Estimation of Heat Transfer Characteristics of Inclined Cylinders with Different Fins Using Machine Learning

Natural convection from annular-finned cylinders depends on coupled effects of fin arrangement, inclination, heat input, and material properties that are difficult to capture using conventional correlations. This study presents a data-driven framework for estimating the Nusselt number (Nu) and thermal resistance (R th) of cylinders equipped with homogeneous aluminum and functionally graded Al– Cu fins. The dataset contains 360 experimental observations covering five fin configurations, inclination angles from 0 • to 75 • , and heat-input levels from 25 to 150 W. Gaussian process regression (GPR), a Bayesian regularized artificial neural network (BRANN), and XGBoost were evaluated. Generalization was assessed using leave-one-power-level-out, leave-one-angle-out, and leave-one-fin-configuration-out validation. Accumulated local effects analysis was used to interpret GPR, and symbolic regression provided explicit correlations for both outputs. GPR achieved the highest accuracy. Under the primary leave-one-angle-out validation, its R 2 values ranged from 0.989 to 0.992 for Nu and from 0.988 to 0.991 for R th , with corresponding MAPE ranges of 1.458–1.518% and 3.273–4.375%. The 95% prediction intervals achieved empirical coverage rates 1 of 92.8–95.6%, with the widest intervals occurring when an entire fin configuration was excluded. The symbolic correlations produced R 2 values of 0.955–0.975. Increasing the Rayleigh number enhanced Nu, whereas increasing inclination weakened convection and raised R th. An intermediate fin-spacing ratio maximized the normalized convective response, while denser arrangements reduced thermal resistance. Functionally graded Al–Cu fins generally increased Nu and decreased R th relative to aluminum fins. The framework combines accurate estimation , calibrated uncertainty, physical interpretation, and compact engineering correlations for inclined annular-finned cylinders.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-18
DOI
https://doi.org/10.21203/rs.3.rs-11491188/v1
Primary Topic
Heat Transfer and Optimization
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Estimation of Heat Transfer Characteristics of Inclined Cylinders with Different Fins Using Machine Learning

Aslı Durmuşoğlu, İbrahim Şafak, Sibel Güneş
2 links
Research Square
Heat Transfer and Optimization
preprint

Estimation of Heat Transfer Characteristics of Inclined Cylinders with Different Fins Using Machine Learning

Aslı Durmuşoğlu, İbrahim Şafak, Sibel Güneş
preprint en

Abstract

Natural convection from annular-finned cylinders depends on coupled effects of fin arrangement, inclination, heat input, and material properties that are difficult to capture using conventional correlations. This study presents a data-driven framework for estimating the Nusselt number (Nu) and thermal resistance (R th) of cylinders equipped with homogeneous aluminum and functionally graded Al– Cu fins. The dataset contains 360 experimental observations covering five fin configurations, inclination angles from 0 • to 75 • , and heat-input levels from 25 to 150 W. Gaussian process regression (GPR), a Bayesian regularized artificial neural network (BRANN), and XGBoost were evaluated. Generalization was assessed using leave-one-power-level-out, leave-one-angle-out, and leave-one-fin-configuration-out validation. Accumulated local effects analysis was used to interpret GPR, and symbolic regression provided explicit correlations for both outputs. GPR achieved the highest accuracy. Under the primary leave-one-angle-out validation, its R 2 values ranged from 0.989 to 0.992 for Nu and from 0.988 to 0.991 for R th , with corresponding MAPE ranges of 1.458–1.518% and 3.273–4.375%. The 95% prediction intervals achieved empirical coverage rates 1 of 92.8–95.6%, with the widest intervals occurring when an entire fin configuration was excluded. The symbolic correlations produced R 2 values of 0.955–0.975. Increasing the Rayleigh number enhanced Nu, whereas increasing inclination weakened convection and raised R th. An intermediate fin-spacing ratio maximized the normalized convective response, while denser arrangements reduced thermal resistance. Functionally graded Al–Cu fins generally increased Nu and decreased R th relative to aluminum fins. The framework combines accurate estimation , calibrated uncertainty, physical interpretation, and compact engineering correlations for inclined annular-finned cylinders.

Research Square
Hakkari University (TR), Kayseri Üniversitesi (TR), Erciyes University (TR)
Heat Transfer and Optimization
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon