Optimization of order allocation problem in a multiple buyer-supplier network: a multi-objective and multi-criteria approach

• A five-step framework developed for multi-agent supply chain optimization. • A hybrid multicriteria–multi-objective (MAUT–CP) model proposed for order allocation in multiple buyer-supplier networks. • Utility functions capture both buyer and supplier preferences within the network. • Equity thresholds ensure the fair distribution of performance across all agents. • Real-world case demonstrates a 45.6 % improvement in network order allocation performance. • The proposed framework supports strategic decision-making in collaborative supply chains. Optimizing order allocation in decentralized networks with multiple buyer-supplier network is essential for enhancing competitiveness, sustainability, and collaboration throughout supply chains. However, the existing literature rarely addresses this challenge while simultaneously incorporating the business perspectives of both buyers and suppliers. To fill this gap, this study proposes a novel decision-support framework to solve the order allocation problem within a multi-agent supplier network that trades a common input. The proposed model integrates the Multi-Attribute Utility Theory (MAUT) with Compromise Programming (CP), a multi-objective optimization technique. Utility functions were elicited to capture individual preferences, and an aggregation process was employed to support collective decision-making. The framework was validated in a real-world case involving the Dairy Local Productive Arrangement (LPA) in Pernambuco, Brazil, which comprises small- and medium-sized milk producers and dairy processors. The results indicate a 45.62 % improvement in the CP metric compared to the initial status quo, evidencing that the final solution moves substantially closer to the ideal performance levels anticipated by each decision-maker. Furthermore, the introduction of a constraint limiting performance equity differences to 10 % resulted in a more balanced distribution of benefits, with only minimal reduction in overall network efficiency. These findings highlight the framework’s capacity to promote fairness without significantly compromising performance, offering valuable insights for collaborative decision-making in multi-agent supply networks.