A passing siding with an engine attached will be ok but iwth the rolling qualities of most car today you are going to have to have a mechanical means of stopping them or invent really tiny air brakes. Best method I have seen is using a tortise to raise a wire that the axle hits.
I can show you lots of sidings that are on a grade around my town so yes they are prototypically correct. On a model railroad however that could be a problem. I have a friend that has a couple that are on grades on his layout and what he does to stop the cars from rolling is he actually glues one set of wheels to the truck frames on a box car that will always be sitting in that siding. When he drops cars in the siding he just couples the cars he is dropping to the box car with the glued wheels. If you want to use this method it is a good idea to add weight to the box car as well, you wouldn’t want to drop say 4 cars and have them weigh more then the one with the glued wheels. It will get pulled down the grade for sure if it was just a standard weight car. Something to think about…
Although this would have been uncommon on the NYC, it wouldn’t have been for the lines out west. I have a DVD of the Durango & Silverton that shows a passing siding on a fairly steep grade.
I can easily reach all the sidings I have on grade and all of these sidings have cork roadbed. Since I am now experimenting with skewers for uncoupling, I am also experimenting with smaller size toothpicks and I simply press the toothpick into the cork to hold the car. This seems to be working so far, even with ballast.
Back in the dear, dead days of steam, the Southern Pacific had an entire division yard on a grade - Dunsmuir, CA. They could gravity-classify cars without needing a hump, but getting them stopped (with hand brakes) was a bear!
Speaking of sidings and grades. Have any of you ever noticed the differene in elevation between the siding and the main track and where a train leaving the siding to re enter the main has to climb a short grade to get back to the level of the main track. I’ve always wondered if this was intentional to keep cars from rolling out onto the main or if it is just the result of years of adding more ballast to the main while neglecting the siding. Looking at old photos, the seems to be less difference in elevation than you see today.
The answer to your question is, “Both of the above.” When possible, track which might hold standing cars is initially laid with a slight upgrade to keep those cars from rolling onto mainline trackage. Then, over the years, main tracks get fresh ballast and realignment far more frequently than do sidings and other low speed trackage. The disparity in height gradually increases over time.
Yard workers sometimes refer to their workplace as, “The bowl,” tacit recognition of the practice of having every track slope down from the ladders at both ends.
I once heard of sticking a coupler spring in one end of an axle on one truck. It sort of acted like a brake. The spring increased the friction between axle and truck. My thought’s are that tuning it right will allow it to roll without problems, yet still stop on a slight grade. I think I read that on this forum. I forget the thread name though…
Nearly all sidings have from a very slight grade to very substantial grades. Here in Kansas where I run trains, some sidings have just a slight maybe .1% grade that may be enough to let you roll into the siding with no power to certain places where you have about .5% grade ascending to a .5% descending grade from one side to the other. others are all up/down hill, or undulating. Remember that in the real world sidings are generally over a mile long and finding room to put that much track in is hard to find a level spot or a spot with no grade crossings.
The prototype used derails for protection. They were common at the ends of passings sidings and the beginning of spurs. If a railroad site had a passing siding and a spur, there would be 3 derails. If a railroad had a passing siding and a double-ended spur (which looks like a passing siding but is used to serve industries rather than train meets), there would be four derails. The variance in heights of tracks was due to differences in performance expectations and level of maintenance. Faster, heavier trains demanded heavier rail, higher density of ties, better drainage, etc.