling techniques intended to reduce the need for tool changes are presented in the literature. One strategy is to sequence parts on each machine, or on the system as a whole, to minimize tool changeover time between part types. In an empirical study, Carrie and Petsopoulos [1985] found that part sequence has little effect on the performance of a modeled FMS. This is because the availability of fixtures largely determines when parts are input; if parts return several times to a few key machines, their progress depends on the utilization of these few machines. The implications of fixture/pallet availability constraints and of priority scheduling on FMS performance are studied by Shalev-Oren et al. [1985]. Another technique is to sequence parts so as to minimize both the part variety and tool variety at any one time. Menon and O'Grady [1984] suggest sequencing parts so as to minimize a weighted sum of deviations from a desired level of six factors: machine hours, due dates, magazine capacity, number of tools available of each type, number of standard tools at each machine, and the number of nonstandard tools required by each part type. While this approach appears promising, it is not clear how to classify tools as standard or nonstandard, nor how to determine appropriate weights for each factor. Carrie and Perera [1986] post-process data from simulation models of a particular FMS in Anderson-Strathclyde, UK, to compute tooling requirements for several schedules and to evaluate these schedules based on the frequency of tool changes driven by part variety and tool wear. They find that tools are changed ten times more often due to wear than due to part mix. This observation indicates that greater consideration should be placed on minimizing tool changes due to wear. This is in contrast to most recent research, which focuses on minimizing tool changes dictated by part variety. 5.3 Process Planning for Economic Production Rates Schedules are sometimes implemented while assuming a given processing time for each operation on a part type using a particular machine tool. Once a throughput target is set, however, the processing times can be manipulated to reduce costs and increase tool lives (as well as improve surface qualities) at no expense to system throughput. This interaction between machining

-16 - conditions and the overall system throughput suggests that improved scheduling performance can be based on a production rate/tool wear tradeoff. Hitomi [1976, 1977] tackles the joint problems of determining the optimal machining speeds and optimal cycle time in a deterministic multi-stage flow line. Unlimited buffer space is assumed between machines. Cost savings are obtained by slowing down noncritical machines until their cycle times match that of the bottleneck machine. McCartney and Hinds [1982] introduce a procedure to review the machining rates of parts which are first scheduled using maximum production rates. Their procedure will slow some machining rates to reduce production costs (on machines which are not on the critical path) while maintaining due-date performance. Their policy is similar to classical PERT/CPM heuristics (see for example, Whitehouse [1973]). Determining buffer capacity along with optimal cutting speed and tool replacement policy in a two-machine system is discussed by Koulamas et al. [1987]. A penalty cost is imposed for tool failures during production. They show that the tool replacement policies determined independently for each operation do not change when these two operations are coupled, and that the buffer size is sensitive to the tool change times. Queueing network models are used in a large aerospace FMS to optimize the process rates and to determine changes in bottlenecks and queue lengths as the processing rates are altered. This problem is particularly intricate due to the problem of shifting bottlenecks. Schweitzer and Seidmann [1988, 1990] present several nonlinear queueing network optimization methodologies which determine the minimum cost processing rates given the FMS throughput target, the work-inprocess level, part routes, transporter delays, and tool cost functions. Using industrial sample data, they show that a slight acceleration of the processing rates at a few economic bottleneck machines allows for significant rate reductions in others. This provides for substantial gains in tool lives as compared with the conventional one-machine process planning models. Their results also prove that it is not optimal to balance utilization of all machines, to balance waiting times at all machines, or to use the processing times to compensate for local transporter delays. Watanabe and Fujii [1988] find that when adaptive control systems adjust machine feedrates and cutting speeds due to changes in workpiece hardness and tool dullness, predetermined schedules are often violated. They propose a heuristic control model which links the operation speed to the order tardiness. The system proposed is likely to result in major operational improvement. Given the heavy computational demands of this control scheme, however, its applicability to full scale, real-time adaptive control is currently unclear. 5.4 Spares Management Ensuring the availability of required tools is critical to system performance. Even if a specific tool is present at a machine tool at atthe start of a day, its unplanned replacement may be necessary upon detecting a problem such as breakage, wear, poor quality finish, or excessive cutting temperature.

