Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Final Research Report: Developing Reliability-Based Inspection Practices P a r t I I
131 S U M M A R Y Introduction The National Bridge Inspection Standards (NBIS) mandate the frequency and methods used for the safety inspection of highway bridges. The inspection frequency specified in the National Bridge Inventory (NBI) is calendar-based and generally requires routine inspec- tions to be conducted at a maximum interval of 24 months. The calendar-based inspection interval applied uniformly across the bridge inventory results in the same inspection interval for new bridges as for aging and deteriorated bridges. Such a uniform inspection practice does not recognize that a newly constructed bridge, with improved durability characteristics and a few years of exposure to the service environment, may be much less likely to develop serious damage over a given time period than an older bridge that has been exposed to the service environment for many years. As such, inspection needs may be less for the newer bridge, and greater for the aging structure, relative to the uniform interval currently applied. Bridges that are in benign, arid operating environments are currently inspected at the same interval as bridges in aggressive marine environments, in which significant damage from corrosion may develop much more rapidly, resulting in increased inspection needs. Further, bridges that are known to possess âgoodâ characteristics or details are treated the same as those with characteristics or details known to perform poorly. Current practices make it difficult to recognize if the same or improved safety and reliability could be achieved by varying inspection methods and frequencies to meet the needs of a specific bridge, based on its design, condition, and operational environment. A more rational approach to inspec- tion planning would better match inspection requirements to inspection needs through reliability-based analysis that considers the design, materials, condition, and operational environment of a bridge. As such, the goals of this project were to develop reliability-based inspection practices to meet the goals of: 1. Improving the safety and reliability of bridges and 2. Optimizing resources for bridge inspection. The objective of this project was to develop a suggested bridge inspection practice for consideration for adoption by AASHTO. The practices developed through the project are based on rational, reliability-based methods to ensure bridge safety, serviceability, and effective use of resources. Final Research Report: Developing Reliability-Based Inspection Practices
132 Findings Reliability theories and practices were applied through the research to develop a guideline for Risk-Based Inspection (RBI) that provides a new approach for bridge inspection. The methodology consists of bridge owners performing a reliability assessment of bridges within their inventories to identify those bridges that are most in need of inspection to ensure bridge safety, and those for which inspection needs are less. This assessment is conducted by an expert panel at the owner level known as a Reliability Assessment Panel (RAP). The RAP conducts a reliability-based engineering assessment of the likelihood of serious damage resulting from common deterioration mechanisms, over a specified time period, and the likely outcome or consequences if that damage were to occur. The reliability-based assessment can be described by a simple, three-step process: Step 1: What can go wrong, and how likely is it? Identify possible damage modes for the elements of a selected bridge type. Consider design, loading, and condition characteristics (attributes), and then categorize the likelihood of serious damage occurring into one of four Occurrence Factors (OFs) ranging from remote (very unlikely) to high (very likely). Step 2: What are the consequences? Assess the consequences in terms of safety and service- ability, assuming the given damage modes occur. Categorize the potential consequences into one of four Consequence Factors (CFs) ranging from low (minor effect on service- ability) through severe (i.e., bridge collapse, loss of life). Step 3: Determine the inspection interval and scope. Prioritize inspection needs and assign an inspection interval for the bridge, based on the results of Steps 1 and 2. This assessment is based on common and well-known design, loading, and condition attributes that affect the durability characteristics of bridges. The attributes are identified and prioritized through expert elicitation processes. A simple reliability matrix, shown in the figure to the left, is used to identify the appropriate inspection interval for the bridge, based on the reliability analysis. Damage modes that tend toward the upper right corner of the matrix, meaning they are likely to occur and have high consequences if they did occur, require shorter inspection intervals and possibly more intense or focused inspections. Dam- age modes that tend toward the lower left corner, meaning they are unlikely occur and/or consequences are low if they did occur, require less frequent inspection. Inspection intervals determined through the RBI process may be longer or shorter than those specified by traditional uniform, calendar-based approaches, depending on needs identified by the reliability-based engineering assessment. Inspections conducted under the RBI process are typically more intense and thorough than traditional inspection practices, and require condition assessment of bridge elements to meet the needs of the reliability- based assessment. Inspection needs are prioritized to improve the reliability of the inspec- tion process, and bridge-specific inspection procedures can be developed based on the reliability analysis. The methodology developed is intended for typical highway bridges of common design characteristics. The methodology developed through the research capitalizes on the extensive body of knowledge and experience with in-service bridge behavior, and the common deterioration mechanisms that cause bridges to deteriorate during their service lives. The process allows for the integration of emerging technologies such as improved data on long-term bridge performance and advanced modeling and analysis techniques, when available. The method- ology was developed with suitable flexibility to address owner-specific needs and conditions, while providing systematic processes and methods to support consistent application of the technology. Reliability matrix for determining maximum inspection intervals for bridges.
133 The methodology developed through the research was tested using two case studies in different states. During these case studies, the processes described in the Guideline for RBI analysis were implemented using state forces to develop RBI intervals for typical highway bridges with superstructures constructed of steel and prestressed members. The RBI inter- vals determined through the RBI were verified through analysis of historical records for a sample of bridges in each state. The reliability-based inspection practices developed through the research differ from traditional, calendar-based approaches. The new approach to bridge inspection pro- vides a methodology to improve the safety and reliability of bridges by focusing inspec- tion resources where most needed. This also leads to optimized allocation of resources, as inspection requirements are better matched to inspection needs through a reliability-based engineering assessment. Conclusions This research developed inspection practices to meet the goals of (1) improving the safety and reliability of bridges and (2) optimizing resources for bridge inspection. The goals of the research have been achieved through the development of a new guideline document entitled âProposed Guideline for Reliability-Based Bridge Inspection Practices,â Part I of this report, which has been developed based on the application of reliability theories. This document meets the project objective of developing a suggested practice for consideration for adoption by AASHTO, based on rational methods to ensure bridge safety, serviceability, and effective use of resources. A reliability-based approach was fully developed and documented through the Guideline. This new inspection paradigm could transform the calendar-based, uniform inspection strategies currently implemented for bridge inspection to a new, reliability-based approach that will better allocate inspection resources and improve the safety and reliability of bridges. The implementation of the Guideline developed through the research was tested by con- ducting case studies in two states. These studies demonstrated and verified the effectiveness of the procedures developed in the research for identifying appropriate inspection intervals for typical highway bridges. It was shown through these studies that the RBI practices identi- fied appropriate inspection intervals of up to 72 months. It was concluded from these studies that implementation of the RBI practices did not adversely affect the safety and serviceability of the bridges analyzed in the study, based on the analysis of historical inspection records. These studies also successfully demonstrated the implementation of the Guideline and the procedures therein using state DOT personnel. The results reported herein demonstrated and verified that inspection intervals of up to 72 months were suitable for certain bridges. Such extended inspection intervals would allow the reallocation of inspection resources toward bridges requiring more frequent and in-depth inspections, resulting in improved safety and reliability of bridges. As such, the project goals of developing a reliability-based bridge inspection practice that could improve the safety and reliability of bridges, and optimizes the use of resources, were achieved through the research. Suggestions The research reported herein has demonstrated the effectiveness of the RBI procedures for determining suitable inspection intervals for typical highway bridges, and as such, broader implementation of the technology is suggested.
134 The procedure, methods, and approach described herein can be applied for atypical bridges as well. For example, non-redundant bridge members can be assessed using this approach, as illustrated in previous research (60). The approach can also be applied to com- plex bridges, or to bridges with advanced deterioration. Analysis requirements may be more detailed and advanced; development of such analysis may be pursued to provide a uniform strategy for bridge inspection across the entire bridge inventory. Finally, the back-casting procedure utilized herein may be considered for implementation when RBI practices are to be used. Back-casting provides a means for verification of models developed by the RAP and quality assurance of the RBI process. As such, the back-casting procedure provides a critical tool for the implementation of RBI technology.
135 Background The periodic inspection of highway bridges in the United States plays a critical role in ensuring the safety, service- ability, and reliability of bridges. Inspection processes have developed over time to meet the requirements of the National Bridge Inspections Standards (NBIS) (1) and to meet the needs of individual bridge owners in terms of managing and maintaining bridge inventories. The inspec- tion frequency mandated by the NBIS requires the inspection interval (maximum time period between inspections) not to exceed 24 months. Based on certain criteria, that interval may be extended up to 48 months with approval from the Federal Highway Administration (FHWA) (2). Maximum inspection intervals of less than 24 months are utilized for certain bridges according to criteria developed by the bridge owner, typically based on age and known deficiencies. Most bridge owners utilize the maximum inspection interval of 24 months, as mandated by the NBIS, for the majority of the bridges in an inventory, and the reduced intervals for bridges with known deficiencies. The uniform inspection interval of 24 months was specified at the origination of the National Bridge Inspec- tion Program in 1971 based on experience, engineering judg- ment, and the best information available at the time. The uniform approach provides a single maximum inspec- tion interval for most bridges, regardless of the bridge age, design, or environment. To date, this mandated inspection interval has provided an adequate level of safety and reli- ability for the bridge inventory nationwide. However, such a uniform inspection practice does not recognize that a newly constructed bridge with improved durability charac- teristics and a few years of exposure to the service environ- ment may be much less likely to developed serious damage over a given time interval than an older bridge that has been exposed to the service environment for many years. As such, inspection needs may be less for the newer bridge, and greater for the aging bridge, relative to the uniform interval currently required. Bridges that are in benign, arid operating environments are inspected at the same interval as bridges in aggressive marine environments, where significant damage from corrosion may develop much more rapidly, requiring increased inspection to ensure that safety and serviceability is maintained. Fracture critical members designed under mod- ern criteria have vastly improved resistance to fatigue than older bridges, and as such, the likelihood of fatigue damage for modern bridges is much lower than for older bridges. Newer bridges in general are designed to higher standards with more durable materials such that their resistance to loading and environmental effects is much greater than older bridges. Current practices make it difficult to recognize if the same or improved safety and reliability could be achieved by varying inspection methods or frequencies to meet the needs of a specific bridge based on its design, condition, and opera- tional environment. Recognizing the variability in the design, condition, and operating environments of bridges would provide for inspec- tion requirements that better meet the needs of individual bridges and improves both bridge and inspection reliabil- ity. A more rational approach to inspection planning would determine the interval and scope of an inspection according to the condition of the bridge and the likelihood that damage would occur. This would allow for resources to be focused where most needed to ensure the safety and reliability. Such inspection planning tools are highly developed in other industries, using the principles of reliability and risk assess- ment to match inspection requirements to inspection needs. These methodologies evaluate the specific characteristics of components, such as resistance to damage modes, anticipated deterioration mechanisms, current condition, and loading, to evaluate the reliability of the component. Appropriate inspection requirements are determined based on this evalu- ation, such that the safety and operation of the component is maintained over its service life, and resources are allocated efficiently. C H A P T E R 1
136 As such, the goals of this project were to develop reliability- based inspection practices to meet the goals of: (1) Improving the safety and reliability of bridges and (2) Optimizing resources for bridge inspection. The objective of this project was to develop a proposed bridge inspection practice for consideration for adoption by AASHTO. The practices developed through the project are based on rational methods that ensure bridge safety, serviceability, and effective use of resources. This report includes an overview of the inspection planning process that is based on the reliability principles developed dur- ing this project, and is documented in Part I of this report: âProposed Guideline for Reliability-Based Bridge Inspec- tion Practices.â
137 Research Approach The Guideline were developed in consideration of modern industrial practices and is the result of an exhaustive review and analysis of current methodologies and practices for the inspection and management of structures and facilities, assessment of needs and capabilities, and the development of methodologies focused on the unique needs of highway bridges. The research was aimed at identifying the most effec- tive strategy regarding development of a reliability-based bridge inspection practice. Through this investigation, a sys- tematic process for determining the frequency and scope of highway bridge inspections has been developed based on reli- ability concepts. Theories and practices for applying âreliability conceptsâ are increasingly popular as a basis for design codes as a means of adopting a more scientific basis for estimating variations in loading and resistance (strength) of components. Applying reliability theories in this context typically includes probabi- listic analysis to deal with uncertainties in the design param- eters and loading. There have been attempts to apply these design reliability concepts to maintenance and inspection activities, and some of this prior work will be discussed in this report. Unfortunately, such probabilistic approaches are, in most cases, found to be exceptionally complex and often require assumptions regarding the future behavior and per- formance of bridges that are difficult to verify. Additionally, probabilistic methods are typically focused on predicting strength, and do not address the serviceability requirements that are important in terms of bridge inspection. As such, alternative methodologies were sought through the course of the research. In industrial applications, the more common terminol- ogy for inspection practices that use reliability theories for development of inspection and maintenance strategies is ârisk-based,â with reliability being one component of a risk analysis that also includes consideration of the consequences of some type of failure or loss of service. Sometimes, reli- ability and risk terms are used interchangeably. An extensive study of the current state-of-the-practice and state-of-the- art for reliability and RBI practice was conducted as part of this project to determine the most applicable methodologies for the inspection of highway bridges. The best practices and the successful implementations of these inspection practices were reviewed, analyzed, and considered by the Research Team. An expert panel meeting/workshop was held that included bridge inspection experts from state departments of transportation to provide bridge-owner perspective on the tools being developed through the research. Several different approaches for developing a reliability- based inspection practice for highway bridges were consid- ered, ranging from âpureâ probabilistic structural reliability theories to fully qualitative risk analysis. The system that was developed is intended to incorporate the best practices and concepts from both schools of thought. The resulting methodology provides a reliability-based inspection practice that is implementable within the existing bridge inspection programs in the United States. Important consideration in developing the methodology included: â¢ The approach should be practically implementable and realistic. â¢ The approach needs to be sufficiently flexible to meet the needs of states with different inspection programs and bridge management approaches. â¢ The approach must be effective in ensuring bridge safety. â¢ The approach should match inspection requirements with inspection needs. â¢ The approach should capitalize on the existing body of knowledge regarding in-service bridge behavior. Based on these considerations, a reliability-based meth- odology was developed for risk-based bridge inspection. In summary, the methodology developed has its foundation based on risk analysis that includes both the anticipated reli- ability of bridges (and their elements) and the consequences C H A P T E R 2
138 of damage to a bridge. The methodology is strongly grounded in existing industrial practice. The methodology described in this report has been devel- oped based on the well-established methods used in other industries for practical inspection planning. Such industrial standards, which are discussed in detail in the project interim report (3), provide a technical foundation for the methodol- ogy developed. The approach has been customized to provide a practical, implementable tool that can be expanded and developed over time. The research resulted in the develop- ment of the Guideline, which documents the tools, method- ologies, and requirements for RBI practices.
139 Findings and Applications 3.1 Introduction The Guideline developed under this project describes the methodology for RBI practices for typical highway bridges. The goal of the methodology is to improve the safety and reli- ability of bridges by focusing inspection efforts where most needed and optimizing the use of resources. The Guideline provides a framework and procedures for developing suitable inspection strategies, based on a rational, reliability-based engineering assessment of inspection needs. The methodology considers the structure type, age, condition, environment, load- ing, prior problems, and other characteristics that contribute to the reliability and durability of highway bridges. Generally, the methodology involves bridge owners per- forming a âreliability assessmentâ of bridges within their bridge inventory to identify those bridges that are most in need of inspection to ensure bridge safety, and those for which inspec- tion needs are less. The assessment is conducted by consider- ing the reliability and safety attributes of bridges and bridge elements. This reliability assessment is conducted by an expert panel assembled by a bridge owner (e.g., state) known as an RAP. This panel conducts an engineering assessment of the likelihood of serious damage resulting from common dete- rioration mechanisms, over a specified time period, applied to key elements of a bridge. This assessment is based on common and well-known design, loading, and condition attributes that affect the reliability characteristics of bridge elements. These attributes influence the likelihood that a particular element will fail over a given time period, i.e., its reliability. The attri- butes are identified and prioritized through an expert elicita- tion process. This process capitalizes on the experience and knowledge of bridge owners regarding the performance of the bridges within specific operational environments, given typical loading patterns, ambient environmental conditions, construction quality, etc. The reliability estimate is combined with an evaluation of the potential outcomes or consequences, in terms of safety and serviceability, of damage progressing to a defined failure state. These data are then used to determine and prioritize inspection needs for specific bridges, or families of bridges with very similar design and condition characteristics. This includes determining a suitable inspection interval and scope, or procedures, to be used in the inspection. Under this pro- cess, the inspection interval is not fixed, such as it is in a uni- form, calendar-based system, but rather is adjusted to meet the anticipated needs of the specific bridge or bridges in a family. Therefore, bridges with highly reliable characteristics, which are unlikely to have serious deterioration over a speci- fied time, typically have a longer inspection interval than a bridge with less reliable characteristics, or for which the con- sequences of a failure may be more severe. For example, a bridge in good condition with highly durable and redundant design characteristics may have a longer inspection interval than a bridge in poor condition, lacking modern durabil- ity characteristics, and/or having a non-redundant design. Through this process, inspection resources can be focused where most needed to ensure the safety and serviceability of bridges. Inspection needs are prioritized to improve the safety and reliability of the bridge inventory overall. The approach developed under the research is a risk-based approach that differs from purely reliability-based approaches in that the likelihood of failure is combined explicitly with the consequences of that failure. Risk can be defined generally as the product of the probability of an event and the associated consequences: Risk Probability Consequence= Ã Probability in this equation is the likelihood of an adverse event or failure occurring during a given time period. This is sometimes expressed quantitatively as a probability of failure (POF) estimate for a given time interval, or as a qualitative assessment of the likelihood of an adverse event based on expe- rience and engineering judgment. Generally, this probability is C H A P T E R 3
140 the complement of the reliability. Consequence is a measure of the impact of the event occurring, which may be measured in terms of economic, social, safety, or environmental impacts. The Guideline developed through this research was focused on the inspection of typical highway bridges of common design characteristics. Atypical structures, such as long-span truss bridges, cable-stayed bridges, suspension bridges, and other unique or unusual bridge designs may require certain considerations not presently captured in the Guideline. Scour and underwater inspections have existing methodologies for evaluation, and as such are not included in the Guideline. Bridges assessed using this methodology are assumed to have a current load rating that indicates that the structural capacity is sufficient to carry allowable loads. 3.2 Overview of Methodology The RBI process involves an owner (e.g., state) establish- ing an expert panel (RAP) to define and assess the durabil- ity and reliability characteristics of bridges within that state. The RAP uses engineering rationale, experience, and typical deterioration patterns to evaluate the reliability characteris- tics of bridges and the potential outcomes of damage. This is done through a relatively simple process that consists of three primary steps: Step 1: What can go wrong, and how likely is it? Identify pos- sible damage modes for the elements of a selected bridge type. Considering design, loading, and condition charac- teristics (attributes), categorize the likelihood of serious damage occurring into one of four Occurrence Factors (OFs) ranging from remote (very unlikely) to high (very likely). Step 2: What are the consequences? Assess the consequences in terms of safety and serviceability assuming the given dam- age modes occur. Categorize the potential consequences into one of four Consequence Factors (CFs) ranging from low (minor effect on serviceability) through severe (i.e., bridge collapse, loss of life). Step 3: Determine the inspection interval and scope. Use a simple reliability matrix to prioritize inspection needs and assign an inspection interval for the bridge based on the results of Steps 1 and 2. Damage modes that are likely to occur and have high consequences are prioritized over damage modes that are unlikely to occur or are of little consequence in terms of safety. An RBI procedure is devel- oped based on the assessment of typical damage modes for the bridges being assessed that specifies the maximum inspection interval. Inspections are conducted according to the RBI procedure developed through this process. The RBI procedure differs from current inspection practices generally, because the dam- age modes typical for the specific bridge are identified and prioritized. The inspection is required to be capable of assessing each of these damage modes sufficiently to support the assessment of future needs. As a result, the inspections may be more thorough than traditional practices, including hands-on access to key portions of a bridge such that damage is effectively identified to support the RBI assessment. The results of the inspection are assessed to determine if the existing RBI procedure needs to be modified or updated as a result of find- ings from the inspection. For example, as a bridge deteriorates over time and damage develops, as reported in the inspection results, inspection intervals may be reduced to address the inspection needs for the bridge as it ages. The overall process for assessment under the developed Guideline is shown schematically in Figure 1. The process begins with the selection of a bridge or family of similar bridges to be analyzed. For the selected bridge or bridges, the RAP identifies common damage modes for elements of the bridge given the design, materials, and operational environ- ment. Key attributes are identified and ranked to assess OFs that categorize the likelihood of serious damage developing over a specified time interval. CFs that categorize the poten- Figure 1. Schematic diagram of the RBI process.
141 tial outcomes or consequences of damage are also assessed. Based on the assessment of the OFs and CFs for the vari- ous elements of the bridge, an inspection procedure is estab- lished, including the interval and scope for the inspection. Criteria for reassessment of the inspection procedure are also developed based on the assessment. The criteria for reassess- ment are typically based on conditions that may change as a result of deterioration or damage, and that may affect the OFs for the bridge. The RBI practice is then implemented in the subsequent inspection of the bridge. Inspection results are assessed to determine if any established criteria have been violated, or if conditions have changed that may require a reassessment of the OFs. If such changes exist, a reassessment of the OFs is completed and the inspection practice modified accordingly. The method of determining the inspection interval, or time period between inspections, is shown schematically in Fig- ure 2. The interval is based on the RAP assessment of the OFs and the CFs, plotted on a simple two-dimensional reliability matrix as shown in the figure. The OFs and CFs are used to place typical damage modes in an appropriate location on the matrix. In this figure, the horizontal axis represents the CF as determined for a particular damage mode for a given bridge element. The vertical axis represents the outcome of the OF assessment for a given damage mode for the given element. Damage modes that tend toward the upper right corner of the matrix, meaning they are likely to occur and have high conse- quences if they did occur, require shorter inspection intervals and possibly more intense or focused inspections. Damage modes that tend toward the lower left corner, meaning they are unlikely to occur, and/or consequences are low if they did occur, require less frequent inspection. This is simply a ratio- nal approach to focusing inspection efforts; inspections are most beneficial when damage is likely to occur and important to the safety of the bridge; inspections are less beneficial for things that are very unlikely to occur, or are not important to the safety or serviceability of the bridge. Through this process, individual bridges, or families of bridges of similar design characteristics, can be assessed to evaluate inspection needs from a reliability-based engineering assessment of the likelihood of serious damage occurring, and the effect of that damage on the safety of the bridge. The meth- odology can be applied throughout a bridge inventory, or to portions of a bridge inventory. Suitable Quality Control (QC) and Quality Assurance (QA) procedures should be utilized to ensure consistency. The RBI approach considers the structure type, age, condi- tion, and operational environment in a systematic manner to provide a rational assessment process for inspection plan- ning. A documented rationale for the inspection strategy uti- lized for a given bridge is developed. The damage modes most important to ensuring the safety of the bridge are identified such that inspection efforts can be focused to improve the reliability of the inspection results. The sections that follow describe the key elements of the RBI practices for bridge inspection. Section 3.3 provides background data underlying the RBI process, including reli- ability concepts such as POF, the reliability theory applied with the RBI process, damage modes and deterioration mechanisms considered in the analysis, and typical lifetime behavior characteristics that support the RBI approach. This section also highlights the differences between the reliability theory applied for inspection planning, and those tradition- ally applied for structural design codes. Section 3.4 discusses key elements of the Guideline developed under the research and initial testing of some of the processes developed. Finally, Section 3.5 describes data needs and resources to support the RBI analysis. 3.3 Reliability A key element in the RBI process is to understand the meaning and role of reliability in the context of determining inspection needs and inspection planning. This section of the report provides supporting data and background informa- tion regarding important aspects of reliability and its under- lying theories, and how these support RBI. Reliability is defined as the ability of an item to operate safely under designated operating conditions for a designated period of time or number of cycles. The inspection practices docu- mented in the Guideline are based on the concepts and theo- ries of reliability. The reliability of a bridge element is defined in terms of its safe operation and adequate condition to sup- port the serviceability requirements for bridges. This defini- tion is broader and more applicable to determining bridge inspection needs than structural reliability estimates, which are typically defined as a function of the load-carrying capac- ity of the structure and notional POF estimates. The challenge Figure 2. Reliability matrix for determining maximum inspection intervals for bridges.
