Building Belah Viaduct Continued

Phil Baines talks us through the process of building the incredible structure that is Belah Viaduct:

The Building of Belah Viaduct - Part 2

  Having built the handrail, the next section to be constructed was the deck. The reason for doing the deck next was to do with the shadow baking. As the upper portion of the supports would be under the deck, it was necessary to have it in place to cast shadows on to the supports.

  As with any major construction it can be daunting looking at the project as a whole, so the best way forward is to break it down into manageable chunks. With 16 spans, each span becomes an obvious chunk and that is how I built the deck, in 61ft sections. The model would also be used for the Deepdale Viaduct. Deepdale is curved so I decided I would need to build it as a Lofted item, therefore, it made even more sense to construct the model in single span sections.

  The start was simple, an 18.9m(61ft) long by 7.4m(24ft) wide deck which was set with its upwards facing poly at zero in the modelling  software. The deck was set as ground level so that when the Deepdale model was attached to the track, using the Offset tool, it would be at the right height.

  With the deck in place, all that was necessary was to replicate the various supports and braces underneath. I would make one brace or support at a time, unwrap it, then duplicate it as required. I make a basic assumption that the light will fall on identical parts in an identical way, so unwrapping then copying saves a lot of unnecessary duplication of effort.

  Although it was decided to make this a high poly model it is still not necessary to model every detail of the bridge, so a certain amount of artistic licence is involved in choosing the elements to re-produce. The initial view of the deck span looked like above and the observant among you will note the missing elements.

  As can be seen in the above photo, the rails crossing the viaduct ran on top of wooden baulks of timber, so the final version of the deck span had these added, then the model was textured and shadow baked with the following result. I also modelled in a check rail while I was on with it.

  Next I made a start on the supports. There are 15 supports and each support is constructed of  5m high sections, ranging from 2 sections in the shortest to 11 sections in the tallest.

The construction starts with a single central column which is displaced 2.25m along the Y axis, and then unwrapped. The column is then copied and the copy moved and inclined to represent the side column, which is repeated for the other side. The UVW maps are then selected and moved to a clear area of the texture sheet, so, when shadow baked  they display the different shadows cast by their  different orientation.

  The three columns were then grouped together and the group copied and rotated 180deg around the 0,0 point. This orients the same polys to the inside of the section so I can use the same UVW map.

  Using identical techniques I then built up all the elements of the section, adding the various beams and braces until I had a complete section ready for duplication.

  You will notice the strange extensions to the cross beams, which were added to help me during the duplication process.

The duplication process was done in three different ways. Firstly the central features were simply copied and lowered by 5m. Secondly, the side elements were copied, lowered 5m then  displaced sideways by 0.45m so their tops lined up with the bottoms of the layer above. Thirdly, the cross braces were lowered then scaled to fit the wider cross section. This stretches the texture slightly, but on such minor items I considered it an acceptable compromise. I didn't think this would be acceptable for the cross beams, as they were much larger and texture stretching would be obvious, therefore, I made them long enough for the widest cross section, but with pieces I could chop off to fit at each level.

I then have a structure like this:

  As mentioned earlier, the top section is in the shadow of the deck span, so I now select all the elements of the topmost support section and move them on the UV map, then shadow bake both sections.

  All that remains is to duplicate the lower section 9 more times and I have replicated the tallest support. This is, in turn, duplicated 14 times to generate all the supports. It is then a simple process of chopping off unwanted sections to create supports of the right height, whilst trimming the cross beams to the right length at the same time and we are nearly  there.

  One further tweak is required. In this picture from the ©Turner collection it is possible to see the lower half of each bottom section was painted black.

 Whilst this was not a consistent feature over the life of the viaduct, it was decided to add it to the model.

  This was achieved by, once more, playing with the UV map. Having selected all the bottom section elements of the model, they were unwrapped and then their UV maps moved to a clear area of the texture where the black paint could be added.

  The main elements off the Viaduct were now finished and in my next blog I'll detail the final bits and pieces to get the viaduct in game.

Heading West

Development of Eden Valley Line - By Ben Yates

Kirkby Stephen East Junction - Stainmore route to the left and the Eden Valley line on the right

  Appleby (latterly Appleby East) was opened in 1862, 14 years before the Midland Railway rolled into town under the duress of Parliament with the railway they didn't want to build - sorry S&C fans, but it’s true, and this makes me wonder if the locals really rang the church bells to celebrate parliamentary approval of the railway coming to town in 1866, when they had an operational line 4 years previously. The Midland station was a stone’s throw across the Schoolyard, but critically was at a higher elevation by some 40 feet, just 200 yards apart. The 2 lines were joined with a north facing junction on the Midland, and a south facing junction on the EVR north of Appleby East. This made the line useful as a diversionary route to the northern half of the S&C, and there are records of the prestigious Thames Clyde Express (behind a Jubilee) having been diverted over
the EVR, so with Class 40s having run to Warcop it can be assumed the EVR was
built to higher loadings than Bouch’s bridges on Stainmore. The link remains to
the day despite the closure of the last section to Warcop, and is still used as an occasional refuge for special trains allowing service trains to pass. The gradient is around 1 in 77, which with a reverse curve presents a challenge to get a train started.

