Está en la página 1de 25

r

10-14

HIGHWAY SAFETY MANUAL

10.6. SAFETY PERFORMANCE FUNCTIONS


In Slep 9 ofthe prediclive method, lhe appropriale safety performance funclions (SPFs) are used lO predicl average
crash frequency for lhe selecled year for specific bas~ conditions. SPFs are regression models for eslimaling lhe predicted average crash frequency of individual roadway segments or intersections. Each SPF in the predictive method
was developed with observed crash data [or a set of similar sites. The SPFs, like aH r~gressionrnodels, estimate the
value ofa dependenl variable as a function ofa sel ofindependenl variables. In lhe Sl>Fs developed for lhe HSM, lhe
dependent variable estimated is the predicted average crash frequency [or a roadway segment or intersection under
base condilions and lhe independenl variables are the AADTs oflhe roadway segmenl or inlersectjon legs (and, for
roadway segmenls, lhe length ofthe roadway segmenl).
The SPFs used in Chapler 10 were originally formulaled byVogt and Bared (13,14,15). A few aspecls oflhe
Harwood el al. (5) and Vogl and Bared (13, 14, 15) work have been updaled lO malch recenl changes lo lhe crash
prediclion module ofthe FHWA lnteractive Highway Safety Design Model (3) software. The SPF coefficienls, default crash severity and collision type distribulions, and defaull nighttime crash proportions have been adjusled lo a
consislenl basis by Srinivasan el al. (12).
The predicled crash frequencies for base condilions are calculaled from lhe prediclive models in Equations 10-2
and 10-3. A delailed discussion ofSPFs and lheir use in lhe HSM is presented in Chapler 3, Seclion 3.5.2, and lhe
Part C-Introduclion
and Applications Guidance, Seclion C.6.3.

Each SPF al so has an associated overdispersion parameter, k. The overdispersion parameter provides an indication of
lhe slalislical reliability of lhe SPE The closer lhe overdispersion parameler is lo zero, lhe more slalislically rehable
lhe SPE This pararneler is used in lhe EB Method discussed in Part C, Appendix A. The SPFs in Chapler lOare summarized in Table 10-2.
Table 10-2. Safely Performance Funclions included in Chapler lO
Chapter 10 SPFs for Rural Two-Lane, Two-Way Roads
Rural two-Iane, two-way roadway segments

SPF Equations

and Figures

Equation 10-6, Figure 10-3

Three-Ieg stop controlled intersections

Equation 10-8 , Figure 10-4

Four-Ieg stop controlled intersections

Equation }0-9, Figure 10~5

Four-Ieg signalized intersections

Equation 10-10, Figure 10-6

Sorne highway agencies may have performed stalislically-sound

sludies lO develop their own jurisdiclion-specific

S~Fsdenved fr,om.local conditions and crash experience. These models may be substituted for models presented in
lhls chapler. Cnlena for lhe developmenl of SPFs for use in the prediclive melhod are addressed in lhe calibralion
procedure presenled in Appendix A lo Part C.

10.6.1. Safety Performance Func1ions for Rural Two-Lane. Two-Way Roadway Segments
The predlcllve model for predicling average crash frequency for base condilions on a particular rural two-Iane
lWo-way roadway segmenl was presenled in Equalion 10-2. Tbe effecl oftraffic volume (AADT) on crash freuency
IS mcorporaled lhrough an SPF, whlle lhe effecls of geomelric design and lraffic conlrol fealures are incorporaled
lhrough lhe CMFs.
The base conditions for roadway segments on rural two-lane, two-way roads are:

.1

Lane widlh (LW)

12 feel

Shoulder widlh (SW)

6 feel

Shoulder type

Paved

.. -.-

-''!::-

i:
CHAPTER ,Q-PREDICTIVE

METHOD FOR RURAL TWO-LANE.

TWO.WAY

ROADS

lO.' S

Roadside hazard rating (RHR)

Driveway density (DD)

5 driveways per mile

Horizontal curvature

None

Vertical curvarure

None

Centerline rumble strips

None

Passing lanes

None

~wo-way left-turn lanes

Nane

Lighting

None

Autamated speed enforcement

None

Grade Level

0% (see note below)

A zero pereent grade is not allowed by most sta tes and presentS issues sueh as drainage. The SPF uses zero pereent
. as a numerieal base eondition that must always be modified based on the actual grade.
The SPF far predicted average crash frequency for rural two-Iane two-way roadway segments is shawn in Equation
10-6 and presented graphieally in Figure 10-3:
l

N
spfn

= AADT

x L x 365 x 10" x

e'-O.3I2)

(10-6)

Where:
predicted total erash frequeney for roadway segment base eonditions;
AADT

average annual daily traflie volume (vehieles per day); and

length of roadway segment (miles).

Guidanee on the estimation of traflie volumes for roadway segments for use in the SPFs is presented in Step 3 of
the predietive method deseribed in Seetion 10.4. The SPFs for roadway segments on rural two-lane highways are
appleable to the AADT range from zero to 17,800 vehicles per day. Applieation to sites with AADTs substantially
outside this range may not provide reliable results.

10-16

HIGHWAY 5AFETY MANUAL

""
5

..!

~
~

"

c.
>.
u
s::

":.,.J

...

-'~
"
j

~
U
."

1,

'ti

t:
"

5,000

10,000

15,000

AAOT (veh/day)

Figure 10.3, Graphical Form ofSPF for Rural Two-Lane, Two-Way Roadway Segments (Equation 10-6)

"

"

The value of the overdispersion parameter associated with the SPF for rural twa-lane, twa-way roadway segments is
determined as a function of the roadway segment length using Equation 10-7. The closer the overdispersion parameter is to zero, the more statistically reliable the SPF. The value is detennined as:
~
k ~ 0.236
L

(10-7)

Where:

,~.
"

l, "
1:

overdispersion

parameter; and

length ofroadway

segment (miles).

Tables 10-3 and 10-4 provide the default proportions for crash severity and for collision type by crash severity level,
respectively. These tables may be used to separate the crash frequencies from Equation 10-6 into components by
crash severity level and collision type. Tables 10-3 and 10-4 are applied sequentially. First, Table 10-3 is used to
estimate crash frequencies by crash severity level, and then rabIe 10-4 is used to estmate crash frequencies by.collision type for a particular crash severity level. The default proportions for severity levels and collision types shown
in Tables 10-3 and 10-4 may be updated based on local data for a particular jurisdiction as part ofthe calibration
process described in Appendix A to Par! C.

... !

CHAPTER lG--PREDICTIVE

METHOD FOR RURAL TWO-LANE, TWO-WAY

10-17

ROADS

Table 10-3. Default Distribu!on for Crasb Severity Level on Rural Two-Lane, Two-Way Roadway Segments
Percentage

Crash Severity Level

ofTotal

Roadway Segment Crashes.


1.3

Fatal
Incapacitating

5.4

Injury

Nonincapacitating

10.9

injury

14.5

Possible injury

32.1

Total fatal plus injury

67.9

Property damage only

100.0

Total
Note: Accident severity distributions are estimated far rural two-Iane roadway segments in Exhibt 10-4.
Based on H51S data far Washington

(2002-2006)

Table 10-4. Default Distribu!on by Collision Type for Speeifie Crasb Severity Levels on Rural Two-Lane, Two-Way
Roadway Segments
Percentage

Total Fatal and lnjury

Collision Type
SINGLE-VEHICLE

orTotal

Roadway Segment

Property

Crashes

Damage Only

by Crash Severity Level"


Total (AH Severity Levels Combined)

CRASHES

Collision with animal

3.8

18.4

12.1

Collision with bicycle

0.4

0.1

0.2

Collision with pedestrian

0.7

0.1

0.3

3.7

1.5

2.5

Overtumed

54.5

50.5

52.1

Ran offroad
Other single.vehicle

crash

0.7

2.9

2.1

Total single.vehicle

crashes

63.8

73.5

69.3

Angle collision

10.0

7.2

8.5

Head.on collision

3.4

0.3

1.6

16.4

12.2

14.2

Rear-end collision
Sideswipe collisionb

3.8

3.8

3.7

MULTIPLE-VEHICLE

CRASHES

Other multiple.vehicle

collision

2.6

3.0

2.7

Total multiple.vehicle

crashes

36.2

26.5

30.7

100.0

100.0

100.0

Total Crashes

~Based on HSISdata for Washington (2002.2006)


b lndudes approximately
70 percent opposite-direction

sideswipe collisions and 30 percent same-direction sideswipe collisions

10.6.2. Safety Performance


Functions for Intersections
The predictive model for predcting average crash frequency at particular rural two-lane, two-way road intersections
was presented in Equation 10-3. Tbe effeet of tbe majar and minor road traffie volumes (AADTs) on erasb frequeney
is incorporated through SPFs, while the effects of geometric design and traffic control features are incorporated
through the CMFs. The SPFs for rural two-lane, two-way higbway intersections are presented in tbis section.

