Evaluation of MCS impacts - MITweb.mit.edu/11.951/oldstuff/albacete/Other_Documents/Europe Transport...

© Association for European Transport and contributors 2006

EVALUATION OF IMPACTS OF THE MOTORWAY CONTROL SYSTEM (MCS) IN STOCKHOLM

Nissan, Albania, Ph.D. Candidate

and Bang, Karl L. Professor em.

Department of Transport and Economics Royal Institute of Technology (KTH)

ABSTRACT Stockholm as many other major urban areas suffers from road traffic congestion due to difficulties to provide much needed capacity expansion of its road networks to cope with the increasing demand. Major arterials operate close to or over capacity during morning and afternoon rush hours, which results in recurring congestion and serious disturbances even by relatively minor incidents. A Motorway Control System (called MCS) has been implemented on the E4 motorway through Stockholm with recommended speed signalling triggered by automatic incident detection alarms for downstream queues. A major research study was undertaken in cooperation with national and regional road authorities to assess the impacts of MCS on traffic performance and traffic safety. The project included extensive field studies with stationary as well as mobile data collection techniques aimed at capturing individual driver behaviour (speed adaptation, deceleration process) as well as traffic characteristics (flow, speed, lane distribution, headways). The impacts of MCS were analysed by comparison of survey results for time periods with high, middle and low traffic flows without and with the system in operation. The study concluded that the main MCS impact consisted of a sharp decrease of the standard deviation of speed, which would indicate an improvement in homogeneity and traffic safety. Furthermore MCS reduced the frequency of very short headways as well as the frequency of lane changes between the middle and the left lane. Driver behaviour impacts were difficult to assess due to the difficulties to perform controlled experiments with or without active MCS for comparable situations. Most drivers seemed to ignore the recommended speed signs since they were not mandatory. However, there was an indication that recommended speed signalling warning for downstream queuing influenced the deceleration process to become more stable.


1. INTRODUCTION

1.1 Background Stockholm as many other major urban areas suffers from road traffic congestion due to difficulties to provide much needed capacity expansion of its road networks to cope with the increasing demand. Many arterial roads operate close or over capacity during morning and afternoon rush hours, which causes serious recurring traffic congestion as well as incident impacts. The Swedish Road Authority (SRA) in cooperation with the Stockholm government have therefore implemented a Motorway Control System (called MCS) based on the Dutch MTM system on the E4 motorway through Stockholm. Although MCS has been in operation since the late nineties and is currently being expanded, there is yet no comprehensive knowledge regarding the impacts of this system on traffic performance and traffic safety. This paper is based on a major research project undertaken at KTH in cooperation with national and regional road authorities in Stockholm aimed at filling this knowledge gap.

Figure 1, Route E4 approaching Stockholm during morning rush hour.


1.2 The Stockholm Motorway Control System (MCS) The northern part of the E4 motorway (8 km) was equipped with MCS in 1996, and the southern part (12 km) in 2004. The latter section has an AADT around 150 000 vehicles, and flows around 5000-6000 veh/h per direction of travel (3 lanes 3.5 m wide). MCS is essentially a main-line link control system, although some ramp metering has also been implemented in 2005 to avoid excessive overloading of critical road segments. The basic function of the system is to detect incidents in the form of slow moving traffic, and to reduce the risk for collisions by display of recommended speed limits using Variable Message Signs (VMS) on gantries upstream of the incident. This involves the following process as shown in Figure 2:

1. Traffic surveillance for recording of the actual traffic situation and

automatic incident detection. Microwave detectors collect information of vehicle passages in each lane as a basis for calculation of flow, speed, occupancy and headway, which is, aggregated as one-minute averages. Smoothed averages of speed are also calculated as a basis for Automatic Incident Detection alarms (AID), which can be verified using closed circuit television (CCVT).

2. Lane signalling using VMS for communication with the drivers.

AID-alarms based on detection of speed using two threshold levels (35 and 50km/h). If one of the values subsides its threshold, an AID legend-request results which triggers pre-determined lane-signalling plans aimed at reducing the speed of traffic approaching the location of the detected incident. These signals are displayed by VMS panels mounted on gantries above each lane every 500 m which can show recommended speed levels (70, 60, 50, 30 km/h) or lane closure.

3. Driver response to the lane signalling and surrounding traffic situation.

Drivers approaching a gantry with activated lane signals can react in different ways; ignore the signal message or change their speed and/or selected lane in response to the signal. The purpose of MCS is that the drivers will change their behaviour in such ways that it will improve traffic safety and traffic performance on the controlled road link.


