How come when a train is going around a curve that is on a grade going up or down, the locomotive looks like it is leaning towards the outside of the curve?
Original Post
Replies sorted oldest to newest
Well, for that question, I give you a grade of C+. Don
Centrifugal force. 
You’re obviously not running an Acela. 🤓
That's why I superelevate (the thickness of a popcycle stick)all my elevated curves.
I try to set track screws on the inside of turns and slightly bank the turns . Not Daytona like but some.
A'cause it's ICRR #382 traveling at warp speed?
A'cause an object traveling straight line is inclined to continue that path until another force overcomes that intial force.... (also see inertia vs momentum)
You are playing "Excentrifugal Forze" much to loudly and your eyeball fluid is rippling the light refractions. (I do that too 
)
C+...... I missed that first pitch... Son
It is leaning out due to centrifigal force. Sometimes you can "bank" the curve with the outside slightly higher than the inside to overcome this issue.
There is actually a geometric component to this phenomenon that is completely independent of physics. That is, even if the loco is *stationary* it will still be leaning outward on any curve with a gradient. The geometric component exacerbates any physics involved and makes your loco more likely to fling itself off the track from centripetal acceleration.
Using popsicle sticks to bank curves in *flat areas* is helpful, but there are diminishing returns when trying to do it on gradients that are curving. It introduces a more severe twist into the rails, and as soon as the twist is greater than the depth of the wheel flanges over a single truck distance, you have problems. It's worse for longer trucks (3 axles or more) and worse for longer engines with greater distances between trucks. And obviously worse as the curve gets tighter and the gradient gets steeper.
The deep flanges of hi-rail and tubular stuff is helpful for this, and allows more severe gradients in curves than the scale guys can use. I'm betting those tiny flanges mean they have to stick to almost prototype gradient limits in curves.
For examples of what Jeff is saying, simply look at how many rollercoaster are actually designed. They often turn first, then transaction to tilt or climb/drop, etc..
Or..don't want to run trains like a car chase scene thru San Francisco. You want all wheel planted firmly all the time.
Superelevation IS a banked curve by the way.
