Atölyede iş çizelgeleme

Abstract

Bu çalışmada atölyede iç çizelgeleme konusu incelenmek üzere bir atölye veri tabanı kurulmuş ve sıralama problemlerine simülasyon yolu ile çözüm aranmıştır. Bilindiği gibi atölyede iş çizelgeleme dahilinde, akiş tipi ve sipariş tipi atölye olmak üzere iki atölye tipi incelenmektedir. Endüstride genelde rastlanılan atölye tipinin sipariş tipi atölye olduğu dikkate alınarak, çalışmada bu tip atölyelerde iş sıralamaya destek verebilecek bir atölye veri tabanı oluşturulmuştur. Hazırlanan veri tabanında malzemelere, bu malzemelere acılan işemirlerine, mamullerin ürün ağacı yapılarına, malzemelerin işlenmesi için gerekli rota bilgilerine ve bu malzemelerin islendiği iş merkezlerine ait temel bilgiler saklanmaktadır. Bu temel bilgilerden başka planlama ve maliyetlendirme acısından da gerekli bilgiler saklanabilmektedir. Geliştirilen bu modelde, en genel atölye hali incelenebilmekte; atölye, hazırlanan paket üzerinde simüle edilebilmektedir. Sıralama için. hazırlanan simülasyon modeli, günlük oluşan atölye hareketlerinin geri beslenmesi ile, fiili atölye durumunu tam olarak göz önüne almaktadır, öncelik tavsiyesinde bulunan sistem, sabit zaman artışlı, sonlu geliş kaynaklı, seri kuyruk sistemi simülasyonu yapmaktadır.

In this study» a job shop data base is established in order to find solutions to job shop scheduling problems. The basic aim of scheduling is to determine the order of sequence for processing a sets of jobs through several machines in an optimum manner. First of all, a detailed resource investment is done about subject. Therefore some parts of the study is dealt with the theory of scheduling. So it is necessary to give an abstract about the subject. Scheduling decisions allocate available capacity or resources (equipment, labor and space) to jobs, activities, tasks, or customers through time. Since scheduling is an allocation decision, it uses the resources made available by facilities decisions and aggregate planning. Therefore scheduling is last and most constrained decision in the hierarchy of capacity planning decisions. In practice, scheduling results in time-phased plan (or schedule) of activities. The schedule indicates what it is to be done, when, by whom, and with what equipment. Scheduling should be clearly differentiated from aggregate planning. Aggregate planning seeks to determine the resources needed, while scheduling allocates the resources made available through aggregate planning in the best manner to meet operations objective. Aggregate planning is done on time frame of about one year, while scheduling is done on time frame of a few months, weeks, or hours. Scheduling seeks to achieve several conflicting objectives: high efficiency, low inventories, and good customer service. Efficiency is achieved by a schedule which maintains high utilization of labor, equipment and space. Of course, the schedule should also seek to maintain low inventories, which-unfortunately-may lead to low efficiency due to lack of available material or high setup times. Thus a tradeoff decision in scheduling between efficiency and inventory levels is required. Customer service can be measured by the speed with which customer demands ars met, either through available stock or short lead times. Fast customer service is in conflict, too, with low inventories and high efficiency. The primary aim of scheduling is, therefore, to make tradeoffs between conflicting objectives so as to arrive at a satisfactory balance. It is not possible to treat scheduling for all types of operations as a single subject. In order to highlight the differences, scheduling can be classified by type of processes: line, intermittent, and project. The scheduling of line processes is required for both assembly lines and so-called process industries. For these line processes the scheduling problem is» at least partly, solved by the design of the process, since the product flows smoothly from one work station to another. For a single product made in one facility, there is no scheduling problem, because the flow of materials is completely determined by the design of the process. A scheduling problem exists only when multiple products are made in a single facility and thus compete for use of limited resources. In the intermittent process terminology ("shop", "job", "work center") comes from traditional manufacturing job shops. The concepts, however, apply equally to intermittent operations of all types, including factories, hospitals, offices, and schools. The intermittent scheduling problem is quite different from that for line processes. First of all, each unit flowing through an intermittent process typically moves along with many starts and stops, not smoothly. This irregular flow is due to the layout of the intermittent process by machine group or skills into work centers. As a result, jobs wait in line as each unit is transferred from one work center to the next. Work-in-process (WIP) inventory builds up and scheduling becomes complex and conflict. The intermittent scheduling problem can be thought of as a network of queues. A queue of WIP inventory is formed at each work center as jobs wait for the facilities to become available. These queues are interconnected through a network of material flows. The problem of