Aurecon's Troy Burton, Mass Transit Expertise Leader, and Michael Gardiner, 3D Visualisation Leader, discuss in this technical paper originally presented at the 2012 Conference of Railway Engineering (CORE2012).
This tool has been used as an integral part of the design process such that the final design ensures that train drivers are afforded adequate advance sighting of the signals and receive a clear and unambiguous message.
It has also significantly reduced the risk of costly construction re-work which often occurs on rail projects at the final commissioning stage when signal sighting issues are first identified. A particular constraint on this project was the background conditions to the signals in the form of glare from road vehicles from parallel roads for a significant section of the rail corridor.
Signal sighting from a train driver’s perspective is a very significant operational and safety issue for railway authorities/operators.
Signals must be positioned so that they afford train drivers adequate advance sighting and convey a clear and unambiguous message. The drivers approach view is the prime consideration but regard must also be given to the signalling arrangements and associated operational speeds.
Typically new or upgraded rail lines are designed with these important considerations in mind with a typical 2D design drawing approach with limited 3D visual aids. In the past this has led to costly construction re-work just prior to final project commissioning.
This comes as a result of signal sighting committees/groups undertaking final and lengthy signal sighting walkovers along the tracks of the new/upgraded lines and identifying obstructions etc. which then need to be rectified, sometimes at significant cost.
With recent emerging computer graphic technology, Aurecon have devised a 3D Visualisation tool which explores the relationship between signal position, the train driver view, and the background conditions including adjacent roads and lineside equipment.
It assesses the impact on the train driver’s ability to sight rail signals from a safe and predetermined distance.
A 3D representation of the overall design was created and then played back from the perspective of the train driver. This provided the ability to instantly recognise problem areas for further investigation simply by visual analysis of the design.
Key areas which were incorporated into the model included a detailed analysis of the visibility of rail signals with regards to OHLE masts, structures and rail signals, particularly on sweeping turns.
As the project included twin tracks and full bi-directional signalling the visualisation was created with four distinct passes, representing the train moving in both directions on both tracks.
There are numerous key issues to consider when designing signal positions. All issues have varying significance depending on the actual application.
A list of key issues applicable to this project is as follows:
The listed issues were considered, discussed and reviewed by the design team in conjunction with the operator and signal sighting committee. The basis for this collaborative effort was the 3D virtual model.
In order to create an accurate environment for analysis, it was important that the design model and existing survey were both physically correct and of sufficient quality to provide a realistic depiction of the finished project.
Once the earthworks model was created in Bentley MX, it was integrated into the survey model that was obtained through a combination of Light Detection And Ranging (LIDAR) and traditional surveying techniques.
A triangulation was then created, and this triangulated irregular network (TIN) mesh was then exported out as an AutoCAD DWG file to be imported into the visualisation package, Autodesk 3D Studio Max.
Upon import, textures were applied to the model and in particular the aerial photography was ‘draped’ over the survey to give a visual indication of the actual environment. In a similar fashion final hard and soft scape was applied to the earthworks model to provide an actual realistic finish.
The typical cross section for the rail formation was then lofted down the control strings, and aligned such that the strings were matched to the top of rail as per the design specification. The control strings sat within the earthworks model and were separated locally around the station island platforms and so they were an accurate reflection of the design.
Stations were originally created by station designers in Autodesk Revit. This information was then imported into 3D Studio Max at the correct location and with the desired textures and material finish applied.
The bulk of the materials finishes were applied by the station designers using the software’s extensive libraries. However custom finishes were created in Adobe Photoshop from photographic reference.
The stations were originally modelled in an extremely high level of detail. This kind of data was not necessary in the visualisation so they were heavily optimised upon import, and care was taken not to affect the silhouette of the structures, but only to remove irrelevant details.
To supplement the design model there were a number of additional 3D models which were created and placed into the visualisation, such as rail masts, signalling and bridges.
These were created based on standard technical drawings to ensure their accuracy and scale.
The standard technical drawings were imported into 3D Studio Max and used as a template to create the model in three dimensions.
The virtual camera needed to be placed precisely at the correct height of the train driver’s point of view, so it would simulate the driver’s vision. This information, sourced from Queensland Rail standards, was determined to be 2.5 m above the top of rail.
The 3D Visualisation artist collaborated directly with rail designers, structural drafters and station designers to ensure the models were of the highest accuracy and latest revision. Frequent review sessions were held to ensure there were no visual discrepancies.
