SM-ALR Monitoring M25 J5-7 Twelve Month Evaluation Report Highways England. January 2016

Transcription

1 SM-ALR Monitoring M25 J5-7 Twelve Month Evaluation Report Highways England January 2016

2 Notice This document and its contents have been prepared and are intended solely for Highways England s information and use in relation to the SM-ALR Monitoring Project. Atkins assumes no responsibility to any other party in respect of or arising out of or in connection with this document and/or its contents. This document has 49 pages including the cover. Document history Job number: Document ref: /06/09 Version Purpose description Originated Checked Reviewed Authorised Date Ver 1.0 Initial Issue Joe Castle Andrew Truscott Ver 2.0 Pre-publication checks Joanna Goulding Freda Rashdi Joanna Goulding Jill Hayden Joe Castle 10/12/15 Mike Wilson 08/01/16

3 Table of contents Chapter Pages Executive summary 5 1. Introduction Scope of Project and Purpose of This Report Background of the Scheme Evaluation Timescales Expected Effects of SM-ALR 9 2. Flows Introduction Daily Flows per Link Flow over Each Time Slice per Link Long Vehicles Percentage Tests for Statistical Significance of the Results Summary Journey Time Analysis Introduction Average Journey Time Journey Time Reliability Results of Analysis of Filtered Data Vehicle Hours Delay Summary Safety Introduction Severity and Severity Index Casualties, FWI and KSI Rate Additional Analysis Summary Conclusions Flows Journey Times Safety 28 Appendices 29 Appendix A. Flow Additional Information 30 A Hour Average Daily Traffic (ADT) 30 A.2. Monthly Average Daily Traffic (ADT) 30 A.3. Flows by Time Slice 31 A.4. Long Vehicles Percentage 32 Appendix B. Journey Time Additional Information 33 B.1. Filtered and Unfiltered Days 33 B.2. Unfiltered Journey Time Data 33 B.3. Analysis of Filtered Journey Time Data 37 B.4. Vehicle Hours Delay 41 Appendix C. Safety Additional Information 42 C.1. Contributory Factors 42 C.2. Red X Compliance 48

4 Tables Table 1-1 Day Type and Time Slice Definitions... 9 Table 2-1 T Tests for Results by Time Slice Table 2-2 T Tests for ADT Results Table 3-1 Journey Time Metrics Table 4-1 Number of Collisions by Severity and Collision Rates Table 4-2 Number of Collisions and Collision Rates Following National Trends Table 4-3 Collisions by Severity and Severity Index Table 4-4 Number of Casualties and FWI Rate Table 4-5 Total KSI and KSI Rate Table 4-6 Collisions Occurring in Daylight and Darkness Table 4-7 Summary of ERA Activity Table 4-8 Vehicle Types Using ERAs Figures Figure 1-1 Geographical Location of the M25 Section 5 SM-ALR Scheme... 8 Figure 1-2 M25 J5 to J7 Layout Schematic... 8 Figure 1-3 Data Collection & Evaluation Periods... 9 Figure 1-4 Speed by Lane Before and After Figure 2-1 Average Daily Traffic by Day Type J5 - J6 Clockwise (Lane Increase) Figure 2-2 Average Daily Traffic by Day Type J6 - J7 Clockwise Figure 2-3 Average Daily Traffic by Day Type J5 - J6 Anticlockwise (Lane Increase) Figure 2-4 Average Daily Traffic by Day Type J6 - J7 Anticlockwise Figure 2-5 Average Flow by Time Slice J5-J6 Clockwise Figure 2-6 Average Flow by Time Slice J6-J7 Clockwise Figure 2-7 Average Flow by Time Slice J5-J6 Anticlockwise Figure 2-8 Average Flow by Time Slice J6-J7 Anticlockwise Figure 3-1 Clockwise Journey Time Comparison - Unfiltered Sample Figure 3-2 Anticlockwise Journey Time Comparison - Unfiltered Sample Figure 3-3 Clockwise Journey Time Reliability Analysis (Unfiltered) Figure 3-4 Anticlockwise Journey Time Reliability Analysis (Unfiltered) Figure 4-1 Example Lane Closure Event... 25

5 Executive summary Background Smart motorways are central to the modernisation of England s motorways and have been designed to reduce congestion, improve journey time reliability and lead to shorter journey times, while at the same time maintaining safety. SM-ALR is part of the key strategic route around London SM-ALR Objectives M25 Section J5 to J7, is part of the key strategic orbital route around London which forms the hub of the English motorway network and also serves as a commuter route for local traffic. It is within the counties of Surrey and Kent and located in the southern segment of the M25. It starts at J5 which is the intersection with the M26, A21 and A25 and finishes at J7; the intersection with the M23. This is an interim report presenting results following 12 months of After evaluation from May 2014 to April 2015 there will be further reports at year 2 and year 3. The report is split into sections to cover each of the following objectives of Smart Motorway All Lane Running (SM-ALR): flows; journey times; safety; and hazards. Traffic flow increased more than / equal to regional trends HGV proportion increased slightly Flows For J5 to J6, the SM-ALR section where a lane has been added, flows have increased by 13% clockwise and 3% anticlockwise. The scheme has experienced traffic growth of approximately 2% between J6 and J7, which was not previously at capacity and did not have any lanes added; this is in line with regional growth trends. The percentage of long vehicles has increased across all time slices, by 3 percentage points for weekdays but only 1 for weekends. Overall journey times have reduced AM peak clockwise and PM peak anticlockwise account for benefits Journey time reliability is better 1,680 VHD saved daily Improved resilience Journey Times Average journey times have reduced by 3% overall in the clockwise direction and 2% anticlockwise. The biggest improvements have occurred in the time periods which were congested in the Before period (AM peaks clockwise and PM peaks anticlockwise); for example average journey time reduction of 1 minute 40 seconds for Friday PM anticlockwise. Some of the previously uncongested periods, particularly on J6 to J7 clockwise, have experienced slight increases (a few seconds). These increases are all negligible compared to the benefits at other times and it is likely that they stem from improved speed limit compliance. The biggest journey time reliability improvements relate to the times when journey times were most unreliable in the Before period, i.e. the clockwise AM peaks and the anticlockwise PM peaks. At less congested times the journey time reliability has not significantly changed. Vehicle Hours Delay (VHD) has reduced by a quarter across the scheme with 550 hours saved daily in the clockwise direction and 1,130 anticlockwise. The improvements in average journey times and journey time reliability during previously congested periods indicate that the SM-ALR scheme is successful in providing better service to customers, being more reliable and resilient.