Figure 2: Flow chart of the MCS control process

1.3 Objectives and Scope

The main purpose of the project was to gain improved knowledge of MCS

implementation for link control, especially focusing on its impact on traffic

safety and traffic performance, two important objectives of transport policy.

1.4 Research Methodology On congested motorways shockwaves are normally frequent causing “stop-and-go” waves, in which vehicles can only travel with reduced speed or come to a complete standstill for periods of time. Furthermore congestion increases the amount of interaction between vehicles and leads to a high amount of lane changes in very short gaps. In order to capture these characteristics as well as individual driver behaviour in general, the data collection has to be very detailed. The selected research strategy for the project described in this paper was therefore “microscopic” assessing traffic performance and safety impacts based on studies of driver response to the displayed MCS lane signals. The project included the following main stages:

A) Study of driver behaviour in response to MCS variable speed sign messages, and related local traffic performance and safety effects.

B) Study of impacts of MCS on traffic process along the road. C) Coding of MCS- functionality and driver behaviour characteristics in

a microscopic simulator. D) Calibration, validation and application of the simulation model for

the usage of MCS impact assessment.

Driver (Traffic

Surveillance system

AID

VMS Speed, lane closure


This paper covers results obtained in stage A and B above. Stage C is planned to be completed in 2007 and results from the empirical study regarding driver behaviour related to MCS operation will be incorporated in a microscopic simulation model (e.g. VISSIM). In addition, the Automatic Incident Detection (AID) algorithm will be implemented in the simulator for further analyses of different strategies and improvement of the MCS functionality. This will enable “controlled experiments” in Stage D for evaluation of system impacts and alternative strategies for application of the system and its operation in the traffic management centre (called TRAFIK STOCKHOLM).

1.5 International Literature Review Motorway link control using advisory or enforced variable speed signs to warn drivers of incidents blocking the driveway such as queues downstream have been applied in many countries since the early seventies (Kühne, R.D., 1991; Smulders, S., 1990, 1992, 1996; Smulders, S.Helleman, D.E., 1998; Zackor, H., 1979). Similar techniques can also be used to warn for adverse weather conditions such as fog (Cooper, B.R.Sawyer, H., 1993; Hogema, J.H.Van der Horst, R., 1997), slippery road (Rämä, Pirkko., 2001) and work zones (Lin, P. W. et al., 2004; Ober-Sudermeier, A. Zackor, H., 2001), In an overview of speed limit systems (Hegyi, A., 2004) a distinction is made between 1) approaches that aim at homogenization, and 2) resolution of shockwaves or jams. Most applications of link control emphasize homogenization or stabilization effects i.e., (Rämä, Pirkko., 2001; Smulders, S., 1990, 1992, 1996; Zackor, H., 1979). For proper operation of such systems real-time information of traffic flow, speed, occupancy and for some type of systems also weather, visibility or road surface condition are required to determine the appropriate speed at which drivers should be travelling. Several authors report that enforced speed limits lead to higher driver compliance which is necessary to ensure efficient system operation (Harbord, B., 1997; Rämä, Pirkko., 2001; Smulders, S., 1990, 1992, 1996; Van den Hoogen, E.Smulders, S., 1994; Wilkie, J. K., 1997) Homogenization measures such as speed control aim to reduce the flow and speed variance between or within lanes. It also aims at decreasing the fraction of small headways in order to reduce the risk for shock waves, thereby minimizing accident risks and upstream congestion. (Remeijn, H., 1982) shows that variable speed signs for driver warning have been successfully applied on many Dutch motorways. Other studies, e.g. (Garber, N.J. Gadiraju, R., 1989), show that accident rates do not necessarily increase with an increase in average speed, but rather with increased speed variance. The aim of homogenization should therefore not be reduction of the mean speed but reduction of the speed differences (Cremer, M., 1978), (Smulders, S., 1990).