HIGHWAY SAFETY MANUAL

10-18

SPFs have bee~ developed for three types of intersections on rural two-lane, two-way road . The tbree types ofintersections

are:

,1

Tbree-leg intersections with minor-road stop contror(3ST)

Four-Ieg int~.rsections with minor-road stop control (4ST)

Four-Ieg signalized intersections (4SG)


SPFs for lbree;leg signalized intersections on rural two-Iane, two-way roads are nol available. Other types ofinlersections may be found on rural two-Iane, two-way highways bul are nol addressed by these procedures.

The SPFs for each ofthe intersection types listed above estimates lotal predicted average crash frequency for
intersection-related crashes within the limits of a particular intersectiorr and on the intersection legs. The distinction
between roadway segment and inlersection crashes is discussed in Section 10.5 and a detailed procedure for distinguishing between roadway-segment-related and inlersection-related crashes is presenled in Section A.2.3 in Appendix A to Par! C. These SPFs address interseclions lhal have only two lanes on bolb the major and minor road legs,
nol including turn lanes. The SPFs for each of lhe tbree inlersection types are presented below in Equations 10-8,
10-9, and 10-10. Guidance on the estimation of traffic volumes for lbe major and minor road legs for use in lbe SPFs
is presented in Seclion 10.4, Step 3.
.

1',,'

The base conditions which apply lo lhe SPFs in Equations 10-8, 10-9, and lO-lO are:

Intersection .skew angle

I
1,. I
1I

Intersection left-turn lanes

Nane on approaches without stop control

Intersection right-turn lanes

None on approaches without stop control

Lighting

Nane

I ~
1

Three-Leg

1;

Stop-Controlled

Interseclions

The SPF for tbree-Ieg stop-controlled intersections is shown in Equation 10-8 and presented graphically in Figure! 04.
N'P/JST

exp[-9.86

(10-8)

Where:

!1.
"

estimate o~ intersec:ion-related
controlled mtersectIons;

NSPfJST

,
1

'!I

+ 0.79 x In(AADT ) + 0.49 x In(AADTm,)]

J ,
'1" ,

predicted average crash frequency for base conditions

AADTmPj

AADT (vehicles per day) on the major road; and

AADTm,.

AADT (vehicles per day) on the minor road.

for three-Ieg stop-

The overdispersion parameter (k) for this SPF is 0.54. This SPF is applicable to an AADT . range from zero to
19,500 vehicles per day and AADT mi. range from zero to 4,300 vehicles per day. Applicati~P~to sites with AADTs
substanllally outslde lhese ranges may not provide reliable results.
.

I
,I

I'

:
)I

j
[

1'1
~i "

I,
I

CHAPTER 1o-PREDICTIVE

METHOD FOR RURAL TWO.LANE,

TWO.WAY

10.19

ROADS

10
9
8

AAOTrn1n=4,300
AADTmir\=4,000

7
AADTrnir\=3,000

~
c:

GI

"

C'
GI

.s:
~

AADTmin=2,OOO

5
AAOTmin=l,OOO

<-g

'1::
'6
GI
~

"-

3
2

o
5,000

10,000
AAOT m.Jo'

15,000

(veh/day)

Figure 10-4, Graphical Representalion ofthe SPF forThree-leg Slop-controlled (3ST) Inlersections (Equalion 10-8)

Four-Leg Slop-Controlled Intersections


The SPF for four-leg slop controlled inlersclions is shown in Equation 10-9 and presenled graphically in Figure 10-5.
N,pf4ST

exp[-8.56 + 0,60 x In(AADT"o) + 0.61 x In(AADTmJJ

(10-9)

Where:
=

estimate ofintersection-related

predicted average crash frequency for base conditions for four-leg stop

controlled intersections;

AADTmOj = AADT (vehicles per day) on lhe major road; 'and


AADT ,
mi

AADT (vehicles per day) on the minor road,

The overdispersion pararneler (k) for.lhis SPF is 0,24, This SPF is applicable lO an AADT moj range from zero lo
14,700 vehicles pr day and AADT mi, range from zero lo 3,500 vehicles per day. Applicalion lo siles with AADTs
substantially outside these ranges r:naynot provide .accurate rsults.

10-20

HIGHWAY 5AFETY MANUAL

10
9

AADTmin=3,500

AAOTmin=3,OOO

8
7

...'"<:

AAOTmin=2,OOO

::l

<T

.t:

'"~

AADTmin=l,OOO

"'C

1;;

'6
~

3
2

, I ~j'

Ji ,,'

5,000

10,000

15,000

AAOTm.jo, (veh/day)

'1; .'11'
l"k

,:Ir ,';1:-

"1', I!,

:ill _,';I!

l'
/,1',1
1

Figure 10-5. Graphieal Representation

ofthe SPF for Four-Ieg, Stop-eontrolled

(4ST) Interseetions (Equation 10-9)

, ,:,

'1",:!j ..

Four-Leg Signalized Interseetions


The SPF for four-Ieg signalized interseetions is shown in Equation 10-10 and presented graphieally in Figure 10-6.

' 'o

.1

:,,;,P,:
l.,.

',1

1'1

11,

11'1'1,1

1"

11"';

N,pf"G ~ exp[-5.I3

.,'

+ 0.20 x In(AADT",,)]

'(lO-lO)

Where,

'111i:!,:,~'
, ~';':,1' :.',:1.,'1

',: ,1
'I:~'
il:
,

+ 0.60 x In(AADTm,)

~l

'J'

AADTm'j

:=
=

SPF estimate of intersection-related

predicted average crash frequency for base conditions;

AADT (vehicles per day) on the major road; and

"'1' ; '

ilii'

,/

'1
I

AADTml" '~ AADT (vehiclesper

day) on the minor road.

The overdispersion parameter (k) for this SPF is 0.11. This SPF is applieable to an AADT ,range from zero to
m'l
25,200 vehicles per day and AADT mi" range from zero to 12,500 vehicles per day. For .instanees when applieation is
made to sites
with AADT substantially outside these ranges, the reliability is unknown.
,
.

CHAPTER 1G-PREDICTIVE

METHOD fOR RURAL TWO-LANE,

TWO-WAY

10-21

ROADS

18
17

AADT

")In=

12, 500

AADTm;n=10,OOO

16

AADTm;,,=8.000

15
AADTrnin=5,OOO

14

13

AAOT,...,=3,OOO

12

.AADTm1n=2,OOO

11

"i!!

10

.<=
~

el>

C'

...

"t>

el>

~
u

'tiel>

AAOT"",,=4,OOO

AAOTmin=1.000

6
S
4
3
2
1
O
O

10,000

5,000

25,000

20,000

15,000

AAOT m'jo, (veh/day)

Figure 10-6. Graphical Representation

..

ofthe SPF for Four-leg Signalized (4SG) Intersections (Equation lO-lO)

Tables 10-5 and 10-6 provide the default proportions for crash severity levels and collision types, respectively. These
tables may be used to separate the crash frequencies from Equations 10-8 through 10-10 into components by severity level and collision type. Tbe default proportions for severity levels and collision types shown in Tables lO-s and
10-6 may be updated based on local data for a particular jurisdiction as part ofthe calibration process described in
Appendix A to Part c..
Table 10-5. Default Distributian far Crash Severity Level at Rural Twa-Lane, Two-Way Intersections
Percentage ofTotal Crashes

C.-ash Severity Level


Fatal
Incapacitating

Injury

Three-Leg
Stop-Controlled Intersections
1.7

1.8

0.9

4.0

4.3

2.1

16.2

10.5

Possible injury

19.2

20.8 .

