There has to be an easier method of grading a route created with TransDem

[Approach_Medium]

Hi;
I created a route in TANE SP2 (actually, it's a beta build for what is being called SP3) using TransDem (ver 2.6.2.2 x64). This route is quite large, with the mainline running nearly 150 miles and two large yards with numerous industries. I used DEM's from the USGS, and created raster images from Google Earth. I created the route using TransDem custom colours, then added track splines from the route I had built in TransDem (also imported from Google Earth).

So now I've got the huge task of grading and spacing the tracks. Spacing is fairly straight forward, but I'm looking for an easier (and faster) method of grading.
Currently, I have been checking spline heights at two points (usually a mile or so apart) and applying a grade. There are inherent problems with this method, as the DEM's I used are only 1/3 arc and therefore not accurate enough to provide actual elevation for the rail line.

Is there an easier method?
It would be really great it TANE had a feature that would allow you to select two splines, then apply a grade between them. As it stands now, I can only apply grade between two spline points, and these are never far enough apart to do what I want.

Thanks for your help

CP

 

[Jayco-man]

The method I used was to use Google Earth, change the setting's to meters, then I just go long the rails and check the meter height of the rails and apply that to the rails in Traniz. Hope this will help some. I know this is not dead accurate, but should to close.

 

[Approach_Medium]

The method I used was to use Google Earth, change the setting's to meters, then I just go long the rails and check the meter height of the rails and apply that to the rails in Traniz. Hope this will help some. I know this is not dead accurate, but should to close.

I didn't think that elevation data in Google Earth was that accurate. If the tracks are in a cut, or elevated above terrain, would GE be able to tell you that? I am going to check it out myself now. I have dual monitors, so I could have TANE running alongside Google Earth.

Edit: Hey, this really works! Thanks.
At first I thought it wasn't working, as the elevation was reading 0m everywhere. Then I remembered that when I created the images for raster maps in TransDem, I had to set the elevation exaggeration to 0. Setting it back to 1 did the trick!

 

[Jayco-man]

The only other way would be to get a track chart of the route if that's possibly.

 

[Approach_Medium]

The only other way would be to get a track chart of the route if that's possibly.

I may have some track charts for Conrail in NJ, but not complete.
While on the subject of gradient, I am finding that placing junctions on a grade is difficult. There appears to be a tolerance for gradient at junctions. Crossovers are especially difficult. I kept getting the red splines at junctions, until I reduced gradient across them to 0. I'll have to experiment with this.

 

[JCitron]

The method I used was to use Google Earth, change the setting's to meters, then I just go long the rails and check the meter height of the rails and apply that to the rails in Traniz. Hope this will help some. I know this is not dead accurate, but should to close.

That's how I go about it. I have found the difference between the DEM in Trainz, compared to Google Earth application to be less than a meter in some cases.

 

[Approach_Medium]

Back around 2012 when I started working with TransDem and Google Earth to create Trainz routes, I don't think GE had such good resolution - for images or elevation.

 

[H222]

I may have some track charts for Conrail in NJ, but not complete.
While on the subject of gradient, I am finding that placing junctions on a grade is difficult. There appears to be a tolerance for gradient at junctions. Crossovers are especially difficult. I kept getting the red splines at junctions, until I reduced gradient across them to 0. I'll have to experiment with this.

Track charts would be close to ideal.

Regards to junctions on gradients. It can be done. I have an entire marshaling yard on a 0.35% grade with junctions literally everywhere. The junctions just have to be a very similar %age.

Marshaling Yard piccy here

Jamie

 

[Approach_Medium]

Track charts would be close to ideal.

Regards to junctions on gradients. It can be done. I have an entire marshaling yard on a 0.35% grade with junctions literally everywhere. The junctions just have to be a very similar %age.

Marshaling Yard piccy here

Jamie

OK. I have experimented, and found that a grade will work, but all segments of track need to be at or very close to the same grade. I think that with all the splines generated by TransDem, I must have missed a few and had "humps" in the track at the junction.
I always have this issue with too many splines when I import a TransDem created route. It happens because the paths I use for tracks in Google Earth are precise, and in order to follow the exact alignment of the tracks, I need to use many spline points - especially on curves. It appears that TransDem converts every one of those little dots that you see on a GE path into spline points.

