This study revisits the turbulent forced‑convection performance of multi‑walled carbon nanotube (MWCNT)/water nanofluids. Building on a Eulerian–Eulerian two‑fluid framework, we couple a second‑order orientation tensor with population balance equations (PBEs) to simultaneously resolve shear‑induced alignment, anisotropic thermal conductivity and aggregation kinetics. A rigorous grid convergence study and quantitative comparisons with published experiments demonstrate root‑mean‑square errors below 5 % in Nusselt numbers and friction factors. Micro-level validation of the constitutive models is provided by comparing predicted orientation factors with rheo-optical measurements and simulated cluster size distributions with dynamic light scattering (DLS) data. The new coupled model predicts an optimal operating window (0.8–1.2 vol% % MWCNTs, Reynolds number 20 000–40 000) where the performance evaluation criterion (PEC) exceeds 1.3. Within this window, axial thermal conductivity increases by ~38 % and Nusselt numbers by ~24 % at 1 vol% %, while hydraulic penalties remain manageable. Concentrations above 1.2 vol% % trigger rapid aggregation that reduces thermal conductivity and increases viscosity, producing PEC values below unity. An economic analysis based on a 500-kW cooling system and realistic nanofluid preparation costs indicates payback periods of less than two years when operated in the high-PEC region. The paper closes with a balanced discussion of modelling limitations and future research directions.