142 with applying theoretical structural reliability concepts, such as those used in modern design specifications, is that the envisioned damage mode (loss of load-carrying capacity) represents only a portion of the required information needed from a bridge inspection. From the perspective of practical bridge inspection, safe operation includes strength consid- erations, but also includes a variety of serviceability limit states that may be related in some way to strength consider- ations, but are not direct measures of strength. Serviceability considerations such as local damage that can affect traffic, deflections and cracking, and loss of durability characteris- tics need to be assessed through periodic inspections, even if the effect on structural capacity, and therefore structural reliability, is nominal. Additionally, existing required load ratings provide structural analysis in terms of load capac- ity for bridges (4). These ratings generally provide limited insight into the inspection needs for a bridge, although the engineering analysis considers certain inspection results, such as section loss, in the analysis. Several methods and processes have been suggested for the assessment of in-service bridge reliability and the estimation of inspection requirements based on structural reliability, and these were studied during the course of the research. Research based on structural reliability theory for the devel- opment of inspection strategies, repair optimization, and updating bridge reliability estimates based on visual inspec- tions has been performed (5â8). Significant work in the area of applying structural reliability theory to highway bridges was reviewed during the course of the research, and detailed review is included in the project interim report (3, 7, 9â15). The conclusion reached based on the review of this literature was that these approaches were not currently implementable for highway bridge inspection, due to several factors. First, structural reliability models and probabilistic analysis does not typically capture the serviceability limit states critical to identifying in-service bridge inspection needs. Second, struc- tural reliability models are highly theoretical in nature, and the complexity of analysis required for even a simple structure makes application to the diversified bridge inventory in the United States impractical. Finally, the results of the structural reliability assessments are often based on POF estimates that are notional and design-based, such that significant uncertainty would result from mapping these results to inspection needs for specific bridges. However, the underlying concepts of reliability could be applied for the purpose of bridge inspection if appropriate and implementable methodologies for estimating reliability of bridges or bridge elements were developed. These method- ologies need to consider the serviceability requirements for bridges and bridge inspection, and define reliability appropri- ately such that it can be assessed based on inspection results and anticipated future deterioration. This analysis could then be applied as one component of an inspection planning pro- cess that includes an assessment of the consequences associ- ated with failure due to specific damage modes (16â19). Based on the analysis of the research on reliability methods, the research team pursued a path to develop a semi-quantitative, reliability-based framework for inspection practices. The key elements of developing that methodology included identi- fying the reliability theories to be implemented to evaluate bridges, and an appropriate description of âfailureâ to assess when a bridge element is no longer performing adequately, and hence has reduced reliability. The following sections describe briefly the underlying reliability theory utilized in the RBI Guideline, and the definition of failure used. Damage modes and deterioration mechanisms that cause a bridge element to deteriorate into the defined failure state are discussed, and the overall concept of matching inspec- tion needs to bridges during different stages of typical in-service behavior are described. 3.3.1 Reliability Theory Reliability is defined as the ability of an item to operate safely under designated operating conditions for a designated period of time or number of cycles. For bridges and bridge ele- ments, reliability typically decreases as a function of time due to deterioration and damage accumulated during the service life of a bridge, for example, corrosion of steel elements in a bridge that develops over the service life of the bridge, result- ing in increasing damage over that service period. The like- lihood of failure typically increases with time such that the reliability of a bridge or bridge element can be expressed as: PrR t T t( )( ) = â¥ Where R(t) is the reliability, T is the time to failure for the item, and t is the designated period of time for the itemâs operation. In other words, the reliability is the probability (Pr) or likelihood that the failure time exceeds the operation time. Sometimes, the likelihood is expressed as a probability den- sity function (pdf) that expresses the time to failure of an item (T) as some generic distribution, such as normal, log normal, etc. (13, 15, 20). This distribution can be used to cal- culate a POF function, F(t), to express the probability that the item will fail sometime up to time t. This time-varying func- tion describes likelihood of failure up to some given time, or the unreliability of the item, and the reliability is then: 1R t F t( ) ( )= â
143 In other words, the reliability is the probability that the item will not fail during the time period of interest. The chal- lenge for RBI was to determine an appropriate and practical method of estimating the probability, or likelihood, of failure described by the function F(t). This requires a definition of what is meant by âfailureâ for a bridge element or structure. It also requires an appropriate time interval over which an effective and meaningful assessment can be accomplished given the diversity in materials, designs, and operational environments included across the bridge inventory. When a large population of test data for identical or near identical components exposed to the same operational envi- ronment are available, a probability function describing the failure characteristics of the component may be deter- mined and verified based on test results. This can provide a quantitative frequency-based estimate of the POF that indicates the number of events (failures) expected during a given time period. However, such test data are generally unavailable for bridges, because design, construction qual- ity, and operational environments vary widely, and failures are rare. A suitable probability distribution may be assumed when test data are not available, but verifying the accuracy of such a distribution can be difficult for complex systems like highway bridges, where design and construction methods are constantly evolving, operational environments vary, and performance characteristics are also evolving. As a result, past performance of similar elements of a bridge may not be indicative of future performance, and the applicability of an assumed function to a specific bridge is unverifiable, since the lifetime failure characteristics described by the assumed function describe events that have not yet occurred. If designs, construction practices, and materials were not evolving over time, this might be more practical, but this is not the case for highway bridges. Under conditions for which data to adequately character- ize anticipated future behavior is limited, or where failure is rare, engineering judgment and experience can be used to estimate the expected reliability of a specific bridge within a given operational environment (21â23). Under these cir- cumstances, the POF is determined based on qualitative or semi-quantitative analysis and the probability is based on degree of belief, rather than frequency. To make such deci- sions, individuals with expertise and experience with typical performance characteristics, under a specific set of opera- tional environments, is required. Utilizing expert judgment and expert elicitation is a common method of characterizing the reliability of components or systems for the purpose of assessing inspection needs (21â24). Such engineering judg- ment and knowledge provides data when quantitative data are missing, incomplete, or inadequate. In the RBI method- ology, expert elicitation is used as a process for estimating the anticipated likelihood of failure for bridge elements, and hence their reliability, over a given time period of 72 months. The following sections describe the definition of failure, damage modes and deterioration mechanisms, and typical lifetime performance characteristics that are underlying the RBI process analysis. 3.3.2 Failure A key step in assessing the reliability of a bridge element is understanding how and why elements âfail,â and the typical deterioration mechanisms that cause the elements to âfail.â The damage modes and deterioration mechanisms that typi- cally affect bridge elements are well known, in most cases. For example, corrosion is obviously a significant deteriora- tion mechanism in concrete and steel bridge elements that causes them to âfail.â The likelihood of the failure occurring in some future time interval depends on attributes of the ele- ment, such as its materials of construction, design, durability, and current condition, as well as what conditions are used to describe an element as âfailed.â For bridges, catastrophic collapse would be one obvious condition that could be used to define failure, but such failures are very rare. Important con- cerns for bridge inspections extend well beyond simply avoid- ing rare catastrophic failures. Ensuring the safety of the bridge, in terms of structural capacity, serviceability, and safety of the traveling public are important factors in determining the inspection needs of a bridge. Therefore, failure requires a suitable definition that cap- tures the need to ensure the structural safety of the bridge, the safety of travelers on or below the bridge, and the service- ability of the bridge. Failure, utilized in this context, is defined as when an element is no longer performing its intended func- tion to safely and reliably carry normal loads and maintain serviceability. For example, a bridge deck with severe spalling may represent a âfailedâ condition for the bridge deck even though the deck may have adequate load-carrying capac- ity, because the ability of the deck to reliably carry traffic is compromised. Therefore, for the case of reliability assessments for determining bridge inspection needs, it was necessary to adopt a commonly understood definition of failure that con- siders common deterioration patterns in bridges and that can effectively be assessed through the inspection process. Addi- tionally, failure must be defined in a commonly understood manner that can be readily assessed, is consistent with the his- torical experiences of bridge managers, and is sufficiently general to be easily applied across the broad spectrum of design characteristics and elements that exists across the bridge inventory. To meet this need, the NBIS condition
144 rating of 3, âserious condition,â was chosen as a general, durable, and readily understood definition of failure. Bridge elements that have deteriorated to this extent may no longer be perform- ing their intended function, and remedial actions are typically planned to address such conditions. It is not envisioned that any bridges or bridge elements assessed using a reliability- based approach are allowed to deteriorate to this condition. Rather, inspection intervals are adjusted to ensure that the likelihood of failure in the time intervals between inspections always remains low. The subjective condition rating of 3 is defined within the Recording and Coding Guide (25) as follows: NBIS Condition Rating 3: SERIOUS CONDITION: Loss of sec- tion, deterioration, spalling or scour have seriously affected pri- mary structure components. Local Failures are possible. Fatigue cracks in steel or shear cracks in concrete may be present. In terms of the AASHTO Bridge Element Inspection Guide, this condition generally aligns with elements in Condition State (CS) 4, âseriousâ (26). These condition descriptions are widely understood and there is significant past experience in the conditions warranting a rating of 3 throughout the bridge inventory for the myriad of different materials and design characteristics that exist. This condition description provides a practical frame of reference for assessing the likelihood of failure in some future time period. For example, one could readily assess if a bridge deck that currently has a condition rating of 7, and has durable attributes such as adequate concrete cover and epoxy-coated reinforcing steel, was very likely, or very unlikely to deteriorate to a condition rating of 3 in the next 72 months. If the deck is very unlikely to deteriorate to a failed state during that time interval, repeated inspections of the deck may yield little or no benefit. On the other hand, if the deck were in poor con- dition, with a condition rating of 4, it may be more likely to fail during this time period, and more frequent inspec- tions are necessary to monitor the deterioration and identify repair needs. 3.3.3 Damage Modes and Deterioration Mechanisms The failure state described above is typically reached as the result of the accumulation of one or more forms of damage. For example, a deck may reach the âfailedâ state because of widespread spalling; a steel beam may reach that state as the result of severe section loss. These typical forms of deterio- ration in bridges are observable in a visual assessment of the bridge, or sometimes with the assistance of a nondestruc- tive evaluation technology (NDE). The observable effects on which a condition assessment is normally based are forms of damage, or damage modes. Damage modes are typically assessable through the inspection process and their extent or degree recorded in the inspection results. Spalling, cracking, scaling, sagging, etc. are damage modes. Damage modes are normally the result or manifestation of a deterioration mechanism, such as corrosion or fatigue. Deterioration mechanisms describe the path to failure, and may occur at different rates depending on factors such as operational environment and loading patterns. For example, a concrete bridge deck may fail due to the damage mode of concrete spalling, and the deterioration mechanism is corro- sion. If the deck is located in an aggressive environment, the corrosion mechanism may be fast acting, if in a benign envi- ronment, the mechanism may be slow acting. Similarly, if the damage mode is cracking in a steel element, and the cracking results from the deterioration mechanism of fatigue, then the rate at which the damage mode will progress depends on the cyclic loading of the bridge. If the bridge has very low average daily truck traffic (ADTT), then the likelihood of the dam- age mode progressing is lower than if the ADTT were high. However, if the damage mode is cracking and the deterio- ration mechanism is constraint-induced fracture (CIF), the progression of the damage mode may only depend on the susceptibility of the weld detail to CIF. Within the RBI process, it is important to separate the damage mode from the deterioration mechanisms such that suitable attributes or characteristics can be appropriately identified. For example, if the damage mode is spalling in a bridge deck, the deterioration mechanism could be corrosion of embedded reinforcing steel, or could be debonding of an overlay. Obviously, the attributes affecting how likely it is that debonding will occur differ from those that affect how likely it is that corrosion damage may occur, even though the result- ing damage may have very similar effects on the serviceability of the deck. 3.3.4 Lifetime Performance Characteristics Part of the overall assessment of the reliability of a bridge element is an understanding of the typical lifetime behavior of engineering components. Generally, failure patterns can be described by a âbathtubâ curve such as that shown in Figure 3, which represents the failure rate, or POF, as a func- tion of the time. The âbathtub curveâ shows the initial failure of new components due to defect (infant mortality), the use- ful life period, and the wear-out period. For bridges, the infant mortality portion of the bathtub curve illustrates the effects of construction errors or flaws, which typically become evi- dent in the early life of a bridge. One of the purposes of QCs and inspections during the construction phase of a bridge is to reduce the infant mortality rate, that is, to ensure there are not defects in the structure from construction errors that
145 will lead to a shorter than expected service life. Following the period in which infant mortality may occur, elements typi- cally have long service lives and failures are rare. Toward the end of the service life, when elements are in advanced stages of deterioration, the likelihood of failure can increase sub- stantially. As a result, more frequent and thorough inspec- tions may be necessary to monitor deterioration and identify repair needs. The bathtub curve shows schematically the typical performance of engineered components: the shape and timeline of the curve for specific bridge elements obviously depends on the attributes of the element, including the design characteristics, typical construction quality, operational envi- ronment, management and maintenance practices, etc. Among the purposes of RBI or any other life-cycle management system that includes inspection is the reduction of the wear- out rate by finding and repairing or replacing components before they fail, reducing unnecessary or unjustified inspec- tion efforts, and optimizing the utilization of inspection resources. Inspection needs are typically lower during the useful life of elements, when failures are rare, and increase as the failure rate increases as the result of deterioration mecha- nisms that manifest in damage. Many different methods are available to model failure pro- cesses and determine failure rate characteristics such as those shown in Figure 3, from qualitative to quantitative including hybrid methods. Qualitative methods would include expert judgment; hybrid methods would include methods like Markov Chain models, which use expert opinions and empir- ical data to estimate transition probabilities (27). Quantita- tive methods can range from fully empirical (using statistical fits to test or field data) to fully physics-based (using physi- cal models of failure processes). Weibull and log-normal statistics have both been used to describe failure processes that are driven by forces such as fatigue, wear, and/or cor- rosion. Given a sufficiently large population of engineered structures and the same driving forces, their rate of failure (or equivalently the POF at any time) can often be described by Weibull or log-normal statistics. Thus, if items are cheap and easy to test, a statistical description of their failures can be created and used to predict the behavior of similar items in the future. However, for bridges, characteristics of the ele- ments and their environment vary widely and are difficult to capture within such models, particularly when considering the needs of a specific bridge. For example, Table 1 shows variables used for probabilistic modeling of bridge reliabil- ity from some common literature resources and indicates the level of data resources that need to be either determined empirically, or estimated using statistical tools and probabil- ity functions (9, 10, 14, 28). As this table indicates, the mag- nitude of data that needs to be either collected or assumed is significant. The assumptions required to effectively estimate such a large number of properties and characteristics require verification, and may vary widely across different Figure 3. Plot of the âbathtubâ probability curve. Concrete cover Corrosion rate Time to corrosion initiation Workmanship Crack width Prestress steel strength and modulus Concrete strength and modulus Critical crack width Prestress losses Reinforcing steel strength and modulus Crack depth Impact factor Shrinkage of concrete Cracking density Area of reinforcing steel in concrete Thickness Loading rate Flexural forces Dead load Surface chloride concentration Shear forces Truck live load Critical chloride concentration Load distribution factors Water-cement ratio Chloride diffusion Reinforcement spacing Table 1. Variables used for probabilistic estimates of time-varying reliability.
146 bridge design, operational environments, and construction practices. Verification of the assumption requires observa- tion of bridge performance over its service life, and therefore, by definition, cannot be determined in time to be usefully applied. It is also notable that among the many parameters assembled to estimate time-dependent reliability, bridge joint condition is not among them. However, practically, this fac- tor alone may outweigh all of the others in terms of assessing the expected deterioration patterns and rate for a bridge. Additionally, because design and construction processes are evolving, elements that have the same role in different bridges often do not share key design features or operational environments that could affect their long-term performance. This makes estimating the many factors shown in Table 1 even more challenging and impractical across an inventory that includes 600,000 bridges and a multitude of operational environments. Therefore, expert judgment is required to consider the role and significance of specific design and envi- ronmental features for specific bridges, and to estimate future performance effectively. 3.4 Key Elements of RBI This section of the report provides an overview of key ele- ments of the RBI process described and detailed in the âPro- posed Guideline for Reliability-Based Inspection Practicesâ developed through the research. This includes a description of the OF, the CF, inspection procedures for RBI, and the RAP. 3.4.1 The OF Within the RBI process, an estimate of the POF for a given bridge element is expressed as an OF. This factor is an esti- mate of the likelihood of severe damage occurring in a speci- fied time interval, considering the likely damage modes and deterioration mechanism acting on the element. Key attri- butes of the element that affect the likelihood are consid- ered and documented to support the estimate. This section describes the approach and methodology for estimating the probability, or likelihood, of failure for bridge elements for the purpose of inspection planning. There are a variety of methodologies for estimating the expected performance of components or elements. These range from fully quantitative methods to fully qualitative methods. For example, the American Petroleum Instituteâs Recommended Practice 581 has, for certain critical com- ponents, empirical equations that estimate the POF for the component given certain attributes of the component and its operational environment (29). These empirical equations include factors associated with the attributes of specific com- ponents and are used to calculate the expected POF over some defined time period. In other cases, physics-based models for damage such as fatigue cracking are combined with industrial modeling tools to estimate the POF for specific components or systems (17, 30â33). For cases in which historical data may be scarce, where systems are complex and/or evolving such that relevant historical data are unavailable, expert judgment and expert elicitations can be used (21). To develop an estimate of the POF over a certain period, several factors need to be considered, including what con- stitutes a practical definition of failure, as described above, over what time period the assessment can be made, and what resolution is required for the estimate. Often, estimates uti- lized in reliability analysis are simply order of magnitude estimates, or even ranges, over which the POF is expected to fall. For example, ASME guidelines suggest first-level qualita- tive analy sis can be achieved using a simple three-level scale shown in Table 2 (21). An estimate of the annual POF associ- ated with a qualitative ranking is also provided. In this context (for this industry) a high POF is intended to represent failure rates on the order of a 0.01 or 1 in 100, while moderate prob- ability (likelihood) is intended to cover 2 orders of magnitude from 0.01 to 0.0001, with low probably being less than 0.0001. In moving from totally qualitative to semi-quantitative analysis, the order of magnitude of the failure rate may be estimated, and these numerical values provide a mapping of qualitative to quantitative rankings. In practical applications, even if quanti- tative methods are used, the estimated POFs are typically con- sidered to be, at best, order of magnitude estimates, due to the inherent variation and uncertainty in engineered systems. For application for the RBI assessment for highway bridges, existing industrial approaches were considered as a basis for developing appropriate methodologies for estimating reliabil- ity for highway bridges elements. This required that an appro- priate time interval be determined over which an assessment for the POF could be made, based on available data and engi- neering factors. Appropriate categorizations or qualitative scales to effectively describe that reliability were developed for use as part of a reliability-based assessment. 22.214.171.124 Assessment Interval Given the typically long service life of a bridge and the slow rate of deterioration mechanism such as corrosion, annual POF estimates such as those described above may have little Possible Qualitative Rank Annual Failure Probability Low <0.0001 (1/10,000) Moderate 0.0001-0.01 (1/10,000 â1/100) High >0.01 (1/100) Table 2. ASME POF rankings using a three level scale.
147 meaning and vary widely according to the assumption made in a particular analysis. Additionally, âfailureâ is not typically well defined as is the case, for example, with a pipe or valve. If the pipe leaks, it is failed, if the valve fails to open when required, it is failed. But with elements in a bridge, the major- ity of deterioration mechanisms extend over long time peri- ods, fracture being an exception, and the âfailureâ state itself can be subjective (34). Elements may reach a state that meets the definition of âfailureâ and stay in that condition for some number of years. Therefore, it is more appropriate to describe how likely it is for deterioration or damage to occur to the extent that an element deteriorates into a âseriousâ or âfailedâ condition. For the RBI process for bridge inspection, an OF is used to represent a qualitative measure of this likeli- hood over a time interval of 72 months. This time period was determined based on engineering factors that included prior research, experience, expert judgment, and data from corrosion, damage and deterioration models (13, 35â43). For example, commonly available corrosion models indicate that significant periods of time transpire between construc- tion of a bridge and initiation of corrosion, particularly in environments that are not aggressive (i.e., little or no use of de-icing chemicals and no marine exposure). Once initiated, corrosion may take a significant period of time to manifest in damage, depending on factors such as bar spacing, cover, con- crete material properties, and environment. Estimates for dam- age progression typically range from 6 years on the low end, for uncoated rebar in typical concrete structures, to 20 years or more for epoxy-coated bars. For steel elements, although corrosion damage can be severe, the rate at which corrosion damage occurs is actually very slow, typically less than 0.004â 0.006 in./year, even in moderately aggressive environments (37, 44, 45). Therefore, the amount of section loss that could occur during a 72-month interval is nominally less than 1â16 of an inch, assuming two sides of a steel plate were corrod- ing equally, at a relatively high rate of 0.005 in./yr. Section loss on this order of magnitude would not be considered serious. Therefore, it is practical to assess the likelihood of damage pro- gression occurring over a time frame of 72 months, because the likelihood is low that these deterioration mechanism could result in a bridge element deteriorating from a âgoodâ condi- tion to a âseriousâ condition during such a time interval. It should be noted that this interval of 72 months is an assessment interval over which the reliability of an element is estimated for the purpose of assessing inspection needs. Inspection intervals may be significantly less than 72 months when existing damage is present, or the attributes of an element suggest the likelihood of damage developing is high. The time period of 72 months is also considered a time period for which an engineer could reasonably estimate future performance within four fairly broad categories, ranging from âremoteâ to âhigh,â based on key attributes that describe the design, loading, and condition of a bridge or bridge element. The interval provides a suitable balance between shorter intervals, when the POF could be unrealistically low do to the typically slow progression of damage in bridges, or longer intervals, where uncertainty would be increasingly high. For example, if an engineer was asked to predict if a deck currently rated in good condition was likely to progress to a serious state in 1 year, that estimate would be very low, since deterioration mechanisms are slow acting. However, if the time period of 10 years were used, the uncertainty could be very high. The time interval was selected in part to provide a suitable balance over which damage progression could be reasonably predicted based on engineering assessment and rationale. 126.96.36.199 OF Categorization The OF is a qualitative ranking or categorization of the likelihood that an element will fail during a specified time interval. A four category, qualitative scale was developed for estimating the OF for RBI practices. The scale ranges from remote, when the likelihood is extremely small such that it would be unreasonable to expect failure, to high, where the likelihood of the event is increased. The categories and asso- ciated verbal descriptions are shown in Table 3. The OF is determined by expert judgment considering key characteristics, or attributes, of bridge elements. âAttributesâ are characteristics of a bridge element that contribute to the elementâs reliability, durability, or performance. These attri- butes are typically well-known parameters affecting the perfor- mance of a bridge element during its service life. These includes relevant design, loading, and condition characteristics that are known or expected to affect the durability and reliability of the element. These attributes are identified and assessed through the expert elicitation process. Numerical ranges that could be used to describe the OF scale quantitatively are shown in Table 4. Such numerical values provide ranges or target values for the qualitative rankings that could be used to map quantitative data, if it is available, to the qualitative rating scales. Failure of a bridge element is a relatively rare event, and design and construction details vary widely. As a result, relevant and verifiable frequency- based probability data are scarce, as previously discussed. The numerical values shown in Table 4 are target values that Level Category Description 1 Remote Remote likelihood of occurrence, unreasonable to expect failure to occur 2 Low Low likelihood of occurrence 3 Moderate Moderate likelihood of occurrence 4 High High likelihood of occurrence Table 3. OF rating scale for RBI.
148 can be used to map such data or models to the qualitative scales used in the analysis. For example, data from PONTIS deterioration curves or from probabilistic analysis, or other deterioration models, could be incorporated directly into the assessment of the OF using these scales. These numerical categories can also provide a framework for future develop- ment of models or data derived from analysis of the deterio- ration patterns in a particular bridge inventory. The quantitative description can be also be used as a vehi- cle for expert elicitation by using common language equiva- lents for engineering estimates. For example, if you asked an expert to estimate the probability of serious corrosion dam- age (widespread spalling, for example) for a particular bridge deck given its current condition, a common engineering response might include a percentage estimate, for example, less than 0.1% chance or less than 1 in a thousand. This esti- mate can then be mapped to the qualitative scale as being âlow.â Such estimates are typically very conservative, particu- larly for lower, less likely events. 188.8.131.52 Method of Assessing OFs OFs are determined through expert elicitation by the RAP assembled by a bridge owner. The RAP provides experience and knowledge of the performance of materials, designs, and construction quality and methods within a specific opera- tional environment. This knowledge and experience is used to categorize the OF considering credible damage modes and deterioration mechanisms for bridge elements. The assessment is conducted by identifying critical design, loading, and condition characteristics, or attributes, that affect the reliability and durability of the elements. For exam- ple, consider the damage mode of spalling due to corrosion damage in a concrete bridge deck. A bridge deck may have âgoodâ attributes, such as being in very good condition, hav- ing adequate concrete cover, epoxy-coated steel reinforcing, and minimal application of de-icing chemicals. Given these attributes of the deck, it may be very unlikely that severe dam- age (i.e., failure) would occur in the next 72 months. This is based on the rationale that the deck is presently in good condition, and has attributes that are well known to provide resistance to corrosion damage. As such, an OF of âLowâ or âRemoteâ might be used to describe the likelihood of failure due to this damage mode. Alternatively, suppose the deck is in an environment where de-icing chemicals are frequently used, the reinforcement is uncoated, and the current rating for the deck is a 5, Fair Condition, indicating that there are signs of distress in the deck. Based on this rationale, the likeli- hood of serious damage developing would be much greater, resulting in an OF rating of âModerateâ or âHigh.â Past expe- rience with decks of a similar design, combined with engi- neering judgment, can be used to support the assessment of the specific OF for a given deck. These attributes can be generally grouped into three catego- ries: Design, Loading, and Condition attributes. Design attri- butes of a bridge element are those characteristics of the element that describe its design. Design attributes are frequently intrin- sic characteristics of the element that do not change over time, such as the amount of concrete cover or material of construc- tion. In some cases, preservation or regular maintenance activ- ities that contribute to the durability of the bridge element may be a design attribute, such as the use of penetrating seal- ers as a preservation strategy. Loading attributes are characteristics that describe the loads applied to the bridge element. This may include structural loading, traffic loading, or environmental loading. Environ- mental loading may be described in macro terms, such as the general environment in which the bridge is located, or on a local basis, such as the rate of de-icing chemical application on a bridge deck. Loading attributes describe key loading charac- teristics that contribute to the damage modes and deteriora- tion mechanisms under consideration. Condition attributes describe the relevant bridge element conditions that are indicative of its future reliability. These can include the current element or component-level rating, or a specific condition that will affect the durability of the ele- ment. For example, if the damage mode under consideration is concrete damage at the bearing, the condition of the bridge Level Qualitative Rating Description Likelihood Expressed as a Percentage 1 Remote Remote probability of occurrence, unreasonable to expect failure to occur â¤1/10,000 0.01% or less 2 Low Low likelihood of occurrence 1/1000-1/10,000 0.1% or less 3 Medium Moderate likelihood of occurrence 1/100- 1/1,000 1% or less 4 High High likelihood of occurrence >1/100 > 1% Table 4. OF categories and associated interval estimates of POF.
