Technical Note for Road Safety Infrastructure Management

download Technical Note for Road Safety Infrastructure Management

of 12

Transcript of Technical Note for Road Safety Infrastructure Management

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    1/12

    Technical NoteFor Road Safety Infrastructure Management-

    Bina Marga

    By

    Tri TjahjonoDepartment of Ciil !ngineering

    "niersity of IndonesiaDepo# $%&'& IND(N!SI)

    $* Introduction

    The issue to be addressed in this technical note is to determine themaximum number of accidents that could be reasonably toleratedafter road improvement programme in Indonesia concluding that astatistically signicant increase or vive versa has occurred. Thequestion will be examined for 2 levels of accident conditions i.e. thenumber of accidents and the severity levels and for 3 dierent timeperiods following improvement i.e. month! " months and annualmonitoring.

    It should be noted that currently tra#c accident database producedby the police is under reporting in particularly for number of in$uries

    and damage only accidents. Therefore the most reliable peace ofinformation should be the number of fatalities that can be correctedthrough a combination of information among police! hospital andinsurance information. %owever! The Indonesia Tra#c &olicecurrently is proposing the Integrated 'oad (afety )anagement(ystem *I'()(+ under the (trategic 'oads infrastructure &ro$ect.Through this pro$ect improvement of tra#c accident database inIndonesia can be expected in the near future.

    '* Basic )nalysis for Monitoring Road Safety

    Indicators'*$* Method

    The general approach is to predict the expected number ofaccidents in the period following implementation if roadimprovements were not introduced and provide the upper ,-condence interval limit for this estimate. If this limit is exceeded bythe actual *observed+ accidents following implementation of thepro$ect then it is reasonable to assume that the pro$ect hassignicantly reducing the accident rate or vice versa. / number of

    caveats go with this conclusion! which are summarised at the end ofthe note.

    01/08/2007 1

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    2/12

    0ollowing implementation! it will then be possible to loo1 moreclosely at the dierences between the observed and expectednumber of accidents and also to ta1e into consideration a widerrange of bac1ground factors than is possible apriori. The appendix

    to this note gives a detailed methodology and illustration of how theactual and observed may be compared! alongside the calculation ofa tra#c safety index of eectiveness.'*'* )ssumptions

    The number of accidents is assumed to follow a &oisson distribution*for rare events+! however! where the accident rate *expectednumber of accidents for a specic period+ exceeds 4! anapproximation to the standard normal distribution is made.

    /n ad$ustment is made to the expected number of accidents ifpro$ects were not introduced for the growth in tra#c 5ow! as thegrowth in tra#c would be expected to increase accidents. 0ourgrowth rates are examined i.e. 4! 2! -! 6 and it should benoted that these are assumed to apply to the period under reviewfollowing implementation. 7learly these are approximate and maynot be appropriate to every period in question. 0ollowingimplementation an estimate of tra#c growth should becomeavailable.

    '*+* Notation

    The following notation applies to the formulae8

    9 the number of accidents in the after period if pro$ects were notintroduced. This is estimated by

    9 the observed number of accidents in the after period when

    pro$ects is introduced. This is estimated by rtf9 the tra#c 5ow rate in the review period with respect to the rate

    prior to implementation. 0or example .42 represents a 2increase in tra#c

    rd 9 period of review after implementation relative to the totalperiod of prior observation. 0or example! :3" represents month after implementation relative to 3 years of priorobservation.

    ; 9 the total number of accidents of a particular group observed inthe before period

    '*&* Calculation of the upper ,. con/dence limit

    The estimate of the number of accidents in the after period ifpro$ects were not introduced is given by8

    9 rd .rtf. ;

    01/08/2007 2

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    3/12

    The ,- 7I for is given by8

    +

    +

    m

    n

    eP

    !

    -%95)]mn,-([

    'ather than calculate the cumulative function! standard statisticaltables can be used to ascertain the upper limit for a given*approximate+ mean rate.

    7onsidering the total number of accidents in all categories *forexample "2 in the 3 year period+. The following results areobtained8

    $ month follo0ing implementation 1rd2 $3+%4

    Tra#c 5owgrowthassumed

    Tra#c 5owrate! *rtf+

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    4/12

    It should be noted that these results must be viewed with caution!as they ta1e no account of a number of sources of variation!including seasonality! weather conditions! special events and thepossibility of a >frea1? occurrence! such as a bus accident! giving

    in5ated actual numbers.

