Core of Rosslyn Transportation Study Existing Conditions Report - Appendix. Appendix E: VISSIM Model Calibration Summary

Transcription

1 Core of Rosslyn Transportation Study Existing Conditions Report - Appendix Appendix E: Model Calibration Summary

2 Core of Rosslyn Transportation Study Model Calibration Summary Model Calibration Summary Introduction and Purpose The following document details the procedures and assumptions used to develop and calibrate the AM and PM Existing Conditions models for the Rosslyn Street Reconfiguration Implementation Study. The calibration process follows the guidance from the Virginia Department of Transportation (VDOT) Traffic Operations and Safety Analysis Manual (TOSAM) Version 1. (Version 9, Build 1) was used to develop the AM and PM existing conditions models of the Rosslyn transportation network. Vehicle, transit, pedestrian, and bicycle modes were included in the models. The models were calibrated to real world conditions, as documented or observed using a variety of field data sources. A separate data collection summary document has been prepared and contains information on the existing conditions data used to calibrate the models. The objective of calibration was to develop models that reasonably reflect existing multimodal traffic operations in the Rosslyn study area. A calibrated set of models will be the basis for evaluation future roadway and multimodal improvements. Characteristics of the Study Area The study area modeled in consists of the N. Lynn Fort Myer Drive corridors, plus one adjacent intersection on each side of the corridor. Freeway mainline segments and ramps that provide access to the arterial network are also included (I-66, Arlington Boulevard, and Route 11). Figure 1 shows the study area. Page 1

3 Core of Rosslyn Transportation Study Model Calibration Summary Figure 1. Study Area A four-hour simulation period was selected to capture the onset and dissipation of study area congestion. This consists of a one-hour seeding period, a two-hour peak period, and a one-hour shoulder period. The AM and PM peak hours span the middle hour of the two-hour peak period. Table 1 summarizes the simulation periods and Figure 2 shows the total network intersection volume profiles that were used to determine the simulation periods. Table 1. Simulation Periods Seeding Period Peak Hour Peak Period Shoulder Period AM 6:3 to 7:3 AM 8: to 9: AM 7:3 to 9:3 AM 9:3 to 1:3 AM PM 3:3 to 4:3 PM 5: to 6: PM 4:3 to 6:3 PM 6:3 to 7:3 PM Page 2

4 Core of Rosslyn Transportation Study Model Calibration Summary , AM SIMULATION PERIOD PEAK PERIOD PM SIMULATION PERIOD PEAK PERIOD TOTAL INTERSECTION VOLUME 1, 8, 6, 4, 2, SEEDING HOUR PEAK HOUR SHOULDER HOUR SEEDING HOUR PEAK HOUR SHOULDER HOUR Figure 2. Simulation Periods and Total Intersection Volume Profiles AM Peak Period Conditions The AM peak period operates under congested conditions and is characterized by: Queuing along N. Lynn Street from Key Bridge back to Arlington Boulevard; congestion on Key Bridge also creates queues onto westbound I-66 (Exit 73) off-ramp Queuing and slow-moving platoons along eastbound Lee Highway, from the intersection with N. Lynn Street, spilling back to N. Rhodes Street Significant traffic flow, vehicle interactions, and queuing from multiple approaches at the northern and southern termini of N. Lynn Fort Myer Drive corridors Queuing and slow-moving platoons along eastbound Clarendon Boulevard into the study area, spilling back to N. Rhodes Street Queuing on eastbound Key Boulevard and eastbound Arlington Boulevard (frontage road) due to preponderance of cut-through traffic Queuing along eastbound Arlington Boulevard back to the 1 th Street interchange from the Theodore Roosevelt Bridge, causing spillback onto N. Meade Street from the on-ramp Queuing and slow-moving platoons along eastbound I-66 from eastern end of Theodore Roosevelt Bridge Significant transit activity along N. Moore 19 th Street N. Significant pedestrian activity throughout the study area, delaying right-turns at intersections and garage entrances PM Peak Period Conditions The PM peak period operates under congested conditions and is characterized by: Page 3

5 Core of Rosslyn Transportation Study Model Calibration Summary Queuing along N. Lynn Street from Key Bridge back to 19th Street N; congestion on Key Bridge also creates queues onto westbound I-66 (Exit 73) off-ramp Queuing along eastbound I-66 and Lee Highway from the weave between the on-ramp from eastbound Lee Highway and the off-ramp to Route 11 Queuing along westbound Wilson Boulevard in the study area Reduced speeds and weaving issues along Fort Myer Drive approaching Arlington Boulevard Significant pedestrian activity throughout the study area Network Development Roadway Geometry The model files created as part of VDOT s Transform 66 Inside the Beltway project were used as base files for the development of the Rosslyn model. The network roadway geometry was updated from the VDOT models, based on aerial imagery and verified with field observations, to better represent the Rosslyn study area network. The updated geometry and network represents roadway configuration as of November 217. As such, the following tactical urbanism or police enforcement strategies are included in the existing conditions models: Restriping of southern leg of Fort Myer Drive and 19 th Street N. Restriping of southern leg of Fort Myer Drive and eastbound Lee Highway Restriping and bicycle lane on N. Lynn Street between 19 th Street N. and eastbound Lee Highway Prohibition of westbound right-turns from Wilson Boulevard to N. Lynn Street Prohibition of southbound left-turns from N. Moore Street to Wilson Boulevard AM and PM model geometry is consistent, with the exception of the reversible lane operation on Theodore Roosevelt Bridge which was uniquely coded for each peak period. The bridge has four eastbound lanes and three westbound lanes in the AM, while in the PM, the bridge has three eastbound lanes and four westbound lanes. Speed Distributions Desired speed distributions, or model free-flow speeds, were set based on posted speed limits and the following guidance provided by TOSAM: From 5 miles per hour (mph) below the posted speed limit to 5 mph above the posted speed limit for arterials From 3 mph below the posted speed limit to 1 mph above the posted speed limit for freeways From 7.5 mph to 15.5 mph for right turns and from 12.4 mph to 18.6 mph for left turns Ramp speed distributions are referenced from posted advisory speeds. Traffic Control Devices A total of 19 signalized intersections were modeled using ring barrier controllers (RBCs). Signal timings were based on Synchro models and timing sheets provided by Arlington County. Page 4

