Machine learning analysis of oscillatory-turbulent heat transfer using carbon-based diamond nanofluids over MHD nonlinear wavy surfaces
At a Glance
Section titled “At a Glance”| Metadata | Details |
|---|---|
| Publication Date | 2026-01-01 |
| Journal | Sustainable Carbon Materials |
| Authors | Danial Habib, Caiyan Qin, Yue Yang, Mingming Zhang, Yuting Guo |
| Institutions | Harbin Institute of Technology, Kyoto University |
| Analysis | Full AI Review Included |
Executive Summary
Section titled “Executive Summary”- Objective: Comprehensive numerical and machine-learning analysis of oscillatory heat transfer and turbulent flow dynamics using diamond-water nanofluids over nonlinear wavy surfaces under magnetohydrodynamic (MHD) effects.
- Key Material Insight: Nanodiamonds (20-50 nm) provide ultra-high thermal conductivity (1,000 W/m·K) and exceptional chemical stability, making them ideal for high-precision thermal management.
- Methodology: Similarity-transformed governing equations solved using the Keller-Box method (KBM) with second-order accuracy, validated via Artificial Neural Networks (ANN) achieving MSE ≈ 10−7.
- Performance Finding: Aggregated nanoparticle configurations (ANP) enhance thermal performance by increasing the Nusselt number (Nu) by up to 30%, despite a 25% increase in viscous dissipation and skin friction (Cf).
- Geometric Effect: Surface undulations induce oscillatory thermal dynamics and turbulent mixing structures, reducing overall Nu by 15%-20% due to boundary layer discontinuity.
- Optimization: Optimal performance identified at 2%-3% nanoparticle volume fraction and moderate magnetic field strengths (M = 2.3-3.4), balancing thermal enhancement with pumping power requirements.
- Application Target: Advanced thermal management for electronics cooling, aerospace systems, and high-performance heat exchangers.
Technical Specifications
Section titled “Technical Specifications”| Parameter | Value | Unit | Context |
|---|---|---|---|
| Base Fluid | Pure Water | - | Carrier medium for nanofluids |
| Nanoparticle Material | Diamond (SCD) | - | 20-50 nm size range, >95% purity |
| Thermal Conductivity (Nanodiamond) | 1,000 | W/m·K | Ultra-high thermal performance |
| Density (Nanodiamond) | 3,510 | kg/m3 | Solid phase property |
| Specific Heat Capacity (Nanodiamond) | 497.26 | J/kg·K | Solid phase property |
| Dynamic Viscosity (Base Fluid) | 0.000803 | Pa·s | Water at reference temperature |
| Density (Base Fluid) | 997.1 | kg/m3 | Water reference |
| Specific Heat Capacity (Base Fluid) | 4,179 | J/kg·K | Water reference |
| Magnetic Field Strength (M) | 1.0 - 5.0 | dimensionless | MHD effect parameter |
| Sinusoidal Waviness (εT) | 0.1 - 0.3 | dimensionless | Surface undulation amplitude |
| Nanoparticle Volume Fraction (ϕ) | 2% - 3% | % | Optimal thermal/hydrodynamic balance |
| Prandtl Number (Pr) | 0.62 | dimensionless | Fluid property |
| Convergence Criterion (KBM) | δ | ||
| ANN Training Split | 70/15/15 | % | Train/Validation/Test |
| ANN Best MSE | ≈ 10−7 | - | Aggregated scenarios, Scenario I-B |
| Grid Resolution | 1,600 × 200 | elements | Computational domain discretization |
| Nusselt Number Enhancement (ANP) | ~30 | % | Aggregated vs. non-aggregated |
| Nusselt Number Enhancement (NANP) | ~22 | % | Non-aggregated configuration |
| Viscous Dissipation Penalty | ~25 | % | Cost of aggregation |
| Heat Transfer Reduction (Waviness) | 15 - 20 | % | Due to boundary layer discontinuity |
Key Methodologies
Section titled “Key Methodologies”- Problem Formulation: Derivation of a governing flow model using similarity transformations for oscillatory MHD flow over vertical nonlinear wavy surfaces.
- Nanofluid Property Modeling:
- Application of classical Maxwell and Brinkman models for effective thermophysical properties.
- Implementation of fractal-based models to capture nanoparticle clustering, irregular geometries, and nanolayer effects.
- Accounting for interfacial resistance, cluster diameter, and nanolayer thickness for nanoscale confinement.