Experiments reported by Van den Hoogen (1994) indicate that Variable Speed Limit (VSL) control does not eliminate congestion at bottlenecks since the capacity of the roadway is not increased, however the effects of the bottlenecks on upstream traffic can be lessened. Other studies use variable speed limits with the aim to reduce the occurrence of shock waves on motorways, (Hegyi, A., 2004; Hegyi, A. et al., 2003, 2003). An attempt was made by Hegyi (2004) to outline the difference between homogenization and breakdown prevention in terms of when to apply speed limits. The homogenization approach uses limits that are above the critical speed (i.e., the speed that corresponds to the maximal flow), while the breakdown preventing approach allows speed limits that are lower than the critical speed in order to limit the inflow to bottleneck areas. Although link control using variable speed signs has been applied on a wide scale in several countries since the seventies, there is a lack of research studies evaluating the impact of these systems. Most of the developed models are not validated with any empirical data (Alessandri, A. et al., 1997; Hegyi, A. et al., 2002). Kostsialos et al. (2002), and Papageorgiou et. al (2003) state that ”very few systematic studies have been conducted to quantify the impact of link control measures. One important component in the control loop in variable speed control management is driver behaviour, which is difficult to detect and forecast adding uncertainty to the problem of traffic management.

2. DATA COLLECTION The study included a pilot study on a stretch of motorway on the northern part of the E4 motorway, where MCS had been active since 1996. The main study was carried out on the southern part of the same motorway where MCS was implemented in 2004. During the “before” surveys in May 2004 MCS was functioning with AID and VMS message selection without displaying any messages to the drivers. The “after” study was performed in September 2004 with fully activated MCS. Both the “before” and the “after” studies included stationary as well as mobile data collection. The following traffic characteristics were observed for each lane:

- Traffic flow distribution and composition. - Headway and headway distribution (headway defined as the time

between passage of the front axel of the leading vehicle and the front axel of the following vehicle).

- Spot speed and speed distribution. - Travel time and delay. - Lane change frequency.


2.1 Stationary Studies Two methods of data collection were used for the stationary studies: A: MCS data collection B: Video recording A: MCS data collection Data provided by the MCS detectors is aggregated by outstations in the field before it is transmitted to the traffic control centre. The resulting information includes total traffic flow and the average speed for all lanes per minute. Furthermore the system records AID alarms and the time for activation and deactivation of the recommended speed signs and lane closure signals. The studied segment for MCS data collection included ten gantries. Figure 3 below illustrates an example of the plotted data. Figure 3. Average speed and displayed recommended speed for a selected gantry (plotting of MCS data) B: Video recording from elevated positions To gain more insight into the behaviour of drivers during congested traffic conditions a very detailed data collection is required. For this purpose video recording from elevated positions was selected as described below. Four bridges were used as vantage points for video camcorders covering the movements of each vehicle in both directions of travel. From the Tram Bridge the road section was observed by two remote controlled video cameras mounted on a 15 m high mast. Four additional cameras were mounted on the other bridges as shown in Figure 4.

Average speed (km/h) and the displayes (VMS) recommended speed in gantry nr. (649)

0

20

40

60

80

100

7:00

7:05

7:10

7:15

7:20

7:25

7:30

7:35

7:40

7:45

7:50

7:55

8:00

8:05

8:10

8:15

8:20

8:25

8:30

8:35

8:40

8:45

8:50

8:55

9:00

Time (min)

Aver

age

spee

d (k

m/h

)

VMS Average speed


Figure 4. Studied segment of E4 equipped with MCS detectors in each lane and four gantries (location in meters) with VMS showing the positions of the video recorders used for research data collection. To capture traffic behaviour on individual vehicle level the collected data and the videotapes were processed using the SAVA video analyzer program developed at KTH (Archer, J., 2003). The SAVA program has been designed to interpret the information from digital films recorded in Digital Audio Video Interleaved (*.avi) format as illustrated in Figure 5. The collected data consists of vehicle passage times, traffic flow, vehicle speed, headway, vehicle type, and lane changes for each vehicle in each lane.

Figure 5. Analysis of video film recorded from the Tram Bridge in the

northbound direction towards Stockholm

47.46548.290 48.620

48.93547.800

47.255

47.635 47.98048.935

48.38549.220

Tram Bridge


2.2 Mobile studies A comprehensive car- following study was undertaken during two weeks in the morning and afternoon peak traffic period 07:00 to 09:00 and 15:00-17:00. The mobile study was carried out using an instrumented vehicle (Volvo V70) equipped with a computer for automatic recording of trip data, which was logged on a laptop. The collected data included travel distance (m), velocity (m/sec), longitudinal acceleration (m/s2), latitudinal acceleration (m/s2), GPS-based information, film recording of the view ahead and behind the vehicle, etc. The mobile data collection was performed using different drivers, male and female, chosen randomly. The purpose of the study was to capture detailed driver behaviour data focusing on the driver response to the recommended speeds displayed on the panels. Figure 6, shows an example of plotted results of vehicle speed and displayed recommended speed.