6 Safety is not worsened 7% of vehicles did not comply with Red X 81% of ERA stops not an emergency Safety There is a small (but not statistically significant) reduction in collision rate, over and above the national background of improved safety. While the reduction is not significant, the results provide an initial indication that safety has not worsened as a result of the scheme. Monitoring of Red X compliance revealed that across all events analysed, an average of 7% of vehicles were non-compliant. Approximately 0.3 Emergency Refuge Area (ERA) stops per hour per ERA were observed during monitoring of CCTV. The majority of these were non-emergency with HGVs and cars representing the highest non-emergency use of ERAs. During the monitoring one hard shoulder misuse event was observed.

7 1. Introduction 1.1. Scope of Project and Purpose of This Report Highways England has commissioned a project to monitor and evaluate the impact of the first SM-ALR scheme, the M25 between Junction 5 and Junction 7. It is crucial that the performance of the scheme is accurately assessed in order to: review the safety performance and ensure that the core assumptions for Smart Motorways were robust; better understand the change in risk to road users and to road workers and ensure objectives were met; quantify and provide evidence of the benefits of the concept to inform future schemes; and provide evidence to help improve the concept of operation and the design requirements. As part of the SM-ALR Monitoring project, an evaluation methodology was designed and approved by Highways England Traffic Appraisal, Modelling and Economics (TAME); this is described in Deliverable 01: SM-ALR Monitoring Design Report. Data has been collected and analysed for the Before period in accordance with the Monitoring Design Report and is described in Deliverable 02: Monitoring Baseline Report. This report presents results following 12 months of After evaluation from May 2014 to April The report is split into sections to cover each of the following objectives of SM-ALR: flows; journey times; safety; In addition to comparisons against the Before period, monitoring has taken place including compliance with Red X, ERA usage, hard shoulder misuse and speed limit compliance Background of the Scheme Location M25 Section J5 to J7, is part of the key strategic orbital route around London which forms the hub of the English motorway network and also serves as a commuter route for local traffic. It is within the counties of Surrey and Kent and located in the southern segment of the M25. It starts at J5 which is the intersection with the M26, A21 and A25 and finishes at J7; the intersection with the M23. The majority of the M25 is Smart Motorway with hard shoulders, which together with the SM-ALR scheme, form an overall long term strategy to manage the existing motorway network more effectively.

8 Figure 1-1 Geographical Location of the M25 Section 5 SM-ALR Scheme The SM-ALR Scheme SM-ALR is a controlled four lane carriageway with no hard shoulder. This is supported by technology in the form of Motorway Incident Detection and Automatic Signalling (MIDAS) traffic detection and traffic control. The signs and signals can be controlled by operators and by automatic algorithms for Congestion Management (CM) and Queue Protection (QP). ERAs are available for broken down vehicles. The M25 J5 to J7 SM-ALR is a mixture of 4 lane ALR and 4 lanes plus hard shoulder, see Figure 1-2. It has been changed from the previous layout which was a mixture of 4 lanes plus hard shoulder and 3 lanes plus Hard Shoulder. As part of the upgrade to Smart Motorway, radar detectors were installed at 500m intervals from J5 to J6. Loop detectors were retained from J6 to J7. Figure 1-2 M25 J5 to J7 Layout Schematic

9 1.3. Evaluation Timescales Figure 1-3 shows the evaluation periods used for the Before and After periods. Figure 1-3 Data Collection & Evaluation Periods For the analysis of flows and journey times it is useful to consider the results separately for different day types and time slices. This is because the traffic conditions are different and therefore so are the impacts. Table 1-1 shows the time slices and day types used for the flow and journey time analysis, in accordance with the Monitoring Design Report. Table 1-1 Day Type and Time Slice Definitions Day Type AM Peak Inter-peak PM Peak Monday Thursday 05:30 10:30 10:30 15:00 15:00 20:00 Friday 05:00 09:00 09:00 13:00 13:00 20:00 Saturday - Sunday 08:00 20: Expected Effects of SM-ALR The SM-ALR concept involves increasing the number of running lanes from three to four by re-allocating the space previously used by the hard shoulder. In addition, other infrastructure is provided to mitigate the potential negative impact associated with hard shoulder removal. The effect of an increase in capacity is that periods of congestion are expected to be less frequent, shorter and less intense leading to reductions in journey time and better journey time reliability. The road effectively becomes more resilient to regular and incident related congestion. In addition safety benefits could be realised, not least because of the number of unnecessary hard shoulder stops avoided but also because traffic speeds become more consistent and the speed differential between lanes reduces. The effect on speeds has been particularly positive as demonstrated by Figure 1-4, which shows a snapshot of data from Before and After collected during the evaluation process. It can be seen that in the Before period, there was significant congestion in the AM and inter peaks. There was a speed differential of approximately 12mph between lanes. The After snapshot shows no congestion in the peaks, and the speed differential between lanes is also lower, in the order of 6mph.

