Performance evaluation and enhancement of multistage manufacturing systems with rework loops

The phenomena of machine failures, defects, multiple rework loops, etc., results in much difficulty in modeling rework systems, and therefore the performance analysis of such systems has been investigated limitedly in the past. We propose an analytical method for the performance evaluation of rework systems with unreliable machines and finite buffers. To characterize the rework flow in the system, a new 3M1B (three-machine and one-buffer) Markov model is first presented. Unlike previous models, it is capable of representing multiple rework loops, and the rework fraction of each loop is calculated based on the quality of material flow in the system. A decomposition method is then developed for multistage rework systems using the proposed 3M1B model as one of the building blocks. The experimental results demonstrate that the decomposition method provides accurate estimates of performance measures such as throughput and Work-In-Process (WIP). We have applied this method to several problems, such as the determination of the optimal inspection location and the identification of bottleneck machines in rework systems.     

Dynamic production control in parallel processing systems under process queue time constraints

This research examines the production control problems in two-station serial production systems under process queue time (PQT) constraints. In these serial production systems, all jobs must be processed at a fixed order in the upstream and downstream stations. There are multiple machines in both stations, and all machines are subject to random machine failures. In the downstream queue, the sum of waiting and processing time for each job is limited by an upper bound. This upper bound of time is called the PQT constraint. Violation of the PQT constraint causes high rework or scrap costs.     

Overlapping decomposition: a system-theoretic method for modeling and analysis of complex manufacturing systems

A system-theoretic method, referred to as overlapping decomposition, is presented to evaluate the performance of a complex manufacturing system with assembly, parallel, rework, feedforward, and scrap operations. The idea of the method is to decompose the complex system into a set of serial production lines, with the first or last machines of each serial line overlapped with another line, and to modify the parameters of overlapping machines to accommodate the effects of machines and buffers in other lines. Iterative procedures are introduced to estimate the system production rate. The convergence of the procedures and the uniqueness of the solutions are proved analytically and the accuracy of the estimates is evaluated numerically. Note to practitioners - Many large volume manufacturing systems consist of complex operations, for instance, assembly, disassembly, rework loop, parallel lines, feedforward lines, scrap, etc. Development of a performance evaluation method to provide fast and accurate analysis of system throughput is important for design and continuous improvements. This paper introduces a system-theoretic method, referred to as, overlapping decomposition, to analyze the performance of such complex manufacturing systems. The complex system is decomposed into overlapped serial production lines and modifications are introduced to accommodate the coupling effects among all these lines. From the theoretical point of view, this paper presents the proofs of the convergence of recursive procedures and the uniqueness of solution. From the application point of view, the method has obtained good results in solving practical problems on the factory floor.     

Optimization of the finite production rate model with scrap,rework and stochastic machine breakdown

This study employs mathematical modeling along with a recursive searching algorithm to determine the optimal run time for an imperfect finite production rate model with scrap, rework, and stochastic machine breakdown. In real-life manufacturing systems, generation of defective items and machine breakdown are inevitable. The objective of this paper is to address these issues and to be able to derive the optimal production run time. It is assumed that the proposed manufacturing system produces defective items randomly, a portion of them is considered to be scrap, and the other portion can be repaired through rework. Further, the proposed system is subject to random breakdown and when it occurs, the abort/resume (AR) policy is adopted. Under such an inventory control policy, the production of the interrupted lot will be resumed immediately when machine is fixed and restored. Mathematical modeling along with a recursive searching algorithm is used for deriving the replenishment policy for such a realistic production system.     

Dynamic production system diagnosis and prognosis using model-based data-driven method

Advanced manufacturing systems are becoming increasingly complex, subjecting to constant changes driven by fluctuating market demands, new technology insertion, as well as random disruption events. While information about production processes has been becoming increasingly transparent, detailed, and real-time, the utilization of this information for real-time manufacturing analysis and decision-making has been lagging behind largely due to the limitation of the traditional methodologies for production system analysis, and a lack of real-time manufacturing processes modeling approach and real-time performance identification method. In this paper, a novel data-driven stochastic manufacturing system model is proposed to describe production dynamics and a systematic method is developed to identify the causes of permanent production loss in both deterministic and stochastic scenarios. The proposed methods integrate available sensor data with the knowledge of production system physical properties. Such methods can be transferred to a computer for system self-diagnosis/prognosis to provide users with deeper understanding of the underlying relationships between system status and performance, and to facilitate real-time production control and decision making. This effort is a step forward to smart manufacturing for system real-time performance identification in achieving improved system responsiveness and efficiency.     

