Since the 1960s, people have built models of traffic volumes based on cordon surveys, traffic counts, etc. that combined origin and destination data with road characteristics and traffic flow. These are used not only for road pricing but to prioritize road-building, upgrade, and renovation projects. I have not followed the literature for some time but I suspect these models now are using tracked cell-phone data to detect origin, destination, and route choice, much as Google uses that method to show you traffic on alternate routes. Indeed, Google (and Mapquest) clearly have the ability to spit out both fastest and shortest expected paths between origin and destination, as well as shortest path in current traffic. 

Dear Saad Touhbi,

refer these papers, it may useful for your research work.

http://www.ctr.kth.se/publications/ctr1998_08.pdf

http://www.sciencedirect.com/science/article/pii/0038012167900122

http://www.rms.nsw.gov.au/documents/projects/guide-to-generating-traffic-developments.pdf

http://www.dot.ca.gov/newtech/researchreports/reports/2009/final_summary_report-calif._infill_trip-generation_rates_study_july_2009.pdf

Since arrival of vehicles is a random process, it can be represented by generating random numbers in the simulation process. By knowing the statistical distribution which fits well with the field observed headway of vehicles, vehicle arrivals can be generated theoretically by generating random numbers as per the chosen statistical distribution. Once a vehicle is generated as per the applicable theoretical distribution, the type of vehicle is identified based on the percentage composition of traffic.

Please refer the below link for further details

http://ascelibrary.org/doi/abs/10.1061/(ASCE)0733-947X(2005)131%3A7(544)

Thank you all for the suggestions.

Kolita Weerasekera I've actually see some of your work and it helped on my analysis of time headway on a specific road type. However I did not understand the part where you concluded that there is no lane dependance in your use case.