10 Figure 1-4 Speed by Lane Before and After M25/4291A Fri Before M25/4290A Fri After

11 2. Flows 2.1. Introduction This section presents the traffic flow analysis for the first year of operation of the SM-ALR scheme between J5 and J7 of the M25. The traffic data has been taken from Highways England s Traffic Database (TRADS 1 ). A full year of data for the Before and After periods has been used for the analysis 2. The results include comparisons of daily flows, flows per day type and time period, and maximum 15 minute flows; also seasonal variations and percentage of long vehicles in the vehicle fleet Data Availability and Quality This scheme is one of the first to use radar detectors as an alternative to the inductive loops that have traditionally been used. The radar technology has been found to provide less accurate results than might be expected, with inconsistencies between the traffic counts of adjacent detector locations. Unfortunately this means that between J5 and J6, loop-based data in the Before period is being compared with radar-based data in the After period, with no certainty that they are directly comparable. In fact there is evidence (from comparing loop flows on the adjacent section in conjunction with off- and on-slip loop flows) that J5 to J6 radar flows are being under-reported by up to 6%. However because the radar units are under reporting traffic flow the positive impact of the scheme is also therefore understated. Analysis across other data sources has identified no evidence that other results presented in this report are affected. This issue is under further investigation and work is underway to improve the data provided by the radar detectors. For the purposes of this report, to overcome the issue the flows for J5 to J6 have been calculated using the J6 to J7 flows and the J6 slip road flows. This means that the flow results can be viewed with more confidence, but it has limited the analysis that could be performed for this section. However since a second and third year of evaluation is planned as part of this project additional analysis of the whole scheme length can be undertaken. Due to a lack of data in February and March between J6 and J7 clockwise, it was not possible to retain the use of the same site used in previous analysis. This meant the data for that section had to be supplemented and infilled with data for an alternative site. As this is not in the section using radar, the sites are reasonably consistent with one another for previous months and so it is expected that the alternative is a reasonable substitute for the original. In addition, TRADS data was not available in April for the whole section as a result of the migration of the TRADS data to the DfT website. MIDAS 3 data has been aggregated appropriately to allow the application of the same processing as was used for the preceding months. As the TRADS data is constructed from MIDAS data anyway, it was considered acceptable to carry out this process, noting the restrictions mentioned above on the reliability of the data Daily Flows per Link The average daily traffic for the Before and After periods is compared in Figure 2-1 to Figure 2-4, with the 24 hour Average Daily Traffic (ADT) flows between each junction plotted for the different day types. The percentage change is shown above the After bar in each case. The corresponding values are shown in Appendix A.15.3.A.1. Appendix A.2 shows the monthly ADTs; there has not been any significant change to the seasonal trend of traffic flows; lower flows in the winter months are consistent with the national picture. Clockwise, there has been a 13% increase in ADT between J5 and J6, where the number of lanes has increased from 3 to 4 as part of the scheme. The only real increase between J6 and J7, which was already 4 lanes, is at the weekend. Before the scheme, J5 to J6 flows were far lower than J6 to J7 but now the flows on both links are similar (around 70,000 ADT) The Christmas, Boxing Day, New Year and other bank holidays have been removed from the data used in both Before and After periods. 3

12 Figure 2-1 Average Daily Traffic by Day Type J5 - J6 Clockwise (Lane Increase) Figure 2-2 Average Daily Traffic by Day Type J6 - J7 Clockwise Anticlockwise (following page), both links have seen a similar increase in flows (around 2% for weekdays and 5% for weekends). This means that J5 to J6 flows were previously lower than J6 to J7 and are still lower, despite the increase from 3 to 4 lanes on this link. A large number of vehicles leave the motorway at J6 anticlockwise which explains why the J5 to J6 flows are lower. It could be that additional capacity is not yet required on this link; it will be useful to continue to monitor flows here to see whether they start to increase more significantly.

13 Figure 2-3 Average Daily Traffic by Day Type J5 - J6 Anticlockwise (Lane Increase) Figure 2-4 Average Daily Traffic by Day Type J6 - J7 Anticlockwise Overall, ADTs have increased by 2% on J6 to J7 where the number of lanes was unchanged. There has been a 13% increase on J5 to J6 clockwise, but only 3% anticlockwise. In comparison, national motorway flows have increased by 3.5% in the same period. All-roads flows have increased by 3.3% nationally and by 2.1% in the South East 4. So the growth in the scheme is roughly in line with regional trends, apart from J5 to J6 clockwise where it is significantly higher as a result of the additional capacity provided. 4

14 2.3. Flow over Each Time Slice per Link Figure 2-5 to Figure 2-8 compare the average flow Before and After in each time slice for each link. The percentage change is shown above the After bar in each case. The corresponding values are shown in Appendix A.3. Clockwise, there have been significant increases in all periods between J5 and J6 where the number of lanes has increased. However PM peaks flows have reduced between J6 and J7, which is unexpected. Figure 2-5 Average Flow by Time Slice J5-J6 Clockwise Figure 2-6 Average Flow by Time Slice J6-J7 Clockwise

15 Anticlockwise, there appear to have been small increases in some weekday time slices between J5 and J6 where the number of lanes has increased. There have been no real changes on weekdays between J6 and J7. Weekend flows have increased on both links. Figure 2-7 Average Flow by Time Slice J5-J6 Anticlockwise Figure 2-8 Average Flow by Time Slice J6-J7 Anticlockwise 2.4. Long Vehicles Percentage The percentage of long vehicles by time slice for J6-J7 anticlockwise is shown in Appendix A.4. There has been a slight increase in the percentage of long vehicles.