scheduling intermittent processes is how to manage these queues. One of the characteristics of an intermittent operation is that jobs or customers spend most of their time waiting in line. The amount of time spent waiting will, of course, vary with the load on the process. If the process is highXy loaded, a job may spend as much as 95 percent of its toWl production time waiting in queues. Under these c1lccums tances, if it takes a week to actually process an ordeK, it will take 20 weeks to deliver it to the customer. On the other hand, if the process is lightly loaded, the waiting time will be reduced, since all the jobs flow through the process rapidly. Regardless of the load on the process, the challenge is to develop scheduling procedures that will effectively manage the flow of jobs, and work. There are a number of scheduling problems for intermittent processes! input-output analysis, loading, sequencing, and dispatching. Each of these problems and their interrelationships will be described below. The purpose of input-output control is to manage the relationship between a work center's inputs and outputs. Before discussing these relationships, a definition of terms will be helpful. Input is the amount work (jobs) arriving at a work center per unit of time. Input may be measured in such units money, number of orders, standart hours of work, or physical units (tons, meters, cubic xi meters) per unit of time. Load is the level of WIP inventory or back orders in the system. Load is the total volume of work still to be processed. It may be measured in the same units as input, but load is not expressed as a rate per unit of time. Output is the rate at which work is completed by work center. Output rate depends on both capacity and load. Capacity is the maximum rate of output which can be produced. Capacity is determined by a combination of physical factors and management policy. The relationships between this four terms may easily be visualized by the hydraulics analogy. Input is represented by the rate at which water flows into the tank and is controlled by the input valve. Load is represented by the level of water in the tank and corresponds to WIP inventory or back orders. Output is the rate at which water flows out of the tank. Capacity is the size of the output pipe, not the size of the tank. While capacity limits the maximum rate of flow, the actual output rate may be less than capacity if the water level is low. The proper way to control this tank system is to regulate the input valve so that the output and load achieve the proper levels. One cannot push more water through the tank simply by openning up the input valve, although this tactic is frequently attempted in factories and service operations. Once capacity is reached, the only way to get more output is to increase the size of the output pipe. Managers are well aware of the consequences of too little input: low machine utilization, idle labor, and high unit costs. What is often not understood are the consequences of too much input. In this case working capital will rise due to a larger WIP inventory, the average processing time to complete an order will increase as orders spend more time in queues, and system performance will generally decline. It is often better to control input by backlogging orders or even turning business away, if necessary, than to make futile attempts to push more through the system. One popular way attempt to increase output without increasing capacity is to expedite the work in process. Expediting is done by identifying critical jobs and rushing them through the facility. For example, an expeditor may place red tags on critical jobs which should be worked on first. This is a shortsighted solution which often does more harm than good. Every job expedited today may cause two jobs to be late tomorrow. Expediting destroys a smooth flow of work; it is the antithesis of planning. Even in the best-managed operations, a little expediting is may be needed when things go wrong; but expediting should not be substituted for proper planning, scheduling and control. One way to tell whether an operation is out of control is to count the number of jobs carrying rush stickers, red tags, or other expediting messages. Expediting indicates a failure to manage the relationships between input and output. xix In a factory, input is controlled by the releasing function. If the factory is in steady state, work should be released to the shop floor only in the same amount as the product is being shipped out the door. The release rate and shipment rate are frequently measured in standart labor hours of work or machine hours, whichever is the most critical resource. A job is not released to the factory simply because the materials are available or because some of the workers are idlej a like amount of work must be shipped out the other end of the factory. Only so much work is put into the system over a given period of time. If more jobs are accepted without a corresponding change in service capacity, jobs will simply wait longer to be served. Loading is a type of scheduling used to develop a "load" profile by work center. In loading, the total hours number of jobs is used to obtain a rough idea of when orders can be delivered or whether capacity will be