Monitoring Guidelines for Railway Bridges

(1)

Monitoring Guidelines for Railway Bridges

0 5 10 15 20 25

-100 -50 0

strain [με]

stringer, location N1

0 5 10 15 20 25

0 50 100

strain [με]

time [s]

stringer, location N2

PRIORITY 6

SUSTAINABLE DEVELOPMENT GLOBAL CHANGE & ECOSYSTEMS

INTEGRATED PROJECT

(2)

This report is one of the deliverables from the Integrated Research Project “Sustainable Bridges - Assessment for Future Traffic Demands and Longer Lives” funded by the European Commission within 6^th Framework Pro- gramme. The Project aims to help European railways to meet increasing transportation demands, which can only be accommodated on the existing railway network by allowing the passage of heavier freight trains and faster passenger trains. This requires that the existing bridges within the network have to be upgraded without causing unnecessary disruption to the carriage of goods and passengers, and without compromising the safety and econ- omy of the railways.

A consortium, consisting of 32 partners drawn from railway bridge owners, consultants, contractors, research institutes and universities, has carried out the Project, which has a gross budget of more than 10 million Euros.

The European Commission has provided substantial funding, with the balancing funding has been coming from the Project partners. Skanska Sverige AB has provided the overall co-ordination of the Project, whilst Luleå Tech- nical University has undertaken the scientific leadership.

The Project has developed improved procedures and methods for inspection, testing, monitoring and condition assessment, of railway bridges. Furthermore, it has developed advanced methodologies for assessing the safe carrying capacity of bridges and better engineering solutions for repair and strengthening of bridges that are found to be in need of attention.

The authors of this report have used their best endeavours to ensure that the information presented here is of the highest quality. However, no liability can be accepted by the authors for any loss caused by its use.

Figure on the front page: Photo of an engine passing on the Keräsjokk Bridge (Sweden) and strain measurements of a train crossing.

Project acronym: Sustainable Bridges

Project full title: Sustainable Bridges – Assessment for Future Traffic Demands and Longer Lives Contract number: TIP3-CT-2003-001653

Project start and end date: 2003-12-01 -- 2007-11-30 Duration 48 months

Document number: Deliverable D5.2 Abbreviation SB-MON

Author/s: G. Feltrin (Editor), Empa Date of original release: 2007-11-30

Revision date:

Project co-funded by the European Commission within the Sixth Framework Programme (2002-2006)

Dissemination Level

PU Public X

PP Restricted to other programme participants (including the Commission Services) RE Restricted to a group specified by the consortium (including the Commission Services) CO Confidential, only for members of the consortium (including the Commission Services)

(3)

Summary

This document provides recommendations how to specify, design, implement and operate monitoring systems in a systematic and coherent way. It defines the actors and their roles within the monitoring activity. The guideline introduces the concept of model monitoring system as the fundamental planning tool for specifying the physical monitoring system. This tool allows bridge owners and structural engineers to specify their requirements on a monitoring system by using concepts that are familiar to them. The concept of model monitoring system permits to separate the roles and re- sponsibilities of the different actors and to clearly define the interface between structural engineer (bridge expert) and monitoring experts. The model monitoring system enhances the role of the structural engineer within the monitoring process by placing his knowledge of bridges in the centre of the process.

This guideline requires that the design of the model monitoring system have to be based on a bridge model. This requirement automatically provides an interpretation scheme for the data generated by the physical monitoring system without it this data would be meaningless.

The task of the monitoring expert is to implement and operate a monitoring system that conforms to model monitoring system specified by the structural engineer. Since monitoring technology evolves rapidly, this guideline does not address in detail the technological oriented components of the monitoring process. Nevertheless, a monitoring toolbox is included, which provides briefly the most relevant information to different methods, algorithms and sensors being in use in monitoring. The goal of the toolbox is to provide condensed technological background information for the structural engineer. This information allows him to influence the design of the physical monitoring system.

The appendices contain guidelines of the novel monitoring techniques developed within the project. The scope of these guidelines is to promote the use of these novel techniques in practical monitoring activities.

Additional background information is provided by 5 documents listed below:

– SB-5.1 Monitoring Instrumentation and Techniques (Feltrin et al., 2007)

– SB-5.2.S1 Guidelines for Monitoring of Steel Railway Bridges (Sedlacek et al., 2007a)

– SB-5.2.S2 Guideline for Estimating Structural Damping of Railway Bridges (Feltrin and Gsell, 2007)

– SB-5.2.S3 Corrosion Monitoring Systems for Reinforced Concrete Bridges (Sørensen and Frølund, 2007)

– SB-5.2.S4 Estimating Reliability of Monitoring Systems for Bridges (Luczynski et al., 2007)

(4)

Acknowledgments

This guideline has been drafted on the basis of Contract No. TIP3-CT-2003-001653 between the European Community represented by the Commission of the European Communities and the Skanska Teknic AB contractor acting as Coordinator of the Consortium. The authors acknowledge the Commission of the European Communi- ties, the Swiss State Secretariat for Education and Research, Empa, University of Oulu, University of Minho, Wroclaw University of Technology and City University London for its financial support.