-17 - Results presented for spares management in multi-echelon inventory systems can be extended to the optimal allocation of spare tools among machines. There is a trade off between spares levels and the capacities of repair facilities which is examined by Gross, Miller, and Soland [1983] using a hybrid queueing network optimization model. Spares may be allocated among machining centers. Vinod and Sabbagh [1986] present a closed queueing network model for this allocation, capturing the availability of tools by requiring that the necessary tools be located at the machine before a part is dispatched to it. The cost of spares and of repair capacity is minimized considering failure rates for multiple tool types. The specific storage locations of spare tools -- in magazines, in racks near machines, or in remote tool cribs -- can affect system performance (Kusiak [1986]). Pan et al. [1986] analyze tooling reliability using models for serial systems performance. They predict the reliability of an automatic tool changing system with various carbide inserts and spares subject to Weibull failures. Many studies of multi-level spare parts allocation have appeared in the management science literature (e.g., Bryant [1983], Baker et al. [1986], and Gerchak et al. [1988]). Most are aimed at generic machine component spares and may be extended to handle the particular characteristics of spare tool allocation. In designing a spare tools management policy, one needs to account for the following major system attributes: (1) the capability of the tool handling system, required and existing, (2) the number of machine tools that can perform the same operations, (3) the ability to substitute non-identical tools, (4) the need to provide alternate part routes, (5) the number of identical tools required, (6) the tool magazine capacities, (7) tool life distributions, and (8) tool costs. 5.5 Tooling Inventory Management A large number of tool types is often required to facilitate operational flexibility. Berr and Falkenburg [1985] provide statistics indicating that in practice, for each tool type, there are at least three duplicate tools required: one in a tool magazine, one as a backup (centralized or on a relevant machine), and one in preparation (i.e., refurbishing, inspection, reconditioning, presetting, or mounting into the tool shank). Moreover, the number of tool types in storage increases over time due to such factors as new product introductions, engineering changes in existing products, and the availability of more advanced cutting materials. The appropriate number of tools to be purchased of each tool type must be determined (Graver and McGinnis [1988]). Optimal tool reorder points and safety stock levels are not addressed in the literature. Custom tools can shorten processing times, but are more expensive and require extensive purchase

-18 - lead times. This tradeoff has not been studied; nor has the tradeoff between tool availability, manufacturing capacity, tool reorder points, and the overall investment in tooling stocks. 6. SUMMARY AND CONCLUSIONS We have outlined several approaches to tool management problems in automated manufacturing. Our analysis clearly indicates that tooling decisions strongly affect plant capacity, scheduling flexibility, and a significant portion of the variable production costs. Several industrial applications have been described whereby integrated tool management systems have resulted in significant operational savings and improved facility performance. We have presented a conceptual framework for tool management decisions in the context of resource planning in automated manufacturing. This framework shows the control and information interfaces between the various manufacturing management tasks associated with tooling. Table 1 presents a taxonomy of the major tool-based research problems discussed here. It conceptualizes the taxonomy around design, planning, and control aspects at the tool, single machine, and system management decision levels. Our analysis of the tool management decision issues explains why decisions made at one level constrain those at lower levels, and how information from lower-level decisions feeds back to higher levels. For example, the choice of machining parameters depends upon tooling economics (a tool level decision) which is used as information in determining the spare tool allocation (a system level decision); but spare tools inventories influence system reliability and the potential for parallel processing of identical parts on several machines (a system level decision). Similarly, a decision to increase the number of identical tools loaded simultaneously into the tool magazine of a single machine reduces its product scope. This limits the grouping and loading strategies (a system level decision), reduces the number of setups needed for changing worn tools (a single machine level decision), and allows the process