A cloud computing architecture for supply chain network simulation

Abstract

This paper presents a Cloud Computing Architecture For Supply Chain Network Simulation (CCAFSCNS). The structure and elements of the CCAFSCNS as well as the relations among elements are introduced. The purpose of the architecture is to facilitate the distributed simulation of large-scale multi-echelon supply chains with an arborescent structure. The simulator in the CCAFSCNS permits the user to specify the network structure, the inventory stocking policies and demand characteristics so that supply chain performance can be estimated (e.g. average inventory on hand, average fill rates, average backorders, etc.) for each stock-keeping-unit. A prototype system that implements the CCAFSCNS is presented and illustrated with three example use cases. In addition, this paper discusses the time issues associated with the cloud computing solution and shows that the cloud computing solution can significantly shorten the simulation time. INTRODUCTIONThe performance of a supply chain impacts the competitive advantage of companies within the chain. How to coordinate the logistics activities in a supply chain, such as inventory control and transportation management, determines performance. Realistic multi-echelon supply chain networks make the analysis by mathematical models problematic for two main reasons: (1) the scale of the problem can be large, and (2) the complicated behaviors of different participants, such as manufacturers, retailers and customers. In addition, decision makers may need system performance during each planning stage to make step-by-step decisions. Therefore, simulation is an excellent tool to assist in supply chain analysis since it can (1) predict supply chain performance, (2) allow users to track the system performance at each time period, (3) perform what-if analysis for specific scenarios, and (4) compare alternative system configurations. One disadvantage of simulation methods is that it may be very time consuming to get statistically accurate results. When the multi-echelon supply chain network contains thousands of stock keep units (SKUs), it might take hours to run a single replication. In order to control the variance or half-width of the statistics generated by the simulation, many replications may be needed. Thus, if a single computer is used to simulate the multi-echelon supply chain network, it may take several hours or days to get the results. The purpose of this paper is to discuss the design of a cloud computing architecture that facilitates the execution of large-scale multi-echelon supply chain network simulations.In this paper, a background section introduces supply chain modeling and supply chain simulation tools. Then, section 3 illustrates the Cloud Computing Architecture For Supply Chain Network Simulation (CCAFSCNS). The structure and elements of CCAFSCNS as well as the relations among elements will be introduced in detail. A prototype system that implements the CCAFSCNS is then presented while 978-1-4673-4781-5/12/$31.00 ©2012 IEEE

Supply Chain Simulation As A Service To Increase Adaptation Capability In Manufacturing

“…This platform allows researchers to disseminate their research and replicate others researchers' artefacts in order to reproduce, enhance, and compare research findings and validity of research results. Rosetti et al (Rossetti et al 2012) presented a prototype architecture, called Cloud Computing Architecture For Supply Chain Network Simulation (CCAFSCNS) that directly deploys simulation on the Cloud to support supply chain networks. The framework allows distributed simulation of large-scale multi-echelon supply chains with an arborescent structure.…”

Distributed Approaches to Supply Chain Simulation

“…Studies which provide a valuable solution to the imposed problem are described in: papers which use DS to speed up the model execution time of the previous monolithic DES model [e.g. (Katsaliaki et al, 2009)] or to enable the execution of a large-complex system (Rossetti and Chen, 2012); a paper which applies DS to the manufacturing and installation of railways switch point assemblies and shows that the improved communication among simulation federates makes the scheduling tool more responsive to the dynamic needs of production and installation, increasing on-time deliveries and reducing the cumulative delays on late deliveries (Bandinelli et al 2004); research which refers to the initialization of a distributed logistics simulation without the need of warm-up period (Dalal, 2003) and two papers which use HLA to achieve reusability, interoperability, data hiding and evaluation of several management policies for decision-making in an extensive aerospace supply chain (Bruzzone et al2005) and in the emergency care (ambulances-hospitals) (Anagnostou & Taylor, 2017). Taylor et al (2009b) report on a study conducted at John Deere & Company, a manufacturer for agricultural, forestry, and construction equipment.…”

Distributed simulation: state-of-the-art and potential for operational research

“…There are several examples where systems have been developed to speed up simulation, some of which use some form of parallel optimization (Frank, Laroque, & Uhlig, 2013;Zhang & Anosike, 2012). There is also some evidence of the use of e-Science / e-Infrastructure architectures to implement this (Kiss et al, 2014;Rossetti & Chen, 2012).…”