149 joint may be a key attribute in determining the likelihood that corrosion will occur in the bearing area. 184.108.40.206 Screening Attributes Attributes can also be identified as screening criteria that identify certain characteristics that have a predominate effect on the reliability of an element. Attributes used for screen- ing may be design, loading, or condition attributes. Screening attributes are used to quickly identify bridges that should not be included in a particular analysis, either because they already have significant damage or they have attributes that are outside the scope of the analysis being developed. Screening attributes are typically attributes that: â¢ Make the likelihood of serious damage occurring very high, â¢ Make the likelihood of serious damage occurring unusually uncertain, and â¢ Identify a bridge with different anticipated deterioration patterns than other bridges in a group. The RAP must identify the appropriate value/condition for the attribute to use as a screening tool. For example, if consid- ering the likelihood that the steel bridge will suffer corrosion damage that reduces its rating to a 3, and the current rating is 4, the RAP may consider that such condition indicates that there is an unusually high likelihood of further damage devel- oping over the next 72-month period, and as such, use the condition rating of 4 as a screen. In such a case, the analysis can move forward to an assessment of the consequences of the damage without further evaluation of the attributes that contribute to the likelihood of damage, based on the screen- ing item. Another example would be to screen steel beam elements in bridges that have open decking. Since the open decking allows drainage directly onto the steel beams, the deterioration of these bridges would not be similar to steel beams with typical concrete decks; these bridges are screened from the analysis of steel beam bridges, as they may require separate analysis. It may be appropriate to treat these bridges as a separate group, developing the analysis to consider key attributes of those bridges with open decking. 220.127.116.11 Ranking Attributes Key attributes for a bridge element are identified by the RAP and used to assess the appropriate OF for the given ele- ment and damage mode being considered. This assessment is supported through an empirical scoring procedure that provides a rational method of estimating the OF category. The attributes identified are ranked according to the impor- tance of each attribute in assessing the reliability of a certain bridge element. For example, for attributes that play a primary role in determining the likelihood of damage, a scale of 20 points could be used, 15 points for an attribute that has a moderate role, and 10 points for an attribute that plays a minor role. For the damage mode of corrosion in a steel beam, for example, a leaking joint which results in drain- age of de-icing chemicals directly onto the superstructure is highly important in assessing the likelihood of serious cor- rosion damage occurring. Therefore, this attribute may be assigned a 20 point scale by the RAP. The RAP may consider age of the structure to contribute to the likelihood of corro- sion damage, but to a much lesser extent relative to a leaking joint, and assign a 10 point scale. Once the overall importance of the attribute is identified, different conditions or situations may be described to distrib- ute points appropriately based on the engineering judgment of the RAP. Again using the joint as an illustration, if the joint is leaking or can reasonably be expected to be leaking during the time interval, it will have the highest effect and be scored the full 20 points. If the joint is debris-filled or exhibiting moderate leakage, a score of 15 points may be appropriate, if there is a joint, but it is not leaking, a score of 5 points may be assigned. If the subject bridge is jointless, a score of 0 points may be used. The exact scoring for a given attribute may vary according to the design characteristics or operational envi- ronment of a particular bridge inventory. The key attributes and ranking scores are then used to develop a simple scoring process that ranks the reliability characteristics of a particular element, for a given damage mode, as a rational means of assessing the appropriate OF. The scoring methodology is intentionally flexible to adjust to the needs and requirements of different bridge invento- ries, while still providing a systematic process to document rationale for the OF assessment. It is not a âone size fits allâ approach, but rather intended to conform to the vary- ing needs of different operational environments and bridge inventory characteristics. The commentary section of the Guideline, Appendix E, provides suggested scoring and ratio- nale for more than 50 common attributes that might be identified by a RAP assessment of concrete and steel bridges. Alternatively, the RAP may identify additional attributes that meet the needs of a particular inventory, and develop ratio- nale explaining the purpose and assessment process for the attribute. Suggested scoring weights for the attributes may also vary according to the needs and experiences within par- ticular operational environments. Calibration of the scoring process is obviously required to ensure the overall assessment of attributes is consistent with engineering judgment. Certain key attributes should be identified as part of criteria for reassessment of bridge inspection requirements, following subsequent inspections. These attributes are typically associ- ated with condition, which may change over the service life of the bridge as deterioration occurs. When changes in these
150 condition attributes can result in a change in the likelihood of a given damage mode resulting in failure (i.e., the OF), reas- sessment of the inspection requirements is necessary. Several illustrative examples of attribute scoring are also provided as guidance in making the assessments. This includes scoring sheets for tabulating scores for different elements of bridges, and using those scores to determine the OF. How- ever, once attributes and attribute rankings for bridge elements are determined by an RAP, the scoring may be more readily accomplished by integrating or developing software for scoring characteristics of bridge elements more efficiently. Some of the attributes identified by the RAP may already be stored in exist- ing databases and bridge management systems; others may need to be acquired from inspection reports, bridge plans, and other sources. 18.104.22.168 Use of Surrogate Data For many bridges, the use of âsurrogatesâ for the attri- butes identified in the reliability analysis may be considered to improve the efficiency of the analysis for larger families of bridges. As used herein, âsurrogateâ refers to specific data that can be used to either infer or determine another piece of information that is required for the reliability assessment. For example, assume a fracture critical bridge was designed and built in the year 2000, which is well after the implementation of the AASHTO/AWS Fracture Control Plan. This informa- tion can be used to determine that the steel must at least meet certain minimum toughness requirements, and the bridge meets modern fatigue design requirements. Note that this was determined only from the date of construction and with no detailed review of the design calculations or specifications. As stated, the use of surrogates is particularly attractive when identifying and assessing a family of bridges. Design and loading attributes identified by the RAP are typically static in nature, that is, they do not change over time. The condition attributes will typically change over time, as damage accumu- lates and deterioration mechanisms manifest. However, when elements are in generally good condition, specific condition attributes identified by the RAP may not require individual assessment for each bridge or family of bridges; the previ- ous inspection results can simply be used as a surrogate for the individual attributes. This will typically allow for larger groups of bridges of similar design to be grouped into a par- ticular inspection interval, based on the criteria developed by the RAP. For example, again considering steel bridges built to modern design standards, it is known that the design attri- butes that would increase the likelihood of fatigue cracking and fracture have been mitigated through improvements in the design, fabrication, and construction process. The con- dition attributes that are required to assess the reliability of the element would include the presence of fatigue cracks due to out-of-plane distortions, fatigue cracking due to primary stresses, and corrosion damage. However, if the component rating is 7, in good condition according the NBIS scale, or CS 1 in an element-level scheme, the existing ratings can be used as a surrogate for the condition attributes. Note: This assumes the inspection result is from an RBI procedure, i.e., the inspection was capable of identifying if fatigue cracks existed. This allows all bridges that are of this same rating (and simi- lar design and condition attributes) to be treated collectively in a process that does not require much detailed analysis of individual bridges. If the condition rating or condition state changes, then the bridges can be reevaluated according to the RAP criteria. If the condition does not change between peri- odic inspections, reassessment may not be necessary. It is important to note that this process is significantly dif- ferent than assigning an inspection interval based simply on the current condition of the bridge; for example, deciding to inspect all steel bridges with rating of 7 on a longer interval than all of those rated a 6. The RAP analysis forms a rationale that identifies not only the current condition attributes that affect the reliability of the element, but also the design and loading attributes of the bridge or bridge element that affect the poten- tial for damage to occur. In other words this RAP evaluation forms an engineering rationale for the decision-making process that considers not only the condition of the element, but also the damage modes and the potential for that damage to occur. For element-level inspection schemes, the attributes iden- tified by the RAP may map directly to an element and element condition state. For example, consider that the RAP identifies leaking joints as an attribute driving the likelihood of section loss in the bearing area of a steel beam. The element condi- tion state (joint leaking) is recorded in the inspection process and can be used as a criterion for that attribute score. In some cases, all of the attributes identified by the RAP as being criti- cal to the likelihood of failure of an element may be included in a comprehensive element-level inspection process; in other cases, they may not. For NBI-based inspection schemes, attributes identified by the RAP may map to sub-element data collected in addition to the required condition ratings for the primary components of the bridge. This data could be used if it is collected under a standardized scheme for rating and data collection for the sub-elements. For the primary components, the generalized nature of the component rating makes this more difficult for specific attributes. 3.4.2 CFs The second factor to be assessed under the RBI process is the CF, a categorization of the likely outcome presuming a given damage mode were to result in failure of the element being considered. The assessment of consequence is geared
151 toward assessing and differentiating elements in terms of the consequences, assuming that failure of the element occurs. It should be noted that failure of an element is not an antici- pated event when using an RBI approach, rather the process of assessing the consequences of a failure is merely a tool to rank the importance of a given element relative to other elements for the purpose of prioritizing inspection needs. The CF is used to categorize the consequences of the failure of an element into one of four categories, based on the antici- pated or the expected outcome. Failure scenarios are consid- ered based on the physical environment of the bridge, typical or expected traffic patterns and loading, the structural char- acteristics of the bridge, and the materials involved. These scenarios are assessed either qualitatively, through necessary analysis and testing, or based on past experience with simi- lar failure scenarios. The four-level scale used to assign the CF is shown in Table 5. The CF ranges from low, used to describe failure scenarios that are benign and very unlikely to have a significant effect on safety and serviceability, through catastrophic scenarios, where the threat to safety and life is significant. Thus, both short-term (generally safety related) and long-term (generally serviceability related) consequences can be considered. In assessing the consequences of a given damage mode for a given element, the RAP must establish which outcome characterized by the CFs in Table 5 is the most likely. In other words, which scenario does he or she have the most confi- dence will result if the damage were to occur. Using the illus- tration of brittle fracture in a girder, it is obvious that the most likely consequence scenario would (and should) be dif- ferent for a 150 foot span two-girder bridge than for a 50 foot span multi-girder bridge. For the short-span, multi-girder bridge, an engineer may state with confidence that the most likely consequence scenario is âHighâ and that âSevereâ con- sequences are very remote for a multi-girder bridge, based on his/her experience and the observed behavior of multi-girder bridges. For the two-girder bridge, the consequence scenario is likely to be âSevere.â As this example illustrates, the CF sim- ply ranks the importance of the damage mode as being higher for a two-girder bridge than for a multi-girder bridge. For many scenarios, qualitative assessments based on engineering judgment and documented experience are sufficient to assess the appropriate CF for a given scenario; for others, analysis may be necessary using suitable analytical models or other methods. A series of more detailed criteria for specific elements (i.e., decks, steel girders, P/S girders, etc.) are provided in the Guideline that can be utilized during the assessment to deter- mine the appropriate CF for a given element failure scenario. These criteria, combined with owner-specific requirements developed in the RAP or from other rational sources for assess- ing bridges and bridge redundancy, are then used to determine the appropriate CF for a given scenario. 3.4.3 Inspection Procedures in RBI Conducting a reliability-based assessment of the inspec- tion needs for bridges requires specific information regarding the current condition of bridge elements that allows for the assessment of expected future performance. For example, to determine the appropriate OF for corrosion damage in a steel bridge element, one would have to know if corrosion damage were currently present and to what degree or extent. Without this information, it would not be possible to assess the likeli- hood of severe damage developing over the next 72 months. Therefore, it is necessary under the RBI approach to perform inspections that are capable of detecting and evaluating rel- evant damage modes in a bridge. The relevant damage modes for specific bridge elements are identified through the RAP analysis of the OF, and this assessment provides foundation for the inspection scope and procedures to be used in the field for future inspections. The thoroughness of the inspection process is typically increased relative to, for example, compo- nent-level approaches that require only a single rating for a component (superstructure, substructure or deck). The methods or procedures used to conduct the inspection must be capable of reliably assessing the current condition of the bridge elements for the specific damage modes identified through the RBI process. In many cases, visual inspection or visual inspection supplemented with sounding may be ade- quate for conducting RBI. The inspections may be hands-on, such that damage is effectively identified to support the reli- ability assessment. For example, when assessing the likelihood of severe fatigue cracking in a bridge (the OF), it would be nec- essary to know if there were currently fatigue cracks. To make Level Category Consequence on Safety Consequence on Serviceability Summary Description 1 Low None Minor Minor effect on serviceability, no effect on safety 2 Moderate Minor Moderate Moderate effect on serviceability, minor effect on safety 3 High Moderate Major Major effect on serviceability, moderate effect on safety 4 Severe Major Major Structural collapse/loss of life Table 5. CFs for RBI.
152 that assessment, sufficient access to the superstructure of a bridge is required to determine if fatigue cracking is currently present, obviously, and the inspection procedure must include reporting the presence or absence of fatigue cracks. In some cases, NDE techniques may be required within the inspection procedure to allow for reliable detection of certain damage modes identified through the RBI analysis. For example, if the RAP identifies cracking in a bridge pin as a credible dam- age mode because a bridge has pin and hanger connections, a visual inspection is inadequate. Because the surface of the pin where cracking is likely to occur is not accessible, due to inter- ference from the hanger plates, beam web and reinforcements, ultrasonic testing (UT) or other suitable NDE technology is necessary to allow for the cracking to be assessed. The RAP analysis of the OFs and CFs provide a basis for the inspection requirements to be used in the field, by identifying credible damage modes and prioritizing these damage modes based on their potential effect on safety and serviceability. Based on the assessment of the OFs and the CFs, damage modes for a bridge can be prioritized based on the product of these factors: IPN OF CF= Ã Where IPN = Inspection Priority Number. For example, if the fatigue cracking has a moderate likelihood of occurring and the consequence is severe, then the IPN would be 3 Ã 4 = 12. If fatigue cracking were moderately likely, but the con- sequence were only moderate (minor service disruption), for example, if the bridge in question is a short-span, multi- girder bridge with known redundancy, the IPN for that dam- age mode would only be 3 Ã 2 = 6. This process highlights the damage modes that are most important, that is, most likely to occur, and have the greater associated consequences if they did occur. This information is included in the inspection pro- cedure for the bridge, providing guidance to the inspectors on emphasis areas for the inspection, based on the engineer- ing analysis and rationale developed by the RAP. It should be noted that the calculation of the IPN for each damage mode identified in the process does not limit the scope of the inspection to only those damage modes. However, it provides a simple method of prioritization of damage modes that are most important, based on a rational assessment that incorporates bridge type, age, design details, condition, etc., as well as the associated consequences. The resulting outcome from the RAP analysis provides inspec- tion requirements that are tailored to the specific needs of the bridge and include a prioritization of the damage modes for that bridge. This provides a more focused inspection prac- tice that is based on an engineering assessment of the specific bridge or bridge type in order to improve the effectiveness and reliability of the inspection. 22.214.171.124 Reliability of Inspection Methods For most RBI planning processes, such as those used for assessing cracking in nuclear power plants or oil and gas facili- ties, the reliability of different inspection strategies or methods is considered the assessment (21, 23, 29). For inspection tech- nologies, reliability is typically defined by a measure of the abil- ity of the technology to perform its intended function. Reliable and effective inspection methodologies reduce the uncertainty in the current condition of components, and therefore can affect future POF estimates and rationale for a given inspection interval. The reliability of specific inspection methods may be quantified using probability of detection (POD) or other reli- ability analysis for a limited number of especially high-risk components and damage scenarios. This may be justified based on the significant risk associated with these facilities, includ- ing both the high cost and high environmental consequences of certain failure modes. However, for more general assess- ments of risk, the effectiveness of inspections is qualitatively described to rank various inspection approaches on a relative scale using engineering judgment. For example, API has cre- ated a five-category rating system used for several components described in API 581 (29). Inspection methods are qualitatively categorized on a scale that ranges from A to E, with A being âhighly effectiveâ and E being âineffective.â A similar approach was taken to develop guidance on the reliability or effectiveness of inspection methods for typical damage modes anticipated for common bridge elements. Tables included in the Guideline indicate the reliability of NDE technology for various damage modes for specific bridge elements, such as steel beams, concrete decks, etc. The reliability of the inspection method is described on a four- level qualitative scale and represented symbolically. Methods that are generally unreliable for a given damage mode or mechanism are described as âLowâ and methods expected to provide high reliability and effectiveness are âHigh.â The assess- ments of the reliability of inspection methods were made using expert judgment, literature review, experience, and data from other industries, where available (46). Information on the rela- tive costs of different methods is also included as guidance. The Technical Readiness Level (TRL) of different methodologies is also provided and describes if the methodology is a com- monly available tool that is readily accessible, if the method is specialized such that specialized expertise is required for implementation, or if the method is experimental in nature. Presently, there is somewhat limited reliability data available for many bridge inspection techniques and NDE technologies applied for bridge inspection. In part this is because histori- cally there has been little motivation to conduct such testing, since the inspection intervals are uniform and generally do not require any formal demonstration of effectiveness of the inspection procedure. However, in an RBI approach, where
153 inspection intervals may be longer based on rational assess- ments of potential damage, inspection scopes may need to be appropriately adjusted. As a result, determination of the reli- ability of the inspection method becomes a factor in the over- all approach to the inspection process. Reliability data such as that provided in the Guideline is expected to be refined and developed over time, as the reliability-based approach is imple- mented for existing bridge inventories. The tables provided in the Guideline provide the framework for including such analy- sis in the RBI methodology. These tables provide user guidance for identifying appropriate inspection methods and/or NDE technologies to address specific anticipated damage modes. 126.96.36.199 Element-Level vs. Component-Level Inspections There exists under the current implementation of the NBIS a variety of approaches to collecting, documenting, and stor- ing data on bridge inventories within individual states. While many states are licensed to use the PONTIS bridge manage- ment system, which is an element-level process for storing inspection information and evaluating future programmatic needs, the degree to which states fully implement the element- level inspection process varies. Other states use the component- based system that is required under the NBIS; still others use a span-by-span approach. However, to implement the RBI pro- cess, more detailed information than that typically required for a component-based system is needed. A component-level approach, which is intended to provide a single average or over- all rating for the three major bridge components, does not pro- vide sufficient data for assessing the likelihood of future damage developing for most cases, and as such will not support an RBI analysis. Information on the specific damage modes present on the bridge, their location, and their extent are needed to assess inspection needs. As a result, inspection needs under an RBI process are more closely aligned with more detailed, element- level systems. The key characteristics that are needed to support the RBI assessment are as follows: â¢ Report the damage mode or modes affecting key elements of the bridge, â¢ Report the location and extent of the damage, and â¢ Report on key damage precursors as developed through the RAP assessment. Precursors identified through the RAP process may include evaluating specific elements of the bridge such as the joints or drainage systems. Specific conditions that are precursors necessary to assess the likelihood of damage in the future will also be needed, such as the presence of rust-stained efflores- cence or fatigue cracking. Many of these may be found in the current AASHTO Bridge Element Inspection Manual (26), in many cases as bridge management elements or defect flags. The bridge management elements and defect flags may need to be more fully developed under the RBI process as needs develop for specific inventories. 3.4.4 RAP The RAP is an expert panel assembled at the owner level to conduct analysis to support RBI by assessing the reliabil- ity characteristics of bridges within a particular operational environment and the potential consequences of damage. The performance characteristics of bridges and bridge elements vary widely across the bridge inventory due to a number of factors. Variations in the ambient environmental conditions obviously have a significant effect, since some states have sig- nificant snowfall, and, as a result, apply de-icing chemicals to bridges frequently, while other states are arid and warm, such that de-icing chemicals may be infrequently or never applied. Design and construction specifications vary between states. Typical details such as drainage features, and use of protective coating or other deterioration inhibitors, for example, seal- ers for concrete, vary between bridge owners as do traditional construction practices, construction details, and materials of construction. In terms of consequences, redundancy rules and traditional policies vary somewhat between bridge owners, with some bridge owners requiring four members to be con- sidered redundant, while others require only three, for exam- ple. Owners may also have policies specifying girder spacing or other configuration requirements. All of these factors contribute to the operational environment of a bridge that affects the likelihood and rate of deterioration of bridges and bridge elements, and, to a lesser extent, the assessment of the potential consequences of that damage. As a result, knowl- edge and expertise of the operational environment, historical performance characteristics, bridge management and main- tenance practices, and design requirements for bridges and bridge elements are essential for conducting reliability-based assessments. The role of such expert knowledge of a specific opera- tional environment is a typical component for reliability or risk-based assessments of inspection needs. It is necessary that individuals with historical knowledge of the operational environment and typical deterioration patterns within that environment participate in the process. This participation is needed to effectively assess reliability characteristics of bridge elements and to identify and prioritize key attributes and factors that support the rational characterization of the OFs and CFs. To utilize this expert knowledge, which is inher- ently local to a specific bridge inventory, a RAP is formed at the owner level to conduct the reliability-based assessment. The RAP panel typically will consist of four to six experts from the bridge-owning agency. This team should include
154 an inspection team leader or program manager that is famil- iar with the inspection procedures and practices as they are implemented for the inventory of bridges being analyzed. The team should include a structural engineer who is famil- iar with the common load paths and the overall structural behavior of bridges, and a materials engineer who is familiar with the behavior of materials in the particular environment of the state and has past experience with materials quality issues. Experts from outside the bridge-owning agency, such as academics or consultants, may be used to fill technical gaps, provide independent review, or simply supplement the RAP knowledge base as needed. A facilitator may also be used to assist in the RAP process. 188.8.131.52 RAP Expert Elicitation Expert elicitation is a method of gathering insight into the probability or likelihood of failure of a component, or of eval- uating associated consequences when insufficient operational data exists to make a quantitative, frequency-based estimate. When failures are rare, or it is necessary to predict future fail- ures, expert elicitation is used to provide quantitative or qual- itative estimates (categories) for use in assessing inspection needs or the likelihood of adverse future events. Processes for expert elicitations are common in nuclear applications and other safety-critical industries for performing risk assess- ments of operating events and assessing in-service inspection needs (21, 22, 47, 48). Key elements of the elicitation process include assembling appropriate subject matter experts and framing the problem to be assessed for the experts in order to elicit objective judgments. Consensus processes are used to aggregate expert judgments and ensure contributions from all of the experts involved (21). For RBI for bridges, expert elicitation is used to: â¢ Categorize the OF based on expert judgment: â Determine credible damage modes for bridge elements and â Identify and prioritize key attributes that contribute to the reliability and durability of bridge elements. â¢ Assess likely consequence scenarios and categorize the CF. The processes to elicit expert judgment from the RAP are simple and relatively straight-forward. The primary purpose of the processes is to provide a systematic framework that allows for efficient, objective analysis, and allows for input from all members the RAP. This allows for their expertise to be utilized and for dissenting judgments or views to be resolved such that issues are addressed as comprehensively as possible. For example, to identify the credible damage modes that are specific to the type of bridge and the element being considered, the problem is framed for the panel by describ- ing the element under consideration and its operational envi- ronment. The following question is then posed to the RAP: âThe inspection report indicates that the element is in serious condition. In your expert judgment, what is the most likely cause (i.e., damage mode) that has produced/resulted in this condition?â This elicits from the panel a listing of damage modes that are likely to occur for that element. Each expert is asked to independently list the damage modes he/she judges are most likely to have resulted in a fail- ure of the element. The expert records each damage mode and provides an estimate of the relative likelihood that each damage mode would have resulted in the element being in serious condition. The expert does this by assigning relative probabilities to each damage mode, typically with a mini- mum precision of 10% (the sum of the ratings should be 100%). The expert may note supporting rationale for the esti- mate. The individual results from each member of the RAP are then aggregated to evaluate consensus among the panel on the most likely damage modes for the element. An iterative process may be necessary to develop consensus on the cred- ible damage modes for a given bridge element and identify damage modes that are not credible. However, for many ele- ments, the damage modes are well known and consensus may be reached quickly. Attributes are then identified through a follow-up process. In most cases, the key attributes for a given damage mode can be identified by posing the following question to the RAP: â¢ Consider damage mode X for the subject bridge element. If you were asked to assess the likelihood of serious damage occurring in the next 72 months, what information would you need to know to make that judgment? This generates input from the RAP on what attributes of the element are critical for decision making regarding future expected behavior. The resulting input from the RAP can be categorized appropriately and ranked according to the rela- tive importance of the attribute for predicting future damage for the identified damage mode and element. While there are potentially many attributes that contribute to the durability and reliability of a bridge element, it is necessary to identify those attributes that have the greatest influence on the future performance of an element. Rationale for each attribute is documented, either by using rationale already provided in the Guideline, or developing suitable rationale through a variety of means including past performance, experience with the given bridge element, input from the RAP members, previous and contemporary research, analysis of historical performance, etc. Expert elicitation is also used for assessment of the CF by providing different potential failure and consequence sce-
155 narios and asking the RAP to assign relative likelihood to the outcome of the failure according to the CF scale. This is a use- ful tool for evaluating the appropriate CF for situations that are not well-matched to the examples and criteria provided in the Guideline, or to establish basic ground rules for the assessment of common situations. The process involves a few basic, but critical steps as follows: 1. Statement of the Problem: The RAP is presented with a clear statement of the problem and supporting infor- mation to allow for expert judgment to be made. Care should be taken to ensure the problem statement does not contain information that could lead to a biased decision. The problem statement typically includes data regarding the bridge design, location, typical traffic patterns, and the failure scenario under consideration. 2. Expert Elicitation: Independently, each member of the RAP is asked, based on judgment, experience, available data, and given the scenario presented, to determine the most likely consequence resulting from the damage mode under consideration. The expert is asked to express this as a per- centage of the likelihood, with the smallest unit of estimate typically being 10%. The experts may provide a statement on what factors they considered in making the estimate. 3. Comparison of results: Once each member of the RAP has rated the situation, the results of the elicitation are aggregated. Generally, there will be consensus regarding the most likely consequence. However, in some cases, the most likely choice may not be clear and there will not be consensus. 4. Identify CF: If there is consensus among the panel regard- ing the appropriate CF, then the rationale for making the determination is recorded. This rationale should be con- sistent with criteria provided in the Guideline and if not, the panel documents the deviation or changes and associ- ated rationale. For cases in which consensus is not reached in the initial elicitation, the experts should discuss their rankings and their assumptions and rationale for their specific judgments. The members of the RAP should then be given the opportunity to discuss the various judgments and to revise their scores based on the discussion. In some cases, additional information may be needed to support developing a consensus regarding the appropriate CF. If consensus cannot be reached, a potential approach would be to adopt the most conservative conse- quence scenario that was included among the revised scores. Exceptions to the selected likelihood scenario should also be documented. The RAP may determine that additional analysis is required to determine the appropriate consequence for a given dam- age scenario. In some cases, additional data collection may be required in order to reach a consensus. Individual RAPs have the flexibility to develop effective methodologies to address cases in which consensus cannot be reached. However, the method must result in the selection of the most appropriate CF, based on the Guideline provided and sound engineering judgment. 184.108.40.206 Example of Expert Elicitation This section provides an example expert elicitation as an illustration of the RAP process. As part of the research for NCHRP Project 12-82, an expert panel was assembled of state bridge engineers and inspection experts from seven differ- ent states and an engineer from the FHWA. The goal of the two-day meeting was to have experts from several state DOTs contribute to the development of reliability and RBI practices for highway bridges by providing owner perspective on the approach and tools being developed. The participants in the meeting represented a good cross section of personnel from state departments of transportation, ranging from personnel responsible for overseeing bridge inspection activities at the district level through the state-wide programs for inspection and maintenance. The meeting covered many of the topics necessary to oper- ate a RAP at the state level, including identifying key damage modes for certain bridge elements, identifying and weighting bridge element attributes that contribute to the durability/ reliability of the element, and evaluating the consequences of various damage modes. Among the activities at the meeting was a trial of the suggested expert elicitations processed uti- lized in the Guideline for conducting the reliability analysis needed as part of RBI practices. This section of the report provides example results from this workshop to illustrate the elicitation process and sample data provided by a cross sec- tion of practicing engineers. Although this panel included individuals from a variety of operational environments, and results of the elicitation process would likely have differ- ences within a specific environment, the results are included here to illustrate the process and provide typical results. The example presented here includes the results for a steel bridge superstructure. These same processes were used during RAP meetings held as part of two case studies of the technology, reported in Section 3.6. 220.127.116.11.1 Identifying Damage Modes. The process for determining credible damage modes based on an expert elici- tation was conducted during the workshop. The goal of the exercise was to identify the most likely and credible damage modes for the element and establish the consensus (or lack of consensus) of the panel regarding the most common dam- age modes for that element. The panel was asked to perform this assessment for a steel girder. The following question was
156 posed to the panel, âYou are told a steel girder is condition rating 3, serious condition, according to the current NBIS rating scale. Based on your experience, what damage is likely to be present?â The expert was provided a form similar to that shown in Table 6, except that the damage modes and likelihood indicators were blank. Each member of the panel completed the form, identifying the damage modes and rela- tive likelihood with a precision of 10%. Table 6 illustrates the results provided by one of the panel members. As shown in the table, this member rated corrosion/section loss as the most likely damage mode to be present, with fatigue cracking and impact damage as less likely, and overload as a possible damage mode. In this case, the panel member identified stress corrosion cracking as a possible damage mode, but one that was very unlikely such that no likelihood was assigned for that damage mode. Figure 4 shows the results from each of the panel members for this elicitation exercise. It was the consensus of the panel that the most common damage mode for a steel girder was corrosion damage/severe section loss. This damage mode was selected by everyone on the panel, typically with values of greater than 50% likelihood. The methodology is simple for many bridge elements for which damage modes are well known, and it establishes the consensus of the panel in regards to the most likely damage modes. It also helps to identify damage modes that may be less well known, but of concern for the particular state or bridge inventory. For example, one member of the expert panel had a different view of the most likely damage modes for a steel girder, marking impact damage (40%) as the most likely damage mode in his/her state. The particular state has large areas of arid environment, and hence a different perspective Damage Mode Likelihood (in 10%increments) Corrosion / Severe Section Loss Fatigue Cracking Impact Damage/ Fire Overload Stress Corrosion Cracking 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Table 6. Example of expert elicitation worksheet for steel girder damage modes. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Corrosion Fatigue Overload Impact/fire Li ke lih oo d o f O cc u rr en ce Damage Mode Participant 1 Participant 2 Participant 3 Participant 4 Participant 5 Participant 6 Participant 7 Participant 8 Figure 4. Results of expert elicitation on steel girder damage modes.
