Title: Enhancing TDF Model Results Using Intersection Control Specific Delays and Turning Movement Level Ma
1Enhancing TDF Model Results Using Intersection
Control Specific Delays and Turning Movement
Level Matrix Estimation for a Downtown
Circulation Study
- Presented to the 11th Conference on
Transportation Planning Applications - May 9, 2007
- By
- Gregory Giaimo, Ohio Department of Transportation
2Motivation of Presentation
- Demonstrate Use of New Techniques to Improve TDF
Model Results for Project Analysis - Emphasis on a Careful and Methodical
Implementation versus Push Button
3Two Key Tools
- Explicit Intersection Delay Modeling
- Turn Movement Level Matrix Estimation
4Background on the Mansfield Traffic and Parking
Study
- Various one way street configurations downtown to
be analyzed - Need opening (2012) and design year volumes for
the AM and PM peak hours - Consultant has already set up and run a Synchro
analysis using the base year counted volumes and
the existing operational configuration - Task to create good forecast turn movement
volumes on these alternatives for further Synchro
operational analysis to produce a final
alternative
5Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
6Network Lane Checking
- Turn lane configurations in the model were
checked vs. Synchro. These were double checked
vs. aerial photos and video log. - Errors in network were corrected and full
validation model rerun. - Some errors in the Synchro data were found and
were sent to the MPO
7Check Count Consistency
- Turn counts were checked for consistency in a
spreadsheet - They were also aggregated to link level and
plotted on the network - Overall, the turn counts from adjacent
intersections were reasonably consistent
PM Peak Turn Volumes Aggregated to Links
8Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
9Updating Signal Operations
- Updated signal timings and progression based on
Synchro files - Moral assumed default signal operation
parameters dont work with operational level
studies
Over-ride signal locations
Signal phasing editor
10Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
11Rerun Base Year Model
Save Paths
12Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
13Extract Sub Area Net and Trip Table
- Not always necessary but helpful in this case
14Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
15Fratar Cordon Volumes
- 24 Hour Link car/truck counts at cordon converted
to period level directional values using full
model assignment - Original sub area trip table Fratared
16Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
17Check for Routing Problems
- Fratared trip tables assigned and assignments
checked for path problems - Some small speed adjustments corrected most
problems - Due to odd routing of SR 13 through CBD, SR 13
through trips were placed in their own volume
group to force them on the SR 13 signed route
(because a much shorter path down Main Street
exists but counts indicate the through traffic
largely remains on the signed route)
18Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
19Prepare Inputs for Matrix Estimation
- Car/Truck are compressed, turning movement counts
were by class but truck volume was too low to
bother with independently - Extract trip ends
- Add confidence levels to data
- Trip table cell 1
- Internal trip end 10
- External trip end 50
- Older turn count 50
- New turn count 75
20Run Matrix Estimation
- Can accept various data items such as
- Link or turn counts
- Trip ends
- Seed trip table
- Etc.
- All inputs require confidence levels
- Relative values
- Order of magnitude difference needed to see much
impact - Adjusts the trip matrix to fit the various
weighted (through confidence levels) observed
data with minimum error
21Run Matrix Estimation (cont.)
2 Matrices required, trips and confidence level
Need a path file from a previous assignment
ASCII, zone, origins, destinations, O confidence,
D confidence
The information contained in the print file is
extremely important and should be well understood
by anyone wishing to use matrix estimation
properly
ASCII, screen number, from-through-to nodes,
count, confidence level
22Run Matrix Estimation Print File
This shows that convergence occurred and the
number of iterations.
This shows input matrix characteristics, the
warning messages show counts that had no trips
assigned in the prior assignment. This is
important, a link or turn with no assignment
before wont get any in matrix estimation either.
This shows how good each zones trip ends match.
Heres information on average confidence levels
and amounts of data, very important for helping
set confidence levels.
This shows how well each count was matched,
notice that the counts match better than trip
ends due to higher confidence.
23Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
24Check Forecast Variables
- Variables checked in spreadsheet and through
plots - Noticed large discrepancy between 2000 and 2005
due to changes in assumptions
25Check Forecast Variables (cont.)
