Analisis dan Simulasi Konduksi Panas Kondisi Steady-State Pada Model 2D Struktur Copper Open-Cell Metal Foam Menggunakan Pendekatan Physics Informed Neural Networks

Armi, Al Habibie (2026) Analisis dan Simulasi Konduksi Panas Kondisi Steady-State Pada Model 2D Struktur Copper Open-Cell Metal Foam Menggunakan Pendekatan Physics Informed Neural Networks. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5011221063-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Copper open-cell metal foam merupakan material berpori berkonduktivitas termal tinggi yang banyak digunakan dalam sistem manajemen termal. Kompleksitas mikrostrukturnya menghasilkan jalur konduksi panas yang heterogen dan sangat dipengaruhi oleh parameter geometrik seperti diameter pori, diameter strut, dan variasi ukuran pori. Penelitian ini menganalisis distribusi temperatur, distribusi fluks panas, dan konduktivitas termal efektif (Keff) pada model dua dimensi copper open-cell metal foam menggunakan pendekatan Physics-Informed Neural Networks (PINN). Geometri struktur pori dibangkitkan menggunakan metode Laguerre–Voronoi Tessellation (LVT) dengan parameter morfologi berbasis data literatur material, serta tiga konfigurasi struktur bi-layered (40–20, 30–20, dan 40–30 PPI(Pore Per Inch)) sebagai analisis terapan. Simulasi dilakukan pada tiga variasi kondisi batas aliran panas, sehingga total 12 variasi yang dianalisis. Model PINN diformulasikan dengan mengintegrasikan residual persamaan konduksi panas steady-state ke dalam fungsi loss melalui mekanisme automatic differentiation, dengan Finite Element Method (FEM) sebagai metode referensi. Hasil evaluasi menunjukkan bahwa PINN mampu memprediksi distribusi temperatur dengan akurasi baik, dengan nilai MAE pada rentang 1,33–2,30 K dan Relative L2 Error pada rentang 5,29×10⁻³–8,57×10⁻³ untuk geometri 1,2, dan 3, serta MAE 0,22–0,50 K dan Relative L2 Error di bawah 0,002 untuk struktur bi-layered. Nilai Keff yang diperoleh berada pada rentang 20,94–24,15 W/m·K untuk geometri 1, 2, dan 3. Serta 56,10–80,14 W/m·K untuk struktur bi-layered, konsisten dengan batas teoritis Voigt–Reuss serta data commercial copper foam pada literatur. Variasi struktur pori terbukti memengaruhi mekanisme perpindahan panas, di mana porositas dan konektivitas ligamen secara bersamaan menentukan nilai Keff. Pada struktur bi-layered, nilai Keff selalu berada di antara nilai kedua lapisan penyusunnya dan mendekati rata-rata penjumlahannya, tanpa adanya loncatan temperatur pada daerah antarmuka. Penelitian ini menunjukkan potensi PINN sebagai metode meshless alternatif yang efektif untuk analisis perpindahan panas pada material berpori bergeometri kompleks.
====================================================================================================================================
Copper open-cell metal foam is a high-thermal-conductivity porous material widely used in thermal management systems. Its complex microstructure produces heterogeneous heat conduction paths governed by geometric parameters such as pore diameter, strut diameter, and pore size distribution. This study analyzes the temperature distribution, heat flux distribution, and effective thermal conductivity (Keff) of a two-dimensional copper open-cell metal foam using a Physics-Informed Neural Networks (PINN) approach. Three homogeneous microstructural geometries were generated via the Laguerre–Voronoi Tessellation (LVT) method using morphological parameters from metal foam literature data, along with three bi-layered configurations (40–20, 30–20, and 40–30 PPI) as applied analyses. Simulations were conducted under three heat-flow boundary condition variations, resulting in a total of 12 case studies analyzed in this research. The PINN model was formulated by embedding the steady-state heat conduction equation residual into the loss function via automatic differentiation, with a Finite Element Method (FEM) serving as the reference. Results show that PINN accurately predicted temperature distributions, achieving MAE values of 1.33–2.30 K and Relative L2 Error of 5.29×10⁻³–8.57×10⁻³ for the homogeneous geometries, and MAE of 0.22–0.50 K with Relative L2 Error below 0.002 for the bi-layered structures. The predicted Keff ranged from 20.94 to 24.15 W/m·K for the homogeneous geometries and 56.10 to 80.14 W/m·K for the bi-layered configurations, consistent with Voigt–Reuss theoretical bounds and published commercial copper foam data. Pore structure variations significantly influenced heat conduction mechanisms, with porosity and ligament connectivity jointly determining Keff. In bi-layered structures, Keff consistently fell between the values of the two constituent layers and approached their arithmetic mean, with no temperature jump observed at the interface. These findings demonstrate the effectiveness of PINN as a meshless alternative method for heat transfer analysis in porous materials with complex geometries.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Copper Open-Cell Metal Foam, Steady-State, PINN, Konduktivitas Termal Efektif, Laguerre–Voronoi Tessellation, Effective Thermal Conductivity.
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QC Physics > QC320 Heat transfer T Technology > TA Engineering (General). Civil engineering (General) > TA 459 Metal foams.
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Material & Metallurgical Engineering > 28201-(S1) Undergraduate Thesis
Depositing User:	Al Habibie Armi
Date Deposited:	23 Jul 2026 09:14
Last Modified:	23 Jul 2026 09:14
URI:	http://repository.its.ac.id/id/eprint/136499

Actions (login required)

View Item