2. Innovative Performance Evaluation Matrix for Solving Urban Freeway and Interchange Congestion
Chengxin Dai, PE
Western ITE Conference
Performance measurement in transportation has different levels ranging from day-to-day operational management of highway systems to long-term capital planning that enhances system operations. This presentation focuses on project-level performance measures to identify design features that improve operation and safety and compare alternatives.

3. Study NEED
Multimodal microsimulation analysis to compare performance of alternatives
Develop useful performance measures in overcapacity conditions
As congestion levels increase over the years, traditional performance measures fail to reveal details of congested corridor performance. New performance measures will be introduced in this presentation. I will use the I-5 Broadway-Weidler Interchange Improvement project to introduce innovative performance measures and matrix to scrutinize alternatives.
Aim: With that said, the project needs are to conduct multimodal micro-simulation analysis to evaluate the performances of alternatives for the facility among one of the most congested freeway segments in Oregon, developed new performance measures for overcapacity conditions.

4. Study Area I-5 Crossroads of Portland freeway systems
150,000 AADT in four-lane cross-section at the pinch point
1,100 feet between each interchange
Critical for job and economic generators
I-5 Broadway/Weidler
Interchange
Moda
Center
The area of focus of this project includes 2.5-mile-long freeway I-5 and ramps, as well as surface street intersections around the Broadway-Weidler interchange. The project is at the crossroads of three Portland area freeway systems including I-5, I-84 and I-405. I-5 is the north-south freeway and extends through the Portland metropolitan area, in fact running the full length along the west coast of the U.S. For the entire run from Canada to Mexico, the project area is one of the most congested segments and has been frustrating commuters and freight haulers for years. That serves 150,000 AADT in four lanes at some pinch point. Traffic are mostly local through traffic and regional interstate traffic. Systems spacings are short. This stretch of I-5 is vital to major job generators and economic generators in Portland.

5. Study Area Arterial Network Broadway & Weidler Moda Center Area
Broadway St
Weidler St
Williams Ave
Vancouver Ave
Flint Ave
Moda
Center
Arterial Network
Broadway & Weidler
Moda Center Area
Multimodal facilities
Winning Way
Zoom into the intersection: the interchange serves a large portion of the city, as well as Moda Center Arena (home of Portland Trail Blazers), and industrial area of N Portland. The project network also includes multimodal facilities to support all modes of travel. Six transit lines run on Broadway and Weidler including buses and streetcar. Heavy bike volumes about nearly 500 bikes during peak hour.

6. Future Alternatives A: No-Build B: Auxiliary Lane C: Single Braid
1 vs. 2 lane SB I-84 flyover
B: Auxiliary Lane
1 vs. 2 lane SB I-84 flyover A1 A2 B1 B2
C: Single Braid
1 vs. 2 lane SB I-84 flyover
D: Double Braid
1 vs. 2 lane SB I-84 flyover
Multiple alternatives were developed from the project facility plan and interactive stakeholder workshops. The alternative improvements were designed to reduce weaving, more efficient access on and off the freeway and safer and smoother biking and walking conditions. A total of eight alternatives were developed and modeled in VISSIM. Each alternative modeled 10 peak analysis hours to account for peak spreading. C1 C2 D1 D2

7. Performance Measures for congested CONDITIONS
Traditional MOEs:
LOS, V/C, Average Travel Time, Brainscan Chart, Percent of Unmet Demand, etc.
Not particularly useful in heavy congestion
Proposed MOEs:
Travel Time Reliability
Lane by Lane Speed
Emergency Braking
Performance measures provide a means to evaluate facilities. For congested facility, traditional freeway MOE alone became less meaningful. Traditional MOE from Highway Capacity Software (HCS), such as density, volume-to-capacity (v/c) and level-of-service (LOS), provide a big picture view about performance; for congested conditions, however, they do not tell a good story of when and where congestion starts to build or when it dissipates, or detailed lane-level performance. These traditional measures do not tell the most important to us under the overcapacity condition. Average travel time is outdated since road users and decision makers need to understand how reliable a transportation system is. That cannot be communicated in simple averages. Brainscan chart is average speeds per segment. Percent of unmet demand only shows the overall facility performance.
With that said, successful operational performance measures for congested facility are related to each individual traveler’s traveling experience, such as travel time reliability, speed and delay. Successful safety predictions also tie to each traveler, such as emergency braking events.

8. New performance measures
Travel Time Reliability
The less variation from the median, the more reliable the travel time.
Use Vehicle Record data (second-by-second) sorted into northbound and southbound travel routes.
Travel time reliability is an increasing concern for practitioners and all road users, whether they are vehicle drivers, transit riders or freight shippers. Most travelers are less tolerant of unexpected delays. They value reliable systems to better plan their trips.
With the current state of traffic modeling practice, even the most prevalent traffic modeling software Vissim and SimTraffic do not directly generate travel time reliability result, they provide average travel time instead. To generate reliability results, Vissim vehicle record output were used. The vehicle record captures all the attributes for each vehicle per time step. We know, at a certain modeling timestamp, where each individual vehicle is located, vehicle type, their speeds, acceleration rate, as well as where they are traveling from and destined to. With that, travel time distribution of each travel route could be derived.
Barchart shows cumulative percentiles. We want to selecte a scenario that bars close together. Option A1 is the least reliable scenario; Option B2 has the most reliable scenario with the lowest deviation from the median.

