購(gòu)買(mǎi)設(shè)計(jì)請(qǐng)充值后下載，，資源目錄下的文件所見(jiàn)即所得，都可以點(diǎn)開(kāi)預(yù)覽，，資料完整，充值下載可得到資源目錄里的所有文件。。。【注】：dwg后綴為CAD圖紙，doc,docx為WORD文檔，原稿無(wú)水印，可編輯。。。具體請(qǐng)見(jiàn)文件預(yù)覽，有不明白之處，可咨詢QQ:12401814
AN AGENT-BASED SIMULATION OF A JIT MATERIAL HANDLING SYSTEM
Qi Hao and Weiming Shen
Integrated Manufacturing Technologies Institute
National Research Council, Canada
800 CoUip Circle, London, Ontario N6G 4X8, Canada
[qi.hao; weiming.shen]@nrc.gc.ca
Material handling is a loose loop in most assembly plants. Jiist-in-time (JIT) is amanagement philosophy that strives to eliminate sources of manufacturing waste by producing the right part in the right place at the right time. We propose to apply JIT principles to material handling in assembly plants. Material Kanbans are introduced as an effective means to control and balance the physical material/part flow in the plant Jloor. An agent-based simulation prototype is implemented using AnyLogic?. The flexibility of the agent-based approach facilitates the simulation of various "what-if" scenarios including different layout designs, objective parameters and dynamic situations in the plant floor.
1. INTRODUCTION
Material handling is a loose loop that is generally neglected in most production plants. From our observation, even in a well designed assembly line, in condition that the whole line is optimized in its layout, processes, buffering, scheduling, and operations, material handling is still laid outside of the scope of control. Managers spend their precious time hunting everywhere for missing parts and arranging their deliveries. They are unaware of material handling/delivery schedules and the related resource information (amount and utilization of resources, such as forklifts and drivers). As a result, material handling becomes the major barrier that results in low efficiency, production breakdowns, and low quality of a production system. Just-in-time (JIT) is a management philosophy that could improve profits and return on investment by reducing inventory levels, reducing variability, improving product quality, reducing production and delivery lead times, and reducing other costs (such as those associated with machine setup and equipment breakdown). The pull mechanism, especially introduced by the Kanban control of JIT manufacturing, enables an optimized production process that benefits from the cutting down of production resources. For a plant that already operates under a pull mechanism, material handling should also employ a pull mechanism rather than a MRP-based push mechanism.
This paper intends to propose a pull material handling system based on principles in JIT manufacturing. In such a system, materials transportation in the plant floor is considered as individual tasks. Material Kanban (M-Kanban) is introduced as a carrier of delivery tasks which is an effective means to control and balance the physical material handling flow in the plant floor. The main principle behind is that a task generated by a production station (cell) requires the occupation of an M-Kanban to be delivered. Another technology used in this material handling system is agent. Agent technology is evolved from the research domain of Distributed Artificial Intelligence in 1990s. From its emergence, agent technology is widely recognized as a promising paradigm for the next generation of design and manufacturing systems (Shen et al., 2001). In the JIT material handling simulation system, multiple agents are implemented to facilitate a collaborative problem solving environment. For example, each transportation vehicle is encapsulated as an agent so that it is manageable on its own parameters and behaviors, such as velocity, local schedule, and the associated scheduling, routing and conflict resolving ndes. The driver of a transportation vehicle can deactivate a vehicle from the system to take personal activities or when the vehicle malfunctions and needs a repair. Moreover, the allocation of transportation task is accomplished though the negotiation of a Kanban scheduling agent and a number of vehicle agents. With such capacities, the system is able to simulate very dynamic situations and get more accurate information of transportation resources in general.
The rest of this paper is organized as follows: Section 2 reviews the background knowledge and literature of this study; Section 3 identifies a sample JIT material handling problem and describes the corresponding requirements; Section 4 proposes an agent-based architecture of the JIT material handling system and discusses two major design aspects: production simulation and material handling simulation; finally. Section 6 draws our conclusions.
2. A TECHNOLOGY REVIEW
There are two classifications of production control systems, namely push and pull. Material requirement planning (MRP) systems and Kanban control systems are the two most popular implementations of the push and pull strategies respectively. In a push production, in order to buffer inaccurate forecasts, inaccurate lead times, inaccurate inventory records, variable production schedules or questionable bill of materials (BOMs), MRP generally incorporates safety lead times and safe stocks. However, in practice, MRP may result in a serious problem of excessive inventories (Shirk, 1998; Hopp and Spearman, 1996). Stock levels and lead times are amplified down throughout the supply chain, from the final distributor down to each hierarchy of suppliers.