(5)

1. Introduction

The overall objective of monitoring is to improve the knowledge of the real behaviour of structures. The information obtained by monitoring can be used for very different purposes:

– condition assessment of structures, – monitoring of structural performance, – remaining life-time prediction,

– model identification and correlation, – long term structural health monitoring.

All these purposes require problem specific approaches and techniques which differ substantially from each other and which impede the formulations of simple and ge- neric recommendations covering each aspect of monitoring. Nevertheless, any monitoring activity is based on a restricted number of key elements that can be isolated and for which useful, abstract recommendations can be formulated.

Monitoring of structures is an interdisciplinary activity that requires expert knowledge from different disciplines: structural and electronic engineering, modeling and simula- tion techniques, data processing, communication and archiving techniques, software design and implementation techniques. Since all these disciplines can hardly be covered by a single expert, monitoring is an activity that heavily based on the cooperation of a group of experts with very different engineering cultures and backgrounds. A common detailed analysis and planning, and communication between the experts are therefore a mandatory prerequisite for assuring a common understanding of the goals and a successful monitoring activity.

This guideline provides a data centric approach to monitoring. Not sensors, data acquisition and communication devices are considered as the key elements of monitoring, but specification in terms of physical concepts and data processing. This approach simplifies the access to monitoring for structural engineers and bridge owners since they do not have to take care of what type of sensor to install, but of which physical or chemical quantities to measure and how the measured quantities have to be processed, combined and presented to produce valuable information for them with respect to the pursued objectives.

This document provides guidelines for monitoring tasks on railway bridges. It provides

– a systematic methodology for specifying, designing, implementing and operating monitoring systems,

– a toolbox with recommendations for the application of methods, data processing algorithms and sensors,

– detailed guidelines how to apply several new technologies that were developed within the project.

(9)

2. Terms and definitions

Assessment is a set of activities performed in order to verify the safety and reliability of an existing structure for future use.

Assessment monitoring is any monitoring activity that is performed with the goal to collect data for bridge assessment.

Data acquisition is the act of mapping measurands to digital data that can be ma- nipulated by computers. The data acquisition process consists of sensing, signal conditioning and digitalization of measurands.

Data processing is the act of manipulating the digital data by a computer. The data manipulation may consist in reducing the size of the raw data in terms of bits (data compression), extracting condensed information from the raw data (data reduction) and transforming or combining the data into new data items.

Inspection is the on-site, mostly non-destructive examination to establish the pre- sent condition of a structure.

Inspection monitoring is any long term monitoring activity that is performed with the goal to increase the efficiency and effectiveness of the bridge inspection process (e.g. inspection scheduling).

Investigation is the collection and evaluation of information about a structure through inspection, monitoring, testing, modelling and document search activities.

Long term monitoring is any monitoring activity that under normal operation condi- tions requires maintenance actions of the physical monitoring system.

Measurands are the physical or chemical quantities that are sensed by the use of transducers during the monitoring process.

A minimal realization of a monitoring system consists of the data acquisition and processing module.

Monitoring is the act of acquiring, processing, communicating and archiving informa-

data acquisition data processing

sensing signal conditioning

data acquisition

data reduction data compression data transformation

data interpretation

data visualization

data archival

output of monitoring system raw data

minimal realization of monitoring system

Figure 2.1: Modules and data flow of a monitoring system.

(10)

tion about the actions on and the action effects of a structure over a period of time with a high level of automation. Monitoring is based on transducers for sensing physical or chemical quantities (measurands), programmable electronic equipment for acquiring, processing and communicating data and algorithms that define how data acquisition, processing and communication is performed.

A monitoring system may consist of data acquisition, processing, interpretation, visualisation and archiving modules.

The output of the monitoring system is the product of the data processing act and represents the input of the data interpretation, visualization and archiving modules.

Performance monitoring is any monitoring activity that is performed with the goal to establish the performance of a structure with respect to serviceability conditions.

Raw data are the final product of the data acquisition process and represents the input of the data processing stage.

Short term monitoring is any monitoring activity that under normal operation condi- tions does not require maintenance actions of the physical monitoring system.

Time series is the digital stream of data that is generated by the data acquisition process of measurands and . Time series has generally a constant time interval between any two subsequent samples. Time series that do not have a constant time step between any two subsequent samples are called non-equally spaced time series.

(11)

3. Methodology

3.1 Introduction

3.1.1 Monitoring plan

A detailed planning of the activities is essential to achieve a common understanding of the objectives of a monitoring task and a coordination of the activities. The planning has to be formalized within a monitoring plan. This plan should contain

– the definition of the monitoring objectives – the theoretical model of the bridge

– the specification of monitoring constraints – the design of the model monitoring system – the design of the physical monitoring system – the validation plan of the monitoring system – the maintenance plan of the monitoring system Figure 3.1 displays a diagram of the general approach.

3.1.2 Actors and their duties

The actors of a typical monitoring task are – the bridge owner,

– the structural engineer, – the monitoring expert.

The duty of the bridge owner is to formulate the monitoring objectives, specify the monitoring constraints and define the budget.

The duty of the structural engineer is to provide a theoretical model of the bridge and an interpretation scheme for the data generated by the monitoring system. He collaborates as leading actor with the monitoring expert in the design of the model monitoring system. The structural engineer is responsible for validating and updating the theoretical model of the bridge and the model monitoring system.

