Crossing the tram lines SLOWLY at a shallow angle is THE GUARANTEED WAY to have the front tyre deflected to line up with the groove and fall off - I'll happily proove that, and have tested it - cross fast and with firm but not rigid control of 'bars' nd the momentum of the bike counteracts the turning moment whan the tyre rests only on 2 smooth steel edges.
When you get slowed down by a vehicle blocking the clear route through (like a bus) then you fall
The other type of fall is when you put a side force on the tyre contact patch when it sits straddling the groove on those 2 smooth steel edges and the bike slides sideways. This happens regardless of the crossing angle, especially when the rail - road transverse profile is substandard relative to the standards both advised by ORR guidance (2006) and set out in the Tramways Act 1870.
The MAXIMUM overall differnce in road-rail levels is 6mm and the case Roe vs Sheffield Supertram further recognised the standard for metalwork standing proud of a road surface is 3mm MAXIMUM*. The Edinburgh street track installation has struggled and largely failed to deliver on this.
* as previously posted a whole raft of standards for vertical discrepancies in road pavements, tactile paving, rumble white lining, thermoplastic road markings, steelwork in the carriageway, dropped kerbs, level crossing panels...
sets the limit at 6mm for high friction materials and 3mm for low friction (road markings/steelwork).
I measured some of the installed rail and embedding system with a ridge of nearly 10mm in addition to the rail groove just perfectly set to bring down a cyclist - without any help from the rail.
And additionally these details hardly help..... with chunks of concrete falling out of the concrete track slab ...
https://www.flickr.com/photos/h52/36211824983/in/datetaken/
https://farm5.staticflickr.com/4285/34987974492_42ac8ecf56_z.jpg (Haymarket - right beside place where folk cycle)