6 Core of Rosslyn Transportation Study Model Calibration Summary Vehicle Routing Vehicular routing was coded as complete origin-destination vehicle routing decisions between network entry and exit locations. The process in Figure 3 was used to develop vehicle routing based on peak period trip tables that were validated as part of travel demand modeling. Rosslyn Subarea Peak Period Trip Tables VISUM Rosslyn Subarea Peak Hour Trip Tables VISUM Area Subnetwork Generation VISUM Import Trip Tables Ratio of peak period to peak hour network traffic counts used as scalar.444 for AM.288 for PM O-D Matrix correction using VISUM TFlowFuzzy to match assignment to peak hour turning movement and link counts Roadway links contained in the network areas extracted as VISUM subnetwork Subnetwork O-D matrix correction VISUM trip tables imported into to create vehicle routing decisions Consistency check between and VISUM Figure 3. Vehicle Routing Development Vehicle Compositions Three vehicle classes were used in the models: cars, small trucks, and heavy trucks. Uniform vehicle compositions were set network-wide based on vehicle classification data provided by Arlington County for locations near the study area. Table 2 summarizes the vehicle compositions used in the AM and PM models. The heavy vehicle compositions were set to zero for network inputs on I-66, George Washington Memorial Parkway, and parking garages because of truck prohibitions. In addition, buses were excluded from the vehicle compositions so that bus operations could be modeled explicitly by route and schedule as discussed in a later section. Table 2. Vehicle Compositions Vehicle Class Composition Example Vehicle Types Car 92.5% Passenger Cars, Pickup Trucks, Vans Small Truck 7.% Single Unit Trucks Heavy Truck.5% Single Trailer Trucks Vehicle Input Distribution Vehicle inputs were coded in 15-minute intervals to reflect variations in traffic volumes across the peak periods. Localized input distributions were created for each network entry point using raw traffic counts that were adjusted to balanced peak hour volumes. The distribution factors represent the flow rate for a 15-minute interval in relation to the volume for the network peak hour at that location. The localized vehicle input distributions allow peaking to occur at different locations at different times as observed in the field while still inputting the proper volume for the network peak hour (i.e., the basis for volume calibration). Page 5

7 Core of Rosslyn Transportation Study Model Calibration Summary Public Transit Bus routes and stops in the network area were coded using posted maps and schedules. Routes included in the models are shown in Table 3. Bus stop dwell times were set using a normal distribution with a mean of 2 seconds and standard deviation of 1 seconds. Table 3. Bus Routes Included in Models Operator Bus Route Arlington Transit 43, 45, 55, 61A, 61B Metrobus 3Y, 4A, 4B, 5A, 1E, 38B Dulles North Pentagon, Loudoun County Dulles North State Dept., Transit Purcellville Pentagon, Purcellville State Dept. Potomac and Dale City - Pentagon/Rosslyn/Ballston, Rappahannock Gainesville-Washington, Transportation Gainesville-Pentagon, Commission Manassas-Pentagon, (PRTC) Manassas-Washington Fairfax Connector 599, 699 DC Circulator Dupont Circle - Georgetown - Rosslyn Pedestrian and Bicycle Modes Pedestrian and bicycle demand was coded based on count data. Crosswalks were modeled at every intersection, including midblock crossings on N. Lynn N. Moore Street between Wilson Boulevard and 19 th Street N. Crosswalks were also modeled at parking garage entrances and exits. Bicycle facilities were modeled along N. Lynn Street, Clarendon Boulevard, Wilson Boulevard, and Key Boulevard (on-street bicycle lanes) as well as north of westbound Lee Highway (Custis Trail). Bicycle routing decisions were used for bicycle facilities, and bicycles were assumed to use crosswalks at other locations, per the recorded count data. Seasonal factors developed from bicycle count station data was used to normalize counts collected from different months of the year. Conflict areas and priority rules were coded for turning vehicles that conflict with non-motorized modes. Model Calibration Methodology Calibration Criteria The models were calibrated to existing traffic conditions generally using guidance provided in the TOSAM Version 1. Volume, travel time and queue length calibration measures deviate slightly from TOSAM per the project scope and agreed-upon Forecasting and Traffic Operations Analysis Framework Document. Table 4 summarizes the measures and criteria used to evaluate the effectiveness of model calibration as well as the specific data sources used to calibrate the models. Page 6

8 Core of Rosslyn Transportation Study Model Calibration Summary Table 4. Calibration Items and Targets Simulated Measure Calibration Threshold Field Data Source Simulated Traffic Volume for Peak Hour At intersections, difference targets must be met for at least 85% of approaches For freeways, difference targets must be met for at least 85% of freeway mainline segments and ramps Simulated Travel Time for Peak Period Difference targets must be met for a minimum of 85% of travel time routes. Four arterial routes and three freeway routes were used for calibration. Maximum Simulated Length for Peak Period Calibration target must be met for a minimum of 85% of the critical locations. impact were used to justify calibration in addition to quantitative comparison of queue length. Visual Review of Bottleneck Locations for Simulation Period Simulated average speed (miles per hour) were compared with probe vehicle data for the freeway segments (I-66, Route 11, and Arlington Boulevard) Within ± 2% for <1 vph Within ± 15% for 1 vph to <3 vph Within ± 1% for 3 vph to <1, vph Within ± 5% for 1, vph Within ± 1 minute for routes with observed travel times that are less than 7 minutes Within ± 15% for routes with observed travel times that are greater than 7 minutes Modeled queues qualitatively reflect the impacts of observed queues in the following area: Spillback to adjacent intersections Spillback from ramp intersection to freeway mainline and vice versa Spillback from turn lanes Speed heat map were plotted for freeway segments to qualitatively review the patterns and duration of congestions Balanced peak hour traffic counts (June 214 November 217) Primary 1 : Google Maps API (November 28 December 14, 217) Secondary 1 : Travel Time Runs (November 15 16, 217) Field observations Wednesday November 29, 217 Field observations Wednesday November 29, 217 INRIX October Google Maps API travel time was used for calibration of all routes that are detailed in the data collection summary of existing report except for westbound Wilson Boulevard. Travel time run data is used for this route because the Google Maps route extends beyond the extents of the model. Travel time data were collected from Google API in 5-minute intervals over the defined period and were aggregated to the peak periods. Error Checking Prior to calibration, error checking was completed through a visual review of the simulation and comprehensive review of the error log to identify coding errors that could impact the model results. This Page 7