20.5

Total fatal plus itijury

41.5

43.1

34.0

Property damage only

58.5

56.9'

66.0

100.0

100.0

Total

100.0

Note: Based on HSIS data far California (2002-2006).

Four-Leg
Signalized Intersections

16.6

~onincapacitating injury

"

Four-Leg
Stop-Controlled Intersections

~
,

HIGHWAY SAFETY MANUAL

10-22

Table 10-6. Default Distribution for Collision Type and Manner of Collision at Rural Two- Way Intersections
Percentage ofTotal Crashes by Collision Type

Fatal

and
Injuey

Collision Type
SINGLE-VEHICLE
Collision

Property
Damage
Ouly

Four-Leg Signalized
Intersections

Four-Leg Stop-Controlled
Intersections

Three-Leg Stop-Controlled
Intersections

Fatal

and
Total

Injury

1.9

0.6

Property
Damage
Ooly

Fatal

and

Property
Damage
Doly

Total

Total

Injury

1.0

0.0

0.3

0.2
0.1

CRASHES
0.8

with animal

2.6

1.4

Collision with bicycle

0.1

0.1

0.1

0.1

0.1

0.1

0.1

0.1

Collision

0.1

0.1

0.1

0.1

0.1

0.1

0.1

0.1

0.1

2.2

0.7

1.3

0.6

0.4

0.5

0.3

0.3

0.3

24.0

24.7

24.4

9.4

14.4

12.2

3.2

8.1

6.4

0.3

1.8

0.5

with pedestrian

Overturned
Ran offroad

1.1

2.0

1.6

0.4

1.0

0.8

28.3

30.2

29.4

11.2

17.4

14.7

4.0

10.7

7.6

27.5

21.0

23.7

53.2

35.4

43.1

33.6

24.2

27.4

Head-on collision

8.1

3.2

5.2

6.0

2.5

4.0

8.0

4.0

5.4

Rear-eRd collision

26.0

29.2

27.8

21.0

26.6

24.2

40.3

43.8

42.6

5.1

13.1

9.7

4.4

14.4

10.1

5.1

15.3

11.8

5.0

3.3

4.2

4.2

3.7

3.9

9.0

2.0

5.2

71.7

69.8

70.6

88.8

82.6

85.3

96.0

89.3

92.4

100.0

100.0

100.0

100.0

100.0

100.0

100.0

100.0

100.0

Other single-vehicle

crash

Total single.vehiele

crashes

MULTIPLE-VEHICLE

CRASHES

Angle collision

Sideswipe

collision

Other muitiple-vehicle
Total multiple.vehicle
Total Crashes

collision
crashes

Note: Based on HSIS data for California (2002-2006).

10.7. CRASH MODIFICATION FACTORS


In Slep lO ofthe predictive method shown in Section lOA, crash modification factors (CMFs) are applied to account
for the effects of site-specific geometric design and traffic control features. CMFs are used in the predictive method
in Equations 10-2 and 10-3. A general overview of crash modification factors (CMFs) is presented in Chapter 3, Section 3.5.3. Tbe Part C~lntroduction
andApplications Guidance provides further discussion on the relationship of
CMFs to the predictive method. This section provides details ofthe specific CMFs applicable to the safety performance functions presented in Section 10.6.
Crash modification factors (CMFs) are used to adjust the SPF estimate ofpredicted average crash frequency for
the effect of individual geometric design and traffic control features, as shown in the general predictive model for
Chapter 10 shown in Equation 10-1. The CMF for the SPF base condition of each geometric design or traffic control
feature has a value of 1.00. Any feature associated with higher crash frequency than the base condition has a CMF
with a value greater Ihan 1.00. Any featufe associated with lower crash frequency than the base condition has a CMF
with a value less than 1.00.
The CMFs used in Chapter lO are consistent with the CMFs in Part D, although they have, in sorne cases, been
expressed in a different form to be applicable to the base conditions. The CMFs presented in Chapter 10 and the
specific site types to which they apply are summarized in Table 10-7.

ri.

CHAPTER 10-PREDICTIVE

METHOD FOR RURAL TWO-LANE,

TWO-WAY

ROADS

::;~-----------------------------------------rabIe 10-7. Summary ofCrash Modification Factors (CMFs) in Chapter lO and the Corresponding
Performance Functions (SPFs)
facilityType

Rural Two-LaneTwo-Way
Roadway Segments

Three- and four-Ieg stop control


intersections and four-leg
signalized intersections

10-23

Safety

CMF

CMF Description

CMF Equations and Tables

eMPI,

Lane Width

Table 10-8, Figure)O-7,


, Equation 1O-1l

CMF1,

Shoulder Width and Type

Table 10-9, Figure 10-8, Table 10-10,


Equation 10-12

CMFJ,

Horizontal Curves: Length, Radius. and


Presence or Absence of Spiral Transitions

TabIe 10-7

CMF4,

Horizontal Curves: Superelevation

Equation 10-14, 10-15, 10.16,

CMF,.

Grades

CMF <ir

Driveway Density

Table 10-11

CMF7,

Centerline Rumble Strips

See text

CMFII,

Passing Lanes

See text

CMF9r

Two.Way Left-Tum Lanes

Equation 10-18, 10-19

CMF/Or

Roadside Design

Equation 10-20

CMFI/r

Lighting

Equation 10-21, Table 10-12

CMF12r

Automated Speed Enforcement

See text

CMF/j

Intersection Skew Angle

Equation 10-22, 10-23

CMF1i

Intersection Left-Tum Lanes

Table 10-13

CMFJ,

Intersection Right- Tum Lanes

Table 10-14

CMF4/

Lighting

Equation 10-24, Table 10-15

-, Table 10-11

10.7.1. Crash Modification Faetors for Roadway Segments


The CMFs for geometric design and traffic control features ofrural two-lane, two-way roadway segments are presented below. These CMFs are applied in Step 10 ofthe predictive method and used in Equation 10-2 to adjust the
SPF for rural two-lane, two-way roadway segments presented in Equation 10-6, to account for differences between
the base conditions and the local site conditions.
CMF,,-Lane
Width
The CMF for lane width on two-lane highway segments is presented in Table 10-8 and illustrated by the graph in
Figure 10-7. This CMF was developed from the work of Zegeer et al. (16) and Griffin and Mak (4J. The base value
for the lane width CMF is 12 ft. In other words, the roadway segment SPF will predict safety performance of a roadway segment with 12-ft lanes. To predict the safety performance ofthe actual segment in question (e.g., one with
lane widths different than 12 ftJ, CMFs are used to account for differences between base and actual conditions. Thus,
12-ft lanes are assigned a CMF of 1.00. CMF" is determined from Table 10-8 based on the applicable lane width and
traffic volume range. The relationships shown in Table 10-8 are illustrated in Figure 10-7. Lanes with widths greater
than 12 ft are assigned a CMF equal to that for 12-ft lanes.
For lane widths with 0.5-ft increments that are not depicted specifically in Table 10-8 or Figure 10-7, a CMF value
can be interpolated using either of these exhibits since there is a linear transition between the various AADT effects.

"
"'1

,
J

HIGHWAY 5AFETY MANUAL

10-24

,1

.1
.I
Table 10.8. CMF for Laoe Width

00

Roadway Segmeots (CMF ro)


AADT (vehicles per day)

:'

LaneWidth
9 ft or less

1.05

1.05+2.81

x IO-4(AADT-100)

1.50

lO ft

1.02

1.02+ 1.75 x lO-4(AADT-400)

1.30

11 ft

1.01

1.01 + 2.5 x 10-' (AADT - 400)

1.05

12 ft or more

LOO

LOO

LOO

l'

i.