The monitoring expert collaborates with structural engineer in the design of the model monitoring system. He designs, deploys, operates, validates, updates and maintains the physical monitoring system on the bridge.

3.2 Definition of the objectives

Any monitoring activity should start with a definition of the objectives. This definition should be formulated in the technical language of civil engineering and in terms of physical or chemical quantities (e.g. displacement, chloride concentration etc.). Any reference to transducers or electronic devices should be avoided. The definition should be as precise as possible based on the information available at the time of its first formulation.

(12)

3.3 Theoretical model of the bridge

Any monitoring task should be based on a theoretical structural model. The theoretical model provides the conceptual basis for designing the model monitoring system and the scheme for interpreting the output of the monitoring activity. The complexity of the theoretical model should be adequate to the monitoring objectives. During the monitoring activity, the theoretical bridge model should be reviewed and revised, if the output of the physical monitoring system does not comply with the predictions of the theoretical model.

3.4 Specification of constraints

Any monitoring task is subjected to different kind of constraints. These may be determined by site specific factors that are required to e.g. guarantee the operability of the bridge and respect security rules of the operator. Furthermore, guidelines aimed

Definition of monitoring objectives

Design of model monitoring system

Design of physical monitoring system

Deployment of physical monitoring system

Validation of model and physical monitoring system

Maintanance of model and physical monitoring system

Theoretical bridge model Specification of constraints

Figure 3.1: General monitoring methodology

(13)

to preserve the integrity of a structure may restrict the way transducers are mounted on the bridge. An additional constraint is the budget. The constraints should be specified in cooperation with the bridge owner and all third parties that may be influenced by the monitoring task. All constraints that are relevant for the monitoring task should be included into the monitoring plan.

3.5 Design of a model monitoring system

The model monitoring system specifies completely every conceptual aspect of the monitoring system and focuses mainly on the data processing aspects of monitoring.

It does not specify the implementation of the monitoring system in terms of transducers and electronic equipment but in terms of physical quantities and data processing and management algorithms.

The model monitoring system should be designed to achieve the objectives of the monitoring task based on the theoretical model and the specification of the constraints. The model monitoring system should be used to define the model data output of the monitoring system. The model data output must provide all the information that is needed to achieve the objectives of the monitoring task. The quality of the model output should be described in terms of output accuracy, resolution, frequency, stability, robustness and reliability.

The model monitoring system should contain at least detailed information about – the physical quantities that should be acquired (measurands), its exact location

on the bridge and, if required, its direction in space with respect to a coordinate system

– the record duration and the sampling rate with which each physical quantity should be acquired

– the triggering of data acquisition process (pre-defined time schedule or event driven)

– the amplitude range, frequency range, accuracy and resolution with which each measurand should be acquired

– the reliability with which the physical quantities should be acquired – the algorithms with which the measurands should be processed

– the data interpretation, visualisation and archival of the monitoring output – the time duration of the monitoring activity

During the monitoring activity, the model monitoring system should be reviewed and revised after each updating of the theoretical model.

3.6 Design of physical monitoring system

The physical monitoring system is the physical realization of the model monitoring system. The physical monitoring system consists of the equipment deployed on the bridge for sensing the physical quantities, for processing the acquired data and for communicating the output data to the logging station, and the software that process the data and controls the overall monitoring process. The physical monitoring system should be specified in terms of

– transducers

(14)

– cabling and wireless communication technology

– data acquisition, processing, communication and archival devices – software for data acquisition, processing and communication.

The hardware (equipment) and software (programs) of the physical monitoring system have to provide at least the same functionality and output quality as the model monitoring system. In particular, the choice of transducers, electronic devices and software has to be dictated by the accuracy, stability, robustness and reliability requirements that are formulated in the model monitoring system.

The effects of environmental influences like temperature, humidity, mechanical impact and electro-magnetic interference on the output quality of the physical monitoring system have to be investigated. If necessary, protection measures against environmental influences have to be applied to assure the output quality.

3.7 Deployment of the physical monitoring system

The physical monitoring system should be implemented on site according to its specification. A physical monitoring system log file should be produced that contains all information that is necessary to reconstruct the implementation and to track back possible error sources: e.g. for each transducer the implementation log file should provide information about its exact location, direction, model, series number and mounting details.

3.8 Validation of model and physical monitoring system

The model and the physical monitoring system have to be validated after the deployment of the physical monitoring system. The validation procedure should be described in the monitoring plan.

The validation of physical monitoring system has to be performed with respect to the output requirements of the model monitoring system. If the validation process demonstrates that the physical monitoring system does not comply with the output requirements, the physical monitoring system has to be upgraded until the output requirements have been achieved.

The validation of the model monitoring system has to be performed with respect to the monitoring objectives. If the validation process demonstrates that the model monitoring system does not achieve the monitoring objectives, the model monitoring system has to be revised. The revision concerns primarily the design of the model monitoring system. However, since the design of the model monitoring system is based on a theoretical bridge model, the revision may also concern the theoretical bridge model. If the validation process demonstrates that the theoretical bridge model does not comply with the output of the physical monitoring system, the theoretical bridge model has to be updated until a suitable correlation has been established.

During operation, the output of the physical monitoring system has to be periodically assessed to guarantee the required output quality.