    It is strongl r!"omm!n#!# t$%t % r!%son%&l! monitoring '!rio# t%(! 'l%"! %)t!r

    im'l!m!nt%tion &!)or! #r%*ing %n )irm "on"l+sion on s%)!t !))!"ts o) t$! 'ro,!"t

    im'l!m!nt%tion-

    Summary of !7pected num8er of accidents 8y seerity and,. upper limit 8y reie0 period

    Tra#c@rowth

    Fatalities

    A month A " months A 2 months

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    5/12

    point 'a'

    point 'c'

    point 'b'

    120

    140

    160

    180

    200

    220

    -6 -5 -4 -3 -2 -1 0 1 2 3 4

    Time (Years)

    NumberofAccidents

    'before' 'after'

    (ituation 2. To estimate the safety of particular road orsegment in the >after? period if pro$ect has beenimplemented *point >c? in 0igure +.

    (ituation re5ects >the do nothing? situation and situation 2 re5ects

    >the do something? situation. / valid comparison can be carried outbetween these two situations since they have the same time frame.The terminology of >before? situation may be misleadingC it is notappropriate to ma1e a comparison on safety between before pro$ectis introduced *point >a?+ and after it is introduced *point >b?+. 'atherthe comparison should be made between the safety indicators ofthe particular road or segment after pro$ect has been implementedwith the estimated safety of particular road or segment if pro$ecthad not been introduced. 0igure shows these three point positions.7omparing between points >a? and >b? would lead to bias in theresult since changes might be occurred between these two timeframes. /t the least tra#c 5ows would have changed over time.

    ig+r! 1 !)or! %n# )t!r 4&s!r%tion%l 6%)!t 6t+#

    +*'*'* Four Step )pproaches

    This simple method is based on the method suggested by %aeur

    *,,6+. 0our steps are advised8

    Step 1. Estimate and predict the number of accidents in thebefore and after implementing the projectafter? period *+! or >the do something situation? and predictthe number of accidents of the entity in the >after? period if no

    pro$ect had been applied *+ or >the do nothing situation?.

    after? period. &rediction of can bedone by assuming the number of accidents will follow the

    01/08/2007 5

    Before

    Project

    AfterProject

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    6/12

    nature in the >before? period sub$ect to correction for changesin tra#c 5ow and the period of >before? and >after?observation.

    The reduction *or increase+ in the expected number of

    accident *+ can be obtained by equation *giving a positivesign for a reduction and negative sign for increasing numberof accidents+.

    = *+

    Step 2. Estimate the variances/ssuming that the estimated values of accidents in the>before? and >after? trepro$ectent * and + are statisticallyindependent! then the variance of variables follows by

    equation 2.

    783:783:783: += *2+

    Step 3. Measurement of the safety eectDne of the ways of measuring the safety eect is to introduce

    the index of eectiveness *+ and this is dened as :.

    Therefore! = /; . %owever! in order to have a statisticallyunbiased estimator for the following equation is usedinstead8

    ]/893:1/[)/(

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    7/12

    +

    +

    m

    n

    !

    %95)]mn,-([> *-+

    where! m)]n,-([> + is the probability mass function of

    accidents! *n+ and *Am+ are the minimum and maximum valuesfor the mass function to be signicant at approximately ,-! and

    is the expected number of accidents. Dften these are the targetaccidents.

    +*'*&* Correction Factors

    It would be naEve to assume that any reduction *or increase+ inaccidents in the after *>treated?+ period would be caused solely bythe >trepro$ectent?. /t least! two factors need to be corrected i.e.8

    the duration between the >before? and >after? period and the tra#c5ow.

    The process for these corrections is describes as follows for dierentpossibilities8. Foth periods of same duration!

    Tra#c 5ow and >remaining factors? are the same in both periods!

    then 9 2. The >before? and >after? periods are in the dierent duration!

    Tra#c 5ow and >remaining factors? are the same in both periods!

    then 9 rd

    3. The >before? and >after? periods and tra#c 5ow are dierent!

    >'emaining factors? are the same in both periods! then 9 rd rtf

    It is important to note that one of the properties of the &oissondistribution is that the variance and the mean should beapproximately equal.

    0inally! estimation of parameters and variances are given byequation " and 68

    &arameter8 ? = ! then ?8:39 =*"+

    &arameter8 rr t#= ! then ]r8:39)r[()(r8:39 t22

    t

    2

    # +=

    *6+

    +*+* Numerical !7ample

    Table shows the number of accidents and severity on study area*observed road segment+ in the 3year >before? period.

    01/08/2007 7

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    8/12

    Table . Gumber of accidents and severity on the case study road)ccidentseerity

    :ear ;+ :ear -' :ear ;$ Total

    0atal 4 2 3 -

    (erious 6 6 2 ";(I 6 , - 2

    (light -= , 3 Total "- -= 3, "2

    Fy way of illustration of the methodology! assuming that followingimplementation of pro$ect! in the rst! sixth and twelfth months!number of accidents is shown in Table 2.