> 2000

400 to 2000

<400

Note: The collision types related to lane width to whch this CMF applies nelude single-vehicle
opposite-direction
sideswipe, and same-direction 5ideswipe crashes.

run-offcthe-road

and multiple-vehkJe

head-on,

1.70
The factor applies to

single-vehicle run-off-the-road
and multiple-vehide
head.on,
opposite-direetion
sideswipe,

1.60

and same-direction sideswipe


crashes.
1.50

9.ft Lanes

1.30

10-ft lanes

1.05

11-ft lanes

1.00

12-ft lanes

1.50

(;

1::

c:.-

.,.-

1.40

.O

''O"

1.30

"

:;;
.t:
~

.-~

1.20

1,10

1.05

1.00
400

O
iJ

Figure

1,200

800

1,600

2,000

2,400

AADT (veh/day)

10-7. Crash Modificatioo Factor for Lane Width

00

Roadway Segmeots

If the laoe widths for the two directioos of travel 00 a roadway segmeot differ, the CMF are determioed separately
for the laoe width in each direction oftravel aod the resultiog CMFs are ilieo be averaged.
The CMFs shown io Table 10-8 and Figure 10-7 apply only to the crash types that are most likely to be affected by
Jarre width: single-vehic1e run-otf-the-road and multiple-vehicle head-on, opposite'-direction sideswipe, and same-directian sideswipe crashes. These are the only crash types assumed to be affected by variation in larre width, and other
crash types are assumed to remain unchanged due to the larre width variation. The CMFs expressed on this basis are,
therefore, adjusted to total crashes within the predictive method. This i5 accomplished using Equation 10-11:
(10-11)

.,

CHAPTER lD-PREDICTIVE

METHOD FOR RURAL 1WO-LANE,

1WO-WAY

10-25

ROADS

Where:
crash modification

factor for the effect of lane width ,on total crashes;

erash modifieation factor for the effeet oflane width on related erashes (i.e., single-vehicle run-off-theroad and multiple-vehicle

head-on,

opposite-direction

sideswipe,

and same-directionJsideswipe

crashes),

sueh as the erash modifieation factor for lane width shown in Table 10-8; and
proportion of total erashes eonstituted by related erashes.
The proportion of related erashes, p , (i.e., single-vehicle run-off-the-road, and multiple-vehiele head-on, oppositedireetion sideswipe,and same-direetion sideswipes erashes) is estimated as 0.574 (i.e., 57.4 pereent) based on lhe
default distribution of erash types presented in Table 10-4. This default erash type distribution, and therefore the
value of p ro' may be updated from local data as part of lhe ealibration proeess.
CMF,,-Shoulder
Width and Type
The CMF for shoulders has a CMF for shoulder width (CMF .ro) and a CMF for shoulder type (CMF,ro)' The CMFs
for both shoulder width an'd shoulder type are based on the results of Zegeer et al. (16, 17). The base value of shoulder width and type is a 6-foot paved shoulder, whieh is assigned a CMF value of 1.00.
CMF ,'ro for shoulder width on two-Iane highway segments is determined from Table 10-9 based on the applieable
shoulder width and traffie volume range. The relationships shown in Table 10-9 are illustrated in Figure 10-8.
Shoulders over 8-ft wide are assigned a CMFwroequal to that for 8-ft shoulders. The CMFs shown in Table 10-9 and
Figure 10-8 apply only to single-vehicle run-offthe-road and multiple-vehicle head-on, opposite-direetion sideswipe,
and same-direction

sideswipe

crashes.

Table 10-9. CMF for Shoulder Width on Roadway Segments (CMF.ro)


AADT (vehicles per day)

<400

400 to 2000

> 2000

1.t9

1.10 + 2.5 x 10~ (AADT - 400)

1.50

2ft

1.07

1.07 + 1.43 x 10-4 (AADT - 400)

1.30

4ft

1.02

1.02+8.125

1.15

6ft

1.00

8 ft or more

0.98

Shoulder Width
Oft

x lO-~(AADT-400)
1.00

0.98 + 6.875 x

to-' (AADT

1.00
- 400)

Note: The cotlision types related to shoulder width to which this CMF applies nclude single-vehic1e run-off the-road
head-on, opposite-direction
sideswipe, and same-direction sjdeswipe crashes.

0.87
and multiple-vehic1e

;.,.,

HIGHWAY SAFETY MANUAL

10-26

1.60
1.50

().ft

Shoulden

Thls factor Ilpplles to single-vehide

1.50

run.off.the-road and multiple-vehlcle


head-on. opposite-directlon sideswipe.
and Sllme-directlon sldeswlpe uashes.

1.40
1.30 2-ftSliouldets

<;

DO

t:

.,

u-

c:

.o

1.20

1.15

40ft

Shoulders

."O

1.10

~
.c:
~

1.10

1.00

C!

6-ft Shoulder5

0.87

8-ft Shoulders

0.98

1.00

0.90

0.80

200

400

800

600

1,000

1,200

1,400

1,600

1,800

2,000

2,200

2,400

AAOT (veh/day)

Figure 10-8. Crash Modification Factor for Shoulder Width on Roadway Segments
The base condition for shoulder type is paved. Table 10-10 presents values for CMF = which adjusts for the safety
effects of gravel, turf, and composite shoulders as a function of shoulder width.

Table 10-10. Crash Modification Factors for ShoulderTypes and ShoulderWidths

on Roadway Segments (CMF =)

Shoulder Width (ft)


O

1.00

1.00

1.00

1.00

1.00

1.00

1.00

Paved

1.00

1.00

1.01

1.01

1.01

1.02

1.02

Gravel

1.00

1.01

1.02

1.02

1.03

1.04

1.06

Composite

1.00

1.01

1.03

1.04

1.05

1.08

J.JI

Turf

Shoulder Type

Note: The values fer composite shoulders in this table represent

a shoulder fer which SO percent of the shoulder width is paved and 50 percent

of the shoulder width is turf.

If the shoulder types aneVor widths for the two directions of a roadway segment differ, the CMF are determined
separately for the shoulder type and width in each directiori of travel and the resulting CMFs are then be aver~ged.
The CMFs for shoulder width and type shown in Table 9, Figure 8, and Table lO apply only to the collision types
that are most likely to be affected by shoulder width and type: single-vehicle run-offthe-road and multiple-vehicle
head-on, opposite-drection sdeswpe, and same-drecton sideswipe crashes. The CMFs expressed on this basis are,
therefore, adjusted to total crashes using Equation 10-12.

;r
11,

CHAPTER 10-PREDICTIVE

METHOD FOR RURAL TWO-LANE,

TWO-WAY

ROADS

10-27

(10-12)
Where:
crash modification factor for the e/fecl of shoulder widlh and type on lotal crashes; ,,
crash modification factor for related crashes (Le., single-vehicle run-off-the-road and multiple-vehicle
head-on, opposite-direction sideswipe, and same-direction sideswipe crashes), based on shoulder width
(from Table 10-9);
crash modification faclor for relaled crashes based on shoulder type (from Table 10-10); and
proportion of total crashes constituted by related eras hes.
The proportion of relaled crashes, Pro' (i.e., single-vehicle run-o/f-lhe-road, and multiple-vehicle head-on, oppositedireclion sideswipe, and same-direction sideswipes crashes) is estimaled as 0.574 (Le., 57.4 percenl) based on the
defaull distribulion of crash types presenled in Table 10-4. This default crash type dislribulion, and therefore lhe
value of pra, may be updated from local data by a highway agency as parl of lhe calibration process.
CMFJr-Horizontal
Curves: Length, Radius, aud Presence or Absence of Spiral Transitions
The base condition for horizontal alignment is a tangent roadway segmento A CMF has been developed to represent
lhe marmer in which crash experience on curved alignments di/fers from thal of tangenls. This CMF applies lO total
roadway segment crashes.
The CMF for horizonlal curves has been determined from !he regression model developed by Zegeer el al. (18).
The CMF for horizonlal curvature is in lhe form of an equalion and yields a factor similar lo the other CMFs in lhis
chapter. The CMF for length, radius, and presence or absence of spiral transitions on horizontal curves is determined
using Equalion 10-13.
_ (1.55 x Le)
ClvfF3r

