Relation between winter road maintenance and road safety

Nr 399A - 1995

Relation between winter road maintenance and road safety

Hans Savenhed

VTI rapport

Nr 399A 0 1995

Relation between winter road

maintenance and road safety

Hans savenhed

Swedish National Road and

Publisher: Publication:

VTI Rapport 399A

Published: Project code:

Swedish National Road and 1995 72338-7, 20150

I anspart Research Institute

S-581 95 Linkoping Sweden Project:

Printed in English in 1996 Relation between winter road maintenance

and road safety

Author: Sponsor:

Hans Savenhed Swedish National Road Administration

Title:

Relation between winter road maintenance and road safety

Abstract (background, aims, methods, results) max 200 words:

To be able to optimise resources for winter road maintenance from a socioeconomic perspective, it is of vital importance to the road administrator to know how traffic safety is in uenced by maintenance type and time of action.

In recent years, the use of salt has changed. Pre-wetted salt and a saline solution are more frequent today. Anti-icing before the occurrence of slippery conditions is also more common. It is interesting to study the effects of different strategies and choice of material.

The National Road Administration has improved the records of winter road maintenance actions by using a computerised system, thus increasing the possibility of follow-up.

These data from the winters of 1988/89 and 1989/90 have been linked and matched with the traffic and accident data from the National Road Administration s road data base.

The accident rate during the hours before and after the action was calculated for a number of factors, e. g. region, type of measure, type of road and speed limit.

The study was commissioned by the National Road Administration.

ISSN: Language: No. of pages:

CONTENTS SUMMARY

1 BACKGROUND

2 AIM

3 DATA

3.1 Action data from follow-up

3.2 Accident and road data

3.3 Traffic data

4 ANALYSIS RECORD

5 RESULTS

5.1 Weather data

5.2 Action data

5.3 Accident rate more than 12 hours before and after

imple-menting the action

5.4 Accident rate 12 hours before and after implementing the

action, all actions

5.5 Accident rate for different types of action

5.6 Accident rate in different regions

5.7 Accident rate for other distributions

5.8 Accident rate according to season

5.9 Injury rate

6 DISCUSSION

REFERENCES

Appendix 1: Questionnaire for positioning actions Appendix 2:

Appendix 3: Appendix 4:

distributed among region and type of action

width, ADT and maintenance class

VTI RAPPORT 399A

Distribution of accidents concerning type of action and region

Page U ) M O O -b -b 13 14 14 15 18 20 24 26 28 35 36

38

40

Number of km with implemented actions and number of jobs started,

PREFACE

Several extensive checks on data quality have been carried out during the project. Special thanks are due to the father of the VVH system (follow up of actions), Owe Widegren, for all his assistance with data transfer and for many valuable views and suggestions.

Hans Velin and Peter Wretling must be given credit for their handling of very large data files and complex data processing.

Valuable opinions were provided by colleagues at the VTI and representatives of

the contractor (Swedish National Road Administration, SNRA). Discussions were held with Lennart Axelson, Hans Danielsson, Osten Johansson and PG Land, all of the SNRA. Ulf Brijde, Staffan Moller, Rein Schandersson and Gudrun Oberg,

all of the VTI, were inexhaustible discussionpartners when difficulties arose. Christina Ruthger was responsible for the English version.

I also wish to thank Siv Britt Franke and Annette Karlsson, who edited and

supervised the typing of the report.

Hans S avenhed

Relation between winter road maintenance and road safety by Hans S'avenhed

Swedish Road and Transport Research Institute (VTI) S 581 95 LINKOPING

Sweden

SUMMARY

To be able to optimise resources for winter road maintenance from a socioeco-nomic perspective, it is of Vital importance to the road administrator to know how traffic safety is in uenced by maintenance type and time of action.

Previous studies have shown that the accident risk increases rapidly with com paratively small quantities of snow. However, there is insufficient knowledge con-cerning the exact duration of the increase in accident risk and the level of increase

in risk.

In this study, the hourly accident rate was calculated for 12 hours before and after implementing the action. The information thus obtained has facilitated a study of the accident rate for different types of actions. A survey of regional differences

was also carried out.

The analysis was based on the follow-up of winter road maintenance measures carried out by the Swedish National Road Administration. During several winters, a considerable number of maintenance areas have used a computerised system, designed exclusively to follow up winter road maintenance. Information on the type of measure, place, start and finishing time contained in the system made it interesting to try to match it with the Road Administration's road data base in order to carry out accident rate calculations in relation to time of action.

The analyses are based on data from the winters of 1988/89 and 1989/90

(October April). Data from 50 maintenance areas were used (25 in the north, 16

in central Sweden and 9 in southern Sweden). During these two winter periods, 7,884 accidents were reported by the police on the road network covered by the

maintenance areas. Of these accidents, 1,766 occurred within 12 hours before or

after the action. When wildlife accidents and accidents at junctions (accidents that occurred at a maximum of 10 m from a junction) were excluded, 941 accidents

remained in the analysis.

II

The accident rate (number of accidents per million axle pair kilometres) increased during the hours before the action and reached its maximum value 1 to 1.5 hours before the action was implemented. This maximum value (2.6) is 12 times higher than the average accident rate calculated from the accidents that occurred more than 12 hours before or 12 hours after the winter road maintenance action had been carried out, i.e. for the complementary period to the 24-hour period included in the action analysis.

It may seem remarkable that the accident rate reaches its maximum about one

hour before the maintenance action. The result is, however, well in accordance

with the accident rates reported in similar studies in Germany and the US. The explanation may be that road users have started to reduce their speed and adapt their driving behaviour to the impaired road conditions, implying that the accident rate decreased to some extent even before the winter road maintenance action had been performed. During the first half hour after implementing the action, the

acci-dent rate was found to be halved.

A distribution according to the type of action gives a maximum value of the acci-dent rate in anti-icing which is half the corresponding rate in de-icing.

Accidents involving one vehicle only reach a maximum accident rate which is three to four times higher than that for other accident types.

As expected, the south of Sweden showed an accident rate with a maximum value

three times the corresponding value for the north of Sweden.

1 BACKGROUND

In Sweden, the Swedish National Road Administration started salting roads in the mid 608. In the mid-70$, ploughing and salting of roads were carried out simulta-neously by a single vehicle (combined salting and snowploughing). Tests with pre-wetted salt started at the beginning of the 805 and became more widespread at the end of the 803. Furthermore, the technique of utilising a saline solution then began to be used.

To be able to optimise resources for winter road maintenance from a socio-eco nomic perspective, it is of vital importance for the road administrator to know how traffic safety is in uenced by the choice of maintenance type and time of action.

Previous studies, Reference 1, have shown that the accident rate (number of

acci-dents per million axle pair kilometres) increases rapidly with comparatively small quantities of snow and stabilises at an increase of 3 5 times when there is more than 5-10 mm of snow precipitation and 2.5-3 times with both snow and rain during one day. Calculations of accident rate before and after winter road mainte-nance actions are reported in Reference 2. These calculations are based on a com-paratively small material, which is also beginning to be out of date and thus unre-liable for any conclusions to be drawn.

In the Road Administration s Effect catalogue (January 1989, Reference 3), the

relations between amount of snow on the road and accident rate were changed

substantially, compared with the earlier version.