157 on the most likely damage modes. This illustrates how RAPs in different operational environments may identify and pri- oritize damage modes differently, depending on their opera- tional environment and experiences managing their bridge inventory. This is an advantage of the methodology, as dam- age modes that are most important to a given bridge inven- tory are identified through the process; these damage modes are not necessarily the same across the diversified operational environments of bridges across the country. It should be noted that this type of expert elicitation is a process for identifying and prioritizing likely damage modes for a given bridge or family of bridges based on expert judg- ment. It is not necessarily repeated over and over again for cases in which damage modes are well known. Rather, it is a tool for establishing that there is agreement on the most likely damage modes, capturing the expert judgment of the panel, and ensuring that the analysis is comprehensive and consid- ers all credible damage modes. Through this process, damage modes for which the likeli- hood is very small or essentially zero can be sorted out from more common damage modes through a rational process. In most cases, as was shown here, likely damage modes are expected to be well known by experienced bridge engineers and consensus can be readily achieved. 18.104.22.168.2 Attributes. Once the primary damage modes were identified, the panel considered the individual dam- age modes identified and the element attribute that con- tributed to the reliability of bridge element. For example, for the damage mode of corrosion/section loss, the expert elicitation consisted of posing the question to the panel, âFor the steel girder, you are asked to estimate how long it will be before significant corrosion/section loss would occur for that bridge. What information would you need to know to make that estimate?â A group discussion was held to identify and discuss the key attributes, and discuss their relative importance to determining the future deterioration pattern for the steel girder. The panel suggested that one of the most important attributes was the maintenance cycle for the bridge, or the maintenance activities that were typi- cally performed as part of normal operations. This includes such activities as bridge washing, cleaning away of debris that may accumulate, and maintenance of joints. The con- sensus of the panel was that this was a highly important attribute that should contribute to the rationale. The panel also identified that the bridge deck type was an important attribute that could potentially be a screening criteria for those bridges that have, for example, open-grated or timber decks. The panel identified that built-up members with the potential for crevice-type corrosion, micro-environments associated with traffic overspray, and condition history (trend data) were other attributes that could be considered in assessing the future performance of steel bridge elements in terms of corrosion. The attributes were ranked according to their importance as high (H), medium (M), or low (L), and if the attribute was potentially a screening criteria (S). Table 7 summarizes the results of the discussion. The attributes identified by a particular RAP in a specific operational environment may differ from those indicated in Table 7; however, these results are provided as an illustration of the process of eliciting expert judgments from a RAP. Once the attributes are identified and ranked appropriately, a sim- ple scoring regime can be developed based on the results and used to categorize the OF based on these attributes. 22.214.171.124.3 Consequence Scenarios. Expert elicitation to determine CFs was also demonstrated. An overview of the process for selecting the appropriate consequence category for a given damage mode was presented to the panel. This overview included several examples of different consequence scenarios that might be experienced during the evaluation process, and a review of the draft criteria for assessing the CF within an RBI process. An exercise was conducted to illustrate and test the use of expert elicitation for evaluating the likelihood of different Design Attributes Loading Attributes Condition Attributes Attribute Rank Attribute Rank Attribute Rank Deck Joints/Drainage H Macro Env. H,S Existing Condition H,S Built-Up Members M Micro Env. H Joint Condition H,S Deck Type M,S Maintenance Cycle H Material Type L,S Condition History Trend M Age L Debris Accum. M Table 7. Summary of attributes identified by the expert panel for steel superstructures.
158 consequence scenarios. The purpose of this exercise was to determine if, given a certain damage scenario, there could be consensus on the most likely outcome of that damage, based on the defined consequence scenarios and applied to a specific bridge. This process can be used by an RAP to develop and illustrate consensus and agreement with the Guideline for assigning consequence categories, to address situations that may not be sufficiently addressed or unclear, or to address unique situations for which expert judgment is required. The bridge presented to the panel was a multi- girder steel bridge with an Average Daily Traffic (ADT) of 2000 and spanning a divided state highway. Photographs of the bridge and descriptions of its structural configuration were provided to each member of the panel in written form for use in assessing different damage modes and associated consequence scenarios. Each panel member was provided with a handout that included basic directions, a bridge description, photographs of the bridge, and nine different damage scenarios to evaluate independently. The panel members were asked to complete the bubble chart for each damage scenario, as shown in Table 8. The results were collected and reduced to summary charts showing the average assigned likelihoods. There were nine damage scenarios presented to the panel, ranging from fracture of primary member to delamination and spalling of piers and abutments. The results of this exer- cise indicated that for certain scenarios, there was strong consensus on the most likely consequence of the indicated damage. For example, for the following scenario: âThe overlay is debonding; approximately 20% of the deck is spalled.â The assessment of the panel was distributed as shown in the Figure 5. As shown in the Figure, there was consensus from the panelâs independent assessments that this scenario represented a low to moderate consequence. Discussion of this scenario indicated that some panel members judged that the consequences could be high, based on their inter- pretation of the failure scenario presented. Discussion of the assessments quickly yielded assessment that the appropriate CF was moderate. A second scenario of interest was a comparison between fatigue cracking due to out-of-plane distortion vs. fatigue cracking due to primary stress. For the former, the panel rated the most likely consequence as moderate (~50% like- lihood), for the latter, the most likely consequence as high (>60% likelihood)âfor the multi-girder bridge utilized in the exercise. It was also interesting that a scenario of one beam fracturing was similar to a primary stress fatigue crack, >60% likelihood that the consequence would be âhighâ according to the consequence categories provided. Figure 6 shows the average outcome for the fracture of one of the steel beams, and as indicated in the figure, the elicitation indicated that the most likely outcome/consequence for this scenario would Consequence Category Likelihood (%) Low Moderate High Severe Table 8. Sample table for assessing likelihood for damage scenarios. Figure 5. Likely consequences of general deck spalling. Figure 6. Likely consequences of beam fracture.
159 be âhigh.â These data, which illustrate the consensus formed from the independent assessments of individuals from a num- ber of different states, would be refined when applied within a specific bridge inventory and operational environment. A series of criteria and requirements are provided in the Guide- line to assist in this process, and in many cases the CFs may be governed simply by the criteria in the Guideline or owner poli- cies regarding the treatment of redundancy or other factors. In other cases, additional analysis or testing needed may be iden- tified through the process. For cases not easily addressed or well defined, this type of expert elicitation is especially useful as a tool for developing rationale to support the categorization of the CF or identified specific analysis needs. These examples illustrate the process and feasibility of expert elicitation for determining the key factors required in the RAP assessment. The decisions regarding the likely damage modes and potential consequences are very similar to decision processes currently utilized by bridge engineers to determine the urgency of repair needs, anticipate future repair needs, and manage bridge inventories to ensure safety and serviceability of bridges. These decision processes are simply collected and aggregated systematically to provide rationale for decision making regarding bridge inspection requirements. Additional testing of the processes and evalua- tion of the consistency of the elicitation outcomes were con- ducted through case studies reported in Section 3.6. 3.5 Data to Support RBI Analysis There are a number of resources available or that could be developed to support the RAP assessment of the OF for bridge elements by providing data to support decision mak- ing. While none of these sources for data provide perfect solu- tions, for example, for calculating quantitatively the OF, they can provide data that supports decision making and ratio- nale developed through the RBI process. This section of the report describes a few of these resources, as well as important consideration for utilizing these data for the reliability assess- ment of bridges. First, use and application of the qualitative and quantitative data is described. 3.5.1 Quantitative vs. Qualitative Analysis Industrial standards for reliability and risk assessment recognize both quantitative and qualitative methods for estimating the POF and consequences of failure. Qualitative data typically are composed of information developed from past experience, expertise, and engineering judgment. Inputs are often expressed in data ranges instead of discrete values and/or given in qualitative terms such as high, medium, and low (although numerical values may be associated with these levels) (21, 23, 29, 49). Quantitative data are data developed through specific probabilistic models, databases of failure rates, or past performance data such as deterioration rate models. These data are typically more in-depth and detailed than qualitative data. This can provide valuable insight and uniformity in approach, but developing such data can be impractical for realistic situations that are too complex to be modeled effectively. Data on past performance are frequently incomplete or inaccurate, and in some cases can provide inef- fective estimates of future performance (50). Additionally, the effort required to collect and analyze the data may far outweigh the value of the data in estimating future perfor- mance, particularly when the data are sparse, include a large uncertainty, or design characteristics are evolving. Qualitative data enables the completion of assessments in the absence of detailed quantitative data. This qualitative data can be augmented with quantitative data when and where available, forming a continuum of data as shown in Figure 7 (23). The accuracy of results from a qualitative assessment depends on background and expertise of the analyst (21). Quantitative data, such as deterioration rate information measured from NBI or bridge management software (BMS) data, can provide supporting rationale for decision making, if handled appro- priately. Estimates of precise numerical values (quantitative analysis) can imply a higher level of accuracy when compared to qualitative analysis, though this is not necessarily the case, particularly when there is a high degree of uncertainty or variation. It is the quality of the data that is most important to support an analysis, and the fact that data are quantitative does not necessarily mean they are more accurate. Difficulty in effectively representing past experience, expert knowledge, and bridge-specific conditions can result in quantitative data that are biased and inaccurate, or whose applicability to a specific situation is unknown due to a complex array of assumptions utilized in developing the data. As a result, care- ful elicitation of expert knowledge from those most famil- iar with the operating environments, historical performance characteristics within those environments, and the expected future performance is used for RBI (21). Formal methods for eliciting expert opinion for the purpose of risk assessment are included in the Guideline and in the literature, as previously discussed (47, 51, 52). High Low Quantitative AnalysisSemi-qualitative analysis Detail of Analysis Qualitative Analysis Figure 7. Continuum of data needed for qualitative to quantitative analysis (23).
160 3.5.2 Data Needed for Assessment To perform a reliability-based assessment, the primary data required include data on bridge design characteristics and details, materials, environment, and current condition. Bridge inventory data describing the overall characteristics of a bridge, such as can be developed from existing NBI data tables, can provide some information. Data on materials and design characteristics are generally available in the bridge files, typical design and detailing practices, and local knowledge of construction practices. Damage data describe the deteriora- tion active or expected on a structure and estimate its effects on the structure and rate of development. For damage data, sources include general data available including the NBI database, inspection reports and supporting data within a DOT, element-level data for many states, and industrial data such as the experience of bridge owners, previous research, historical data, and historical experience. In some case, dete- rioration rate data or trends may be available and used as part of the assessment of future performance. Additional data on the anticipated performance of bridge elements is developed through the RAP process based on expert judgment. 126.96.36.199 Deterioration Rate Data and Previous Failure Histories Deterioration rate data such as that developed by Agrawal (35) and others (40, 53, 54) can be used to support estimates of future performance of bridge elements. However, there are challenges to applying these data exclusively to determine appropriate inspection strategies for bridges. First, data on bridge deterioration is often not specific, expressed normally in subjective condition ratings that may not capture specific characteristics of the bridge or the deterioration mechanisms that led to a certain condition rating. As a result, making accurate predictions regarding future performance can be challenging. Second, variation in the data is high, such that estimating deterioration curves typically requires advanced probabilistic analysis that develops mean estimates for the population. These mean or average values provide information on expected average performance of an overall population, but not for a specific item within that population. Deterioration rate data may need to be modified to adjust the data to local operating and management conditions to be used effectively to estimate the future performance of specific bridges or bridge elements within a population. However, deterioration curves and probabilistic failure estimates are valuable to the RBI analysis process in several ways. Deterioration curves can provide background and sup- port rationale for engineering judgment regarding future performance of bridge elements, based on past performance when combined with an assessment of the key attributes for the elements identified through the RAP process. If a bridge owner had a population of bridge elements that were very similar in design, and constructed at the same time and to the same specifications and quality, and exposed to the same environment, then accurate probabilistic estimates of future performance could be developed. Generally, this would be atypical of the bridge inventory. Consequently, the method developed for RBI practices provides a means for incorporat- ing such analysis, but does not rely on these data alone. Considerations for utilization of deterioration rate data include: â¢ Similarity of operational environment: The RAP should consider if the particular bridge under consideration shares the same operational environment as the elements from which data was obtained. Key elements of the operational environment include the ADT, ADTT, macro-environment of the bridges (severe environment vs. benign environment), micro-environment (salt application, joint and drainage conditions, exposure to overspray), and typical maintenance and management (among others). â¢ Similarity of Key Attributes: Key attributes that affect the damage modes and mechanisms for the bridge element should be similar for the bridge under consideration to those from which deterioration rate data was obtained. This may include materials of construction, design attributes, and condition attributes. Quality of construction and years in service may also be a factor. Component ratings for superstructure, substructure, and deck (and culverts) are provided for all bridges under the NBIS scheme providing general information on the deteriora- tion of the structural components over time, based on visual observations. Element-level data are documented for states using PONTIS or other element-level inspection schemes. Obviously, these condition data are an important component to evaluating the current condition of a bridge, at least in a general way, and identifying bridges with low or high condition ratings. These data can also be used to construct the deteriora- tion curve data to support assessments, or to make estimates of typical performance characteristics for bridges of a particular design, as described below. 188.8.131.52 Inventory Data Analysis Data from the NBI database can be analyzed to support the rationale for bridge inspection intervals developed through the RBI process. For example, historical NBI data can be ana- lyzed to determine the average period of time a particular bridge element remains in a certain condition rating. These data can be utilized to support rational decision making and the use of surrogate data, such as utilizing condition ratings
161 of 7, âGood Condition,â as a surrogate for condition attributes associated with a certain bridge type. For example, Figure 8 shows the time-in-condition for prestressed bridge super- structures in the state of Oregon. These data were developed by examining 20 years of NBI data, and determining from these data the time period (no. of years) individual bridges remained in a certain condition rating, according to the inspection results documented in the NBI. Weibull distri- butions were used to characterize the distribution of years in rating for this population of bridges, and these Weibull distributions are shown in the figure. Simply summing the mean (average) number of years, historically, that a pre- stressed bridge has remained in each certain condition rating, assuming a bridge component is currently rated a 7 and changes to a 6 immediately, the average number of years to progress to a condition rating of 3 is ~15 years. Given a maximum inspec- tion interval of 72 months (6 years), at least two inspection cycles would be completed within this 15 year period. During these inspections, if deterioration occurs more rapidly than initially envisioned, the inspection interval is appropriately reduced. These data support the rationale that significant margin exist when considering a bridge currently in a condition rating of 7. When considered within an RBI process, which identifies attributes of bridges that are likely to cause more rapid deterioration, such rationale is well-founded and based on quantitative, historical data. More complex analysis of such data may also be used, including deterioration curves, probability calculations, etc. BMS, such as the PONTIS program, may provide data on tran- sition probabilities or lifetime estimates based on Weibull statistics, which can be utilized to provide quantitative data to support the RAP analysis. These data can be used to complement the RAP analysis. However, to effectively use these data, information provided through the RAP process is needed to ensure the relevance of the data as discussed in Section 3.5.2. 3.5.3 Industry Data The RBI practices rely on engineering judgment and expe- rience with performance of engineered structures under actual conditions to estimate future performance. So-called âindustry dataâ are developed from the existing body of knowledge across the industry, frequently contained in the body of research literature available, to inform and support expert judgments. These data may include specific, quantita- tive data such as would be provided from models, or the com- bined or collective knowledge based on the existing body of research and past experience across the industry. This section provides two examples of âindustry dataâ that can be used to support analysis under the RBI process: a simple, com- monly available modeling example and a collective knowl- edge example. There exists a significant body of research concerning the degradation of highway bridges by common deterioration modes. There are two primary modes of deterioration that cause bridge damageâcorrosion of reinforcing steel in con- crete, and corrosion of steel bridge components. Certainly there are others, such as fatigue cracking, but corrosion and its effects can be associated with much of the damage occurring in bridges over time. Methods of determining the remain- ing life of elements and details based on fatigue mechanisms Figure 8. Graph showing Weibull distributions for time-in- condition for prestressed bridges in Oregon.
162 are documented and well known. Because of the significant importance of corrosion-based deterioration modes to the degradation of bridges, there exists a significant founda- tion of knowledge regarding corrosion and its effects on bridges, which can be leveraged to develop estimates of future behavior based on the age, current condition, and design attributes of a bridge. The rate of corrosion of steel and steel embedded in concrete varies widely according to localized conditions, with the local environment being a key factor. Geographical regions where de-icing chemicals are regularly applied generally have significantly higher corro- sion rates than regions where de-icing chemical use is low or even nonexistent. The local environment at the bridge, such as leaking joints or poor deck drainage, also has a significant effect. This section discusses generalized data regarding the corrosion rates in steel, for both steel members and steel embedded in concrete. This data is provided to illustrate the type of âindustry dataâ that can be used to support the rationale used by an RAP during the assessment process, and could be further developed if needed to address specific situations, or utilized as current industrial knowledge for general cases. 184.108.40.206 Corrosion in Concrete Structures The rate at which corrosion damage may develop varies widely for different geographical regions, depending on the level of exposure of the concrete to corrosive agents such as air-borne chlorides, marine environments, and the use of de-icing chemicals. The main factors that contribute to steel corrosion are the presence and amount of chloride ions, oxy- gen, and moisture. To illustrate how these factors affect struc- tures located in different geographical regions, commercial software was used to generate benchmark corrosion effects models for different regions of the country. One of the objectives of the modeling was to illustrate the variation in the likelihood of corrosion damage occur- ring in different geographical locations across the United States. Given that the inspection interval is uniform under the existing system, and that corrosion presents one of the most common and significant forms of damage to bridges, this study was intended to examine how much variation there might be in corrosion rates, and hence inspection needs, to assess corrosion damage across the United States. The results of the study are reported in terms of time to the initiation of corrosion. The time to the propagation of damage varies somewhat but can be considered to be on the order of 6 years for uncoated reinforcement to 20 years for epoxy-coated reinforcement, based on the rate that damage is expected to propagate once initiated in the reinforcing steel (36). Design parameters such as the amount of concrete cover, rebar spacing, and concrete material properties obviously affect the rate at which damage will propagate for a specific concrete compo- nent. These factors were assumed constant for the purposes of evaluating how quickly the effects of corrosion might be realized across different geographic regions. Fickâs second law of diffusion was used as the govern- ing equation to account for differences between geographic locations, such as temperature levels and ambient chloride concentrations. Fickâs second law of diffusion is generally stated as: 2 2 dC dt D d C dx = â Where C = the chloride content D = the apparent diffusion coefficient x = the depth from the exposed surface, and t = time The chloride diffusion coefficient, D, is modeled as a func- tion of both time and temperature, which represents the rate at which chloride ions travel through uncracked concrete. Higher temperatures allow for an increase in chloride diffu- sion as the ions have more energy to move, as compared to those in cooler temperatures. For the modeling, the benchmark concrete mixture assumed contained only Portland cement with no special corrosion protection strategies. The value of 0.05 percent by weight of concrete was used as the threshold chloride level for corrosion initiation for the uncoated rebar. This was done to represent a worst case scenario for corrosion initiation, given that no corrosion mitigation strategies were employed. Complete details on the analysis process are available in the literature. Six states across the United States that represented differ- ent geographical regions and thus different chloride build-up rates on the surface of the concrete, resulting from chlorides in the environment and de-icing chemical application, were modeled. These states included Arizona, Arkansas, Florida, New York, Washington, and Wisconsin. For each state, chlo- ride diffusion rates were modeled for rural highway bridges, urban highway bridges, and also for marine zones, where appropriate. Cover depths of 1 inch and 3 inches were used to illustrate the effect of concrete cover over the range of typi- cal cover. Representative results of the analysis for an urban highway bridges are presented here. Figure 9 visually illustrates the difference in the modeled time to corrosion initiation for different geographic regions. As shown in Figure 9, there are vast differences in the model time to corrosion initiation for different locations across the country. For aggressive climates, such as New York and Wisconsin, corrosion initiated in as little as ~7 years, while in less aggressive environments, such as Arizona, corrosion
163 initiation is not anticipated for almost 70 years. While this model does not consider localized effects, such as cracking of the concrete that can greatly increase the rate of chlorides intrusion into the concrete, it does illustrate that the time to corrosion for a generic, uncracked case varies significantly across geographic regions. As shown in the figure, New York, Wisconsin, and Florida have very similar behavior in terms of time to corro- sion initiation. These environments would fall more toward the severe or aggressive side of the exposure environment scale. Washington falls within a more moderate exposure environment. Arizona and Arkansas, with the slowest chlo- ride diffusion rates, are more mild environments. What is most notable in this data is that the time to corrosion for the simple, benchmark situation varies over an order of magni- tude across the different geographic regions modeled. This data illustrates that given the important role of corrosion in the time-dependent deterioration of bridges, uniform inspection intervals are unlikely to be the most efficient solu- tion to the inspection problem. Bridges in aggressive envi- ronments are likely to deteriorate more rapidly, and thus require more frequent inspections than bridges located in benign environments. This is only one among a multitude of factors that contribute to the need for inspections; how- ever, it is one of the most important and widespread. Data such as those provided through this simple modeling can be used, among other inputs, to provide supporting rationale for categorizing the OF with the RBI system. Element attri- butes that contribute to increased corrosion resistance, such as the use of epoxy-coated rebar or concrete mixes intended to resist the effects of corrosion are also needed for the analy- sis. This is particularly true in aggressive environments in which corrosion mitigation strategies might greatly increase the time to corrosion if they were used, supporting ratio- nale for a lower OF, or conversely a higher factor if they were not used. Such modeling is relatively simple, widely avail- able (the application used was available free-ware), and can include other relevant attributes to provide quantitative data to support the RAP assessment. 220.127.116.11 Corrosion in Steel There is also a significant amount of available literature related to the corrosion of steel bridges and the use and per- formance of protective coatings for steel bridge corrosion control (37, 44, 45, 55â57). During periods of active cor- rosion, it is generally accepted that corrosion rates of steels under most natural exposure conditions follow a linear rate to a point where the corrosion rate slows and flattens to a steady state rate less than that of the initial few years of cor- rosion. During the initial stages of corrosion, the rust scale builds up at the steel surface at a fairly consistent rate. Once the scale covers the entire exposed surface in a uniform man- ner, the rate of corrosion is limited by the rate of oxygen dif- fusion through the intact rust layer. Although this pattern of a âdeteriorating linearâ corrosion rate is dominant for boldly exposed steel, the rate itself is highly dependent upon the spe- cific exposure conditions. The corrosion rate tends to abate over time for many environments, but for the most aggres- sive environments (marine) this reduction in corrosion rate may not occur. Also, the corrosion rate at localized areas on the same structure, or even the same steel member, can vary. Therefore, it is prudent to view long-term corrosion rates as maintaining a near linear corrosion rate over time and to assume corrosion rates that are in the range documented for steel exposed to high moisture, high chloride environments. These corrosion rates tend to be in the range of 0.004 inches to 0.006 inches per year, per side of exposed steel, and these values can be used as a conservative planning rate to predict the impact of corrosion on a deteriorating member. Because of this relatively slow rate of corrosion section loss in steel, the accumulation of damage in the near future is predictable, particularly in a relatively short time frame such as the next 72 months. The condition of the structural steel and pro- tective coatings relative to corrosion can be easily assessed during inspections. If the current condition is not well understood, for example, the amount of section loss present in the bridge is not known, an effective assessment may not be possible. However, under an RBI scheme, the inspection process to be used must ascertain the level of section loss present, enabling the effective assessment of the likelihood and severity of future damage occurring. This data provides an example of the collective knowledge available and eas- ily accessible that can be used to provide a basis for RAP assessments. Figure 9. Time to corrosion initiation for different states based on a diffusion model.