16 2.5. Tests for Statistical Significance of the Results A sequence of t tests was performed in order to assess whether the changes in the flows measured has been significant. In order to carry out these t tests, the standard deviation, average flow and number of observations were calculated for each link in the section. An observation was considered to be a full day of data for each site on the link which was considered. This was required as the data has had to be collected and averaged for a number of sites in order to address the variability in the observations. The t tests assumed that there was no change in the flow and tested for a statistically significant change using a twotailed test at a 95% confidence level. Unfortunately the method used to calculate the J5 to J6 flows means the t tests could not be performed for this link. The results for J6 to J7 are presented in Table 2-1 for the flows by time slice and in Table 2-2 for the ADTs; a series of arrows denote the type of change experienced in that time period. An up arrow denotes a statistically significant increase in flow, a down arrow denotes a statistically significant decrease, and a dash denotes no significant change in the time period. Table 2-1 T Tests for Results by Time Slice Direction Location Mon-Thurs Friday AM Peak PM Peak AM Peak Interpeak Interpeak PM Peak Saturday -Sunday Clockwise J6 - J7 Anti-Clockwise J6 - J7 Table 2-2 T Tests for ADT Results Location Value Mon-Thurs Friday Sat-Sun ADT Clockwise J6 - J7 Anti-Clockwise J6 - J7 There was no significant change in flows in most time slices for J6 J7, which is not unexpected as there was no increase in lanes on this link. The ADT has increased in both directions, suggesting that there has been an overall increase; from this analysis it is clear that this is driven by weekend traffic Summary For J5 to J6, the SM-ALR section where a lane has been added, flows have increased by 13% clockwise and 3% anticlockwise. The scheme has experienced traffic growth of approximately 2% between J6 and J7, which was not previously at capacity and did not have any lanes added; this is in line with regional growth trends. The percentage of long vehicles has increased across all time slices, by 3 percentage points for weekdays but only 1 for weekends.

17 3. Journey Time Analysis 3.1. Introduction This section outlines the changes in journey times and reliability on the M25 J5-7 SM-ALR between the Before and After periods. The data used has been taken from TomTom satellite navigation devices, which provide anonymised data of journeys through the scheme during the Before and After periods. The journey time data is at a very spatially disaggregate level, allowing speed analysis to be undertaken at regular intervals along the scheme. Before interrogating the TomTom database two analyses were undertaken in order to define the data that should be extracted: A review of severe incidents and road works was undertaken to identify any days that should be removed from the analysis because they would not represent normal operating conditions. A review of the journey time database (JTDB 5 ) data was used to identify days with a high standard deviation of journey times (i.e. days with large variability in journey times), again, to identify days where operating conditions were atypical. With the above analysis complete, two forms of data were extracted from the TomTom database, which will be referred to as unfiltered and filtered 6 : Unfiltered data is data with just the road works periods and severe incidents removed. Other than these events, all days within the year are included in the dataset. The unfiltered data gives a view of the average performance of the scheme before and after the scheme. Filtered data has the key incidents and maintenance days removed, but also days identified with a high standard deviation of journey time. The filtered data is therefore more indicative of what occurs on a day with less demand or incident related congestion. Essentially, abnormal events such as high demand for sports events or incidents that have caused atypical delay are removed from the filtered data. The unfiltered journey time data is a more accurate representation of real life performance than the filtered data. These results are presented in this section and form the main conclusions on the performance of the scheme. The filtered journey time data provides information about scheme performance on those days which are not affected by events or incidents (only around 60% of days on this very busy stretch of motorway) and the results of that analysis are contained in Appendix B Average Journey Time The analysis of average journey times from junction to junction demonstrates the change in journey times on a link level. The results of the analysis are summarised visually in Figure 3-1 for clockwise and Figure 3-2 for anticlockwise. Detailed results are provided in Appendix B.2.1. Clockwise, only the weekday AM and Friday inter-peak periods suffered from congestion in the Before period (shown by journey times in excess of the free-flow reference journey time 7 of just over 11 minutes). There has been an overall journey time reduction of 3%, and almost all periods demonstrate a saving, highest in the previously congested periods (e.g. being 9% or 1 minute 20 seconds for Friday AM peak). The Mon-Thu inter-peak experienced a 2% increase (13 seconds longer in the After period) and Mon-Thu PM peak has no significant change; neither of these were congested previously. The overall findings are positive but there is a strong contrast between J5 to J6 and J6 to J7. The longer SM-ALR link, J5 to J6, experiences savings in almost all time slices (between 2 and 80 seconds, average 5% reduction). However the short J6 to J7 link shows slight increases in almost all time slices (between 3 and 14 seconds, average 3% increase) The filtered and unfiltered date ranges used are shown in Appendix B.1. 7 Based on a free-flow reference speed of 67mph

18 Figure 3-1 Clockwise Journey Time Comparison - Unfiltered Sample Figure 3-2 Anticlockwise Journey Time Comparison - Unfiltered Sample

