Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] Flow definition disrepancy
Hello,
the main reason for the discrepancy comes from the randomness of the jtrrouter algorithm. You will get different counts when running with different seeds.
Here are two alternate approaches to get exact numbers:
1) define the flows directly. (define 12 flows with attribute number instead of 4 flows and and 12 probabilities)
2) use routeSampler with turn counts as input and set option --optimize full
The first approach is fine for a single intersection whereas the second will also work for complex networks.

When looking at your scenario, I noticed that it also suffers from issue https://github.com/eclipse/sumo/issues/6068
This leaves North->East vehicles stranded on the intersection and partially blocking the stream from the East.
I recommend upgrading to https://github.com/eclipse/sumo/suites/512680527/artifacts/2764645

regards,
Jakob


Am Mo., 9. März 2020 um 18:17 Uhr schrieb Chedly Bourguiba <chedly@xxxxxxxxxxx>:
Hi Jakob,

Thank you for your answer! Attached are the files used for the simulations. The faulty detectors are 614623 and 614624 going WB thru. According to the flow and turns files they should be 259 * 0.335 = 87  vehicles during the half an hour period but I got around 94 which is around 10% higher than it should be.
Thank you again for your help!

Best,

Chedly 

On Sat, Mar 7, 2020 at 1:56 PM Jakob Erdmann <namdre.sumo@xxxxxxxxx> wrote:
You could model this with a closingReroute (for the jammed section) and set the rerouter probability to the percentage of drivers that would try to avoid a congested road.

Am Sa., 7. März 2020 um 09:34 Uhr schrieb ing. Massimiliano Petri <m.petri@xxxxxxxxxxxx>:
Dear all,
does anyone know how it is possible to simulate the impact of variables
message signs giving informations about road congestions in following
streets?
Thanks,
Massimiliano

---
-------------------------------------------------
Eng.Massimiliano Petri

mobile:+39.3929169328
e-mail: m.petri@xxxxxxxxxxxx

University of Pisa
University Center 'Logistic Systems'
Via dei Pensieri, 60
57128-Livorno

Il 07-03-2020 00:10 Chedly Bourguiba ha scritto:
> Hi Jakob,
>
> Thank you for your quick response.
>
> We are currently defining the flows for one single time period, for
> one single intersection. We have specified the number of vehicles
> entering an approach (259). The approach has 5 lanes (2 for right
> turns, 2 for thru movements and 1 for the left turn). We have also
> defined the turning probabilities for left, thru and right.
>
> When we run the simulation, however, the output numbers are
> significantly higher for the thru movement lanes than expected (based
> on the number of total vehicles at the approach multiplied by the
> probability for the thru movement). In addition, the right turn
> movements and the left turn movements are close to what they should
> be. Consequently, the total number of vehicles at that approach is
> higher than the number we expect from the flow file. We're struggling
> to understand how this behaviour could be possible. We've checked that
> the flow number in the flow file and the turning probabilities in the
> turns file are correct.
>
> Do you know any possible reasons why this should be happening? Let me
> know if you need any further information to diagnose.
>
> Thanks,
> Chedly
>
> On Thu, Mar 5, 2020 at 10:52 PM Jakob Erdmann <namdre.sumo@xxxxxxxxx>
> wrote:
>
>> Hello,
>>
>> The traffic demand given to SUMO typically does not specify the
>> exact lane sequence. This means individual lanes could see higher
>> flows and other lanes could see lower flow due to lane choice.
>> I would expect the simulated edge flow to be no higher than the
>> defined flow.
>> An exception might happen if vehicles cannot be inserted for some
>> time and then are released into the simulation at a later time,
>> thereby increasing the flow beyond the specified value.
>> Things to check for:
>> - edge flow vs expected flow at different time aggregations
>> - depart delay (trip info), delayed departures per lane (gui
>> coloring style)
>> - departLane definition for your flows
>>
>> regards,
>> Jakob
>>
>> Am Fr., 6. März 2020 um 01:52 Uhr schrieb Chedly Bourguiba
>> <chedly@xxxxxxxxxxx>:
>>
>>> Hi,
>>>
>>> I have an intersection where I know the exact vehicle count per
>>> lane and movement per hour (a specific hour). When running sumo, I
>>> define the flow XML file using the vehicle per hour attribute and
>>> then I compare the count I had for the specific hour that I
>>> modeled and some lanes would have a count higher than the value I
>>> used to model the flow.
>>> Do you have any leads/idea where the discrepancy can come from?
>>> Thank you in advance for your help!
>>>
>>> Best,
>>>
>>> Chedly
>>>
>>> _______________________________________________
>>> sumo-user mailing list
>>> sumo-user@xxxxxxxxxxx
>>> To change your delivery options, retrieve your password, or
>>> unsubscribe from this list, visit
>>> https://www.eclipse.org/mailman/listinfo/sumo-user
>> _______________________________________________
>> sumo-user mailing list
>> sumo-user@xxxxxxxxxxx
>> To change your delivery options, retrieve your password, or
>> unsubscribe from this list, visit
>> https://www.eclipse.org/mailman/listinfo/sumo-user
> _______________________________________________
> sumo-user mailing list
> sumo-user@xxxxxxxxxxx
> To unsubscribe from this list, visit
> https://www.eclipse.org/mailman/listinfo/sumo-user
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

Back to the top