Bridge live load models from weigh-in-motion data

Heywood, Robert J. (1992). Bridge live load models from weigh-in-motion data PhD Thesis, School of Engineering, The University of Queensland. doi:10.14264/uql.2014.611

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
THE8039.pdf Thesis full text application/pdf 43.75MB 1

Author Heywood, Robert J.
Thesis Title Bridge live load models from weigh-in-motion data
School, Centre or Institute School of Engineering
Institution The University of Queensland
DOI 10.14264/uql.2014.611
Publication date 1992
Thesis type PhD Thesis
Supervisor Colin O'Connor
Total pages 383
Language eng
Subjects L
290800 Civil Engineering
Formatted abstract

This thesis presents the development of a methodology for deriving statistical highway bridge live load models for short span bridges. The methodology is applied to Australian weigh-in-motion data to establish live load models for Australian bridges. The methodology can readily be applied to weigh-in-motion databases collected in other locations of the world.

The recent development of weigh-in-motion technology provided the opportunity to collect and investigate an extensive vehicle database which was collected in an unbiased manner. The basis of the live load models is a database of some 697,547 vehicles collected from 31 sites around Australia using the CULWAY weigh-in-motion system. This represents 7.7 years of continuous recording of Australian roads.

The bridge live load models in use around the world are reviewed, and demonstrate a wide variety of solutions to representing the traffic loads applied to bridges. It is demonstrated that some of these models do not satisfactorily represent Australian traffic.

The live load models derived are presented in a form which is consistent with reliability analyses. Thus the live load models are suitable for use in the calibration of the Australian AUSTROADS Bridge Design Code.

Statistical models of effects such as bending and shear are considered the most appropriate models for live load. Their disadvantage lies in their dependence on the very extensive variety of structural forms evident in the bridge infrastructure. This is counteracted through the good correlations between the effects induced in bridges and span. These correlations are demonstrated for the six simple and continuous influence lines and eight spans between 5 m to 40 m chosen to represent the bridge infrastructure.

Empirical power relationships between effect and span are presented for spans between 10 m and 40 m. The resulting coefficients, described as bridge indices and bridge exponents, facilitate the reduction of the live load models for a selected influence line to two parameters. For general purposb traffic and midspan moments in simple spans, the two parameters may be reasonably approximated by a single parameter known as the bridge site index. The bridge site index facilitates quick comparisons of the intensity of loadings of different routes. Its many applications in the area of bridge management are discussed.

The data is analysed using statistical and structural engineering procedures to develop statistical live load models for each site and for Australia. Live load models are developed for normal traffic and for remote outback Australian sites where road trains of up to 150 t dominate bridge loading considerations.

The tendency for critical vehicles at a site to be long or short can be rationally determined from the bridge exponent. This is utilised to develop a consistent methodology for ensuring only sites of similar traffic characteristics are combined during the development of the Australian models. Three groups of sites are identified although these are subsequently reduced to two groups - general access and remote.  Statistical distributions of live load effects are presented as cumulative distribution functions plotted on probability paper for each site and for the combined national groupings of sites. Summary means and coefficients of variation provide a compact basis for future research into the reliability of Australian bridges.

A methodology based on the equivalent base length concept has been used by the Ministry of Transport, Ontario to derive a bridge design vehicle from vehicle data. This methodology is demonstrated but differs the live load models of effect developed in this thesis. An optimisation routine is developed to design vehicles that best represent the statistical models of moments, shears and reactions calculated from the weigh-inmotion database. The optimisation process identified a 55 t vehicle with fixed axle spacings to represent the general access traffic with a root mean squared error of 2.4%. A 90 t second vehicle was selected (with similar accuracy) for the remote sites which are dominated by road trains.

The resulting vehicles and the statistical model of effects are compared with the current Australian design live load model and the notional strength of Australian bridges. The data demonstrates that vehicles are regularly overloaded. This raises questions as to the capacity of old Australian bridges to sustain the levels of load applied, especially after consideration of their age, often substantially smaller design loads and level of deterioration. One major element in their favour, is the low volumes of traffic on many of the older bridges. 

The influence of traffic volume on multiple presence effects in two-way bridges ofshort span is investigated through a case study. The traffic stream is reproduced from the weigh-in-motion records for a site and the effects induced by vehicles in one and two lanes calculated by simulation. The arbitrary-point-in-time distribution for a two lane bridge, derived using a convolution procedure, demonstrates good correlation with that calculated by simulation. The methodology developed facilitates research into the influence of traffic volume and recurrence interval on the effects induced in the bridge under investigation. It concludes, that for short spans and the ultimate limit state, the influence of the second lane of traffic is smaller than presented in existing codes. This is especially marked for the low traffic volumes that are characteristic of the routes containing the majority of old Australian bridges. The sensitivity of lane reduction factors to traffic volume is shown to reduce with the recurrence interval, illustrating the possibility that lane reduction factors should be reduced for serviceability and fatigue considerations.

The major contributions of the research presented in this thesis are summarised below:

♦    A review of the literature relating to bridge live load models, multiple presence and reliability theory.

♦    A methodology for deriving live load models for bridges from weigh-in-motion data.

♦    Statistical live load models for Australian short span bridges.

♦    The determination of multipliers which relate one day maxima to values for longer periods. 

♦    The development and application of an optimisation procedure to derive vehicles to represent the average extreme daily effects induced in Australian bridges.

♦    The derivation of two vehicles with fixed axle spacings to represent Australian traffic.

♦    A methodology and case study to calculate the multiple presence effects in two-lane two-way bridges and investigate the influences of traffic volume.

♦    The evidence that multiple presence reduction factors are different for serviceability and ultimate limit states and that current factors may be conservative, especially for
low traffic volumes.

There remains a need to extend the research to develop live load and multiple presence models for medium and long spans bridges. The representative vehicles developed for short spans demonstrate that the current T44 loading is inconsistent with the traffic loadings applied to Australian bridges. The live load models are suggested as the basis for the development and calibration of a new bridge design load for Australia. The statistical live load model for Australia provides many opportunities to undertake reliability based research associated with the evaluation of the nation's bridge infrastructure.

Keyword Bridges -- Live loads

Document type: Thesis
Collection: UQ Theses (RHD) - UQ staff and students only
Citation counts: Google Scholar Search Google Scholar
Created: Mon, 08 Dec 2014, 12:15:49 EST by Ms Christine Heslehurst on behalf of Research Management Office