19 Anticlockwise, there has been an overall journey time reduction of 2%. Only the PM peak time slices were congested in the Before period and these experienced large benefits, 6% (50 seconds) for Mon-Thu and 11% (1 minute 40 seconds) on Friday. Other periods experienced slight reductions or slight increases, with the changes being less than 16 seconds. Both links experienced overall journey time reductions anticlockwise. In summary, the main benefits are in the AM peak clockwise and the PM peak anticlockwise, tying in with the Before period delays and indicating tidal commuter traffic. Some of the previously uncongested periods have experienced slight increases. These increases are all negligible compared to the benefits at other times and it is likely that they stem from improved speed limit compliance Journey Time Reliability Reliability of journey times is a critical measure of a roads utility and function for road users. Percentile data has been used to understand the distribution of journey times through the scheme. Four metrics have been used, as shown in Table 3-1. Table 3-1 Journey Time Metrics Metric Description 5 th Percentile One in 20 vehicles are completing the journey faster than this, so it is a good measure of the best time achievable. 25 th Percentile One in four vehicles are completing the journey faster than this and it is known as the lower quartile. The further this value from 5th percentile the more variability in the fastest journeys, it is an indicator that delays are experienced by a high proportion of all users 75 th Percentile Three quarters of vehicles complete the journey faster than this and it is a good measure of general variability from day to day of in journey times. 95 th Percentile 95% of vehicles complete the journey faster than this, the remaining journeys are likely to be affected by incidents or heavy congestion. The further the 95th percentile journey time is from the 75 th percentile the more heavily congested a journey is, this is an indication of incident related variability. These four metrics are shown below in Figure 3-3 and Figure 3-4 as box and whisker diagrams for each time slice, Before and After. The box contains the 25 th to 75 th percentile range and the whiskers show the 5 th and 95 th percentile values. The 75 th percentile and 95 th percentile journey times are annotated on the plots. Clockwise, the most unreliable journey times (Before and After) are in the AM peaks, Friday inter-peak and to a lesser extent at weekends. These periods have all seen an improvement in journey time reliability, particularly Friday AM with a 6 minute reduction in 95 th percentile journey time. The uncongested time slices have similar results Before and After. Anticlockwise, the most unreliable journey times (Before and After) are in the weekday PM peaks. These periods have experienced an improvement in reliability, e.g. 95 th percentile journey times reduced by 4.5 minutes Mon-Thu and 7.5 minutes on Friday. The uncongested time slices have similar results Before and After.

20 Figure 3-3 Clockwise Journey Time Reliability Analysis (Unfiltered) Figure 3-4 Anticlockwise Journey Time Reliability Analysis (Unfiltered) These results indicate that in previously congested periods, journey time reliability has improved meaning that the route is now more resilient with all lane running.

21 3.4. Results of Analysis of Filtered Data Filtered data has the key incidents and maintenance days removed, but also days identified with a high a standard deviation of journey time. The filtered data is therefore more indicative of what occurs on a day with less demand or incident related congestion. More detail about the filtered data can be found in Appendix B.3. In summary, when the most variable days are removed, the level of scheme benefits is lower although there are still benefits. Where small journey time increases do occur it is possibly due to the increased instances of Variable Mandatory Speed Limits or improved driver compliance Vehicle Hours Delay VHD gives an indication of the overall user delay within a study area. Vehicle hours are calculated by multiplying the average journey time (in hours) by the number of vehicles making the journey. This value is then compared to a reference vehicle hour calculation based on a journey time with no delay at 108kph (67mph). Unfiltered journey times are used as these best represent the average network performance before and after the scheme taking into account all types of days. Detailed results of the VHD analysis can be seen in Appendix B.4. Clockwise the VHD has reduced by 14%, or 550 hours daily. The anticlockwise VHD has reduced by 38%, with 1,130 hours saved daily Summary Average journey times have reduced by 3% overall in the clockwise direction and 2% anticlockwise. The biggest improvements have occurred in the time periods which were congested in the Before period (AM peaks clockwise and PM peaks anticlockwise); for example average journey time reduction of 1 minute 40 seconds for Friday PM anticlockwise. Some of the previously uncongested periods, particularly on J6 to J7 clockwise, have experienced slight increases (a few seconds). These increases are all negligible compared to the benefits at other times and it is likely that they stem from improved speed limit compliance. Speed by distance plots demonstrate that speeds have increased predominantly around the J5 clockwise merge and between J6 and J7 anticlockwise, most likely as a result of road layout changes at these locations. Speeds have reduced in the second half of the scheme on the clockwise carriageway, probably due to better compliance with speed limits. The biggest journey time reliability improvements relate to the times when journey times were most unreliable in the Before period, i.e. the clockwise AM peaks and the anticlockwise PM peaks. At less congested times the journey time reliability has not significantly changed. VHD has reduced by a quarter across the scheme with 550 hours saved daily in the clockwise direction and 1,130 anticlockwise. The improvements in average journey times and journey time reliability during previously congested periods indicate that the SM-ALR scheme is successful in providing better service to customers, being more reliable and resilient.

22 4. Safety 4.1. Introduction This section compares the Before and After safety performance of the M25 J5-7 SM-ALR scheme. STATS19 data has been used to identify the number and rate of personal injury collisions, however it must be remembered that the After period (one year) is a relatively short period; a larger data set is required before the findings will become statistically significant and confidence can be placed in them. The desirable minimum period for the analysis of collision data is three years. STATS19 collates all injury collision data in a consistent manner each year and is a generally reliable source for numbers of injury collisions. Damage-only collisions are not recorded in STATS19 so it is not a record of all collisions. Recording collision details relies on police input at the collision scene, therefore there is some scope for inconsistencies when the information is recorded. Operational STATS19 was used because it includes free-text collision descriptions. This data is robust to the extent that it is unlikely to change significantly when the validated results are produced. Further analysis of STATS19 shows the location of collisions and the user groups involved. Table 4-1 shows the number of collisions during the Before and After periods along with the rate of collisions. Overall the results show a reduction in the collision rate; this is a positive finding however not statistically significant due to the small After sample size, see Section To fully understand the results we also need to take into account the background trend in collisions see Section However it is fair to say that of the collisions occurring during the After period analysis of the contributory factors leading to the collision has failed to identify any factors attributable to SM-ALR features. Table 4-1 Number of Collisions by Severity and Collision Rates Period Before After Fatal Serious Fatal & Serious Slight Total Year Year Year Total Collision Rate (collisions per hmvm) (18.0 hmvm) Collision Rate (collisions per mvkm) (2906 mvkm) Total Year Collision Rate (collisions per hmvm) (6.3 hmvm) Collision Rate (collisions per mvkm) (1017 mvkm) The two fatal collisions in the Before period included a vehicle losing control and leaving the carriageway and a vehicle colliding with an overbridge. Both were single vehicle accidents. There have been no fatalities in the After period, however there have been a total of nine serious collisions in the After period. Eight of these collisions have occurred between J5 to J6 (SM-ALR) and one between J6 to J7 (SM-CM). The serious collisions between J5 and J6 include three collisions involving motorcycles, two of these due to loss of control for unknown reasons and one where a motorcycle filtering through lanes during stationary traffic lost control as a result of a driver opening a vehicle door. There were two collisions associated with lane changing and/or failing to look. A single vehicle collision as a vehicle lost control on a bend colliding with the central barrier and one collision a result of a vehicle hitting debris (tyre) in the road. There was also a serious collision where a vehicle braked sharply for an unknown reason causing a nose to tail collision involving a total of three vehicles, the vehicle at the front and back left the scene.

