Ansers inline...

*JTR router*
>
> -       Does it make sense to define smaller interval sizes, but do not
> change the turning ratios in these intervals?
>
no.

> -       What’s the effect of “route-steps”? Does it only affect the
> result when different turning ratios are specified for different interval
> sizes?
>
no effect on output at all (it only affects memory usage while processing
input files) .

>
>
> *RouteSampler*
>
> When I apply the RouteSampler.py, I indeed get more uniformly distributed
> turning flows. As input, I take the routes previously computed by the JTR
> router. However, the overall desired accumulated counts are not correct. I
> confirmed this by applying the “generateTurnRatios.py”. For example, at a
> specific intersection (two unidirectional links crossing) the turning
> ratios should be 50% for each leg. In contrast, I got 47% and 53% after
> applying the routeSampler. I apply the “—optimize full” option, and the
> interval size does not seem to have a big impact.
>

Does routeSampler report on the mismatch (GEH > 0)? If so, the candidate
routes did not allow satisfying all constraints even using optimization.

> -       Does it make sense to reduce the counts for early intervals at
> intersections where it takes some time to load the network?
>
No. because routeSamper does not take into account the time it takes for
vehicles to reach the intersection. It assumes that flows are stationary in
each interval.
This may be a source of counting deviations looking at the simulation
counts in smallish intervals.

> -       Are there any other ways to reach the desired turning ratios?
>
First you need to figure out the source of the mismatch:
- constraint effects:routeSampler reports that it could not match all the
counts
- temporal effects: only simulate a single interval and sum up counts over
the whole simulation, if the counts fit then the effect is temporal


regards,
Jakob
_______________________________________________
sumo-user mailing list
[email protected]
To unsubscribe from this list, visit 
https://www.eclipse.org/mailman/listinfo/sumo-user

Reply via email to