Hello Peibo,

If you want to calibrate both speeds and volumes for a single location given the observed speed and volumes, you may have to make a few changes. In general, you may have to change

1) Driving Behavior Parameters: If you use Wiedemann 99, the first and most important parameter that affects capacity, and subsequently, the volume and speed is the headway time (CC1). Reducing the headway time will increase the capacity. If you use Wiedemann 74 try changing the additive and multiplicative parts of safety distance. You may also change the maximum lookahead distance. Reducing the maximum lookahead distance may increase capacity.

2) Lane change Parameters: the safety distance reduction factor affects both volumes and speeds. Reduction of safety distance reduction factor leads to more aggressive driving and increase in speed and volume.

3) Simulation Resolution: reducing the simulation resolution (e.g. to 1) may lead to reduction in speed (at the same volume).

You may take a look at my dissertation in which I have calibrated the speeds and volume at a single point for a freeway (https://digitalcommons.lsu.edu/gradschool_dissertations/4170/). If you keep not getting the speed and volumes that you desire you will have to use COM programming to automatically change different parameters to make your speed and volumes look similar to those measured in real-world. You may find this source pretty useful: Article Macroscopic and microscopic analyses of managed lanes on fre...

Don't change volume as they are observed values. Vehs(ALL) in data collection results list does not match the volumes the do following steps: 

> Change Driving behaviour: lateral gap problem.

> Change number of lanes: If there is higher volume of traffic on the link (Because virtual lanes increase, that means 3 lane act as 4 lane or in india it act as 5 lanes effectively)

> Link speed and behaviour: add behaviour to link and change the mode wise link speed. 

> Reduce speed Areas: incorporate reduce speed areas in such a manner that slow moving vehicles avoid 1& 2 lane and high speed vehicles avoid  2 or 3 lane (here I consider 3 lane road) Right now might be all modes are mixing

Regards

 Hi Tarun,

Thank you for your answering.

I did what you said above, but I still could not get the exact volume from data collection results.

> Change Driving behaviour: I adapted the parameters calibrated by other researchers for car-following and leave lane change as default.

> Change number of lanes: I am simulating a given study area, so the number of lane is as same as the study site.

> Link speed and behaviour: In link behavior, I add the link behavior for different vehicle class, but I cannot find the option of wise link speed. (the capture shows the settings for link behaviors). After I did this, the volume reduced dramatically, and still could not match the observed one.

> Reduce speed Areas: there is an off-ramp in my network, but it is located in the downstream of my data collection point. Is there any effects if I do not define the reduce speed area?

Regards

Dear Peibo,

why do you need to reproduce "exactly" the volume you observed? How big is the error? In general, I think you observed the average value of a random phenomenon. In addition, you should always get an error assuming also the ones related to the simulation models. Regards

Hi Michele,

Thank you for your answering.

Actually, I do not need to reproduce the "exact" value. I am using GEH to indicate the goodness-of-fit of volume, but I can not reproduce the volume with GEH less than 5.  I found there is one option in VISSIM to input the exact volume by time intervals, so I ried to match the exact values.

I am now thinking if there are some parameters that I can change to reproduce the reasonable volume and speed with goodness-of-fit.

Regards

Peibo,  just checking that you are using the GEH values with 1-hour volumes, not a smaller unit?  If so, you need to modify they GEH, but that is not recommended for units less than 1 hour.  I've seen it used for 2 hour volumes (modified GEH) , but the assumptions made by Geoff Havers will not hold for volumes less than one hour.

Dear Peibo,

As @Urie and Michele rightly mentioned, observed and simulated values will differ because of obvious reasons. As you also know, for the acceptable simulation model, GEH statistic should be less than 5. Firstly, If you are not satisfied with your results, a review of your modelling process might help you to find some weak links (if exist). Secondly, if you have less than 1-hour data, then you can compute equivalent hourly flow and thereby, GEH (this may result in less reliable results).

Talking about the parameters you can change (you might have taken care of all these):

CC0 (Standstill distance): (Desired distance

between lead and following vehicle at v = 0

km/h)

CC1 (Headway Time):(Desired time in

seconds between lead and the following vehicle)

Look Ahead Distance (Min)

Look Ahead Distance (Max)

Look Back Distance (Max)

Look Back Distance (Min)

Waiting time before diffusion

Minimum Headway (front/rear)

Safety distance reduction factor

Minimum lateral distance @0 km/h

Minimum lateral distance @50 km/h

Desired speed distribution

Try the results with "different random seeds" and take the average values (10 different seeds would be fine)

Below is a link to my paper where I have used both VISSIM for modelling and GEH statistic for validation of my BRTS model.

Regards,

Anshuman

https://www.jstage.jst.go.jp/article/easts/11/0/11_1575/_pdf

Hi Uire,

Thank you for your answering. The observed data I used is the hourly volume, and the simulated one I used is the equivalent hourly volume (60 times of the 1-minute volume). Is this valid for GEH test?

Regards

Hi Anshuman,

Thank you for your answering and your reference.

Hi Peibo,

From my experiences using VISSIM, I found out that Simulation Resolution also has an effect on a simulated result. So I would recommend you to try using 20 Sim Resolution. And let me know how the result came out. 

Regards,

Thodsapon

Iam a new user of vissim software ,can anybody send me the procedure of using it to evaluate drivers behaviour under rainfall and snow.i have traffic data e.g speeds,volume and density i used pneumatic tubes to collect them