164 3.6 Case Studies of the Methodology Two case studies were conducted to evaluate the effective- ness of the RBI method. The objectives of the case studies were as follows: â¢ Demonstrate the implementation of the methodologies with state DOT personnel and â¢ Verify the effectiveness of RBI analysis in determining suit- able inspection intervals for typical highway bridges. To demonstrate the implementation of the methodologies for RBI, two state DOTs were selected to be trained for and execute an RAP analysis for a family of bridges in their states. This included training in RBI technologies and executing expert elicitation according the procedures described in the Guideline. These RAP meetings resulted in data models for determining the RBI requirements for a family of bridges. These results were then tested to verify that the RBI practice developed through the RAP process was effective in deter- mining an acceptable inspection interval for the subject bridges. This was achieved through a back-casting process that utilized historical inspection records for specific bridges. These inspection records were used to assess if the inspection intervals identified through RBI would have been effective in maintaining the safety and serviceability of the bridge, were the RBI procedures applied in the past. This process com- pared the outcome of the RBI analysis with actual perfor- mance data for specific bridges, providing a validation of the RBI approach. The first case study was conducted for a sample of pre- stressed bridges in Oregon and the second one for steel bridges in Texas. In each case a group of bridge experts were gathered to conduct the RBI analysis during a 1.5 day RAP meeting in the host state. The composition of the RAP panels consisted primarily of state department of transportation engineers involved in the inspection, maintenance, and management of bridges within the state. The goals of RAP meetings were to develop RBI practices for the subject family of bridges. The objectives of the meeting were to identify and rank dam- age modes for each bridge component (deck, superstructure, and substructure), discuss deterioration mechanisms that lead to those damage modes, and identify related attributes. These attributes were then ranked according to their impact on the likelihood of severe damage occurring within a speci- fied time interval. CFs associated with these damage modes were also assessed. This section of the report describes the outcome of the case studies conducted in Oregon and Texas. This includes an overview of the RAP meeting agenda, resources used in the RAP meetings, and the results of back-casting completed to verify the RBI approach. 3.6.1 Summary Overview of RAP Meeting The RAP meeting consisted of a series of designed expert elicitations intended to develop comprehensive data mod- els for RBI. Processes implemented during the case studies were as described in Section 3.4. During the RAP, credible damage modes pertaining to the family of bridges being analyzed were identified through consensus of the RAP. Rel- evant attributes that contribute to likelihood of those damage modes progressing or occurring were also developed through the designed elicitations. Following the identification of the damage modes and relevant attributes, these attributes were ranked according to their impact on the likelihood for that damage mode (high, medium, or low) as a means of establish- ing an initial scoring approach. CFs for each damage mode and bridge component are also developed through a designed elicitation and consensus of the panel. Data from the RAP meetings were subsequently analyzed by the research team, organized into scoring models for each damage mode based on the RAP results, and utilized in the back-casting procedure to verify the effectiveness of the RAP results. 3.6.2 RAP Meeting Attendees The RAP meetings were attended by a variety of individu- als from participating states, as shown in Table 9. The RAP meeting in Oregon was attended by nine individuals, includ- ing DOT engineers, academics, and industrial representatives. The industrial representative participating on the Oregon RAP was from a fabricator that provided precast members for bridge projects in the state. The RAP also included a uni- versity professor with active research in the area of bridge evaluation and condition assessment. There were three indi- viduals with Ph.D.âs. In contrast, the RAP in Texas was comprised of only five indi- viduals, and all of the participants were employed by the Texas DOT. The participants generally held Bachelor of Science (B.S.) degrees, with one member holding a Master of Engineering (M.E.) degree. Most of the participants in the RAP meeting held at least B.S. degrees in civil engineering. A little more than 70% of the participants were registered Professional Engineers (P.E.). 3.6.3 Schedule and Agenda The RAP meeting in each state consisted of a 1.5 day face to face meeting in the host state. A webinar was presented approximately 1 week prior to the RAP meeting, to famil- iarize participants with the overall process, field any ques- tions participants may have, and identify any resources that may be needed for the meeting. This teleconference consisted of presenting overview slides introducing the concepts and
165 approach of the research and the planned activities during the RAP meeting. Most of the individuals that participated in the RAP meeting also attended this webinar to be introduced to the technology and prepare themselves for participation. 18.104.22.168 RAP Meeting Agenda The meeting agenda was developed to establish an effec- tive training pattern for the reliability assessment to be con- ducted. The previous expert panel meeting held during the initial phase of the project acted as the model for the RAP meeting agenda to be carried out in each state. However, in developing the RAP agenda, it was decided that the training goals would be best met by reorganizing the session into dis- tinct training and execution phases. In other words, train- ing associated with each of the aspects of the analysis, such as CFs, OFs, etc. were provided for the entire process before tasks to identify the parameters specifically for the family of bridges to be examined in the case study. This is in contrast to the expert panel meeting held during the initial phases of the research, during which the elicitations for each factor were conducted following training for that particular factor. The primary motivation for this decision was to ensure that the participants had a full and complete picture of how data would fit together in the final analysis, before making any decisions on what the parameters or attributes should be for the particular family of bridges being analyzed. The meeting began with an overview of the research approach, describing the goals and objectives of the RAP of the workshop and the overall research approach. This over- view session was followed by a training session on how to identify damage modes and attributes for bridge elements, for the purpose of estimating the OF required for the analysis. This session includes three exercises to illustrate the process to be undertaken in the expert elicitation for identifying dam- age modes and key attributes, and ranking the importance of those attributes in terms of the reliability of the element under consideration. In these exercises, a typical two-span steel bridge was presented as the example to pose questions regarding the typical damage modes that would be anticipated Name Emp. Current Position Highest Degree P.E. Oregon Participant A Oregon DOT Bridge Program Unit Manager B.S.M.E. Y Participant B Oregon DOT Structural Service Engineer B.S.C.E. Y Participant C Oregon DOT Senior Engineer Ph.D. Y Participant D Oregon DOT Bridge Operation and Standards Managing Engineer B.S.C.E. Y Participant E Oregon St U. Professor Ph.D. Str.Eng. N Participant F Knife River Corp Chief Engineer Ph.D. Y Participant G Oregon DOT Bridge Maintenance - N Participant H Oregon DOT Bridge Planner & Financial Analyst M.S. of Economics N Participant I Oregon DOT Senior Bridge Inspector B.S.C.E., AEStruct. Eng. Y Texas Participant A TX DOT Director of Field Operations-Bridge Division B.S.C.E. Y Participant B TX DOT State Bridge Constr/Maint Engr B.S.C.E. Y Participant C TX DOT Senior Bridge Const. and Maint.Engr M.E.C.E. Y Participant D TX DOT State Inspection Engineer B.S.C.E. Participant E TX DOT Bridge management Engineer B.S.C.E. Y Table 9. Listing of RAP meeting attendees in Oregon and Texas.
166 for this element. The members of the RAP recorded their responses on the bubble sheets and subsequently discussed the identified damage modes as a group. During these dis- cussions, credible damage modes were identified for further analysis. This exercise was followed by an elicitation of attributes related to the reliability/durability associated with the primary damage modes identified by the group and prioritization of those attributes from high to low. This exercise illustrated the process of the developing attributes and a semi-quantitative scoring scheme for a particular family of bridges as a means of identifying the OF for the RBI analysis. The process illus- trated in this example is later repeated by the RAP for the superstructure, substructure, and deck components for the subject family of bridges (i.e., prestressed superstructures in Oregon and steel superstructures in Texas). Training was also provided on the CF categories that are part of the analysis. A group exercise expert elicitation for consequences was administered to illustrate the process of identifying a consequence ranking for a particular damage mode scenario. During this exercise, panel members consid- ered the likely consequences of an identified damage mode progressing to the defined failure state (e.g., serious condi- tion) in terms of safety and serviceability of the bridge. Following these exercises, the expert elicitation for the fam- ily of bridges under consideration was conducted. Separate sections of the meeting address the superstructure, substruc- ture, and deck components of the bridge. The same process implemented in the illustrative examples was conducted for each component to identify the likely damage modes, attri- butes contributing to the reliability considering those dam- age modes, and prioritization of the attributes. These data were used to identify criteria and develop the initial scoring scheme to be implemented for assessing the OF for the vari- ous damage modes identified through the process. Consequence scenarios for each damage mode were also developed through group discussions. During this task, each damage mode identified in the earlier exercises was consid- ered, and an expert elicitation was conducted to identify the appropriate CF for each damage mode, and key factors that affect the factor selected. For example, if the damage mode is spalling damage on a deck, the CF may be high or even severe if ADT and traffic speeds are high, but moderate if the ADT and traffic speeds are low. Group discussion was used to develop consensus on these factors. Policies and common practices in the particular state also contributed to these discussions. The balance of the agenda was used to refine and com- plete the criteria and rankings for attributes, OFs, and CFs for the subject family of bridges. Screening criteria, surrogate data, and available data on attributes from existing inspection practices were identified. For example, if the subject state col- lects element-level data, how do various element ratings and damage flags correspond to the attributes and damage modes identified through the RAP process. At the completion of the meeting, it was anticipated that the damage modes, ranking for attributes, and basic scoring approach would be completed, as well as the CFs for various scenarios. However, discussions of the CFs revealed that cer- tain descriptions of the various CF levels were problematic, and these descriptions were subsequently modified to address these concerns. As a result, the RAP meetings provided pre- liminary data on the CFs to be used for the analysis, and these were later refined during the analysis process. The data from the RAP meeting were compiled and ana- lyzed by the research team following the meeting. These data were utilized to developed scoring models, or data models, reflecting the input from the RAP. These data models were then used in the back-casting process to evaluate the histori- cal performance of a sample population of bridges in each state to verify the effectiveness of the data models developed through the RAP process. 22.214.171.124 RAP Participants Notebook A participantâs notebook was prepared for distribution to members of the RAP. This notebook provided a refer- ence for use during the meetings. This notebook included standard information regarding the meeting, such as the agenda and copies of the slides to be presented during the training portions of the meeting, including space for partici- pantâs personal notes. In addition, copies of the forms to be completed during the meeting are included for future ref- erence following the meeting. The notebook also included color copies of the risk matrices to be used in determining the inspection interval based on the RBI analysis conducted by the RAP. The notebook also included key appendices from the Guide- line. These appendices include the guidance for identifying damage modes and attributes (i.e., OFs), CFs, determining the inspection interval, and the complete index and commen- tary of attributes identified in the Guideline. These portions of the handbook were included to act as references for the RAP participants to use during the RAP meeting for conducting the RBI analysis. 126.96.36.199 Software Development A software application was developed to support the RBI analysis of bridges based on the results of the RAP meetings. This application was developed within a spreadsheet pro- gram, and provides a simple and rapid means of implement- ing the damage modes, attributes, and scoring methodology for estimating the OF.
167 In this application, the user selects the attributes identified by the RAP for a particular damage mode, as shown in Fig- ure 10. A check box is used to select screening, design, load- ing, and condition attributes as described in the Guideline. Reserved attributes are included so a user can easily add addi- tional attributes that may not be included in the Guideline. Once the attributes are selected from the appropriate list- ing, the application organizes the selected attributes into a scoring page as shown in Figure 11. On this screen, pull- down menus are used to score the individual attributes for a particular bridge according to the scoring scheme developed. These pull-down menus allow a user to quickly select the appropriate ranking for a particular attribute based on the criteria developed through the RAP. The individual scoring for any attribute can be easily modified on an editing page to meet the requirements of a particular user. A hot-link is provided to the attributes com- mentary included in the Guideline, such that a user can easily refer to the rationale for a particular attribute and the envi- sioned scoring mechanism. After each attribute is scored, the OF score and guidance is automatically calculated for that damage mode. This software application was developed for use in the case studies to implement the analysis of the RAP from each state, and for testing that analysis against the historical perfor- mance of bridges during the back-casting. Looking forward, this software application provides a model for future, more sophisticated computer applications to allow for efficient and simple application of the RBI technology. For example, such a software module could be an add-on to the PONTIS pro- gram or other BMS, where many aspects of the scoring could be automatically obtained based on element ratings already collected as part of a routine inspection. 3.6.4 Back-Casting Procedure The case studies conducted in Texas and Oregon devel- oped a set of criteria and attributes for determining the OF and the CF, resulting in inspection intervals based on the risk matrix. These criteria and attributes produced a risk-based Figure 10. Example screen from software application showing selection of attributes. Figure 11. Example screen from software application showing pull-down menus for scoring attributes.
168 data model to be used to determine the appropriate maxi- mum inspection interval for a specific bridge or family of bridges. To verify if the use of these models provided a suit- able inspection interval that did not compromise the safety and serviceability of bridges, a back-casting procedure was used. In the back-casting procedure, the data models devel- oped by the RAP were applied to individual bridges based on historical inspection records. For example, the data model may be applied to a bridge based on the year 2000 inspection records for the bridge, resulting in an RBI interval that would have been determined in the year 2000, were RBI practices applied at that time. These results were then compared with the actual performance of the bridge, based on the inspection records for the years 2002, 2004, 2006, etc. to determine if the RBI inspection interval would have adequately addressed the inspection needs for the bridge. The criteria for determining the effectiveness of the data model included: 1. Did the condition rating for any component change sig- nificantly during the RBI interval in a manner that was not captured or anticipated effectively, but would have been captured (or detected sooner) by a standard, 24-month interval? 2. Were there any significant maintenance or repair actions completed that would have been delayed as a result of imple- menting an RBI interval (relative to a standard, 24-month interval)? 3. Were there any significant factors or criteria not identi- fied through the RAP analysis that were needed in the data models to provide suitable results? The procedure for back-casting consisted of obtaining the element-level inspection reports ranging back to approxi- mately 1998, depending on the availability of data for the spe- cific bridge. The data model was applied at each inspection year to assess the appropriate inspection interval based on the inspection data. As a result, the RBI interval may be consis- tent over the time period examined, decrease over that time period, or even increase during the time period as a result of a repair or improved condition rating or condition state. The overall concept of back-casting is shown schematically in Figure 12. This figure shows NBI ratings for an example bridge component over time. The RBI data model is applied to the bridge component based on inspection results from 1998. Assuming this results in an inspection interval of 72 months, the inspection results from each biannual routine inspection (24 months) is examined to see if there were any significant changes to the condition, or other events or circumstances detected by the routine inspection that may have been missed or delayed due to the RBI interval of 72 months. The RBI inter- val is calculated for each year there is an inspection result, indi- cated by the numerical results shown on the diagram. A change in the RBI inspection interval to 48 months is also shown in the figure. Assessment of the results includes determining if the change of inspection interval identified through the RBI criteria was effective in capturing the appropriate inspection interval, considering changes in the condition of the compo- nent reflected in the inspection results. It should be noted that the RBI inspection interval does not necessarily reflect NBI condition rating changes; however, since both depend on the condition of the component, they may be similar. Figure 12. Graph of condition ratings for a bridge component over time, showing schematic example of the back-casting procedure.
169 188.8.131.52 Inspection Data for Back-Casting Inspection data from each state were reviewed in detail to implement the data models developed through the RAP pro- cess, i.e., evaluation of the attributes identified by the RAP. This included design and loading attributes, which typically do not change over the life of the bridge and condition attri- butes that change as the bridge ages or undergoes repair or rehabilitation. Inspection data from Oregon consisted of PONTIS data file outputs, including photographs, notes, and standard Structural Inventory and Appraisal (SI&A) sheets. Avail- able inspection data from 1997 to present were assessed. In Texas, inspection data consisted of inspection reports that included standard SI&A sheets, NBI component rating sheets, and element-level data collected at the time of the inspection. Available data from 1999 through present were assessed for the Texas case study. 184.108.40.206 Review of Work As part of the back-casting analysis, a database of work projects maintained by the Oregon DOT was queried to determine if any significant repairs had been completed on the subject bridges during the interval of the back-casting process, and results were provided to the research team. This was done to ensure no significant events occurred on the bridge that resulted in major work or repair between inspection intervals, which may have been missed due to an extended inspection interval or may not be reflected in the inspection reports. In Texas, inspection records were more diversified. Work during the intervals between inspections was determined from the element-level data collected as part of the bridge inspection process. This was achieved by reviewing each inspection report for notes that would indicate that an improvement or repair was made to the bridge, or that an improvement or repair was urgently needed. Unexplained changes in the condition rating for a component were also investigated to determine if an urgent repair or rehabilitation activity was the source of the improvement. 220.127.116.11 Sampling To complete the back-casting verification study of the result of the RAP assessment, a sample population of bridges was assessed over a time period dating back 15 to 17 years. To determine the number of bridges to be assessed to develop a statistically significant result, a statistical analysis of popu- lation sampling was completed. Generally, such statistical models require some a priori knowledge of the anticipated variance in the population to be sampled to estimate the number of samples required to represent the overall popu- lation, considering the parameter to be measured. It was anticipated that the RBI criteria developed by the RAPs would include the current condition rating for a bridge as one of the criteria (attributes). Therefore, it was desired to select a sam- pling of bridges that has the same variation as the population overall, namely, that the natural variation of the inspection results of the overall population is represented in the sam- pling selected, based on the condition ratings provided in the inspection files. Experimental data from the FHWA visual inspection study (58) was used as a basis for the estimate, assuming that the variance of condition rating for all com- ponents in the FHWA study. Based on population sampling statistics, assuming that the desired accuracy was Â±0.5 con- dition ratings with 95.5% confidence resulted in a desired sample size of 17 bridges. For a confidence interval of 95%, the sample size for back-casting would be 10 bridges. Based on these results, the sampling of bridges included a minimum of at least 10 bridges; in the study, 17 bridges were selected from Texas and 22 bridges were selected in Oregon. 3.6.5 Statistical Analysis of NBI Data Statistical analysis of NBI data for the participating states was conducted to identify the characteristics of the each stateâs inventory and to support the RBI analysis. Analysis of NBI data was completed with the following objectives: 1. To determine the typical characteristics of the bridge inventories in the participating states of Texas and Oregon. 2. To develop quantitative data based on NBI condition rating history to be used to support the RAP analysis and ratio- nale for RAP-developed criteria. The objective of providing quantitative statistical data to support anticipated criteria that may be developed by the RAP during the course of the case studies can be illustrated as follows. Consider that the RAP identifies an attribute/criteria (among others) that a bridge has a superstructure rating of 7, based on the rationale that such a condition rating would indicate little deterioration or damage presently, and a low likelihood (i.e., OF) that severe damage would occur over the ensuing 72-month period. Analysis of the time-in-condition data from the NBI records provides quantitative data to sup- port this rationale, as discussed in Section 3.5. To conduct these analyses, data from the NBI dating back to 1992 were obtained from the Federal Highway Admin- istration (http://www.fhwa.dot.gov/bridge/nbi.cfm); these data are publicly available via the web site indicated. These data were used to develop data on the past performance of bridges in each of the participating states and to characterize the overall inventory in each state.
170 3.6.6 Bridge Inventories in Texas and Oregon The families of bridges selected for the two participat- ing states were based on the bridge inventories in each state. It was as desirable to have a sufficient inventory as to have a large inventory from which to draw sample bridges, and a representative population of bridges in terms of age. Table 10 shows bridge population statistics for the partici- pating states based on data available in the NBI. The bridge families selected are highlighted for both Oregon and Texas. Prestressed bridges were selected for analysis in Oregon because this bridge superstructure type made up almost 50% of the bridge inventory in that state, making it a significant popula- tion of bridges. This population of bridges has an average age of almost 29 years, consistent with the era of prestressed bridge construction, and there are more than 3,600 bridges of this material type. In Texas, the overall number of bridges is large, such that any family of bridges of similar superstructure materials would provide a suitable population of bridges for analysis. In this case, steel bridges were selected for analysis for three reasons; first, they provided suitable number of bridges for analysis, second, it was desirable to do one analysis for con- crete and the other for steel bridges, and, finally, the average age of the population was much older than the prestressed bridge population in Oregon, providing diversity in the ages of populations in these states. Figure 13 illustrates the age distribution for bridges in each state, as well as the age distribution for the bridge sample selected for analysis. Vertical lines on the figure indicate the mean ages for each population. As these distributions illus- trate, the mean or average age of bridge selected for analysis were older than the overall populations. This was considered desirable, because relatively new bridges are generally less challenging for RBI analysis, because they are usually in good condition and have good durability attributes. Therefore, selecting a population that was slightly older than the overall population presented a greater challenge for testing the RBI processes. Bridges included in the sample were generally randomly selected, with the exception that the desired sample of bridges for analysis had a geographic distribution across the subject state, and emphasis was placed on including bridges with suf- ficient historical data to make the back-casting meaningful. In Oregon, several bridges had limited historical data because the bridge was constructed after the year 2000; however, the sample of bridges was larger such that there were at least 17 bridges with the desired historical data available. 18.104.22.168 Bridge Sample Locations Bridges selected for back-casting were distributed geo- graphically within the states. Figure 14 shows the distribution of bridges in each state. As shown in the figure, bridges were selected from different regions of each state, although the geographic distributions of the sample bridges are affected by the population characteristics of each state. For example, in Oregon, population density is significantly higher in the western part of the state, and as such, the majority of bridges are in the western part of the state; the sample of bridge reflects this effect. 3.6.7 Time-in-Condition Rating The NBI data for Texas and Oregon were analyzed to deter- mine the typical lengths of time that a bridge component was in a particular condition rating. These data were derived from the NBI database, with some data trimming to accommodate the fact that the data sets are incomplete. That is, there are no data prior to 1992 or after 2011, so some trimming of these data are needed to improve the certainty of the derived time intervals. Data were trimmed from the data set if there were Bridge Inventory in Oregon Description No. Length(m) % of No. % of Length Average Age (year) Concrete 2,050 87,000 28 24 55.2 Steel 1,089 109,160 15 30 48.5 Prestressed concrete 3,612 154,877 49 43 28.9 Other 602 12,408 8 3 49.3 Total 7,353 363,444 100 100 40.8 Bridge Inventory in Texas Concrete 29,098 704,514 56 23.40 48.0 Steel 7,423 776,717 14 25.90 38.1 Prestressed concrete 13,781 1,392,706 27 46.30 23.6 Other 1,576 131,465 3 4.40 33.0 Total 51,878 3,005,403 100 100 39.6 Table 10. Bridge population statistics for Texas and Oregon.
171 5 years or less of consecutive data in a condition rating at the beginning or the end of the available time interval. The trimming value of 5 years was selected based on study of dif- ferent possible trimming values, ranging from 3 to 7 years, performed by the research team. This study indicated that the specific trimming value only had modest effects on the outcome of the analysis, and as such, 5 was selected as an acceptable value that ensured sufficient data were available for a meaningful statistical analysis. This method of trim- ming the data provides a suitably conservative result, because the analysis indicates that time-in-condition ratings are typi- cally much larger than 5 years for components in reasonably good condition (rated 6, 7, or 8). Data presented within this report include the superstructure and deck condition ratings; data for substructures were also analyzed. However, the deck and superstructure condition ratings typically change more frequently than substructure ratings, and as such, the deck and superstructure are the focus of the data reported herein. Figure 15 shows the time-in-condition results for pre- stressed bridges and decks of prestressed bridges in the state of Oregon. As shown in the figure, bridge superstructures rated in good condition tend to have longer intervals in that rating; as Figure 13. Age distributions of sample bridges and overall populations for (A) prestressed bridges in Oregon and (B) steel bridges in Texas.
172 the rating decreases, the time in a particular rating is reduced. For example, for the prestressed bridge superstructures illus- trated in the figure, the average time period a super structure was rated 8 was almost 14 years (s = 4.9 years), but the time period a superstructure is rated a 5 is less than 5 years (s = 2.7 years). The average time period for a prestressed super- structure is rated a 6 is 6.5 years (s = 3.8 years). For bridges in condition ratings of 4 or 3, these data are not particularly useful for two reasons; first, there are very few bridges in this cat- egory, and second, the bridges get repaired, and as such, the time interval in the condition is really more representative of a measure of how quickly these bridges may be improved or repaired rather than how long they remain at this condition rating. Figure 15B shows the time-in-condition rating for decks of bridges with prestressed superstructures. Similar obser- vations can be made, as shown. For example, a deck remains in condition rating of 7 for 10.2 years (s = 5.03 years); the time period a 6 remains a 6 is 6.4 years (s = 4.8 years), on average. Figure 16 shows the results of the trimming analysis for steel bridges in Texas. In this case, steel superstructures and bridge decks on steel superstructures were analyzed. For steel superstructures in Texas, the average time-in-condition rating of 7 was 10 years (s = 5.4 years), for decks of steel bridges, the average time-in-condition rating was found to be 11 years (s = 5.6 years). These data are useful as they reinforce and support the supposition that a bridge in good condition tends to stay in good condition for a long time interval (i.e., longer than the maximum inspection interval recommended using the pro- posed methodologies). For example, if one used the surrogate data of condition rating of 7 for superstructure, substructure, and deck to identify bridges with an appropriate inspection interval of 72 months, these data provide quantitative evidence to support that rationale, as discussed in Section 3.5. These data were used in the case studies to support âsurrogate dataâ analysis based on the data models developed by each RAP. In this analysis, the condition rating of 7 was used as âsur- rogate dataâ for the condition attributes to assume the OF would be low for condition-related damage modes. For these cases, the inspection interval of 72 months may be applied, based on these data. 22.214.171.124 Inspection Intervals Inspection intervals were determined based on the reli- ability matrix introduced in the Guideline. Figure 17 shows the proposed reliability matrix that is used for typical highway bridges. This matrix illustrates the appropriate Figure 14. State maps showing geographic distribution of sample bridges. Figure 15. Time-in-condition rating for (A) prestressed bridge superstructures and (B) decks based on NBI data for Oregon.