Inspection allocation in manufacturing systems using stochasticsearch techniques

Quality is the hallmark of a competitive product. It is necessary to use inspection stations to check product quality and process performance. In this paper, the authors are concerned with the problem of location of inspection stations in a multistage manufacturing system. The authors present two stochastic search algorithms for solving this problem, one based on simulated annealing and the other on genetic algorithms. These algorithms are developed to determine the location of inspection stations resulting in a minimum expected total cost in a multistage manufacturing system. The total cost includes inspection, processing and scrapping cost at each stage of the production process. A penalty cost is also included in it to account for a defective item which is not detected by the inspection scheme. A set of test examples are solved using these algorithms. The authors also compare performance of these two algorithms     

A multi-stage production-inventory model with learning and forgetting effects,rework and scrap

This paper studies a serial production line where a proportion of defective items is produced at each stage. Defective units enter a rework process, which is imperfect as well. Twice defective items are scrapped. This paper also considers learning and forgetting in production and rework processes and studies how the number of shipments of a lot from a production stage to the next influences the overall performance of the system. A model for a multi-stage production-inventory system is developed and optimized against an aggregate performance measure of four partial measures that are based on production time, process yield, in-process inventory and shipment frequency. Each of these partial performance measures is weighed by the system’s decision maker in accordance to importance. The numerical results show how the values of learning rates, weights assigned to the partial performance measures and the number of production stages influence the overall performance of the system.     

Product Completion Time Prediction Using A Hybrid Approach Combining Deep Learning and System Model

The prediction of product completion time is critical in real time production scheduling and control to achieve customer demand satisfaction. However, it is a challenging task due to the increasing complexity of production systems and greater diversity of products. The recent advancement in data-driven approach and machine learning algorithms have provided unprecedent opportunities to tackle such problems that otherwise very difficult to solve using conventional methods in the manufacturing industry. However, most existing studies on product completion time prediction adopt a purely data-driven approach while ignoring the prospect of integrating domain knowledge in their machine learning models. In this paper, we propose a hybrid approach to predicting product completion time by combining the strengths of both machine learning techniques and analytical system model. A mathematical model for multi-product serial production line is proposed to describe the real-time dynamics of the system. With this model, the strict lower bound of product completion time can be efficiently computed for given system status, where the lower bound represents the least possible product completion time when assuming no random downtime in the system. Instead of directly predicting the product completion time, a deep learning model is developed to only predict the distance between the lower bound and actual product completion time. Guided by properties of production system, we discover a recurrent sequence in the prediction problem by modeling each machine and product as recurrent units. The Long Short-Term Memory (LSTM) method, a prominent variant of recurrent neural network (RNN), is used to combine with the system model to predict the product completion time in a real-time fashion.     

基于退化机器模型的分布式柔性生产系统:性能分析、任务调度及预测性维护

在近些年的制造环境中,由于市场对多品种、小批量定制产品需求的增加,生产制造更加深入地向着柔性方向发展.如何利用现有资源,提高生产效率,实时地对系统性能进行评估与预测,并对基于小批量生产的实时调度进行优化改进,在分布式柔性生产系统中具有重要的研究意义.因此,基于退化机器模型的多批次串行生产线的性能进行分析,并对分布式生产系统进行任务调度及预测性维护.具体地说,对于具有退化机器模型及有限容量缓冲区的生产系统,首先采用马尔科夫分析方法建立数学模型;随后,提出精确方法来计算此生产系统模型实时的性能指标,并针对该模型下的调度问题,设计最优完成时间指标优化算法;此外,提出基于退化机器模型的预测性维护策略以减少完成时间;最后,通过数值实验验证该算法的可行性和有效性.     

Reducing rigidity by implementing closed-loop engineering in adaptable design and manufacturing systems

Traditional engineering design and manufacturing systems function as an arrangement of individual engineering activities with predefined rigid interfaces. In contrast, agile design and manufacturing engineering strategies require flexible processes that can be adapted rapidly for fast response to dynamic market demands without compromising the cost and quality of the product. In an attempt to address this deficiency of traditional systems, this paper proposes and explores a novel model of product lifecycle management (PLM) in which instead of studying engineering activities exclusively with respect to a temporal variable, the processes are deliberately managed to form closed loops of two or more activities with respect to two independent dimensions: time and rigidity. The model is referred to as closed-loop engineering (CLE). The CLE model eliminates the need to specify solid interfaces between the engineering activities. Instead, flexible interfaces are generated by real-time analysis of respective actions and the reactions between them. In order to study the performance and efficiency of CLE, case studies in three different levels of sub-activity, inter-activity and infra-activity loops in the fields of design, manufacturing and inspection are presented. The implementation results revealed that individual closed loops of different engineering activities in the product development process can significantly enhance the quality of the final products. Also, the flexibility of the interfaces between the activities results in adaptable production systems.     