The SNRA has attempted to improve its record of winter road maintenance actions by using a computerised system, VVH (Reference 4). The aim of the system was to improve the possibility of follow-up. The system has been available for a few years in the maintenance areas. Type of action, road network, start and finishing time of each action, etc. were stored in the PC program.

By using the follow-up of actions in the VVH system, it was possible to specify the accident rate before and after winter road maintenance actions more accu-rately. The road network in the VVH system is larger than in previous studies. This also means that the number of accidents used as a basis for calculations is more comprehensive, which reduces the uncertainty of the results. The possibility of dividing the material into several classes is also improved.

The type and time of action are probably also more accurately reported than in

previous studies. It should be observed, though, that neither the report in the VVH

system nor the data used in previous studies were originally collected with the aim of calculating accident risk.

2 AIM

In recent years, the usage of salt has changed. Pre-wetted salt and saline solutions are more frequent today. Anti-icing (before the occurrence of slippery conditions) is also more common. It is interesting to study the effects of different strategies

and choice of material.

The aim has been to increase knowledge and improve the precision of information on how the accident rate (number of accidents per million axle pair km) varies during the hour before and after winter road maintenance actions and to quantify both the magnitude of the increase in accident rate at the action and its duration. Even subsets of the material will be studied. For example, regional and seasonal accident rate and differences in accident rate during day and night will be

calcu-lated. Furthermore, the accident rate will be calculated for variables connected

with the road, such as type of road, road width, speed limit and maintenance class.

3 DATA

The analysis record is based on the data collected from the VVHfollow-up of the

National Road Administration s winter road maintenance actions. These data are further described in Section 3.1.

Furthermore, the accident analysis is based on data collected from the SNRA s

road data base. Accident and road data are described in Section 3.2, while traffic

data are reported in Section 3.3.

The linking of these data to form an analysis record is described in Chapter 4.

The data and calculations will be described in more detail in a VTI Notat,

Refer-ence 5.

3.1 Action data from follow-up

In order to improve the follow-up of winter road maintenance actions, a computer ised system has been developed. All SNRA s maintenance areas have beenable to use the system.

Immediately after the action has been completed, the person who carries out the action fills in a form. These action data are then entered in the VVH programme. Data from four separate files collected from the VVH data base are used.

0 The rst le contains data on working periods. The file contains a job number, start and finishing time for the whole working period, snow depth

(occasionally), cause and measure, and other details.

0 Using these basic data, the second le is then created, which consequently contains calculated values. Working periods can be long (more than 10 hours) and are thus divided into smaller parts. These parts each form an entry in the new file. The entries contain details such as the same job number as in the first

file: a job sub-number, calculated start and finishing time for all sub sections,

type of action and an identification of section and area.

0 The third le contains the above mentioned identification of section and area,

thus giving a serial number for each entry.

0 The fourth le contains detailed information on sub-sections, e.g. road num-ber, place name at start and finishing points, length of section, maintenance

class, etc.

The transfer of data from Dbase in the PC environment to ASCII codes in the VAX/VMS environment led to a number of unexpected problems concerning computer technology.

The coding of places for actions proved to be a greater problem. The VVH is designed for a local follow-up. Unfortunately, the VVH system does not utilise the reference system in the road data base where each junction has a unique

identifi-cation number. Instead, place names, estate names or in the worst case (for the uninitiated) local names and definitions are used. In order to be able to link acci-dent and ow data from the road data base with action data from the VVH, all

local names in the VVH have to be translated to the reference system of the road

data base.

To obtain a translation into the SNRA s reference system, aquestionnaire was

sent to all maintenance areas using the VVH system and included in the study (Appendix 1). This procedure demanded a great deal of work and reduced the planned working material since not every maintenance area could manage the extra work required for the recoding. The questionnaire was sent to 82 mainte-nance areas and the response rate was 60% (50 maintemainte-nance areas).

Action data from the VVH follow-up of two winters (1988/89 and 1989/90) were

used. Twenty-eight maintenance areas with data from both winters and 22 mainte-nance areas with data from one winter were included in the analyses. Data from approximately one sixth of the SNRA s maintenance areas were thus available.

Action data were obtained from 25 maintenance areas in northern Sweden, 16

maintenance areas in central Sweden and 9 maintenance areas in southern

Sweden.

The length of the road network was 25,970 km: 20% in southern Sweden, 47% in

central Sweden and 33% in northern Sweden. During the two winters, 17,084

actions were started and 832,720 km were treated. Distribution among region and type of action is reported in Appendix 2.

Northern Sweden

_{..., Central Sweden}

- Southern Sweden

E: Counties not analysed!

Figure 3:1 Regional division in northern, central and southern Sweden VTI RAPPORT 399A

When dividing the entire working periods from the first file in the second file, a simple interpolation is used. The entire time of action is spread proportionally over the length of the sub-sections. This means a loss of quality when indicating the time of action for a sub-section of the road. Two examples of errors that may occur are given below.

C

B

Figure 3:2 Example of action where a transport section is necessary (salting). In the example in Figure 3:2, the vehicle starts from B, salting the sub-section BC. The transport section CA is then covered in order to finish the working period by salting the sub-section AB. All working hours are spread proportionally over BC and AB, while no time is given to the transport section AC when the work sub-file

is created.

Salt refilling, food breaks or other types of interruptions during the working period were disregarded by the programme creating the second job sub-file.

D

Figure 3.3 Example of action carried out in both directions (ploughing)

In this example, the road is ploughed from A to D and then back from D to A. This situation is treated by the VVH programme by dividing time proportionally;

time is measured first for AB and then for BC and CD in turn, i.e. as if actions had

been implemented on both road halves on one section before the next sub-section was treated in the same way.

In both examples, the times of action will be incorrect on the sub-sections.

A close quality check has been carried out in one maintenance area in the county of Cstergotland around the municipality of Atvidaberg. The real times for a number of jobs on the sub-sections were evaluated and compared with the times interpolated in the second VVH file (job sub-file).

The result was that 50% of the job sections had a time error in absolute values of

11 minutes at a maximum and 75% of 27 minutes at a maximum.

Furthermore, time errors were smaller on main roads with the heaviest traffic and

most accidents. This is due to the fact that actions are nearly always implemented

first on these roads.

After these comparisons, it was agreed that (with reasonable effort) a better

divi-sion of the entire work into job sections cannot be made than that between the

files in the VVH follow up.

3.2 Accident and road data

Accident and road data were collected from the Swedish National Road Admini

stration s road data base both for personal injury accidents and property accidents. All traffic accidents reported to the police, registered in the road data base during the two winters and occurring in the road network included in the maintenance areas where production data were collected, were stored in a file. The same coor-dinate system as in the road data base was used.

A total of 7,884 accidents reported to the police occurred on the road network

during the two winters. Of these accidents, 1,766 occurred in the period included

in the accident rate analysis, 12 hours before or 12 hours after actions were

implemented on the road. When wildlife accidents (678) and accidents at

junc-tions (147) were excluded, 941 accidents remained in the analysis.

Accidents were distributed among the factors used in the analysis according to the following tables.

