If you experiment with parameter ranges (as the answer showed in your previous question, link), the integral can be done. For example, your integrand f[d,w,r] is singular at d=2, so we avoid this point; as a test, I then plotted the integrand for d=2.5, and found that it is real for r > 50 (roughly). Then,
In[1]:= Integrate[ f[2.5, 1, r], {r, a, b}, Assumptions -> {a > 50, b > a}, GenerateConditions -> False]
Out[1]= (1.35316 - 0.625 Sqrt[1. - 26.103 a^0.5 + 0.08 a^2])/a + (-1.35316 + 0.625 Sqrt[1. - 26.103 b^0.5 + 0.08 b^2])/b
So my comment was wrong, Mathematica can evaluate the integral, as long as the integrand is real.
EDIT: 1) The dependence on $\omega$ is trivial, 1/$\omega$, 2) for a fixed value of $d$, the valid range of $r$ is determined by the range where the square root in the denominator is real - Mathematica cannot solve for this explicitly (at least in my attempts) because of the $\Gamma$ function. You will have to get around this somehow by working out each case.