3.9 Maintenance of physical monitoring system

If the physical monitoring system has to be operated over a period of time that requires maintenance actions, a maintenance plan should be prepared. The mainte-

(15)

nance plan describes all actions that are required to assure the output quality of the physical monitoring system. The maintenance plan should be based on a reliability analysis of the physical monitoring system. All maintenance actions should be reported in the physical monitoring system log file. After each maintenance action, a verification of the output quality of the physical monitoring system should be performed and reported.

(16)

4. Monitoring objectives

Typical objectives of monitoring are

– gathering of real life data that is used as input in structural assessment – performance monitoring

– post-strengthening monitoring

4.1 Input data for structural assessment

In cases where purely model based design oriented structural assessment is unable to provide the wanted verifications, monitoring may be considered as an additional tool for assessing the real behaviour of a bridge. Typical objectives are

– monitoring of strain, displacement, velocity and accelerations for assessing the global behaviour and for structural model calibration

– monitoring of local strains for assessing the critical spots against fatigue – monitoring of loads for determining the probability density functions of loads

and the structural effects of loads

– monitoring of temperature and temperature effects for assessing real temperature profiles and their effects on a bridge.

Monitoring is particularly recommended if the assessment is performed with probabil- istic models. In this case, monitoring provides detailed and validated information about the probability density functions of loads, structural responses etc. This information reduces the effects of modelling, physical, prediction and statistical uncertainties in the assessment process.

4.2 Performance monitoring

The goal of performance monitoring is assess the current performance of a bridge with respect to serviceability conditions. Before starting performance monitoring, the normal performance of a structure should be characterized. Since the performance of a structure can only be defined with respect to an action or a group of simultaneous actions, the normal performance should be characterized by a set of performance testing events. A performance testing event consists of a well defined single action or a group of simultaneous actions that produce a reaction of the structure. The reaction is defined by measurands or data generated by data processing steps from the

measureands. The normal performance is then characterized by a bounded set of allowable reactions. Any reaction outside of this set designates an abnormal performance of the structure. The recurrent performance testing events that are used for characterizing the performance of the structure have to occur often with respect to the monitoring period.

Normal performance relies often on regulations in codes, guidelines, recommendations etc. In these cases, a detailed structural model may not be necessary since the type of physical quantities, its location and direction may be already defined by regulations.

(17)

4.3 Life time prediction monitoring

The goal of life time prediction monitoring is to collect real life data that contributes to estimate the residual life time of a bridge or structural component thereof subjected to a long term degradation process. The monitoring results reduce the uncertainties with respect to actions and structural responses and hence increase the reliability of the residual lifetime analysis. In general, the monitoring process is focused on meas- uring the relevant actions and strains (and/or stresses) of critical components of the structure over a period of time. Long term as well as periodic short term monitoring can be applied. The duration of short term monitoring is selected according to the data requirements of the residual lifetime analysis tools. Advanced residual lifetime analysis techniques that use monitored data have been developed for steel bridges subjected to fatigue. Detailed recommendations are given in the guideline SB-5.2.S1 Guidelines for Monitoring of Steel Railway Bridges (Sedlacek et al., 2007a).

4.4 Post-strengthening monitoring

The goal of post-strengthening monitoring is to verify the effectiveness of strengthening. Post-strengthening monitoring is performed in two steps. A first step is performed prior to the strengthening of the structure with the goal to establish a reference char- acterization of structural behaviour. The second step is performed after strengthening with the goal to verify the structural behaviour with respect to the reference charac- terization. Post-strengthening monitoring should be based on well defined action events. Since monitoring is unable to provide any direct information about the the strength of a structure, the quality of strengthening should be qualified by indirect and measurable parameters (e.g. increase of structural stiffness provides information about the quality of bonding between the strengthening components and the structure).

(18)

5. Preliminary investigations

5.1 Information gathering

An existing railway bridge is usually a complex, individual structure, which is situated in a specific site and which was subjected to very individual loading history and environmental conditions. Therefore, all existing information about the structure, which is relevant for the monitoring tasks, should be available before starting the design process of the model monitoring system. This information includes the

– geometry of the structure,

– physical and chemical characteristics of the materials, – physical and chemical characteristics of the soil, – tracks type and its position with respect to the bridge,

– thickness and mechanical characteristics of the ballast layer, – type and extension of structural damages,

– load and stress history,

– local environmental conditions,

– existing drawings, inspection and test reports,

– local infrastructure conditions (on site power supply, wireless mobile telephony and wired or wireless communication infrastructure)

Missing information, which is relevant for achieving the monitoring tasks, should be provided by additional investigations.

5.2 Structural model

Any monitoring task should be based on a structural model. Its scope is to provide – in the planning phase: a theoretical framework for selecting measurands, its

exact location and its direction in space with respect to a coordinate system.

– in the operating phase: a correct and consistent interpretation of the output of a monitoring system.

The complexity of the structural model should be appropriate with respect to the monitoring objectives. In a first stage, a simple model can be used. If a significant disagreement persists between the model and the field measurements, the complexity of the model should be incrementally increased until the measurements on the bridge correlates sufficiently well with the model.

The structural model should include the effects of environmental parameters, in particular the temperature, if these parameters have an important effect on the measurands.