9 Core of Rosslyn Transportation Study Model Calibration Summary included identifying missing conflict area and locations where signal timing were not be operating correctly. Given the highly-saturated traffic conditions, numerous closely spaced intersections, and heavy pedestrian activity, the simulation and coding was reviewed in detail to prevent unrealistic network gridlock. Number of Runs Requirement Microsimulation results inherently vary between runs due to random seed numbers, so results are an average of multiple microsimulation runs. The VDOT Sample Size Determination Tool was used to determine the appropriate number of simulation runs at a 95 th percentile confidence level. Initial model runs with ten seeds were conducted, and the tool was used with the following measures of effectiveness to determine the required number of runs for the AM and PM models: Travel time on northbound N. Lynn Street, southbound Fort Myer Drive, eastbound Clarendon Boulevard, and westbound Wilson Boulevard Throughput on northbound N. Lynn southbound Fort Myer Drive at 19 th Street N. Throughput on eastbound and westbound I-66 between Lee Highway and Route 11 The resulting required number of runs was ten for both AM and PM peak periods. As such, all results reported are an average of ten microsimulation runs. Calibration Parameters and Adjustments Calibration of the models to meet the calibration targets required adjusting specific parameters to achieve the desired throughput, travel time, queue lengths, and speeds. Lane Change Distance Lane change distance is the distance in the model where a vehicle will start attempting to make a lane change to a target lane prior to an off-ramp, a lane drop, or change in direction in travel. This lane change distance is a parameter on every connector in the network, and its default change distance value is 656 feet. This parameter was adjusted on a case-by-case basis at different locations with the goal of replicating observed field conditions. It was increased on freeway exits and oversaturated arterial movements to replicate the advanced lane changing that was observed in the field. In some cases, lane change distance was reduced from the default value to achieve better lane utilization at locations with closely spaced intersections or on-ramp lane drops, such as on 19 th Street N., Wilson Boulevard, and the eastbound I-66 on-ramp from Lee Highway. Driver Behavior incorporates two different car-following models one for freeways and one for arterials. Aside from roadway geometrics, driver behavior has the greatest impact on roadway capacity by adjusting vehicle spacing, headways, and lane changing behavior. Car-following and lane-changing parameters were modified for selected locations during the calibration process based on recommended values from guidance documents, previous experience with similar types of networks and operations, engineering judgment, and field observations. The models contain link display colors that correspond to the various link behaviors to assist with review and calibration. Table 5 provides a list of driver behaviors used in the models, associated parameters some of which (in bold) were modified from the default values, and example application areas. Page 8

10 Core of Rosslyn Transportation Study Model Calibration Summary Table 5. Driver Behaviors Car Following Parameters Lane-Changing Parameters Parameters Freeway (Default) Freeway Medium Aggressive Freeway High Aggressive Freeway Reduced Capacity Arterial (Default) Arterial Aggressive Arterial Aggressive High Capacity Arterial Aggressive Very High Capacity Look Ahead Distance: Num. of Observed Vehicles CC (Standstill Distance) (ft) N/A N/A N/A N/A N/A CC1 (Headway Time) (s) N/A N/A N/A N/A N/A CC2 (Following Variation) (ft) N/A N/A N/A N/A N/A Arterial Reduced Capacity Average Standstill Distance (ft) N/A N/A N/A N/A Additive Part of Safety Distance N/A N/A N/A N/A Multiplicative Part of Safety Distance N/A N/A N/A N/A Maximum Deceleration (Own Vehicle) (ft/s2) Maximum Deceleration (Trailing Vehicle) (ft/s 2 ) Accepted Deceleration (Own Vehicle) (ft/s 2 ) Accepted Deceleration (Trailing Vehicle) (ft/s 2 ) Minimum Headway Safety Dist. Reduction Factor Maximum Deceleration for Cooperative Braking (ft/s 2 ) Advanced Merging Off On On Off Off On On On Off Cooperative Lane Change Off On On On Off On On On Off Use Cases Weave or Oversaturated Short arterial Segments with Segments with Default for all merge which Short weave or Major bottleneck Default for all segments with weaving or observed close significant freeway links see high merge areas locations arterial links high merging areas car following weaving throughputs throughput Example Locations Most basic freeway links Arlington Blvd approaching N. Meade St I-66 weave from Route 11; I-66 merge from GW Parkway Theodore Roosevelt Bridge Most arterial links NB/ Key Bridge, / Wilson Blvd in PM NB Lynn St, Wilson Blvd in AM Lee Hwy at N. Lynn St Fort Myer Dr at Fairfax Dr in PM Page 9