I had forgotten most of what I had done several years ago when building another version of this same route. When I wanted to set the grade over a section with many spline points (such as at junctions and on curves), I would lay out a separate section of track with only two vertices and set each end to the desired height, and measure the gradient. I would then apply that gradient over the tracks between the two points. But of course I still need the heights, and until recently I don't think GE provided so much accuracy. I am very pleased to find that now they do. But I don't expect that to be the case in all areas. It appears that GE has still not updated all of their maps to the high-res. But mostly those older maps are areas we don't need much resolution anyway - like deserts... where we don't really have much to worry about with gradients... ha ha.

Another thing that is different between my last rendition of this route and the one I am working on today is that I was using only 10m grid in Traniz where I am now using only 5m. That makes quite a difference with accuracy of elevation. I may have also been using 1arc rather than 1/3arc DEMs back then. I would have used 1/9 DEMs, but they weren't available everywhere, and TransDem didn't seem to be able to handle them for such a large route. That may have been before the x64 version of TransDem was released, or before I upgraded my system's RAM from 8 to 16Gigs. In any case, I don't think DEM accuracy of better than 1/3arc is necessary for a Traniz route. Using GE maps, and 'road trips' along the route, I can determine exactly where the cuts and bridges are.

Edit: I took a look on my PC and found that I do indeed have track charts for the entire route I am working with. Because the CSX River Sub (North Bergen NJ to Selkirk NY) was owned and operated by Conrail when this chart was printed, it includes the whole route. I had never really taken a good look at it, but now seeing all the details such as gradient, horizontal track alignment, and super elevation. The chart I have was printed in 1986. It is very interesting to see that the gradient is not constant as I would have thought. There are minor (and some major) gradient changes over short distances, especially where one line goes underneath another. Conrail improved track clearances during the 1980's to accommodate double-stacks.

 

[anathoth71]

While on the subject of gradient, I am finding that placing junctions on a grade is difficult. There appears to be a tolerance for gradient at junctions. Crossovers are especially difficult. I kept getting the red splines at junctions, until I reduced gradient across them to 0. I'll have to experiment with this.

I struggled with many a junction on a gradient before adopting this technique:


1. Lay a temporary dummy track parallel, but close to, your modelled trackage which has the problem junctions.

2. Set the gradient of this dummy track to the gradient you want to maintain along the modelled trackage. You can do this by applying a gradient to it in the usual fashion, copying the height of the first and last spline points across, or just using your eyes and the terrain. Whichever takes your fancy.

3. Now use the Insert Spline Point button to add spline points to the dummy track at locations which correspond directly to the location of the spline points along the modelled trackage.

4. Now you can use the spline vertex heights in your dummy track as a "template" for the modelled trackage, so go ahead and copy the height of these spline points from the dummy track to the modelled trackage.

5. Delete the dummy track when you are satisfied with the modelled trackage.

I have had maybe a 99% success rate with this technique, and it can be applied to any track formation at all - crossovers, multiple tracks, simple siding junctions, etc etc. Junctions on gradients became a non-issue after I started doing it this way, maybe it will work for you too. An extra tip I would add is to have extra spline points on your modelled trackage as it approaches and departs the junction area, maybe 50-100 meters out. These provide a good reference point to start your parallel dummy track.

Sorry if I haven't explained this well enough, I can elaborate further if you wish me too, just let me know.



EDIT: I must have missed your last post on my reading of this thread, it seems you have already used a somewhat similiar technique for longer sections of track. I am describing something like that here, but used only "locally" over the section of track containing the problem junctions, not the whole climb.

 

[Approach_Medium]

anathoth71
Yes, I did lay the dummy track alongside, but I never thought of adding the spline points corresponding to the modeled track. It makes sense and will make things a lot easier for me.
I am going to start using the track charts I found. I found a newer (1999) version of the same chart, but I don't think much has changed, except that CSX added a new passing track and extended some of the others.
It's a huge project, and this is all before I begin texturing and placing buildings, trees, roads, water, etc, etc.

 

[Approach_Medium]

The process of grading a large route, regardless of method used is tedious and boring!
I found myself trying to use the track chart, but it is more difficult (and tedious) doing that than using Google Earth elevation data. But I think the least tedious is just what anathorth71 said. Just lay a dummy track alongside the modeled track and go with the terrain that already exists. That, unless the terrain is obviously off. If that happens, I just modify the terrain to match the grade of the track.
In the end, no model, no matter how hard we try to make it so, is going to be an exact match to the prototype.
Time for the "fudge factor..."