System identification and model updating techniques should be used to estimate the model parameters and to correlate the model with the measurements. The sensitivity of the model parameters on the measurands should be analyzed. The effect of measurement noise, model uncertainties and systematic model errors on the model

(19)

parameters should be assessed. The major sources of model uncertainties and systematic model errors are:

– the lack of data or poor quality data,

– the interaction between bridge and soil (boundary conditions),

– the interaction between bridge and non-structural components (e.g. ballast, tracks),

– the effects of environmental parameters.

If significant disagreements persist between an updated model and the field measurements, the effects of the disagreements on the monitoring objectives should be estimated.

5.2.1 Preliminary inspections and tests

Since most railway bridges are prototypes, the model based prediction of the response of bridges to loads may be affected by great uncertainty. Therefore, the structural model should be calibrated and validated with preliminary field tests.

The scope of preliminary inspections and tests is

– to validate or calibrate the theoretical structural model,

– to provide specific information for the design of the model monitoring system, – to tests or validate parts of the physical monitoring system (e.g. sensors, com-

munication infrastructure)

Preliminary tests should be performed if essential information is missing or if, based on the available information, a sufficiently reliable prediction of bridge response is not feasible.

(20)

6. Design of model monitoring system

6.1 Measurands

Each measurand of a monitoring system needs to be specified in detail. The specification of a measurand should contain:

– the physical or chemical quantity that should be measured – the location of the measurand on the bridge

– the direction in space if the measurand is a vector quantity – the reference value of the measurand

– the expected amplitude range of the measurand

– the record duration and the sampling rate with which the measurand should be acquired

– the condition that triggers the data acquisition process

– accuracy and resolution with which a measurand should be acquired – temperature range

A measurand can be either a local or non-local physical or chemical quantity. A local measurand is characterized by sensing the physical or chemical quantity in a very limited space region, so that the averaging effect of the measurand is negligible. A non-local measurand is characterized by sensing the physical or chemical quantity in an extended space region, so that the averaging effect of the measurand can not be neglected.

The location of measurands should be specified in a drawing with respect to a global or local coordinate system. The position of a local measurand is completely defined by its location. A non-local measurand should be specified by its location, direction and spatial dimensions (length or area).

Measurands can be scalar or vector quantities. Scalar quantities are e.g. temperature, humidity, chloride concentration. Vector quantities are e.g. displacements, ve- locities, accelelerations, strains. If the measurand is a vector quantity, its direction in space with respect to a global or local coordinate system has to be defined.

Measurands can be either absolute or relative. An absolute measurand do not refer to any particular state of the structure (e.g temperature, accelerations). A relative measurand always refer the change of the measurand with respect to a particular state of the structure. Relative measurands have to be specified with respect to a reference value that is associated to the reference state. The effect of variable loads and temperature on the reference value has to be considered. Generally, a suitable reference state is a bridge subjected only to dead loads.

The amplitude range of the measurands defines the lower and upper limits of the measurand. The amplitude range should be estimated using existing information, a preliminary investigation or predictions computed with the structural model. In order to avoid a deployment of sensors with insufficient range, the amplitude range of the measurands should be defined with care by adequately considering the level of uncertainty.

(21)

The record duration is the minimum time period that values of the measurands should be acuired. The record duration should be specified in time units (e.g. seconds, minutes).

The sampling rate defines the number of readings per unit of time (e.g. seconds) with which the measurand will be acquired. When monitoring dynamic processes, the sampling rate should be specified to cover the relevant frequency range of the process. If the frequency range of the process is subjected to a great uncertainty or unknown, the sampling rate has to be determined with preliminary investigations or it should be specified with care taking into account the uncertainty.

The readings of the measurand should be triggered either according to a predefined time schedule or a threshold. If the readings are triggered according to a threshold, the measurand associated to the threshold and the threshold value has to be specified. When specifying the threshold value, the effects of false triggering due to sensor drift should be considered.

The accuracy should be characterized by quoting, for a given confidence level, the uncertainty, as a percentage of the reading (% rdg). The resolution, the smallest change of the measurand that should be detectable, should be characterised using the unit of the physical or chemical measurand. To avoid the deployment of expensive sensors, a balance between resolution and amplitude range should be achieved.

The temperature range in which the measurand should be acquired should be specified. The information about the temperature range is important for sensor choice.

6.2 Specific recommendations

6.2.1 Displacement measurands

For each displacement measurand a reference point has to be defined. The displacement is measured with respect to the reference point. The reference point may be fixed (non-movable) or movable. For a reference point that are supposed to be fixed, an assessment should be made regarding possible movement due to environmental effects, loadings etc. and its effect on the displacement measurand has to be quantized.

Direct displacement measurements of the bridge superstructure or part of it with respect to the soil are generally expensive. This aspect has to be considered in the design of the model monitoring system. Alternative measurands that provide the same final information, e.g. strain measurands, or indirect methods for acquiring displacement information should be investigated. Indirect methods for displacement monitoring can be achieved by inclination or curvature measurements (see Toolbox, section 7.1).

6.2.2 Corrosion monitoring

Panning and implementation of corrosion monitoring should be guided by the background document SB-5.2.S3 Corrosion Monitoring Systems for Reinforced Concrete Bridges (Sørensen and Frølund, 2007).