11 Core of Rosslyn Transportation Study Model Calibration Summary Conflict Areas and Priority Rules Some conflict areas parameters were modified at locations in which field observations suggested that drivers were willing to accept shorter gaps or were more inclined to block other movements. This involved reducing the front gap, rear gap, safety distance factor, and avoid blocking parameters. Priority rules were used to help prevent complete intersection blocking. They were also used for preventing turning vehicles from conflicting with heavy pedestrian movements when use of conflict areas would cause unrealistic network gridlock. External Congestion Some locations in the study area operate under constrained conditions due to queue spillback from outside of the study area, namely northbound on Key Bridge and eastbound on Theodore Roosevelt Bridge. A traffic signal was added to the edge of the network on Key Bridge to replicate queue spillback from the intersection with M Street NW. Signal phasing and timing was approximated based on field observations of the signal operations. Congestion was replicated on the Theodore Roosevelt Bridge leaving the network using time-dependent and lane-dependent reduced speed areas and by traffic routing based on known weaving patterns at the edge of the network. Speeds were set based on probe vehicle average speeds and local knowledge of traffic operations. Signal Adherence Yellow change intervals of signal timings were changed slightly in some locations where vehicles, pedestrians, or cyclists were observed to use a greater amount of the yellow interval than what was simulated in. The yellow clearance interval of the congested vehicular movements (#1, #2, and #3 below) were shortened to provide more green time to achieve greater throughput. Crosswalk or trail crossing pedestrian clearance intervals were reduced at one location (#4 below) to provide more WALK time. The crossings were previously observed to take multiple cycles to clear the pedestrian demand in whereas in the field pedestrians were observed to cross the intersection in each cycle. By default, pedestrians and bicycles strictly adhere to flashing DON T WALK signal indications in. 1. Eastbound Lee Highway and N. Lynn Street 2. Eastbound Lee Highway and Fort Myer Drive 3. Westbound Fairfax Drive at N. Lynn Street 4. Eastbound/westbound pedestrian and bicycle crossing of Fort Myer Drive along westbound Lee Highway (Custis Trail) Pick-Up and Drop-Off Activity Frequent passenger pick-up and drop-off activity was observed in the rightmost lane of N. Lynn Street between 19 th St N. and eastbound Lee Highway in both AM and PM. This activity was replicated in using parking lot vehicle routes and a parking area in this lane. Vehicles were coded to stop here at a rate of approximately one vehicle per minute and park for an average duration of 2 seconds and up to two minutes. Lane Utilization In the AM, buses from southbound N. Moore Street were observed using the eastbound through lane (rightmost lane) at the intersection of 19 th Street N. and N. Lynn Street to turn left to head northbound toward Lee Highway and I-66. To replicate this de facto turning movement, transit buses destined for Page 1

12 Core of Rosslyn Transportation Study Model Calibration Summary eastbound I-66 were allowed to make this movement in the AM model. Replication of this behavior was also required for volume and throughput calibration along these streets. Calibration Results Existing AM Model An overall summary of the calibration results of the existing AM model is included in Table 6. Overall, all calibration criteria were met. Detailed calibration results can be found in Attachment A. Table 6. AM Calibration Summary Calibration Item Basis Total Percent Target Target Met Simulated Traffic Volume Approaches (Intersections) (n = 79) 74 94% 85% Yes Simulated Traffic Volume Segments (Freeways) (n = 35) 34 97% 85% Yes Simulated Travel Time Routes (Arterials and Freeways) (n = 1) 9 9% 85% Yes Maximum Simulated Length Approaches (Critical Locations) (n = 18) 18 1% 85% Yes Visual Review of Bottleneck Locations Freeway Segments Yes AM Volume Calibration Throughput volumes produced from the model for the AM peak hour were compared with traffic counts for intersection approaches and freeway segments. Intersection volume calibration was accomplished for 94% of approaches. The approaches not meeting the calibration criteria are primarily low volume approaches where a small difference is a large overall percentage of the target volume. Freeway volume calibration was accomplished for 97% of segments. The only segment not meeting the calibration criteria is the eastbound Arlington Boulevard on-ramp from N. Meade St, which is constrained by heavy congestion along Arlington Boulevard. AM Travel Time Calibration Average travel times produced from the model for the AM peak period were compared with field measurements for four arterial routes and six freeway routes. Travel time calibration was accomplished for 9% of travel time routes. The only route not meeting the calibration criteria is eastbound Arlington Boulevard, which has a slightly longer model travel time than the average peak period field travel time. The model travel time is within the magnitude of travels times observed for shorter durations within the peak period, and the visual review of the eastbound Arlington Boulevard bottleneck calibration is comparable to INRIX speed data. AM Length Calibration Average and maximum queue lengths produced from the model for the AM peak period were compared against observed field measurements at 18 critical intersection or ramp approaches, as agreed upon by Arlington County. Both measures were observed to qualitatively assess whether modeled queues reflect the impacts of observed queues (e.g., spillback from a ramp intersection, turn Page 11

13 Core of Rosslyn Transportation Study Model Calibration Summary bay, or downstream intersection). It was determined, based on engineering judgement and knowledge of area travel patterns, that all critical locations adequately represent field conditions. In several locations, field measurements were constrained by sight distance or an upstream intersection, while the queue length (as measured using queue counters in the model) will account for queues spilling back through upstream intersections. For example, queues along eastbound 19 th St N. at N. Lynn Street spill back to the intersection with Key Boulevard, at which point queues extend back along Key Boulevard to the network extents. This was observed in the field and is known to occur from local knowledge of the study area. AM Bottleneck Calibration Speed heat maps for freeways (I-66, Arlington Boulevard, Route 11) qualitatively reflect the patterns and duration of congestion. Model speeds in 15-minute interval by freeway segments are comparable to INRIX speed data for the magnitude and duration of the slow-downs. The model speeds for eastbound I-66 are slightly faster than the INRIX speeds approaching the Arlington Boulevard ramps later in the simulation period (3 mph compared to 15-2 mph). Existing PM Model An overall summary of the calibration of the existing PM model is included in Table 7. Overall, all calibration criteria were met. Detailed calibration results can be found in Attachment A. Table 7. PM Calibration Summary Calibration Item Basis Total Percent Target Simulated Traffic Volume (Intersections) Simulated Traffic Volume (Freeways) Simulated Travel Time (Arterials and Freeways) Maximum Simulated Length (Critical Locations) Visual Review of Bottleneck Locations Approaches (n = 79) Segments (n = 35) Routes (n = 1) Approaches (n = 18) Freeway Segments Target Met 68 86% 85% Yes 3 86% 85% Yes 1 1% 85% Yes 17 94% 85% Yes Yes PM Volume Calibration Throughput volumes produced from the model for the PM peak hour were compared with traffic counts for intersection approaches and freeway segments. Intersection volume calibration was accomplished with 86% of approaches. The approaches not meeting the calibration criteria are primarily due to congestion entering the network onto southbound Fort Myer Drive from Key Bridge, which constrains southbound demand. Freeway volume calibration was accomplished with 86% of segments. The segments not meeting the calibration criteria are segments along eastbound I-66 impacted by congestions on Theodore Roosevelt Bridge and are all less than 1 percent different from target volumes. Page 12

