Abstract
Carbon nanotube (CNT) nanofluids are promising working media for next-generation thermal management, yet reliable prediction of their effective thermal conductivity remains difficult because classical models inadequately represent CNT anisotropy, realistic three-dimensional (3D) dispersion, and interfacial heat-transfer barriers. This study aims to improve the predictive capability of a widely used CNT-nanofluid model by explicitly incorporating (i) true 3D random CNT orientation and (ii) Kapitza (interfacial) thermal resistance. The methodology derives the orientation-averaged effective CNT conductivity using isotropic 3D averaging (yielding an α = 1/3 projection factor) and introduces a physically motivated Kapitza resistance term into the CNT contribution, producing a modified closed-form expression for the nanofluid thermal conductivity ratio. The model is validated against three representative experimental datasets spanning polar and non-polar base fluids (water, ethylene glycol, and R113). Across these cases, the proposed model reduces the mean absolute percentage error to 13.92% compared with 27.09% for the reference formulation, and decreases the root mean square error from 0.362 to 0.225, indicating both improved accuracy and reduced prediction variability. The results show particularly strong improvement for systems where interfacial effects are influential, supporting the model's physical realism. Overall, the proposed framework provides a more defensible and practical tool for designing CNT-based nanofluids in applications where Kapitza-dominant heat-transfer limitations must be captured.