(8~)-

(0.012 x S)
(10-13)

- -----~---~-------

(1.55

Le)

Where:
CMF"

crash modification factor for the effect ofhorizontal

alignment on total crashes;

L,

length ofhorizonlal

radius of curvature (feel); and

1 if spiral transition curve is present; O if spiral transition curve is not present; 0.5 if a spiral transition
curve is present at ane but not bath ends of the horizontal curve.

curve (miles) which includes spiral transilions, ifpresenl;

Sorne roadway segments being analyzed may include only a portian of a horizontal curve. In this case, L represents
the length of the entire horizontal curve, including portions of the horizontal curve that may lie outside the roadway
segrnent of interest.
In applying Equalion 10-13, iflhe radius ofcurvature (R) is less !han 100-ft, R is sel lo equal lo lOO ft. Iflhe length
ofthe horizontal curve (L) is less than 100 feel, L, is set to equal 100 ft.
CMF values are computed separately for each horizontal curve in a'horizontal curve set (a curve set consists of a
series of consecutive curve elements). For each individual curve, the value of L used in Equation 10-13 is the total
lenglh oflhe compound curve sel and lhe value ofR is the radius oflhe individ~al curve.
Ifthe value ofCMF"

is less than 1.00, !he value ofCMF,,is

set equal to 1.00.

HIGHWAY

10.28

SAFETY MANUAL

.!
CMF 4,-Horizontal Curves: Superelevation
The base condition for the CMF fer the superelevation of a horizontal curve is the amollnt of superelevation identified in A Policy on Geometric Design o/ Highways and Streets-also
called lhe AASHTO Green Book (1). The superelevation in the AASHTO Green Book is determined by taking into account the value of maximum superelevation
rate, emax' established by highway agency policies. Policies concerning maximum super~levation rates [OI horizontal
curves vary between highway agencies based on climate and other considerations.
The CMF for superelevation is based on the.superelevation variance ofa horizontal curve (i.e., the difference
between lhe actual superelevalion and lhe superelevation identified by AASHTO policy). When the actual superelevalion meels or exceeds thal in the AASHTO policy, lhe value ofthe superelevation CMF is 1.00. There is no effect
of superelevation variance on crash frequency until the superelevation variance exceeds 0.'01. The general functional
form of a CMF for superelevation variance is based on the work of Zegeer el al. (18, 19).
The following relationships present the CMF for superelevation variance:

CMF"

CMF"

(10-14)

1.00 for SV < 0.01

(10-15)

1.00 + 6 x (SV - 0.01) for 0.01" SV < 0.02

(10-16)

CMF" ~ 1.06 + 3 x (SV - 0.02) for SV? 0.02


Where:
CMF4r

crash modification factor for the effect of superelevation variance on total crashes; and

superelevation variance (ftlft), which represents the superelevation rate contained in the AASHTO
Green Book minus the actual superelevation of the curve.

SV

CMF" applies to lolal roadway segment erashes for roadway segmenls located on horizontal curves.

CMF ,,-Grades
The base condition for grade is a generally level roadway. Table 10-11 presents the CMF for gr~des based on an
analysis ofrural two-lane, two-way highway grades in Ulah conducted by Miaou (8). The CMFs in Table 10-11 are
applied to each individual grade segment on the roadway being evaluated without respect to the sign of the grade.
The sigo ofthe grade is irrelevant because each grade on a rural two-hme, two-way highway is an upgrade for one
direction of travel and a downgrade for the other. The grade factors are applied to the entire grade from one poiot of
vertical intersectioo (PVI) to the next (i.e., there is no special account taken ofvertical curves). The CMFs in Table
J 0-11 apply lO total roadway segment crashes.
>

rabie 10-11. Crash Modificalion Faclors (CMF ,) for Grade of Roadway Segmenls
Approximate Grade (%)
Level Grade
(:5 3%)

Moderate Terrain
(3%< grade :5 6%)

Steep Terrain
(> 6%)

1.00

1.10

1.16

CMF,-Driveway
Density
The base eondition for driveway density is five driveways per mile. As with the olher CMFs, the model for lhe base
condition was established for roadways with this driveway density. The CMF for driveway density is detennined
using Equation 10-17, derived from lhe work ofMuskaug (9).
CMF.
6,

0.322 + DD x [0.05 - 0.005 x In(AADT)]


0.322 + 5x [0.05 - 0.005 x In(AADT)]

(10'17)

CHAPTER lQ---PREDICTIVE METHOD FOR RURAL TWO-LANE,

TWO-WAY

10-29

ROADS

Where:
crash modification

factor for the effect of driveway density on total crashes;

AADT

average annual daily traffie volume ofthe roadway being evaluated (vehieles per day); and

DD

driveway density eonsidering driveways on both sides ofthe highway (driveways/mple).

If driveway density is less than 5 driveways per mile, CMF" is l.00. Equation 10-17 can be applied to total
roadway erashes of all severity levels.
Driveways

serving

a11 types of land use are considered

in determining

the driveway

density. AH driveways

that

are used by traffic on at least a daily basis for entering or leaving the highway are considered. Driveways that
receive only occasional use (less than daily), such as field entrances are not considered.
CMF ,,-Centerline
Rumble Strips
Centerline rumble strips are installed on undivided highways along the eenterline of the roadway whieh divides
opposing direetions of traftie flow. Centerline rumble strips are ineorporated in the roadway surfaee to alert
drivers who unintentionally crass, Dr begin to cross, the roadway centerline. The base condition for centerline
rumble strips is the absence of rumble strips.
The value of CMF 7~ for the effect of centerline rumble strips for total erashes on rural two-lane. two-way
highways is derived as 0.94 from the CMF value presented in Chapter 13 and erash type pereentages found in
Chapter 10. Details of this derivation are no! provided.
The CMF for eenterline rumble strips applies only to two-Iane undivided highways with no separation other than
a eenterline marking between the lanes in opposite direetions oftravel. Otherwise the value ofthis CMF is l.00.

,
l.
I

!
I
f

CMF ,,-Passing
Lanes
The base eondition for passing lanes is the absenee of a lane (i.e., the normal two-Iane eross seetion). The CMF
for a eonventional passing or elimbing lane added in one direetion oftravel on a rural two-Iane, two-way highway
is 0.75 for total erashes in both direetions of travel over the length ofthe passing lane from the upstream end of
the lane addition taper to the downstream end of the lane drop taper. This value assumes that the passing lane is
operationally warranted and that the length of the passing lane is appropriate for the operational eonditions on the
roadway. There may al so be sorne safety benefi! on the roadway downstream of a passing lane, but this effeet has
not been quantitied.
The CMF for short four-Iane seetions (i.e., side-by-side passing lanes provided in opposite direetions on the
same seetion ofroadway) is 0.65 for total erashes over the length ofthe short four-Iane seetion. This CMF
applies to any portion of roadway where the eross section has four lanes and where both added lanes have been
provided over a limited distance to inerease passing opportunities. This CMF does not apply to extended fourlane highway seetions.
The CMF for passing lanes is based primarily on the work of Harwood and St.John (6), with eonsideration also
given to the results ofRinde (11) and Nettelblad (10). The CMF for short four-Iane seetions is based on the
work ofHarwood and Sto John (6).
CMF,,-1\vo-Way
Left-Turn Lanes
The instal!ation of a eenter two-way left-turn lane (TWLTL) on a rural two-Iane, two-way highway to ereate a
three.lane cross-section can reduce crashes related to turning maneuvers at driveways. The base condition for
two-way left-turn lanes is the absenee of a TWLTL. The CMF for installation of a TWLTL is:

'V

.,..'
,

1
\1

10-30

HIGHWAY SAFETY MANUAL

(10-18)

CMF" ~ 1.0 - (0.7 x P <fwy x P LTlD)


Where:

CMF"

~ crash modification factor for!he effect oftwo-way left-turn lanes on total crashes;
,

P dwy

dri.veway-related crashes as a proportion of total crashes; and

PLTlD

left-turn crashes susceptible to correction by a TWLTL as a proportion of driveway-reIated crashes.