23 The serious collision between J6 and J7 is described as a vehicle losing control, collided with the central barrier and struck other vehicles. The contributory factors by severity for the collisions are shown in Appendix C Background Trend in Collisions There is a trend over time leading to a reduction in the number of personal injury collisions against a trend of increasing traffic volumes. The reasons for the reduction are wide ranging and include improved safety measures in vehicles. This trend needs to be accounted for when comparing the Before and After periods. The best way to take into account the national trend is to assume that, if the scheme had not been built, the number of collisions on the roads in the study area here would have dropped at the same rate as they did nationally during the same time period. This provides what is known as a counterfactual without scheme scenario and can be compared on a like-for-like basis with the observed After data which is the with scheme scenario 8. The difference between the numbers of collisions in these two scenarios can then be attributed to the scheme rather than the wider national trends. Table 4-2 shows that, there has been a reduction in the collision rate over and above the background reduction in collisions. Table 4-2 Number of Collisions and Collision Rates Following National Trends Period Annual Average Before Period Counter Factual Before Period Number of Collisions Collision Rate (collisions per hmvm) Collision Rate (collisions per mvkm) After Statistical Significance A Chi squared test compared the number of Before and After collisions and Annual Average Daily Traffic flows (AADTs) against expected values if there was no change. The test result indicates that the reduction in the collision rate is not statistically significant and therefore not necessarily a direct impact of the scheme Severity and Severity Index The Severity Index is calculated based on fatal and serious collisions as a proportion of all collisions. The results in Table 4-3 show an increase in the Severity Index which is due to the reduced slight and increased serious collisions. Again the small After sample size means that no conclusions should be drawn at this stage. Table 4-3 Collisions by Severity and Severity Index Period Number of Collisions by Severity Fatal Serious Slight Total Severity Index Before (36 months data) After (12 months data) The counterfactual factor is calculated using the national collision data for motorway class roads After period (2014) and for the middle year in the Before period (2011). The calculated factor between these years is 0.97 for the number of collisions and 0.93 for collision rate.

24 4.3. Casualties, FWI and KSI Rate The Fatal Weighted Injury (FWI) 9 is calculated based on the numbers of fatal, serious and slight casualties as weighted proportions, to adjust for the severity. The FWI rate allows a comparison between road sections with different flows and lengths. The reduction of FWI shown in Table 4-4 is attributable to the smaller number of fatal casualties recorded in the After period, however this is based on a small sample sizes so is not statistically significant. Table 4-4 Number of Casualties and FWI Rate Period Before (36 months data) (18.0 hmvm, 2.91 bvkm) After (12 months data) (6.3 hmvm, 1.02 bvkm) Severity Fatal Serious Slight Total FWI FWI Rate per hmvm FWI Rate per bvkm There has been an increase in the Killed or Seriously Injured (KSI) rate, shown in Table 4-5; this is attributable to the higher number of serious casualties in the After period. However this is based on a small After sample size so, again, is not statistically significant. Table 4-5 Total KSI and KSI Rate Period Total KSI KSI Rate per hmvm KSI Rate per bvkm Before (36 months data) (18.0 hmvm, 2.91 bvkm) After (12 months data) (6.3 hmvm, 1.02 bvkm) Additional Analysis Darkness and Daylight Collisions The results in Table 4-6 shows the number of collisions recorded for daylight and darkness, including information relating to lighting. The provision of lighting is the same in the Before and After periods, and is generally at J5 and between J6 and J7. Approximately 75% of the scheme is unlit. The rate of collisions has reduced in daylight and in darkness. The Contributory Factors for the collisions which occurred in darkness no lighting, did not include any reference to lack of lighting as being relevant to the collisions. Table 4-6 Collisions Occurring in Daylight and Darkness Number of collisions in Period Collisions per hmvm Lighting Condition Before After (36 months data) (12 months data) Before After Daylight Darkness - lights lit FWI is defined as: (number of fatalities) x (number of serious casualties) x (number of slight casualties).

25 Darkness - lights unlit Darkness - no lighting Darkness - lighting unknown Darkness - total Total Red X (Lane Closed) Analysis An analysis of Red X compliance was undertaken using HALOGEN for Sign and Signal settings and MIDAS TCD files for minutely flows per lane. The two data sets were combined to identify lane closures and flows along the lane during the restriction. An example of a Red X event is presented in Figure 4-1. Figure 4-1 Example Lane Closure Event Key: 5381B Signal M25/5381B Detector site L1 Lane 1 L2 Lane 2 L3 Lane 3 L4 Lane 4 (60) 60mph VMSL (50) 50mph VMSL (40) 40mph VMSL LDR Lane Divert Right Arrow xiii Wickets on MS4 Sign showing lane 1 closure /III Wickets on MS4 Sign showing lane divert right NR National Speed Limit A total of 39 lane closures have been assessed and the results are summarised in Appendix C.2. The perlane minutely flow is provided to give an indication of how busy the motorway was; a flow of 30 vehicles per minute per lane is a high flow (one vehicle every 2 seconds). Non-compliance in this sample ranges from 1 to 11 vehicles per minute, 1% to 16% of total flow. Across all Red X events analysed the minutely average flow of non-compliance was 4 vehicles per minute and the average percentage of total flow was 7%. The percentage non-compliance was compared to the incident duration and traffic flow; no correlation was found with either. This suggests that the subset of drivers who choose not to comply with Red Xs do so regardless of how busy the motorway is or how long the incident duration is.