14 Core of Rosslyn Transportation Study Model Calibration Summary PM Travel Time Calibration Average travel times produced from the model for the PM peak period were compared with field measurements for four arterial routes and six freeway routes. Travel time calibration was accomplished all travel time routes. PM Length Calibration Average and maximum queue lengths produced from the model for the PM peak period were compared against observed field measurements at 18 critical intersection approaches. Both measures were observed to qualitatively assess whether modeled queues reflect the impacts of observed queues. It was determined that 17 of the 18 critical locations (94%) adequately represented field conditions. Eastbound Key Boulevard at N. Nash Street has slightly longer model queues than observed, and calibration of this queue length is more critical for AM and not PM. PM Bottleneck Calibration The speed heat maps for freeways (I-66, Arlington Boulevard, Route 11) qualitatively reflect the patterns and duration of congestion. Model speeds in 15-minute interval by freeway segments are comparable to INRIX speed data for the magnitude and duration of the slow-downs. The model speeds for eastbound Arlington Boulevard are slightly faster than field observed speeds approaching Theodore Roosevelt Bridge (3 mph compared to 15-2 mph). Conclusions Based on the results obtained from the AM and PM models and comparison to field data for volume, travel time, queue length, and bottlenecks/speeds, it is concluded that the models are reasonably calibrated to replicate existing multimodal traffic operations as well as to meet the required standards established for the project. These models will be used for future scenario assessment upon the development of future volumes and concepts. Page 13

15 AM Calibration Summary AM Peak Period Calibration Summary Calibration Item Basis Criteria Total Percent Target Target Met Within ± 2% for <1 vph Simulated Traffic Volume (Intersections) Approaches (n = 79) Within ± 15% for 1 vph to < 3 vph Within ± 1% for 3 vph to < 1, vph Within ± 5% for 1, vph 74 94% 85% Yes Within ± 2% for <1 vph Simulated Traffic Volume (Freeways) Simulated Travel Time Segments (n = 35) Routes (n = 1) Within ± 15% for 1 vph to < 3 vph Within ± 1% for 3 vph to < 1, vph Within ± 5% for 1, vph Within ± 1 minute for routes with observed travel times that are less than 7 minutes Within ± 15% for average observed travel time that are greater than 7 minutes 34 97% 85% Yes 9 9% 85% Yes Maximum Simulated Length Visual Review of Bottleneck Locations Approaches (n = 18) Freeway Segments Modeled queues qualitatively reflect the impacts of observed queues Speed heat maps qualitatively reflect patterns and duration of congestions 18 1% 85% Yes Yes

16 AM Volume Calibration (Intersections) Volume Calibration (Intersections) AM Peak Hour (8:-9: AM) Approaches (n = 79) Volume Criteria Within ± 2% for < 1 vph Within ± 15% for 1 vph to < 3 vph Within ± 1% for 3 vph to < 1, vph Within ± 5% for 1, vph Subtotal Total Percent Target % 85% Target Met Yes # Intersection Approach Movement Southbound Key Bridge and George Washington Parkway Fort Myer Drive and Lee Hwy Fort Myer Drive and Lee Hwy Fort Myer Drive and 19th Street N../N. Nash Street Fort Myer Drive and Wilson Blvd N. Nash 17th Street N. Balanced Count (vph) Throughput (vph) Difference (vph) Difference (%) TH 1,37 1, % 1,947 1,949 2 RT % LT % TH % % -8% Intersection 2,697 2, % TH 1,98 1,18 1 1% 1,43 1,42-1 RT % LT % TH % % 4% Intersection 1,716 1, % LT % 1,217 1, TH % TH 1,621 1,623 2 % 1,914 1, RT % 2% 1% Intersection 3,131 3, % LT % TH 689 1, , % 3% RT % TH % RT % LT % TH % -4% % Intersection 1,832 1, % LT % TH % Tunnel % -4% RT % TH % RT % -1% U-Turn % LT % -2% TH % Intersection 2,544 2, % LT % RT % LT % TH % TH % RT % 2% -1% % Intersection %

17 AM Volume Calibration (Intersections) # Intersection Approach Movement Balanced Count (vph) Throughput (vph) Difference (vph) Difference (%) Fort Myer Drive and Fairfax Drive Fort Myer Drive/N. Meade Street and Arlington Blvd Ramp N. Moore Lee Hwy N. Moore 19th St N. N. Moore Wilson Blvd N. Lynn Lee Hwy N. Lynn Lee Hwy NB LT % TH % RT % TH % RT % LT % % TH % Intersection 1, % LT % 1,333 1, TH 1,235 1, % TH % RT % LT % TH % % RT % Intersection 2,64 2, % NB RT % -13% TH 1,854 1,854 1,847 1, % % Intersection 1,885 1, % LT % TH % 41% RT % LT % TH % 6% RT % LT % TH % 2% RT % Intersection % RT % 2% TH 1,51 1, % -7% TH % -3% Intersection 1,825 1, % NB LT % 2,741 2, TH 2,626 2, % TH % RT % -3% 1% Intersection 3,463 3, % NB TH 1,759 1, % 1,941 1, RT % LT 982 1,4 22 2% 1,885 1, TH % -4% -1% Intersection 3,826 3, % -8% -5% 1% -8%