- In contrast the 2005 and 2030 variables are
consistent and show growth in reasonable areas - Since the forecast variables arent official it
was decided to use these to derive growth factors
for the base year
26Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
27Run Forecast Year Model
New Scenarios added for forecast land use
Base Year Model with Forecast Variables
28Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
29Fratar to Forecast Year
- Trip ends extracted from forecast year trip
tables - Fratar factors (not targets) calculated in
spreadsheet - Separate Origin and Destination factors (excerpt
below is from the Origin factor computation)
30Study Flow Chart
Code alternative networks
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Check variables, reformat and fix as necessary
Assign both periods for all three years (base,
opening, design) to all networks
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
31Code Alternative Networks
- Four alternatives
- Alt. 1 Two way operation on Main and Diamond
- Alt. 2 Two way operation on Main and Walnut
- Alt. 3 Two way operation on Diamond, Main,
Mulberry and Walnut - Alt. 4 As Alt. 3 plus two way operation on First
and Second - In addition to adding two way operation, lane
configurations and signal operations updated
appropriately
32Code Alternative Networks (cont.)
Alternative 1
Nobuild (Existing)
TURNLANE annotated, LANES colored
33Code Alternative Networks (cont.)
Alternative 3
Alternative 2
TURNLANE annotated, LANES colored
34Code Alternative Networks (cont.)
Alternative 4
TURNLANE annotated, LANES colored
35Study Flow Chart
Compress car/truck, get trip end file and
confidence levels
Turn counts and Synchro files from MPO
Forecast zonal variables from MPO
Code alternative networks
Assign both periods for all three years (base,
opening, design) to all networks
Check variables, reformat and fix as necessary
Check network count coding
Reassign, correct obvious path problems, reassign
Add detailed signal timing and progression
Create scenarios in full model for forecast years
Run matrix estimation to create final base year
demand volumes
Take DBF turn volumes to XLS and convert to
Synchro CSV format
Rerun base year model using revised network
Extract sub areas for each period for each
forecast year
Extract sub area network and trip table for AM
and PM periods
Calculate Fratar factors by zone to adjust base
year demand to future
Calculate Fratar targets at sub area cordon
Fratar base demand to future
36Assign Export to Synchro
- Save Turn Volumes to DBF
- Lookup tables of Synchro direction codes and
intersection IDs required in spreadsheet to
convert Cube From-Thru-To node
- A pivot table in spreadsheet then easily converts
turn movements to the tabular format required by
Synchro
37Impact of Adjustments on Volume
38Impact of Adjustments on Volume (cont.)
39Turn Volumes From Several Methods
- The obvious question After all this effort what
is the impact on the final results? - Different turning movement volumes are obtained
using the adjusted sub-area model vs. simply
coding the network changes/variables into the
original model as shown below - What impact do these differences have?
402000 Alternative 1, PM Peak Assignment,
Intersection of Diamond Park Ave West
Validated Regional Model
Adjusted Sub-area Model
236 Left 859 Through 353 Right
263 Left 389 Through 378 Right
41Turn Volumes From Several Methods (cont.)
- As expected the adjusted model matches counts
better - While the standard link based RMSE of this model
is 33, the CBD turning movements have a
substantially higher value of 73 for the
validation year model
42Turn Volumes From Several Methods (cont.)
- The regional model using junction based
assignment performs marginally better than the
regional model using equilibrium when compared to
counts
43Turn Volumes From Several Methods (cont.)
- However, more importantly, the sensitivity of
traffic to the coded alternative is much closer
to that obtained with the adjusted model
44Turn Volumes From Several Methods (cont.)
- Changes resulting from the forecast year are much
smaller than the model error (without
adjustments) due to the low growth in the area
45Turn Volumes From Several Methods (cont.)
- Changes resulting from the alternative are
similar magnitude to the model error without
adjustment - Thus, the adjustment process and junction
modeling is most likely needed to reasonably
predict turning movement level outputs in an area
highly regulated by traffic control devices such
as the CBD
46Comparison of Model LOS
How do these volume differences impact the
decision making process? Different Volumes, Same
Methods (Year 2000 No Build Shown)
Original Model
Final Sub Area Model
47- Questions
- Please Use Microphone