designer to increase the cutting speed for certain operations (a tool level decision). The increase in the number of automated facilities, and the corresponding increase in the number of scientific publications associated with modeling the impact of tool management decisions, clearly attests to the increasing concern of properly integrating tooling considerations within production management. Many of the studies discussed here prove that significant operational benefits can be realized with proper tool selection and allocation policies. Our study points at several promising research directions in this area: * Analysis of particular decision problems: These include such research issues as tool inventory levels and the dynamic allocation of duplicate tools to machines (see Table 1). We need to study the impact of design decisions, such as tool magazine capacity, on tooling costs and the effect of the subsequent tooling constraints on system capabilities. * Integration of various tool management decision levels: To date, most research efforts in tool management focus on single-level decisions. Ignoring the impact across levels leads to suboptimal

-19 - results. The current research incorporating tooling economics within production scheduling exemplifies the benefits of integrating decision levels. * Improved modeling of actual industrial practice: Certain studies still make unrealistic assumptions, for example (a) that all tools require only one magazine slot, (b) that each operation requires exactly one tool, or (c) that tool costs are independent of machining rates. * Field driven empirical studies: There is a need for systematic evaluation of current industrial tool management policies, and of various approaches recently suggested in the academic literature. In addition, useful tool-related data is scanty, particularly when it comes to estimating the expected tool life (and costs) for tools shared by multiple part types and machines. * Strategic role of tool management in product designs: In an era when product life cycles continue to shrink, managers are searching for better means to integrate product design, testing and manufacturing functions. Incorporating appropriate selection and loading models within the tooling information systems will facilitate concurrent tool/product design efforts, reduce the timeto-market for new products, and economize on existing manufacturing resources. Table 1

-20 - BIBLIOGRAPHY 1. AFENTAKIS, PANOS, SOLOMON, MARIUS, and MILLEN, ROBERT A., (1989), "The Part Type Selection Problem," Proceedings of the Third ORSA/TIMS Conference on Flexible Manufacturing Systems: Operations Research Models and Applications, K. E. Stecke and R. Suri (eds.) Elsevier Science Publishers B.V., Amsterdam, pp. 141-146. 2 ALBERTI, N., LA COMMARE, U., and LA DIEGA, S. NOTO, (1989), "Cost Efficiency: An Index of Operational Performance of Flexible Automated Production Environments," Proceedings of the Third ORSA/TIMS Conference on Flexible Manufacturing Systems: Operations Research Models and Applications, K. E. Stecke and R. Suri (eds.) Elsevier Science Publishers B.V., Amsterdam, pp. 67-72. 3. AMMONS, JANE C., LOFGREN, C. B., and McGINNIS, LEON F., (1985), "A Large Scale Work Station Loading Problem," Annals of Operations Research, Vol. 3, pp. 319-332. 4. ARBEL, A. and SEIDMANN, ABRAHAM, (1984), "Performance Evaluation of Flexible Manufacturing Systems," IEEE Transactions on Systems, Man and Cybernetics, Vol. 14, No. 4, pp. 606-612. 5. ARMAREGO, E. J. A. and BROWN, R. H., (1969), The Machining of Metals, Prentice-Hall, Inc., Englewood Cliffs, NJ. 6. AYRES, ROBERT V., (1988), "Future Trends in Factory Automation," Manufacturing Review, Vol. 1, No. 2, pp. 93-103. 7. BAKER, KENNETH, MAGAZINE, MICHAEL AND NUTTLE, HENRY, (1986), "The Effect of Commonality on Safety Stock in a Simple Inventory Model," Management Science, Vol. 32, No. 8. 8. BAO, HAN, (1980), "Application of Dynamic Programming to Optimize Tool Replacement Schedules for Multi-Tool Operations Involving Distributed Tool Lives," ASME, Journal of Mechanical Design, Vol. 102, pp. 446-451, July. 9. BARD, JONATHAN and FEO, THOMAS A., (1989), "The Cutting Path and Tool Selection Problem in Computer Aided Process Planning," Journal of Manufacturing Systems, Vol. 8, No. 1, pp. 17-26. 10. BATRA, J. and BARASH, MOSHE M., (1972), "Computer Aided Planning of Optimal Machining Operations for Multiple Tools Setups with Probabilistic Tool Life," Report No. 49, Purdue University, Laboratory for Applied Industrial Control, W. Lafayette, IN. 11. BERR, A. and FALKENBURG, DONALD R., (1985), "Tool Management for Flexible Manufacturing Systems," Annals of the CIRP, Vol. 34, No. 1, pp. 387-390. 12. BERRADA, MOHAMMED and STECKE, KATHRYN E., (1986), "A Branch and Bound Approach for Machine Load Balancing in FMS," Management Science, Vol. 32, No. 10, pp. 1316-1335. 