Junction dynamics - stuck vehicles

[gac55]

We are debugging an issue whereby we are seeing extreme delays occurring to vehicles. This is across both standard vehicles and bus, with no clear difference between modes.

We spent some time debugging PT vehicle scaling and are confident nothing is wrong there.

We have turned to investigating the flow vs speed relationships for links with unusually high travel times. Looking at individual links, the answer was not clear: we can see vehicles getting stuck without traffic being particularly high, and without any buses traversing the link.

Things however become more clear if we overlay at the "stuck vehicles" effect with volume/capacity of opposing traffic links (ie stuck vehicles on secondary links trying to enter traffic on a main link). There is a clear pattern there:

Example 1 Study link: 7885199382074589921_7885199382018096261 (in red in the figure below) Opposing traffic link: 316662

Extreme delays on the study link appear once the opposing link reaches capacity, and disappear afterwards:

Vehicles are still able to travel the main link (316662) (although at low speed), however the vehicles on 7885199382074589921_7885199382018096261 get stuck there for many hours. We observe a similar pattern for another example:

Study link: 268430 (in red in the figure below)

Opposing traffic link: 7857997944793576039_7857997944592433397

again, while the opposing (and priority link) is at capacity, vehicles from the secondary link cannot enter and get stuck.

We are using Hermes at present and this does not (as we understand it) support the 3 types of link dynamics QSIM does - LinkDynamics {FIFO, PassingQ, SeepageQ}. However, we are unsure if this is even the dynamic in question, as the problem we are seeing is more related to junction priority?

In this case, the simulation is for a left hand drive country if that has any relevance.

Any observations would be very welcome