 RWA member SAR704 provides this useful tutorial for laying tracks according to a gradient profile.

I'm not sure how many folks lay their own track in a route, with their own track rules nowadays. But a little while ago, I decided that it was too tedious to estimate lengths based on guesswork through profiles of random resolutions. The reason I use this term is because it is essential towards the technique that I have used to lay track lately.

I tend to use Irfanview for simple image editing, resizing etc. The program can also be useful just to zoom into a section such as in the image attached. The selection (arrow) tool lets you select an area of the image and Irfanview displays the size in pixels of the selection at the bottom of the screen. You can convert this number of pixels to meters if you know the scale of the image. Here we have a gradient profile. Whilst the original chart was in poor condition, and is no longer in my possession, I took some lousy albeit adequate photos of it in 2014 which prevented the need to go hunting for another copy of the original again.

Note that the miles amount to an average of 400 pixels, give or take a few. The reason for this is to enable me to multiply the values from Irfanview by 4 to determine the length of the gradient. For example, there is a downward gradient of 1 in 80 near 144 mile point. If I highlight it with the freehand selection tool, then the width amounts to about 136 pixels. Multiplied by 4, it is 544 metres of length.

If I drag between the half circular shapes in the image just below the 143 and 144 mileposts, the amount of pixels amounts to approximately 472, which is roughly 1888 metres on the actual line. The only issue that we need to determine the length of each gradient.

The short 227.6 portion is roughly 25 pixels, so about 100 metres long. The level portion is about 664 metres (176 pixels roughly). If we combine these values, and determine the total length of the straight section, then there is still a risk that the total does not add up to exactly what's expected.

There can also be differences when laying the actual straight, which is why it is useful to use the 'follow network' tool within the editor interface to get an idea of the exact length of the straight section. What I mean by this is determine exactly where the next curve starts, place a straight piece of track directly after this, then use the snap to track tool to place a straight section which could be 3300 metres in length depending on the situation.

The only catch here is that it is NOT IDEAL to attempt to link sections where a length LONGER than 4500 is required. Also, TS does not seem to link copying and pasting these straights, despite the face that they render. So for longer sections, it is recommended to place multiple sections that do not exceed 4000 metres in length. This is provided that you have a midrange PC or better.

Now, given that I specified 472 (before I went off track, no pun intended), and after measuring the section, it is actually (as an example) 2024 metres long. So what I then do is count the number of differing gradient values in this section and also calculate the overflow which is 136 metres.

Given that there are 4 different gradients in this short section, it makes sense to relay the 4 sections of track between the curves with a 34 metre increase in length. At the end, it may be a tiny bit off (10 metres possibly). This is where you can choose to add the 10 metres to the last section, or add 2.5 metres to each of the 4 portions.

But regardless of the above, this technique has saved me several hours of hard work with track laying, and turns out to be relatively precise, provided that you are happy with the sections of track amounting to a power of 2, which is an ideal habit with this practice.