Robust planning in optimization for production system subject to random machine breakdown and failure in rework

This study is concerned with robust planning in optimization, specifically in determining the optimal run time for production system that is subject to random breakdowns under abort/resume (AR) control policy and failure in rework. In most real-life production processes, generation of defective items and breakdowns of manufacturing equipment are inevitable. In this study, random defective rate is assumed and all manufactured items are screened. The perfect quality, reworkable and scrap items are identified and separated; failure-in-rework is assumed. The system is also subject to random machine breakdown; and when it occurs, the AR policy is adopted. Under such policy, the production of the interrupted lot will be immediately resumed when the machine is restored. Mathematical modeling and derivation of the production-inventory cost functions for both systems with/without breakdowns are presented. The renewal reward theorem is used to cope with the variable cycle length when integrating cost functions. The long-run average cost per unit time is obtained. Theorems on convexity and on bounds of production run time are proposed and proved. A recursive searching algorithm is developed for locating the optimal run time that minimizes the expected production-inventory costs. A numerical example with sensitivity analysis is provided to give insight into the optimal operational control of such an unreliable system.     

Contribution of simulation to the optimization of inspection plans for multi-stage manufacturing systems

In this paper we develop a new mathematical model to optimize inspection plans for multi-stage manufacturing systems with possible misclassification errors. The presented model minimizes total inspection related costs while still assuring a required output quality. Because of the complexity of the proposed mathematical model, a simulation algorithm is presented to model the multistage manufacturing system subject to inspection and to estimate the resulting inspection costs. We use the popular Arena simulation software to implement our simulation algorithm and then we utilize OptQuest, Arena’s built-in optimization package, to find the optimal inspection plan. Finally, a numerical example is presented and simulation experiments are also conducted in order to examine the effects of several model parameters.     

Quantitative automated inspection of standard parts using machine vision

On line automated visual inspection for quality and process control is becoming a very important requirement in an automated manufacturing environment. This paper examines the possibility of real-time inspection of standard part using machine vision.     

Data-driven dynamic bottleneck detection in complex manufacturing systems

Production (throughput) bottlenecks are the critical stations defining and constraining the overall productivity of a system. Effective and timely identification of bottlenecks provide manufacturers essential decision input to allocate limited maintenance and financial resources for throughput improvement. However, identifying throughput bottleneck in industry is not a trivial task. Bottlenecks are usually non-static (shifting) among stations during production, which requires dynamic bottleneck detection methods. An effective methodology requires proper handling of real-time production data and integration of factory physics knowledge. Traditional data-driven bottleneck detection methods only focus on serial production lines, while most production lines are complex. With careful revision and examination, those methods can hardly meet practical industrial needs. Therefore, this paper proposes a systematic approach for bottleneck detection for complex manufacturing systems with non-serial configurations. It extends a well-recognized bottleneck detection algorithm, “the Turning Point Method”, to complex manufacturing systems by evaluating and proposing appropriate heuristic rules. Several common industrial scenarios are evaluated and addressed in this paper, including closed loop structures, parallel line structures, and rework loop structures. The proposed methodology is demonstrated with a one-year pilot study at an automotive powertrain assembly line with complex manufacturing layouts. The result has shown a successful implementation that greatly improved the bottleneck detection capabilities for this manufacturing system and led to an over 30% gain in Overall Equipment Effectiveness (OEE).     

Simulation of a production line system with machine breakdowns using network modeling

In a manufacturing system, several factors can affect the reliability of the system in producing expected output levels, including resource input rates, labor rate variability, product quality, and machine failures. Although all of these factors will be reflected in the modeling process presented in this paper, the primary focus will be on the effect of machine breakdowns on system output. Network modeling and simulation with Q-GERT is the vehicle of analysis that is employed. This technique will be demonstrated via a realistic case example encompassing a complex production system consisting of several assembly lines, each containing several machines. The statistical results of the simulation of the example system are presented and discussed. In addition, examples of how the simulation model can be used to test changes in machine repair times and breakdown rates will be presented.     

Throughput analysis in automotive paint shops: a case study

In this paper, an overlapping decomposition method is used to estimate the throughput of a production system with multiple rework loops. The idea of the method is to decompose the system into a couple of serial lines and modify the parameters of overlapping machines to accommodate the effects of other lines. Using this method, the throughput of an automotive paint shop is analyzed and continuous improvement procedures are described.

