An overview of AI-driven methodologies for the development of nanoparticle-enhanced fluids
File(s)
Published version
Author(s)
Minea, Alina Adriana
Leon, Florin
Markides, Christos N
Type
Journal Article
Abstract
Nanoparticle-enhanced fluids are increasingly investigated for thermal management, energy conversion, lubrication and process applications, but their formulation is complicated by coupled changes in thermal conductivity, viscosity, stability and pressure-loss characteristics. This structured narrative review critically examines how artificial intelligence, machine learning, physics-informed modeling and adaptive experimental design are being used across the nanoparticle-enhanced-fluid research workflow. The review organizes methods by research task and data structure rather than by an assumed performance ranking: classical regression and feed-forward networks for tabular property data; convolutional models for images and spatial fields; recurrent and transformer-based models for time-dependent stability data; graph models for molecular representations; Gaussian-process and Bayesian methods for uncertainty-aware optimization; and physics-informed or reduced-order approaches for constrained field prediction. The evidence indicates that supervised models for formulation-specific thermophysical-property prediction are comparatively mature, whereas transferable benchmarks, external validation and uncertainty reporting remain limited. Physics-informed methods can improve conservation and boundary-condition consistency, but they do not automatically eliminate data requirements, optimization difficulties or interpretability concerns. Autonomous discovery loops, reinforcement learning and large language models are promising for experiment planning and knowledge integration, but remain exploratory for end-to-end nanofluid discovery. The review therefore proposes validation and reporting practices, identifies industrial implementation barriers, and presents a staged roadmap toward reproducible, physically credible and scalable artificial intelligence assisted fluid development.
Date Issued
2026-08-10
Date Acceptance
2026-08-08
Citation
AI Thermal Fluids, 2026
ISSN
3050-5852
Publisher
Elsevier BV
Journal / Book Title
AI Thermal Fluids
Copyright Statement
©2026 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Publication Status
Published online
Article Number
100047
Date Publish Online
2026-08-10