Table 3:1 Number of accidents distributed among region

Region Number of % accidents Southern 223 23.7 Central 507 53.9 Northern 2 1 1 22.4 Total 941 100

Table 3:2 Number of accidents distributed among type of action

Type of action Number of %

accidents Combined salting 154 16.4 and snowploughing Salting 584 62. 1 Anti icing 86 9. 1 Snowploughing 1 17 12.4 Total 941 100

10

Table 3:3 Number of accidents distributed among accident type

Accident type Number of % Including wildlife

accidents and junction

-accidents

Number %

Single 506 53.8 566 32.0

Turning, junct.,

rear-end, oncom., overtak 93 9.9 162 9.2

oncoming, overtak. 235 25.0 248 14.0

Cycle, pedest, miscel 107 l 1.4 1 12 6.3

Wildlife - - 678 38.4

Total 941 100 1,766 100

Table 3:4 Number of accidents distributed among speed limit (R1 = direction 1,

R2 = direction 2)

Speed limit Number of %

accidents 50 59 6.3 70 169 18.0 70/R1 90/R2 16 1.7 90 538 57.2 90/R1 110/R2 1 0.1 110 158 16.8 Total 941 100

Table 3:5 Number of accidents distributed among type of road

Type of road Number of %

accidents

Two-lane road 775 82.4

Motorway (MW) 53 5.6

Express road (ER) 105 11.2

Four-lane road (4 LR) 8 0.9

Total 941 100

11

Table 3:6 Number of accidents distributed among maintenance class

Maintenance class Number of %

accidents MW + 4LR + ADT 2 7,000 367 39.0 Nation. rd + 1,500 S ADT < 465 49.4 7,000 Regional roads 109 1 1.6 Total 941 100

Table 3:7 Number of accidents distributed among ow class

ADT class Number of %

accidents ADT < 500 14 1.5 500 S ADT < 2,000 163 17.3 2,000 S ADT < 8,000 476 50.6 8,000 S ADT < 16,000 264 28.1 2 16,000 24 2.6 Total 941 100

Table 3:8 Number of accidents distributed among road width

Road width Number of %

accidents < 6 m 28 3.0 6 - 8.5 m 377 40.1 8.5 - 10 m 150 15.9 > 10 m 378 40.2 Unknown 8 0.9 Total 941 100

12

Appendix 3 presents the distribution among sub groups for region and type of

action.

3.3 Traf c data

ADT data were collected from the SNRA s road data base. The average annual daily traffic for the year reported in the road data base was used for each road section in the analyses. These values were then used to calculate the hourly traffic for the two winters in questions. Index series for calendar year, month, day of the week and hour, distributed among road categories based on many years of traffic

measurements, were used to calculate the hourly traffic then used in the analyses (References 6 and 7).

Using these index series, a quantity of data was created, containing the hourly traffic for all hours during the two winters in question and for each road section in the road network included in the analysis.

The time for the action rarely coincides with an exact hour. The vehicle mileage forming the basis of the accident risk calculations were weighted by means of two consecutive hourly traffic ow from the data base described above. If an action was implemented at 1.04 pm, the vehicle mileage for the half-hour just before action was calculated in the following way: 26/30 of half the hourly traffic calculated between noon and 1 pm. was added to 4/30 of half the hourly traffic calculated for the period 1 pm. to 2 pm. In order to calculate vehicle mileage, this volume of traffic was then multiplied by the length of the section in question. The in uence of weather and road condition on traffic ow has been disregarded.

13

4 ANALYSIS RECORD

The analysis record contains action data, accident and road data and hourly traffic variations in general. These data are linked through the reference system used in the SNRA s road data base. Each junction has a unique identification number. These three files were then linked and matched for each road section in the main-tenance areas. The basic values of the analyses were road section and time of action. In order to calculate the accident rate, accidents and vehicle mileage were linked, 12 hours before and 12 hours after implementing the action.

In the case where two actions were implemented so close in time that the after and before periods overlapped, time was distributed according to the example below. In this example, the after period of action 1 overlaps the before-period of action 2 by 5 hours. These 5 hours are divided equally between actions. The after-period of

action 1 is thus reduced to 9.5 hours (12 - 5) + 2.5 hours. In the same way, the

before-period of action 2 is reduced to 9.5 hours, 2.5 + (12 - 5) hours.

X 19 hours X

action 1 action 2

12 hours after

12 hours before

7 hours 5 hours 7 hours

5.1

RESULTS

Weather data

14

The weather conditions prevailing throughout the country during the two studied

winters are described in brief, i.e. October April 1988/89 and 1989/90. The data

were collected from Weather and Water , published by the Swedish Meteoro-logical and HydroMeteoro-logical Institute, for the months in question, Ref. 8.

The tables below present the average temperature and precipitation for each month compared with normal values. Snow depth is for the end of the month. In the

tables, G stands for Gotaland, S for Svealand and N for Norrland. The mountain

area of Norrland was not included.

October November December January February March April Summary: Average temp. O- 1 OC colder l 4°C colder

coldest in the north

O-3°C colder

WINTER 88/89

Compared with normal values

Precipitation Snow depth

Somewhat less 0 cm

Some 0 25 cm

Somewhat less 0 cm G

0-50 cm S+N

coldest in the north 4-7°C warmer

5 7°C warmer

4-5°C warmer

0-2°C warmer

G extremely little 0 cm G+S-southN S little - normal

N normal 0-50 cm upperN

Normal double 0 cm G+S 0-100 cm N

Normal - 2-3 timesO cm G+S+southN

0 150 cm upperN

G small 0 cm G+S+southN

S+N normal - 0-150 cm upperN

2-3 times

A cold start with little precipitation, followed by a warm period with considerable precipitation. Gotaland was snowless for a large part of the winter period. There were exceptionally long periods with bare ground in Svealand and south east Norrland.

15

WINTER 89/90

Compared with normal values

Average temperature Precipitation Snow depth

October 0 1.5°C warmer G 1.5 2 times 0 cm

N O-1°C colder S+N 0.5-1.5 times

November 0-1.5°C warmer Gotland+coast of N 0-20 cm

sG O-1°C colder normal-double

rest half

December 0-3°C colder Half - normal 0 cm G+S

coldest in the north 0-20 cm N

January O-6°C warmer wG+N 2-3 times 0 cm G+S

warmest in the north 0-100 cm N

February 6 9°C warmer Normal - 2-3 times 0 cm G+S

warmest in the north O-lOO cm N

March 4-6°C warmer Half - normal 0 cm G+S+sN

0-100 cm

in upper N

April 2-3°C warmer G 0.5-1 time 0 cm

S 0.5-2 times N 0.5-1.5 times

Summary: The winter began with changeable weather but turned warm with considerable precipitation. Gotaland and southern Svealand were snowless for a considerable part of the winter.

5.2 Action data

Figures 5:1 and 5:2 present the exact time and weekday when the actions were

started.

The somewhat higher proportion of actions on Mondays and Tuesdays indicates that instructions are not always observed but normal working hours are still partly decisive for performance.

The problem has been even greater and is known to the SNRA. In recent years, the level of ambition to implement actions according to existing instructions has been improved.

16

Pe

r

ce

nt

Figure 5:1 Starting time for actions distributed among hours of the day VTI RAPPORT 399A

N O CO (.0 ' V N 1 1'

19

20

21

22

23

9

1O

11

12

13

14

15

16

17

18

8

Ho

ur

17

Mo

nd

ay

20

~

15

10

*

Pe

r

ce

nt

25

Figure 5:2 Starting time for actions distributed among weekdays VTI RAPPORT 399A

Sa

tur

da

y

Fr

id

ay

We

dn

es

da

y

Th

ur

sd

ay

TU

es

da

y

Sun

da

y

We

ek

da

y

18

5.3 Accident rate more than 12 hours before and after implementing

the action

First, the accident rate was calculated for the part of the two winters when no winter maintenance action had been implemented and which consequently had not been included in the analysis (complementary period), i.e. more than 12 hours

before and 12 hours after time of action.

