TOPl review - arterial model

[Andy]

Please find the modified slides for Aurora - arterial model in
'TOPL_arterial_260908_2.ppt'.

Please note:

1. Slides 13 - 14, which is on the calculation of travel times in
Aurora,
could be skipped in the presentation;

2. Slide 18 (suggestions of future work) can also be skipped and I can
just mention them orally during the presentation;

3. I added the statistics (mean + SD of travel time estimates) to the
PDFs
on slide 16 (which I have presented to you last Thursday). The
statistics
shows that the differences between the Aurora estimates and the ones
obtained by Sensys's vehicle matching algorithm is within 10%;

4. Following the discussion last Thursday, I loaded all 250 vehs
(including matched and unmatched vehs) detected at Fairmount, instead
of
only those 45 matched vehs, and get the results on slide 17. The
result
shows a smoother distribution of travel time estimates (due to larger
sample size) but a higher mean. It is due to the higher traffic volume
loaded into Aurora since in the real case some traffic is dissipated
due
to turnings or lane-changings.

5. I tried to take the dissipated traffic into account: I estimated
the
amount of dissipated traffic based on the difference between the
numbers
of matched and unmatched vehs. However, doing so will make traffic at
the
entry and the exit of the arterial do not conserve and hence cause a
problem in calculating the arterial travel time (if I use the method
of
cumulative curves...).

6. I didn't include the delay plots in the slides as they show similar
pattern as the travel time plots...

That's for now. Please comment.

Andy