173 inspection intervals based on the estimates of the OF and the CF from the RAP analysis. In the figure, the inspection intervals are I =12 months, II = 24 months, III = 48 months, IV = 72 months, and V = 96 months. For example, when an OF is âLowâ and CF is âHigh,â the proposed inspection inter- val is 48 months. This matrix was applied to the results of the OF analysis, based on attribute scoring, and the appropriate CF for the given bridge component and damage mode. Each damage mode for each bridge component was analyzed using the RBI procedure, resulting in a data pair (OF, CF) for each damage mode for each component. These data were located on the risk matrix to determine the inspection interval for each bridge, as illustrated in the results section for each state. 3.6.8 Overview of Case Study Results The objective of this section of the report is to provide an overview of the results of the RAP meetings in each state. This section includes a summary of the damage mode and attri- butes identified in each state, and the consequence analysis that was conducted during the RAP meetings. 126.96.36.199 Summary of Damage Modes and Attributes This section summarizes the damage modes and attributes identified through the RAP process. These data provide the data model for assessing the OF as part of the RBI process, and as such, are documented here to illustrate how the data model was developed and what was considered. Due to the detailed nature of many of the attributes and description, most of these data have been placed in Appendix A for the Oregon case study and Appendix B for the Texas case study. These appendices document the attributes and attribute scoring for each damage mode that was used during the back- casting analysis. 188.8.131.52 Damage Modes and Attributes The expert elicitation process described in the Guideline and implemented during the case studies generally worked effectively to ascertain credible damage modes and identify key attributes affecting those damage modes. The process consists of having participants complete forms identifying credible damage modes, and then using a consensus process to list the damage modes, pare down those that are repeti- tive or irrelevant, etc. During the consensus process, dam- age modes identified by participants were recorded on a white board, along with the data from the likelihood esti- mates made by the participants. An example of this process is shown in Figure 18. This figure illustrates the beginning of the expert elicitation process, when the data from each member of the RAP is collected for discussion. The orange numbers shown in the figure indicate the number of panel members recording a particular likelihood (10%, 20%, 30%, etc.) for a given damage mode. As shown in the figure, this initial process included a number of damage modes for decks, including rebar corrosion, delamination, and spalling, which were pared down through discussion to a corrosion-related damage mode of spalling. Figure 16. Time-in-condition rating for steel bridge (A) superstructures and (B) decks based on NBI data for Texas. Figure 17. Reliability matrix for RBI.
174 Rutting was also identified as a credible damage mode for decks by the RAP in Oregon. This damage mode illustrates one benefit of a RAP consisting of bridge owners. Rutting of decks is related to the use or over-use of studded tires, and occurs along particular corridors in Oregon. It is unlikely many other states would identify this damage mode, but regardless, in Oregon such damage occurs and affects the ser- viceability of some bridges. It was the consensus of the panel that this damage mode was credible and required consider- ation in an RBI process. Texas identified punch-through as a credible damage mode. In this case, punch-through is not a corrosion-related dam- age mode, but rather related to the construction of thin decks, sometimes with poor quality concrete, that results in punch- through as a result of repetitive loading and age. Because much of the state is relatively arid, and use of de-icing chemicals is minimal, decks may have longer lives than they might in an area where corrosion is a significant issue. If the deck is thin and concrete quality is poor, punch-through can occur. Like rutting, this damage mode is due to local (state) policies and construction practices, namely that very thin decks were used during certain historical time intervals, and concrete quality was not well controlled at the time. In a state where corrosion damage was more prevalent, such a deck would deteriorate severally due to corrosion before such punch-through could occur. Like rutting, this damage mode is not likely common in other states. These relatively unique damage modes illus- trate the utility of the RAP approach. A summary of identified damage modes is shown in Table 11 for Oregon and Texas. It can be seen that damage modes of concrete deck and substructure are similar for Oregon and Texas. For superstructures, only the impact damage mode was common between prestressed and steel bridges analyzed in the two states, as would be expected, since the superstructures are of different material types. During the Oregon RAP, the panel expanded its assess- ment from open prestressed shapes, such as typical AASHTO shapes and Bulb-Tees, to include adjacent box girders bridges and prestressed slabs. The consensus of the panel was that the damage modes and attributes were Figure 18. Example of RAP data for damage modes in decks. Bridge Element Oregon (Prestressed or Post-Tensioned) Texas (Steel) Deck Spalling Rutting Cracking (Non-corrosion Induced) Spalling Punch-Through Cracking Delamination Superstructure Cracking (Shear) Strand Corrosion Fire Damage Impact Rebar Corrosion within the Span Bearing Seat Problems Adjacent Box Girders Rebar Corrosion/Section Loss Strand Corrosion (Fracture) Flexural Cracking Shear Key Failure Impact / Fire Fatigue Cracking Section Loss Fire Damage Impact Deflection Overload Bearing Failure Substructure Settlement Corrosion Damages Fire Overload Damages ASR Settlement Corrosion Damage Overload Damage ASR Table 11. Summary of damage modes in Oregon and Texas.
175 essentially identical for these families of bridges, with the exception that adjacent box girder bridges had a shear key damage mode that would need to be assessed as a screen- ing tool. These data are reflected in the summary of damage modes shown in Table 11. For each damage mode identified by the RAP, attributes that contributed to the likelihood of that damage mode occurring and progressing were identified through the RAP survey process and consensus of the panel. An example result of the consensus process is shown in Table 12 for deck spall- ing, summarized from the Oregon RAP. In this table, the attributes identified by the panel are shown in the left col- umn, followed by the rank that each attribute was assigned by the panel. This rank shows the unanimous vote on the rank for each attribute; this represents consensus developed among the panel, not necessarily initial results of the elicita- tion process. In some cases, individual members may have ranked these attributes differently, but consensus was devel- oped through discussion. Once consensus was developed, limits or parameters for scoring each attribute was developed through open discussion among the RAP member and results are shown in the table. For Oregon, which utilizes element- level inspection processes, many of the attribute parameters could be described using existing models from their element- level inspection manual. For example, for deck cracking, the element manual already has quantitative description of condition states 1, 2, 3, and 4, and therefore additional description was not neces- sary. For other attributes, for example ADTT, limits for high, medium, and low were developed through discussion. A comprehensive listing of the damage modes, attributes, and limits/parameters used in the back-casting analysis are included in Appendix A. The potential source of the data, based on state-specific inspection processes, is also tabu- lated in Appendix A. It should be noted that in some cases, the RAP identified attributes that were later correlated with existing element data following a more detailed review of the element-level manual. In other words, the RAP identified a given attribute and appropriate scoring limits during the meeting, and these were later found to match existing ele- ment descriptions in the Oregon element manual. A similar process was followed for Texas, which collects more limited element data during inspections. 3.6.9 CFs Designed expert elicitations were also used to develop CFs for each of the damage modes during the RAP meetings. For most damage modes, singular failure scenarios were assessed for each bridge component. The failure scenarios considered consisted of the component condition rating being serious (CR = 3), not necessarily structural failure. For decks, for example, the scenario considered in that the deck deterioration would typically be considered âseriousâ (CR = 3) during a normal inspection. For superstructure components (i.e., prestressed girders or steel girders), loss of load-carrying capacity for one member was considered. For Oregon, the CF for deck damage and substructure damage was considered to be generally Moderate. For superstruc- ture components, the initial CF developed in the RAP was High for most damage modes (except bearing area damage); this factor was sub sequently discretized during the analysis process. For Texas, issues were identified during the RAP meet- ing with the CF descriptions, as previously described, and these CF descriptions were subsequently adjusted during the back-casting to address these issues. These revisions adjusted the descriptions of different CF levels, but not the levels themselves. The CFs were subsequently assessed during the back- casting according to a series of scenarios to test and evaluate the influence of different parameters on the analysis. These focused largely on the CF assigned to the bridge superstruc- ture. The scenarios included considering the CF as uniformly Attributes Rank Limits H M L H M L Cracking 8 Existing model Delamination 8 >25% 11%-24% <10% ADTT 8 >5000 501-4999 <500 Location / Environment 8 Coastal and Mountain Valley (general environment) Desert Age 8 >50 10-49 <10 Dynamic Loading 8 Existing Model Rebar Corrosion 8 Rust/Black/Low Cover No stains, Epoxy/high cover De-icing 8 High Low Table 12. Example attributes rankings for deck spalling from the Oregon RAP.
176 High and considering the CF as uniformly Moderate, or deter- mining the CF based on structural redundancy and feature under the bridge. For the latter, the CF was based on the fol- lowing criteria: The CF was Moderate for the superstructure if: â¢ Superstructure consisted of more than four members AND â¢ Beam spacing of 10 ft or less AND â¢ Bridge not over a roadway. The CF was considered High if: â¢ Superstructure consisted of four members or fewer OR â¢ Beam spacing was greater than 10 ft OR â¢ Bridge was over a roadway. These criteria were based in part of the result of previous NCHRP research on redundancy of bridges and on discussions with engineers from the RAP panel (59). These dis cussions included previous experience with impact damage on struc- tures that resulted in loss of load-carrying capacity for a prestressed bridge member. The feature under the bridge, i.e., if the bridge were over a roadway, was included as a factor to consider based on the perceived risk of affecting the feature under the bridge. For example, if a primary bridge member lost load-carrying capacity or deteriorated to a serious condition, consequences may be increased either as a result of falling debris or signifi- cant displacement, or emergency shoring that may be required that would affect the serviceability of the roadway below the bridge. Additional factors considered for determining the CF included considering the traffic volumes; in these analyses, bridge decks with ADT greater than 10,000 were considered to have High CFs. This is intended to reflect a case where deck damage resulted in a major serviceability consequence. 3.7 Back-Casting Results for Oregon This section summarizes the results of the back-casting analysis conducted as part of the study. The State of Oregon provided 22 bridges from around the state for the analysis, as shown in Figure 14A. As shown in this figure, bridges were obtained from across the state to represent different envi- ronmental conditions surrounding the sample bridges. The damage modes, attributes, and data scoring models used in the back-casting process are documented in Appendix A. 3.7.1 Environments The environmental conditions considered in the analy- sis of bridges in Oregon differed according to the damage mode being considered. For example, for corrosion of super- structure metals (rebar or strands), the RAP identified three separate areas with coastal and mountainous regions being the most aggressive environment, while desert portions of the state represented the least aggressive environment, obviously. However, for spalling of bridge decks, the panel identified areas of the state where de-icing chemical use was highest because these areas are urban areas with high traffic volumes. For the damage mode of rutting, travel corridors that experi- ence high traffic volumes likely to be using studded tires were identified. Generally, these corridors were identified because they connected major urban areas and resort locations. The environments identified by the Oregon RAP are summarized in Table 13. 3.7.2 CFs There were six different CF cases considered in analyzing results in Oregon, as shown in Table 14. These different cases were selected to illustrate how different criteria established by an RAP might affect the outcome of the analysis. These included considering all superstructure damage modes as Damage Mode Environment Reason Corrosion Coastal and Mountainous Aggressive environment, high humidity and/or use of de- icing chemicals Valley or General Environment Desert Spalling Portland High application of de-icing chemicalsSalem Bend La Grande Rutting I-5 Presence of traveling traffic with studded tiresI-84 Table 13. Environments identified by the Oregon RAP for different damage modes.
177 âHighâ consequence, considering all superstructure dam- age modes as âModerateâ consequence, and determining the CF based on the redundancy of the bridge, as described in Section 3.6.9. Additional analysis was done to test the effect of including, or not including, the screening criteria of ele- ments with a condition state identified as CS 4 or 5. It should be noted that including this screening factor affects the OF, making the likelihood âHighâ for any element with any por- tion of the element reported in CS 4 or 5, a failed condition. Using these screening criteria does not change the CF, but may change the inspection interval. This case, which includes redundancy, feature under, and condition screenings is appli- cable for the subject bridges, and is shown in bold in Table 14. Finally, the CF was adjusted to consider the consequences for deck damage modes as âHighâ for bridges with high ADT, in this case determined by bridges with ADT of 10,000 or greater (according to NBI data). This case demonstrates the consideration of traffic volumes in the consequence analysis of a deck, which may be applicable in certain urban areas. 3.7.3 Back-Casting Results for Oregon Figure 19 illustrates the results of the back-casting proce- dure as done on one of the subject bridges. Shown in this fig- ure is the NBI condition rating history for the bridge, showing how the condition ratings have varied over the course of the back-casting period. On this graph, the inspection interval Case No. Description 1 High consequence for superstructure damage modes 2 Moderate consequence for superstructure damage modes 3 Superstructure damage mode CF is determined by redundancy and facility under bridge (screening not used) 4 Superstructure damage mode CF is determined by redundancy and facility under bridge â screening for CS 4 or 5 is used 5 All criteria in scenario 3 plus deck damage has high consequence if ADT > 10000, screening not used 6 All criteria in scenario 5 plus considering screening factors for CS 4 or 5 Table 14. CF cases used for back-casting in Oregon. Figure 19. Example of the back-casting process showing NBI condition ratings over time and the inspection interval determined through RBI analysis.
178 determined through the RBI analysis for each year there was an available element-level inspection report is shown enclosed in a box near the bottom of the figure. In a few isolated cases, there were not element-level reports available for every year, though NBI data was available. This example was selected as an illustration of applying the RBI analysis for each historical inspection result, and how that outcome may vary over the course of the life of a bridge. In this case, the inspection inter- val was reduced and then later increases following a repair, based on the condition of the bridge. This was not common occurrence, but it is useful as an illustration of how the results of the back-casting are summarized in the figure, and how the RBI inspection interval could vary over the life of the bridge based on the RBI analysis. Also shown on the graph are any repairs that had been completed on the bridge, and the year that these repairs were completed. It is very important to recognize that the RBI process is not intended to predict or track the NBI ratings. In some cases, changes in the inspection interval determined from the RBI analysis may coincide with changes in the NBI condition rating, because either these ratings are included in the analysis or the rating changes coincide with changes in the element condi- tion states that are included in the analysis. In other cases, these may not coincide, because the RBI analysis depends not only on the current condition, but also the potential for seri- ous damage to occur looking forward based on the bridge attributes (as expressed through the OFs) and the conse- quence of that damage. For example, a bridge rated in good condition according to the NBI condition rating may have a relatively short inspection interval, either because the poten- tial for damage is high based on the attributes of the bridge, or the consequences are high based on the redundancy or other circumstances influencing the CF. The research team believes this feature, i.e., the ability to look forward with an RBI analy- sis, is a significant advantage over the present calendar-based system. At present, in the current calendar-based approach there is no rational way to attempt to address the negative (or positive) attributes associated with future condition of a given specific bridge or family of bridges. Overall, the results of back-casting verified that the meth- odology was capable of determining an effective and safe inspection interval. There were no instances of bridge dete- riorating to a serious condition during the RBI inspection intervals recommended using the proposed methodology. The process was effective in differentiating inspection inter- vals based on the risk profiles developed through the RAP process, i.e., the OFs stemming from attribute scoring and the CFs. In some cases, bridges that were in generally good condition according to the NBI ratings resulted in short inspection intervals, indicating that the process was sensitive to risk factors that are not necessarily revealed through con- dition ratings. In other words, even though the condition of the bridge at the present time was generally good, there was a high likelihood of deterioration based on the design, envi- ronment, or loading of the bridge. In other cases, bridges that included components rated in fair condition were assigned longer intervals. Table 15 shows the overall results for each of the CF cases for the last inspection record analyzed for the 22 bridges, typ- ically from an inspection conducted sometime between 2011 and 2013. The CF Case 4 is highlighted in the table because this case, which includes consideration of the redundancy of the bridge, traffic under the bridge, and screening any bridges with elements with CS 4 or 5 reported, is a durable and widely applicable category. These data are based on the consequence cases described above and the data models developed through the RAP. The year of construction, superstructure type (sim- ple span or continuous), the facility under the bridge, and the scour rating are also shown in the table. These data were obtained from the NBI data for these bridges. This table also presents results for Cases 1 and 2, with CF for the superstruc- ture always high or always moderate, respectively. These data represent the simplest analysis of the CF for a superstructure. Cases 5 and 6, which included an ADT criteria for deck CF are also shown, to illustrate how a more restrictive criteria for the deck would affect the analysis. Scour ratings were not a part of the RBI analysis, as scour generally has its own evaluation procedures. Additionally, the scour rating was not considered in the overall analysis because this is a unique characteristic of the specific bridge, and there- fore may skew the results for a population of bridges selected at random. A bridge owner may choose to screen bridges with poor scour ratings as a policy; however, screening bridges in this manner in the current analysis would not be beneficial in measuring the overall effectiveness of the RBI procedures. Table 16 shows the summary of the RBI results for the Oregon bridges in terms of percentage of the sample popula- tion. Based on these analyses, again focusing on CF Case 4, approximately 41% of bridges would remain on a biennial inspection schedule, while just over 59% of bridges would have a larger interval of 48 or 72 months. These data illustrate the effect of using different criteria to identify the CF for the population of bridges, and results were as expected: relatively simple but conservative use of CF of âhighâ for the super- structure results in fewer bridges identified with extended intervals, using a less conservative âmoderateâ factor results in more bridges on extended intervals, etc. The overall results of the back-casting, considering each of the analyses conducted at each existing inspection record, are shown in Table 17. These results include 157 separate analyses done based on the inspection records, and for each of the six cases for determining the CF and OF described in Table 14 above. CF Cases 5 and 6, which include consideration of the ADT on the bridge deck, show only a modest differ-
179 Bridge ID Year Built Facility Under Simple span (SS) or Cont. (C) Scour Rating C as e 1 C as e 2 C as e 3 C as e 4 C as e 5 C as e 6 02376B 1975 Water SS 3 48 48 48 48 24 24 07801A 1973 Highway C N 24 24 24 24 24 24 01741B 1962 Relief for waterway SS 9 24 24 24 24 24 24 07935A 1973 Water SS 3 24 48 48 48 48 48 07935B 1973 Water SS 3 24 48 48 48 48 48 17451 1996 Water C 8 48 48 48 24 48 24 16454 1987 Highway SS N 24 48 24 24 24 24 16453 1987 Highway SS N 48 48 48 48 48 48 9546 1967 Highway/ waterway C U 24 48 24 24 24 24 00988A 1967 Water C 5 24 48 48 48 48 48 01056A 1970 Water C 5 24 48 48 24 24 24 9358 1965 Highway SS N 24 48 24 24 24 24 16873 1991 Water SS 8 48 72 72 72 48 48 18175 1999 Water C 8 48 48 48 48 48 48 01895A 1995 Railroad waterway C 8 24 48 48 48 48 48 9915 1970 Highway C N 24 48 24 24 24 24 8994 1962 Water SS U 24 48 24 24 24 24 8896 1963 Water SS 3 48 48 48 48 48 48 20666 2009 Water SS 8 48 72 72 72 48 48 19739 2007 Railroad waterway C 5 24 48 48 48 24 24 19738 2006 Railroad waterway C 5 24 48 48 48 24 24 19284 2005 Other C N 48 48 48 48 48 48 Note: N = not over waterway, U = bridge with âunknownâ foundation. Table 15. Overall results for each of the CF cases in Oregon. CF Case No. Inspection Interval24 month 48 month 72 month 1 64% 36% 0% 2 9% 82% 9 % 3 32% 59% 9% 4 41% 50% 9% 5 45% 55% 0% 6 55% 45% 0% Table 16. Summary of final back-casting intervals for 22 bridges in Oregon. CF Case No. Inspection Interval 24 month 48 month 72 month 1 62% 38% 0% 2 8% 82% 10% 3 28% 62% 10% 4 34% 57% 9% 5 39% 58% 3% 6 44% 53% 3% Table 17. Summary of back-casting intervals for 22 bridges in Oregon (all analyses). ence. The results shown in this table are generally consistent with those shown in Table 16, considering that the bridges are aging with time, and consequently the inspection intervals may be reduced. For example, at the end of the back-casting period, 50% of the bridges had a 48-month inspection interval assigned, as shown in Table 16. However, 68% of the bridges had a 48-month interval assigned at some point in the back- casting period, and 57% of all of the analyses conducted indi- cated a 48-month interval, as shown in Table 17. These data are significant in showing the consistency of the process when applied over 17 years of historical data through the back-casting process. Significantly, there were no instances of bridge deteriorat- ing to a serious condition between inspection intervals, and those with poor condition rating generally were assigned inspection intervals of 24 months based on the RBI analysis. For example, Figure 20 presents the condition rating history and RBI inspection interval for Bridge 16454. This bridge was
180 constructed in 1987, less than 30 years ago, and the back-casting assessment for the bridge was initiated in 1998, when the bridge was only 11 years old. However, the RBI inspection interval was determined to be 24 months, due to damage modes related to corrosion susceptibility of the superstructure. For this bridge, cracking in the superstructure was present early in the service life, resulting in increased likelihood of corrosion damage to the strands in the prestressed members. A repair completed in 2007 consisted of epoxy injection of the superstructure cracking. Looking forward from 1998, the superstructure condition deteriorated relatively rapidly as the bridge aged. For this bridge, the RBI assessed interval was 24 months throughout the back-casting period, an appropriate interval given the susceptibility to corrosion damage for this bridge. The validity of the short interval is also supported by the fact that the CR decreased from 6 to 4 around 2002. Again, the ability of the RBI method to identify the attributes that would suggest the superstructure is susceptible to damage resulted in the shortened interval. There were several bridges that had reported poor condi- tion ratings, and typically those had inspection intervals of 24 months assigned. There were some exceptions: for exam- ple, Bridge 07935A (Figure 21) had a reported condition rat- ing of poor (CR = 4) in 2013 and had an overlay installed, and the inspection interval assigned by RBI was 48 months. This may seem like a long inspection interval considering this deck apparently required an overlay. However, the element- level condition state for the deck was 100% in CS 2 (CS 2 = Patched areas and/or spalls/delaminations exist on either side of the deck. The combined distressed area is 10% or less of the Figure 20. Condition rating history and RBI inspection interval for Bridge 16454.
181 total deck area); the soffit element was 95% in CS 1 and 5% in CS 2, and the deck cracking element was 100% in CS 1. In this case, the assigned NBI condition rating appears not to be well correlated with the element condition states. Given that the NBI condition rating typically have a variability of Â±1, and in this case was not consistent with the element-level data, it may be that the condition rating is not reflective of the overall conditions. Since these element-level condition states contribute significantly to the likelihood estimate, a longer inspection interval was assigned. Arrows superimposed on the figure illustrate when an RBI inspection would have been conducted, assuming the start year of 1999. In this case, the year of the RBI inspection would not coincide with the year that the condition rating of 4 occurred, though the schedule year is somewhat arbitrary, being based herein on the earliest date of available data. This example was the most problematic of the 22 sample bridges included in the back-casting, in terms of the RBI interval assigned for the bridge. However, as described above, the apparent incongruity between the RBI inspection interval and the condition rating was explained by the element-level inspection results. 184.108.40.206 Risk Matrices The results of the analysis can be illustrated on the risk matrix to summarize the data and indicate the control- ling damage modes, i.e., those damage modes representing the highest risk or IPN. Table 18 shows the damage modes assessed in the Oregon case study, along with an alpha-numeric Figure 21. NBI condition rating history and RBI inspection intervals for Bridge 07935A. Deck Superstructure Substructure Spalling (D1) Cracking (S1) Settlement (F1) Rutting (D2) Strand Corrosion (S2) Corrosion (F2) Cracking (D3) Impact (S3) Rebar Corrosion Within the Span (S4) Bearing Seat Problems (S5) Table 18. Key to risk matrix summaries of RAP analysis.