The intention was to compare this basic level of the accident rate with the increases obtained when winter maintenance actions are required by road

condi-tions.

Figure 5:3 shows a distinct increase in accident rate during the dark hours of the day.

It should be observed that the reported maximum values of accident rate are probably an underestimation of the real values. In bad weather and road condi-tions, vehicle mileage is probably reduced to some extent compared with the values calculated by means of an index. It is also probable that the concentration of accidents in worse weather and road conditions is reduced slightly because of the uncertainty of the connection between accidents and time of action. In both

cases, the calculated accident rate is in uenced and the maximum value is lower than the real value.

Ac

ci

de

nt

ra

te

6

.

19 1

-Figure 5:3 Distribution of accident rate among the hours of the day, more than 12 hours before and 12 hours after time of action (complementary period). The broken line is the 95% con dence interval.

VTI RAPPORT 399A

16

20

24

12

De

fi

ni

te

ti

me

20

5.4 Accident rate 12 hours before and after implementing the action,

all actions

Dry salting, pre-wetted salt and saline solutions were combined into a single class (salting). The types of action reported in the analyses are snowploughing, com-bined salting and snowploughing, salting and anti-icing. The action of sanding was excluded in the analyses since the amount of data was too small and too uncertain. When sand was used, the division of time for subsections is very uncer-tain as the time for loading new sand was also included in the working hours. The accident rate was calculated 12 hours before and 12 hours after implementing the action, respectively. The accident rate was calculated for each half-hour during the six hours just before and after the time of action. Between 6 and 12 hours after the action, the accident rate was calculated for every hour. A 95% confidence

interval was calculated, based on the uncertainty of the number of accidents.

Wildlife and junction accidents were excluded from the calculations.

At a junction, winter road maintenance may be performed at different times or with different types of action on the connecting roads. To be able to link accidents and time to the correct type of action, junction accidents were excluded from the analysis.

Both wildlife and junction accidents are reported in Figure 5:12, where the acci-dent rate for all types of acciacci-dents has been calculated. The figure shows that wildlife accidents are not in uenced by winter road maintenance. The result is well in accordance with experience from earlier studies, for example the MINSALT study.

In Figure 5:4, all actions throughout Sweden are reported. The figure is based on 941 accidents.

21

I l l I l I 1 l 1 1

Ac

ci

de

nt

ra

te

6

-5i

4_

Figure 5:4 Accident rate (accidents per million axle pair km) twelve hours before and after implementing the action (all actions, 941 accidents).The broken line is the 95% confidence interval.

VTI RAPPORT 399A

12

10

Ho

ur

s

af

te

r

1

0

Ho

ur

s

be

fo

re

1

2

Ac

ti

on

22

Figure 5:4 shows the increase in accident rate the hours before implementing the action, reaching its maximum value 1 to 1.5 hours before implementation. The maximum accident rate of 2.6 is 6 times higher compared with the average acci-dent rate calculated for the period between 12 hours and 6 hours before the action.

The increase in risk is double, 12 times, compared with the accident rate

calcu-lated for the complementary period, i.e. all hours during the two winters which are

at least 12 hours from the time of action.

Directly after implementing the action, the accident rate is halved from 1.8 to 0.9, see Figure 5:4.

There is a slight increase in accident rate for the period 12 to 6 hours before action compared with the period 6 to 12 hours after action. One explanation could be that many actions start in the morning, Figure 5:1. This means that the before-period comprises more hours with darkness than the after-period. Since the accident rate is higher in darkness than in daylight, this could explain the increased level of accident rate calculated for the period before action.

It may seem remarkable that the accident rate reaches its maximum about one

hour before the maintenance action. However, the result is well in accordance

with the results reported in similar studies in Germany and the US (References 9 and 10). One explanation could be that road users have started to reduce their speed and adapt their driving behaviour to the impaired road conditions, implying that the accident rate decreased to some extent even before the winter road main-tenance action had been performed.

Figure 5:5 describes the accident rate for the isolated actions. When the before and after periods of two consecutive actions do not overlap, they are defined as isolated (i.e. actions are implemented at least 24 hours before or after another

action). The accident rate is similar for the whole material, but with the difference

that the maximum level is considerably higher. Earlier studies show that there is an increase in accident rate when slippery conditions are rare, compared with the case when slippery roads and bad road conditions have a long duration or are fre-quent (References 11, 12 and 13).

Ac

ci

de

nt

ra

te

23 h- - -- - u u

-1

0

Ho

ur

s

be

fo

re

1

2

Figure 5:5 Accident rate for actions implemented at least 24 hours before or after

another action ( isolated actions, 359 accidents). The broken line is the 95% confidence interval.

VTI RAPPORT 399A

Ho

ur

s

af

te

r

Ac

ti

on

24

5.5 Accident rate for different types of action

Figure 5:6 presents the accident rate for the types of action included in the analysis.

The accident rate has the same pattern for all types of actions, with a maximum

value one or a few hours before the action is implemented. More than half (62%)

of the analysed accidents occurred when the action was de-icing.

The maximum value (1.6) of the accident rate for anti-icing is not quite half as large as the corresponding value (3.4) for de-icing. However, the difference is not

significant since each confidence interval (95%), based on accidents, overlaps.

If anti-icing is effective, ideally there should be no increase in accident rate. The figure shows that the method seems to be fairly effective since the maximum level is considerably lower compared with conventional methods. It is also confusing that the maximum value of the accident rate for anti-icing occurs before imple-menting the action. If the action had failed in any respect, the accident rate would be highest after the action, since it should be performed before the occurrence of slippery conditions. The most probable explanation is that the working periods are so long that when anti-icing is started and registered, there is insufficient time to salt every subsection before the slippery condition is a fact, a situation that was to be prevented. Salting of some of the sub-sections after the occurrence of slippery conditions has thus been incorrectly registered as anti icing. The use of pre wetted salt may also have been incorrectly registered as anti-icing because this is the method utilised in anti-icing.

These facts were reported to the SNRA s contacts, who in turn announced that the anti-icing method has been improved since the start. The experience gained since the method was introduced has been put into practice and today there is a

convic-tion that the result is even better.

The difference in maximum level of the accident rate for different types of action is mainly explained by the fact that the choice of action is in uenced by differ

ences in road condition and weather, which in turn give rise to different levels of

accident rate.

Typ

e

of

ac

ti

on

Ac

ci

de

nt

ra

te

Sn

owp

lo

ug

hi

ng

""

""

"

an

d

sa

lt

in

g

25

m

Sa

lt

in

g

*

-An

ti

ic

in

g

Sn

owp

lO

Ug

hi

nQ

Figure 5:6 Accident rate distributed according to type of action. (Snowploughing

and salting 154 accidents, salting 584 accidents, anti-icing 86

acci-dents and snowploughing 117 acciacci-dents.)