18 AM Volume Calibration (Intersections) # Intersection Approach Movement N. Lynn St and 19th St N. N. Lynn Wilson Blvd N. Lynn St and 17th St N. N. Lynn Fairfax Drive N. Nash Lee Hwy N. Nash Lee Hwy N. Nash Key Blvd Balanced Count (vph) Throughput (vph) Difference (vph) Difference (%) LT % NB TH 1,435 1,639 1,347 1, % -5% RT % LT % TH % TH % RT % 6% -2% Intersection 2,624 2, % LT % NB TH 1,288 1,742 1,37 1, % -1% RT % LT % 1, TH % TH % RT % -8% 3% Intersection 3,281 3, % NB TH % 6% LT % 9% Intersection % LT % NB TH 1,6 1,689 1,65 1, % % RT % LT % TH % TH % RT % -8% -3% Intersection 1,997 1, % NB LT % TH % TH % RT 2 2 % 5% 8% LT % TH % -4% RT % Intersection % NB TH % RT % LT % TH % -7% -18% LT % TH 1,679 1,772 1,742 1, % 2% RT % Intersection 2,83 2,86 3 % NB LT % TH % TH % RT % LT % RT % -11% -4% -1% Intersection %

19 AM Volume Calibration (Intersections) # Intersection Approach Movement N. Nash St and Wilson Blvd N. Nash Arlington Blvd N. Meade St and Arlington Blvd N. Kent 19th St N. N. Kent Wilson Blvd Balanced Count (vph) Throughput (vph) Difference (vph) Difference (%) LT % NB TH % -12% RT % LT % TH % -9% RT % LT % TH % -8% RT % U-Turn % LT % TH % 3% RT % Intersection 1,923 1, % NB LT % RT % TH % RT % LT % TH % -6% -14% -3% Intersection % LT % NB TH % -1% RT % LT % TH % -1% RT % LT % TH % -14% RT % Intersection 1,213 1, % LT % NB TH % 1% RT % LT 3 3 % TH % -2% RT % LT % TH % -3% RT % LT 8 8 % TH % -1% RT % Intersection % LT % TH % TH % RT % -7% % Intersection 1,413 1, %

20 AM Volume Calibration (Intersections) # Intersection Approach Movement N. Arlington Ridge Road and Wilson Blvd N. Oak Street and Wilson Blvd N. Oak Street and Clarendon Blvd Balanced Count (vph) Throughput (vph) Difference (vph) Difference (%) LT % NB TH % -5% RT % LT % TH % % RT % LT % TH % -7% RT % LT % TH 781 1, , % 1% RT % Intersection 1,856 1, % NB LT % TH % TH % RT % TH % RT % -5% -3% % Intersection % NB TH % RT % LT % TH % LT % TH % -3% -6% -8% Intersection 1,182 1, %

21 AM Volume Calibration (Freeways) Volume Calibration (Freeways) AM Peak Hour (8:-9: AM) Segments (n = 35) Volume Criteria Within ± 2% for < 1 vph Within ± 15% for 1 vph to < 3 vph Within ± 1% for 3 vph to < 1, vph Within ± 5% for 1, vph Subtotal Total Percent Target Target Met % 85% Yes Interchange I-66 Eastbound I-66 Westbound US 5 Eastbound US 5 Westbound Route 11 Northbound Route 11 Southbound Segment Upstream On-ramp from Lee Hwy Between Ramps Off-ramp to Route 11 Between Ramps On-ramp from GW Pkwy Between Ramps On-ramp from US 5 Downstream Upstream Off-ramp to US 5 and GW Pkwy Off-ramp to GW Pkwy Off-ramp to US 5 Between Ramps On-ramp from NB Route 11 Between Ramps Off-ramp to Lee Hwy/Key Bridge Downstream Upstream Off-ramp to NB N Meade St Between Ramps On-ramp from N Meade St Between Ramps Off-ramp to I-66 Off-ramp to GW Pkwy On-ramp from I-66 On-ramp from GW Pkwy On-ramp from I-66 and GW Pkwy On-ramp from NB GW Pkwy Between Ramps Off-ramp to NB N Lynn St Between Ramps On-ramp from Ft Myer Dr Downstream Upstream Off-ramp to I-66 Downstream (Wilson Blvd) Upstream (Wilson Blvd) On-ramp from I-66 Downstream Type Balanced Count (vph) Throughput (vph) Difference (vph) Difference (%) Basic 2,576 2, % Ramp 1,85 1, % Weave 3,661 3, % Ramp % Basic 2,72 2, % Ramp % Merge 3,579 3, % Ramp 1,74 1, % Basic 4,653 4, % Diverge 2,855 2,853-2 % Ramp 1,177 1, % Ramp % Ramp % Basic 1,678 1,678 % Ramp 1,23 1,22-1 % Weave 2,71 2, % Ramp % Basic 1,979 1,978-1 % Diverge 2,922 2,92-2 % Ramp % Basic 1,985 1, % Ramp % Weave 2,555 2, % Ramp See I-66 Eastbound above Ramp 1,481 1, % Ramp See I-66 Westbound above Ramp % Ramp 1,2 1,195-5 % Ramp % Weave 1,966 1, % Ramp % Basic 1,459 1, % Ramp % Merge 1,847 1, % Basic 2,365 2,373 8 % Ramp Basic See Westbound I-66 above See Wilson Blvd/Arlington Ridge Rd intersection Basic % Ramp See Eastbound I-66 above Basic 1,321 1, %