 

[Vern]

There seems to be a basic misunderstanding of what Transdem actually gives you. It is a model of the terrain overlaid with mapping and is only as good as the DEM or mapping data and its match in the flat earth of Trainz. That does not include an accurate gradient profile of the route you're building or the placement of embankments and cuttings for you to the exact track profile. You need to engage a bit of thought and do some surveying. Look at maps, look at aerial views, grab the gradient profile (if one exists) and adapt that to the terrain. That's why Trainz Surveyor has fixed spline points and a gradient input field. A rise of 10m in 1000m is 1 in 100 or 1.0%. No calculus or advanced mathematics required!

 

[Approach_Medium]

There seems to be a basic misunderstanding of what Transdem actually gives you. It is a model of the terrain overlaid with mapping and is only as good as the DEM or mapping data and its match in the flat earth of Trainz. That does not include an accurate gradient profile of the route you're building or the placement of embankments and cuttings for you to the exact track profile. You need to engage a bit of thought and do some surveying. Look at maps, look at aerial views, grab the gradient profile (if one exists) and adapt that to the terrain. That's why Trainz Surveyor has fixed spline points and a gradient input field. A rise of 10m in 1000m is 1 in 100 or 1.0%. No calculus or advanced mathematics required!

I understand all of that. All I was saying in my last post is that it can become tedious to properly grade a route. This is especially true where there are many junctions, crossovers, etc.
It's just something that I should not be doing when I'm tired or just 'not in the mood'. After all, I don't think many modelers do their 'work' when it becomes real work, and not the enjoyable hobby that it was intended to be.
I have all the tools to properly grade my route. I just need to have the right mindset for doing it.
I know a lot of this route by firsthand observation, as I have taken many road trips along the route. I am going to try using the track charts, but the ones I have do not include absolute elevation, so I need to get that from GE (or another source).

One thing that was causing me some frustration last night before I quit my session was that as I trace up a line on Google Earth, I see the ground elevation changing up and down by 1-2m on one track, but on an adjacent track (not the same line) it is a bit lower. That's fine, except where a long crossover (several hundred feet) occurs between the two lines. So the elevation needs to be very close at that point or the junctions won't work. I don't think in reality that there would be any elevation change on such a crossover, although I have seen serious elevation changes going into yards or industries.

While the DEMs don't show cuts and fills, it does appear that Google Earth does, which helps greatly.

 

[JCitron]

There are going to be a lot of places like this as I too have discovered as well. No matter how much we try to be 100% true to the terrain, there is no way we can due to the resolution of the data so we lose the subtle terrain differences as you've noted. In a similar situation, I worked backwards from the crossover and ended up using some retaining walls and embankment splines to fill stuff in. There was no way to do that with mesh without flooding the lower track, or pulling the ground out from under the upper track.

On another area of the same route, is a short grade leading to an engine facility and storage tracks. The grade is short leading to the facility and too short to fit within the resolution we can achieve without producing a 12.5% grade between the two points! In order to put in the facility, I had to move the engine house a bit east to give a long enough grade from the upper level and change the lead to the engine house a bit to accommodate the longer track.

And in another situation, the difference between factories next to a raised mainline, all located next to a hill, produced a mushy lump because the original DEM data didn't have the resolution. I even tried a 1/9 arc-second DEM instead of the 1/3 arc-second resolution along with a 5-meter grid, but nope it was still a mushy lump. What happened is the factories were interpreted as part of the grade, which as all located very close to the nearby hill. The same thing occurred farther up the line with some overpasses as well.

 

[Approach_Medium]

I don't think anyone playing a Traniz route is going to notice (or care) whether the route is precisely to spec. What matters is that it works from a Trainz POV. Another thing I have always had trouble with is slip switches. They can be done in Trainz (even in older builds), but it takes some work.
I will be using 5m grid for all my routes from now on, considering that my hardware seems to be able to handle it.

I think that I can work with the track charts, but I need Google Earth elevation info to 'check' my absolute elevations. If I were to go only with the track charts (which do not show elevation - only grade) then by the time I get 130 miles to the end of my route I will be off by hundreds of feet. It's kind of like trying to place mileposts by measuring distance between each one. So I started a a known point and drew a poly line along the route, watching the distance readout in GE. I noticed one strange thing though with GE distance readings:
If you start at 0.0mi, and you draw a path along the route, watching the distance reading, you will see it go from 0.99 to 1.00, to 1, to 1.0, then 1.10, and so on. Once you reach 10.0, you lose the 1/100's digit, but the effect is the same. At 100 miles, you lose the 1/10 digit.
At first I was placing mileposts at the point where the distance rolled over from 0.99 to 1.00, but found that I was off when compared to a known milepost (which I observed firsthand). I found that for distances under 10 miles (not sure about km) the point at which you have reached exactly 5280 feet is when the distance changes from 1.00 to 1 (no decimal point). When you reach 10 miles, the point is also where it rolls from 9.99 to 10. After 10 but under 100 miles, the exact mile point is still where the digits roll to the next whole number (without decimal). Once you reach 100 miles, it gets really tough. The exact mile point no longer corresponds with the change in reading from 100 to 101. You would have to watch the distance in feet to know where the milepost should be placed.
To solve this problem, I simply place my 100 mile marker, then start over with 0.0.
None of this is an issue for short routes, but if you go 100+ miles, you are going to be off by a considerable distance if you are looking to place mileposts at their true locations.

