The way to show (and understand) the need for guard rails is quite easy. I'll probably make a video demo eventually, but for now, here's a descriptive text story that should allow anyone can visualize (or construct) to see it more clearly.

1 Imagine a sharp 18" (45 cm) radius curve made of sectional track, and a loco pushing few stiff cars in reverse around it. Also try and imagine how much (what length and depth) of the outer wheel flange is below the rail head on the outside wheels. That's the only thing keeping the train from just going in a straight line and falling off the curve. Agreed?

2 Now imagine if you cut away sections of the outside rail, say 1-2" (2-5 cm) long, in the outside curve rail at around 6" (15 cm) intervals. Now will the train stay on the curve or just leave the track? If you think the train will leave the track, go to step 3. Otherwise repeat step 1.

3 Now imagine that although each rail section has been cut away, we have somehow put a smooth flat metal plate in the place of each cut out section, but just far enough below the rail height, so that the wheel flange tip can just smoothly roll onto and along it, without dropping down at all. Now we can imagine the outside wheel just carrying on smoothly once it enters the cut away gap, then just as smoothly rolling back onto the rail at the end of the gap, it the wheel stays lined up with the next intact rail section.

But do you believe that the wheel will somehow align it itself with the beginning of the next rail at the end of each gap, if we are on a curve? If not, got to step 4. Otherwise repeat step 1.

Also, at this point, please note that frogs do not have a plate to carry the wheels smoothly over their gap. But they do have "wing rails" which serve the same purpose. To keep things simple, I'll leave that explanation out for now, or you can look it up on the NMRA website.

4 By now, we should all agree that the only reason any car goes around a curve is the pressure of the outside rail acting on the flange of the outside wheel to steer that wheel away from it's natural tendency to keep going in a straight line, and force it instead to change direction to match the curve. Agreed? Then let's move on.

5 At this stage, someone is going to say that this is all fine and dandy, but the frog sections of (most) turnouts are straight, so in the case of a frog gap, the unguided wheel will just automatically keep going and end up exactly in the right place to go back onto the frog vee point.

And that of course is mostly true, except that most cars are much longer than the straight section of most frogs. So while in a simplistic theory, (at this stage) the guard rails could be ineffective on the straight side of a turnout, the diverging side of any turnout is the same case as the plain curve visualization above. Why?

Well, before going on to step 6, think carefully about what you think the angle of the wheel is in relation to the rail it is rubbing against, and what the shortest length the cut away section could be to avoid any problems. Wheel axles don't automatically turn with the track, they are fixed to a truck, or in the case of a four wheel car, to the car body

6 So now imagine if the curved example we just covered was modified so that, wherever we cut out the rail sections, the track was made straight, and also straight for an extra inch on either side, just like a straight frog section. Although the short track section is now straight, the back end of the car whose front is just entering that straightened section, is of course, still some way back on the curve, and so is not yet pointing exactly in the direction of the straight section rails. So the car body (and thus even its front truck pivot, if it has one) is still moving at a slight outwards direction, relative to the straight section. This means there is then still then pressure needed on the outside front wheel flange(s) to hold and correct the direction of the car and/or front truck to line up with the new angle of the straight section. And with the outside rail missing at the gap, that pressure can't happen.

And as you can well imagine, the reversing train is still going to leave the track, even if the gapped sections in the curve are themselves relatively short straights. So what do you think is an easy way to fix that?

7 At this point, we should all be in agreement that the train is going to leave the curve through the gaps, unless we have a method for redirecting the wheels, so they automatically still follow the track when the outer rail is missing.

The simple method to do that is a guard rail. This is placed a short distance in from the INSIDE rail, and replaces the missing pressure of the outside rail, with a new, almost identical pressure on the BACK of the flange of the INSIDE wheel instead.

If we now imagine our curve, with guards rails fitted opposite the gapped section, we can see that the train will round the curve safely in either direction, no matter how many gaps there are in the outer rail, or how long or short they are. In fact if the imaginary support plate (or equivalent system) was continuous, we wouldn't need the outer rail at all. (and there was once a prototype for that)

8 Let's now go back and address the situation of the guard rail on the STRAIGHT route of a turnout. Although in step 5, we said it might not be needed in theory, there are some very practical aspects to consider both for the models and the real railroads.

Both models and real railroads have coned wheels. And both have a degree of sideways slop in the gauge, so that the wheels can run sideways, up and down on their coning, to help steer the wheels continually towards the center, when running on straight and slightly curved track. It also keeps the flanges from continually bumping and rubbing against the side and edge of the rail head.

But no mechanical system is perfect, so real wheel sets do swing back and forth sideways across the track slightly, when running, even on the straightest track, and of course whenever the radius changes, and after the occasional track bumps and imperfections. So to some small extent, trucks and cars can actually be temporarily twisted away from the straight direction part of the time. The is one cause of the "sway" that is very common on older passenger cars and freight wagons.

If we consider a train that is being pushed in reverse, then the stiffness (friction) of the leading cars will always cause considerable "zig-zagging" of the intermediate cars, as they will take the paths of least resistance, within the "gauge slop" before the couplers impart the full pushing force.

So on any scale railroad, the wheels on some cars may happen to be close to one rail edge and others more toward the opposite side, as well as many approximately centered. This isn't a problem normally on plain track, but can make the straight route of a frog, be as much of a problem as the curved route. The risk is that a wheel already at the frog side rail edge will then continue to move far enough over at the beginning of the frog gap, that it continues to go even more sideways through the gap and the flange will then hit or cross the frog vee point, causing a major derailment.

Again, placing a guard rail so that it just prevents the extreme sideways flanges hitting the frog vee is essential. However, positioning it so that it also re-centers the passing cars wheel sets that are just slightly wrong, reduces bumpiness and gives much smoother overall running. (See step 9.) The stresses on the straight route guard rail will be statistically less than on the curved route, so often, an extra heavy clamp is placed on the curved route guard rail, to save using stronger bolts, etc., on the curved route side of the turnout.

see http://www.proto87.com/media/frog_dwcopy.jpg

9 There are other issues with model and real railroads that have to do with preventing excess bumpiness of the cars' rides and real railroads worry about heavy weight pounding damage on the frog point, that guard rails also reduce and partly prevent. e.g. Because a wheel that moves slightly sideways while passing through a frog gap will have a different coning radius entering than when leaving. Then of course that the wheel, axle and to a lesser extent the (very heavy) loco/car body will be be either lifted or dropped a small height amount. But track maintenance prevention and reduction on real railroads is a whole other story.

10 In conclusion, guard rails can be clearly seen to be necessary on both routes, wherever the effective length of the wheel flange is less than the length of the frog gap. But regardless, guard rails also make the the passage of the average wheelset in the train much smoother.

So your operations session running will be quieter and derailments fewer (hopefully non-existent.)

A small set of Pictures showing the Heavily "Rubbed" Guard Rails on switch #175, North of Grover Beach Station, on the LA-SF Amtrak Coast route.

Note these are large pictures around 80K each and may take time to fully download

Although the marks look dark gray in the pictures, it is because of the their vertical angle to the sky. They are in fact bright polished steel, just like the railheads

General view looking North

General view looking South

Straight Route Guard Rail

Curved Route Guard Rail

Close up of leading edge of Guard rail