In contrast, using a pull strategy, a JIT system uses underutilized capacity instead of buffer inventories to hedge against problems that may arise. Production is initiated in response to real customer orders and the removal of items from the final distributor buffers triggers production upstream to replenish exhausted inventories layer by layer. Krishnamurthy et al. (2004) quantitatively compares the performance of MRP and Kanban for a multi-stage, multi-product manufacturing system. They reached the conclusion that pull strategies are handicapped for manufacturing facilities that produce a number of different products with distinct demands and/or processing requirements, as well as for facilities that make highly engineered products in small batches (even one-of-a-kind) for their customers.
The ideal industries that JIT production applies include automobile because it is where the JIT concept originated. The automotive industry is characterized by low product variety, and high-volume production. In an automotive assembly line, although there are some sub-lines using push strategies (sometime it is called hybrid production), such as the body shop, paint shop, and engine line, however, once cars are lining up to be processed on the main assembly line, the production is under control of a pure pull mechanism. Buffers are set at offline sites of sub-lines to tickle uncertainties and better serve the optimized production rate of the main assembly line.
The concepts of JIT and Kanban are never new. JIT were firstly developed by Toyota in the 1950's and adopted in the United States in the 1980's. Lean manufacturing and lean enterprise, proliferating in western countries, are also evolved based on JIT principles. Many small and medium sized businesses have embraced these concepts along with some of the major corporations such as, Mercedes/Benz, Pratt & Whitney, Porsche and General Electric to name a few. Womack and Jones (1996) provide a thoughtful expansion upon their value-based business system based on the Toyota model. Along the way they update their action plan in light of new research and the increasing globalization of manufacturing, and they revisit some of their key case studies from the automotive, aerospace, and other manufacturing industries.
Many analytical, mathematical or experimental models are proposed to address the Kanban based operational planning and control issues (Berkley, 1992; Uzsoy and Martin-Vega, 1990). Simulation has been by far the methodology of choice in the majority of studies reported in the literature (Gupta and Al-Turki, 1998). Theoretically, the number of Kanbans and allocation of Kanbans in a system significantly affects the performance of a pull system. Instead of optimization of these Kanban arrangement which leads to a fixed number of Kanbans, Martins and Lewandrowski (1999) proposed a mathematical buffer stocks dimensioning approach using a dynamic kanban strategy. Gupta and Al-Turki (1998) compared the performance of a traditional kanban system (TKS) and a flexible kanban system (FKS). Through the simulation of two simple JIT models, they proved that FKS outperforms TKS under real-time manufacturing environments, such as sudden breakdown of a material handling system.
From application point of view, the researches of pull technologies could be classified in three categories: I) production control; 2) inventory management; and 3) supply chain management (Kusiak, 2000). Material handling is a topic being previously researched in the literature (Gupta and Al-Turki, 1998; Askin, 1999; Venkataramanaiah et al., 2001), however, they all deal with the automatic material (specifically, the Work-In-Process, which belongs to the production line itself) transfer problem between production cells. The material we emphasize here does not refer to the WIP going through the assembly line, rather, it is the supply fiow of material or parts subordinating to the main production line. None of known literature touched the topic of material handling from this aspect.
Moreover, in most plants, material inventories are only virtually under control of either a MRP II system or a Kanban system at the enterprise level, but not physically at the dynamic plant floor. Missing parts, wrong part delivered, parts not at right place at right time are common occurrence in almost all mainstream production plants, including GM, Ford, and Sterling Truck. Material handling is a frustrating problem faced by production managers. Production managers are feeling nervous everyday and are blamed for lack of ability to control the manufacturing process. As a result, analysis of material handling and dynamic simulation will be of great help to industries.
3. MATERIAL HANDLING SPECIFICATION OF A
SIMPLIFIED ASSEMBLY LINE
Simulation of material handling in a pull production setting is the primary purpose of this research. A Kanban-based material handling will be investigated to make it in line with the pull production line. For the convenience of a common understanding, a sample scenario is chosen as the background problem, as shown in Figure 1 （Figure slightly）.
The JIT-based material handling approach we proposed borrows similar principles from JIT-based production control and JIT-based inventory management in that: the right material is delivered from its inventory to the right production site, at the right time and in the right amount. Here, material transportation in the plant floor is considered as individual tasks. A task requires a material Kanban (M-Kanban) to be delivered. In figure 1, a material request signal is firstly generated by an assembly station running out of a part supply; after occupying a material Kanban, the request is broadcasted to a number of vehicles moving in the scope of a wireless network; then through negotiation, the task is confirmed by a vehicle and being delivered finally to the right station.
In our view, JIT material handling based on Kanban concept is not merely a pure event based system. In an event system, an event calls for a system response immediately; while in a Kanban based system, a generated event gets processed only after obtaining a physical object - M-Kanban. In other words, the processing (transportation) of a material requirement event holds until the system releases a M-Kanban and the event is qualified to occupy this free Kanban among all other events. Based on a Kanban control mechanism, we believe that the material handling system is able to reach a natural balancing of material requirements and transportation activities through delicate arrangement and management of Kanbans.
