Hi,
The Curvature Diagram in OpenRail produces different output compared to the Curvature Diagram in BRT.
We recently had a training where we noticed this.
It seems that in OpenRail this diagram does not actually displays curvature (1/R) but a different parameter called deflection. Details here: Horizontal Curvature Diagram (bentley.com)
The issue with this new concept is that, for constant curvature, if the chord length is variable then the deflection varies.
See the graph below - in blue is the curvature diagram and in dashed amber the deflection. the area where the two differ is where I changed the chord length, the curvature remaining the blue line.
This is an issue for a track designer because the survey file can often include rail strings with variable chord lengths. For example, in the UK, the turnout area is surveyed at 10m chord (5m between points) and plain line R>500m at 20m chord (approx 10m between points). Some laser surveys have a continuously variable chord length depending on the curvature of the string.
In these cases the ORD "curvature diagram" provides us the wrong image of the site, suggesting that we have variable curvature when in fact we have just a variable distance between points.
I find strange that a straight forward concept, well known in railway track design, is re-invented in this strange, misleading and un-useful way.
I think this needs to be corrected and the correct curvature should be displayed instead of this alternative concept.
And this new approach, of using the deflection, is erroneous also because the versine can vary, for the same curvature, depending on where along the chord is measured (as shown also in the sketch from the help page - snapshot below).
In my opinion this is a fundamental bug (flaw?) of the software from track alignment design perspective.
So, if this is to be corrected, the correction should not be on based on these versines, but on the equation of the circle passing through three consecutive points.
This is how the old BRT is doing it, being therefore able to display correct curvature values even for points at irregular interval.
See below a snapshot from BRT for three consecutive points placed at irregular interval - the radius is computed right and the curvature diagram indeed displays the curvature.
It is worth appreciating the new option ORD has of smoothing the diagram - it would be good this to be implemented for curvature.
Hi Constantin,
100% agree, this appears to be a bug in the software. I would suggest this new 'curvature' diagram is wrong and misleading.
As a test, I traced over a single curve with a line of varying point spacings, ranging from 1m to 25m and created a curvature diagram in both ORLD and BRT.
This first screenshot is the curvature diagram from ORLD. The points which are spaced closer together (1m) on the left-hand side implies a flatter curve, whereas the points spaced further apart (25m) in the middle seem to suggest a tighter curve radius. This is very misleading because in reality this is a single continuous curve, just with varying point spacing!
This is the correct, perfectly smooth curvature diagram from BRT -
(Of course, in the real world you will not see a perfectly smooth curvature diagram, but this manufactured example is just to prove the point that the ORLD diagram is misleading)
I would be interested to hear the rationale from Bentley behind this new approach.
Update: I've reported this as Service Request no. 7001486410.