Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Modelling of Traffic Loads on Bridges Based on Measurements of Real Traffic Loads in Sweden

Carlsson, Fredrik LU (2006)
Abstract
This thesis treats traffic load effects in road bridges and in particular bending moment and shear force in different types of bridges. All traffic load effects are described in terms of statistical distribution functions for annual extreme values. The statistical distributions are based on measurements of real traffic, made in Sweden during 2002-2003. The measurements are managed by the Swedish Road Administration and accounted for in Vägverket (2004b). The data base contains information about 280000 vehicles out of which approximately 31000 are classified as heavy vehicles. Only vehicles that are defined as heavy are decisive for the design of bridges. The data base contains information about the time when the vehicles passed the... (More)
This thesis treats traffic load effects in road bridges and in particular bending moment and shear force in different types of bridges. All traffic load effects are described in terms of statistical distribution functions for annual extreme values. The statistical distributions are based on measurements of real traffic, made in Sweden during 2002-2003. The measurements are managed by the Swedish Road Administration and accounted for in Vägverket (2004b). The data base contains information about 280000 vehicles out of which approximately 31000 are classified as heavy vehicles. Only vehicles that are defined as heavy are decisive for the design of bridges. The data base contains information about the time when the vehicles passed the measuring station, the number of axles on the vehicle, the mutual distance between axles and weight of every single axle. With such information it is possible to determine section forces in different types of bridges very accurately. Information about when vehicles pass the measuring station is used to verify assumptions regarding for example the mutual distance between two consecutive heavy vehicles. This is of interest when section forces in longer bridges are studied.



For variable loads the characteristic value is used to define the load. Codes used to design bridges, usually define the characteristic value as the 98th percentile of the annual maximum distribution, i.e. a 50 year return period. The data material used here is collected during 51 days at 8 different locations in Sweden. Using the measured vehicles as input the statistical distribution functions for traffic load effects in different types of bridges are determined. The information of interest here is to describe the maximum load effects during the reference period, so called extreme value distributions. Collecting the necessary amount of data to determine extreme value distributions directly would be too costly and time consuming. Instead some kind of extrapolation method must be applied. In this thesis the POT-method, (Peaks Over Thresholds) and simulations are used to determine the extreme value distributions for traffic load effects. The POT-method implies that only the section forces which exceed a certain level (threshold), are fitted to a standard distribution. From that distribution the extreme value distributions for the section forces are determined. The advantage of using this method is that it is possible to determine the extreme value distributions and the characteristic values at different traffic intensities in an easy way. This report contains tools to determine the extreme value distributions for section forces in bridges with different traffic intensity.



In probability based analysis of structures statistical information about all the basic variables in the limit state function must be available, i.e. the statistical distribution functions must be known. The most dominating variable load on bridges is the load generated by traffic; see e.g. Das (1997) and Bailey (1996). These loads affect the result of the bridge analysis to a great extent and accurate information about these loads is therefore very important.



When a new bridge is designed the consequence of using too high loads is in most cases negligible, i.e. the extra cost for a larger cross section is manageable. In contrast, if the safety level in an existing bridge is considered too low, the load level is of major importance. Strengthening or in the worst case replacement of the bridge is very expensive. Evaluation of existing bridges are often based on a deterministic analysis where the loads are taken from some bridge code, e.g. Vägverket (2004a), i.e. the actual traffic load for the specific bridge is not taken into account.



The traffic loads in deterministic codes are generally very conservative because they are intended to cover all types of bridges. Instead of using deterministic methods to determine the safety level of an existing bridge, more sophisticated methods that can account for specific information about a bridge should be applied. Probability based methods are very applicable in these cases.



An advantage with probability based methods is that specific information about the bridge can be included in the analysis. An example of such information is measurements of the material properties in the bridge. Another possibility more seldom used is to determine the traffic load effects based on the traffic intensity at the actual bridge site. The traffic intensity is measured by the Swedish road administration on many locations on the Swedish road net. With the methods described in this report it is possible to determine the extreme value distributions for the section forces with the traffic intensity as input. Two case studies are used to illustrate the method.



Of course, a probability based method is not the whole solutions to bridges with insufficient safety. The method shall be seen as one out of several possible ways to evaluate the condition of a bridge. Every bridge that can be saved for a new period of time by use of more sophisticated methods implies substantial saving of money. (Less)
Abstract (Swedish)
Popular Abstract in Swedish

Denna rapport behandlar trafiklaster på broar och de lasteffekter som de ger upphov till. Böjmoment och tvärkraft för olika brotyper beskrivs i form av statistiska fördelningar med tillhörande parametrar. Dessa är baserade på mätningar av verklig trafik gjorda i Sverige under åren 2002-2003 i Vägverkets regi. Resultaten finns redovisade i Vägverket (2004b), och finns lagrade i en databas som innehåller information om ca. 280 000 fordon varav ca. 31000 är klassade som tunga fordon. Intressanta fordon i detta sammanhang är de tunga fordonen och databasen innehåller mycket information om dessa. Exempel på sådan information är, tiden då fordonen passerade mätstationen, fordonets antal axlar, varje... (More)
Popular Abstract in Swedish