    Table 2 Gumber of /ccidents and severity after implementation of

    the pro$ect)ccidentseerity

    Month

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    9/12

    78:39 2 *:3"+2J.4-2"2A"22 4.42K 9 4.-accidents:month

    5.-=5.-0=8:39 =+= accidents:month

    =-5.8@ = 9 2.-"

    (T

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    10/12

    A" 2 3" 26.- =."6.3 Go .264.2,

    - 3" 2=.3- 6."-6.- Go .24.2=

    6 3" 2=.=, 6.6.6 Go .224.26

    A2 2 "= --.4=

    2.,22.4

    Go .244.2

    - "= -".64

    .342.4

    ,

    Go .64.23

    6 "= -6.6=

    4.222.

    2

    Go .-4.22

    Table . (ummary of ;(I /ccidents)onth

    Tra#c@rowth

    Gumber of /ccidents (ignicantMierence

    )A5s(

    Index ofafter?

    A>before?

    Mierences

    B

    A 2 3 4."4 2..66 Go 3."2.23

    - 3 4." 2.3,.66 Go 3.-2."

    6 3 4."2 2.3=.66 Go 3.62.2

    A" 2 " 3.-6 2.3.-6 Nes 3.23.-"

    - " 3."= 2.33." Nes 3..-2

    6 " 3.6- 2.2"."3 Nes 3.4,.,

    A2 2 34 6. 22.="6.4- Nes 3.43.2

    - 34 6.3- 22."-6.3 Nes 2.,-.3=

    6 34 6., 22.-6.= Nes 2.=,.3-

    Table -. (ummary of (lightly In$ured /ccidents)onth

    Tra#c@rowth

    Gumber of /ccidents (ignicantMierence

    )A5s(

    Index ofafter?

    A>before?

    Mierences

    B

    A 2 3 .44 .44.=- Go 4.634.3

    - 3 . ..=" Go 4.64.2

    6 3 .2, .,.=" Go 4.644.

    A" 2 24 23.,6 3.,6-.,= Go 4.=4.22

    - 24 2."= ."=".44 Go 4.6,4.2

    6 24 2-.- -.-".42 Go 4.6=4.2

    A2 2 3= 6., ,.,4.4- Go 4.664.6

    - 3= ,.3- .3-4.4

    Go 4.6-4.6

    6 3= -4.2, 2.2,4.

    Go 4.64."

    Disclaimer

    This method result should be treated with caution since simple>before? and >after? observation does not address all aspects that

    can in5uence the nature of road accidents.

    01/08/2007 10

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    11/12

    %auer *,,6+ quoted that

    . 7hange in tra#c safety re5ects not only the eect of pro$ectbut also the eect of factors such as tra#c! weather! driverbehaviour as well as changes in the way that accidents are

    reported *particularly damage only accidents+. /s a result! it isnot possible to conclude how much of the change can beattributed to the pro$ect system.

    2. 7hange in safety may be result from the regressiontomeanphenomenon and not result from the pro$ect system.

    It is possible to address these two issues by employing multivariategeneralised linear models *point + in con$unction with

  • 8/11/2019 Technical Note for Road Safety Infrastructure Management

    12/12

    MONITORING OF TOTAL NUMBER OF AI!ENT"+1 +1 +1 +6 +6 +6 +12 +12 +12

    Traffic Growth 2 5 ! 2 5 ! 2 5 !

    "TE# $ Estimate number of accidents %before% and %after% treatment

    n"#ber of acci$ent%6 6 6 36 36 36 68 68 68

    162 162 162 162 162 162 162 162 162

    $"ration ob%er&ation

    before 36 36 36 36 36 36 36 36 36

    after 1 1 1 6 6 6 12 12 12

    ratio 003 003 003 01! 01! 01! 033 033 033

    traffic f(ow

    )rowth 102 105 10! 102 105 10! 102 105 10!

    *A 002 002 002 002 002 002 002 002 002

    e%ti#ate of n"#ber of acci$ent%

    ,a#b$a 6 6 6 36 36 36 68 68 68

    phi 45 4!3 482 2!54 2835 288 5508 56!0 5!!8

    $e(ta -141 -128 -11 -846 -!65 -!11 -122 -1130 -1022

    "TE# & Estimate of 'ariances

    *A ,a#b$a 6 6 6 36 36 36 68 68 68

    *A phi 054 054 055 126 154 1!3 !!05 !81! !83

    *A $e(ta 654 654 655 5526 5554 55!3 14505 1461! 1463./ $e(ta 256 256 256 !43 !45 !4! 1204 120 1212

