1) if you set your frequency to ~half the cycle time, you should see some data points with higher flow. When averaging over a full cycle you would still mix green and red phases 2) yes, each interval would give you one datapoint. For network-wide density you could use the total number of vehicles (sampledSeconds / freq) and divide it by the total network lane length (shown in the network property dialog in sumo-gui, accessible when clicking on the car icon in the bottom bar). 3) I haven't looked systematically at urban fundamental diagrams but I think your plot is plausible
Am Do., 11. Juni 2020 um 17:15 Uhr schrieb Amirhosein Karbasi < [email protected]>: > Thank you, dear Jakob. > 1.now if I use freq equal to traffic light cycle the problem will be > solved? > 2.you said "the flow data points you are using are for individual edges. > If you need a whole-network diagram you need to sum up the edges flows and > compute an average density." for flow I should only aggregate flows in > the intervals and should I compute average ? > 3. you said " in your scenario, the flow is mostly affected by the traffic > lights rather than the vehicle density. Thus it will look different from a > motorway diagram". is this normal in intersections? > > On Thu, Jun 11, 2020 at 4:55 PM Jakob Erdmann <[email protected]> > wrote: > >> 1) in your scenario, the flow is mostly affected by the traffic lights >> rather than the vehicle density. Thus it will look different from a >> motorway diagram >> 2) the flow data points you are using are for individual edges. If you >> need a whole-network diagram you need to sum up the edges flows and compute >> an average density. Also, the flow is averaged over 300s (more than one >> traffic light cycle). This means you are not seeing peak flow on green but >> the average flow including stopped-on-red >> 3) I recommend that you use 'entered + departed' for you flow >> computation. Also, there are probably edges that have low vehicle counts >> but downstream congestion prevents vehicles from advancing and thus gives >> zero flow. >> >> regards, >> Jakob >> >> Am Do., 11. Juni 2020 um 12:20 Uhr schrieb Amirhosein Karbasi < >> [email protected]>: >> >>> Hello, >>> >>> I want to draw a flow density diagram. My network includes 64 >>> intersections. I ran a simulation (1800s) and use an additional file for >>> Edge-based output (freq is 300). I drew a flow density diagram in excel >>> based on the Edge-based output file and I calculated flow according to “ >>> *entered* * 3600 / *freq*”. Now I have 3 problems: >>> >>> 1. I think my flow density diagram is not similar to regular flow >>> density diagrams. (there are a png in attach file)(Simulation time = 1800s). >>> >>> 2. When I drew a plot my max flow is about 852 (for 1800s, 2400s, and >>> full-time simulation max flow is approximately 1000) and I think max flow >>> should be bigger than 20000 for a network like this. >>> >>> 3. I have too many zero flows. >>> >>> I saw various topics about this on the Sumo website but my problems are >>> not solved yet. >>> >>> I sent my whole files too. please say me what I am doing wrong? >>> >>> Thank you so much. >>> _______________________________________________ >>> sumo-user mailing list >>> [email protected] >>> To unsubscribe from this list, visit >>> https://www.eclipse.org/mailman/listinfo/sumo-user >>> >> _______________________________________________ >> sumo-user mailing list >> [email protected] >> To unsubscribe from this list, visit >> https://www.eclipse.org/mailman/listinfo/sumo-user >> > _______________________________________________ > sumo-user mailing list > [email protected] > To unsubscribe from this list, visit > https://www.eclipse.org/mailman/listinfo/sumo-user >
_______________________________________________ sumo-user mailing list [email protected] To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