The value ofPd"" can be estimated using Equation 10-19 (6).


(0.0047 x DD) + (0.0024 x Dd2))

.
Pdwy =

1.199

+ (0.0047 x DD) + (0.0024

Dd2))

(10-19)

.'

Where:

Pdwy

driveway-related crashes as a proportion oftotal crashes; and

DD

driveway density considering driveways on both sides of the highway (driveways/mile).

The value

ofPLTID

is estimated as 0.5 (6).

Equation 10-18 provides the best estimate of the CMF for TWLTL installation that can be made without data on
the left-turn volumes within the TWLTL. Realistically, such volumes are seldom available for use in such analyses
though Section A.I. of Appendix A to Part C describes how to appropriately calibrate this value. This CMF applies
to total roadway segment crashes.
The CMF for TWLTL installation is not applied unIess the driveway density is greater than or equal to five driveways
per mile. If the driveway density is less than five driveways per mile, the CMF for TWLTL installation is 1.00.

"1 '

CMFIO,-Roadside
Design
For purposes of the HSM predictive method, the level of roadside design is represented by the roadside hazard rating
(1-7 scale) developed by Zegeer et al. (16). The CMF for roadside. design was developed in research by Harwood et
al. (5). The base value ofroadside hazard rating for roadway segments is 3. The CMF is:
e(-o.6869 + 0.0668 xRHR)
e(-Q.4865)

:I

"i

(10-20)

Where:

CMFJ(k
RHR

crash modification factor for the effect of roadsid~ design; and


roadside hazard rating.

This CMF applies to total roadway segment crashes. Photographic examples and quantitative definitions for each
roadside hazard rating (1-7) as a function of roadside design features such as sideslope and elear zone width are
presented in Chapter 13, Appendix 13A.
!,';

CMFIl,-Ligbting
The base condition for lighting is the absence of roadway segment lighting. The CMF for lighted roadway segments
is determined, based on the work ofElvik and Yaa (2), as:

CHAPTER 1Q-PREDICTIVE

CMFII,

METHOD FOR RURAL TWO-LANE,

TWO-WAY

10-31

ROADS

(10-21)

1.0 - [(1.0 - 0.72 x p,", - 0.83 x pp") x P,,]

Where:
CMFII,
P

'"'

PP"'
p",

crash modificalion factor for lhe effecl of lighting on lotal crashes;

proportion of total nighttime crashes for unlighled roadway segments lhat involve a fatality or injury;

proportion oflotal nighttime crashes for unlighted roadway segmenls thal involve property damage only; and

~ proportion oftolal crashes for unlighted roadway segmenls lhat occur at night.

This CMF applies lo total roadway segmen! erashes. Table 10-12 presents default values for lhe nighttime erash proporlions PI", Pp", and p,,' HSM users are encouraged lO replace lhe estimates in Table 10-12 with loeally derived volues. If
lighling inslallation inereases the density ofroadside fixed objeets, lhe volue ofCMFiO, is adjusled aeeordingly.

rabie 10.12. Nighttime Crash Proporlions for Unlighted Roadway Segmenls


Proportion ofTotal NighUime Crashes by Severity Leve)
Roadway Type

Proportion ol Crashes iba! Occur a Night

Fatal sud Injury Pln,

PDOppn'

P"

0.382

0.618

0.370

2U
Note: Based on HSIS data for Washington

(2002-2006)

CMF 11,-Au!omated Speed Enforcement


Automated speed enforcement systems use video or photographic identification in conjunction with radar or
lasers to detect speeding drivers. These systems automatical1y record vehicle identification information without
the need for police officers at the scene. The base cooditioo for automated speed enforcement is that it is absent.
The value ofCMF12, for the effecl ofautomated speed enforeement for lotal crashes on rural two-lane, two-way
highways is derived as 0.93 from lhe CMF value presented in Chapter 17 and crash type percentages found in
Chapter 10. Delails of lhis derivation are nol provided.

10.7.2. Crash Modification Factors for lntersections


The effecls of individual geometrie design and traffic control features of nlerseclions are represented in the
predietive models by CMFs. The CMFs for inlerseelion skew angle, left-tum lanes, righl-tum lanes, and lighting
are presented below. Eaeh ofthe CMFs applies to tolal crashes.
CMF ,,-Intersection
Skew Angle
The base condition for intersection skew angle is zera degrees ofskew (Le., an intersection angle of90 degrees).
The skew angle for an intersection was defined as the absolute value of the deviation from an intersection angle of
90 degrees. The absolute value is used in the definition of skew angle because postive and negative skew angles are
considered lo have similar detrimenlal effeet (4). This is illustraled in Chapler 14, Section 14.6.2.

Three-Leg Intersections with Stop-Controlan the Minar Approach


The CMF for interseetion angle al three-leg inlerseelions wilh stop-control on the minor approaeh is:

CMF"

(O.004~skew)

Where:
CMFIi

= crash modification factor for the effect ofintersection skew on total crashes; and

(10-22)

HIGHWAY SAFETY MANUAL

10-32

intersection skew angle (in degrees); the absolute vaJue ofthe difference between 90 degrees and the
actual intersection angle.

skew

This CMF applies to total intersection crashes.


Four-Leg Intersections with Stop-Control on (he Minor Approaches
The CMF for intersection angle at four-leg intersection with stop-control
CMF , = e

(0.0054

J
Oil

the minar approaches is;


(10-23)

skew)

1,

Where:
CMF"

crash modification factor for the effect of intersection skew o~ntotal crashes; and

skew

intersection skew anglo (in degrees); the absolute value ofthe difference between 90 degrees and the
actual intersection angle.

This CMF applies to total interseclon crashes.


lfthe skew angle differs for the two minor road Jegs at a four-leg stop-controlled
computed separateJy for each minor road leg and then averaged.

intersection, values ofCMF"

is

Four-Leg Signalized Intersections


Since the traffic signal separates most movements from conflicting approaches, tbe risk of collisions related to the
skew angle between the intersecting approaches is limited at a signalized intersection. Therefore, the CMF for skew
angle at four-leg signalized intersections is 1.00 for al1 cases.
CMF 1i-Intersection Left- Turo Lanes
The base condition for intersection left-tuTn lanes is the absence of left-tum lanes on the intersection approaches.
The CMFs for the presenee of left-turn lanes are presented in TabJe 10-13. These CMFs appJy to installalon of
left-turn lanes on any approaeh to a signalized interseetion, but only on uneontrolled major road approaches to a
stop-controlled interseclon. The CMFs for installation of left-turn Janes on mulliple approaches to an intersee-.
tion are equal to the eorresponding CMF for the installalon of a left-turn lane on one approach raised to a power
equal to the number of approaches with left-tuTn lanes. There is no indication of any safety effect of providing
a left-turn lane on an approach eontrolled by a stop sign, so the presence of a Jeft-turn Jane on a stop-controlled
approach is not considered in applying Table 10-13. The CMFs for installalon of left-turn Janes are based on researeh by Harwood et al. (5) and are consistent with the CMFs presented in Chapter 14. A CMF of 1.00 is always
be used when no left-tum lanes are present.

rabIe 10-13. Crash Modification Factors (CMF,;J for lnstallation ofLeft-Turn

Lanes on lnterseetion Approaches

Number of Approaehes with Left~Turn Lanes.


lnterseetion Type

Interseetion Traffie Control

Three-leg Intersection
Four-leg Interseetion

OneAppmaeh

Two Appmaches

Minor road stop control

0.56

0.31

Minor road stop controlb

0.72

0.52

Traffie signa!

0.82

0.67

Three Appmaehes

Four Approaches

0.55

0.45

Stop~controlJed approaches are not considered in determining the number of approaches with left.turn lanes
Stop signs present on minor road approaches only.

