Additive Manufacturing (AM) has advanced sustainable product creation in the past decade, reducing our CO2 footprint. This influences optimized heat exchanger design, relying on predictive simulations. However, AM rough surfaces differ from prior research on turbulent flow over rough walls. Existing 3D Navier-Stokes models neglect these unique characteristics. To address this, we developed a high-fidelity LES database for AM wall models. Studying rough pipes made with AM and Nickel Alloy IN939, we examined turbulence at Re = 11,700. Turbulence and drag forces impacted the skin friction coefficient significantly. Surprisingly, even with a high value of equivalent sand-grain roughness, ks, a logarithmic layer was observed. Roughness affected turbulence, including mean flow, velocity fluctuations, and Reynolds shear stresses. We introduced an effective wall-normal distance to compare turbulence statistics, collapsing shear stresses, and velocity fluctuations outside the roughness sublayer. This showed Townsend's similarity for the roughness. We also investigated surface roughness's impact on heat transfer using LES and RANS at Re = 1,000 – 20,000. The SST k-omega RANS model with enhanced wall treatment captured mean flow characteristics for smooth and rough channels. Caution is advised for higher Reynolds numbers due to grid resolution dependence.