A key purpose of the visualisation was to analyse the signal structures, in particular if they would become obscured to the driver’s vision against background traffic on adjacent roads. This would be an issue particularly at night when there would be a large number of headlights visible to the driver.
The design team needed to ensure that at no point would there be background interference or distraction to the drivers advance sighting of the signals from adjacent road traffic head lights.
A number of possible solutions were discussed, including populating the roads with a high degree of traffic in order to simulate a peak hour scenario.
This idea was discarded because the design team needed a solution that would show all possible lighting scenarios at every relevant point in space, not just snapshots of traffic at random times.
The final solution was developed to represent the headlights as a band of light stretching down the surrounding road lanes.
The band would be of sufficient height to encompass the light of everything from the lowest car to the highest truck, resulting in an environment where every possible scenario and combination of options could be displayed at once.
The exact heights for these light bands were obtained through manual measurements of a wide range of vehicles, and were found to be 0.5 m to 1.37 m for headlights, and 0.8 m to 1.15 m for taillights. This is considered to cover the majority of road vehicles and appropriate for this application.
Early in the process it became apparent that the virtual environment would need to be constructed with a physically correct lighting space – the rail signals needed to be visible from a distance with the same level of intensity as they would be in real life, as would the bands of light representing road vehicle headlights.
This necessitated the adoption of what is known as a ‘Linear Workflow’. Traditionally, 3D movies are generated in a linear colour space, while monitors and television displays show their content in an alternate colour space, known as sRGB (standard RGB).
The resulting difference when converting between these two colour spaces means that lights and shadows are displayed incorrectly. In particular, brightness no longer shows in a physically realistic manner.
In order to display colour and light correctly, it was necessary to construct the scene and lighting solution to match the two colour spaces and eliminate any conversion discrepancies. This process is known as adopting a Linear Workflow.
Once this was set up, real world lighting values could be assigned to the various elements, measured in candelas per metre squared. Candelas are a standard method of measuring luminous intensity, or the power emitted by a light source in a particular direction – a common candle emits one candela.
Through research and information provided on standard drawings the correct lighting intensities were found for the coloured LED lighting in the signal masts – 300 cd/m² for red, 350 cd/m² for yellow, and 500 cd/m² for green. This was applied through 3D Studio Max’s object properties, allowing a light intensity value to be added to any object – in this case the LED globes.
In this manner the design team was assured that all lighting information, from the ambient moonlight to the rail signals and road vehicle headlights, is displayed at a physically correct level of illumination throughout the project.
This became particularly important when analysing signal positions in the far distance. As in real life, the LED lighting was bright and clear even over long distances, allowing the designers to easily recognise when a signal mast first became visible to the train driver.
With the twin tracks being bi-directional there were four separate movies created. For easy reference and orientation it was decided to display the track orientation and current chainage in the corner of the screen, with the chainage dynamically updating as the camera moved along the virtual design.
Ordinarily a straight conversion between the current time and start chainage would yield the correct chainage at any point. However, for visual convenience, the virtual train increased and decreased speed at the start and end of the visualisation over an exponential curve, simulating a smooth start up and stop. On this basis a script was created that calculated the chainage on a per-frame basis by analysing the current percentage travelled along the rail control strings, and relating this back to the known start and end chainages of these strings through a mathematical formula.
In this manner, the correct chainage could be displayed at any point of the visualisation and this, together with the track orientation, would allow the user to instantly understand their location.
It also aided during presentations and other instances where the visualisation would need to be paused. The movie could be stopped and the exact stopping location would be known as the chainage would be displayed in the corner. This allowed designers to easily measure distances to known features – for instance when a signal mast first comes into view.
For the purposes of the visualisation, a nominal speed setting was selected in order to reduce the overall run time of the movies. The speed adopted is far above normal train speed. The actual train operating speeds could be implemented quite easily if required.
A potential further benefit of this 3D visualisation tool could be the use by the railway operator in train driver training. The model is accurate and could be used as an additional training tool for new drivers to this section of the rail network.
Familiarity could be quickly gained of the overall line topography, alignment, station locations, bridge locations, speed boards and most importantly signal positions.
The 3D visualisation achieved all the desired outcomes, providing an effective tool for the design team, operator and signal sighting committee to analyse and resolve key signal sighting issues along the corridor such that the final design ensures that train drivers are afforded adequate advance sighting of the signals and receive a clear and unambiguous message.
It has also significantly reduced the risk of costly construction re-work which often occurs on rail projects at the final commissioning stage when signal sighting issues are first identified.