Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
3Activity
0 of .
Results for:
No results containing your search query
P. 1
Reliability Indices for Road Geometric Design

Reliability Indices for Road Geometric Design

Ratings: (0)|Views: 163 |Likes:
Published by youkesi

More info:

Published by: youkesi on Mar 19, 2010
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

10/23/2012

pdf

text

original

 
Reliability indices for road geometric design
FRANCIS . D. NAVIN
Department of Civil Engineering, The University of British Columbia,
2324
Main Mall, Vancouver, B.C.,Canada V6T
I
W5
Received February
8,
1991Revised manuscript accepted January 3 1, 1992Highway engineers, when asked to state the safety of a particular design, are usually at a loss to give a single mean-ingful measure as is possible in structural or geotechnical engineering. This paper outlines a method to estimate themargin of safety and reliability index for isolated highway components. The stopping sight distance is used to demon-strate the method. The method uses the basic highway design equations. On the assumption that the variables are ran-dom, the expected value of the mean and the variance are estimated; and from these the margin of safety and thereliability index are calculated. The most likely combination of variables for the existing design condition may alsobe estimated. The variables included represent the characteristics of the driver, the vehicle, and the road surface.
A
method is proposed to specify the design parameter's value representing a road's strategic importance, the users,the vehicles, the drivers, the environment, the terrain, and the standard of design and construction. The apparent advan-tage of the proposed reliability-based method is that the designer must explicitly specify the importance of the modify-ing factors and may also more closely investigate the behaviour of the variables in the design parameters in the criticalregion near noncompliance.
Key words:
limit states design, stopping sight distance, safety, highway design, reliability.Lorsqu'on leur demande de se prononcer sur la securiti d'un concept particulier, les ingknieurs routiers ont habituel-lement de la peine
a
fournir une seule mesure valable comme cela est le cas en genie geotechnique ou en genie desstructures. Cet article dicrit une mithode d'evaluation de la marge de securite et de I'indice de fiabiliti de composantesroutikresisolies. La distance de visibiliti d'arrEt est utilisie pour dimontre la methode. Celle-ci a recours aux iquationsde conception routikre de base et retient comme hypothkse que les variables sont aleatoires; la marge de securitk etI'indice de fiabiliti peuvent Etre calcules
a
partir de ces iquations. La combinaison de variables la plus probable pourles conditions de conception existantes peut egalement Etre estimie. Les variables incluses representent les caractiristi-ques du conducteur, du vihicule et du revstement routier.Une methode est proposie afin de spicifier la valeur du paramktre de conception reprisentant I'importance stratigiquede la route, les utilisateurs, les vihicules, les conducteurs, I'environnement, le terrain et la norme de conception etde construction. L'avantage de la mithode proposee est que le concepteur doit spicifier clairement I'importance desfacteurs de modification et qu'il peut igalement analyser de plus prks le comportement des variables dans les parame-tres de conception qui se trouvent dans la region critique
a
proximite de la dirogation.
Mots clis
:
calcul aux Ctats limites, distance de visibiliti d'arrEt, securiti, conception des autoroutes fiabiliti.[Traduit par la redaction]
Can.
J.
Civ. Eng.
19,
760-766
(1992)
Introduction
Road geometric design procedures have been criticized byOlsen
et
al.
(1984), Navin (1986), Hauer (1988), Hutchinsonand Parker (1989), and Glennon (1989) for failing to meetsome of the established operational safety standards undercertain conditions. Road design procedures have evolved inmuch the same manner as those in other engineering fields,i.e., years of practical experience backed by theory whenpossible. The end result is an implicit assumption that if thepublished standards have been correctly applied, then theroad has an adequate margin of safety. This assumption isalso accepted by the courts when ruling on the designer'sliability for vehicular accidents when road geometry oroperation is suspected. Road geometric design standards arefound in RTAC (1976) and AASHTO (1984), the opera-tional standards are in UTCD (1976) and the ITE Handbook(1982). These standards have developed a very efficientdesign and building process, but have generally laggedbehind the reality of the actual drivers and vehicles operatingon the nation's roads.
NOTE:
Written discussion of this paper is welcomed and will bereceived by the Editor until February28, 1993 (address insidefront cover).
Primed in Canada
/
Imprimc
au Canada
The purpose of this paper is to outline a new road designmethod based on the reliability concept of limit state designthat is used in structural engineering. The proposed methodwill allow for the direct input into design equations ofchanges that may characterize the future of either the vehi-cle, the driver, or the pavement surface. As well, the char-acteristics of the design parameters in the region of interestmay also be explored to gain additional engineering insights.
Limit states design definitions
The structural engineer's design objectives are similar tothose of the highway engineer, which are to provide a safereliable structure that does its job; to provide a structurethat is economical to build and maintain; and to providea structure that is aesthetically pleasing (see Kulak
et
al.
1985). Design, to the structural engineer, means the geo-metrical arrangement of structural members as well as theselection of adequately large members to resist the imposedforces and bending moments determined during analysis.As structures have become more complicated and engineer-ing knowledge more complete, the analysis has advancedfrom the relatively simple procedures for statically deter-minate structures to those of limit state design.
 