Figure 6, Floating car measurement with KTH instrumented vehicle.

3. DESCRIPTIVE RESULTS Selected results from data reduction of the “before” and “after” surveys are documented below.

3.1 Traffic Flow and Composition Table 1 records traffic flow, speed and headway data for each of the three lanes during time periods representing a range of traffic flow conditions: High traffic 07:00-09:00, Low traffic 11:00-13:00, and Middle traffic 15:00-16:00.

Speed km/h of the instrumented Car-followingand the displayed recommended speed 07:43 - 07:52

613618622626630

663 658 654 649 646 640 635 630 626 622 618 613

663 658 654 649 646 640 6350

20

40

60

80

100

120

07:43 07:44 07:45 07:46 07:47 07:48 07:49 07:50 07:51 07:52

Time (min)

Spee

d [k

m/h

]

Vehicle speed(km/h)

Detector Stations

Displayed speed,VMS

Designed speed


Table 1, Traffic flow (veh/2 hours) during different time periods in each lane (LL = left Lane; ML = Middle Lane; RL = Right Lane) During the morning peak traffic period (north bound) the left lane carries considerably more traffic than the other two lanes as can be seen from the table. The lane distribution becomes more balanced after the implementation of MCS as shown for the southbound direction for the same time period. During periods with lower traffic the middle lane carries the highest traffic load before as well as after implementation of MCS.

3.2 Lane Changes Lane changes were observed through video recordings over a stretch of 200 m length. Table 2 shows results for the two hour morning peak traffic period 07:00 – 09:00 regarding frequency of lane changes and their distribution.

Table 2, Frequency and ratio of lane changes per two hours and 200 m between different pairs of lanes during the morning peak period without and with MCS operation. The number of observed lane changes was more than twice as high without as with MCS in operation. Most lane changes without MCS occurred from the middle lane to the left lane, with MCS in operation this occurred from the right to the middle lane.

Lane LL ML RLNorth bound 4040 3360 3384

South bound before MCS 3324 3129 2778South bound after MCS 2483 2272 2235





Midday period 11:00-13:00

Morning peak period

07:00-09:00

Afternoon period 15:00-17:00

Lane Changing Without MCS Share With MCS ShareLL-ML 110 15% 100 30%ML-LL 221 31% 18 5%ML-RL 162 22% 67 20%RL-ML 180 25% 125 38%LL-ML-RL 20 3% 8 2%RL-ML-LL 30 4% 15 5%Total 723 100% 333 100%


3.3 Speed Characteristics Speed characteristics for different lanes and time periods with and without MCS in operation periods were studied. Figure 7 illustrates individual vehicle speeds between 15:30 and 17:30 in the outbound direction towards Stockholm before and after MCS implementation. The figure shows that drivers in left lane maintained much higher speed than those in the middle and right lanes, and drove faster than the speed limit (90 km/h) during the whole period. This was also evident from the speed distribution chart shown for each lane for low traffic conditions in Figure 8 below.

Figure 7, Individual vehicle speed observations (km/h) for three lanes during the afternoon period “before” (upper figure) and “after” (lower figure) after MCS implementation.

Speed (km/h) for all lanes (after study 15:30-17:30)

010203040506070

8090

100110120130140

15:3

0:00

15:3

5:00

15:4

0:00

15:4

5:00

15:5

0:00

15:5

5:00

16:0

0:00

16:0

5:00

16:1

0:00

16:1

5:00

16:2

0:00

16:2

5:00

16:3

0:00

16:3

5:00

16:4

0:00

16:4

5:00

16:5

0:00

16:5

5:00

17:0

0:00

17:0

5:00

17:1

0:00

17:1

5:00

17:2

0:00

17:2

5:00

17:3

0:00

Time (sec)

Spe

ed (k

m/h

)

LL ML RL

Speed (km/h) for all lanes (before study 15:30-17:30)

0.0010.0020.0030.0040.0050.0060.0070.0080.0090.00

100.00110.00120.00130.00140.00

15:3

0:00

15:3

5:00

15:4

0:00

15:4

5:00

15:5

0:00

15:5

5:00

16:0

0:00

16:0

5:00

16:1

0:00

16:1

5:00

16:2

0:00

16:2

5:00

16:3

0:00

16:3

5:00

16:4

0:00

16:4

5:00

16:5

0:00

16:5

5:00

17:0

0:00

17:0

5:00

17:1

0:00

17:1

5:00

17:2

0:00

17:2

5:00

17:3

0:00

Time (sec.)

