China
Africa
Explore chapters and articles related to this topic
Overview of Production Flow Analysis and Simplification Toolkit (PFAST)
Published in Shahrukh A. Irani, Job Shop Lean, 2020
Production Flow Analysis (PFA) provides an effective framework for analyzing the material flows at different levels of resolution in a factory. Historically, the four stages of PFA were implemented manually. That limited the detection and elimination of many instances of chaotic material flows in the existing factory. Therefore, PFAST (Production Flow Analysis and Simplification Toolkit (PFAST) was developed to implement a variety of computer algorithms (Figure 6.1) that help to partially automate the manual PFA methods. Thereby, the implementation of PFA can be done for large datasets in significantly shorter periods of time. Most of the algorithms in PFAST are versatile, and can be used to automate different stages of PFA, as shown in Table 6.1. Essentially, PFAST is an academic software that combines algorithms implemented in (i) facility layout software like FactoryFLOW (www.plm.automation.siemens.com) and Flow Planner (www.proplanner.com/en/products/flow_planner/) and (ii) statistical analysis software like SAS (www.SAS.com) and Minitab (www.Minitab.com). For example, if you Google “Cluster Analysis Freeware”, you will find software available online that is equivalent to two of the PR Analysis (Product-Routing Analysis) modules in PFAST – PR Analysis I and PR Analysis II – that help to determine the machine groups and part families to form manufacturing cells.
A new aggregation algorithm for performance metric calculation in serial production lines with exponential machines: design, accuracy and robustness
Published in International Journal of Production Research, 2021
Yishu Bai, Jiachen Tu, Mengzhuo Yang, Liang Zhang, Peter Denno
Production flow analysis, control, and optimisation are among the most important problems in manufacturing research and practice. To study these problems, production systems are typically modelled as queueing networks under various classes of assumptions and conventions on job arrival behaviour, machine reliability and processing capability, buffering constraint, etc. Regardless of the models used, system performance metric calculation is one of the most important problems in production systems research. To solve this problem, two approaches are generally considered: discrete-event simulation and analytical calculation. While the former becomes a popular choice in recent years thanks to the rapid advancement in computing powers, its shortcomings (e.g. random errors, relatively long computation/simulation time) are still difficult to overlook. On the other hand, while the analytical approach generally relies on much more computationally efficient algorithms, exact analysis is only feasible for small-scale systems (e.g. one- or two-machine lines). For larger systems, only approximations are possible. Over the last several decades, a great amount of research on the analytical calculation of steady-state performance of production systems have been carried out (see, for instance, monographs by Buzacott and Shanthikumar 1993; Askin and Standridge 1993; Papadopoulos, Heavy, and Browne 1993; Yao 1994; Perros 1994; Gershwin 1994; Altiok 1997; Curry and Feldman 2009; Li and Meerkov 2009 and review papers by Dallery and Gershwin 1992; Papadopoulos and Heavy 1996; Li et al. 2009). The recent development in this direction includes extensions of these methods to systems with machines having multiple failure/degradation states (e.g.Levantesi, Matta, and Tolio 2003; Tan and Gershwin 2009; Colledani and Gershwin 2013), multiple job types (e.g.Colledani et al. 2008; Zhao and Li 2014; Alavian, Denno, and Meerkov 2017), production control (Ambani, Meerkov, and Zhang 2010; Jia et al. 2016b), more complex structures (e.g. Li 2005; Jia et al. 2016a), etc. Moreover, analysis of production systems' transient performance has also attracted growing research attention since the last decade (see, for instance, Mocanu 2005; Zhang et al. 2013; Chen et al. 2013, 2016).