Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] Modelling two bus lines of equal frequency

Thank you Mirko. I made a mistake while writing the mail (sorry!), here is the correct probability of insertion, corresponding to the graphs I sent:
<flow id="517" type="vehDist_IDM_bus" line="517" begin="0.00" route="Ruta_517" end="4500" probability="0.0155" />
<flow id="518" type="vehDist_IDM_bus" line="518" begin="0.00" route="Ruta_518" end="4500" probability="0.0155" />


With these probabilities, it's "expected" to get around 110 buses per hour.

I will try the solution provided, using --random-depart-offset <TIME>..

Thank you,

Kind regards

El mié, 27 jul 2022 a las 14:43, Mirko Barthauer (<m.barthauer@xxxxxxxxxxx>) escribió:

Although I don't get how you obtain several dozens of buses with twice the insertion probability of 0,0015. With very small flows, the relative variance can get quite large, so for some simulations, it might just be 10 vehicles in total and 30 in others, but never 110.

Anyway, see the docs for some suggestions how to procede. If you only have buses, maybe just insert them at regular intervals and add an offset --random-depart-offset <TIME>.

Best regards
Mirko

Am 27.07.2022 um 15:35 schrieb Jaime Moya:
Hi Mirko,

The scenario consists of a one lane road where only buses are being modelled. The flows vary between 10 b/h and 110 b/h.

The buses are inserted at the route begin and there is no spillback. On the following image, I plotted the cumulative number of buses for each line (each graph is a replication)
image.png
As you can see, more 517 buses are inserted in almost all cases. Could it be related to randomness? How can I balance the number of buses, without inserting them at regular intervals?

Thanks in advance,

Kind regards

--
Jaime Moya
Pontificia Universidad Católica de Chile


El mar, 26 jul 2022 a las 3:13, m.barthauer@xxxxxxxxxxx (<m.barthauer@xxxxxxxxxxx>) escribió:

Hi Jaime,

 

that's difficult to guess without knowing more about your scenario. Did you check whether the buses are inserted at the route begin at all? Maybe there is a spillback queue or many other vehicles want to enter the simulation at the same place, such that SUMO cannot place new vehicles on the road safely. If your network is too big to do this manually, think of placing a detector (with attribute vTypes="vehDist_IDM_bus") at the route begin or adding FCD output to your bus vehicle type to get relevant output for further diagnosis.

 

Mirko

 

 

 

 

 

-----Original-Nachricht-----

Betreff: [sumo-user] Modelling two bus lines of equal frequency

Datum: 2022-07-25T18:59:30+0200

Von: "Jaime Moya" <eljaimo@xxxxxxxxx>

An: "Sumo project User discussions" <sumo-user@xxxxxxxxxxx>

 

 

 

Hi,
 
I am modelling two bus lines of equal frequency, but different stop times at bus stops. To avoid buses being inserted at perfect regular times, I am using probability:
 
<flow id="517" type="vehDist_IDM_bus" line="517" begin="0.00" route="Ruta_517" end="4500" probability="0.0015" />
<flow id="518" type="vehDist_IDM_bus" line="518" begin="0.00" route="Ruta_518" end="4500" probability="0.0015" />
 
However, in many scenarios I am obtaining big differences in the number of simulated buses of one line with respect to the other (line 517 vs line 518). For example, the following graph shows the difference (number of 517 buses minus number of 518 buses) for a number of simulated scenarios:
image.png
 
These differences are important because I am getting results with noise. I don't want to use "vehsPerHour" or "number" because the buses would be equally spaced. Should I use period [exp(x)]? Doing so would balance the number of 517 and 518 simulated buses?
 
Thanks in advance.
 
Kind regards,
 
--
Jaime Moya
Pontificia Universidad Católica de Chile

Libre de virus. www.avast.com

 


_______________________________________________
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