We just started doing treadmill work last week and we are really wondering about this. The machines they have at our gym don’t appear to actually tilt…the belt doesn’t seem to move upward when we change the incline. How do machines that don’t actually tilt simulate an incline?
I would posit that your premise is flawed. All treadmill machines with an ‘incline’ feature do in fact tilt. I have a machine from the early 1980s with this functionality, so I can hardly imagine why more recent models wouldn’t be able to do it. More over, as you have suggested, there is no way to replicate an incline artificially.
Does the belt move, or are you moving the belt? If the latter, increasing roller friction would (IMHO) simulate walking up increasing inclination.
Put an airplane on it, and it’ll tilt. 
They do tilt, just not a lot. (Did you really want to simulate walking up a 45-degree ramp?)
There is more than just changing rolling resistance on the belt - the changing incline affects how far you have to pick up your feet, and no amount of variation in rolling resistance can simulate the height of your step.
Okay. I guess it’s a matter of the slope being a lot more visible over 1/2 mile than over the 3 feet or so that the machine covers.
These are motorized treadmills, to answer Nanoda’s question.
The ones at my gym certainly tilt. It’s really obvious by the time you set it to 5 degrees or so. It also helps if the treadmill you’re on is side-by-side with a treadmill that is not in use. You can see the difference better.
Now that I’ve read this twice, here’s the probable answer: The belt isn’t supposed to tilt independently of the treadmill. But when the entire treadmill tilts, usually up to a 10 percent incline, the belt tilts with it.