CMF ,,-Intersection
Right- Turn Lanes
The base condition for intersection right-turn lanes is the absence of right-turn lanes on the intersection approaches.
The CMF for the presence ofright-turn lanes is based on researeh by Harwood et al. (5) and is consistent with the
CMFs in Chapter 14. These CMFs apply to installation ofright-turn Janes on any approach to a signalized intersec-

CHAPTER 1(}-PREDICTIVE

METHOD FOR RURAL lWO-LANE.

lWO-WAY

10-33

ROADS

tioo, but only on uncontrolled majar road approaches to stop-controlled intersections. The CMFs for installation of
right-tum lanes 00 multiple approaches to an intersection are equal to the corresponding CMF for installaon of a
right-turn lane 00 ene approach raised to a power egual to the nl;lmber of approaches with right-turn lanes. There

is no indieation of any safety effeet for providing a right-turn lane on an approaeh eontrolled by a stop sign, so the
presenee ofa right-turn lane on a stop-eontroIled approaeh is not eonsidered in applyingTable 10-14. The CMFs
in the table apply to total interseetion erashes. A CMF value of 1.00 is always be used when no fight-turn lanes are
presento This CMF applies only to right-turn lanes that are identified by marking or signing. The CMF is not applicable to long tapers, fiares, or paved shoulders that may be used informaIly by right-turn traffie.
Table 10.14. Crash Modifieation Faetors (CMF)
Two-Lane, Two-Way Highways

for Right-Turn Lanes on Approaehes to an Intersection on Rural


Number oC Approaches with Right-Turn Lanes'
Oue Approach

Two Approaches

Intersection Type

Intersection Traffie Control

Three-Leg

Minar road stop controlb

0.86

0.74

Minar road stop controlb

0.86

0.74

Traffie signal

0.96

0.92

lntersection

Four-Leg Intersection

Stop-cantrolled

Stop signs present on minar road approaches

approaches

are not considered

in determining

Three Approaches

the number of approaches with rlght.turn

Four Approaches

0.88

0.85

lanes.

only.

CMF ,,-Lighting
The base eondition for lighting is the absenee of interseetion lighting. The CMF for lighted interseetions is adapted
from the work of Elvik and Vaa (2), as:
CMF" = 1 - 0.38

(10-24)

p.;

Where:
CMF,;
p.,

erash modifieation factor for lhe effeet oflighting

on total erashes; and

proportion of total crashes for unlighted interseetions that oecur at night.

This CMF applies to total interseelion crashes. Table 10-15 presents default values far the nighttime erash proportion
.. HSM users are eneouraged to replaee the estimates in Table 10-15 with laeally derived values .

.,

Table 10-15. Nighttime Crash Proportions for Unlighted Interseetions


Proportion oC Crashes lhal Occur al Night

p.,

[ntersection Type

0.260

3ST

0.244

4ST

0.286

4SG
Note: Based on H51S data for California

(2002-2006)

10.8. CALlBRATION OF THE SPFS TO LOCAL CONDITION5


In Step 10 of the predietive method, presented in Seetion lOA, the predietive model is ealibrated to local state or
geograpbic conditions. Crash frequencies, even for nominally similar roadway segments or intersections, can vary
widely from one jurisdiction to another. Geographic regions differ markedly in climate, animal population, driver
populations, crash reporting tbreshold, and crash reporting practices. Tbese variations may result in sorne jurisdictions

10-34

HIGHWAY SAFETY MANUAL

experiencing a different number ofreported traffic crashes on rural two-lane, two-way roads than others. Calibration
factors are included in the melhodology to allow highway agencies to adjust the SPFs to match actual local conditions.
The calibration factors for roadway segments and intersections (defined as C, and Cj, respectively) will have values greater
than 1.0 for roadways lhat, on average, experience more crashes lhan lhe roadways used in lhepevelopment oflhe SPFs.
The calibration factors fo! roadways that experience fewer crashes on average than the roadways used in the development
oflhe SPFs will have values less than 1.0. The calibration procedures are presented in Appendix A to Part C.
Calibration factors provide one method of incorporating local data to improve estimated crash frequencies for individual agencies or locations. Several other default values used in the predictive melhod, such as.collision type distribution, can also be replaced with locally derived values. The derivation of values for these pararneters is addressed in
the calibration procedure in Appendix A to Par! C.

10.9. L1MITATIONS OF PREDICTIVE METHOD IN CHAPTER 10


This section discusses limitations ofthe specific predictive models and the application afthe predictive method in
Chapter 10.
Where rural two-lane, two-way roads intersect access-controlled facilities (Le., freeways), the grade-separated
interchange facility, including the two-lane road within the interchange area, cannat be addressed with the predictive
method for rural two-Iane, two-way roads.
The SPFs developed for Chapter 10 do not include signalized three-leg intersection models. Such intersections are
occasionally found on rural two-Iane, two-way roads.

10.10. APPLlCATION OF CHAPTER 10 PREDICTIVE METHOD


The predictive method presented in Ch'pter 10 applies to rural two-lane, two-way roads. The predictive method
is applied to a rural two-Iane, two-way facility by following the 18 steps presented in Section IDA. Appendix IDA
provides a series of worksheets for applying lhe predictive method and lhe predictive models detaHed in lhis chapter.
All computations within lhese worksheets are conducted with values expressed to three decimal places. This level of
precision is needed for consistency in computations. In the last stage of computations, rounding the final estimate of
expected average crash frequency to one decimal place is appropriate.

10.11.SUMMARY
The predictive method can be used to estimate the expected average crash frequency for a series of contiguous sites
(entire rural two-lane, two-way facility), or a single individual site. A rural two-lane, two-way facility is defined in Section 10.3, and consists of a two-lane, two-way undivided road wmch does not have access control and is outside of cities
or towns with a population greater lhan 5,000 persons. Two-lane, two-way undivided roads lhat have occasional added
lanes to provide additional passing opportunities can also be addressed with lhe Chapter 10 predictive melhod.
The predictive method for rural two-Iane, two-way roads is applied by following the 18 steps ofthe predictive
method presented in Section IDA. Predictive models, developed for rural two-lane, two-way facilities, are applied
in Steps 9, 10, and II of lhe method. These predictive models have been developed to estimate the predicted average crash frequency of an individual site which is an intersection or homogenous roadway segmento The facility is
divided into these individual sites in Step 5 of lhe predictive melhod.
E.ch predictive model in Chapter 10 consists of a safety performance function (SPF), crash modification factors (CMFs), and a calibration factor. The SPF is selected in Step 9 and is used to estimate the predicted average crash frequency for a site with base conditions. The estimate can be for either total crashes or organized by
crash-severity or collision-type distribution. In arder to account for differences between the base conditions and
the specific conditions ofthe site, CMFs are applied in Step 10, which adjust the prediction to account for the

CHAPTER lO-PREDICTIVE

METHOD FOR RURAL TWO-LANE,

TWO-WAY

10-35

ROAD5

geometric design and traffic control features of the site. Calibration factors are also used to adjust lhe prediction
to local conditions in the jurisdiction where the site is locate. The process for determinipg calibration factors for
the predictive models is described in Part C, Appendix A.l.
Section 10.12 presents six sample problems which detail lhe application of the predictive method. Appendix lOA
contains worksheets which can be used in the calculations for the predictive method steps.

10.12. SAMPLE PROBLEMS


In this section, six sample problems are presented using the predictive method for rural two-lane, two-way roads.
Sample Problems I and 2 illustrate how to calculate lhe predicted average crash frequency for rural two-lane roadway segments. Sample Problem 3 illustrates how to calculate the predicted average crash frequency for a stop-controlled intersection. Sample Problem 4 illustrates a similar calculation for a signalized intersection. Sample Problem
5 illustrates how to combine the results from Sample Problems I through 3 in a case where site-specific observed
crash data are available (i.e., using the site-specific EH Method). Sample Problem 6 illustrates how to combine the
results from Sample Problems 1 through 3 in a case where site-specific observed crash data are not available but
project-level observed crash data are available (i.e., using the project-Ievel EH Method).
Table 10-16. List of Sample Problems in Chapter la
PageNo.

Description

10-)5

Predicted

10-42

Problem No.