22 AM Travel Time Calibration Travel Time Calibration Travel Time - AM Peak Period (7:3-9:3 AM) Routes (n = 1) Segment ID Travel Time Criteria Subtotal Total Percent Target Target Met Within ± 1 minute for routes with observed 9 travel times that are less than 7 minutes 9 9% 85% Yes Within ± 15% for average observed travel time that are greater than 7 minutes Route Field (MM:SS) Peak Period Travel Time Difference (MM:SS) (MM:SS) Difference (%) Arterial Routes 1 Northbound Lynn St (Arlington Blvd to George Washington Memorial Pkwy) 5:47 5:27 -:2-6% 2 Southbound Fort Myer Dr (George Washington Memorial Pkwy to Arlington Blvd) 2:25 3:1 :36 25% 3 Eastbound Clarendon Blvd (N. Rhodes St to N. Arlington Ridge Rd) 4:27 5:17 :5 19% 4 Westbound Wilson Blvd (N. Arlington Ridge Rd to N. Oak St) 2:58 2:23 -:35-2% Freeway Routes 5 Eastbound I-66 ( Lee Hwy off-ramp to Roosevelt Bridge) 3:24 4:17 :54 26% 6 Westbound I-66 (Roosevelt Bridge to 21st St N. overpass) 1:41 1:44 :3 3% 7 Eastbound Arlington Boulevard (N. Rolfe St to Roosevelt Bridge) 8:13 9:38 1:26 17% 8 Westbound Arlington Boulevard (Roosevelt Bridge to N. Rolfe St) 1:21 1:28 :7 9% 9 Northbound Route 11 (Memorial Ave to I-66) :5 :56 :6 11% 1 Southbound Route 1 (N. Arlington Ridge Rd to Memorial Ave) :32 :32 -:1-3%

23 AM Length Calibration Length Calibration AM Peak Period (7:3-9:3 AM) Approaches (n = 18) Criteria Modeled queues qualitatively reflect the impacts of observed queues (e.g., spillback from ramp intersections, turn bay, or downstream intersection) Total Percent Target Target Met 18 1% 85% Yes Location NB N. Lynn St at Fort Myer Dr/GW Parkway on-ramp NB N. Lynn St at Key Bridge GW Parkway off-ramp to Key I-66 off-ramp at N. Lynn St Lee Hwy N. Lynn St Fort Myer Dr at 19th St N. Key Blvd at N. Nash St 19th St N. at N. Lynn St NB N. Lynn St at Wilson Blvd Observed Max (feet) Max (feet) Average (feet) Max Difference (feet) Max Difference (%) Field Conditions Represented (Yes/No) Notes 36 1, , % Y spills back upstream into Rosslyn core 2,64 1, % Y spills back upstream into Rosslyn core 92 1,1 1, % Y spills back to GWP 1,38 2, % Y spills back into mainline weave area 3,2 2, % Y reaches N. Scott St Bridge % Y Average queue represents observed typical condition within the block of 19th St N. and Lee Hwy; maximum queue length reflects occasional rolling queue spillback back to Lee Hwy 1,52 1, % Y spills back along Key Blvd to edge of network (N. Quinn St) 2,2 2, % Y spills back past Fort Myer Dr 1,275 1, % Y representative of occasional spill back onto Arlington Blvd onramp, consistent with field observations.

24 AM Length Calibration Location Wilson Blvd at N. Lynn St Wilson Blvd at Fort Myer Dr Wilson Blvd at Fort Myer Dr Clarendon Blvd at N. Oak St Wilson Blvd at N. Oak St Observed Max (feet) Max (feet) Average (feet) Max Difference (feet) Max Difference (%) Field Conditions Represented (Yes/No) 9 1, % Y 1,33 1, % Y Notes is metered by signal at Fort Myer Dr; max queue can spill back upstream through several closelyspaced intersections to N. Oak St as can be observed within sight distance in the field is metered by upstream intersections; max queue can spill back along Wilson Blvd/Clarendon Blvd out of Rosslyn core (west of N. Oak St) as observed in the field % Y fills block back to N. Lynn St 9 2, , % Y % Y Maximum queue extends to N. Pierce St per observation; slow-moving platoon (maximum rolling queue) is observed to extend to N. Rhodes St (1,8 feet) fills block back to Fort Myer Dr N. Moore St at Wilson Blvd % Y Average queue represents observed typical condition with minimal queuing impacts on N. Moore St NB N. Lynn St at Fairfax Dr 68 1, % Y ing back to Arlington Boulevard off-ramp to N. Meade St

25 AM Length Calibration Location Observed Max (feet) Max (feet) Average (feet) Max Difference (feet) Max Difference (%) Field Conditions Represented (Yes/No) Notes Arlington Blvd on-ramp from N. Meade St Arlington Blvd off-ramp to N. Meade St 1,93 2,651 1, % Y 1,6 2, % Y measurement in includes Arlington Blvd frontage road; queues along on-ramp spill back from Arlington Blvd back to N. Meade St as observed in the field; the maximum queue on Arlington Blvd frontage road is 1,2 feet spills back onto Arlington Blvd mainline as observed in the field. Observed queue length is limited to sight distance.

26 AM Bottleneck Calibration Eastbound I-66 Freeway Average Speed Comparison Westbound I-66 N Scott Street N Lynn Street Route 11 US 5 Ramp D.C. N Scott Street N Lynn Street Route 11 US 5 Ramp D.C. 7:3 AM 7:3 AM 8: AM 8: AM 8:3 AM 8:3 AM 9: AM 9: AM 9:3 AM 9:3 AM SPEED SPEED 1: AM 1: AM 1:3 AM 1:3 AM 7:3 AM 7:3 AM 8: AM 8: AM 8:3 AM 8:3 AM 9: AM 9: AM 9:3 AM 9:3 AM 1: AM 1: AM OBSERVED SPEED OBSERVED SPEED 1:3 AM 1:3 AM Average Speed (mph) Average Speed (mph) 7

27 AM Bottleneck Calibration Freeway Average Speed Comparison Eastbound Arlington Blvd Westbound Arlington Blvd 7:3 AM 7:3 AM 8: AM 8: AM N Rhodes Street N Meade Street Roosevelt Bridge Roosevelt Bridge N Meade Street N Rhodes Street 8:3 AM 8:3 AM 9: AM 9: AM 9:3 AM 9:3 AM SPEED SPEED 1: AM 1: AM 1:3 AM 1:3 AM 7:3 AM 7:3 AM 8: AM 8: AM 8:3 AM 8:3 AM 9: AM 9: AM 9:3 AM 9:3 AM 1: AM 1: AM OBSERVED SPEED OBSERVED SPEED 1:3 AM 1:3 AM Average Speed (mph) 7

