Computational analysis of Darcy–Forchheimer flows in a stenosed arteries: Medical applications with ANN-CFD coupling

• Published in: Journal of thermal analysis and calorimetry Vol. 150; no. 18; pp. 14203 - 14217
• Main Authors: Hira, Fatima Shafiq, Rubbab, Qammar, Ahmad, Irshad, Majeed, Afraz Hussain
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Nature B.V 01.09.2025
• Subjects: Algorithms; Arteries; Artificial neural networks; Blood flow; Brownian motion; Cavitation; Cooling; Coronary vessels; Drug delivery systems; Finite difference method; Heart; Heat generation; Heat sources; Heat transfer; Hydraulic fracturing; Influence; Investigations; Lubricants & lubrication; Mechanics; Microorganisms; Multilayer perceptrons; Multilayers; Nanoparticles; Ohmic dissipation; Partial differential equations; Researchers; Resistance heating; Stroke; Thermal radiation; Veins & arteries; Viscosity
• ISSN: 1388-6150; 1588-2926
• DOI: 10.1007/s10973-025-14281-x

A potential application of this research involves optimizing nanoparticle-based solutions to enhance blood flow in stenosed arteries, thereby improving drug delivery systems and therapeutic treatments for cardiovascular diseases. This approach could lead to more effective and targeted therapies. In this study, artificial neural networks (ANNs) were employed to examine the effects of heat generation and activation energy on the Darcy–Forchheimer flow of hybrid nanoparticles containing magnetized fluid in blood flowing through stenosed arteries. This computational approach provides valuable insights into the complex interactions within the system. Heat transfer analysis considers several key factors, including thermal radiation, viscous dissipation, heat sources, and Joule heating. These elements collectively significantly impact the overall heat transfer process within the stenosed arteries. Applying similarity transformations simplifies the governing Partial Differential Equations (PDE), reducing it to an Ordinary Differential Equations (ODE) more amenable to computation. The resulting ODE is then solved using the finite difference method, a common numerical technique. The Levenberg–Marquardt Algorithm (LMA) is utilized within multi-layer perceptron models, specifically those configured with 10 neurons in each hidden layer. This choice of algorithm is motivated by its efficiency in training such networks. The ANN model’s architecture consists of several distinct layers designed for specific functions. It includes nine input layers to facilitate data entry and three output layers to present the results of the analysis.