Modelling decision making in digital supply chains: insights from the petroleum industry

Purpose: This paper aims to overcome some of the limitations of previous works regarding automated supply chain formation (SCF). Hence, it proposes an algorithm for automated SCF using multiple contract parameters. Moreover, it proposes a decision-making mechanism that provides means for incorporating risk in the decision-making process. To better emphasize the features of the proposed decision-making mechanism, the paper provides some insights from the petroleum industry. This industry has a strategic position, as it is the base for other essential activities of the economy of any country. The petroleum industry is faced with volatile feed-stock costs, cyclical product prices and seasonal final products demand. Design/methodology/approach: The authors have modeled the supply chain in terms of a cluster graph where the nodes are represented by clusters over the contract parameters that suppliers/consumers are interested in. The suppliers/consumers own utility functions and agree on multiple contract parameters by message exchange, directly with other participant agents, representing their potential buyer or seller. The agreed values of the negotiated issues are reflected in a contract which has a certain utility value for every agent. They consider uncertainties in crude oil prices and demand in petrochemical products and model the decision mechanism for a refinery by using an influence diagram. Findings: By integrating the automated SCF algorithm and a mechanism for decision support under uncertainty, the authors propose a reliable and practical decision-making model with a practical application not only in the petroleum industry but also in any other complex industry involving a multi-tier supply chain. Research limitations/implications: The limitation of this approach reveals in situations where the parameters can take values over continuous domains. In these cases, storing the preferences for every agent might need a considerable amount of memory depending on the size of the continuous domain; hence, the proposed approach might encounter efficiency issues. Practical implications: The current paper makes a step forward to the implementation of digital supply chains in the context of Industry 4.0. The proposed algorithm and decision-making mechanism become powerful tools that will enable machines to make autonomous decisions in the digital supply chain of the future. Originality/value: The current work proposes a decentralized mechanism for automated SCF. As opposed to the previous decentralized approaches, this approach translates the SCF optimization problem not as a profit maximization problem but as a utility maximization. Hence, it incorporates multiple parameters and uses utility functions to find the optimal supply chain. The current approach is closer to real life scenarios than the previous approaches that were using only cost as a mean for pairwise agents because it uses utility functions for entities in the supply chain to make decision. Moreover, this approach overcomes the limitations of previous approaches by providing means to incorporate risk in the decision-making mechanism. © 2019, Emerald Publishing Limited.