I'll give it a shot. While I despise US units, that's what I'm used to....
Ixx (I) from a CAD sketch (I assumed an internal radius of 5mm and no tapers): 2.46 in^4
Max stress (σ): 36 ksi
c (distance from neutral axis): 1.969 in
Mmax (M) = σ*I/c = 44,977 in*lbs
Length (l): 24 in
Simply supported beam with a single centered load (P); M = P*l/4 or P = M*4/l = 7,496 lbs
Assuming 2 pieces, not connected you'd get about 15 kips. However, there is almost certainly something connecting the 2 channels together, back to back, and that would increase the load carrying capacity.
Since we're already here, deflection at the proposed load of 20,000 lbs (20 tons total capacity divided by 2 channels) is easy:
δ = P·L^3 / (48·E·I) = 0.081 in - a little over 1/16"
This only looks at bending, doesn't include anything for flange or torsional buckling. Feel free to scrutinize, I'm not on the clock today.