VTI RAPPORT 399A

12

10

Ho

ur

s

af

te

r

1

0

Ho

ur

s

be

fo

re

1

2

Ac

ti

on

26

5.6 Accident rate in different regions

Figure 5:7 shows the accident rate distributed among regions. As expected, the maximum accident rate for southern Sweden was considerably higher than the

corresponding rate for central and northern Sweden (Reference 14). The rate is 5

for southern, 3 for central and 1.6 for northern Sweden. Part of the explanation is probably that slippery conditions and snow are much more frequent and persist for longer periods in northern Sweden, while at the same time, studded tyres are used more frequently compared with southern Sweden. In southern Sweden, precipita-tion is more frequently connected with slippery condiprecipita-tions (= impaired vision). In northern Sweden, the proportion of accidents reported to the police is lower than

in southern Sweden.

Re

gi

on

Ac

ci

de

nt

ra

te

27 m So ut he rn Ce nt ra l No rt he rn

Figure 5:7 Accident rate divided among region (southern 223 accidents, central

507 accidents and northern 211 accidents).

VTI RAPPORT 399A

12

10

Ho

ur

s

af

te

r

1

0

1

2

Ac

ti

on

Ho

ur

s

be

fo

re

28

5.7 Accident rate for other distributions

Figure 5:8 presents the accident rate according to speed limit, Figure 5:9 type of road, Figure 5: 10 road width, Figure 5:11 maintenance class and Figure 5: 12 type

of accident. Wildlife accidents (678) and junction accidents (147) are included,

making a total of 1,766 accidents. These figures show that the increase in accident

rate is the same during the hours around the action, with a maximum value about one hour before the action is implemented. Furthermore, the maximum value of

the accident rate is lowest for large roads. The accident rate has its lowest maximum value for roads with a speed limit of 110 km/h, motorways, roads wider than 10 metres and the maintenance class including roads with an ADT exceeding 7,000.

The accident rate for single accidents is substantially higher (3 4 times) compared with other types of accidents. The accident rate for wildlife accidents is at a con-stant level and is not in uenced by weather and road conditions. These types of accident constitute the greater part of the accidents in the analysis, single 32% and wildlife 38%.

The material was divided into several categories and some of these divisions are reported in Appendix 4. The figures must be studied with great care since the accident material was not as comprehensive as planned due to the fact that not all the action data could be entered into the reference system in the road data base.

When the whole material is studied, the increase in accident risk is reduced by

50% for anti-icing compared with de-icing. When the material is distributed among various speed limits, road width or ow class, the increase in accident risk for anti-icing is lowest on large roads. The speed limit 110 km/h compared with 70 km/h, road width exceeding 10 m compared with a road width of 6 to 8.5 m and a ow class ranging from 8,000 to 16,000 compared with 500 to 2,000 vehicles per day, all show the same tendency when anti-icing is compared with de icing.

The above-mentioned results could support the discussion in Chapter 5.5. The jobs are so lengthy that some, started and registered as anti-icing, will be carried out too late on the concluding subsections and should rightly be registered as salting. Actions are often implemented on main roads first of all. On these roads, it is most probable that a job registered as icing is really carried out as

29

icing and thus registered correctly. On minor roads, there is a risk that a job regis-tered as anti-icing is carried out when the slippery condition is already a fact and should thus have been registered as de-icing.

To sum up, the above reasoning should strengthen the positive picture of the acci-dent material concerning anti icing compared with de-icing with reference to the

increase in accident risk.

When the type of action is combined salting and snowploughing or ploughing, the case seems to be exactly the opposite. Roads with low speed limits, narrower carriageways or lower ows show a lower increase in accident risk. This is hardly due to the fact that the slush was removed more efficiently or faster on minor

roads, but rather because speeds are higher on main roads.

Sp

ee

d

li

mi

t

Ac

ci

de

nt

ra

te

30

Figure 5:8 Accident rate according to speed limit (70 km/h 169 accidents, 90

km/h 538 accidents and 110 km/h 158 accidents).

VTI RAPPORT 399A

12

10

Ho

ur

s

af

te

r

00

co

w

N

C O

on:

0

<1:

Ho

ur

s

be

fo

re

Typ

e

of

ro

ad

Ac

ci

de

nt

ra

te

31

8

e

8

_o a)

:3;

5.3g ,

IQQ

00g. EEG) .J

I]

ll

l l I I I I l T 1 I I I l I l I l I

12

10

Ho

ur

s

af

te

r

1

0

Ho

ur

s

be

fo

re

1

2

Ac

ti

on

Figure 5:9 Accident rate according to type of road (two-lane roads 775

accidents, motorway 53 accidents and express roads 105 accidents). VTI RAPPORT 399A

Ro

ad

wi

dt

h

Ac

ci

de

nt

ra

te

32

12

wi dt h < 8. 5m wi dt h < 1 0 m

10

Ho

ur

s

af

te

r

wi dt h

Ii

1

0

Ho

ur

s

be

fo

re

1

2

Figure 5:10 Accident rate according to road width (6-8.5 111 377 accidents, 8.5 10 In 150 accidents and >10 In 378 accidents).

VTI RAPPORT 399A

Ac

ti

Ma

in

te

na

nc

e

cl

as

s

Ac

ci

de

nt

ra

te

M W + 4L R + Na t. ro ad s + m 15 00 < A DT < 70 00

-AD

T>

70

00

-Re

gi

on

al

ro

ad

s

-1

0

Ho

ur

s

be

fo

re

i

5

i

₁

2

Figure 5:11 Accident rate according to maintenance class (MW+4LR+ADT>

7,000 367 accidents, national roads+1,500<ADT<7,000 465 acci

dents and regional roads 109 accidents). VTI RAPPORT 399A

Ho

ur

s

af

te

r

Ac

ti

on

VTI RAPPORT 399A

accidents and wildlife 678 accidents).

turning +junction+rear end 162 accidents, oncoming+overtaking 248(Various+cyc1e+pedestrians 112 accidents, single 566 accidents,

Figure 5:12 The accident rate according to type of accident.

Ac

ci

de

nt

_1.9

4

_1.8

?

_1.7

~

1.6:

_1.5

_1.4

9

1.3:

_1.2:

1.1

a

1.0-

:

_0.9: _0.8_

_0.7

-

_0.6

-:

_0.5

5

_0.4

_0.

31

_0.2

?

_0.1

?

_0.

09

ra

te

Typ

e

of

ac

ci

de

nt

in

cl

ud

in

g

wi

ld

li

fe

an

d

jun

ct

io

n

ac

ci

de

nt

s

Va

ri

ous

+

cyc

le

+

pe

de

st

ri

an

s

m

Si

ng

le

Tur

ni

ng

+

jun

ct

io

n

+

rear -e nd

On

co

mi

ng

+

w

ove

rt

ak

in

g

Wi

ld

li

fe

1

2

1

o

8

Ho

ur

s

be

fo

re

I l l I I I I l i I I r I l I I I

10

12

Ho

ur

s

af

te

r

34

35

5.8 Accident rate according to season

Earlier studies based on inventories of road conditions (Reference 12) showed an

increase in accident rate in icy/snowy road conditions in the early winter and late winter. The occurrence of slippery road conditions is not so frequent as during mid-winter. Furthermore, road users are not so used to winter road conditions and fewer cars use winter tyres.

The winter periodwas divided into three parts for a comparison of results.

- Autumn/early winter October and November 224 accidents

- Mid-winter December, January and February 639 accidents

- Spring/late winter March and April 78 accidents

The accident rate was then studied concerning the same variables as in earlier reports. The same basic pattern was observed, i.e. the maximum value appears one or a few hours before implementing the action.