13. BOUCHER, T. 0., (1987), "The Choice of Cost Parameters in Machining Cost Models,"The Engineering Economist, Vol. 32, No. 3, pp. 217-230, Spring. 14 BRYANT, JOHN L., (1983), "Stocking Repair Kits for Systems with Limited Life," Management Science, Vol. 29, No. 5, pp. 546-558. 15. BURBIDGE, JOHN L. (1975), The Introduction of Group Technology, Heinemann, Ltd., London, UK. 16. BURBIDGE, JOHN L., (1990), "Production Control: A Universal Conceptual Framework," Production Planning & Control, Vol. 1, No. 1, pp. 3-16. 17. CARRIE, ALAN S., and BITICI, UMIT S., (1989), "Tool Management and its Implication for Material Flow," in Advanced Information Technologies for Industrial Material Flow Systems, Shimon Y. Nof and Colin L. Moodie (eds.), Springer-Verlag, Berlin. 18. CARRIE, ALAN S. and PERERA, D. T. S., (1986), "Work Scheduling in FMS under Tool Availability Constraints," International Journal of Production Research, Vol. 24, No. 6, pp. 1299-1308. 19. CARRIE, ALAN S. and PETSOPOULOS, A. C., (1985), "Operation Sequencing in an FMS," Robotica, Vol. 3, p. 259. 20. CHAKRAVARTY, AMIYA K.and LIU, JOHN J., (1989), " Optimal Job-order Release in a Robotic Assembly Cell," Proceedings of the Third ORSAITIMS Conference on Flexible

-21 - Manufacturing Systems: Operations Research Models and Applications, K. E. Stecke and R. Suri (eds.), Elsevier Science Publishers B.V., Amsterdam, pp. 205-210. 21. CHAKRAVARTY, AMIYA K. and SHTUB, AVRAHAM, (1984), "Selecting Parts and Loading Flexible Manufacturing Systems," Proceedings of the First ORSA/TIMS Special Interest Conference on Flexible Manufacturing Systems: Operations Research Models and Applications, K. E. Stecke and R. Suri (eds.), Ann Arbor, MI, August 15-17. 22. CHAKRAVARTY, AMIYA K. and SHTUB, AVRAHAM, (1986), "Production Planning with Flexibilities in Capacity," Proceedings of the Second ORSA/TIMS Conference on Flexible Manufacturing Systems: Operations Research Models and Applications, K. E. Stecke and R. Suri (eds.), Elsevier Science Publishers B.V., Amsterdam. 23. CHANG, TED C. and WYSK, RICHARD A., (1985), An Introduction to Automated Process Planning Systems, Prentice-Hall, Englewood Cliffs, NJ. 24. CHANG, TED, WYSK, RICHARD A., DAVIS, R. P., and CHOI, B., (1982), "Milling Parameter Optimization Through a Discrete Variable Transformation," International Journal of Production Research, Vol. 20, No. 4, pp. 507-516. 25. COHEN, P. H. and BLACK, J. T., (1977), "Tool Life Distribution Discussion," Journal of Engineering for Industry, Vol. 97, No. 3. 26. CONARD, C. J. and McCLAMROCK, N. HARRIS, (1987), "The Drilling Problem: A Stochastic Modeling and Control Example in Manufacturing," IEEE Transactions on Automatic Control, Vol. 32, No. 11, pp. 947-958. 27. COOK, NATHAN H., (1966), Manufacturing Analysis, Addison-Wesley, Reading, MA. 28. COOK, NATHAN H., (1973), "Tool Wear and Tool Life," ASME, Journal of Engineering for Industry, pp. 931-938, November. 29. CROOKALL, J. R. and VENKATARAMANI, N., (1971), "Computer Optimization of Multipass Turning," International Journal of Production Research, Vol. 9, No. 2, pp. 247-259. 30. CUMINGS, S., (1986), "Developing Integrated Tooling Systems: A Case Study at Garrett Turbine Engine Company," Proceedings of the Fall Industrial Engineering Conference, Boston, MA. 31. DASKIN, MARK, JONES, PHILIP C., and LOWE, TIMOTHY J., (1990), "Rationalizing Tool Selection in a Flexible Manufacturing System for Sheet-Metal Products,"Operations Research., forthcoming. 32. DROZDA, T. J. and WICK, C. (eds.), (1983), Tools and Manufacturing Engineering Handbook, SME Publication, Dearborn, MI. 33. ERMER, D. S., (1970), "A Bayesian Model of Machining Economics for Optimization by Adaptive Control," Journal of Engineering for Industry, Vol. 92, pp. 628-632. 34. ERSCHLER, JACQUES, LEVEQUE, DIDIER, and ROUBELLAT, FRANCOIS, (1982), "Periodic Loading of Flexible Manufacturing Systems," Proceedings of the IFIP Congress, APMS, Bordeaux, France, pp. 327-339, August 24-27. 35. FENTON, R. G. and JOSEPH, N. D., (1979), "The Effects of Statistical Nature of Tool life on the Economics of Machining," International Journal of Machine Tool Design Research, Vol. 19, p. 43. 36. GAYMON, D. J., (1986), "Computers in the Tool Crib," Manufacturing Engineering, pp. 41 -44, September. 37. GERCHAL, Y., MAGAZINE, MICHAEL J., and GAMBLE, B., (1988), "Component Commonality with Service Level Requirements," Management Science, Vol. 34, No. 6, pp. 753-760. 38. GERSHWIN, STANLEY B., HILDEBRANT, RICHARD R., SURI, RAJAN, and MIl TER, SANJOY K., (1986), "A Control Perspective on Recent Trends in Manufacturing Systems," IEEE Control Systems Magazine, Vol. 6, No. 2, pp. 3-15. 39. GILBERT, W. W., (1950), "Economics of Machining," in Machining - Theory and Practice, American Society for Metals, Cleveland, OH. 40. GRAVER, THOMAS W. and McGINNIS, LEON F., (1989), "A Tool Provisioning Problem in FMS," International Journal of Flexible Manufacturing Systems, Vol. 1, No. 3, pp. 239 -254, June.