NAVIN
76
1
Much the same has occurred in highway design. The maindifference for highways is that road geometry design hasremained relatively stable while vehicle design, and in par-ticular truck design, has changed greatly. The highwayengineer must determine a road's size, specify the detailgeometry as well as locate the road. In a way the highwayengineer is both architect and engineer of the road.Limit states design as used by the structural engineer,according to Kulak
et al.
(1985), is a design method in whichthe performance of a structure is checked against the limitingconditions at the appropriate load levels. Two limits areused: the ultimate limit state which concerns safety fromexceeding some structural boundary and the serviceabilitylimit state where the structure does not behave as intended.In the analysis process, the structure is subject to reasonablemaximum loads and the members selected so a reasonablemargin of safety remains. The serviceability limit subjectsthe structure to "service loads." The ultimate limit statecriterion is shown in Fig. 1. The loading demand is somerandom variable, D, distributed about a mean value of
D
and the resistance of the structure, R, a random variablewith mean
R.
In this example the supply R is exceeded bythe demand D in the shaded area where the structure is con-sidered to have failed. Noncompliance or failure is definedby the engineer or an appropriate standard. The objectiveof limit state design is to reduce the area of overlappingdemand and supply to some appropriately small value.The basic structural engineering equation for checking theultimate limit state condition iswhere
4
is the resistance or performance factor; R is thenominal resistance of a structural element;
y
s the impor-tance factor;
$
is the load combination factor; and
a
areload factors with subscripts D for dead load, L for live load,
Q
for wind and earthquake, and T for all the others.The resistance factor is applied to the nominal strengthof the member and reflects the anticipated variation due tomaterial properties, dimensions, and workmanship. Theimportance factor relates the consequences of a collapse touse and occupancy of the structure. The load combinationfactor reduces the probability of failure if a number of loadsfrom different sources are acting simultaneously. The loadfactor takes into account the possibility of loads larger thanthose anticipated to act on the structure, the uncertaintyinvolved in predicting the loads, and the approximations inthe analysis of the effects of the loads on the structure.This technique has been shown to underestimate actualstructural failures by a factor of 10 and is considered of lit-tle value for real-world structural processes. Madsen
et al.
(1986) argue that the purpose of the theory is not a descrip-tion of the performance of the structures but the
control
of the design process intended to efficiently produce reliablestructures. The underestimation is the result of human errorsin the design and construction process which are notexplicitly considered in the equations.
Measures of safety
The simplest measure of safety is the central factor ofsafety, which is defined as the ratio of the average resistancesupplied, R, and the average demand, D (see Hart 1982).The more usual measure is the conventional factor of safety.
MAGNITUDE
(unlts
I
FIG.
1.
Frequency distribution curves for demand and supply.
In this case the average demand is increased by some multi-ple,
k,
of the standard deviation, and the supply is reducedby some multiple of the standard deviation. The ratio ofthese two values is the conventional factor of safety. Thisapproach implies that designers are uncertain about the exactvalues and allow for more demand and a less supply of theengineering design parameter.Extending this idea of uncertainty further, it is possibleto have a finite chance that the demand will exceed supply;for example, the stopping sight distance required by thedriver and the vehicle exceeds that provided by the highwaydesign. For the moment there is no need to go into thereasons why the demand may exceed supply, but simplyaccept it as an event that may occur. Ang and Tang (1984)give a method to derive the expected value and variance froman algebraic equation and also the derivation of twomeasures of safety. The margin of safety (eq. [2]) is the dif-ference between the expected value of the supply and thatof the demand. The ratio of the margin of safety and thecombined variance shown in eq. [3] is defined as thereliability index.These definitions will now be applied to the stopping sightdistance on a level road as an illustration of how reliabilitymethods may be applied to highway geometric elements.
Stopping sight distance
Stopping sight distance is fundamental to all highwaydesign and the current state of the art has been well coveredby Olsen
et al.
(1984) and, more recently, in TransportationResearch Record 1208 (TRB 1989). To calculate the stoppingsight distance, suitable values are assigned to the variablesof eq. [4]. This equation represents the mechanics in thestopping sight distance, and includes the vehicle speed,
V,
the driver's perception-reaction time,
T,
and the vehicle'sability to stop,
a?.
v2
[4]SSD
=
VT
+
7
2%This basic equation can be manipulated to calculate theexpected demand by the driver-machine, E(SSDDM),and
 
