Over at this question about A&A variants I suggesting changing heavy bombers such that each bomber is worth two dice, choose the higher value. For attacking this means the probability of successful attack increases 33 percent: from 0.67 (4/6) to 0.89 (1 - 2/6^2). For strategic bombing the expected value of IPCs would goes up by 28 percent.
So my rule makes bombers about 33 percent better than they were before, which using Tom's methodology is worth about 5 IPCs, since each bomber is costs 15. This is a 6-turn pay-back with 1 bomber, or two turns with 3--not accounting for the time-value of IPCs. This seems much more in line with the increases in the other techs. (Also see the footnote at the bottom explaining how this rule reduces the likelihood of bad rolls in addition to increasing the average. This is additional value not included in the 5 IPC figure.)
Derivation of the increase in expected strategic bombing value
For 1-die the value is 3.5. For rolling two dice and picking the higher one, the math is much more complicated. From this paper on dice probabilities, the probability that the highest number rolled with n dice is k is given by (sorry, no tex formatting on the board game site):
(k^n - (k - 1)^n) / 6^n
which for n = 2 reduces to
(k^2 - (k - 1)^2) / 36
Now we calculate the probability that our two rolls will give us each value from 1 - 6.
Value Probability Equation
1 1/36 (1^2 - (1-1)^2)/36
2 3/36 (2^2 - (2-1)^2)/36
3 5/36 (3^2 - (3-1)^2)/36
4 7/36 (4^2 - (4-1)^2)/36
5 9/36 (5^2 - (5-1)^2)/36
6 11/36 (6^2 - (6-1)^2)/36
36/36 <--- just checking the sum
Now to calculate the expected value we have to weight these probabilities by the likelihood (i.e. multiple the value and probability columns and sum), which results in 4.47--a 28 percent increase over 3.5.
This table also makes clear that rolls are now going to be biased on the high side. Where as before the chances of getting a 1 were 1/6 = 17 percent, now they are 3 percent. And the probability of getting a 6 has increased from 17 percent to 30 percent. So the average has moved up and moved up in a way that makes it very unlikely you will get a bad roll, removing some of the luck from strategic bombing, further increasing their value above the postulated value of 5 IPCs per bomber.