-22 - 41. GRAVES, STEPHEN C. and REDFIELD, CAROL HOLMES, (1988), "Equipment Selection and Task Assignment for Multiproduct Assembly System Design," International Journal of Flexible Manufacturing Systems, Vol. 1, No. 1, pp. 31-50, September. 42. GROSS, DONALD, MILLER, D. R., and SOLAND, RICHARD M., (1983), "A Closed Queueing Network Model for Multi-Echelon Repairable Item Provisioning," IIE Transactions, Vol. 15, No. 3, pp. 344-352. 43. GRUVER A. and SENNINGER M.T., (1990), "Tooling Management in FMS," Mechanical Engineering, Vol. 112, No. 3, pp. 40-44.. 44. HARTLEY, JOHN R., (1984), FMS at Work, IFS Publications, Bedford, UK. 45. HATI, S. K. and RAO, S. S., (1976), "Determination of Optimum Machining Conditions - Deterministic and Probabilistic Approaches," Journal of Engineering for Industry, Vol. 98, pp. 354-359. 46. HITOMI, K., (1971), "Studies of Economical Machining," Bulletin of the Japan Society of Mechanical Engineers, Vol. 14, No. 69, pp. 294-302. 47. HITOMI, K., (1976), "Analysis of Production Models; Part I: The Optimal Decision of Production Speeds; Part II: Optimization of a Multistage Production System," AIIE Transactions, Vol. 8, No. 1, pp. 96-112, March. 48. HITOMI, K., (1977), "Optimization of Multistage Production Systems with Variable Production Times and Costs," International Journal of Production Research, Vol. 15, No. 6, pp. 583-597. 49. HITOMI, K., (1989), "Analysis of Optimal Machining Speeds for Automatic Manufacturing," International Journal of Production Research, Vol. 27, No. 10, pp. 1685-1691. 50. HOUGH, C.L., (1986), "Sufficient Conditions for Cutting Rate Tool Life Characteristic Functions for Metal Removal,"Journal of Engineering for Industry, Vol. 108, No. 4, pp. 235 -237. 51. HWANG, SYMING, (1986), "Part Selection Problem in Flexible Manufacturing Systems Planning Stage," Proceedings of the Second ORSAITIMS Conference on Flexible Manufacturing Systems: Operations Research Models and Applications, Ann Arbor, MI, Elsevier Science Publishers B.V., Amsterdam, pp. 297-309, August 12-15. 52. IWATA,K. and SUGIMURA, N., (1987), "An Integrated CAD/CAPP System with "KnowHows" on Machining Accuracy," Journal of Engineering for Industry, Vol. 109, No. 2, pp. 128-133. 53. JAIN, V.K. and GUPTA, B.K., (1987), "Effects of Accelerated Tests on Shear Flow Stress in Machining," Journal of Engineering for Industry, Vol. 109, No. 3, pp. 206-212. 54. KENDALL L. A. and BAYOUMI, A., (1988), "Automated Tool-Wear Monitoring and Tool Changing Using Intelligent Supervisory Control," International Journal of Production Research, Vol. 26, No. 10, pp. 1619-1628. 55. KIRAN, ALI S. and KRASON, R. J., (1988), "Automating Tooling in a Flexible Manufacturing System," Industrial Engineering, pp. 52-57, April. 56. KOULAMAS, C. P., LAMBERT, B. K., and SMITH, MILTON L., (1987), "Optimal Machining Conditions and Buffer Space Size for the Two Stage Case," International Journal of Production Research, Vol. 25, No. 3, pp. 327-336. 57. KUSIAK, ANDREW, (1983), "Loading Models in Flexible Manufacturing Systems," Working Paper No. 05/83, Department of Industrial Engineering, Technical University of Nova Scotia, Halifax, Nova Scotia, Canada, March. 58. KUSIAK, ANDREW, (1986), "Parts and Tools Handling Systems," Modelling and Design of Flexible Manufacturing Systems, A. Kusiak (ed.), Elsevier Science Publishers B.V., Amsterdam, pp. 99-109. 59. La COMMARE, U., DIEGA, L., NOTA, S., and PASSANNANTE, A., (1983), "Optimum Tool Replacement Policies with Penalty Cost for Unforeseen Tool Failure," International Journal of Production Research, Vol. 23, p. 237. 60. LAMBERT, B. K. and WALVEKAR, A. G., (1978), "Optimization of Multi-pass Machining Operations," International Journal of Production Research, Vol. 16, No. 4, pp. 259-265.