26 ERA Monitoring The ERA monitoring was undertaken to identify the causes of ERA stops, vehicle types and risks of entering, stopping or exiting the ERA. Six ERAs between J5 and J6 were identified for continuous monitoring over long periods covering peaks and inter-peak. (There are no ERAs between J6 and J7 because there is hard shoulder on this section.) A total of 220 hours was divided between the following ERAs: ERA 1 M25 anticlockwise between J5 and J6 at 4288B; ERA 2 M25 clockwise between J5 and J6 at 4286A; ERA 3 M25 anticlockwise between J5 and J6 at 4313B; ERA 4 M25 clockwise between J5 and J6 at 4311A; ERA 5 M25 anticlockwise between J5 and J6 at 4328B; and ERA 6 M25 clockwise between J5 and J6 at 4365A. In total during the ERA monitoring 69 unique ERA stops were observed. This means stops by a lead vehicle; further related vehicle activity such as Highways England Traffic Officer services or recovery vehicles are not counted. A summary of ERA activity can be seen in Table 4-7. Table 4-7 Summary of ERA Activity Activity Number Percentage of All Stops Emergency Refuge Telephone (ERT) Used 3 4% Highways England Traffic Officer Attended 3 4% Non-Emergency (e.g. drove off without exiting vehicle, comfort break etc.) Genuine Reason (e.g. problem with vehicle) 56 81% 13 19% The 69 unique ERA stops over 220 hours of ERAs monitored gives a rate of approximately 0.3 stops per hour per ERA. From the sample observed it was judged that 81% were non-emergency. A breakdown of the types of vehicles which were the lead vehicles stopping in ERAs and whether they were genuine emergencies is shown in Table 4-8. It can be seen that cars make the majority of ERA stops. HGVs and cars represented the highest non-emergency use of ERAs, at 86% and 84% respectively. Table 4-8 Vehicle Types Using ERAs Vehicle Type Number of ERA Stops Percentage of Total Non-Emergency Genuine Emergency Car 32 46% 84% 16% Car & Caravan 1 1% 0% 100% Van 14 20% 71% 29% HGV 22 32% 86% 14% Total 69 81% 19% 4.5. Summary The 12 month After period STATS19 sample size is small so the results are not conclusive. There is a small (but not statistically significant) reduction in collision rate, over and above the national background of improved safety. While the reduction is not significant, the results provide an initial indication that safety has not worsened as a result of the scheme. Monitoring of Red X compliance revealed that across all events analysed, an average of 7% of vehicles were non-compliant.

27 Approximately 0.3 ERA stops per hour per ERA were observed during monitoring of CCTV. It was judged that 81% were non-emergency. HGVs and cars represented the highest non-emergency use of ERAs, at 86% and 84% respectively. During the monitoring one hard shoulder misuse event was observed.

28 5. Conclusions 5.1. Flows For J5 to J6, the SM-ALR section where a lane has been added, flows have increased by 13% clockwise and 3% anticlockwise. The scheme has experienced traffic growth of approximately 2% between J6 and J7, which was not previously at capacity and did not have any lanes added; this is in line with regional growth trends. The percentage of long vehicles has increased across all time slices, by 3 percentage points for weekdays but only 1 for weekends Journey Times Average journey times have reduced by 3% overall in the clockwise direction and 2% anticlockwise. The biggest improvements have occurred in the time periods which were congested in the Before period (AM peaks clockwise and PM peaks anticlockwise); for example average journey time reduction of 1 minute 40 seconds for Friday PM anticlockwise. Some of the previously uncongested periods, particularly on J6 to J7 clockwise, have experienced slight increases (a few seconds). These increases are all negligible compared to the benefits at other times and it is likely that they stem from improved speed limit compliance. The biggest journey time reliability improvements relate to the times when journey times were most unreliable in the Before period, i.e. the clockwise AM peaks and the anticlockwise PM peaks. At less congested times the journey time reliability has not significantly changed. VHD has reduced by a quarter across the scheme with 550 hours saved daily in the clockwise direction and 1,130 anticlockwise. The improvements in average journey times and journey time reliability during previously congested periods indicate that the SM-ALR scheme is successful in providing better service to customers, being more reliable and resilient Safety The 12 month After period STATS19 sample size is small so the results are not conclusive. There is a small (but not statistically significant) reduction in collision rate, over and above the national background of improved safety. While the reduction is not significant, the results provide an initial indication that safety has not worsened as a result of the scheme. Monitoring of Red X compliance revealed that across all events analysed, an average of 7% of vehicles were non-compliant. Approximately 0.3 ERA stops per hour per ERA were observed during monitoring of CCTV. Of these 81% were non-emergency. HGVs and cars represented the highest non-emergency use of ERAs, at 86% and 84% respectively. During the monitoring one hard shoulder misuse event was observed.