Denna rapport behandlar trafiklaster på broar och de lasteffekter som de ger upphov till. Böjmoment och tvärkraft för olika brotyper beskrivs i form av statistiska fördelningar med tillhörande parametrar. Dessa är baserade på mätningar av verklig trafik gjorda i Sverige under åren 2002-2003 i Vägverkets regi. Resultaten finns redovisade i Vägverket (2004b), och finns lagrade i en databas som innehåller information om ca. 280 000 fordon varav ca. 31000 är klassade som tunga fordon. Intressanta fordon i detta sammanhang är de tunga fordonen och databasen innehåller mycket information om dessa. Exempel på sådan information är, tiden då fordonen passerade mätstationen, fordonets antal axlar, varje enskild axels inbördes läge och vikt. Med denna information är det möjligt att bestämma snittkrafter i olika brotyper med mycket hög precision. Information om när fordonen passerade mätstationen utnyttjas i denna rapport för att t.ex. undersöka inbördes avstånd mellan tunga fordon, vilket är av intresse då snittkrafter i längre broar studeras.



Variabla laster definieras i allmänhet genom sina karakteristiska värden, som normalt definieras som 98% fraktilen av årsmaximum. Datamaterialet som används för att bestämma fördelningsfunktioner för snittkrafter är insamlat under 51 dygn fördelat på 8 olika geografiska platser. För att bestämma snittkrafternas extremvärdesfördelningar, dvs. fördelningar som beskriver hur de maximala snittkrafterna under en given referensperiod är fördelade, måste någon typ av extrapolationsmetod utnyttjas. I denna rapport utnyttjas POT-metoden, (Peaks Over Thresholds) och simuleringar. POT-metoden innebär att bara de snittkrafter som överskrider en på förhand bestämd nivå (Threshold), passas till en känd fördelning. Utifrån denna fördelning kan snittkrafternas extremvärdesfördelningar bestämmas. Fördelen med denna metod är att det är möjligt att bestämma fördelningar och karakteristiska värden vid olika trafikintensiteter. I avhandlingen presenteras en metod för att bestämma extremvärdesfördelningar för snittkrafter som funktion av trafik intensiteten på en specifik bro.



Vid sannolikhetsbaserade analyser av konstruktioner måste alla grundvariablerna i brottekvationen beskrivas med statistiska fördelningar. För broar är trafiklasten den dominerande variabla lasten, se t.ex. Das (1997) och Bailey (1996), varför god information om dessa laster är mycket viktig. Vid nykonstruktion av en bro spelar det ingen större roll om lasterna är något överskattade. Marginellt ökande dimensioner innebär en relativt liten extra kostnad. Så är inte fallet om en existerande bros säkerhet kan anses vara för låg. Lasternas storlek är då avgörande eftersom en förstärkning eller ersättning av en bro kostar mycket pengar. Utvärdering av en befintlig bros säkerhetsnivå baseras normalt på en deterministisk analys med lasterna föreskrivna i normregler, t.ex. BRO 04. Lasterna är i dessa fall konservativa eftersom att de är valda för att täcka olika typer av broar, och specifik information om den aktuella bron kan inte beaktas i analysen. Sannolikhetsbaserade analyser är mycket tillämpbara i detta fall eftersom det är möjligt att implementera specifik information angående bron i analysen. Exempel på sådan information är mätresultat av materialegenskaper i bron. Denna information är ofta tillgänglig eller möjlig att ta fram. Trafiklasternas storlek för en specifik bro beror på trafikintensiteten som finns uppmätt på många ställen på det Svenska vägnätet. Med metoder beskrivna i denna rapport är det möjligt att med trafikintensitet som indata bestämma extremvärdesfördelningar för snittkrafter. Metodiken illustreras med hjälp av fallstudier för två specifika broar.



