An ARIMA supply chain model with a generalized ordering policy

Abstract

This dissertation develops models to understand and mitigate the bullwhip effect across supply chains. The models explain the bullwhip effect that is caused by using the up to target ordering policy in standard Material Requirement Planning (MRP) systems. In the up to target ordering policy, the orders are directly driven by actual demand oscillations. We develop the models in AutoRegressive Integrated Moving Average (ARIMA) forms for a single demand item in a tandem line supply chain model. Different from supply chain models in current literature that are based on the assumption of the up to target ordering policy with some specific ARIMA models and specific numbers of stages in supply chain, the up to target ordering policy models in this dissertation can be applied to any ARIMA demand, any ordering lead time, and any number of stages in supply chains to derive the closed form expressions of the variation in inventory and the variation in orders. In addition, we propose the generalized ordering policy in which the up to target ordering policy is a special case. The generalized ordering policy permits manufacturers to smooth orders with the guaranteed stationary inventory in which smoothing orders is regarded as an effective way to mitigate the bullwhip effect. With the generalized ordering policy, manufacturers can control the tradeoffs between the variation in inventory and the variation in differencing orders which is stationary due to differencing. The generalized order models can be applied to any ARIMA demand, any ordering lead time, and any smoothing period. Two special cases of the v generalized ordering policy are also illustrated. One is the previously mentioned up to target ordering policy that minimizes the variation in inventory. Another is the smoothing ordering policy that minimizes the variation in differencing orders. We also provide generic formulas to determine the optimal smoothing weights in the smoothing ordering policy for ARIMA(p, 0, q) and ARIMA(p, 1, q) orders. Finally, this dissertation introduces the bounded MRP following the rate based planning concept. We propose a simulation based technique to set the bounds into standard MRP systems for exponential smoothing or ARIMA(0,1,1) demand. With this bounded MRP, we can mitigate the bullwhip effect and reduce the conflict between production planning and infeasible capacity planning.

Demand forecasting at retail stage for selected vegetables: a performance analysis

“…Researches have been conducted on forecasting for various supply chain commodities and situations for every stage (Zhang and Qi, 2005;Ali et al, 2009;Guanghui, 2012;Eksoz et al, 2014;Ruiz-Aguilar et al, 2014;Wang and Xu, 2014;da Veiga et al, 2016). A plethora of research studies are initiated with conventional and time-series forecasting techniques such as moving average, exponential smoothing, linear regression, hybrid regression models, state space and auto-regressive integrated moving average (ARIMA) (Gilbert and Chatpattananan, 2006;Ramos et al, 2015), seasonal ARIMA (SARIMA) and quantile regression (QR) hybrid model (Arunraj and Ahrens, 2015) for various products and domains at various stages of the supply chain following different linear and non-linear trends, namely seasonality, promotions, stock-outs, product substitution, perishability and non-perishability. There are various studies found on random forests for species distribution (Araujo and New, 2006), multivariate decision trees (Sok et al, 2016), forecasting for solar radiation (Voyant et al, 2017) and healthcare sector (Rubin et al, 2018), etc.…”

Investigating the impact of inventory inaccuracy on the bullwhip effect in RFID-enabled supply chains using colored petri nets

“…Verification of modeled system properties arises with multi-agent systems (MAS) modeling, such as shared resource conflicts, or deadlocks, because of the distributed nature of MAS. Besides simulation modeling, analytical models have been used to study the impact of inventory policies on the bullwhip effect (Chen et al , 2000; Gilbert and Chatpattananan, 2006; Dai et al , 2016).…”

A study on behaviour of bullwhip effect in (R, S) inventory control system considering DWT-MGGP demand forecasting model

“…According to Lee et al (1997), demand forecasting, order batching, price fluctuations, supply shortages and non-zero lead-time causes BWE within a supply chain. Generally, demand of a product follows a time series pattern; hence, different time series models like Behaviour of bullwhip effect 385 autoregressive (AR) (Luong, 2007;Luong and Phien, 2007), moving average (MA) (Chen et al, 2000a;Hong and Ping, 2007;Ma et al, 2013), exponential moving average (EMA) (Chen et al, 2000b), exponentially weighted moving average (EWMA) (Hong and Ping, 2007), autoregressive moving average (ARMA) (Zhang, 2004;Duc et al, 2008a;Duc et al, 2008b;Bandyopadhyay and Bhattacharya, 2013) and autoregressive integrated moving average (ARIMA) (Gilbert, 2005;Gilbert and Chatpattananan, 2006) models are proposed to predict demand to reduce the BWE through regulating different model parameters such as AR or MA coefficient and associated lead time between the supply chain entity. However, it is difficult to control the model parameters.…”