    "TE# Measurement of safet effect (Inde* of Effecti'eness)

    theta 12! 124 122 12! 124 122 120 11! 115

    "TE# + Estimate of Inde* of Effecti'eness ,ariance

    *A theta 030 028 02! 008 008 00! 006 005 005

    ./ *A theta 054 053 052 02 028 02! 024 023 022

    "TE# - .ec/in "tatistica00 "i1nificant !ifferent

    After 6 6 6 36 36 36 68 68 68

    Before 5 5 5 28 28 2 55 5! 58

    no no no no no no no no no

    onth

    .tati%tica(( /ifferent

    ''after' (a#b$a

    ''before' phi

    MONITORING OF 2"I AI!ENT"+1 +1 +1 +6 +6 +6 +12 +12 +12

    Traffic Growth 2 5 ! 2 5 ! 2 5 !

    "TE# $ Estimate number of accidents %before% and %after% treatment

    n"#ber of acci$ent%

    3 3 3 16 16 16 30 30 30

    21 21 21 21 21 21 21 21 21

    $"ration ob%er&ation

    before 36 36 36 36 36 36 36 36 36

    after 1 1 1 6 6 6 12 12 12

    ratio 003 003 003 01! 01! 01! 033 033 033

    traffic f(ow

    )rowth 102 105 10! 102 105 10! 102 105 10!

    *A 002 002 002 002 002 002 002 002 002e%ti#ate of n"#ber of acci$ent%

    ,a#b$a 3 3 3 16 16 16 30 30 30

    phi 060 061 062 35! 368 3!5 !14 !35 !4

    $e(ta -241 -23 -238 -1243 -1233 -1226 -2286 -2265 -2251

    "TE# & Estimate of 'ariances

    *A ,a#b$a 3 3 3 16 16 16 30 30 30

    *A phi 014 014 015 43 521 540 1!1 2083 215

    *A $e(ta 314 314 315 203 2121 2140 4!1 5083 515

    ./ $e(ta 1!! 1!! 1!! 45! 461 463 !05 !13 !18

    "TE# Measurement of safet effect (Inde* of Effecti'eness)

    theta 364 354 34! 323 314 30 303 25 28

    "TE# + Estimate of Inde* of Effecti'eness ,ariance

    *A theta 45 468 451 244 230 222 201 18 182

    ./ *A theta 223 216 212 156 152 14 142 138 135

    "TE# - .ec/in "tatistica00 "i1nificant !ifferent

    After 3 3 3 16 16 16 30 30 30

    Before 1 1 1 4 4 4 ! ! !

    no no no e% e% e% e% e% e%

    onth

    .tati%tica(( /ifferent

    ''after' (a#b$a

    ''before' phi

    MONITORING OF "LIG3TLY IN4URE! AI!ENT"+1 +1 +1 +6 +6 +6 +12 +12 +12

    Traffic Growth 2 5 ! 2 5 ! 2 5 !

    "TE# $ Estimate number of accidents %before% and %after% treatment

    n"#ber of acci$ent%

    3 3 3 20 20 20 38 38 38

    141 141 141 141 141 141 141 141 141

    $"ration ob%er&ation

    before 36 36 36 36 36 36 36 36 36

    after 1 1 1 6 6 6 12 12 12

    ratio 003 003 003 01! 01! 01! 033 033 033

    traffic f(ow

    )rowth 102 105 10! 102 105 10! 102 105 10!

    *A 002 002 002 002 002 002 002 002 002

    e%ti#ate of n"#ber of acci$ent%

    ,a#b$a 3 3 3 20 20 20 38 38 38

    phi 400 411 41 23! 2468 2515 4!4 435 502

    $e(ta 100 111 11 3! 468 515 4 1135 122

    "TE# & Estimate of 'ariances

    *A ,a#b$a 3 3 3 20 20 20 38 38 38*A phi 044 044 045 15!3 1601 1620 621 6403 64!

    *A $e(ta 344 344 345 35!3 3601 3620 1001 10203 102!

    ./ $e(ta 185 186 186 58 600 602 1005 1010 1014

    "TE# Measurement of safet effect (Inde* of Effecti'eness)

    theta 0!3 0!1 0!0 081 0! 0!8 0!! 0!5 0!4

    "TE# + Estimate of Inde* of Effecti'eness ,ariance

    *A theta 018 01! 01! 005 005 004 003 003 003

    ./ *A theta 043 042 041 022 021 021 01! 01! 016

    "TE# - .ec/in "tatistica00 "i1nificant !ifferent

    After 3 3 3 20 20 20 38 38 38

    Before 4 4 4 24 25 25 48 4 50

    no no no no no no no no no

    onth

    .tati%tica(( /ifferent

    ''after' (a#b$a

    ''before' phi

    Appendix

    01/08/2007 12