-23 - 61. LEVI, R. and ROSSETTO, S., (1978a), "Machining Economics and Tool Life Variation, Part 1: Basic Considerations and Their Practical Implications," Journal of Engineeringfor Industry, Vol. 100, pp. 393-396. 62. LEVI, R. and ROSSETTO, S., (1978b), "Machining Economics and Tool Life Variation, Part 2: Applications to Models for Machining Processes," Journal of Engineering for Industry, Vol. 100, pp. 397-414. 63. MALAKOOTI, B. and DEVIPRASAD, J., (1989), "An Interactive Multiple Criteria Approach for Parameter Selection in Metal Cutting," Operations Research, Vol. 37, No. 5, pp. 805-818. 64. MARTIN, JOHN M., (1989), "Managing Tools Makes the Cell Click," Manufacturing Engineering, Vol. 102, No. 4, pp. 59-62, April. 65. MAZZOLA, JOSEPH B., NEEBE, ALAN W., and DUNN, CHRISTOPHER V. R., (1989), "Production Planning of a Flexible Manufacturing System in a Material Requirements Planning Environment," International Journal of Flexible Manufacturing Systems, Vol. 1, No. 2, pp. 115-142, April. 66. McCARTNEY, J. and HINDS, B. K., (1982), "Tooling Economics in Integrated Manufacturing Systems," International Journal of Production Research, Vol. 20, No. 4. pp. 493 -505. 67. McCULLOUGH, E. M., (1963), "Economics of Multitool Lathe Operations," ASME, Journal of Engineering for Industry, pp. 402-404, November. 68. MENON, U. and O'GRADY, PETER J., (1984), "A Flexible Multiobjective Production Planning Framework for Automated Manufacturing Systems," Engineering Costs and Production Economics, Vol. 8, pp. 189-200. 69. MITTAL, R. 0. and LEWIS, R. L., (1989), "A 'Micro' Process Planning System Based on Integer Programming for Prismatic Parts Produced on Horizontal Machining Centers," Annals of Operations Research, Vol. 17, pp. 273-290. 70. NAHMIAS, S., (1989), Production and Operations Management, Irwin, Boston, MA. 71. NA, Y. K., HAN, MIN H., and HOGG, GARY Z., (1987), "Tool Loading and Control in a Flexible Manufacturing Cell," Proceedings of the IIE Fall Conference, Nashville, TN. 72. PAN, J. N., KOLARIK, W. J., and LAMBERT, B. K., (1986), "Mathematical Models to Predict the System Reliability of Tooling for Automated Machining Systems," International Journal of Production Research, Vol. 24, No. 3, pp. 493-501. 73. PANDIT, S. M., (1978), "Data Dependent Systems Approach to Stochastic Tool Life and Reliability," ASME, Journal of Engineering for Industry, Vol. 100, pp. 318-322, August. 74. PRIMROSE, P. L. and LEONARD, R., (1986), "Reappraising Cutting Tool Economics within the Bounds of Accountancy Theory," International Journal of Production Research, Vol. 24, No. 2, pp. 269-278. 75. RAJAGOPALAN, S., (1986), "Formulation and Heuristic Solutions for Parts Grouping and Tool Loading in Flexible Manufacturing Systems," Proceedings of the Second ORSAITIMS Conference on Flexible Manufacturing Systems: Operations Research Models and Applications, K. E. Stecke and R. Suri (eds.), Elsevier Science Publishers B.V., Amsterdam. 76. RAMALINGAM, S., (1978), "Tool Life Distributions. Part 2. Multiple-Injury Tool life Model," ASME, Journal of Engineering for Industry, Vol. 99, No. 3, pp. 523-531. 77. RAMALINGAM, S., PENG, Y. I., and WATSON, J. D., (1978), "Tool life Distribution. Part 3. Mechanism of Single Injury Tool Failure and Tool Life Distribution in Interrupted Cutting," Journal of Engineering for Industry, Vol. 100, No. 1, pp. 193-200. 78. RAMALINGAM, S. and WATSON, J. D., (1978a), "Tool Life Distribution. Part 1. SingleInjury Tool life Model," Journal of Engineering for Industry, Vol. 99, No. 3, pp. 519-522. 79. RAMALINGAM, S. and WATSON, J. D., (1978b), "Tool Life Distribution. Part 4. Minor Phases in Work Material and Multiple-Injury Tool Failure," Journal of Engineering for Industry, Vol. 100, No. 1, pp. 201-209. 80. RANKY, PAUL, (1983), The Design and Operation ofFMS, IFS Publications, Bedford, UK. 81. RHODES, J. S., (1988), "FMS Tool Management Systems," in Flexible Manufacturing Systems, Thomas J. Drozda (ed.), SME Publication, Dearborn, MI.