In northern Sweden, the maximum value of the accident rate is 3.5 for early winter, 1 for mid-winter and 2 for late winter. This is well in accordance with the

experience above.

In central Sweden, the maximum value of the accident rate is 2 for early winter, 3 for mid-winter and 6 for late winter. In southern Sweden, the maximum value is 6

for all three periods. The reason why the pattern is not so distinct in these parts of the country may probably be explained by the fact that icy/snowy road conditions have been more or less equally unusual during each part of the mild winters 1988/89 and 1989/90.

Studies of the distribution among type of action show that the maximum values of

the accident rate are highest during late winter. The maximum values are double, 6

and 3 respectively, compared with mid-winter.

The maximum value of the accident rate is highest for single accidents, 2 for early

winter, 1.5 for midwinter and 2.5 for late winter.

36

5.9 Injury rate

The injury rate, number of persons injured per million axle pair km, was also studied.

Figure 5:13 shows the injury rate for all types of actions. Just as in the correspond-ing figure for accident rate, the injury rate also increases durcorrespond-ing the hours before the action to reach its maximum value 1 to 1.5 hours before the action was imple-mented. The figure reports an approximate confidence interval of 95% based on the uncertainty in the number of persons injured.

The injury rate and accident rate show a similar pattern when distributed among the variables: type of action, region, speed limit, type of road, road width, mainte-nance class and type of accident.

As an example, the injury rate and accident rate for salting and combined salting and snowploughing both have a maximum value which is twice as high as the value of anti-icing. As was the case with the accident rate, northern Sweden has a maximum value of the injury rate considerably lower than the corresponding

value for central Sweden, which in turn is lower than the maximum value for

southern Sweden.

Wildlife accidents are one exception to this pattern. The injury rate is considerably lower than the accident rate. The reason is naturally that wildlife accidents very seldom result in personal injury.

37

Figure 5:13 Injury rate (number of persons injured per million axle pair km)

twelve hours before and after action (all actions, 941 accidents). The broken line is a confidence interval of 95%.

VTI RAPPORT 399A

Ho

ur

s

af

te

r

Ac

ti

on

Ho

ur

s

be

fo

re

38

6 DISCUSSION

As was the case with earlier pilot studies, this analysis was based on action data (in this case from the VVH follow-up system), which were primarily produced to facilitate a simple follow-up. Consequently, there is a limited possibility of com-paring accident and ow data for accident risk calculations. Action data have neither been collected nor filed for a correct and simplified connection. Preferably, all data collected by the Swedish National Road Administration should be filed to facilitate multiprogramming with data such as accident and road records in the road data base. It is always difficult to decide the applications of a data base in advance, and which connections to other data bases it is interesting to make. The number of actions and the shortage of action data in the report meant that the accident material was not quite sufficient for a worthwhile report on the function of accident risk for all distributions (for example salting in northern Sweden on

roads with a speed limit of 90 km/h) as was planned. In most cases, it was,

how-ever, possible to discern one and the same basic structure of the function of

acci-dent risk.

The inaccuracy in data in uences the result, i.e. the maximum values of the

acci-dent rate are an underestimation of the real values. Compared with earlier

studies, both the increased amount and precision of action data above all resulted

in an increase in risk in this analysis. Consequently, the reported increases in acci-dent rates are to be regarded as minimum values of the increase. A more exact

registration of winter road maintenance actions, where time is linked to the place

of action, is desirable in order to perform more accurate analyses.

Satellite positioning of the vehicles in action offers possibilities of a very close analysis in the future.

Winter road condition in uences road users accident risk, trafficability

(speed/travelling time), vehicle costs etc. Winter road condition in turn is in u enced by the choice of action and time of implementation in combination with

actual weather conditions.

I This means that further studies should concentrate on both the connection between road condition and effects and the connection between action and road condition.

39

To be able to state the demands on the road surface standard for various types of roads and road networks, it is decisive for the road authority to know the

connec-tion between road condiconnec-tion and road user effects.

Since the winter of 1992/93, there has been a systematic follow-up of road condi-tion. A basis for profound studies of the in uence of road condition on road safety

should be available.

It is important for contractors and producers to know more about the in uence of various action strategies and efforts on the intended result, i.e. road condition. Anti-icing is a new method which is very interesting to study. In this study, promising results of the method have been obtained. The method is new and has just been developed, which means that there is a need for continued studies. For example, it is important to follow up when anti-icing is carried out, thus preventing unnecessary salting which may result in negative effects on the environment. Studies of other effects should be of great interest to both the road

authority and the road user. Effects of the methods on road condition, friction,

trafficability and road safety are examples of variables calling for closer studies.

Studies of road conditions and friction before and after road maintenance actions

should be of interest concerning every type of action, since knowledge is currently

rather limited.

It would obviously be interesting to make the same study in a few years not only, to improve accuracy but also to follow up possible changes/improvements con-cerning winter road maintenance.

40

REFERENCES

Schandersson, Rein. Samband mellan trafikolyckor, viiglag och vinter-vaghallningsatgarder. Olycksrisker vid olika mangd snonederbiird. (Relation between traffic accidents, road surface condition and winter maintenance actions. Accident risks at varying amounts of snow.) Statens V'ag- och Trafikinstitut, VTI Meddelande 514.

Schandersson, Rein. Samband mellan trafikolyckor, viiglag och vinter-vaghallningsatgarder. En pilotstudie av olycksriskens niva timmarna fore och efter atgard. (Relationships between traffic accidents, road sur-face conditions and winter maintenance measures. A pilot study of the accident risk level before and after measure.) Statens V'ag- och

Trafik-institut, VTI Meddelande 483.

Effektkatalog. Drift- och underhallsatgarder. (Effect catalogue. Road

Maintenance Measures.) V'agverket, 1989: 18.

Manual till vinteruppfoljningsprogrammet VVH. Version 1-89-03. (Manual to the VVH winter follow-up programme. Version 1 89-03) Vagverket, Bygg- och Driftproduktion.

Velin, Hans. Samband mellan vintervaghallning och tra ks kerhet. Programdokumentation och kvalitetskontroll av data (Relation between winter road maintenance and road safety. Documentation of programme and quality check of data.). Statens vag- och transportforskningsinstitut. Notat 73, 1994.

Schandersson, Rein. Tra kens veckovariation under aret. En

samman-stallning grundad pa Vagverkets tra kmatningar 1978-83. (Weekly variations of traffic during the year. Compilation based on the SNRA measurements 1978-83.) Statens Vag- och Trafikinstitut, VTI Notat T73. Tra karbetets forandring 1989-1990. Redovisning av tra karbetets forandring baserad pa data insamlade i VV system for uppfoljning av trafikforandringen. (Change in vehicle mileage 1989 1990. Report of change in vehicle mileage based on data collected in the SNRA s follow-up system concerning changes in traffic.) Vagverket 19912309.

Vader och vatten. Manatlig statistik over temperatur, nederbord och sno-t'acke. (Weather and water. Monthly statistics of temperatures, precipitation and snow cover.) Sveriges meteorologiska och hydrologiska institut. Third International Symposium on Snow Removal and Ice Control Technology. Sept 1992. Minnesota Transportation Research Board.

10 11 12 13 14 41

Hanke, H; Levin,Chr. Eingeschrankte Salzstreuung auf

Landstrassen-Auswirkungen auf Verkehrsablauf und Verkehrssicherheit Strasse

und Autobahn 39 (1988) Nr 10.