exceeded. A precise schedule or sequence far each job is not developed. Loading uses an average waiting time for jobs at each work center to determine the progress of jobs through the facility. This is in contrast to detailed scheduling or sequencing, where job interference is taken into account and the waiting time of each job in each queue is precisely calculated. There are two types of loading: forward and backward. Forward loading begins with the present date and loads jobs forward in time. The processing time is accumulated against each work center, assuming infinite or finite capacity. Backward loading begins with the due date for each job and loads the processing-time requirements against each work center by proceeding backward in time. The capacity of work centers may be exceeded if necessary. Sequencing is concerned with developing an exact order (or sequence) of job processing. In sequencing, job interference and queuing times are computed by laying out a schedule for each job. An average queuing (or waiting) time is not assumed, as in the case of loading. One of the oldest sequencing methods, the Gannt chart, was proposed by Henry L. Bannt in 1917. The Bannt chart is a table with time across the top and a scarce resource, such as machines, people, or machine hours, along the side. This chart is constructed by first scheduling job 1 on all machines. The process of scheduling is continued until all jobs have been placed in the Bannt chart. After a Bannt chart has been constructed, it should be evaluated with respect to both job and machine performance. One way to evaluate machine performance is on the basis of time it takes to complete all work - the make? span. Another measure of the Gannt chart performance is machine utilization. A measure of job performance is the sum of the delivery times for each job. Minimizing this measure would also be equivalent to minimizing job waiting time since the two times are complementary. Xlll Bannt chart scheduling is more precise, since it considers job interference and computes the waiting times for each job. Nevertheless, Bannt chart scheduling becomes very complex when there are several machines in each work center, job times are not precise, and workloads are shifting. In this case, loading gives a good forecast of capacity needs or due-date predictions, without obtaining a precise schedule. Although these sequencing rules and others have a great deal of theoretical interest, they have not been applied much in practice. This is because real sequencing problems involve a great deal of variability in processing times, multiple objectives, and other complicating factors. Nevertheless, the rules are useful for gaining insight into scheduling problems and for suggesting heuristics which might possibly be of value in practice. In practice, schedules are difficult if not impossible to maintain because conditions often change; a machine breaks down, a qualified operator becomes ill, materials do not arrive on time, and so on. As a result, some operations are run without a detailed schedule. In this case, dispatch rules are used. A dispatch rule specifies which job should be selected for work next from among a queue of jobs. When a machine or worker becomes available, the dispatch rule is applied and the next job is selected. A dispatch rule is thus dynamic in nature and continually adjusts to changing conditions. Unlike a schedule, a dispatch rule cannot be out date, and it answers the worker's immediate questions "What should I do next?" In using dispatching rules, a shop-floor control system which updates the status of each job in process is needed. Using the feedback on job status, the supervisor receives a priority report each morning (or in real time). The priority report ranks all jobs waiting in that work center {or due to arrive that day) in priority order. These priorities can be calculated with any of the dispatching rules available. The supervisor then schedules work on the jobs in the work center on the basis of the priorities given. If the highest-priority job cannot be done (owing to machine breakdown, operator illness, or other reasons), the second job is done, and so on down the list. The priority list does not specify a rigid schedule and thus allows flexibility for local work-center conditions. Established job shop data base is designed to give a detailed look at the basic functions which form the foundation of the production control system. This system shows the basic shop floor information and how it can be used to make timely decisions, and monitor an order's progress in the shop. A computer-based Production Control System has several functions which are the basis for the rest of production control. Some of these functions are releasing the order, producing shop documentation, shop floor reporting, order status and control. xiv Simulation is used for determining which priority is the best in a time period. Because, whenever a scheduling problem cannot easily be structured for, or is not amenable to, the analytical approach, the simulation approach provides an effective means to analyze and solve the problem. That is, by simulating the scheduling system, we can study its various operating characteristics and then take appropriate actions to improve the system.