-24 - 82. ROSSETTO, S. and LEVY, R., (1975), "Fracture and Wear as Factors Affecting Stochastic Tool Life Models and Machining Economics," Winter Annual Meeting of the ASME, Houston, TX. 83. RYAN, N. E., (1987), "Bar Coding Labeled Efficient and Flexible," Manufacturing Engineering, pp. 34-38, June. 84. SCHWEITZER, PAUL J. and SEIDMANN, ABRAHAM, (1988), "Capacity Range Analysis and Processing Rate Optimization for FMS's with Distinct Multiple Job Visits to Work Centers," Working Paper, W. E. Simon Graduate School of Business Administration, University of Rochester, Rochester, NY. 85. SCHWEITZER, PAUL J. and SEIDMANN, ABRAHAM, (1990), "Optimizing Processing Rates for Flexible Manufacturing Systems," Management Science, forthcoming. 86. SHALEV-OREN, S., SEIDMANN, ABRAHAM, and SCHWEITZER, PAUL J., (1985), "Analysis of Flexible Manufacturing Systems with Priority Scheduling PMVA," Annals of Operations Research, Vol. 3, pp. 115-139. 87. SHANKER, KRIPA and SRINIVASULU, A.,(1989), "Some Solution Methodologies for Loading Problems in a Flexible Manufacturing System," International Journal of Production Research, Vol. 27, No., 6, pp. 1019-1034. 88. SHARIT, JOSEPH and ELHENCE, SHARAD, (1989), "Computerization of ToolReplacement Decision Making in Flexible Manufacturing Systems: a Human-Systems Perspective," International Journal of Production Research, Vol. 27, No. 12, pp. 2027-2039. 89. SHEIKH, A. K., KENDAL, L. A., and PANDIT, S. M., (1980), "Probabilistic Optimization of Multitool Machining Operations," Journal of Engineering for Industry, Vol. 102, p. 239. 90. SILVER, EDWARD, (1990), "Deliberately Slowing Down Output in a Family Production Context," International Journal of Production Research,"Vol. 28, pp. 17-27. 91. STECKE, KATHRYN E., (1983), "Formulation and Solution of Nonlinear Integer Production Planning Problems for Flexible Manufacturing Systems," Management Science, Vol. 29, No. 3, pp. 273-288, March. 92. STECKE, KATHRYN E., (1985a), A Hierarchical Approach to Solving Machine Grouping and Loading Problems of Flexible Manufacturing Systems," European Journal of Operational Research, Vol. 24, No. 3, pp. 369-378, March. 93. STECKE, KATHRYN E., (1985b), "Procedures to Determine Both Appropriate Production Ratios and Minimum Inventory Requirements to Maintain These Ratios in Flexible Manufacturing Systems," Working Paper No. 448, Division of Research, GSBA, The University of Michigan, Ann Arbor, MI, October. 94. STECKE, KATHRYN E., (1989), "Algorithms for Efficient Planning and Operation of a Particular FMS," International Journal of Flexible Manufacturing Systems, Vol. 1, No. 4, pp. 287 -324, September. 95. STECKE, KATHRYN E. and KIM, ILYONG, (1991), "A Flexible Approach to Part Type Selection in Flexible Flow Systems using Part Mix Ratios," International Journal of Production Research, forthcoming. 96. STECKE, KATHRYN E. and KIM, ILYONG, (1988), "A Study of FMS Part Type Selection Approaches for Short-term Production Planning," International Journal of Flexible Manufacturing Systems, Vol. 1, No. 1, pp. 7-29, September. 97. STECKE, KATHRYN E. and KIM, ILYONG, (1989), "A Study of Unbalancing and Balancing for Systems of Pooled Machines of Unequal Sizes," European Journal of Operational Research, Vol. 42, No. 1, pp. 22-38. 98. STECKE, KATHRYN E. and MORIN, THOMAS L., (1985), "The Optimality of Balancing Workloads in Certain Types of Flexible Manufacturing Systems," European Journal of Operational Research, Vol. 20, No. 1, pp. 68-82, April. 99. STECKE, KATHRYN E. and SOLBERG, JAMES J., (1981), "Loading and Control Policies for a Flexible Manufacturing System," International Journal of Production Research, Vol. 19, No. 5, pp. 481-490. 100. STECKE, KATHRYN E. and SOLBERG, JAMES J., (1985), "The Optimality of Unbalancing Both Workloads and Machine Group Sizes in Closed Queueing Networks of Multiserver Queues," Operations Research, Vol. 33, No. 4, pp. 882-910, July-August.

