Was bouncing minecarts again, 5 on a S-shaped six tile track with two ramps and came across something quite thoroughly odd that I'm complaining about:
╔▲
╚
╗▲╝
|
|
Tower cap and willow minecarts rotate alternatively back and forth, pink tiles are low (50) friction track stops.
First finding was that unlike with ramps, a
cart hitting another after a corner still pays 1000 to cross over; unintuitively like with ramps a cart with less than 1000 speed can still move over a corner (and then stop).
The third cart - second tower-cap cart - The first hit it took from the first willow (weight 15) cart, as it tried to go over tile edge and take a turn gave it 2390, when it was moving westward at 1275 (subtiles/step) from retained velocity from coming back up. Being southward only, this didn't impact the westward velocity.
Every subsequent gave it 3830, which is exactly ramp impulse - all frictions - turn cost of 1k. Which means the collision changed from inelastic to elastic.
This means that
cart direction on crossing a corner is to the side only if the cart is already at tile edge.
Next one after that, similar issue with the 4th minecart (willow) having remaining 2680 velocity north: First hit when moving over was unelastic 2780 (=3830-1000 turn-50 low stop) which also stopped its westward momentum, then...nothing, no resetting the northeast moving cart to 1340 east each step. Larix showed that a light high-speed cart hitting medium-speed heavy cart from behind gives it low speed, but here it is heavy low-speed cart not overwriting light high-speed cart.
Of course, once the 2nd willow cart going northeast tried to cross over north, it gave unelastic push of 831(=[2380-1050]*15/24) to the second tower-cap cart supposedly pushing on it, which zeroed out its speed (though it didn't recenter it either direction).
Afterwards, it started taking 1340 hits as predicted.
The third tower-cap cart, having stood completely still, took exactly predicted 202 (202,5=(1340-1000)*15/24-10 on first step.
Every subsequent one, it took 330, indicative of an elastic collision (330=(10 friction+202,5)*24/15-10 friction) or cart weight difference not being accounted for (1340-1000-10). Mind, it's willow(15) striking tower-cap (24).
So, I assume it is a case where southward heading light cart striking a heavier cart with heading?? with same or lesser speed results in cart weight difference not being taken into account despite being light on heavy.
(It being nonelastic would indicate ~1544 2. willow speed, which ...does not work with the rest of the numbers at all.)
...
Well, it moved over 22 st positions into ramp in 152 steps with zero speed (due crossing corner), and the whole dance started the other way around; elastically giving the willow moving southward with 340 4890 from tower-cap.
Overall: I shouldn't have calced out month long repeaters without doing simple tests of design first. Stupid diagonal elastic collisions ruining my precious predictions >_>