Sannolikhetsbaserade analys skall ses som en av flera möjligheter som konstruktörer kan använda för att visa att en bros säkerhetsnivå är tillfredställande. Varje bro som befinner sig i gränslandet mellan säker och osäker och som kan räddas med en mer sofistikerad analysmetod innebär att stora pengar kan sparas. (Less)
Please use this url to cite or link to this publication:
author
supervisor
opponent
  • Professor Sundquist, Håkan, KTH, Stockholm
organization
publishing date
type
Thesis
publication status
published
subject
keywords
heavy vehicles, traffic intensity, Peaks Over Thresholds, axle and bogie pressure, traffic load effects, probability based classification, Technological sciences, Teknik
pages
246 pages
publisher
Structural Engineering, Lund University
defense location
Room V:C, V-building, John Ericssons v 1, Lund Institute of Technology
defense date
2006-04-07 10:15:00
external identifiers
  • other:ISRN: LUTVDG/TVBK-1032-SE(268)
language
English
LU publication?
yes
id
1794dc14-3076-4a0c-ae0c-3c9c724fdeba (old id 25826)
date added to LUP
2016-04-01 16:59:17
date last changed
2018-11-21 20:45:44
@phdthesis{1794dc14-3076-4a0c-ae0c-3c9c724fdeba,
  abstract     = {{This thesis treats traffic load effects in road bridges and in particular bending moment and shear force in different types of bridges. All traffic load effects are described in terms of statistical distribution functions for annual extreme values. The statistical distributions are based on measurements of real traffic, made in Sweden during 2002-2003. The measurements are managed by the Swedish Road Administration and accounted for in Vägverket (2004b). The data base contains information about 280000 vehicles out of which approximately 31000 are classified as heavy vehicles. Only vehicles that are defined as heavy are decisive for the design of bridges. The data base contains information about the time when the vehicles passed the measuring station, the number of axles on the vehicle, the mutual distance between axles and weight of every single axle. With such information it is possible to determine section forces in different types of bridges very accurately. Information about when vehicles pass the measuring station is used to verify assumptions regarding for example the mutual distance between two consecutive heavy vehicles. This is of interest when section forces in longer bridges are studied.<br/><br>
<br/><br>
For variable loads the characteristic value is used to define the load. Codes used to design bridges, usually define the characteristic value as the 98th percentile of the annual maximum distribution, i.e. a 50 year return period. The data material used here is collected during 51 days at 8 different locations in Sweden. Using the measured vehicles as input the statistical distribution functions for traffic load effects in different types of bridges are determined. The information of interest here is to describe the maximum load effects during the reference period, so called extreme value distributions. Collecting the necessary amount of data to determine extreme value distributions directly would be too costly and time consuming. Instead some kind of extrapolation method must be applied. In this thesis the POT-method, (Peaks Over Thresholds) and simulations are used to determine the extreme value distributions for traffic load effects. The POT-method implies that only the section forces which exceed a certain level (threshold), are fitted to a standard distribution. From that distribution the extreme value distributions for the section forces are determined. The advantage of using this method is that it is possible to determine the extreme value distributions and the characteristic values at different traffic intensities in an easy way. This report contains tools to determine the extreme value distributions for section forces in bridges with different traffic intensity.<br/><br>
<br/><br>
In probability based analysis of structures statistical information about all the basic variables in the limit state function must be available, i.e. the statistical distribution functions must be known. The most dominating variable load on bridges is the load generated by traffic; see e.g. Das (1997) and Bailey (1996). These loads affect the result of the bridge analysis to a great extent and accurate information about these loads is therefore very important.<br/><br>
<br/><br>
When a new bridge is designed the consequence of using too high loads is in most cases negligible, i.e. the extra cost for a larger cross section is manageable. In contrast, if the safety level in an existing bridge is considered too low, the load level is of major importance. Strengthening or in the worst case replacement of the bridge is very expensive. Evaluation of existing bridges are often based on a deterministic analysis where the loads are taken from some bridge code, e.g. Vägverket (2004a), i.e. the actual traffic load for the specific bridge is not taken into account.<br/><br>
<br/><br>
The traffic loads in deterministic codes are generally very conservative because they are intended to cover all types of bridges. Instead of using deterministic methods to determine the safety level of an existing bridge, more sophisticated methods that can account for specific information about a bridge should be applied. Probability based methods are very applicable in these cases.<br/><br>
<br/><br>
An advantage with probability based methods is that specific information about the bridge can be included in the analysis. An example of such information is measurements of the material properties in the bridge. Another possibility more seldom used is to determine the traffic load effects based on the traffic intensity at the actual bridge site. The traffic intensity is measured by the Swedish road administration on many locations on the Swedish road net. With the methods described in this report it is possible to determine the extreme value distributions for the section forces with the traffic intensity as input. Two case studies are used to illustrate the method.<br/><br>
<br/><br>
Of course, a probability based method is not the whole solutions to bridges with insufficient safety. The method shall be seen as one out of several possible ways to evaluate the condition of a bridge. Every bridge that can be saved for a new period of time by use of more sophisticated methods implies substantial saving of money.}},
  author       = {{Carlsson, Fredrik}},
  keywords     = {{heavy vehicles; traffic intensity; Peaks Over Thresholds; axle and bogie pressure; traffic load effects; probability based classification; Technological sciences; Teknik}},
  language     = {{eng}},
  publisher    = {{Structural Engineering, Lund University}},
  school       = {{Lund University}},
  title        = {{Modelling of Traffic Loads on Bridges Based on Measurements of Real Traffic Loads in Sweden}},
  year         = {{2006}},
}