There's a quite some difference between empirical and **phenomenological**: All classical mechanics is based on phenomenological observations, and seems to me that's quite accurate at least in my line of work. if you mean that some coefficients multiplying terms derived from phenomenological observations in some turbulence models are calibrated experimentally, the yes, I agree these might have some influence in turbulence features observed in complex scenarios; however, flow around a sphere/pillar is an archetype flow: most models are tested and its coefficients "calibrated" (or derived in some instances) against such archetypes!