Thursday, September 8, 2011

128> Learning from Roller Coaster Design


I have previously written about how we need a three dimensional approach to transportation, and one of the main themes I have returned to, design-wise, is of a PRT system that can be routed with minimum of restrictions – a track that can go up or down steeply or around curves tightly, coupled with a vehicle to match. 

It is very tempting to model raised track after bridges.  After all, in most cases, that is essentially what we’re dealing with.  But bridges, as we know them, weren’t really designed for the task at hand. Bridges for ordinary vehicles are necessarily gradual in any curvature because vehicles are generally heavy, and so have great momentum, or are not running on tracks, so they may skid off course.  With a vehicle that is both light and locked onto rails, turning radius or elevation changes can be taken at speeds that are as fast as passenger comfort will allow.  In the past, it was necessary to keep all vehicles going at a single, constant speed.  At least from a computer/control standpoint, that is no longer the case.  If there’s a place where only a hairpin turn will work, the entire system need not be held hostage to the speed for that turn.  In the case of empty vehicles, it makes little sense to run the vehicles unnecessarily.  Garaging the vehicles in shaded (or even climate controlled) storage locations could be advantageous, but would be particularly so if its feeder track, from where it diverges to where it rejoins, is as compact as possible.  Staging, garaging, and repairing vehicles takes space, and space is expensive.  I have seen little in current PRT designs that acknowledges this reality.  As a matter of fact, little attention has been paid to the amount of track involved in acceleration/deceleration lanes for stations.  Perhaps this is because the systems have traditionally been designed to be slow.  Unfortunately, going even reasonably fast opens a whole can of worms, design-wise.  But a really smooth, quiet and fast ride is what will make believers out of the passengers.  A slow, clunky implementation is what will ensure that PRT doesn’t catch on.    
Here are some thoughts regarding actual construction of track:  First of all, the track will no doubt be made in sections in a fabrication shop and trucked to the site, where it must fit together. Any on-site welding, if any, will be minimal, especially considering that expansion joints will be required between sections or groups of sections. Steel can expand nearly an inch per hundred feet between record temperature lows and highs for many areas. 

Although it is possible to bend any shape of structural steel, pipe is by far the least troublesome, at least when it comes to complex curves, where the steel must bend, at once, both up or down and sideways.  Squared stock, having a top and bottom that should remain level and sides that should remain plumb, presents a challenge that does not exist with round pipe.  Squared profiles can be produced with precision from welded flat stock however, although that is a lot of welding.  Pipe joints can easily accommodate expansion with a tightly fit inner sleeve that is only welded on one side.  The outer, running surfaces can be angle-cut or even finger-jointed to ensure a smooth ride.   

Presumably track would be assembled on some sort of scaffolding – a big jig that would establish the endpoints and angles while supporting the pieces for welding.  Pipe bending is an imprecise business, as there is some tendency for steel to spring back. Requiring radii of absolute precision is a recipe for very high costs, so any design should accommodate this fact. 

Luckily, such challenges have been faced before by the makers of roller coasters, and I think that their design conclusions apply here as well.  In the top picture, it appears that the large pipe may actually be many segments of straight pipe with only the small pipe being actually bent, although we can’t be sure.  That certainly is a possibility for eliminating some bending altogether. Note the periodic bolted flanges. This universal connection scheme greatly simplifies assembly in the field. I have looked for, but not found, expansion joints.  I believe this is because the loops and curves can enlarge in terms of radius, eliminating the need.  This system is clearly not as strong as it would be with the same weight of steel used in a triangulated truss design, but the simplicity of fabrication more than makes up for it.  Actually, triangulated trusses are not unheard of in roller coasters, as my Google image search revealed, but I think the point here is that with sufficient support they can be removed.  Consider, for example, the track as it approaches the docking area.  Depending on the situation, the track might curve in complex ways, while supports might be quite closely spaced.  Here you would need no triangulating trusses, and, indeed, they would be all different lengths and an unnecessary complication.  A straight run over a highway, on the other hand, would call for a stiffer design. In that case the trusses would be all the same length and can be easily added.  The picture below shows variations with and without trusswork. My apologies if the design looks a bit half-baked… It is a work in progress. 

The idea of standardized, modular lengths brings up a question.  How long should the sections be and why? At the moment I am leaning toward shorter lengths for curves than I had originally thought, principally because I worry that longer lengths might not easily fit together in the field. (It looks like the roller coaster designer concurs.)  Also shorter lengths would seem to be more versatile, enabling a number of transition options. For example, a higher speed turn might incorporate several radii so as to not be too abrupt. Straight sections, I suppose, could be designed around what would fit on an 18 wheeler.  A pair of trailer length segments, bolted together, would easily span a four lane road with a turning and bike lanes.   

The problem reminds me of the slot car set I had when I was a kid.  I had several types of curved and straight track, all in short lengths, and these could be assembled into any number of layouts.
Finally, one advantage to dividing the track into “bite-size” pieces is that it would be easier to put a price tag on this whole thing. There is very little guidance on how much PRT hardware will cost, especially broken down in terms of stations, track, and vehicles. At least this would be starting point for the infrastructure part of it.