182 identifier (D1, D2, etc.). Figure 22 includes the risk matrix for Bridge 16454 with the damage modes located on the matrix according to the results of the RBI analysis (OF and CF). As shown in this figure, the results of the RBI analysis are plot- ted in appropriate locations on the diagram. The locations on these plots describe the inspection interval identified, and can also be used to calculate the IPN to identify the most impor- tant damage modes as identified through the RBI process. For example, in the plot shown, the IPN for S1, S2, and S4 = 9, indicating that these damage modes (cracking, strand corro- sion, and rebar corrosion) have high importance related to the risk profile for the bridge. These data are useful for iden- tifying emphasis areas for the inspection of the bridge, and could be included in inspection procedures or guidance as a normal outcome of the RBI assessment. Such risk-based inspection procedures may improve the reliability of inspec- tion and communicate the engineering-based RBI assessment of the key damage modes for a bridge to inspectors in the field. Appendix C includes the controlling damage modes for the RBI analysis of bridges in Oregon. Frequently, several of the damage modes had similar risk profile, such that there is not a âcontrollingâ damage mode. This is typical for bridges in good condition, such that inspection intervals are typically longer. These controlling damage modes evolve during the service life of the bridge, as damage develops and affects the OF. 220.127.116.11 Surrogate Data for a Family of Bridges An analysis was conducted of the overall inventory in Oregon based on the results of the RAP analysis. The objec- tive of this analysis was to identify the population of low-risk bridges that were in very good condition and that could be assessed in an entirely data-driven process that did not require individual assessments of a bridge. Such bridges could be con- sidered for extended inspection intervals throughout the RBI analysis, based only on a screening process that utilized data in existing databases. These included a series of 22 items that were readily available, such as NBI items or bridge elements included in standard inspection reports. Table 19 indicates the individual items that were analyzed and the accepted values from the screening. Each of the criteria was based on attributes or items developed from the RAP analysis. Each of these items is shown in Table 19, along with the screening criteria used to analyze the inventory data. Generally, these criteria include bridges that have NBI condition ratings of 7 or higher and have no elements with any condition states of 3 or higher reported. In this case, scour ratings were consid- ered as shown in the table, eliminating bridges with unknown foundations, bridges without scour analysis, or bridges that are scour critical. Screening the Oregon databases was performed by the Oregon DOT, which provided a listing of all bridges meeting the element-level screening criteria included in the table. For the NBI criteria, filtering of the data was performed by the research team. Generally, these database searches and filter- ing took only a short time intervalâa matter of 1 hour or less, consisting primarily of inputting screening or filtering criteria in search functions and yielding immediate results. The results of the analysis indicated that 18% (652/~3600) of the prestressed bridge inventory met all of the criteria indi- cated in the table. For these bridges, the likelihood of serious damage developing in the next 72 months interval could be considered low or even remote, based on the RAP analysis. Assuming the CF to be moderate for this population of bridges, an inspection interval of 72 months could be assigned. If the effect of scour is not considered, or considered as a separate inspection requirement, the number of bridges meeting the other criteria was 970 bridges, or about 1 in 4 bridges. These data indicate that the RAP process can be used to develop criteria for an entirely data-driven process for identi- fying bridges that are very low risk, and the number of bridges meeting these criteria is significant (almost 1 in 5 prestressed bridges in Oregon). Such analysis takes only a matter of a few hours to complete, once the data items are identified through the RAP process. 3.8 Back-Casting Results for Texas This section of the report describes the results of back- casting for steel bridges in the state of Texas. This includes a description of the environments identified by the RAP for use in the OF analysis, the CF used in the back-casting analy- Figure 22. Risk matrix for Bridge 16454 illustrating results of the RBI analysis.
183 sis, overall results, and specific examples selected to illustrate implementation of the technology. 3.8.1 Environments The environmental conditions considered in the analysis of bridges in Texas also differed depending on the damage mode being considered for the RAP. Generally, the RAP identified an east-west interstate highway, I-20, as dividing the state into areas where de-icing chemical were likely to be used (north) and areas where they are very unlikely to be used (south). These environments were applied for most damage modes, such as spalling of bridge decks. For the damage mode of sec- tion loss in steel members, the RAP identified that the coastal areas were the most aggressive environment, followed by areas north of I-20 and a moderately aggressive environment, and all other areas being the least aggressive environment. 3.8.2 CFs There were four different CF cases considered in analyzing results in Texas, as shown in Table 20. These different cases were selected to illustrate how different criteria established by a RAP might affect the outcome of the analysis. These included considering all superstructure damage modes as âhighâ Prestressed Bridges (5, 6)* Deck No. Item Criteria Damage Mode Notes 1 #358 Deck Cracking SF CS 2 or less Deck Cracking 2 #359 Soffit Cracking SF CS 2 or less Spalling 3 Deck Elements CS 2 or less Spalling 4 Age Less than 50years Spalling Deck Condition 5 NBI Item 58 7 or greater Spalling Deck Condition 6 # 325 CS 2 or less Spalling Dynamic Loading 7 #370-374 Coded 1 or uncoded Fire Fire or Incident 8 #326 CS 1 only Rutting Deck Wearing Surface Condition Superstructure 9 NBI Item 54 17 ft. or greater Superstructure Impact Bridge Height 10 NBI Item 70 Coded 5 Cracking Legal Load Capacity 11 NBI Item 71 Coded 4 orgreater Impact No Overtopping 12 NBI Item 41 Coded A Cracking Open, No Restrictions 13 #362 Impact(SF) None Rebar Corrosion Traffic ImpactSmart Flag 14 Superstructure Elements#104, 109, 115 CS 2 or less Strand and Rebar Corrosion, Bearing 15 NBI Item 59 7 or greater Superstructure Condition Superstructure Condition Rating 16 Deck Joint Items (All) CS 2 or less Bearing Area Damage Failed Deck Joint 17 Bearing Elements (All) CS 2 or less Bearing Area Damage Bearing Issues 18 NBI Item 34 30 degrees or less Bearing Area Damage Bridge Skew Substructure 19 #360 Settlement SF CS 1 or uncoded Settlement Settlement 20 NBI Item 60 7 or greater Corrosion Damage Substructure Condition 21 NBI Item 113 Not U, 6 or 0-4 Settlement Scour 22 Substructure Elements CS 2 or less Corrosion Damage Sub. ElementConditions * 5 = prestressed concrete, 6 = continuous prestressed concrete: from NBI database. Table 19. List of criteria for data-driven screening process based on RBI. Case No. Description 1 High CF for superstructure 2 Superstructure CF is determined by redundancy and facility under bridge (screening not used) 3 Superstructure damage mode CF is determined by redundancy and facility under bridgeâscreening for pin and hanger used 4 All criteria in scenario 3 plus deck damage has high consequence if ADT > 10000 Table 20. CF cases used for back-casting in Oregon.
184 consequence, and determining the CF based on the redun- dancy of the bridge, as described in Section 3.6.9. Additional analysis was done to test the effect of including, or not includ- ing, the screening criteria for a bridge with a pin and hanger connection. This screening criteria were not identified during the RAP process, although it would likely have been identified during the course of a full-scale implementation of RBI. Again, this screening factor affects the OF, making the likelihood âhighâ for any component containing a problematic detail such as a pin and hanger. Using these screening criteria does not change the CF, but may change the inspection interval. Finally, the CF was adjusted to consider the consequences for deck damage modes as âhighâ for bridges with high ADT, again determined by bridges with ADT of 10,000 or greater (according to NBI data). The CF Case 3 is considered the most appropriate case for the analysis, and is highlighted in the following tables. 3.8.3 Back-Casting Results for Texas Table 21 lists the bridges analyzed in this portion of the study. This table includes data on the year of construction, the facility under the bridge, the structure span type (simple or continuous), etc. The CF Case 3 is highlighted in this table to illustrate the most likely or commonly applicable CF case that would be utilized to evaluate the bridges. Data models developed through the RAP process were used to analyze each bridge and determine the appropriate RBI inter- val. Table 22 shows the results of the analysis for the most recent year for which inspection results were available. As shown in the table, for the most recent analysis year, 12% of the bridges had a 72-month inspection interval and 53% with a 48-month inspection interval, while 35% were found to have a 24-month maximum interval. The maximum interval found during the analysis indicated that 24% of the bridges had an RBI interval of 72 months at some point during the back-casting period, indi- cating that the RBI practice included shorter intervals as these bridges became older and deterioration progressed. Appendix C includes the controlling damage modes for the RBI analysis of bridges in Texas. Frequently, several of the damage modes had similar risk profile, such that there is not a âcontrollingâ damage mode. This is typical for bridges in good condition, such that inspection intervals are typically longer. Table 23 shows the overall results from each of the 117 analyses conducted during the back-casting procedure. These data illus- trate the relative consistency of the process and the application of the attribute criteria to the steel bridge population in Texas. 18.104.22.168 Examples This section provides two examples from the analysis of bridges in Texas. The first example is a bridge that included Bridge ID Y ea r Bu ilt Facility Under Structure Type Scour Condition C as e 1 C as e 2 C as e 3 C as e 4 01-139-0-0769-01-007 1956 Waterway C 5 24 24 24 24 02-127-0-0014-03-194 1963 Highway C N 24 24 24 24 02-127-0-0094-04-057 1939 Waterway SS 8 48 72 72 48 02-220-0-1068-02-058 1957 Highway C N 24 24 24 24 05-152-0-0067-11-188 1990 Highway, Railroad C N 48 48 48 48 08-030-0-AA01-31-001 1985 Waterway SS 5 48 48 48 48 12-085-0-1911-01-003 1943 Waterway SS 8 24 48 48 48 12-102-0-0027-13-195 1979 Highway SS N 48 48 48 48 12-102-0-0500-03-320 1990 Highway C N 48 48 48 24 15-015-0-0025-02-162 1967 Highway C N 48 48 48 48 15-015-0-B064-55-001 1964 Waterway C 5 48 72 72 72 18-057-0-0092-14-210 1973 No Feature Under C N 48 48 24 24 18-061-0-0196-01-133 1960 Highway C N 24 24 24 24 19-019-0-0610-06-162 1971 Highway C N 24 24 24 24 23-141-0-0251-05-020 1934 Waterway C 8 48 48 48 48 23-215-0-0011-07-056 1948 Waterway C 8 48 48 24 24 24-072-0-0167-01-059 1970 Highway C N 48 48 48 48 Table 21. List of bridges analyzed in Texas. CF Case No. Inspection Interval24 month 48 month 72 month 1 35% 65% 0% 2 29% 59% 12% 3 35% 53% 12% 4 47% 47% 6% Table 22. Results of back-casting for bridges in Texas.
185 a pin and hanger connection. In the back-casting analysis, the presence of a pin and hanger connection was used as a screening factor that made the OF high, regardless of other attributes of the bridge. This screening factor is based on the historical experience that pin and hanger connections fre- quently present maintenance challenges. Figure 23 indicates the inspection intervals determined for the structure during the back-casting, along with the NBI condition rating history. As can be seen in the figure, the superstructure condition rating dropped 3 ratings, from 7 to 4, over a single inspection interval. According to the inspection records reviewed during the back-casting, this reduction was due to damage to the pin and hanger connection. Rehabilitation of this pin and hanger joint was required and was ongoing in 2013. This example is important for illustrating the importance of identifying screening factors in the RBI process. Screening factors are intended to identifying bridge attributes that make the likelihood of serious damage unusually high, unusually uncertain, or otherwise different than other bridges in a group. As shown in this example, the screening factor of bridges with pin and hanger connections was needed to capture the unusual behavior of this bridge. The second example was a steel multi-girder short-span bridge constructed in 1943. For this bridge, located in a coastal environment, back-casting indicated an inspection interval of 72 months between the years of 2001 and 2006, changing to a 48-month interval based on the results of the 2008 inspection. The change in the inspection interval for this bridge resulted from corrosion-related deterioration of the superstructure, i.e., likelihood for severe section loss. As shown in Figure 24, even though this bridge was 70 years old, the overall condi- tion of the superstructure was satisfactory at the beginning of the back-casting period, and subsequently reduced to fair, where the structure condition rating remained. The inspec- tion interval is also reduced during this period. Again, the RBI practice does not necessarily reflect the NBI condition ratings, as the data model includes specific information regarding the CF Case No. Inspection Interval24 month 48 month 72 month 1 29% 71% 0% 2 27% 58% 15% 3 32% 53% 15% 4 44% 50% 6% Table 23. Results of back-casting including all analysis. Figure 23. Historical NBI data and RBI inspection intervals for a steel bridge in Texas with a pin and hanger connection.
186 element condition state and other factors. It should also be noted that the deck and substructure are generally in satisfac- tory condition, and the superstructure is in âFairâ condition due to the damage mode of section loss caused by corrosion, known to be a slow-acting deterioration mechanism. Table 24 indicates the damage modes evaluated for the steel bridges in Texas, including damage modes for the superstruc- ture, deck, and substructure. Figure 25 indicates these damage modes plotted on the standard risk matrix, with Figure 25A being the risk matrix for the bridge including a pin and hanger connection, and Figure 25B the bridge with section loss. Considering Figure 25A, the data plotted on the figure illustrate that according to the damage modes identified, the inspection interval for this bridge would be 48 months. Recall that this bridge included a pin and hanger connection, used as a screening criteria to identify the OF as high for the super- structure. In other words, the data in Figure 25A indicates the inspection interval for the bridge if the bridge did not include a pin and hanger connection. This illustrates how screening criteria affect the analysis; for this bridge, the overall condi- tion based on the condition rating, notes, and element-level data suggest an inspection interval of 48 months. However, the bridge includes an attribute, i.e., a pin and hanger con- nection, that makes the anticipated behavior of the bridge Figure 24. Example bridge in Texas with decreasing inspection interval resulting from section loss. Deck Superstructure Substructure Spalling (D1) Section Loss (S1) Settlement (F1) Punch Through (D2) Impact (S2) Corrosion (F2) Cracking (D3) Fatigue Cracking (S3) Delamination (D4) Overload Damages (S4) Table 24. Damage modes for the steel bridges in Texas.
187 Figure 25. Risk matrices for steel bridges in Texas. unusually uncertain, and not typical of other bridges in the family. As such, this screening criterion is critical to deter- mining the effective interval for this bridge. Figure 25B indicates the risk matrix for Bridge 1922-01-003. As shown in this figure, the damage mode of section loss con- trols the inspection interval for the bridge. These data illustrate how individual damage modes can control the inspection interval for the bridge. In this case, the bridge is 70 years old, and in a relatively aggressive coastal environment. As such, it is rational that a shorter inspection interval would be required than if the bridge were in an arid environment. 22.214.171.124 Surrogate Data Surrogate data was analyzed for Texas based on the data models developed from the RAP. Because Texas has not tradi- tionally used its element-level data for bridge management pur- poses, and as such these data are not maintained within a single database, the surrogate data relied solely on NBI data to scan the inventory and identify bridges for the extended interval of 72 months. Table 25 indicates the parameters used in the analysis. Based on this analysis, it was found that 927 bridges, or 12.5% of the inventory in Texas, met all of these parameters Steel Bridges (3, 4) Deck No. Item Criteria Damage Mode Notes 4 Age Less than 50years Spalling Deck condition 5 NBI Item 58 7 or greater Spalling, Cracking Deck Condition Superstructure 10 NBI Item 70 Coded 5 Cracking Legal Load Capacity 11 NBI Item 71 Coded 4 orgreater Impact No Overtopping 12 NBI Item 41 Coded A Cracking Open, NoRestrictions 7 NBI item 54 17 ft or greater Superstructure Impact Bridge Height 15 NBI Item 59 7 or greater Superstructure Condition Superstructure Condition Rating 18 NBI Item 34 30 degrees or less Bearing Area Damage Bridge Skew Substructure 20 NBI Item 60 7 or greater Corrosion Damage Substructure Condition 21 NBI Item 113 Not U, 6 or 0-4 Settlement Scour Table 25. List of criteria for data-driven screening process based on RBI for Texas.
188 for the extended inspection interval. These data were also ana- lyzed without regard to the age criteria identified in Table 25. This resulted in a slight increase: 1068/7423 = 14.38%. 3.9 Discussion of the Case Studies in Texas and Oregon The case studies were used to verify the effectiveness of the RBI procedure developed through the research. Overall, the back-casting illustrated that the RBI process was effective in determining a suitable inspection interval for each bridge in the study. 3.9.1 Back-Casting Results The back-casting procedure was used to verify the effective- ness of the RBI process, and there were three primary ques- tions addressed, as discussed in Section 3.6.4. The following discusses each question individually in terms of the outcome of the back-casting. 1. Did the condition rating for any component change signifi- cantly during the RBI interval in a manner that was not cap- tured or anticipated effectively, but would have been captured (or detected sooner) by a standard, 24-month interval? A detailed review of the condition ratings for each of the bridges included in the study was conducted, as illus- trated in the examples presented herein. This review and analysis indicated that there were no cases where the con- dition rating changed unexpectedly in a manner that was not captured or reflected in the RBI inspection interval identified when screening criteria were used. Recall that screening criteria of CS 4 or 5 for prestressed bridges in Oregon, and a pin and hanger connection in Texas, were implemented in the analysis. 2. Were there any significant maintenance or repair actions com- pleted that would have been delayed as a result of implementing an RBI interval (relative to a standard, 24-month interval)? Reviews of the repair histories for the subject population of bridges were conducted based on available records. This review did not indicate any instances where there were sud- den or unexpected repairs required that would have been delayed as a result of RBI intervals. There were cases where routine maintenance or repair, such as a deck overlay, may not coincide with an RBI interval; however, this depends on when the RBI cycle was initiated. There were also several cases where repair or rehabilitation activities were performed during the back-casting window; however, the activities were generally consistent with the RBI analysis, and would not be adversely affected by the RBI implementation. For example, a bridge identified by RBI as being susceptible to corrosion damage had epoxy injection performed, consistent with the RBI analysis. In most cases, there were no significant repairs during the back-casting window. 3. Were there any significant factors or criteria not identified through the RAP analysis that were needed in the data models to provide suitable results? There was one case in each state where there were fac- tors that were not identified through the RAP processes were needed for the data models. In Oregon, a screen for elements with CS 4 or 5 was needed in the data models; in Texas, a screen for pin and hanger connections was needed. In both cases, these are relatively obvious addi- tions to the data models that were overlooked during the RAP meetings, but would likely be identified by anyone implementing the back-casting procedures. The overall objectives of the case studies were to demonstrate the implementation of the methodologies with state DOT per- sonnel, and verify the effectiveness of RBI analysis in determin- ing suitable inspection intervals for typical highway bridges. In terms of these objectives, the RAP meeting in each state, and the effectiveness of the data models developed through that RAP process, indicated that the processes developed for RBI analysis were effective, practical, and implementable using state DOT personnel in Texas and Oregon. The results of the back-casting process described above verified the effectiveness of the RBI procedures, and demonstrated that implementa- tion of the RBI practice did not adversely affect the safety and serviceability of the sample bridges analyzed. It should also be noted that for bridges where an inspection interval of 72 months was proposed using the RBI analysis, there were no cases of sudden repair, unexpected progression of damage, or sudden changes to the condition ratings for the bridge. These results indicated that the RBI procedures were effective in identifying a portion of the inventory, typically on the order of 10% of the sample bridge population, where an inspection interval of 72 months provided a suitable inspec- tion interval that did not compromise the safety and service- ability of these bridges. It should be noted that the sample population of bridges was older than the average age of the inventories in each state, such that the identified rate (~10%) would likely be higher for a population of bridges constructed more recently.
189 Conclusions, Recommendations, and Suggested Research This research developed inspection practices to meet the goals of (1) improving the safety and reliability of bridges and (2) optimizing resources for bridge inspection. The goals of the research have been achieved through the development of a new guideline document entitled âProposed Guideline for Realibility-Based Bridge Inspection Practices,â which has been developed based on the application of reliability theo- ries. This document meets the project objective of develop- ing a recommended practice for consideration for adoption by AASHTO, which is based on rational methods to ensure bridge safety, serviceability, and effective use of resources. A reliability-based approach was fully developed and docu- mented through the Guideline. Background information and foundation for key elements of the process have been further expanded in the present report, to provide additional details and perspectives on the research conducted as part of the proj- ect. However, the primary outcome of the study is the compre- hensive Guideline developed, which provides a new paradigm for bridge inspection. This new paradigm could transform the calendar-based, uniform inspection strategies currently implemented for bridge inspection to a new, reliability-based approach that will better allocate inspection resources and improve the safety and reliability of bridges. The implementation of the Guideline developed through the research was tested by conducting case studies in two states. The objectives of the case studies were to demonstrate the implementation of the methodologies with state DOT person- nel, and verify the effectiveness of RBI analysis in determining suitable inspection intervals for typical highway bridges. The verification of the methodology was analyzed using a back- casting procedure that compared historical inspection records and the results of RBI analysis. These studies demonstrated and verified the effectiveness of the procedures developed in the research for identifying appropriate inspection intervals for typical highway bridges. It was shown through these studies that the RBI practices identified appropriate inspection inter- vals of up to 72 months. It was concluded from these studies that implementation of the RBI practices did not adversely affect the safety and serviceability of the bridges analyzed in the study, based on the analysis of historical inspection records. These studies also successfully demonstrated the implementa- tion of the Guideline and the procedures therein using state DOT personnel. The results reported herein demonstrated and verified that inspection intervals of up to 72 months were suitable for certain bridges. Such extended inspection intervals would allow the reallocation of inspection resources toward bridges requiring more frequent and in-depth inspections, resulting in improved safety and reliability of bridges. As such, the project goals of developing a reliability-based bridge inspection practice that could improve the safety and reliability of bridges, and opti- mizes the use of resources, were achieved through the research. The following sections describe specific recommendations and suggested research, including detailed suggestions for con- ducting key elements of an implementation strategy intended to support the broader implementation of the research. 4.1 Recommendations The research reported herein has demonstrated the effec- tiveness of the RBI procedures for determining suitable inspection intervals for typical highway bridges, and as such, implementation of the RBI technology is recommended. The research also demonstrated that inspection intervals of up to 72 months were suitable for certain bridges and did not affect the safety and serviceability of bridges analyzed in the study. Such extended inspection intervals would allow the realloca- tion of inspection resources toward bridges requiring more frequent and/or in-depth inspections, resulting in improved safety and reliability of bridges. Based on these results, imple- mentation of RBI technology and inspection intervals of up to 72 months for certain bridges should be pursued. The procedure, methods, and approach described herein can be applied for atypical bridges as well. For example, C H A P T E R 4
190 non-redundant bridge members can be assessed using this approach, as illustrated in previous research (60). Specific attributes may differ for such an application; examples and illustrations of applying the RBI technology for these applica- tions should be pursued. The approach can also be applied to complex bridges, or to bridges with advanced deterioration. Analysis requirements may be more detailed and advanced; development of such analysis should be pursued to provide a uniform strategy for bridge inspection across the entire bridge inventory. Additional research and testing may be used to broaden the application of the RBI technology. Finally, the back-casting procedure utilized herein should be considered for implementation when RBI practices are to be used. This recommendation is based on the research result that indicated a screening criteria in each state was not identified during the RAP. Additionally, the RAP process may be subject to variability not observed in the research when applied over a broader platform. Back-casting provides a means for verification of models developed by the RAP and QA tool for assessing the RBI process. As such, the back- casting procedure provides a critical tool for the implementa- tion of RBI technology. 4.2 Suggested Research Suggested research stemming from this project includes developing applications of the technology for atypical highway bridges, including non-redundant members, complex bridges, and bridges with advanced deterioration, as described above. Additional research to demonstrate the consistency of the process across a larger population of bridge owners, and for families of bridges not examined herein, should be undertaken. In addition, efforts will be required to support implemen- tation of the technology. A comprehensive implementation plan, which includes additional research on such factors as economics of applying the methodology, is included below. 4.2.1 Implementation Strategy The implementation of a reliability-based inspection plan- ning process such as described herein will be a difficult challenge. As with any existing established procedures, specifications, or policy, change is difficult. The current U.S. bridge inspection program and associated procedures have been the standard since the early 1970âs, and considerable âinfrastructureâ has been developed to support the program. Well-established training, experience, and organizational structure within state departments of transportation will need to be modi- fied to meet the needs of an RBI practice. The workforce will need to be retrained to meet the needs of the new approach to inspection and inspection planning. Existing legislative requirements, including the NBIS will need to be modified to allow for the new methodology to be implemented. There- fore, a strategy for converting the established bridge inspec- tion programs from a uniform, calendar-based system to a reliability-based system is required. This section of the report describes implementation strategies and tasks to establish a new paradigm for bridge inspection based on the RBI pro- cesses described in the Guideline. A number of implementation challenges exist looking for- ward toward the adoption of the RBI methodology. Inspection program organizational structures and personnel may need to be modified to accommodate the larger role of engineer- ing and inspection planning required for RBI compared to a uniform, calendar-based approach. Personnel with suitable experience and knowledge to effectively conduct the necessary assessment will be required. In an era where government agen- cies are suffering significant fiscal challenges, often resulting in staff reductions, developing and retaining the necessary resources may be a challenge. A strong technical foundation for RBI will need to be developed to justify maintenance of the resources needed. Training and knowledge development to support RBI will also be needed to implement the technology on a widespread basis. Developing the necessary tools to train individuals in the various aspects of the technology and processes will be an important part of technology transfer and implementation. There will also be a significant political challenge to modify- ing an existing inspection system, which has been in effect for many years, with a process that may result in fewer inspections for certain bridges, even if the process results in an improve- ment in the safety of bridges overall. Engineers, inspectors. and maintenance personnel are likely to perceive the ben- efits of a more rational system, but the non-technical audi- ence may be more difficult to convince. Data from additional case studies or pilot implementation, economic impacts, and safety analysis will be required to provide evidence to support the new approach for inspection planning. However, because deterioration patterns for bridges typically require a long time period to manifest, and failures are rare, generating empirical data to measure improvements in safety will be a significant challenge. The implementation strategy described in this sec- tion is designed to address these issues, and will require some investment of resources to execute and complete effectively. Given these challenges, the implementation strategy has been developed to meet the following goals: â¢ Provide a technical foundation for widespread implemen- tation of the technology and â¢ Develop community support for the new inspection approach. Activities in the implementation strategy to provide a technical foundation for widespread implementation of the
191 technology include conducting additional case studies in cer- tain states to test and develop the technology further, develop- ing training modules and software to support the technologies, and conducting a study focused on the economic and safety impacts of transitioning to the new inspection approach. To develop community support, the implementation plan proposes developing an oversight committee to monitor and develop the Guideline and address bridge-owner needs, and developing an effective communication strategy. Throughout the implementation activities described herein, the FHWA can play an important role in assisting with moving the tech- nology toward eventual acceptance. 126.96.36.199 Implementation Tasks The strategy developed for implementing RBI for bridges in the United States will require a number of steps be com- pleted to test and refine the technology, develop support for transition to a new approach to inspection planning, and eventually gain widespread acceptance of the new technology. The implementation plan developed for the project consists of a number of individual tasks to be completed to achieve the desired goals. Task 1. Establishment of an oversight committee. Task 2. Additional case studies. Task 3. Development of training modules. Task 4. Develop a communications strategy. Task 5. Economic and safety impact study. Task 6. Software development and integration The sections that follow address each of the implemen- tation tasks to be completed toward widespread adoption of the RBI technology. 188.8.131.52 Task 1. Establishment of an Oversight Committee An important element of the longer-term implementation of RBI practices will be the establishment of a committee structure to oversee the development and maintenance of the technology. Implementation of the proposed methodology will require a significant shift in paradigm for inspection planning for high- way bridges. Consequently, there will need to be a long-term commitment with respect to maintaining and implementing the new methodologies contained in the RBI Guideline. As is common with many design codes and standards, a committee is needed to oversee, maintain, and further develop the Guide- line. This committee may be a subcommittee of the AASHTOâs Standing Committee on Bridges, or a subcommittee of the existing T-18 committee on Bridge Management, Evaluation and Rehabilitation, or even the committee itself. The committee should have the goal of providing objec- tive oversight and management of the Guideline and require- ments for RBI. Committee membership should be diversified, and include representatives from states in different geographic regions and with different types of bridge inventories. Par- ticipation of the FHWA in this committee would be desirable. During the transitional stages, which should be anticipated, the committee should include both states implementing or developing RBI processes, and states that are not yet utiliz- ing RBI. Care should be taken to ensure the participation of the community as a whole in the committee, including both bridge owners adopting the RBI Guidelines and those who have not yet made the transition. An important aspect of the proposed methodology is transparency, and any committee overseeing progress will require critical voices to be effective. The role of the committee should be as follows: 1. Oversee implementation of the technology across different states and act as a focal point for information interchange regarding statesâ experience, research, and developments. 2. Identify and recommend research and development needs to support the technology. 3. Recommend and approve changes to the RBI Guideline document. It is envisioned that the oversight committee or sub- committee will be a long-term or even permanent organiza- tion that will serve the larger bridge community. 184.108.40.206 Task 2. Additional Case Studies Additional case studies may be needed to test the applica- tion of the Guideline, identify implementation challenges, and provide additional data on the impact of transitioning to an RBI approach. The objectives of the case study should include: â¢ Assess the effect of RBI outcomes on the inspection prac- tice for different families of bridges. â¢ Evaluate implementation challenges. â¢ Assess the repeatability and consistency of the process. â¢ Provide baseline data for economic and safety impact study. â¢ Revise the RBI guideline as needed. The focus of these case studies will be to evaluate processes and methods described in the Guideline. To meet the objec- tives shown above, case studies should be conducted in several states to evaluate different families of common bridge types. 220.127.116.11 Task 3. Development of Training Modules Implementation of the RBI process will obviously require the development of training modules for those that will be
192 involved in the process. The inspection planning process is more involved and complex under an RBI scheme relative to a calendar-based inspection planning process. The assess- ment of reliability characteristics requires an understanding of the approach and the assessment needs. Therefore, train- ing for both members of the RAP and for inspectors that will implement the results of the RBI planning process may be necessary. This section provides an overview of the training needs for implementing RBI practices. 18.104.22.168.1 Training for RAP Member. The development of an expert panel like the RAP is relatively new to the indus- try, and will require a substantial commitment from stake- holders to identify and train individuals to participate in the process. Individuals that can provide expertise objectively are needed, and training in the tools and mechanism for providing that objective expertise will be required to ensure the method- ology is effectively implemented. Although these panels must be objective, they should also collectively possess an intimate knowledge of the inventory of the given agency or DOT. This group must also be isolated from the political and manage- ment pressures that could undermine the objectivity and effec- tiveness of the process (e.g., pressure to use the RAP process to simply extend intervals to save money). Effective training for RAP members will be one of the most critical parts of the implementation of RBI on a broad scale. This training should provide sufficient knowledge in the theory and underlying approach to RBI planning, deterioration and reliability science, methodologies for expert elicitation, and processes for determining the factors required for the analysis. Training modules developed during the case studies that were conducted as part of the research were shown to be effective, based on the results of the case studies, and they provide a strong foundation for the development of more formal train- ing for widespread implementation of the technology. 22.214.171.124.2 Training of Inspectors. Implementation of RBI practices will require training for bridge inspectors to develop the necessary understanding of the RBI process. RBI assessments for inspection planning provide a prioritization of inspection needs for a bridge based on the anticipated or expected damage modes and the importance of that damage in terms of safety of the bridge. Criteria developed through the RAP process identify key condition attributes used to determine the reliability of individual elements of the bridge and related criteria for reassessment of the inspection inter- val and scope. Additionally, the IPN identified through the RAP analysis prioritizes damage modes for inspection in a man- ner that is significantly different than the traditional, âdetect all the defectsâ approach. It is critical that the underlying approach and methodologies used in the planning process are understood by the inspectors implementing the practices to ensure adequate inspection in the field to support the over- all process. Training of this type was not addressed during the course of the research. Significant resources for inspector training already exist, and are generally implemented through the National High- way Institute (NHI). Existing training modules will need to be adjusted to accommodate the focus on damage and dam- age precursors that are a part of the RBI process. Inspector training modules like the 2-week inspector training course and supporting Bridge Inspectors Reference Manual will need to be modified to be more focused to incorporate the per- spective of the RBI process and its approach to ensuring the safety and reliability of bridges. Entirely new training modules could be developed to sup- port the RBI Guideline. However, while the approach to inspec- tion planning and the required reporting and inspection results differ for RBI, the damage modes that affect bridges are typi- cally well-covered in the existing training modules. Developing entirely new training for RBI including all of the information and examples already included in the existing modules would be a duplication of effort that is likely unnecessary. However, there are certain subjects not currently included in the exist- ing training modules that will be required to effectively imple- ment RBI. Table 26 describes two training modules that may be necessary for implementing RBI. This training for inspec- tors describes the underlying concepts and methodologies for RBI. It is intended that the training modules be presented at the appropriate technical level to develop sufficient back- ground knowledge for an inspector that will implement the RBI process. Ultimately, these specific modules may be added to the existing curriculum to provide training continuity and avoid duplication. The training modules included in Table 26 are intended to include enhanced training in specific inspection methodolo- gies required for implementing RBI, including those iden- tified in the RBI Guideline. For example, increased training for visual inspection to detect fatigue cracking, appropriate lighting and distance requirements, and thoroughness of inspection should be addressed through the training. Imple- mentation of the other basic techniques, such as sounding or concrete, should also be included in the training. More advanced technologies, such as advanced NDE tech- niques that may be specified for certain damage modes, will also require training. For example, if the RAP identifies the use of infrared thermography as a means of assessing delami- nations in concrete, then specialized training in applying this technology will be needed. Training in advanced NDE tech- nologies is typically advanced and focused, and utilization of the technology is specialized in nature. Specific training in these technologies should be developed as needed to meet specific owner needs.
