You should get running algo to get answers, or calculate it somehow.
And its good solution to hardcode something like this in production.

I feel like it is actually more dificult to hardcode the answers than actually usign a algorithm.

I disagree. I had lots of fun with this Kata

I disagree that it leads to wrong code. I managed to solve it without any hacks or anti-patterns. I do agree that the description could be a lot clearer. THe only issue I have right now is that one part of the random test is sometimes failing, the AverageConsumptionByDistance test are flacky but that could be a rounding error on my part so not blaming the test yet. What I liked about this Kata was the seperation of concerns. You could solve it using passing lambdas but I managed to fix it without