28 AM Bottleneck Calibration Freeway Average Speed Comparison Northbound Route 11 Southbound Route 11 7:3 AM 7:3 AM 8: AM 8: AM SPEED Memorial Drive I-66 I-66 Memorial Drive 8:3 AM 8:3 AM 9: AM 9: AM 9:3 AM 9:3 AM SPEED 1: AM 1: AM 1:3 AM 1:3 AM 7:3 AM 7:3 AM 8: AM 8: AM OBSERVED SPEED 8:3 AM 8:3 AM 9: AM 9: AM 9:3 AM 9:3 AM 1: AM 1: AM OBSERVED SPEED 1:3 AM 1:3 AM Average Speed (mph) 7

29 PM Calibration Summary PM Peak Period Calibration Summary Calibration Item Basis Criteria Subtotal Total Percent Target Target Met Within ± 2% for <1 vph 11 Simulated Traffic Volume (Intersections) Approaches (n = 79) Within ± 15% for 1 vph to < 3 vph Within ± 1% for 3 vph to < 1, vph Within ± 5% for 1, vph % 85% Yes Within ± 2% for <1 vph Simulated Traffic Volume (Freeways) Simulated Travel Time Segments (n = 35) Routes (n = 1) Within ± 15% for 1 vph to < 3 vph Within ± 1% for 3 vph to < 1, vph Within ± 5% for 1, vph Within ± 1 minute for routes with observed travel times that are less than 7 minutes Within ± 15% for average observed travel time that are greater than 7 minutes % 85% Yes 1 1% 85% Yes Maximum Simulated Length Visual Review of Bottleneck Locations Approaches (n = 18) Freeway Segments Modeled queues qualitatively reflect the impacts of observed queues Speed heat maps qualitatively reflect patterns and duration of congestions 17 94% 85% Yes Yes

30 PM Volume Calibration (Intersections) Volume Calibration (Intersections) PM Peak Hour (5:-6: PM) Approaches (n = 79) Volume Criteria Within ± 2% for < 1 vph Within ± 15% for 1 vph to < 3 vph Within ± 1% for 3 vph to < 1, vph Within ± 5% for 1, vph Subtotal Total Percent Target % 85% Target Met Yes # Intersection Approach Movement Southbound Key Bridge and George Washington Parkway Fort Myer Drive and Lee Hwy Fort Myer Drive and Lee Hwy Fort Myer Drive and 19th Street N../N. Nash Street Fort Myer Drive and Wilson Blvd N. Nash 17th Street N. Balanced Count (vph) Throughput (vph) Difference (vph) Difference (%) TH 2,21 1, % 2,455 2, RT % LT % TH % -6% -7% Intersection 3,189 2, % TH 1,663 1, % 2,61 1, RT % LT % TH % -7% 2% Intersection 2,788 2, % LT % 1,785 1, TH 1, % TH % 1,21 1, RT % -7% 2% Intersection 2,995 2, % LT % TH 1,11 1,391 1,25 1, % -8% RT % TH % RT % LT % TH % 15% -5% Intersection 2,22 1, % LT % TH % 1,427 1, Tunnel % -6% RT % TH % RT % 1% U-Turn % LT % 5% TH % Intersection 3,71 3, % LT % RT % LT % 6 6 TH % TH % RT % -3% % -6% Intersection %

31 PM Volume Calibration (Intersections) # Intersection Approach Movement Balanced Count (vph) Throughput (vph) Difference (vph) Difference (%) Fort Myer Drive and Fairfax Drive Fort Myer Drive/N. Meade Street and Arlington Blvd Ramp N. Moore Lee Hwy N. Moore 19th St N. N. Moore Wilson Blvd N. Lynn Lee Hwy N. Lynn Lee Hwy NB LT % TH 1,22 1,425 1,159 1, % RT % TH % RT % LT % TH % Intersection 1,73 1, % LT % TH % TH % 1,411 1, RT % LT % TH % 4% RT % Intersection 2,828 2, % NB RT % -26% TH 1,58 1,58 1,581 1, % % Intersection 1,762 1, % LT % TH % 14% RT % LT % TH % 9% RT % LT % TH % -19% RT % Intersection % RT % -1% TH 1,8 1,8 1,86 1, % 1% TH % 3% Intersection 1,85 1,811 6 % NB LT % 2,184 2, TH 1,892 1,895 3 % TH % RT % 1% -2% Intersection 3,118 3,123 5 % NB TH 1,48 1,485 5 % 1,81 1, RT % LT % 1,762 1, TH 1, % -2% -3% Intersection 3,563 3, % -4% % -6% % -4%

32 PM Volume Calibration (Intersections) # Intersection Approach Movement N. Lynn St and 19th St N. N. Lynn Wilson Blvd N. Lynn St and 17th St N. N. Lynn Fairfax Drive N. Nash Lee Hwy N. Nash Lee Hwy N. Nash Key Blvd Balanced Count (vph) Throughput (vph) Difference (vph) Difference (%) LT % NB TH 1,323 1,487 1,296 1, % -3% RT % LT % TH % TH % RT % 12% -15% Intersection 2,23 2, % LT % NB TH 92 1, , % 2% RT % LT % 1,8 1,86 6 TH % TH % RT % 1% -1% Intersection 3,86 3, % NB TH % -8% LT % 19% Intersection % LT % NB TH 1,15 1,224 1,175 1, % 2% RT % LT % TH % TH % RT % 1% -1% Intersection 1,483 1, % NB LT % TH % TH 1 1 % RT % -27% -7% LT % TH 962 1, % -2% RT % Intersection 1,112 1, % NB TH % RT % LT % TH % % 4% LT % TH 1,89 1,21 1,11 1, % % RT % Intersection 1,438 1, % NB LT % TH % TH % RT % LT % RT % -13% -6% 1% Intersection %