In addition to the standard function in a simulation environment, such as discrete event generation, simulation clock generation and an animation interface, this system should have special functional modules that try to model and simulate the dynamics in the plant floor. Three groups of functions make up the JIT material handling system: scenario generation, simulator, and graphical user interface. Scenario generation maintains a large variety of configuration information relating to: 1) static scenario, such as plant layout, moving tracks, and a central part inventory; 2) changeable simulation parameter, such as number of vehicles and number of Kanbans. Simulator is the core of the software in that it controls the simulation of both the production and material handling processes. Each transportation vehicle has a separate vehicle simulation module to make its own decisions, including task sequence, task schedule, moving control, loading andunloading operations, or even collision resolution decisions if necessary. Station simulation is a module to simulate a simplified production process taking into consideration only the consumption and replenishment activities of materials/parts at each station. Kanban Simulation manages buffering and circulation (life-cycle) of M-Kanbans. It makes two kinds of decisions 1) allocation of material requirement signals to empty M-Kanbans; 2) allocation of M-Kanbans to vehicles. In the simulator, basic simulation facilities such as timer and random number generator should be provided to simulate synchronize events or discrete events. Graphical User interfaces are supposed to timely update the graphical simulation, system event, exceptions, etc, and provide timely response upon users' requests for any kind of simulation and statistical data.
A distinctive feature of the designated JIT material handling simulation is "quasi-realism". The proposed simulation system possesses functions that surpass traditional simulations. The most distinguishing one is its ability to facilitate run-time reconfiguration. For example, the arrangement of assembly tasks to manufacturing stations could be adjusted during the execution of a simulation, so that the bottleneck (of the line) and system responses could be constantly changing. Another example is that each component is manageable not only in its configuration parameter, but also controllable in its behaviors individually (for example, each vehicle is able to make schedules, control status, and choose its own delivery route). In contrast, other simulation systems read a batch file before each simulation launch. It is difficult for people to analyze dynamic system behaviors by changing system configurations in separate simulation launches.
4. THE AGENT-BASED JIT MATERIAL HANDLING MODEL
We use agent technology to model major components in the JIT material handling simulation system. Agents are sophisticated computer programs that act autonomously on behalf of their users, collaborate across open and distributed environments, to solve a growing number of complex problems. There are four kinds of agents designed in the simulator.
1) Main Control Agent (MCA)
Main Control Agent (MCA) is responsible for simulation initialization, simulation termination, agent (thread) management, and thread synchronization. MCA also includes a timer and an event generator along with its main thread.
2) Station Agent (SA)
Station Agent (SA) is a running thread simulating material requirement activities at stations. It is dynamically generated and destroyed by the MCA. A simple production rate of the assembly line is set for all stations to consume their required parts in certain amounts. So, with the progress of one production step, the material balances at stations may reach the requirement levels or the urgent levels. In extreme occasion, materials may be exhausted which causes the whole assembly line to stop.
3) Kanban Schedule Agent (KSA)
Kanban Scheduling Agent (KSA) is a separate thread whose role is to take care of 1) the scheduling of material requirements to M-Kanbans, and 2) the assignment of M-Kanbans to vehicles. KSA is dynamically created anddestroyed by the MCA. It applies regular and emergent scheduling strategies. Regular scheduling is fulfilled by the negotiation carried out between KSA and participating VAs.
4) Vehicle Agents (VA)
Each Kanban assigned to a vehicle is confirmed by its VA and served by the VA through a series of actions. A vehicle agent is able to handle its local schedule, maintain its status, and controls its movement, repair, and resume actions. The threads for all vehicles are generated or destroyed by the MCA at the same time.
5. CONCLUSION
Based on requirements of industrial partners, material handling has been recognized as a loose loop in most assembly plants. Just-in-time is a pervasive paradigm implemented in nowadays manufacturing plants that strive to survive in a global competition. The principles of JIT bring forward an optimized production environment and a mechanism for waste less inventory replenishment. However, material handling problem, especially the material/part supply at the plant floor level, is seldom addressed in research literature and in practices. In this paper, we propose a material handling simulation system that applies JIT principles. Material Kanban is an entity that carries a material request and represents a material transportation task. An agent-based simulation environment is designed and a prototype system is implemented using AnyLogic. Many experiments will be performed based on the simulation model build for this purpose.
The JIT-Based material handling is expected to bring forward a number of advantages, such as optimization of stocks levels at production stations / cells, balancing of transportation load in the whole plant floor, obtaining manageability on material handling performance and accurate prediction and optimization of transportation resources. The major difference of this simulation from others is that the flexibility of the agent-based approach facilitates the simulation of various "what-if scenarios including different layout designs, objective parameters and dynamic situations in the plant floor.