The real problem is that the lengths of the ease-in/ease-out sections are too short, for a given radius of curve. As a result the train passes through the eased section way too fast, and so rolls into/out-of the curve too fast... lurch! Why would this be? They couldn't have designed it that lame! RW is a monumental tour de force of software engineering.
I burned a load of time searching, reading, etc. and then I landed on this old post on the UK Train Sim forum. As I read more of @hertsbob's posts, I realize that he knows what he's doing. Anyway, the following set off a chain of thinking for me:
First of all it’s extremely important to have your Track Rule set up correctly for easements before you start laying your track. Duff numbers in the Track Rule can make using easements far more fiddly than it needs to be. As a broad rule of thumb the Max speed tolerance value should be 1.4 times the maximum line speed. These are the sorts of numbers I tend to use. Others may well disagree, but these work for me.
So I learned how to create a cloned/modified track rule that I could play with. Immediately I found that the Max Speed Tolerance in the track rule controls the easing rate, both the rate increase of curvature and the rate of increase of cant, with respect to the distance into the curve. I want the ease to be much slower so as the train enters the curve, it rolls into the curve more slowly as well as tightening its curve more slowly. To do that, increase the Max Speed Tolerance. It is dramatic.
Keep in mind that a tighter curve needs a longer length of ease-in, and a tighter curve results in more cant up to the track rule maximum. I do not know how these numbers relate and the Dovetail help is useless. There are (apparently) three factors that control superelevation in the track rule. Besides MaxSpeedTolerance (one for each Main, Yard, Passenger, Freight) , there are two values: MaxCantAngle which is obvious and CurveToAnglePercent. The latter factor is inscrutable to me. Dovetail says
The track rule defines the maximum tilt and the tilt to radius ratio. The whole ribbon is superelevated. The curve must be eased. The easement must also have tilt selected.
Tilt to radius ratio? That is the ratio of an angular value to a linear (distance) value. Apples and oranges. I don't get it. I wish I did. So I just left that at 40%.
Here are some results: For a tight eased curve of 150m radius, the StevensPass MaxSpeedTolerance of 40 results in an easement that is only 52 meters (170 ft) long. Its MaxCant is 3 degrees so at 60 mph (88fps) the easement is passed in only 1.9 seconds. That's a roll rate of 1.6 degrees per second. Wow! So clearly this is the issue. So I then changed the Max Speed Tolerance to 70. The easement length increased to 283 meters (928 ft). At 60 mph it takes 10.5 seconds to pass through the easement for 0.28 degrees per second. You never feel it. It is smooooth. So then I checked a much shallower curve, 1200 meters. At MaxSpeedTol of 40 the easement is only 6.6 meters, and at 60 the easement lengthens to 35.5 meters. Here, though, I have no idea what cant is calculated for the snallow curve, so I can't calculate the roll rate. How to you ratio an angle to a distance??? Oh well, trust me the result is still super mooth, even through the easement lengths are much shorter for the shallow (long radius) curve.
There is a downside. The easements will be longer, making it harder to follow decals and Google Earth (unless the real curve is also eased ha ha). I have no idea why such low Max Speed Tolerances were used in some of the routes. Anyway, have made up my own variant of the SR_SP track rule with altered values including a Max Cant of 5 degrees for my project (I have the UP track chart and there are 5.5 deg curves). Here are my values (and I need to work in a test route a lot before laying too much "final" track)
- Code: Select all
Cima Cima Yard
DefPrimSPeed 79 15
DefSecSpd 70 15
MainSpeedTol 60 20
MainMinRadius 175 125
YardSpeedTol 20 20
YardMinRadius 125 125
PassSpeedTol 80 20
PassMinRadius 190 125
FrgtSpeedTol 50 20
frgtMinRadius 175 125
SEMaxCantDeg 5 0
SECurvToAng% 40 *
Roughness 15 30