29 Appendices

30 Appendix A. Flow Additional Information A Hour Average Daily Traffic (ADT) The table below shows the values for ADTs Before and After. Statistically significant changes are shown in bold for J6 to J7 (see Section 2.5 for a description of the statistical significance testing). Location Value J5-6 (radar, lane increase) J6-7 (loop) Mon- Thurs Clockwise Friday Sat-Sun ADT Mon- Thurs Anti-clockwise Friday Sat-Sun ADT Before 64,400 69,800 57,100 63,000 65,800 72,100 57,000 64,200 After 72,200 77,400 65,300 71,000 67,000 73,300 60,200 66,000 Change 7,800 7,600 8,200 8,000 1,200 1,200 3,200 1,800 % Change 12% 11% 14% 13% 2% 2% 6% 3% Before 72,300 77,700 63,100 70,500 71,200 76,900 61,300 69,200 After 73,100 77,800 66,000 71,700 72,100 78,200 64,500 70,800 Change ,900 1, ,300 3,200 1,600 % Change 1% 0% 5% 2% 1% 2% 5% 2% A.2. Monthly Average Daily Traffic (ADT) The two figures below show the variation in monthly ADT for clockwise and anticlockwise traffic respectively. It can be seen that there has not been a major change to the seasonal trend of traffic flows. The trend of lower flows in the winter months is consistent with the national picture. ADT by Month Clockwise

31 ADT by Month Anticlockwise A.3. Flows by Time Slice The table below shows the flows for each time slice in the clockwise direction. Location J5 J6 (radar, lane increase) J6 J7 (loop) Value AM Peak Mon-Thurs PM Peak AM Peak Friday Interpeak Interpeak PM Peak Saturday- Sunday Before 22,800 16,000 17,600 15,800 17,200 27,700 44,000 After 25,900 17,900 19,300 18,100 19,100 29,700 50,100 Change 3,100 1,900 1,700 2,300 1,900 2,000 6,100 % Change 14% 12% 10% 15% 11% 7% 14% Before 26,400 17,900 19,600 18,100 19,100 30,900 48,800 After 26,700 18,100 19,100 18,300 19,400 29,700 50,700 Change ,200 1,900 % Change 1% 1% -3% 1% 2% -4% 4%

32 The table below shows the flows for each time slice in the anticlockwise direction. Location J5 J6 (radar, lane increase) J6 J7 (loop) Value AM Peak Mon-Thurs PM Peak AM Peak Friday PM Peak Saturday- Sunday Before 16,800 17,200 23,300 10,600 15,900 34,800 44,500 After 17,200 17,800 23,300 11,000 16,400 34,800 46,800 Change ,300 % Change 2% 3% 0% 4% 3% 0% 5% Before 18,300 18,500 25,600 11,600 17,000 37,300 48,100 After 18,100 18,900 25,700 11,600 17,200 37,600 50,200 Change ,100 % Change -1% 2% 0% 0% 1% 1% 4% A.4. Long Vehicles Percentage The percentage of long vehicles by time slice for J6-J7 anticlockwise is shown in the table below. Issues with vehicle classification by radar detectors prevent reporting for J5 to J6. Data was infilled for J6 to J7 clockwise which also prevents reporting. It can be seen that there has been an increase in the percentage of long vehicles across all time slices in the anticlockwise direction between J7 and J6. It is expected that this would be mirrored on the other three links. Location J6 J7 (loop) Value AM Peak Mon-Thurs PM Peak AM Peak Friday Interpeak Interpeak PM Peak Interpeak Interpeak Saturday- Sunday Before 13% 18% 10% 14% 16% 10% 5% After 16% 22% 13% 17% 18% 13% 6% Change (% points)

33 Appendix B. Journey Time Additional Information B.1. Filtered and Unfiltered Days The table below shows the date ranges and number of days removed from the analysis for the filtered and unfiltered data sets. The start date for both data sets is 7 th November, when the scheme fully opened. Data for April could not be analysed for the filtered data because of issues with the TRADS JTDB migration to a new web server, which meant the data was unavailable. Period Before After Clockwise Filtered Anticlockwise Filtered Clockwise Unfiltered Anticlockwise Unfiltered From 1 Sept 11 1 Sept 11 1 Sept 11 1 Sept 11 To 31 Aug Aug Aug Aug 12 Days removed Days in sample From 1 May 14 1 May 14 1 May 14 1 May 14 To 31 March March April April 15 Days removed Days in sample B.2. Unfiltered Journey Time Data B.2.1. Average Journey Time The tables below show the results. Where After period journey times have become longer they are highlighted in red. Clockwise Journey Time Comparison Unfiltered Sample Before After Section Distance (miles) M-T AM M-T IP M-T PM F AM F IP F PM SS J5 to J :56 09:02 08:50 11:19 09:58 08:58 09:14 J6 to J :57 02:32 02:30 02:47 02:37 02:30 02:30 Total :54 11:34 11:20 14:06 12:36 11:28 11:44 Period Average % Change J5 to J :10 09:09 08:48 09:59 09:23 08:45 08:53-5% J6 to J :11 02:38 02:33 02:47 02:45 02:33 02:34 3% Total :21 11:47 11:20 12:46 12:08 11:17 11:27-3% % Change -3.4% 2.0% 0.1% -9.5% -3.7% -1.5% -2.4%

34 Anticlockwise Journey Time Comparison Unfiltered Sample Before After Section Distance (miles) M-T AM M-T IP M-T PM F AM F IP F PM SS J5 to J :35 02:40 03:24 02:34 02:43 04:08 02:35 J6 to J :31 08:44 09:49 08:27 08:51 10:23 08:30 Total :07 11:23 13:13 11:02 11:34 14:31 11:05 Period Average % Change J5 to J :37 02:44 03:03 02:36 02:41 03:13 02:38-5% J6 to J :40 08:55 09:22 08:35 08:48 09:38 08:31-1% Total :17 11:39 12:25 11:11 11:29 12:51 11:09-2% % Change 1.5% 2.3% -6.1% 1.4% -0.8% -11.5% 0.7%

35 B.2.2. Speed (kph) Over Distance plot Clockwise Speed Over Distance Clockwise Speed Vs Distance Profiles (Unfiltered) AM Monday Thursday IP PM AM Friday IP PM Saturday & Sunday Day

36 Anticlockwise Speed Over Distance Clockwise Speed Vs Distance Profiles (Unfiltered) AM Monday Thursday IP PM AM Friday IP PM Saturday & Sunday Day