- Numerical Solution (KBM):
- Transformation of higher-order PDEs to first-order systems via variable substitution.
- Discretization using central differences with step sizes ∆η = ∆ξ = 0.005.
- Linearization of algebraic equations via Newton’s method.
- Solution of the linear system through block tridiagonal elimination.
- Machine Learning Validation (ANN):
- Feedforward network architecture: input layer (M, εT), one hidden layer (10 neurons, tanh activation), output layer (Nu, Cf).
- Training algorithm: Levenberg-Marquardt backpropagation.
- Overfitting prevention via premature stopping based on validation error.
- Grid Independence Study: Progressive mesh refinement from 100 × 10 to 1,600 × 200 elements, confirming grid independence at differences < 0.0001%.
- Scenario Construction: Systematic variation of M (1.0-3.0) and εT (1.0-3.0) across aggregated (ANP) and non-aggregated (NANP) configurations.
6CCVD Solutions & Capabilities
Section titled “6CCVD Solutions & Capabilities”-
Applicable Materials:
- For experimental validation of this computational study, Optical Grade Single Crystal Diamond (SCD) or High-Purity Polycrystalline Diamond (PCD) powders/nanoparticles are required to accurately model the 1,000 W/m·K thermal conductivity assumption.
- Boron-Doped Diamond (BDD) may be considered for modified studies investigating electrical conductivity effects under MHD conditions.
- Diamond substrates (up to 10mm thickness) are available for fabricating model test surfaces that replicate the nonlinear wavy geometries used in this simulation.
-
Customization Potential:
- The paper’s wavy surface geometry (εT = 0.1-0.3) can be physically realized using 6CCVD’s custom diamond deposition and laser cutting services to produce precisely undulated test specimens.
- For experimental heat transfer validation setups requiring embedded sensors or electrodes, 6CCVD offers custom metalization (Au, Pt, Pd, Ti, W, Cu) with in-house capability for thin-film deposition on diamond substrates.
- Polishing services (Ra < 1nm for SCD, < 5nm for inch-size PCD) ensure surface finish quality critical for accurate boundary layer characterization.
-
Engineering Support:
- 6CCVD’s in-house PhD engineering team can assist with material selection for advanced thermal management projects, including nanofluid formulation guidance and diamond substrate specification for heat exchanger prototyping.
- Consultation is available for integrating MPCVD diamond components into aerospace cooling systems, power electronics thermal interfaces, and high-performance heat exchanger designs.
-
Call to Action: For custom specifications or material consultation, visit 6ccvd.com or contact our engineering team directly.
View Original Abstract
This study presents a comprehensive numerical analysis of the oscillatory heat transfer and of turbulent-flow dynamics of carbon-based diamond-water nanofluids on a vertical nonlinear wavy surface under magnetohydrodynamic (MHD) effects. The work highlights the pivotal role of nanodiamonds, a class of carbon nanomaterial characterized by ultra-high thermal conductivity (1,000 W/m·K) and exceptional stability, in enhancing thermal performance. The resulting governing flow model, derived using similarity transformations, is solved by the Keller-box method (KBM) and validated with exceptional accuracy (MSE ≈ 10<sup>−7</sup>) using an artificial neural network (ANN) for machine learning (ML). Aggregated (ANP) and non-aggregated nanoparticle (NANP) configurations are systematically studied under the rising magnetic field strengths (M = 1.0-5.0) and sinusoidal conditions ($ {\varepsilon }_{T} $= 0.1-0.3). The results show that the formation of ANPs leads to a substantial increase in thermal conductivity due to the formation of conductive clusters, which increase the Nusselt number Nu by 30%, albeit at the considerable cost of a 25% increase in viscous dissipation and skin friction <italic>C<sub>f</sub></italic>. Conversely, non-aggregated nano-diamonds promote smooth velocity profiles that yield moderate enhancement in <italic>Nu</italic> (up to 22%) and high hydrodynamic functionality. Surface undulations generate oscillatory thermal dynamics and turbulent mixing structures, resulting in a 15%-20% reduction in the overall Nu due to boundary layer discontinuity. However, this negates the reduction resulting from the increase in overall thermal bridging. An optimal nanoparticle volume fraction <italic>C<sub>f</sub></italic> of 2%-3% is identified, balancing the thermal enhancement against the pumping power requirements. These results provide important design insights for advanced thermal management systems, especially in the cooling of electronics and high-performance heat exchangers, where magnetic field control and surface geometry optimization are critical for operational efficiency.