Spee

d (k

m/h

)

LL ML RL


Speed and headway distributions for each lane during middle traffic flow conditions different periods are exemplified in Figure 8. From this figure it can be seen that the drivers in the left lane maintained higher speed and shorter headways than those in the other lanes. Figure 8. Example of speed and headway distribution in different lanes during middle traffic flow conditions “before” MCS When the traffic demand increased shockwaves and congestion occurred as shown in Figure 9 for “before” and Figure 10 for “after” conditions (without and with MCS in operation). The speed dropped considerably and the traffic discharge occurred in a queuing mode with shock waves of different severity at the tail of the queue. MCS was implemented to warn the oncoming traffic for such queuing conditions downstream through display of lower recommended speed as shown in Figure 10.

Frequency of headways for all lanes (15:30-15:45)

0

20

40

60

80

100

120

0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5

Headway (Sec.)

Freq

uenc

y

LLMLRL

Frequency of speeds for all lanes (15:30-15:45)

0

50

100

150

200

250

0 20 40 60 80 100 120 140

Speed (km/h)

Freq

uenc

y

LLMLRL


Figure 9, Vehicle speed (km/h) during morning peek traffic “before” implementation of MCS

Figure 10. Vehicle speed (km/h) during morning peek traffic “after” implementation of MCS

Speed km/h for three lanes (before study 07.00-09:00)

0.00

20.00

40.00

60.00

80.00

100.00

120.00

140.00

07:0

0:00

07:0

5:00

07:1

0:00

07:1

5:00

07:2

0:00

07:2

5:00

07:3

0:00

07:3

5:00

07:4

0:00

07:4

5:00

07:5

0:00

07:5

5:00

08:0

0:00

08:0

5:00

08:1

0:00

08:1

5:00

08:2

0:00

08:2

5:00

08:3

0:00

08:3

5:00

08:4

0:00

08:4

5:00

08:5

0:00

08:5

5:00

09:0

0:00

Time (sec.)

Spee

d (k

m/h

)

LL ML RL

Speed (km/h) for three lanes (after study 07.00-09.00)

0

20

40

60

80

100

120

07:0

0:00

07:0

5:00

07:1

0:00

07:1

5:00

07:2

0:00

07:2

5:00

07:3

0:00

07:3

5:00

07:4

0:00

07:4

5:00

07:5

0:00

07:5

5:00

08:0

0:00

08:0

5:00

08:1

0:00

08:1

5:00

08:2

0:00

08:2

5:00

08:3

0:00

08:3

5:00

08:4

0:00

08:4

5:00

08:5

0:00

08:5

5:00

09:0

0:00

Time (Sec.)

Spee

d (k

m/h

)

VMS LL ML RL


Figure 11 shows headway distribution and Figure 12 speed distributions during periods “before” and “after” MCS operation with data from the same survey periods as the data in figures 9 and 10.

Figure 11, Headway distribution for three lanes before and after implementation of MCS.

Figure 12, Speed distribution for three lanes before and after implementation of MCS. From Figure 11 it can be seen that the frequency of very short headways was considerably higher in all lanes before implementation of MCS than after. This phenomenon is closely related to the instability of the traffic stream, particularly for the left lane where congestion usually set in first when MCS is not in operation (Smulders, S. 1990).

Headway distribution before (07:55-08:10) and after MCS control (07:15-07:30)

0

50

100

150

200

250

300

0 1 2 3 4 5 6 7 8 9 10

Headway (Sec.)

Freq

uenc

y LL With controlML With controlRL With controlLL No MCS controlML No MCS controlRL no control

Speed distribution before (07:55-08:10) and after MCS control (07:15-

07:30)

0

50

100

150

200

250

300

350

400

0 10 20 30 40 50 60 70 80 90 100 110 120 130

Speed (km/h)

Freq

uenc

y LL With MCS controlML With MCS controlRL With MCS controlLL No MCS controlML No MCS controlRL no control


3.4 Driver behaviour The mobile data collection for study of driver behaviour focused on the impact of MCS recommended speed signalling triggered by downstream queues. The deceleration process and speed adaptation when approaching a slow-moving queue was studied using three different methods:

1. Study from the moment the driver noticed the end of the queue (controlled from video recording from instrumented vehicle).

