Digital twin-enabled resilient coconut supply chains

Abstract

Agricultural supply chains are increasingly exposed to complex and interdependent risks, yet risk management in perennial crop systems remains largely reactive. This research addresses this gap by developing a Digital Twin (DT)-enabled model to enhance resilience in the Sri Lankan coconut supply chain, with a specific focus on production-stage risks. Three objectives guide the study: (1) identify the critical production risks influencing coconut yield, (2) design a comprehensive DT framework for proactive risk management, and (3) implement and validate a functional DT model capable of early risk detection and mitigation. Structured content analysis of recent literature and semi-structured interviews with domain experts from prominent national institutions were employed to identify and priorities production risks. These were classified into three primary categories: weather and climate variability, pests and diseases, and soil conditions. Building on these insights, a Digital Twin framework was developed using Azure Digital Twins, integrating physical asset representation, data ingestion, virtual modeling, analytics, and decision support services. Historical yield, meteorological data train a Bi-directional Long Short-Term Memory (BiLSTM) model for district-level coconut yield prediction. Soil data was analyzed and incorporated as input to the model as a risk indicator. The model achieved strong performance, with an R2 of 0.9387 and a MAPE of 2.31% on test data, demonstrating its capability to capture nonlinear and lagged relationships between environmental drivers and yield outcomes. The research contributes theoretically by extending Digital Twin and resilience concepts into agricultural supply chains through a structured, scalable framework for perennial crops. Practically, it delivers a decision support tool that enhances visibility, supports proactive interventions, and informs policy and resource allocation for the coconut sector. Overall, this study demonstrates the feasibility and value of DT-based, data-driven resilience strategies for securing the performance and sustainability of coconut supply chains in an increasingly uncertain environment