(22)

6.2.3 Steel bridges

Panning and implementation of corrosion monitoring should be guided by the con- cepta and recommendations of the background document SB-5.2.S1 Guidelines for Monitoring of Steel Railway Bridges (Sedlacek et al., 2007a). Additional recommendations regarding strain and stress measurements are found in SB-3.4 Condition As- sessment and Inspection of Steel Railway Bridges, Including Stress Measurements in Riveted, Bolted and Welded Structures (Sedlacek et al., 2007b).

6.2.4 Environmental effects

Environmental effects like temperature, humidity etc. can have a significant effect on a structure (e.g. temperature may produce large displacements, strains and stresses etc. and humidity may influence the electrochemical potential). Temperature and humidity should be always monitored. The number and distribution of temperature and humidity measurands should be chosen according to the importance of their effect on the other measurands taking into account the monitoring objectives.

6.3 Data processing

Data processing is an intrinsic and essential part of the monitoring process and should be planned with the same care as the set-up of the measurands.

The readings of a measurand are referred as raw data. This raw data can be processed in a chain of data processing steps into a new form of data. A data processing step is defined by an algorithm which transforms one or several data inputs into one or several data outputs. The input of a data processing step can be raw data, that is readings of measurands, as well as output of a previous data processing step.

Each data processing step should be defined by completely specifying the input data and the algorithm.

6.3.1 Common data processing steps Very common data processing steps are

– filtering of time series, – integration of time series, – spectral analysis of time series, – data reduction.

Filtering

The scope of filtering is the removal of unwanted frequency components from a time series. Filters should be specified by its pass and stop bands, and the stop band at- tenuation.

Each filtering process generates a particular phase shift (time shift) in the processed time series. In applications where a precise time synchronisation is important, the effect of time shift has to be evaluated. Time desynchronisation can be avoided by processing all data inputs with the same filter.

(23)

Time integration

Integration of time series is applied for computing velocity time series from accelera- tion time series and displacement time series from velocity time series. Integration of time series should only be applied if

1. the direct measurement of the velocity or displacement is extremely difficult or too expensive to implement.

2. only the dynamic process is of concern and the energy content of the dynamic process below the fundamental frequency of the bridge is small compared to the total energy content.

In case 1), the accelerations should be measured with high sensitive, low noise ca- pacitive or force balanced accelerometers.

In case 2), the accelerations should be measured with accelerometers, whose lower cut-off frequency is at least 0.5 Hz smaller than the fundamental frequency of the bridge.

Since integration enhances the very low frequency components of time series, due to a low signal to noise ratio of sensors, the integrated time series displays a significant drift. This drift can be removed either with a high pass filter or a baseline correction method.

Any data processing step that includes a time integration step has to be validated by independent measurements within a preliminary investigation. The goal of this investigation is to

– evaluate the range of validity of the integration step with respect to the monitored process (e.g. velocity range of trains),

– evaluate the accuracy of the integrated time series.

Spectral analysis

The scope of spectral analysis is to decompose a time series into frequency components. Spectral analysis should be performed using an averaged periodogram method to reduce the effect of noise. The spectral analysis should be specified by

– frequency interval fΔ or the time interval of the periodogram T , where _p 1/ _p

f T

Δ = ,

– the frequency band f_min...f_max 0... f_max, where f_max is limited by the sampling rate according to f_max≤ f_s,

– overlapping in percentage of the segments in the time series – the type of windowing applied to the segments

6.3.2 Data reduction

A data reduction strategy should be considered for every measurands that needs to be sampled with a high sampling rate.

There are three methods to reduce the amount of raw data:

– data compression,

(24)

– data transformation, – data evaluation.

By data compression, the original data is encoded into a new representation that uses fewer bits than the original not encoded data. This data reduction is done by using specific encoding algorithms, which are either lossless or lossy. If a lossy data reduction algorithm is used, care has to be taken, to not compromise a later data processing step by the data reduction process.

By data transformation, the original data is processed by an algorithm with the goal to transform it into a new kind of information that requires less data in terms of bits.

Each data transformation algorithm is intrinsically lossy and prevents the reconstruc- tion of the original data. Examples of simple data transformation algorithm are extracting maxima, minima, mean values, and root mean square of a time series.

By data evaluation, the original data is evaluated according to given criteria with the goal to decide if the data can be totally discarded or not. The criteria may be based on the raw data of a number of measurands (e.g. threshold value) or on the outputs of a data transformation step.

Any data reduction step can chained to build a series of data reduction steps.

6.3.3 Estimation of natural frequencies

If the damping ratio of the associated vibration mode of the bridge is smaller than 10%, the natural frequencies can be estimated by applying the response spectrum method (see section 7.2.2) or the transfer function method (see section 7.2.1).

If the damping ratio of the associated vibration mode of the bridge is greater than 10%, a system identification algorithm should be used for estimating natural frequencies.

The natural frequencies should be estimated using time series that are recorded when no train is on the bridge. If this condition can not be met, the effect of the mass of the trains on the natural frequencies has to be investigated and, if the effect is expected to be greater than 1%, a correction to natural frequencies has to be applied.

For bridges with an U-shaped cross section that is filled with ballast, the effect of penetrated water or freezing of the penetrated water on natural frequencies should be considered.

