Research Note — Computational Fluid Dynamics
Nanofluid Heat Transfer in a Flat‑Tube Automotive Radiator
A numerical study of CuO–water/ethylene‑glycol nanofluid flow through a flat radiator tube, reproducing and validating the laminar multiphase CFD model of Vajjha, Das & Namburu (2010). Geometry was built in SolidWorks, meshed in Ansys Mechanical across four grid densities, and solved in Fluent as a laminar, multiphase (mixture) flow with the energy equation on.
Geometry, Mesh & Setup
- Modeled the radiator's flat tube (8.09 × 2.54 mm cross‑section, 500 mm long) in SolidWorks and imported it into Ansys
- Generated four structured meshes of increasing density (Mesh 1–4, from ~59k to ~363k elements) for a grid‑independence check
- Set up Fluent for a laminar, multiphase (mixture) solution with the energy equation enabled
- Defined the base fluid (60% water / 40% ethylene glycol) and CuO nanoparticles as the secondary phase, at volume fractions of 1–6%
- Applied a convective wall boundary condition and solved with the Coupled scheme, patching the volume fraction before each run
Governing Equations
Continuity, momentum and energy were solved for the nanofluid mixture (subscript nf), discretized with a first‑order upwind scheme:
Continuity: (∇·V) = 0
Momentum: ρ_nf(∇·V)V = −∇P + μ_nf∇²V
Energy: ρ_nf C_p,nf (V·∇)T = k_nf ∇²T
Nanofluid density, viscosity, thermal conductivity and specific heat were computed from correlations for CuO/Al₂O₃ nanoparticles in a water–ethylene‑glycol base fluid, as functions of particle volume fraction φ and temperature.
Post‑Processing & Results
- Extracted a sample line along the tube wall in CFD‑Post to evaluate the average skin‑friction coefficient (C_f,avg) for each CuO concentration
- Ran the case for CuO volume fractions of 1–6% across the finest mesh (Mesh 4) and compared trends against the reference paper
- C_f,avg increased with CuO concentration in every run, matching the trend reported in the source article
- Cross‑checked volume‑averaged density and viscosity against the article's values — mixture density matched exactly; viscosity was within 0.6–1.8%
Validation against the reference article (Mesh 4, volume‑averaged properties):
| CuO vol. % | 1% | 2% | 3% | 4% | 5% | 6% |
|---|---|---|---|---|---|---|
| Density — article [kg/m³] | 1101.49 | 1156.02 | 1210.55 | 1265.08 | 1319.61 | 1374.14 |
| Density — this project [kg/m³] | 1101.49 | 1156.02 | 1210.55 | 1265.08 | 1319.61 | 1374.14 |
| Density error | 0% | 0% | 0% | 0% | 0% | 0% |
| Viscosity — article [kg/m·s] | 0.00109 | 0.00138 | 0.00173 | 0.00217 | 0.00273 | 0.00343 |
| Viscosity — this project [kg/m·s] | 0.00111 | 0.00137 | 0.00172 | 0.00219 | 0.00275 | 0.00331 |
| Viscosity error | 1.83% | 0.72% | 0.57% | 0.92% | 0.73% | 0.58% |
Maximum deviation from the reference article was under 2%, confirming the model reproduces the published nanofluid behavior. Reference: R. S. Vajjha, D. K. Das & P. K. Namburu, "Numerical study of fluid dynamic and heat transfer performance of Al₂O₃ and CuO nanofluids in the flat tubes of a radiator," International Journal of Heat and Fluid Flow, vol. 31, no. 4, pp. 613–621, 2010.