-25 - 101. STECKE, KATHRYN E. and TALBOT, F. BRIAN, (1983), "Heuristic Loading Algorithms for Flexible Manufacturing Systems," Proceedings of the Seventh International Conference on Production Research, Windsor, Ontario, Canada, August 22-24. 102. TANG, CHRIS S. and DENARDO, ERIC V., (1988), "Models Arising from a Flexible Manufacturing Machine, Part I: Minimization of the Number of Tool Switches," Operations Research, Vol. 36, No. 5, pp. 767-777, September-October. 103. TANG, CHRIS S. and DENARDO, ERIC V., (1988), "Models Arising from a Flexible Manufacturing Machine, Part II: Minimization of the Number of Switching Instances," Operation Research,Vol. 36, No. 5, pp. 778-784, September-October. 104. TARN, J.H. and TOMIZUKA, M., (1989), "On-Line Monitoring of Tool and Cutting Conditions in Milling," Journal of Engineering for Industry, Vol. 1 1 1, No. 3, pp. 206-212. 105. TAYLOR, FREDERICK W., (1907), "On the Art of Cutting Metals," ASME Transactions, Vol. 28, pp. 310-350. 106. TOMEK, PAVEL, (1986), "Tooling Strategies Related to FMS Management," The FMS Magazine, Vol. 5, pp. 102-107, April. 107. TRAPPEY, J. F. C., LIU, C. RICHARD, and CHANG, TED C., (1987), "Fuzzy Nonlinear Programming: Theory and Applications in Manufacturing," Report TR-ERC87-7, School of Engineering, Purdue University, West Lafayette IN. 108. VEERAMANI, D., UPTON, D. M., and BARASH, M. M., (1989), "Tool Management in Computer- Integrated Manufacturing," Working Paper, School of Industrial Engineering and Engineering Research Center for Intelligent Manufacturing Systems, Purdue University, West Lafayette, IN, 47907. 109. VINOD, B. and SABBAGH, M., (1986), "Optimal Performance Analysis of Manufacturing Systems Subject to Tool Availability," European Journal of Operational Research, Vol. 24, No. 7. pp. 398-409. 110. WAGNER, J. G. and BARASH, MOSHE M., (1971), "Study of the Distribution of the Life of HSS Tools," Transactions of the ASME, p. 1044, November. 111. WALAS, R. A. and ASKIN, R. G., (1984), "An Algorithm for NC Turret Punch Press Tool Location and Hit Sequencing," IIE Transactions, Vol. 16, No. 3, pp. 280-287. 112. WATANABE, TOHRU and FUJII, R., (1988), "Determining the Job Operation Speed and Schedule for Machine Tools in FMS by Forecasting using a Simulator and a System Performance Index," ASME Proceedings of the USA-Japan Symposium on Flexible Automation, Minneapolis, MN, pp. 751-756. 113. WAYNE, S.F. and BULJAN, S.T., (1989), "The Role of Thermal-Shock on Tool Life of Selected Ceramic Cutting-Tool Materials," Journal of the American Ceramic Society, Vol. 72, No. 5, pp. 754-760. 114. WHITEHOUSE, GARY E., (1973), System Analysis and Design Using Network Techniques, Prentice-Hall, Englewood Cliffs, NJ. 115. WHITNEY, CYNTHIA K. and GAUL, THOMAS S., (1985), "Sequential Decision Procedures for Batching and Balancing in FMS," Annals of Operations Research, Vol. 3, pp. 301 -316. 116. WICK, C., (1987), "Advances in Machining Centers," Manufacturing Engineering, pp. 24 -32, October. 117. YAOHUA, W., ZEMIN, P., and WANG, L., (1987), "Application of Surface Science to Metal Cutting Theory," Proceedings of the IX International Conference on Production Research, August, Cincinnati, OH, pp. 781-787. 118. YELLOWLEY, I., (1983), "A Fundamental Examination of the Economics of the Two-pass Turning Operation," International Journal of Production Research, Vol. 21, No. 5, pp. 617 -626. 119. ZOMPI, A., LEVI, R., and RAVIGNANI, G. L., (1979), "Multi-tool Machining Analysis, Part I: Tool Failure Patterns and Implications," Journal of Engineering for Industry, Vol. 101, pp. 230-241.

-26 - Table 1. Tool Related Research Areas DESIGN PLANNING CONTROL standardization of tool types [10,15,44] tool tracking technology [2,4,39] tool information requirements [30,36,43,83,116]....... TOOL LEVEL -—..............-............assignment of tool types to operations tool life [9,52] [10,25,27,49,53,73,76,77, economic process planning 78, 79,110,113,117] [13,24,26,29,27,32,33,35,39,46,45, 47,50,60,67,74,82,89,107,118,119] monitoring and control technology [*] tool magazine capacity [*] tool changing technology [*] SINGLE MACHINE LEVEL tool replacement strategy due to expected wear - [5,8,28,59,67] sequencing parts/scheduling tools [69,75,90,91,94,102,103,111] sequencing operations/assigning toolk to slots [75,91,94,111]................................ tool replacement strategy due to actual wear[54,88] adaptive control at one machine [32] numbe [*] tool lo [*] loadinj [*]...........-......................... SYSTEM LEVEL ----.... --- —-------- ----—..-... --- ar and type of machines production planning adaptive control strate [65,91] [104,112],ading & handling technology part type selection (batching vs. flexible) [1,51,75,95,96,98,115] g of duplicate tools cell grouping and facility loading [3,12,21,57,71,87,91,92,93,94, 95,97,98,115] tool change times and detailed scheduling [18,19,20,22,68,86] processing rate determination and bottleneck control [47,48,56,66,84,85] spares levels [*] allocation of spares [7,14,37,42,58,72,109] tool inventory control [11,30,36,40] egies [#]: reference number corresponding to related research [*]: open research issue