Note to Practitioners-Painting is an important element of vehicle production. A paint shop has been a system bottleneck in many automotive assembly plants due to its complexity. Fast and accurate analysis of its system throughput is important for design and continuous improvements. This paper introduces an iterative method to analyze the performance of paint shop type production systems, i.e., systems with multiple rework loops. The method has obtained good results in both theoretical study and applications on the factory floor. In addition, a case study at an automotive paint shop is introduced and continuous improvements process to identify and eliminate system bottlenecks is described. The presented method can also be applied to other production systems with similar structures.     

Analysis of a linear walking worker line using a combination of computer simulation and mathematical modeling approaches

It has become increasingly important in the last few years to develop rapid, dynamic, responsive and reconfigurable manufacturing processes and systems. This is because manufacturing enterprises are now being forced to develop and constantly improve their production systems so that they can quickly and economically react to unpredictable conditions such as varying production volumes and product variants with small lot size, high quality and low costs. One effective method to achieve this is to create a more flexible, highly skilled and agile workforce capable to perform multiple or all the required tasks in a production area where the system can be reconfigured easily as needed to accommodate changes of production requirement on a daily or weekly basis.This paper presents a study of a so-called linear walking worker assembly line based on a combination of computer simulation and mathematical analysis. The linear walking worker assembly line is a flexible assembly system where each worker travels down the line carrying out each assembly task at each station; and each worker accomplishes the assembly of a unit from start to finish. This design attempts to combine the flexibility of the U-shaped moving worker assembly cell with the efficiency of the conventional fixed worker assembly line. The paper aims to evaluate one critical factor of in-progress waiting time that affects the overall system performance providing a dynamic simulation outlook as well as an insight into the mechanism of such a flexible and reconfigurable manufacturing system.     

Optimization of quality costs

The main goals of quality management in all industries are customer satisfaction by delivery of defect-free products and the radical reduction of defect rates and quality costs in the production. Controlled technological processes are the most important way to reach these goals. These principles are standard in mechanical engineering and are in use with great success. The properties in electronics production are different from the properties in mechanical engineering. During the assembly of electronic devices, processes are sensitive to the influences (environmental, parameter variation, etc.), that act on these processes. These influences are very strong, especially in the production of small batches of assemblies and a high mix of products. Processes can become uncontrolled, the defect rates can rise, and it is necessary to have inspection and repair processes after the different technological processes. But what about the quality costs? Is it sensible to do these quality control steps from the economical point of view? What is the right inspection strategy—no inspection or 100% inspection or statistical process control? To answer these questions new simple and powerful quality cost models have been developed at the Electronics Technology Laboratory and are in use in electronic producing industries. The quality costs are the “measurement system” to compare different inspection strategies with each other. The costs are calculated by the use of mathematical models—the quality cost models. To analyze and optimize the quality processes of a complete production line we use the method “Dynamic Programming”, developed by the American scientist, Bellman in the early 1950s of the 20th century.     

Manufacturing System Design to Improve Quality Buy Rate: An Automotive Paint Shop Application Study

Manufacturing system design has an impact on product quality. In this paper, we investigate this impact through an application study at an automotive paint shop. Specifically, for repair and rework systems in paint operations, we develop a model to quantify paint quality [in terms of quality buy rate (QBR)] as a function of repair capacity. We show that the QBR can be improved and unnecessary repaints can be reduced by increasing the repair capacity. Note to Practitioners-Manufacturing system design and quality management are important in many manufacturing industries. Although they have attracted substantial research effort, little attention has been paid to address the coupling or interactions between system design and product quality. Empirical evidence and analytical studies have shown that manufacturing system design does impact quality. In this paper, through an application study at a repair and rework system in an automotive paint shop, we show that paint quality, as measured by the quality buy rate, can be improved by designing the system more effectively. Similar problems are also often encountered in other manufacturing systems. Results obtained in this work, along with other results, demonstrate both the theoretical and practical importance of the analysis of manufacturing system design on product quality, and suggest a largely unexplored, but promising research area     

Decomposition program for allocating interstage storage along an automatic production line

A computer program called decomposition is presented for generating an approximate solution for the output rate of a serial, synchronized, automatic production line. Each pair of consecutive machines in the production line is separated by a storage buffer. Each machine is assumed to be subject to random failure if it is up or random repair if it is down each time that parts are transferred. The program can be used to allocate a fixed total storage capacity to the buffers between machines so as to maximize the steady state output rate. Decomposition is an iterative algorithm which requires only a small fraction of the computer time and memory needed by an exact solution. Although the accuracy of the approximate solution decreases as the line is lengthened, the accuracy increases as the total storage capacity of the production line increases. Approximate solutions can be obtained for lines which are too large to be treated by exact methods.