I thought that I would share my experience using 18 inch radius track with you.
I am working on my first HO layout with 90% of the track work complete. (My last layout was N scale.) I am using 18 inch radius curves on the mainline and number 4 turnouts in most cases.
Here are my reasons for using them: Less reaching in the corners; Space saving using the smaller turnouts; I am running 4 axle diesels but I do have one six axle diesel; My max car length is set at 50 foot; Passenger cars are 60 foot long. (OK, so they don’t look prototypical on the 18 R but I can live with that.) I also have some industrial sidings and a couple of yard tracks that are 15R. Once out on the main line between the corners of the room I have flowing curves that really look good.
What I did when building it: All curves have easements into them; All track and turnouts were checked with an NMRA gauge and the track was made to conform to it; All equipment was checked with the NMRA gage and was made to conform to it. Plastic car wheels were replaced with metal ones. I consider my track to be almost perfect. I can run a 15 car train backwards at full speed through the entire mainline, No 4 turnouts and all, and it does not derail, going in either direction.
The point being, that IF you take you time, and IF you build it right, and FIX any problems as they come up, all will be well in the end. Too many times modelers try to rush things, and that just doesn’t work.
If you are going to, or want to use small radius curves because perhaps you have a small layout, you can still make it work. Use easements into all of your curves. Layout your curves so they are no less than 18R when using flex track. Limit the size of the equipment that you run. Build it slowly and check everything. Don’t have any kinks in your track work anywhere. Use an NMRA gauge and check everythin
I can’t quote it from memory, but John Armstrong wrote something along the lines of “the closer you get to the minimum radius of your equipment, the more perfect your trackwork has to be to avoid problems”.
One thought for folks using tight curves it to use something like Kato Unitrack where you don’t have the problems associated with bending flextrack around tight curves. HO Unitrack has narrow-profile code 83 that to my eye looks more realistic than the popular code 83 flextrack lines like Walthers / Shinohara or Atlas. They don’t have exactly an 18" or 15" R curve, but their 19-1/4" and 16-7/8" curves are close (and give you a touch more “breathing room”.
FWIW I’ve found some engines that will go thru an 22" or even 18" radius curve won’t go thru a no.4 turnout, especially if an S-curve is involved like a siding. If the line of track you use has No.5’s they’re often a good compromise - not as gradual as a No.6, but allow a more gradual turn in almost the same space as a No.4
A #4 turnout built to NMRA RP-12.3 has a 15" radius closure rail. This can be a problem for some locomotives that need a larger radius. Not all #4’s are built to the the RP - Atlas #4’s are #4.5 for example. So some brands will work and some not. Interestingly, Fast Tracks offers turnout making tools for a 4.5 as well as for 4 and 5.
This can be crucial. John Armstrong’s Trackplanning for Realistic Operation has a discussion of what he called, with tongue in cheek, the “co-efficient of lurch” – wherein he actually tested equipment on tight curves with and without easement curves. And what he found is that equipment can run on, say, an 18" minimum radius with easement curves that would balk at 22" radius meeting a pure tangent. True, easement curves do eat away into some of the space saved with the tighter curves.
Armstrong was also a strong advocate for #4 turnouts because of the sheer amount of additional trackage they allowed vs #6s or #8s. of course not all #4s are actually #4 and some might be 4 1/2s. I have some recollection that the old “snap track” turnouts were actually closer to being #3 1/2s.
When you carefully study derailment prone areas on most model railroads you usually, or at least often, find that the modeler cheated just a little at track joints, whether using sectional or flex track, and introduced a jolting hard angle. Sometimes impossible to “see” but clearly evident just by running a freight car over the joint and watching it, or “feeling” it by having your finger on the top of the car. Often the layout owner has self hypnotized himself into believing the jolt is not there.