2. Study from the moment the driver started braking (automatically registered by the instrumented vehicle).

3. Study from the moment when the driver noticed the first MCS sign until he/she slowed down to approach the end of the queue.

All three methods considered deceleration while the third method also considered speed adaptation before the drivers approached the queue. The data reduction process included segmentation of the observations as follows

- based on whether the MCS had or had not been activated; - based on if the traffic situation was free-flowing or congested; - based on the initial speed in of the vehicles in three groups, under 50

km/h, between 50-70 km/h and over70 km/h. Most drivers approached a congested road section exceeding the speed limit, irrespective of whether the MCS displayed a lower recommended speed or not. As the driver approaches the end of the queue, traffic grows denser which makes it difficult to maintain a high speed. By the time the end of the queue has been noticed, the speed has decreased to a level, which is between the speed limit and the speed recommended by MCS. Figure 13 illustrates a commonly recurring deceleration processes for drivers in restrained traffic conditions with a lower speed than the recommended by MCS. Figure 13. Illustration of typical deceleration process during restrained traffic conditions when the MCS recommended speed signs are active

Deceleration during restrained flow condition with active MCS speed signs

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

0 49 96 141

185

228

273

315

356

398

440

483

527

571

612

652

692

731

768

804

841

878

914

951

986

1021

1061

1099

1136

1174

1215

1255

1296

1339

1382

1423

1463

1494

1518

1535

1542

Stretch length (m)

Spee

d (k

m/h

)

Speed km/h Speed LimitMCS recommended speed V. brake pressure [Bar]Point when driver notice and approach the queue


Speed adaptation during free-flow conditions The driver behaviour during free-flow conditions was similar as during restrained traffic flow considering speed levels higher than the speed limit before approaching the end of the queue, even though MCS displayed lower recommend speed. Figure 14 shows a typical deceleration pattern during free-flow conditions from the point where the driver first spots the MCS recommended speed until the completion of the deceleration. Figure 14. Illustration of deceleration process during free-flow conditions with active MCS speed signs.

4. ANALYSIS OF MCS IMPACTS

4.1 Impacts on Traffic Conditions The impacts of MCS can be looked at statistically with analysis of variance methodology, particularly for speed and headway distributions during different traffic flow conditions (Rämä, Pirkko., 2001). The purpose was to test whether observed differences in means speeds and headways could be explained by the natural sampling variability, or whether it could be attributed to the impact of MCS speed signalling. Results of this analysis are exemplified in Table 3 below for time periods with or without MCS with similar traffic flow.

Deceleration from free-flow condition with active MCS speed signs

0.0

10.020.0

30.040.0

50.0

60.070.0

80.090.0

100.0

0 36 73 110

146

182

217

252

288

323

357

392

426

461

496

532

569

605

641

675

709

743

774

802

826

848

866

880

894

907

919

930

940

948

955

960

964

Stretch length (m)

Sp

eed

(km

/h)

Speed km/h Speed LimitRecommended speed (MCS) V. brake pressure [Bar]Point when the driver notice and approach the queue


Table 3, Comparison of average speed and headway before and after

activation of MCS. The hypothesis that MCS had no effect on speed was tested. The same test was made for the impacts of MCS on headways. Each analysis was undertaken for high as well as low traffic conditions. The standard deviation of speed was also studied as suggested as an indicator for traffic safety and traffic condition homogeneity by Solomon (1994) and Baruya (1998). The results can be summarized as follows: Impacts of MCS during periods of high traffic flow: (upper part of Table 3)

- The average speed increased significantly in the left and right lanes. - The standard deviation for speed in each lane was reduced (30-40%). - The average headway increased significantly in the right lane. - The standard deviation for the headway decreased only in the left lane.

Impacts of MCS during periods of low traffic flow: (lower part of Table 3) - The results show small differences for speed and headway averages and

standard deviations with or without MCS in operation. The most likely reason for this is that there were very few time periods with activated recommended speed signs during low traffic conditions.

4.2 Impacts on Driver Behaviour In general analysis of MCS impacts on individual driver behaviour is complex due to the difficulty to find comparable traffic conditions during periods with or without active MCS. There is normally more congestion in the former case. However, the following observations could be made based on the study.

