I fully agree with Schro, total time should matter, not speed. And you should add 2 hours or so(time to go to the airports and other things) before making variations.
Imagine Kansas City- Los Angeles, 1182NM
A320, time airborne, 3h15. Total travel time, 5h15 - 315 minutes
E145, time airborne, 3h25. Total travel time 5h25, 325 minutes - penalty 3% against the A320
Q400, time airborne 4h19. Total travel time 6h19 - 379 minutes - penalty 20% against the A320
A140, time airborne 5h18. Total travel time 7h18 - 438 minutes - penalty 39% against the A320
It's not a travel through the whole world. now imagine a much shorter flight : Kansas-City Des Moines. 143NM
A320, time airborne 0h29. Total travel time 149 minutes
E145, time airborne 0h31. Total travel time 151 minutes - penalty 1.3 % against the A320
Q400, time airborne 0h36. Total travel time 156 minutes - penalty 4.7% against the A320
A140, time airborne 0h40. Total travel time 160 minutes - penalty 7.4% against the A320
Maybe it's still too much, but it's clearly more realistic than a pure speed-based penalty. It makes the slow A140 completely unefficient on long routes, but still useful on very short routes.
Or maybe the penalty should be proportional to the TIME difference. 10% penalty per hour difference. On a very long flight, 15 hours against 8 hours would be deadly - 70% penalty. On a very short flight, it would not even be noticed. Between Kansas City and Des Moines, the A140 would have a 11/60*10% = 1,8% penalty. Peanuts. And it would fairly take in account the tech stops. A plane quick to refuel as the E170 would have less penalty linked to the tech stop than a IL62. A tech stop that costs 1 hour would always cost 10%.