For bridges with a strong interaction with the soil, the effect of freezing of the soil water on natural frequencies should be considered.

6.3.4 Estimation of structural damping

Structural damping should be estimated according to the methods that are described and recommended in the guideline SB-5.2.S2 Guideline for Estimating Structural Damping of Railway Bridges (Feltrin and Gsell, 2007). A brief description of the methods is found in section 7.2.3.

6.3.5 Estimation of modal parameters

The modal parameters should be determined by using one of the methods briefly described in section 7.1.

(25)

If an exciter is applied, the best location of the exciter has to be determined in order to excite all vibration modes within the frequency band that are of interest. The location can be determined with a computational modal analysis by using a suitable structural model.

The effect of the exciter’s mass on the modal parameters should be investigated. If a significant effect is found, the modal parameters should be corrected.

6.4 Reliability of monitoring output

The reliability of each data stream of the monitoring output should be specified. The specification should be performed in terms of the admissible number of failures per reference number of scheduled measurement or events (e.g. train passage).

A failure occurs when the data stream produces no data or corrupted data.

The reliability should be specified taking into account the costs.

The reliability of the physical monitoring system should be investigated by using accepted methods of system reliability theory. The background document SB-5.2.S4 Estimating Reliability of Monitoring Systems for Bridges (Luczynski et al., 2007).

(26)

7. Monitoring Toolbox

The goal of the monitoring toolbox is to provide briefly basic information about methods, algorithms and sensors that may be applied when monitoring a bridge. The au- diences addressed by the toolbox are bridge owners and structural engineers involved in the design of the model monitoring system. The toolbox can be used to guide and support the specification of a feasible and cost efficient monitoring system.

The toolbox is subdivided into three sections: Testing methods, algorithms and sensors. In each section, the tools are described with a standardized data sheet. At the beginning of the section, each item of the template is briefly explained.

More detailed background information on instrumentation and techniques are available in SB-5.1 Monitoring Instrumentation and Techniques (Feltrin et al., 2007) and in the appendices A, B, C and D.

(27)

7.1 Testing methods

Legend:

Name of method

Field of applications Specifies a list of quantities that can be determined by the method.

Testing principle Specifies a brief description of the basic principle of the method.

Frequency range Specifies quantitatively the typical frequency range that is covered by the method.

Accuracy Specifies qualitatively the typical accuracy that can be achieved by this method.

High: The quantitative results are reliable and precise within a few %.

Medium: Between high and low.

Low: The results are either qualitative or quantitative with a significant uncertainty.

Equipment costs

Specifies qualitatively the typical equipment costs that are necessary for applying the method.

High: > 100 k€

Low: 20 k€<

Costs of test Specifies qualitatively the typical costs of a test. These costs do not include the equipment costs. Amortization of equipment, data analysis and reporting are included.

High: > 50 k€

Low: 10 k€<

Required skills Specifies qualitatively the typical skills and experience that is required to apply successfully the method.

High: Requires advanced theoretical skills, good practical skills and some year of experience.

Low: Requires little theoretical skills, good practical skills and little experience.

Supplier Specifies qualitatively the typical supplier of the method. Specialized laboratory stands for a public laboratory or private company with significant infrastructure and experienced personnel. Specialists stands for a small private company with at least one personnel owing specialist knowledge.

Required equipment Specifies the required equipment for applying the method.

Advantages Specifies the most relevant advantages of the method.

Disadvantages Specifies the most relevant disadvantages of the method.

Comments Specifies additional comments regarding the method.

References Specifies references for additional information

(28)

Ambient vibration testing

Field of applications – Determination of natural frequencies – Determination of structural damping – Determination of mode shapes – Model calibration

– Damage detection

Testing principle Ambient vibration testing is based on sensing accelerations on a structure, which are generated by ambient sources (e.g. wind, micro-

tremors, traffic). By capturing the accelerations in a dense grid, information about the mode shape can be extracted. The captured accelerations are post-processed with specific algorithms to extract natural frequencies, structural damping and mode shapes. The algorithms are based on the assumption that the ambient sources are generated by wide band stochastic process.

Frequency range 0.5…100 Hz (depends on the size and weight of structure)

Accuracy high medium low

Equipment costs high medium low

Costs of test high medium low

Required skills high medium low

Supplier specialized laboratory specialist

Required equipment – Very high sensitive accelerometers

– Data acquisition device with many input channels

– Software for data processing (commercial software is available) Advantages – Cheap method since there is no need for artificial excitation (shakers,

impact hammers)

– Suitable for long term monitoring – Suitable for large bridges.

Disadvantages – Needs large records to achieve the desired accuracy.

– Needs expensive, very high sensitive accelerometers because of the small accelerations involved.

– The lack of information about the excitation forces may lead to inter- pret the response of a small band process as a vibration mode.

Comments Train traffic on railway bridges are generally not suited as excitation source since the weight and the dynamic properties of the train affect the modal parameters.

References Wenzel, H. & Pichler, D. Ambient Vibration Monitoring (Wenzel and Pichler, 2005).

