Advances in Cycle Time Management

This chapter first defines the basic concepts, such as cycle time, step cycle time, and remaining cycle time, and discusses the relationship between these times. Subsequently, cycle time management is defined and its six activities are introduced: cycle time prediction, cycle time modeling and analysis, management support, cycle time control, cycle time reduction, and cycle time learning. Relevant cases or references are reviewed to illustrate the importance and implementation of each activity. From the review results, the most studied cycle time management activities include cycle time reduction, cycle time modeling and analysis, and management support. In addition, difficulties faced by cycle time management are also summarized. Recently, some advanced information technologies have emerged that can be used to overcome the difficulties of cycle time management and improve its effectiveness: Industry 4.0, big data analytics, edge computing, cloud computing and ubiquitous computing, explainable artificial intelligence, etc. How these advanced information technologies can help improve the effectiveness of cycle time management are discussed. References supporting these potential applications are also reviewed.

This chapter first highlights the importance of cycle time prediction by mentioning the management activities supported by cycle time prediction results. Existing cycle time prediction methods are then divided into six categories: statistical methods, production simulation, machine learning or deep learning methods, case-based reasoning, fuzzy modeling methods, and hybrid methods. For each type of methods, the mathematical background behind it is first explained, and then numerical examples and program codes are provided. In recent years, some advanced information technologies have emerged that can be used to enhance the performance of cycle time prediction: Industry 4.0, big data analysis, edge computing, cloud computing and ubiquitous computing, explainable artificial intelligence, etc. How these advanced information technologies cope with the challenges faced by traditional cycle time prediction methods, and some applications of these advanced information technologies in cycle time prediction are introduced. Numerical examples and program codes are given as well.

This chapter starts by defining the concept of cycle time modeling and analysis. Various cycle time modeling and analysis techniques and tools are then introduced. First, value stream mapping is applied to identify the operations of a job with the longest waiting times and should be improved first. Subsequently, after training the cycle time prediction model using machine learning or deep learning methods, a causal cycle time relationship analysis is performed to evaluate the impact of each job attribute on the cycle time forecast of a job, so as to find out the most important job attributes for the job. To this end, Shapely analysis and related tools are introduced with numerical examples and program codes. Subsequently, to explain and communicate the cycle time prediction mechanism and results, several explainable artificial intelligence (XAI) techniques and tools can be applied. First, for cycle time prediction problems involving big data, job classification is usually performed. Therefore, before predicting the cycle times, jobs need to be classified. To this end, traditional and XAI tools for explaining the job classification process and results are reviewed. Seven requirements that need to be met for excellent explanations are also listed. Subsequently, existing and XAI techniques and tools are applied to explain the cycle time prediction process and results, such as random forest-based incremental explanation. In addition, a systematic procedure is also established to confirm whether a trained cycle time prediction model conforms to domain knowledge, and on this basis, the improved random forest-based incremental explanation technique is proposed.

The results of cycle time prediction, analysis, and modeling can support various management activities, such as internal due date assignment, output projection, job sequencing and scheduling, competitiveness enhancement, customer relationship management, human resource management, and supply chain management. This chapter describes how to apply the results of cycle time prediction, analysis, and modeling to these management activities. Two management activities are especially focused on: internal due date assignment and output projection. Internal due date assignment refers to determining the date on which the manufacturing system can complete and deliver a potential order. The internal due date of a potential order can be determined by adding a certain allowance to the estimated completion time of the potential order. Advanced information technologies, such as cloud and edge computing, can also be applied to internal due date assignment. Therefore, a hybrid cloud and edge computing approach to enhance the effectiveness of internal due date assignment in advanced manufacturing systems is reviewed. In addition, imprecise output projection forecasts can lead to delivery delays, loss of potential orders, or unnecessary inventory accumulation, which is a critical challenge for manufacturing systems to overcome. Therefore, various machine learning or deep learning methods have been applied to output projection. Advanced information technologies, such as Industry 4.0 and explainable artificial intelligence, can also be applied to enhance the effectiveness and interpretability of output projection.

For many manufacturing systems, cycle time reduction is obviously the key to improving competitiveness. Therefore, cycle time reduction is regarded as the most critical and urgent task in production control. This chapter first summarizes the benefits of cycle time reduction on manufacturing systems from existing references and practices, such as reducing work-in-process levels, improving productivity, reducing unit costs, responding to market demand changes, and improving yield. This chapter also points out that cycle time reduction does not necessarily mean speeding up the execution of all operations. In addition, the cycle time reduction target of a product can be decomposed into the targets of its individual components based on the results of the bill of materials or material requirement planning. Subsequently, some methods for reducing cycle time are introduced, including more efficient job scheduling, yield improvement, controlling production conditions, 3D printing, leveraging cloud manufacturing capabilities, pull manufacturing, value stream mapping, and lean six sigma. Obviously, cycle time can be reduced in all aspects of system planning, control, and management. Finally, this chapter also provides numerical examples to illustrate the application of these cycle time reduction methods and discusses their potential effects.