Use this URL to cite or link to this record in EThOS:
Title: Rogue seasonality detection in supply chains
Author: Shukla, Vinaya
ISNI:       0000 0004 2751 3172
Awarding Body: Cardiff University
Current Institution: Cardiff University
Date of Award: 2010
Availability of Full Text:
Access from EThOS:
Access from Institution:
Supply chains face disturbances in the provision of goods and services to customers. A key disturbance which is endogenously generated from the nature of the ordering process used is rogue seasonality, which is characterised by orders and other supply chain variables showing cyclicality in their profiles and this cyclicality not present in exogenous demand. It is observed in many supply chains and is a cause of significant economic loss for entities in these chains. A useful way to manage rogue seasonality could be by detecting its presence and intensity in a system and then taking appropriate and timely action for its mitigation. Called "sense and respond", this approach has been used in various domains extensively, but its application in supply chain management has been limited. This thesis explores the application of this approach for managing rogue seasonality, with the findings from it particularly relevant for a context where many multiple echelon supply chains are being managed by a focal company. Multiple methods are used to analyse each of the rogue seasonality generation and detection components of the thesis. For understanding rogue seasonality generation, system dynamics simulations of single and three echelon linear and four echelon non linear (Beer game) systems are used. The impact of different demand processes and parameters, delays, order of delays, ordering processes, backlogs and batching on rogue seasonality is assessed. The simulation analysis is supported with empirical contexts from the steel and grocery sectors. The understanding gained on rogue seasonality together with the related contextual data is used in the sense or detection part of the thesis. The signature based approach, with the signature derived from clustering of time series data of variables is explored for detection, with the data represented in alternative domains such as amplitudes of Fourier transforms, autocorrelation function, coefficients of autoregressive model, cross correlation function and coefficients of discrete wavelet transform. The thesis determined the signature and index for detecting rogue seasonality. While the signature, which is based on the cluster profiles of the system variables indicates the presence of rogue seasonality, the intensity of rogue seasonality is indicated by the index. In a multi supply chain context, the index could be used to identify problematic supply chains from a rogue seasonality perspective and initiate appropriate management action. At present there is no measure for rogue seasonality and defining an index for the same constitutes a major contribution of this thesis. Among alternative time series representations, the frequency domain representation based on Fourier transform was found to be the most appropriate for deriving the signature and index. This is also a major contribution of the diesis, as the comprehensive assessment of time series representations carried out in this study has not been done in many studies across domains, and those that have done so, have not used any supply chain related data in the assessment. Finally, the framework for exploiting the index for managing rogue seasonality is proposed.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available