Low traffic flow, 11:00-11:15 (same period for before and after (MCS)

Before After Average std. dev. Average std. dev. Average std. dev. Average std. dev. LL 173 347 113.20 9.46 114.20 9.68 0.26 5.61 6.16 5.52 6.48 0.88ML 342 347 97.21 8.00 99.66 8.55 0.00 2.77 2.10 2.87 2.05 0.54RL 297 347 85.91 6.90 86.76 7.67 0.14 3.23 2.36 2.77 2.15 0.01

Lane Traffic flow Speed (km/h) Headway (Sec.)Before After Before After Sig

(95%)Sig

(95%

High traffic flow, 07:40-07:55 before (MCS) 07:00-07:15 after (MCS)

Before After Average std. dev. Average std. dev. Average std. dev. Average std. dev. LL 448 444 42.05 12.00 54.05 7.50 0.00 2.10 0.97 2.02 0.83 0.06ML 410 376 41.00 9.13 45.62 5.06 0.00 2.35 1.14 2.39 1.25 0.63RL 406 349 41.89 9.92 45.96 5.23 0.00 2.38 1.09 2.56 1.29 0.00

Lane Traffic flow Speed (km/h) Headway (Sec.)Before After Before After Sig

(95%)Sig

(95%)


Speed adaptation during free-flow conditions Most drivers did not follow the recommended speed signalled by MCS, normally they would drive much faster. Slightly lower speeds were observed with active MCS signals, but it could not be established whether this was caused by denser traffic downstream or by the speed signals themselves. Speed reduction occurred when the traffic became so dense that this became a necessity. Deceleration process when approaching a queue More in-depth studies of MCS impacts on driver behaviour in this regard are needed. However from the present study it was observed that MCS speed signs seemed to make drivers more cautious with regard to the possibility that there would be a queue ahead. Drivers who had been warned by active MCS had a significantly higher but more even deceleration, which increased closer to the queue. Drivers approaching a queue without pre-warning from active MCS speed signs both accelerated and decelerated before they arrived at the end of the queue, see Figure 15.

Figure 15. Deceleration process in retrained traffic conditions with and without active MCS recommended speed signs, (Szabo, J. 2006, unpublished KTH M.Sc. thesis) The braking distance was also longer in the “before” MCS case with a range between 50-450m as compared to 50-250 m for the “after” MCS case. A questionnaire regarding the car following, showed that 45% of the drivers adapted their speed to the MCS recommended speed-whilst the speed choice of the other 47% was influenced by the vehicle in front (Ax and Werner, 2004).

Influence of MCS on deceleration

-0,80 -0,60 -0,40 -0,20 0,00 0,20 0,40

Distance to queue (m)

m/s2

MCS activet No MCS

MCS active 0,02 -0,06 0,00 -0,06 -0,05 -0,14 -0,32 -0,68 No MCS 0,11 0,16 0,05 -0,24 -0,13 -0,15 -0,06 -0,34

400-350 350-300 300-250 250-200 200-150 150-100 100-50 50-0


5. CONCLUSIONS Traffic flow characteristics for comparable time periods were analysed before and after implementation of MCS with recommended speed sign control. The main MCS impact consisted of a sharp decrease of the standard deviation of speed, which would indicate an improvement in homogeneity and traffic safety. Furthermore MCS reduced the frequency of lane changes between the middle and the left lane. Driver behaviour impacts were difficult to assess due to the inability to perform controlled experiments with or without active MCS for comparable situations. Most drivers seemed to ignore the recommended speed signs, which often did not correspond to what could be maintained during the actual traffic situation. However, there was an indication that active speed signs warning for a queue downstream influenced the deceleration process to become more stable. REFERENCES Alessandri, A., Di Febbraro, A., Ferrara, A., and Punta, E. (1997), "Optimal control of freeways via speed signalling and rampmetering,"in Preprints 8th IFAC/IFORS Symp. Transportation Systems 1997, pp.1066-1071. Archer, J. (2003), User Manual Semi-Automatic Video Analyser (SAVA) Version 4.4. Unpub-lished User Manual, Royal Institute of Technology, Department of Infrastructure, Division of Traffic and Logistics. Ax, J., Werner, D (2004), Driver behaviour impacts of Motorway Control Systems with recommended Speed Signalling. Royal Institute of Technology, Department of Infrastructure, Division of Traffic and Logistics. TRITA-INFRA EX 04-050. Stockholm Cooper, B. R. and Sawyer, H. (1993), "Assessment of M25 automatic fog-warning system". Crowthorne: Transport Research Laboratory. (Final report, Project Report 16). Cremer, M. (1978), "Astate feedback approach to freeway traffic control," in Preprints 7th IFC World Congress, 1978, pp.1575-1582. Garber, N. J. and Gadiraju, R. (1989), "Factors Affecting Speed Variance and Its Influence on Accidents." Transportation Research Record 1213, pp 64-71. Harbord, B. (1997), "Success over 'stop-start'- Results of thr M25 Controlled Motorway pilot, Traffic Technology international, 57-60. Hegyi, A. (2004), "Model Predictive Control for Integrating Traffic Control Measures," TRAIL Thesis Se T2004/2riesThe Netherlands TRAIL Research School ISBN 90-5584-053-X.