All of the above being true, I find that not all mileposts on a rail line are exactly 5280 feet apart. For example, on the CSX (Conrail) River sub track chart for 1999, the distance between mileposts 2 and 3 is 5240 ft, and between 3 and 4 is 5320 ft. As you can see, the distance is off by -40ft and then by +40ft, so there is no loss in accuracy up the line. It should be obvious to anyone why some mileposts cannot be placed at their precise locations.

I recall that someone on this forum once told me he uses a train of exactly one mile in length to measure distances. That will work, but I am placing my mileposts on the map in Google Earth, and capturing them as raster maps for TransDem, so they appear as ground textures in TANE. The only thing is I need to remember to place my mileposts before I paint over the TransDem generated ground textures.
I have also used the ruler in TANE, but it is difficult to place around curves, and unless you continue to extend one ruler to the end of the route, you will have the inaccuracies add up.

I guess all of this is why it's called Surveyor!

 

[ctclark1]

Mileposts are not always 100% accurate in the real world. (in fact if you look in some ETT's you may see a handful of "Marked miles" where, unlike the mileposts on most of the route, these two mileposts ARE exactly 5280 feet apart and can be used for checking accuracy of speedometers etc). Additionally, as you noticed, Conrail marked the distance between mileposts in most instances. These varied due to a variety of reasons, either the originals were not placed accurately, or the route of the tracks was modified slightly, cutting a curve shorter, or making a longer curve to avoid a gradient, and instead of replacing mileposts for the whole route, the distance between two mileposts may be stretched.

On the point about measuring with rulers within Trainz, I don't know if they ever fixed the meter-feet conversion bug from years back (Trainz does everything in meters internally, but calculates to DISPLAY feet only if you select Imperial...) At one point in time it was well know that there was a rounding issue that would cause the displayed feet to be off by a LOT when you get to higher distances. That said, in the last few routes I've worked on, I've used meters exclusively, despite being in New York State, because I find it easier to use Google as a conversion tool.


To answer your other question, I too use a combination of TCs and GE where possible. I also use information I know of about roads crossing the track, etc. If I know a road crosses at grade and my grading has taken my track above the height of the terrain by 2 or 3 meters, I refigure. I'll ease the grade going back a few hundred feet, or even readjust the last point where I changed grade. The other thing to remember is that even though most of the TCs have fairly defined places where the gradient changes, this is not often the case, but an averaged grade over that distance, so this will cause some variation in your model too. What might be listed as a 0.43 grade over the course of 1/2 a mile could vary between 0.4 and 0.46. There's also the case of easing the grades as they change. Rail doesn't immediately change from a 0.43 grade to a -0.1 grade, there's some variation there that is hard to approximate in trainz without inserting a LOT of spline points.

 

[Approach_Medium]

I always use metric in TANE, but since CSX and Conrail use American units, that is what I used as a reference when placing my own mileposts on my GE images. I wish that the U.S. would get with the system and start using metric by default.
As far as grading is concerned, I think that using GE elevation data and laying down a dummy track next to the route, as anathoth71 described is the best option.

BTW: What do these little bumps up and down below the "Horiz. Alignment" line on the TC indicate? Since they have numbers in deg-mins associated with them, I understand they represent the curves, but as I read the TC from south to north, these bumps appear to be going in the opposite direction as the actual curve.

 

[ctclark1]

Yes those are curves. Following their direction is sometimes counterintuitive, especially in Conrail's older charts that sometimes read backwards.

And even if the US were to magically change to metric overnight, that wouldn't fix all of these scanned track charts from 30 years ago.

 

[Vern]

I understand all of that. All I was saying in my last post is that it can become tedious to properly grade a route.

I hear you but from my perspective, laying the track out is the quickest and easiest and most disciplined part of building a route. There are many more time consuming and mind numbing processes such as trying to lay out towns and villages - which is why mine almost invariably end up as an assorted collection of houses and trees, copy 'n' pasted Railroad Tycoon style on the terrain!!