Do as someone else suggested: calculate the Carnot-efficiency of the contraption, and weep.
Carnot eff. = (Thigh-Tlow)/Thigh
(all temperatures in Kelvin)
note that a small temp difference (a few degrees) lowers Carnot efficiency to nearly zero. Let's assume a temp difference (Thigh-Tlow) of 10 deg. K (=10 deg.C); Thigh, assuming to be the Curie temp, of steel, is 1043K.
Carnot eff = 10/1043 = .01 = 1%.
Now, even for a heat engine, that is lousy.
And yes, I know that with a parabole you could get 'free' energy, so some may say efficiency doesn't matter. It does to me though.