Thank you for the answers / insights, guys.
Your excel was really helpful, mikeywith!
If blocks are completely full for a longer period of time, it might fuel a fee bidding war, thus increase the average fee... What do you mean by mining capacity? The difficulty is not adjusted in relation to the number of transactions or fees, nor is the block size... So i'd say that there is no correlation between them, altough i don't think this was your actual question...
At first glance it was trivial for me as well. But I compared my calculator with public calculators and differences occured, though I used the exact same input datas.
ASIC hashrate
Network difficulty
Exchange rate
Power consumption
Cost / kWh
I concluded that some minor and hidden variable is missing. I must add, that differences between various online calculators are also common. These are 15-20% differences (mine included). These contradictions are not relevant enough to assume that the nature of the problem is algorithmic, but not insignificant enough not to frustrate myself.
If you believe in bitcoin, you can probably take a little bit bigger risk, if you don't believe in bitcoin you should play it safe...
I have an ongoing research / experiment. I want to use the Bitcoin concept as a method of proof. My personal views are not relevant, though the technological aspect of the protocol fascinates me.