(29)

Free vibration testing

Field of applications – Determination of natural frequencies – Determination of structural damping – Determination of mode shapes – Determination of transfer functions – Model calibration

Testing principle Free vibration testing is based on sensing the accelerations of a freely vibration structure (no forces are acting on the structure). The free vibrations can be generated by a train or by releasing the structure from a deformed state. By capturing the accelerations in a dense grid, information about the mode shape can be extracted. The captured accelerations are post-processed with specific algorithms to extract natural frequencies, structural damping and mode shapes.

Frequency range 0.5…100 Hz (depends on the vibration modes excited by the train)

specialized laboratory specialist Required equipment – Medium sensitive accelerometers.

– Data acquisition device with many input channels.

– Software for data processing (commercial software is available) Advantages – Cheap method since there is no need for artificial excitation (shakers,

impact hammers)

– Suitable for long term monitoring – Suitable for large bridges.

Disadvantages – Needs repeated tests with a fixed set-up to achieve the desired accuracy.

Comments

References McConnell, K. G. Vibration Testing: Theory and Practice (McConnell, 1995)

(30)

Impact hammer testing

Testing principle The structure is subjected to an impact by using a suitable device (impact hammer). The impact force and the vibration response of the structure at different positions are captured with accelerometers. With a specific set-up, the impact hammer tests are repeated several times to reduce the effects of errors. The recorded data is post-processed to transfer functions and, in a further step, with specific algorithms to determine modal parameters (natural frequencies, modal damping and mode shapes).

Frequency range 0.5…50 Hz (depends on the impact hammer)

specialized laboratory specialist Required equipment – Impact hammer with force transducer

– High sensitive accelerometers.

– Software for data processing (commercial software is available) Advantages – The method allows to determine transfer functions.

– Information about exciting force and frequency range is available.

– Relatively simple and easy to handle excitation device.

Disadvantages – Not suitable for long term monitoring

– Needs repeated tests with a fixed set-up to achieve the desired accuracy.

– Not suitable for large bridges because of the limited energy of the impact.

– The use of an impact hammer may affect the operability of the bridge during the tests.

Comments

(31)

Linear exciter testing

Testing principle The structure is forced to vibrations by an exciter, which moves a mass in vertical or horizontal direction. The inertial forces of the mass produce reaction forces on the bridge. The exciter force and the vibration response of the structure at different positions are captured with accelerometers. The recorded data is post-processed to transfer functions and, in a further step, with specific algorithms to determine modal parameters (natural frequencies, modal damping and mode shapes).

Frequency range 2…30 Hz (strongly dependent on the exciter)

specialized laboratory specialist Required equipment – Exciter equipped with force transducer

– Adjustable frequency range.

– Different type of excitations (e.g. random, harmonic, sweep).

– Suitable for large bridges because of the relative high forces ex- certed by the shaker.

– The use of an exciter affects the operability of the bridge during the tests.

– Needs expensive shakers (e.g. servo-hydraulic shakers), large power supply and expensive control devices

Comments

(32)

Rotating unbalanced ex- citer testing

Testing principle The structure is forced to vibrations by an exciter with a rotational mass with eccentricity. The centripetal forces of the rotating mass produce a harmonic reaction forces on the bridge. The frequency of the harmonic force is given by the rotation speed. By operating the exciter at different rotation speeds, harmonic forces with different frequencies can be achieved. The exciter force and the vibration response of the structure at different positions are captured with accelerometers. The recorded data is post-processed to transfer functions and, in a further step, with specific algorithms to determine modal parameters (natural frequencies, modal damping and mode shapes).

Frequency range 3…30 Hz (strongly dependent on the exciter)

specialized laboratory specialist Required equipment – Exciter with force transducer

– Different type of excitations (random, harmonic, sweep)

– Relatively simple and moderately expensive shaker, usually with electro-motor.

– The amplitude of the harmonic force increases with the square of the rotational speed.

– Not suitable for bridges with low natural frequencies because of small forces at low frequencies.

– The use of an exciter affects the operability of the bridge during the tests.

Comments

References SB-5.6 Prototype of Exciter for Vibration Tests and Concept of Monitor- ing System (Bien et al., 2007).

McConnell, K. G. Vibration Testing: Theory and Practice (McConnell, 1995)

(33)

Displacement measure- ments via inclinometers

Field of applications Determination of displacements

Testing principle The method uses a network of high precision inclinometers that measure the inclination of the structure at the mounting position. These inclination measurements are collected by a central computer. The displacements are computed by using a linear combination of canonical displacement curves that are obtained by a structural model subjected to specific load configurations. The inclinations of the linear combination of canonical displacement curves are then fitted in the least square sense with the measured inclinations.

Frequency range Static measurements

specialized laboratory specialist Required equipment – High precision inclinometers.

Advantages – The method allows to determine deflections of a structure using the abutments as reference points.

– Attractive method if reference points below the bridge are not feasible (road, river, bridge height)

– Suitable for short and long term monitoring of static deflections.

Disadvantages – Very few experience in the practical implementation and therefore few knowledge about application range and accuracy

– Unknown if suitable for monitoring of dynamic processes

Comments For long term monitoring, equipment costs have to be added to the tests costs, which are per year.

References Burdet, O. & Zanella, J.-L. Automatic Monitoring of Bridges Using Elec- tronic Inclinometers (Burdet and Zanella, 2000).

Monitoring Guidelines for Railway Bridges