An overall view looking down at Appleby East, to the
left, and the Midland route to the right

  This brings us onto track-laying, and the process of how to go about capturing this in the sim. The starting point is to take the gradient profile and blow it up as large
as possible – this one is done to A3. After checking for copier distortion, work out the distance of each section to the metre. I then check this by measuring the whole length of the section and checking against the cumulative lengths of each gradient. This has usually been within 2-3% accuracy, and individual lengths are adjusted accordingly. In scientific terms, there is bound to be a significant margin for error when dealing with conversions of 1mm = 160m, but the method seems to work pretty accurately in practice.

To get a match between the profile and the DEM, it is necessary to find a known point of altitude of the railway. This is best done by finding an OS Map contour crossing the line where the line is level with the surround terrain. This method can be used to check periodically as track is laid.

The track is laid with easements, which allow use of superelevation to add cant to the track on curves. Easements are the gradual increase in radius of a curve, which gives a smooth transition to the correct radius. The tool in TS13 is, unfortunately, a swine to use and impossible to lay track under the ground, so large parts of the line have to be laid conventionally, and the terrain formed
to allow the track to be relaid with easements. This is all rather time consuming, but the end result is worth it. Another limitation (though sometimes ignored) is that applying superelevation to junctions means the don’t render properly and provide an awful ride from the cab. I make sure the easement section ends before crossovers to avoid this. 

A super-elevated curve just west of Kirkby Stephen, on the Stainmore line

  Appleby was something of a challenge. The key thing was to get the S&C trackwork over the EVR (now the S&C bridge over the A66T) at the correct height for the overbridge, and alos to match the road bridge that goes under the platform. Unfortunately there are 2 changes of gradient on the S&C between there and Appleby Station – level, climbing 1 in 200, then 1 in 400 and it is
impossible to get an accurate measurement from the tiny Ian Allen gradient book. Trial and error matched the track with both bridges (a good hour’s work!) which allowed the joining line to be worked out.

A further complication was the complicated trackwork at Appleby North yard, which includes a double slip – notoriously hard to render correctly, and impossible to do so when using correct track spacing for the 6 foot as we do (the observant of you may have noticed that default tracks have a wider spacing, presumably down to this). To solve this, tracks are laid singly and offset at a
slightly wider spacing, which did the trick. Once the junctions are laid at each end of the link, a piece of track was laid and manipulated to establish the ruling gradient of 1 in 77 (I’ve seen 1in 80 quoted) and after much fiddling I got both lines to join. I’m not totally happy with how the curves match in radius, so there may be some further work to do on this. The junction was singled around 1959, but I wanted to include the full trackwork to reflect the line at its operational peak.

The junction at Appleby

There’s still a turntable to install – it appears on the maps so I assume it lasted until the end of steam – answers on another postcard.

Stainmore: Lines and Triangles

Lines and Triangles by Ben Yates

  When I first started building routes (and as per my last blog, I see simulation as a means of recreating what isn't there any more) , I had no contemplation that route building involves most of the time not building routes at all. Stainmore is no different.

Credit: NER Society

  To build something modern has the huge advantage of it being there. There's likely a cab ride video, you can ride the thing first person, if you've got a friendly contact in the industry it's not too hard to get the engineers profile of the route with every conceivable detail of crossovers, signals and gradient changes to the metre.

  Stainmore is well documented thanks to Peter Walton's superb 1992 book, but so much is unphotographed, the gradient profile is a good effort but won't win any awards for accuracy, so a great source of information is my battered pair of walking boots!  

  Phil made a wonderful discovery whist visiting the good folks at Kirkby Stephen East that the North Eastern Railway Society publish the engineer's line drawings for the route.

Credit: NER Society

  These show the track layout, and every bridge, gradient sign, field crossing, platform, PWay hut and platform measured to the chain and yard. Gold dust to route builders, and a great reminder that the "fer" in "chemain de fer" is such a fundamental part of railways we readily forget - its not far back that a team of men "walked the line" on first name terms with each fishplate and piece of ballast. The drawings here were a working document, with crossings out as sidings, junctions and stations were taken out of use. This makes them and extraordinary historical reference, and I doff my hat to the NERS for their preservation.

  It also means we have no excuse for getting the details wrong!

  I mentioned gradient diagrams - a source of fascination usually reserved for locomotive performance aficionados, but a fundamental tool to the route builder. The RailSimulator/Railworks/TS franchise uses NASA DEM data to create the terrain, which can be quite stunning, but is not without error, particularly satellites assuming the treetops are the ground! There is something magical about running a length of track through DEM terrain to find the cuttings and embankments form just as per the map. Equally, measuring gradient, distance and checking reference heights to an OS map and GPS altitude data, and measuring to within inches doesn't guarantee it looking right - I always start with science, but route building is definitely more of an art.

  I reckon we've got this one pretty close - the descent from the Summit to Barnard Castle was within inches.

  If you've done any route building, the inadequacies of the 8m triangular terrain mesh will be readily apparent. I  suspect that a smaller mesh and some more tools could easily halve the time taken to build a route. Even with a mastery of tools, it is incredibly time consuming to get a decent look on any route with cuttings, turns and embankments, and there we picked the absolute worst prototype! I can think of many routes where you can get away with what's there, but I realised early on this wouldn't cut the mustard. (curiously, this is why there is almost no narrow gauge in this sim, which is a crying shame. Scaling up to full size was a clever trick of the MSTS era, maybe the pics RS posted of the WHR may come to something? Holding breath may result in asphyxiation.)

  Phil, as so often the case, tied his white horse up outside, hung up his armour and delivered a set of embankment and cutting lofts that make the route feel like it is to walk the line.

Here's some progress pics:

Next time: track laying, Smardale and the Lune Valley to Tebay.