Roundabout Modeling in CORSIM. Aaron Elias August 18 th, 2009

Transcription

1 Roundabout Modeling in CORSIM Aaron Elias August 18 th, 2009

2 Objective To determine the best method of roundabout implementation in CORSIM and make recommendations for its improvement based on comparisons with field data.

3 Tasks 1. Field Data Collection Acquire video of two roundabout sites Extract operational parameters 2. CORSIM Simulation Network Implementation Origin-Destination Coding 3. Results Comparison Control Delay Maximum Queue 4. Conclusions and Recommendations

4 Field Data Collection

5 Site 1 Port Orchard, WA

6 Site 2 Lothian, MD

7 Data Extraction Overview Two hours of data were extracted from each of the two roundabout sites Extracted data can be grouped into two categories: Data needed for CORSIM simulation: Approach volumes Origin-Destinations Geometric information Speed data Data needed for comparison s to CORSIM output: Approach delay Maximum queue

8 Camera Coverage Available for Data Extraction

9 Site 1 Entry and Conflicting Flow Northern Approach Eastern Approach Southern Approach Time Interval Entering Flow Conflicting Flow Entering Flow Conflicting Flow Entering Flow Conflicting Flow Period Period Period Period Period Period Period Period Intersection Volume: 1855 veh. (1 st hour), 1722 veh. (2 nd hour)

10 Site 1 Control Delay and Queue Northern Approach Time Interval Period 1 Period 2 Period 3 Period 4 Average Delay (s) 00: : : :09.4 Maximum Delay (s) 00: : : :34.0 Average Queue (veh) Maximum Queue (veh) Eastern Approach Time Interval Period 1 Period 2 Period 3 Period 4 Average Delay (s) 00: : : :06.6 Maximum Delay (s) 00: : : :20.8 Average Queue (veh) Maximum Queue (veh)

11 Site 2 Entry and Conflicting Flow Northern Approach Eastern Approach Southern Approach Western Approach Time Interval Entering Flow Conflicting Flow Entering Flow Conflicting Flow Entering Flow Conflicting Flow Entering Flow Conflicting Flow Period Period Period Period Period Period Period Period Intersection Volume: 1300 veh. (1 st hour), 1499 veh. (2 nd hour)

12 Site 2 Control Delay and Queue Northern Approach Time Interval Period 1 Period 2 Period 3 Period 4 Average Delay (s) 00: : : :06.3 Maximum Delay (s) 00: : : :28.4 Average Queue (veh) Maximum Queue (veh) Eastern Approach Time Interval Period 1 Period 2 Period 3 Period 4 Average Delay (s) 00: : : :04.5 Maximum Delay (s) 00: : : :19.7 Average Queue (veh) Maximum Queue (veh)

13 CORSIM Simulation

14 Vehicle Entry Headway An Erlang 1 distribution was determined to be the best fit for these two roundabouts. Lothian Arrival Headway Distributions Probability Arrival Headway (s) Actual Distribution Erlang 1 Distribution Port Orchard Arrival Headway Distributions Probability Arrival Headway (s) Actual Distribution Erlang 1 Distribution

15 Network Implementation

16 Internal Roundabout Links

17 External Roundabout Approach Turn Movement Coding

18 Internal Roundabout Link Turn Movement Coding

19 Conditional Turn Movements

20 Origin-Destination Replication Time Period Link CORSIM Field Data Percent Diff. 0.90% 1.20% 1.70% 0.90% 2.50% 0.20% CORSIM Field Data Percent Diff. 2.90% 0.30% 3.80% 1.80% 1.40% 2.80% CORSIM Field Data Percent Diff. 2.40% 0.20% 0.80% 1.00% 11.10% 3.60% CORSIM Field Data Percent Diff. 3.50% 1.30% 1.40% 1.50% Example from Site % 2.40%

21 Modification to Gap Acceptance Driver Type Default Gap Acceptance Roundabout Gap Acceptance

22 Completed Network Animation

23 Results Comparison

24 Site 1 Control Delay Comparison Time Period Control Delay (s/veh) Eastern Approach Northern Approach CORSIM Actual Percent Diff. CORSIM Actual Percent Diff % % % % % % % % % % % % % % % %

25 Site 1 Maximum Queue Comparison Time Period Cumulative Maximum Queues (veh) Northern Approach Southern Approach CORSIM Actual Percent Diff. CORSIM Actual Percent Diff % % % % % % % % % % % % % % % %

26 Site 2 Control Delay Comparison Control Delay (s/veh) Time Period Southern Approach CORSIM Actual Percent Diff. Eastern Approach CORSIM Actual Percent Diff % % % % % % % % % % % % % % % %

27 Site 2 Maximum Queue Comparison Cumulative Maximum Queues (veh) Time Period Southern Approach CORSIM Actual Percent Diff. Eastern Approach CORSIM Actual Percent Diff % % % % % % % % % % % % % % % %

28 Control Delay Averages Across all Time Periods Approach Avg. CORSIM Value Avg. Field Value Percent Diff. Eastern % Northern % Southern % Site 1 Approach Avg. CORSIM Value Avg. Field Value Percent Diff. Southern % Eastern % Northern % Site 2

29 Summary Control delay has a wide range of error between the different periods. However, when control delay is averaged over the whole simulation, the field data and CORSIM s values are within 1 second. Maximum queue also has a wide range of error between the different periods. This is a result of when a maximum queue may occur and differences in the definition of a queued vehicle.

30 Conclusions

31 Overview CORSIM does reasonably well when predicting control delay over an average of multiple time periods. However, on a period-by-period basis it has wide ranging variability. In addition to determining the best way to model a roundabout in the current CORSIM version, this research has also made recommendations to CORSIM s developer for future releases.

32 Recommended CORSIM Changes: User Friendliness and Animation Ability to group nodes as a roundabout. Once grouped, circulating speed is automatically calculated based on radius and superelevation. Conditional turn movements are automatically input and internal links are coded for counterclockwise curvature. Incorporation of splitter islands into the animation.

33 Recommended CORSIM Changes: Traffic Simulation Ability to define the definition of a queued vehicle. Approach and time specific gap acceptance at roundabout approaches. Approach and time specific follow-up time at roundabout approaches.

34 Limitations of this Research Implementation of a roundabout as part of a bigger network was not modeled. CORSIM s ability to model two-lane roundabouts was not investigated.

35 Questions?