762
CAN.
J.
CIV.ENG.
VOL.
19,
1992
TABLE
.
Values for stopping sight distance variablesHighway, AASHTOVariable High Low Truck SourceAASHTO 1984Olsen 1984UMTRI 1987Olsen 1984Olsen 1984Olsen 1984EstimatedEstimatedEstimatedEstimatedEstimatedEstimatedthe corresponding variance, VAR(SSDDM).The process tobetween the driver's ability to brake the vehicle and the vehi-derive the equations is given in Ang and Tang (1984). Thecle's ability to stop and that all the variables are correlatedvalues of interest arein some known degree.The general relationship between the stopping sightE( v;)distance "supplied" by the highway design may be either151E(SSDDM)
=
E( VD)E(TD)
+
-
~(a?)a single value from the design manual or the actual valuesupplied because of construction procedures and changes
2
over time and with the quality of the road surface. For the
+
COV( VD, TD)
+
2~ (a?)following analysis, a single standard design value, SSDH,is used for the highway's supplied value. Noncompliance
-
E'
'D
[COV( vD, a?)is defined as the demanded stopping sight distance exceeding~EI(~?)~Ithat supplied by the highway standard. To compare the
"SUDD~V"
and "demand" of the measures of interest. theE( vk)mar3neof safety and the reliability index are
+
COV(CY?, VD)]
+
ZE[(CY?)
1
'"
[7]M(SSD)
=
SSD,
-
E(SSDDM)The stopping distance demanded by the driver-machinesystem depends on the speed that the driver selects, his orher perception reaction time, and the stopping capabilitiesof the vehicle on a particular type of road surface. Thedemand equation also assumes that there is a relationshipand
Data
The mean, standard deviation, and correlation coefficientof the variables used in the stopping sight distance calcula-tion are given in Table 1. There are few sources of good dataavailable. The following summarizes those found. The dis-tribution of highway speeds has been shown by Olsen
et al.
(1984) for the U.S. and Lay (1986) for Australia to be nor-mally distributed with a standard deviation of 0.10-0.15 ofthe mean. Gerlough and Huber (1975) fitted a log normalcurve to spot speed data from a midwest U.S. freeway. Thisanalysis assumes a normal distribution for speed.The reported brake reaction time has been shown byBrown (1990) to have increased dramatically from about0.6 s for 1935 U.S. laboratory tests to between 1.25 and1.60 s for 1985 on-road tests by Triggs and Harris (1982)in Australia. Olsen
et al.
(1984) also investigated theperception-reaction time for the alerted and surpriseddrivers. Using his data, the estimated mean is 1.10 s and thestandard deviation is 0.24 s for surprised drivers. A normaldistribution was found by the author to be adequate forOlsen's data.Data from many sources are reported by UMTRI (1987)for the deceleration rate of trucks on wet pavement. Values

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->