Andersson, Kjell. Kemisk halkbekampning. Effekt pa tra kolyckor. (Chemical de-icing of roads - effect on road accidents.) Statens Vag och Trafikinstitut, VTI Rapport 145.

Briide, Ulf; Larsson, Jorgen. Samband vintertid mellan Vaderlek-

vag-lag- trafikolyckor. (Wintertime interrelationsships between weather, road conditions and road accidents.) Statens Vag- och Trafikinstitut, VTI Rapport 210.

Cberg, Gudrun; Gustafson, Kent. VTI; Axelson, Lennart VV. Effektivare

halkbekampning med mindre salt. MINSALT-projektets huvudrap-port. (More effective de icing with less salt. Final report of the MINSALT project.) Statens Vag- och Trafikinstitut, VTI Rapport 369.

Nilsson, Goran. Olyckskvot som trafiksakerhetsmatt. Olyckskvotens variation under olika vaglags- och ljusforhallanden. (Accident rate as a measure of traffic safety. The variation of the accident rate in various road and lighting conditions.) Statens V'ag- och Trafikinstitut, VTI Rapport 73.

Appendix 1

Page 1(3)

Questionnaire for positioning actions

Relation between winter road maintenance actions and road safety

From the diskettes with action data (VVH), which the VTI obtained from you, we

have now compiled start and finishing points for all action sections.

Since these start and finishing points are given as place names, estate names and the like, we cannot interpret these to exact places in the road network. This inter-pretation must be continued with a study of when and where in the road network actions are implemented and accidents occur.

The study is based on the fact that we assume that the accident risk is high up to a

few hours before specific actions are taken, then decreases substantially when the

action has been implemented. Examples of such actions are snowploughing, combined salting and snowploughing, and salting. If, however, anti-icing is per formed in time, the slippery conditions will not occur. This is proved by the fact

that the accident risk is the same during the hours before and after anti-icing, when

the action was successful.

To be able to continue our work, we ask you to interpret the place names, estate names, etc. to points in the road network in the SNRA s reference system. The technique can be seen in the enclosed examples. State the junctions enclosing the section and how far from the first junction the section starts and ends. If the sec-tion starts or ends exactly at the juncsec-tion, only the juncsec-tion is reported and no distance is given. A junction is defined through the number of the map plus the point A number, e. g. 1342 A63. The names of the junctions (intersection, county boundary or end of road) can be seen from reference maps and lists of roads. If you have any questions, please do not hesitate to call one of the following persons at the Swedish Road and Transport Research Institute:

Peter Wretling 013 -204176

Hans Velin 013 204267

Hans Savenhed 013-204170

Thank you for your cooperation.

VTI RAPPORT 399A E . xa mp le 1' Sq ua re 85 2

HM

GZ

FM

f

91

0

Al B 1

93

Appendix 1 Page 2(3) T t / I U A Q E R G I 7 0 0 m

85

De no mi na ti on of se ct io n: At vi da be rg -H ac ke fo rs Ro ad num be r: 35 .0 0 Le ng th of se ct io n: 28 .5 No . of se ct io n: 1 FI NI SH : Ha ck ef or s T O F R O M T O Ma p Po in t A Ma p Po in t A Di st an ce Ma p

ST

AR

T:

At

vi

da

be

rg

F R O M M a p Po in t A Di st an ce o o o a o a o o a o a o a t n o o c o o o o o o o c o o a o n y o o o a o a I o c o o o a o I o o o a o a o o o o n o c n I o c o u n n u n c o n o o u c o o a n o c o o u . . o o c o n o o n o u o o a u . o u o n o -u o o c o ~ o o c o n o a . o o . o u u n o ¢ an u o n o o o o c o -n o o c n a o a o a a u o .

-VTI RAPPORT 399A Exa mp le 2: Sq ua re 84 3

A?

eR

oK

iN

D

. . . . _ I. I ' . -' " " " " _ _ -v& -. ~vO _ ~ Appendlx 1 Page 3(3)

3:

on Q Q

Q

U L L S T A M M A

41

No . of se ct io n: 4 Ro ad num be r: 68 7. 00 Le ng th of se ct io n: 23 .2 De no mi na ti on of se ct io n: Ul ls t' am ma -B ro ki nd S T A R T : Ul ls t' am ma FI NI SH : Br ok in d F R O M T O F R O M T O Ma p Po in t A Di st an ce Ma p Po in t A Ma p Po in t A Di st an ce Ma p Po in t A

37

/

4/

87

3

as

34

/0

87

3

a

;

o a o o o n n a o o o a o u c o o o I o o o o o o c a a o I o u o o o o c o o o a o n o o o o o o o o o o o a n a o o u o o n a a o o o g n a n o u o o o o o n u c o o o o o a o a u oo a a a o o o n a o a a a n o a n o n o o a a o o o o o o o o a a a o u o c o o o o n o o A a o a o o c a o o o I o o o o o o o o n

Appendix 2

Page 1(1)

Number of km with implemented actions and number of jobs

started, distributed among region and type of action

Kilometres with implemented actions

Type of Southern Central Northern Total

action Numb. % Numb. % Numb. % Numb %

Salting 7 n. 6,522 16 22,507 56 11,124 28 40,153 100 % 61 57 34 48 Combined n 734 6 6,636 55 4,656 39 12,026 100 plough. & % 7 17 14 15 salt Ploughing n. 1,337 5 6,939 28 16,764 67 25,040 100 % 12 18 50 30 Anti-icing n. 2,103 35 3,326 55 624 10 6,053 100 % 20 8 2 7 Total n. 10,696 13 39,408 47 33,168 40 83,272 100 % 100 100 100 100 Started jobs

Type of Southern Central Northern Total

action Numb. % Numb. % Numb. % Numb. %

Salting n. 1,548 19 4,272 53 2,291 28 8,111 100 % 59 58 32 47 Combined n. 173 8 1,065 53 782 39 2,020 100 plough. & % 7 15 11 12 salt. Ploughing n. 299 5 1,417 26 3,862 69 5,578 100 % 11 19 5-5 33 Anti-icing n. 614 45 622 45 139 10 1,375 100 % 23 8 2 8 Total n. 2,634 15 7,376 43 7,074 42 17,084 100 % 100 100 100 100

VTI RAPPORT 399A

Appendix 3 Page 1 (5)

Distribution of accidents concerning type of action and region

Type of action

Region Combined Salting Anti-icing Ploughing

Acc. % Acc. % Acc. % Acc. %

Southern 23 10.3 155 69.5 36 16.1 9 4.0

Central 94 18.5 334 65.9 42 8.3 37 7.3

Northern 37 17.5 95 45.0 8 3.8 71 33.6

Total 154 16.4 584 62.1 86 9.1 117 12.4

Maintenance Combined Salting Anti-icing Ploughing

class Acc. % Acc. % Acc. % Acc. %

MW + 4LR + 85 23.2 203 55.3 35 9.5 44 12.0 ADT 2 7,000 NR + 1,500 S 63 13.5 314 67.5 40 8.6 48 10,3 ADT < 7,000 Regional rs 6 5.5 67 61.5 11 10.1 25 22.9 Total 154 16.4 584 62.1 86 9.1 117 12.4

