Cloud-based architecture of digital twins applied to the optimal design of large-scale supply chains

The design of large-scale supply networks is an NP-hard MILP problem that requires dynamic representation and efficient optimization in industrial contexts. This study proposed, implemented, and validated a four-layer cloud-based architecture that integrates Microsoft Azure Digital Twins (ADT), a multi-echelon MILP model, a hybrid PSO, and a Python-based integration engine with bidirectional synchronization. The main algorithmic contribution consisted of using the linear relaxation of the MILP as the PSO fitness function, providing a formal lower bound for the original problem. A total of 74 experimental runs were conducted using a capacity repair operator, achieving a zero-infeasibility rate. The evaluated instances, across S, M, and XL scales and five complexity levels, showed an inverse relationship between the integrality gap and the quality of the PSO solution. Likewise, instances with low combinatorial complexity achieved over 90% consistency in opening decisions, preserving the decision state of the digital twin. The proposed framework integrates cloud-based architecture, MILP optimization, and hybrid PSO for supply chains with experimental validation. © The authors © Revista Científica de Sistemas e Informática.

Subjects

Azure Digital Twins

Supply chain

Network design

Digital twin

PSO

MILP

Python

License

Acceso Abierto

URL License

https://creativecommons.org/licenses/by-nc-sa/4.0/

How to cite

Marmolejo-Saucedo, J. A., Rodríguez-Aguilar, R., & Mendoza, A. (2026). Cloud-based architecture of digital twins applied to the optimal design of large-scale supply chains. Revista Científica De Sistemas E Informática, 6(2), e1682. https://doi.org/10.51252/rcsi.v6i2.1682