Intelligent supervised learning for viscous fluid submerged in water based carbon nanotubes with irreversibility concept

Abstract

This article examined the Silver based Di‑hydrogen carbon nanotubes flow model (SDH-CNTFM) between two stretchable coaxially disks by utilizing the Method of Levenberg Marquardt with Back-propagated Neural Networks (MLM-BPNN). Here the base liquid is silver (Ag) and the nanoparticles are SWCNTs and MWCNTs (single and multiwall carbon nanotubes). The governing PDEs for SDH-CNTFM are transformed into ODEs by utilizing similarity transformation. Energy equations are developed through heat generation and viscous dissipation joule heating. Also calculated the total entropy optimization. Flow parameters velocity, entropy optimization, temperature, Nusselt number and Bejan number are discussed for both single and multi-walls carbon nanotubes (SWCNTs and MWCNTs) graphically and in Tabular form. The reference dataset is calculated through implementation of Optimal Homotopy Analysis method (OHAM) for variants of SDH-CNTFM. For the variation of different parameters this reference dataset is utilized in MATLAB to clarify the solution and error analysis plots. Moreover, the approximated solution is assessed through adopting training/testing/validation procedure and comparing it with standard solution which is endorsed by performance study based on MSE convergence, error histogram and regression studies. Heat transfer rate and surface drag force are discussed for both SWCNTs and MWCNTs numerically by using different flow parameters. From obtained outcomes, it is observed that entropy rate boosts up for higher approximation of nanoparticles of volume friction and Brickman number (Br) which is controlled due to the minimization of Brickman number.