Road Combined Salting Anti-icing Ploughing

width Acc. % Acc. % Acc. % Acc. %

< 6 2 7.1 14 50.0 3 10.7 9 32.1 6-8.5 44 11.7 261 69.2 37 9.8 35 9.3 85-10 24 16.0 89 59.3 9 6.0 29 18.7 > 10 79 20.9 218 57.7 36 9.5 45 11.9 unknown 5 62.5 2 25.0 1 12.5 0 0 Total 154 16.4 584 62.1 86 9.1 117 12.4

Appendix 3

Page 2 (5)

Speed limit Combined Salting Anti-icing Ploughing

Acc. % Acc. % Acc. % Ace. %

50 11 18.6 33 55.9 8 13.6 7 11.9 70 28 16.6 97 57.4 13 7.7 31 18.3 70 R1 / 90 R2 4 25.0 10 62.5 1 6.3 1 6.3 90 80 14.9 355 66.0 48 8.9 55 10.2 90 R1 / 110 R2 1 100 0 0 0 0 0 0 110 30 19.0 89 56.3 16 10.1 23 14.6 Total 154 16.4 584 62.1 86 9.1 117 12.4

Type of Combined Salting Anti-icing Ploughing

road Acc. % Acc. % Acc. % Acc. %

2-lane road 116 15.0 498 64.3 69 8.9 92 11.9

motorway 15 28.3 21 39.6 8 15.1 9 17.0

express road 23 21.9 58 55.2 8 7.6 16 15.2

4-lane road 0 0 7 87.5 1 12.5 0 0

Total 154 16.4 584 62.1 86 9.1 117 12.4

Flows Combined Salting Anti-icing Ploughing

Acc. % Acc. % Acc. % Acc. %

< 500 1 7.1 4 28.6 2 14.3 ' 7 50.0 500-2,000 13 8.0 113 69.3 15 9.2 22 13.5 2,000-8,000 67 14.1 312 65.5 42 8.8 55 11.6 8,000-16,000 65 24.6 143 54.2 27 10.2 29 1 1.0 > 16,000 8 33.3 12 50.0 0 0 4 16.7 Total 154 . 16.4 584 62.1 86 9.1 117 12.4

Appendix 3

Page 3 (5)

Accident Combined Salting Anti-icing Ploughing

type

Acc. % Acc. % Acc. % Acc. %

cycl., pedest 12 10.7 74 66.1 9 8.0 17 15.2 miscel. single 80 14.1 375 66.3 54 9.5 57 10.1 turn., junct., 32 19.8 88 54.3 17 1.5 25 15.4 rear-end . head-on, 56 22.6 131 52.8 19 7.7 42 16.9 overtaking wildlife 80 11.8 437 64.5 62 9.1 99 14.6 Total 260 14.7 1105 62.6 161 9.1 240 13.6 Region

Maintenance class

Southern- .-

Central

Northern

Acc.

%

Acc.

%

Acc.

%

MW + 4LR + ADT .>_ 7,000 95 25.9 220 59.9 52 14.2

Nat + 1 500 S ADT < 7.000 93 20.0 236 50.8 136 29.2

Regional roads 35 32.1 51 46.8 23 21.1

Total 223 23.7 507 53.9 211 22.4

Road width Southern Central Northern

Acc. % Acc. % Acc. %

< 6 m 9 32.1 14 50.0 5 17.9 6 - 8.5 79 21.0 222 58.9 76 20.2 8.5 - 10 19 12.7 70 46.7 61 40.7 > 10 114 30.2 197 52.1 67 17.7 Unknown 2 25.0 4 50.0 2 25.0 Total 223 23.7 507 53.9 211 22.4

Appendix 3

Page 4 (5)

Speed limit Southern Central Northern

Acc. % Acc. % Acc. %

50 15 25.4 30 50.8 14 23.7 70 47 27.8 88 52.1 34 20.1 70 R1 / 90 R2 8 50.0 8 50.0 0 0 90 93 17.3 316 58.7 129 24.0 90 R1 / 110 R2 0 0 1 100.0 0 0 110 60 38.0 64 40.5 34 21.5 Total 223 23.7 507 53.9 211 22.4

Type of road Southern Central Northern

Acc. % Acc. % Acc. %

Two lane road 162 20.9 426 55.0 187 24.1

Motorway 11 20.8 31 58.5 11 20.8

Express road 49 46.7 43 41.0 13 12.4

Four-lane road 1 12.5 7 87.5 0 0

Total 223 23.7 507 53.9 211 22.4

Flows Southern Central Northern

Acc. % Acc. % Acc. %

< 500 5 35.7 6 42.9 3 21.4 500-2,000 49 30.1 85 52.1 29 17.8 2,000 8,000 109 22.9 228 47.9 139 29.2 8,000 16,000 59 22.3 165 62.5 40 15.2 > 16,000 1 4.2 23 95.8 0 0 Total 223 23.7 507 53.9 211 22.4

Appendix 3 Page 5 (5)

Accident type Southern Central Northern

Acc. % Acc. % Acc. %

Cycl., ped., miscel. 25 22.3 55 49.1 32 28.6

Single 165 29.2 297 52.5 104 18.4

Turn, junct, rear-end 34 21.0 94 58.0 34 21.0

Head-on, overtaking 43 17.3 137 55.2 68 27.4

Wildlife 91 13.4 437 64.5 150 22.1

Total 358 20.3 1,020 57.8 388 22.0

Sp

ee

d

li

mi

t

70

km

/h

Ac

ci

de

nt

ra

te

Appendix 4 Page 1(8)

Accident rate for type of action, distributed among speed limit,

road width, traffic class and maintenance class

C) U) E E .C .C C) O) C) CD C 2 t '5 2

o

0- 3

U3

'

Q-2:-_O 3_{O 'O} .E_{+_-* 2.:}I 3_O (U C c (U C C .._ U) (U (I) < U) 0 q; : |._ CO LO <l' 00 N

VTI RAPPORT 399A

Ho

ur

s

af

te

r

Ac

ti

on

Ho

ur

s

be

fo

re

VTI RAPPORT 399A

Ac

ci

de

nt

ra

te

6

i

Sp

ee

d

lim

it

11

0

km

/h

Typ

e

of

ac

ti

on

Sn

owp

lo

ug

hi

ng

"""

"""

"""

an

d

sa

lt

in

g

m

Sa

it

in

g

Appendix 4 Page 2(8)

m

An

ti

ic

in

g

Sn

owp

io

ug

hi

ng

i

1

2

._ i . HI RJ F. 9 . . a ; I . , I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I T I l I I r

1

0

8

6

4

2

o

2

4

6

8

10

12

Ho

ur

s

be

fo

re

Ac

ti

on

Ho

ur

s

af

te

r

VTI RAPPORT 399A

Ac

ci

de

nt

ra

te

Ro

ad

wi

dt

h

6

m

to

8.

5

m

6?

4

Typ

e

of

ac

ti

on

Sn

owp

lo

ug

hi

ng

m

an

d

sa

lt

in

g

3

{

w'

''

"

Sa

lt

in

g

Appendix 4 Page 3(8)

m

An

ti

ic

in

g

,... .... .... Showp loug hing l i I I I i Y I I 1 i I I I I I l I I l r I I I I 1 I I I l l I I l l I f l I I T T l I I l l I

1

2

1

0

-8

6

4

2

o

2

4

6

8

10

12

Ho

ur

s

be

fo

re

Ac

ti

on

Ho

ur

s

af

te

r