Validate freeways against Caltrans data

[GoogleCodeExporter]

Freeway speed flow results from Dynqmeq to be validated against Caltrans data.  
Neema to do analysis/data comparison.  If you need extractions from Dynameq 
results, we'll get Michalis to do that.  

Original issue reported on code.google.com by greg.erh...@ucl.ac.uk on 12 Jun 2012 at 5:07

[GoogleCodeExporter]

Neema and I talked, and he has already calculated the speed/flow curves for 
several PeMS sensor locations and we compared it to the Dynameq parameters that 
we used as well as the resulting simulation "sensor time series".

He is going to calibrate the flow model for freeways and ramps based on this 
data and will update the dtaAnyway flow model in the python script.

We can test how these changes worked in the next run.

Original comment by eas...@gmail.com on 13 Jun 2012 at 12:34

[GoogleCodeExporter]

Goal is to have the generalized freeway speed/flow curves completed for a 
validation run that starts today (Tuesday).

Original comment by elizab...@sfcta.org on 19 Jun 2012 at 5:41

[GoogleCodeExporter]

The triangular speed/flow diagrams are derived for 59 lanes at 15 sensor 
locations in SF city. Results can be viewed in "Q:\Model 
Development\DTA\dtaAnyway\TrafficFlowModelCalibration\PeMS FreewayData". 
The summary file gives an overview of the trends. 

Out of 59 lanes, 26 do not show a triangular pattern (like: Q:\Model 
Development\DTA\dtaAnyway\TrafficFlowModelCalibration\PeMS 
FreewayData\Fundamental diagrams\VDS 401470\lane 1-Mainline VDS 401470 Mission 
St OC.xlsx) that are excluded  for further processing. 3 lanes show unusually 
large maximum flows (over 4000 per lane like : Q:\Model 
Development\DTA\dtaAnyway\TrafficFlowModelCalibration\PeMS 
FreewayData\Fundamental diagrams\VDS 401255\lane 1-Mainline VDS 401255 15th 
street - Vermont.xlsx), therefore, they are also excluded  for final 
processing. 

The remaining 30 lanes are kept in data set for calibration. The resulting 
averages for maximum flow, freeflow speed , and backward wave speed are 
respectively 2182 vphpl, 65.22 mph and 8.36 mph, where statistics T test 
implies that the means are reliable representatives of the freeways with 0.9 
confidence level and expected error of 160 vphpl for maximum flow, 1.84 mph for 
freeflow speed, and 0.92 for backward wave.

Based on that I updates the FFS for freeways and expressways (FT = 2,3) to 65 
for area type CBD , Urban Biz , and Urban. Since the effective lengths are kept 
to 20.49 (Dynameq project setting), in order to maintain 2182 maximum flow in 
these area types, the backward wave is chosen 9.75 instead of 8.36. Therefore, 
the RTF is updated to 1.15.

Original comment by neema.na...@sfcta.org on 20 Jun 2012 at 8:31

[GoogleCodeExporter]

http://code.google.com/p/dta/source/detail?r=9887f619d9f14b3cb7dd2997a5cece40a3e
1e731&name=dev

Original comment by neema.na...@sfcta.org on 20 Jun 2012 at 8:47

[GoogleCodeExporter]

Original comment by neema.na...@sfcta.org on 22 Aug 2012 at 5:20

• Changed state: Fixed