Hegyi, A., De Schutter, B., and Hellendoorn, H. (2003), "MPC-based optimal coordination of variable speed limits to suppress shock waves in freeway traffic," in proceedings of the American Control Conference, Denver, Colorado , USA, June 4-6 2003. Hegyi, A., De Schutter, B., and Hellendoorn, H. (2003), "Optimal coordination of variable speed limits to suppress shock waves," Tech.res. CSE02-015, Control Systems Engineering, Fac. of Information Technology and systems, Delft University of Technology, Delft, The Netherlands, Aug.2003. Accepted for publication in Transportation Research Record. Hegyi, A., De Schutter, B., Hellendoorn, H., and Van den Boom, T. (2002), "Optimal Coordination of Ramp Metering and Variable Speed Control- An MPC Approach,"in proceedings of the 2002 American Control Conference, Anchorage, Alaska, pp. 3600-3605, May 2002. Hogema, J. H. and Van der Horst, R. (1997), "Evaluation of A16 motorway fog-signalling system with respect to driving behaviour". Transportation Research Record, 1573, pp. 63.67. Kühne, R. D. (1991), "Freeway control using a dynamic traffic flow model and vehicle reidentification techniques," Transportation Research Record, vol.1320, pp. 251-259, 1991. Lin, P. W., Kang, K. P., and Chang, G. L. (2004), "Exploring the Effectiveness of Variable Speed Limit Controls on Highway Work-Zone Operations" ISSN:1547-2450 print/1547-2442 online. Ober-Sudermeier, A. and Zackor, H. (2001), "Prediction of Congestion due to Road Works on Freeways." IEEE Intelligent Transportation Systems Conference Proceedings_Oakland (CA),USA- August 25-29, 2001. Remeijn, H. (1982), "The Dutch Motorway Control and Signalling System". Traffic Engineering Division, Dutch Ministry of Transport, The Hague. Rämä, P. (2001), Effects of weather-controlled variable message signing on driver behaviour Helsinki University of Technology (Otakaari 4, Espoo, Finland). Smulders, S. (1992), "Control by variable speed signs- the Dutch experiment," in proceedings of the Sixth International conference on Road Traffic Monitoring and Control, IEE Conference Publication, London,pp.99-103,IEE,april.28-30 1992. Smulders, S. (1990), Control of freeway traffic flow by variable speed signs. Transportation Research, B, vol.24, pp.111-132,1990 or vol. 25, pp. 117-122.


Smulders, S. (1996), "Control of Freeway Traffic Flow. CWI Tract no.80,CWI (Dutch institute for research in Mathematics and computer Scince), 1996.ISBN:90-6196-451-2. Smulders, S. and Helleman, D. E. (1998), " Variable speed control: State- of- the art and synthesis," in Road Transport Information and Control, no. 454 in Conference Publication, pp. 155-159, IEE, Apr. 21-23 1998. Van den Hoogen, E. and Smulders, S. (1994), "Control by variable speed signs: results of the Dutch experement." In proceedings of the 7th International Conference on Road and Traffic Monitoring Control. London: the Institute of Electrical Engineer IEE. pp. 145-149. Wilkie, J. K. (1997), "Using variable speed limit signs to mitigate speed differentials uppstream of reduced flow locations," Tech. rep., Department of Civil Engineering, Texas A&M University, College Station, Texas 77845,Aug. 1997.Prepared for CVEN 677 Advanced Surface Transportation Systems. Zackor, H. (1979), "Self-suffecient control of speed on freeways," in Proceedings of the International Symposium on Traffic Control Systems, vol.2A,Berkeley, Califorinia, pp.226-249,Califorinia University,Aug.6-9 1979.

Evaluation of MCS impacts - MITweb.mit.edu/11.951/oldstuff/albacete/Other_Documents/Europe Transport...

Documents

Transcript of Evaluation of MCS impacts - MITweb.mit.edu/11.951/oldstuff/albacete/Other_Documents/Europe Transport...