Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] Calibration of mesoscopic simulation
At edge lengths of 10m you will see strong artefacts from segment discretization (1 vehicle fits into the segment, 2 vehicles (5m+2.5m) do not fit into a segment. I think this is pretty far away from what Eissfeldt had in mind.

- jam density is given by vehicle length and minGap unless (modified by segment discretization as noted)
- critical density cannot be calculated from the tau values. It's a free model parameter (https://sumo.dlr.de/wiki/Simulation/Meso#meso-jam-threshold)
- maximum flow is  1 / (tau_ff + vehLengthAndGap/edgeSpeed). This was changed from the Eissfeldt model to accomodate heterogeneous vehicle lengths.
- backward wave: See the long comments in MESegment::recomputeJamThreshold
- free flow speed depends on the speed distribution of the vehicles as well as density so its not straight forward to compute this analytically.

I'm not quite sure where the value of tau_jj = 1.4 in the sumo default comes from (this pre-dates my work on sumo)
If your calibration on the default segment length of 100m suggests better values, please let me know. For the usage of tau_jj in SUMO, refer again to the comments in recomputeJamThreshold. The model was changed here due to
https://github.com/eclipse/sumo/issues/2244

regards,
Jakob



Am Mo., 2. Sept. 2019 um 14:43 Uhr schrieb Sasan Amini <sasan.amini@xxxxxx>:

Dear SUMO developer team,

 

I am trying to calibrate a mesoscopic simulation for an urban arterial and get results as close as possible to the microscopic version of the same scenario. I get very good results when I use high values for taus (tau_ff=tau_fj=2, tau_jf=3.9 and tau_jj=3) and small meso-edgelength=10m. These values seems to be very extreme in comparison to the defaults ones and the suggested range, which makes me wonder if I have understood the meaning of each parameter correctly.

In the dissertation of Nils Eissfeldt tau is defined as service rate 1/q which can be obtained from the fundamental diagram. I think it is a bit different from its implementation in SUMO as it is discussed in https://github.com/eclipse/sumo/issues/5709 these values are net-time gaps between vehicles while leaving segments. This brings me to the question how the points on the fundamental diagram i.e. jam density, critical density, maximum flow, backward wave speed and free-flow speed can be calculated using the tau values?

If I understood it correctly, it seems that the parameter minGap of the car-following model affects the meso simulation as well which then sets the value of jam density =  1000/(vehicle_length+minGap). How is this then interlinked with tau_jj?

The other point is that in the dissertation on page 70 formula 3.24 there is a relationship between tau_jf and tau_jj defined indicating that tau_jj is bigger than tau_jf but in SUMO it is the other way around.

 

Thank you for you r support,

Sasan

 

_______________________________________________
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

Back to the top