COi

a tangent roadway segment

Predicted average crash frequency

fOi

a curved roadway segment

10-49

Predicted

foc a three.leg

10-55

Predicted average crash frequency

HHiO

Expected

average crash frequency

for a facility when site-specific

observed

crash data are available

10-62

Expected

average crash frequency

tor a facility when site-specific

observed

crash data are not available

average crash frequency

average crash frequency

stop-controlled

foe a four-leg signalized

intersection

intersection

10.12.1. Sample Problem 1


The Site/Facility
A rural two-lane tangent roadway segmento

The Question
What is the predicted average crash frequency of lhe roadway segment for a particular year?

The Facts

1.5-mi length

Tangent roadway segment

10,000 vehlday

2% grade

6 driveways per mi

10-ft lane width

4-ft gravel shoulder


Roadside hazard rating

=4

-------HIGHWAY SAFETY MANUAL

10-36

Assumptions
Collision type distributions used are the default values presented in Table 10-4.
The calibration factor is assumed to be 1.10.

Results
Using the predictive method steps as outlined below, the predicted average crash frequency for the roadway segment
in Sample Problem 1 is determined to be 6.1 crashes per year (rounded to one decimal place).
Steps
Step 1 Ihrough 8
To determine Ihe predicted average crash frequency ofthe roadway segment in Sample Problem 1, only Steps 9
through 11 are conducted. No other steps are necessary beca use only ane roadway segment is analyzed foc ane year,
and Ihe EB Method is not applied.

~
:

Step 9-For the selected site, determine and apply the appropriate safety performance
site's facility type and traffic control features.
The SPF for a single roadway segment can be calculated from Equation 10-6 as follows:
N

~:i
.I

sp,if

Nsprrf

= AADT
=

x L x 365 x lO- x

function (SPF) for the

..(l.312)

10 ' 000 x 1.5 x 365 x 10-' x -"m)

= 4.008

crashes/year

Step 10-Multiply
the resull obtained in Step 9 by the appropriate CMFs to adjust the estimated crash
frequency for base conditions to the site-specific geometric design and traffic control features.
Each CMF used in the calculation ofthe predicled average crash frequency ofthe roadway segment is calculated below:

II

Lane Width (CMF,J


CMF" can be calculaled from Equation 10-11 as follows:

l'
I

,1'

'),'

CMF"

(CMFro - 1.0) x p,. + 1.0

For a 10-ft lane width and AADT of 10,000, CMF ro = 1.30 (see Table 10-8).

r.:
'-1

, nI

The proportion ofrelated crashes, Pro' is 0.574 (see discussion below Equation 10-11).
CMF"

(1.3 - 1.0) x 0.574 + 1.0 ~ 1.17

Shoulder Width and Type (CMF,,J


CMF" can be calculated from Equation 10-12, using values from Table 10-9, Table 10-10, and Table 10-4 as follows:
CMF"

= (CMFwro

x CMF,ro - 1.0) x Pro + 1.0

~ ,1

For 4-ft shoulders and AADT of 10,000, CMFwro


For 4-ft gravel shoulders, CMF,ro
The proportion ofrelated
CMF"

= (1.15

1.15 (see Table 10-9).

1.01 (see Table lO-lO).

crashes, Pro' is 0.574 (see discussion below Equation 10-12).

x 1.01 - 1.0) x 0.574 + 1.0 = 1.09

CHAPTER lD-PREDICTIVE

METHOD FOR RURAL TWO-LANE. TWO-WAY ROADS

10-37

Horizontal.Curves: Length, Radius, and Presence or Absence ojSpiral Transitions (CMFJ)


Sinee the roadway segment in Sample Problem 1 is a tangent, CMF, = 1.00 (Le., the base eondition for CMF k is
no curve).
_.
Horizontal Curves: Superelevaton (CMF,.J

Sinee the roadway segment in Sample Problem 1 is a tangent, aiul, therefore, has no superelevation, CMF, .= 1.00.
Grade (CMF,.)
From Table 10-11, for a two pereent grade, CMF ,. ~ 1.00
Driveway Density (CMF.)
The driveway density, DO, is 6 driveways per mile. CMF,. ean be ealculaled u~ing Equalion 10-17 as follows:

CMF6, =

0.322 + DD X [0.05 - 0.005 x In(AADT)]


0.322 + 5 X [0.05. - 0.005 x In(AADT)]

+ 6 X [0.05 - 0.005 x In(lO, 000)]


0.322 + 5 x [0.05 - 0.005 x In(IO,OOO)]
0.322

= 1.01
Centerline Rumble Strips (CMF,.J
. Sinee there are no eenterline rumble strips in Sample Problem 1, CMF"
is no centerline rumble strips).

.
1.00 (Le., the base eondition for CMF ,.

Passing Lanes (CMF,.J


Sinee lhere are no passing lnes in Sample Problem 1, CMF,. ~ 1.00 (i.e., the base eondition for CMF" is the
absenee of a passing lane).
1>vo-Way Left- Turn Lanes (CMF ,,)
Sinee there are no two-way left-turn lanes in Sample Problem 1, CMF"
is the absenee of a two-way left-tum lane).

1.00 (Le., lhe base eondition for CMF

,.

Roadside Design (CMFIO.J


The roadside hazard raling, RHR, in Sample Problem 1 is 4. CMF". ean be ealculaled /Tom Equalion 10-20 as follows:

CME:

__ e

(-o.6869+0.0668xRHR)

10, -

(-0.4865)

e( -0.6869 + 0.0668x 4)
e( -0.4865)

= 1.07

;'

Lightng (CMFII.J
Sinee lhere is no lighting in Sample Problem 1, CMFII
of roadway lighling).

1.00 (i.e., lhe base eondilion for CMFII is lhe absenee

Automated Speed Enjor-cement (CMFJ,.J


Since there is no automated speed enforcement in Sample Problem 1, CMF 12r
CMF 12. is lhe absenee of aulomaled speed enforeemenl).
The eombined CMF value for Sample Problem 1 is ealeulaled below.
CMF,omb~ 1.17

1.09

1.01

1.07 ~ 1.38

1.00 (Le., the base condtion for

.----------------

r
1
HIGHWAY SAFETY MANUAL

Step ll-Multiply
the resuIt obtained in Step 10 by the appropriate calibration factor.
It is assumed a calibration factor, e" of 1.10 has been determined for local conditions. See Par!
for further discussion on calibration of the predictive models.

e, Appendix

A.I

Calculation o/ Predicted Average Crash Frequency


.
The predicted average crash frequency is calculated using Equation 10-2 based on the rdults obtained in Steps 9
through II as follows:

~ 4.008 x 1.10 x (1.38)

= 6.084

crashes/year

WORKSHEETS
The step-by-step' instructions above are provided to illustrate the predictive method for calculating the predicted
average crash frequency for a roadway segmento To apply the predictive method steps to multiple segments,' a series
offive worksheets are provided for determining predicted average crash frequency. The five worksheets include:

Work.sheel SP lA (Corresponds lo Work.sheet lA)-Generallnformation

and Input Data for Rural Two-Lane,

Two-Way Roadway Segments

Work.sheet SP lB (Corresponds lo Work.sheet lB)-Crash

Modification Factors for Rural Two-Lane, Two- Way

Roadway Segments

Work.sheet SP 1C (Corresponds lo Work.sheel 1C)-Roadway

Segment Crashes for Rural Two-Lane, Two- Way

Roadway Segments

Work.sheel SP lD (Corresponds to Work.sheet lD)-Crashes

by Severity Level and Collision Type for Rural

Two-Lane, Two-Way Roadway Segments

Work.sheelSP 1E (Corresponds lo Work.sheel1E)-Summary

Results for Rural Two-Lane, Two-Way Roadway Segments

Details of these sample problem worksheets are provided below. Blank versions of corresponding worksheets are
provided in Chapler 10, Appendix lOA.
.

Worksheet SP1A-Generallnformation
and Input Data for Rural Two-Lane. Two-Way Roadway Segments
Worksheet SPIA is a summary of general information about the roadway segment, analysis, input data (i.e., "The
Facts"), and assumptions for Sample Problem 1.

ii

,.
I
I

.'

,1

1-,

j'

También podría gustarte