9. New performance measuresB1
New performance measures
Lane-by-Lane Speed Chart
Compare speeds between scenarios
clearly illustrates speeds in each lane and demonstrates how much of speed improvement is made under the alternative.
During 4:00-5:00 PM in the No-Build Alternative, severe congestion and slow speed (less than 20 mph) occur across all SB lanes near the existing lane drop as far north as I-405. The slow right lane speeds extend north to the Greeley on-ramp. Slower speeds (20-40 mph) are between I-405 off-ramp and N Wheeler On-ramp.
Under the B1 Alternative, speed would improve on I-5 SB: at the Broadway/Weidler interchange, speeds in the left and middle lanes would exceed 50 mph; from the I-5 SB entry to the I-405 on-ramp, speeds increased by at least 10 mph. Weaving congestion would result in slow speed (20-40 mph) between the Greeley on-ramp and the N Broadway off-ramp. Slow speed (less than 20 mph) would occur in the right lane of the I-5 SB weaving segment between the NE Weidler on-ramp and the I-84 off-ramp, but speed in the middle lanes in this area would improve. Speed on I‑5 NB would increase substantially throughout the API, with most lanes approaching 50 mph or higher between I-84 and I-405.
Speeds on I‑5 SB would be considerably improved
Under the Build Alternative The I-5 NB would also substantially improve under the Build Alternative, with most lane speeds at or above 40 mph.

10. New performance measures
Emergency Braking Events
Emergency braking threshold ≥ 14.8 fps2
AASHTO Green Book (2011) emergency braking rate
Emergency Braking data obtained from Vissim Vehicle Records
Emergency Braking Events
vs. Crash Data
Heat Map Production
(Hot Spots)
Safety Performance Prediction
for Alternatives
Is a safety-performance measure developed by the project team. The AASHTO Green book notes that a majority of drivers decelerate at rates between 11.2 and 14.8 feet second squared when confronted by an unexpected object. It’s about driver’s reaction to an unexpected event. Take highway shock wave for example, when a driver crosses a lane quickly, the first car behind it brakes a little, the second car behind didn’t notice immediately and brake a little harder than necessary, and then the drivers behind them all do the same. Someone upstream approaching at highway speed stops as well). Several emergency braking instances could occur.
High resolution vehicle record data was applied for this measure. You may wonder whether EB is truly related to the real-world crashes, so I’ll talk about correlation analysis followed by two in-depth analyses.

11. Existing ODOT Crashes vs. Existing Calibrated VISSIM Results
Pearson’s correlation coefficient:
ram=0.56
rmd=0.95
rpm=0.73
To verify EB is positively and strongly correlated to crashes, the project team downloaded five-year historical crashes and compared with EB of the existing condition model. We noticed MD & PM have a strong and positive correlation. With that, we conclude a higher frequency of EB correlate with a greater likelihood of crashes. This strong correlation opens the door for our next two in-depth analyses.

12. Braking Density Heat Maps
Southbound
Visualize high risk sections
High concentration of emergency braking at:
Southbound Entry
Southbound between I-405 merge and Broadway off-ramp
Going Ramps
Broadway/
Weidler
I-405
I-84
The vehicle braking density map visualizes high risk spots. The project team geocoded data, then used point density raster analysis to count how many events occurred at each spatial point. A high risk segment is I-405 to the lane drop. One mitigation strategy could be installing variable speed sign to warn and slow down the I-405 merge traffic and reduce potential crashes. After the merge is the main pinch point bottleneck where speed and maneuverability are severely restricted and less emergency braking instances will occur.
Future No Build 5-6 p.m.

13. Emergency Braking event
Future PM Southbound Event Analysis
Adding two-lane flyover improves the safety.
North of Broadway: all the “build” scenarios reduce emergency braking events by more than 60%.
South of Broadway: only braid options reduce emergency braking events.
Overall significant reductions in emergency braking for all the build options
Next, the project team then compared the emergency braking event frequency across the alternatives. The No-Build alternative had a total of 80,000 emergency braking events. Option B1 had a roughly 60% reduction because the auxiliary lane increased the capacity and improved operations. The single and double braided ramp alternatives had fewer events as it reduces weaving conflicts.

14. Operation AND SAFETY PERFORMANCE MATRIX
Finally, a performance matrix was developed, including a series of seven measures. New measures complemented with the traditional measures together capture system-wide and individual traveling components. The team quantified them in Likert scale for easier comparison. For instance, under emergency braking, all the two-lane flyover alternatives perform better. Overall, B2 is projected to perform the best as it maximizes the system throughput, minimizes travel time and delay, and potential crashes.

15. summary
Select appropriate performance measures for comparative purposes
Travel time reliability & emergency braking analysis reveal details of congested facility performance.
Combination of new and traditional MOEs for congested facility evaluation
In summary, MOE may vary project by project. select the appropriate MOEs for your comparative studies.

16. Questions?
Contact Information:
Chengxin Dai Traffic Engineer HDR
Jeremy Jackson, PE (co-author) Senior Traffic Engineer HDR