193 126.96.36.199 Task 4. Develop a Communications Strategy As discussed, the proposed method is a significant change in paradigm for the bridge inspection community, and as such developing an effective communications strategy will be a key element of overall success. Education of policy makers and DOT administrators as to the benefits of the proposed methodologies will be needed. Although it is anticipated that owners (i.e., state bridge engineers) will likely embrace the proposed methodologies, it will require the buy-in of policy makers to actually implement any changes to the bridge inspection program. This group includes DOT administra- tors as well as appointed or elected officials. Since the current inspection program is covered by the Code of Federal Regula- tions (CFRs), lack of approval of policy makers could restrict any proposed methods from being implemented. Constant communication with the FHWA regarding the methodology and its progress will be very important to providing decision makers the information necessary to support future changes that may be needed. Additional interactions with state and local government rules, and potential conflicts with other specifications will also need to be assessed. There also exists the challenge that policy makers may have difficulty separating gross numbers of inspections from qual- ity and effectiveness of inspection. Numbers of inspections are frequently equated to safety, even though these two factors may be unrelated, particularly when the method of inspec- tion is ineffective for a particular damage mode or deteriora- tion mechanism. There will be a need for clear explanation of the approach to achieve buy-in from the policy makers, and even then challenges should be expected. There also exist the potential that the cost reallocations will be misinterpreted as a reduction in inspection require- ments to save money, rather than a reallocation of resources to be most effective in ensuring bridge safety. If viewed as a cost saving measure, the practice could lead to reduction in available resources for inspection, which is undesirable. Care needs to be taken to illustrate the enhanced reliability realized through allocating resources more effectively, and the ben- efits in terms of supporting the inspection and repair needs of an aging bridge inventory. Public and political acceptance of a system that may result in fewer inspections will rely on clearly communicating the benefits (more in-depth and focused inspections), not any cost saving. The communication strategy developed should include the development of non-technical publications that describe the RBI approach and highlight the benefits such as increased resources to focus where most needed and reductions in the risks and costs associated with unnecessary inspections. The improved reliability and safety of bridges that can be realized through improved inspection practices should be described as well as the improved management and responsible utiliza- tion of public funds highlighted in these publications. Technology transfer to the broader engineering commu- nity should also be developed as part of the communications strategy. 188.8.131.52 Task 5. Assessment of the Economic and Safety Impacts of RBI A key element in pursuing widespread implementation and acceptance of the RBI technology will be a critical assess- ment of the economic and safety impacts of converting from a uniform, calendar-based system to the RBI methodology. Such a study will likely be a required component of gaining the support of AASHTO and policy makers, who would natu- rally question the cost and safety impact of such a transition. Because inspection resources are reallocated and opti- mized under the RBI process, an organized and systematic assessment of the effect of the process on bridge safety will be required. This study should examine both the benefits of increased inspection thoroughness and assess any real or Module I Background Topics Notes Deterioration mechanism for bridges Overview of typical deterioration patterns Fundamentals of reliability theory and application to inspection Background overview of the underlying theories for RBI, reliability matrices, and likelihood Reliability assessments for RBI RAP process and basis for inspection procedures Module II Practices Understanding the IPN Required thoroughness of inspection and prioritization of damage modes Inspection needs, criteria, and reporting Focus and scope of inspections for RBI, access requirements, reassessment criteria, documentation, and reporting requirements Enhanced inspection methods for RBI Technologies and methods for detecting identified damage modes, enhanced methods for RBI, sounding and crack detection Table 26. Outline of training for inspectors.
194 potential diminishment of safety or safety effects on a given bridge inventory associated with varying inspection inter- vals to match the needs for bridges. The study should also examine the safety effects of continuing the current status quo, addressing both the cost and safety implications of the âdo-nothingâ approach, including the effects of decreasing available resources for inspection. Data from the case studies can be used to assist in this assessment process for this study. Study of the economic impact of transitioning to RBI is also needed. Because the methodology requires the invest- ment of increased resources for the planning of inspections, bridge owners may need to restructure traditional responsi- bilities and staffing to address the needs of full implemen- tation of the technology. Increased engineering efforts are required to complete RAP analysis, particularly in contrast to uniform, calendar-based approaches. Inspections under RBI Guidelines typically have increased scope and increased access requirements relative to traditional routine inspec- tions, and as such are likely to have increased costs. On the other hand, the inspection may be less frequent, such that the overall costs may be unaffected. The economic implications for transitioning to RBI obviously will vary according to the current inspection practice currently used in a state, and additional information on the actual or estimated economic impacts will be needed. This study of the economic and safety impacts of transition- ing to RBI practices will likely be a key tool to the eventual political and policy acceptance of the new technology. This implementation activity is high priority as a means of address- ing the issues associated with achieving acceptance of the new technology among the public, with policymakers, and with stakeholders. 184.108.40.206 Task 6. Software Development and Integration An important element of widespread implementation and acceptance of the new technology will be the development of software tailored to meet the needs of the process, and intended to integrate current or future data collection and storage approaches used by bridge owners. The processes for assess- ing the OFs, such as identifying and scoring key attributes of bridges, can be repetitive once established, and therefore lend themselves to software implementations. Many of the attributes identified through the analysis process may already be stored in existing databases and bridge management systems. Condition attributes and screening criteria may be implemented through existing software developed for bridge inspection and storing bridge inspection data, or in new software developed with RBI in mind. Such software is widespread in other industries and used for risk assessment and condition-based maintenance of facilities and components. The process of implementing a RBI practice can be simplified by the development of software to more rapidly utilize the methodology. Integration with exist- ing software and databases that store relevant information will be beneficial for efficiency in implementation. The case studies conducted as part of the research reported herein developed some basic software tools for these purposes; these tools will need to be integrated into existing software. Devel- oping software to assist in the RBI process will be necessary for implementation efforts to be successful.
195 References 1. National Bridge Inspection Standards, in 23 CFR part 650. 2004: USA. p. 74419â74439. 2. FHWA, Revisions to the National Bridge Inspection Standards (NBIS). 1988: p. 21. 3. Washer, G., Connor, R., Ciolko, A., Kogler, R., Fish, P., and Forsyth, D., Developing Reliability Based Inspection Practices: Interim Report. 2009, NCHRP: Washington, D.C. p. 76. 4. AASHTO, The Manual For Bridge Evaluation. 2008, AASHTO Pub- lications: Washington, D.C. 5. Estes, A. C., D. M. Frangopol, and S. D. Foltz, Updating Reliability of Steel Miter Gates on Locks and Dams Using Visual Inspection Results. Engineering Structures, 2004(26.3), Elsevier: p. 319â333. 6. Estes, A. C. and D. M. Frangopol, Updating Bridge Reliability Based on Bridge Management Systems Visual Inspection Results. Journal of Bridge Engineering, 2003. 8(6), ASCE: p. 374â382. 7. Estes, A. C. and D. M. Frangopol, Bridge Lifetime System Reliabil- ity under Multiple Limit States. Journal of Bridge Engineering, 2001. 6(6), ASCE: p. 523â528. 8. Sommer, A. M., A. S. Nowak, and P. Thoft-Christensen, Probabili- ty-Based Bridge Inspection Strategy. Journal of Structural Engineer- ing, 1993. 119(12), ASCE: p. 3520â3536. 9. Stewart, M. G. and J. A. Mullard, Spatial time-dependent reliability analysis of corrosion damage and the timing of first repair for RC structures. Engineering Structures, 2007. 29(7), Elsevier: p. 1457â1464. 10. Darmawan, M. S. and M. G. Stewart, Spatial time-dependent reli- ability analysis of corroding pretensioned prestressed concrete bridge girders. Structural Safety, 2007. 29(1), Elsevier: p. 16â31. 11. Stewart, M. G., D. V. Rosowsky, and D. V. Val, Reliability-based bridge assessment using risk-ranking decision analysis. Structural Safety, 2001. 23(4), Elsevier: p. 397â405. 12. Chung, H. Y., Fatigue Reliability and Optimal Inspection Strategies for Steel Bridges in Civil Engineering. 2004, University of Texas: Austin, Texas. 13. Enright, M. P. and D. M. Frangopol, Service-Life Prediction of Deteriorating Concrete Bridges. Journal of Structural Engineering, 1998. 124(3), ASCE: p. 309â317. 14. Estes, A. C. and D. M. Frangopol, Repair Optimization of Highway Bridges Using System Reliability Approach. Journal of Structural Engineering, 1999. 125(7), ASCE: p. 766â775. 15. AkgÃ¼l, F. and D. M. Frangopol, Time-dependent interaction between load rating and reliability of deteriorating bridges. Engineering Struc- tures, 2004. 26(12), Elsevier: p. 1751â1765. 16. Straub, D. and M. H. Faber, System Effects in Generic Risk Based Inspection Planning. ASME Conference Proceedings, 2002. 2002(36126): p. 391â399. 17. Goyet, J., A. Rouhan, and M. H. Faber, Industrial Implementation of Risk Based Inspection Planning Lessons Learnt From Experi- ence: Part 1âThe Case of FPSOs. ASME Conference Proceedings, 2004. 2004(37440): p. 553â563. 18. Poulassichidis, T. Risk-Based Inspection as a Reliability-Engineer- ing Tool for Fixed Equipment Decisions, in 2004 ASME Pressure Vessel and Piping Conference. 2004. San Diego, California: Dow Chemical. 19. Faber, M. H., et al., Field Implementation of RBI for Jacket Struc- tures. Journal of Offshore Mechanics and Arctic Engineering, 2005. 127(3), ASME: p. 220â226. 20. Frangopol, D. M., J. S. Kong, and E. S. Gharaibeh, Reliability-Based Life-Cycle Management of Highway Bridges. Journal of Computing in Civil Engineering, 2001. 15(1), ASCE: p. 27â34. 21. ASME, Inspection Planning Using Risk-Based Methods. 2007, The American Society of Mechanical Engineers. p. 92. 22. Regulatory Guide 1.178âAn Approach to Plant-Specific Risk- informed Decisionmaking for Inservice Inspection of Piping, N.R. Commission, Editor. 2003: Washington, D.C. p. 15. 23. API, API Recommended Practice 580, Risk-Based Inspection. 2002 American Petroleum Institute: Washington, D.C. p. 45. 24. Bowles, D. S., et al., Portfolio Risk Assessment: A Tool for Dam Safety Risk Management, in 1998 USCOLD Annual Lecture. 1998, Utah State University: Buffalo, New York. 25. FHWA, Recording and Coding Guide for the Structural Inventory and Appraisal of the Nationâs Bridges, U.S.DOT, Editor. 1995, U.S. Department of Transportation. 26. AASHTO, AASHTO Bridge Element Inspection Manual. 2010, AASHTO Publications: Washington, D.C. p. 170. 27. Cambridge Systematics, I., PONTIS 4.4 Userâs Manual. 2004, AASHTO: Washington, D.C. 28. Stewart, M. G. and D. V. Rosowsky, Structural Safety and Service- ability of Concrete Bridges Subject to Corrosion. Journal of Infra- structure Systems, 1998. 4(4), Elsevier: p. 10. 29. API, Risk-Based Inspection Technology, API Recommended Practice 581. 2008. 30. Straub, D., et al., Benefits of Risk Based Inspection Planning for Off- shore Structures. ASME Conference Proceedings, 2006. 2006(47489): p. 59â68. 31. ABS, Surveys Using Risk-Based Inspection for the Offshore Industry. 2003, American Bureau of Shipping: Houston, TX. 32. Faber, M. H., et al., Fatigue Analysis and Risk Based Inspection Planning for Life Extension of Fixed Offshore Platforms. ASME Conference Proceedings, 2005. 2005(41952): p. 511â519.
196 33. Straub, D. and M. H. Faber, Computational Aspects of Risk-Based Inspection Planning. Computer-Aided Civil and Infrastructure Engineering, 2006, Wiley: p. 179â192. 34. Moore, M., et al., Reliability of Visual Inspection for Highway Bridges. 2001, Federal Highway Administration: McLean, VA. 35. Agrawal, A., A. Kawaguchi, and C. Zheng, Bridge Element Deterioration Rates, TIRC/NYSDOT, Editor. 2009, The City College of New York, Department of Civil Engineering: Albany, New York. p. 105. 36. Ehlen, M. A., M. D. A. Thomas, and E. C. Bentz, Life-365 Service Life Prediction Model Version 2.0.1 Userâs Manual. 2009, Life-365 Consortium II. 37. Steel Bridge Design Handbook, Chapter 23, Corrosion Protection of Steel Bridges. 2010, National Steel Bridge Alliance. 38. Albrecht, P. and J. T. T. Hall, Atmospheric Corrosion Resistance of Structural Steels. Journal of Materials in Civil Engineering, 2003. 15(1), ASCE: p. 2â24. 39. Vu, K. A. T. and M. G. Stewart, Structural reliability of concrete bridges including improved chloride-induced corrosion models. Structural Safety, 2000. 22(4), Elsevier: p. 313â333. 40. Sun, X., et al., Analysis of Past National Bridge Inventory Ratings for Predicting Bridge System Preservation Needs. In Transportation Research Record: Journal of the Transportation Research Board, No. 1866, Transportation Research Board of the National Academies. 2004. p. 36â43. 41. Russell, H. G., NCHRP Synthesis 333: Concrete Bridge Deck Perfor- mance. 2004, Transportation Research Board of the National Acad- emies, Washington, D.C. p. 188. 42. Ramey, G. E. and R. L. Wright, Bridge Deterioration Rates and Durability/Longevity Performance. Practice Periodical on Struc- tural Design and Construction, 1997. 2(3): p. 98â104. 43. A. Sohanghpurwala, NCHRP Report 558: Manual on Service Life of Corrosion-Damaged Reinforced Concrete Bridge Superstructure Ele- ments. 2006, Transportation Research Board of the National Acad- emies, Washington, D.C. 44. Albrecht, P. and J. Terry T. Hall, Atmospheric Corrosion Resistance of Structural Steels. Journal of Materials in Civil Engineering, 2003. 15(1), ASCE: p. 2â24. 45. Klinesmith, D. E., R. H. McCuen, and P. Albrecht, Effect of Envi- ronmental Conditions on Corrosion Rates. Journal of Materials in Civil Engineering, 2007. 19(2), ASCE: p. 121â129. 46. Administration, NASA, Nondestructive Evaluation Require- ments for Fracture-critical Metallic Components. 2008, NASA: Washington, D.C. 47. Boring, R. L. A Review of Expertise and Judgement Processes for Risk Estimation, in European Safety and Reliability Conference. 2007. Stavanger, Norway. 48. Boring, R., Gertman, D, Joe, J., Marble, J., Galyean, W., Blackwood, L., Blackman, H., Simplified Expert Elicitation Guideline for Risk Assess- ment of Operating Events. 2005, Nuclear Regulatory Commission: Washington, D.C. p. 16. 49. API, Risk-Based Inspection, API Recommended Practice 580. 2002: p. 45. 50. Andersen, G. R., et al., Risk Indexing Tool to Assist in Prioritizing Improvements to Embankment Dam Inventories. Journal of Geo- technical and Geoenvironmental Engineering, 2001. 127(4), ASCE: p. 325â334. 51. Ayyub, B. M. Uncertainties in Expert-Opinion Elicitation for Risk Studies. 2000. Santa Barbara, California, USA: ASCE. 52. Hetes, B., et al., Expert Elicitation Task Force White Paper. 2009, EPA Science Policy Council: Washington, D.C. 53. Bulusu, S. and K. Sinha, Comparison of Methodologies to Predict Bridge Deterioration. Transportation Research Record: Journal of the Transportation Research Board, No. 1597, 1997, p. 34â42. 54. Hong, T.-H., et al., Service life estimation of concrete bridge decks. KSCE Journal of Civil Engineering, 2006. 10(4), Springer: p. 233â241. 55. ASTM, Corrosiveness of Various Atmospheric Test Sites As Measured by Specimens of Steel and Zinc, in Metal Corrosion in the Atmo- sphere, W.H.A.S.K. Coburn, Editor. 1968. p. 397. 56. FHWA, Characterization of the Environment. 2000, Washington, D.C. 57. FHWA, Characterization of the EnvironmentâRevisit of Exposure Sites in the Continental US. 2003. Washington, D.C. 58. Moore, M. E., Phares, B. M., Graybeal, B. A., Rolander, D. D., and Washer, G. A., Reliability of Visual Inspection of Highway Bridges, FHWA, Editor. 2001, U.S.DOT: Washington, D.C. 59. Ghosn, M. M., Redundancy in highway bridge superstructures. 1998, Washington, D.C. 60. Connor, R. J. and M. J. Parr, A Method for Determining the Interval for Hands-On Inspection of Steel Bridges with Fracture Critical Members. 2008, Purdue University. p. 32.
197 Abbreviations ADT Average Daily Traffic ADTT Average Daily Truck Traffic BME Bridge Management Element BMS Bridge Management Software CF Consequence Factor CFR Code of Federal Regulation CIF Constraint-Induced Fracture CS Condition State CVN Charpy V-Notch DOT Department of Transportation EMAT Electromagnetic-Acoustic Transducer GPR Ground Penetrating Radar HPC High Performance Concrete HPS High Performance Steel HS High Strength IE Impact Echo IPN Inspection Priority Number IR Infrared Thermography LFD Load Factor Design LIBS Laser-Induced Breakdown Spectroscopy MT Magnetic Particle Testing NBE National Bridge Elements NBI National Bridge Inventory NBIS National Bridge Inspection Standards NDE Nondestructive Evaluation NHI National Highway Institute OF Occurrence Factor PCI Precast/Prestressed Concrete Institute PDF Probability Density Function POD Probability of Detection POF Probability of Failure PT Dye Penetrant Testing QA Quality Assurance QC Quality Control RAP Reliability Assessment Panel RBI Risk-Based Inspection SIP Stay-in-Place SI&A Structural Inventory and Appraisal TRL Technical Readiness Level UPV Ultrasonic Pulse Velocity UT-T Ultrasonic Thickness Gauge
198 Developing Reliability-Based Inspection Practices: Oregon Pre-Stressed Bridges 199 Bridge/Deck/Spalling 199 Bridge/Deck/Rutting 200 Bridge/Deck/Cracking (Non-Corrosion Induced) 200 Bridge/Superstructure/Cracking (Shear) 201 Bridge/Superstructure/Strand Corrosion 201 Bridge/Superstructure/Fire Damage 202 Bridge/Superstructure/Impact 202 Bridge/Superstructure/Rebar Corrosion within the Span 202 Bridge/Superstructure/Bearing Seat Problem(s) 203 Bridge/Substructure/Settlement 203 Bridge/Substructure/Corrosion Damages (Spalling/Delamination/Cracking/Rust) A P P E N D I X A
199 Bridge/Deck/Spalling Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data C.9 and C.12 Cracking/Spalling Condition #358 CS 4 Or #359 CS 4 or CS 5 #358 CS 3 Or #359 CS 3 #358 CS 2 Or #359 CS 2 #358 CS 1 Or #359 CS 1 M 15 #358 Deck Cracking Smart Flag #359 Soffit Cracking Smart Flag (Oregon Coding Guide Pages 79 and 80) C.10 and C.11 Delamination/Patch Condition >25%CS 4 or CS 5 11%-24% CS 3 <10% CS 2 CS 1 H 20 Concrete Decks and Slabs without an Overlay : #12 - #26 -#27 -#38 - #52 -#53 Concrete Decks or Slabs with a Thin or Rigid Overlay: #18 - #22- #44 - #48. L.1 ADTT Loading >5000 501-4999 <500 M 15 Item 29 NBI L.3 (Exposure Environment) Location /Environment Loading Coastal and Mountain Valley (general environment) Desert H 20 Bridge File D.6 (Year Built) Age Design >50 years 10-49 years <10 years H 20 Item 27 NBI (Year Built) L.2 Dynamic Loading Loading >40 mph +CS 3 <40mph + CS 2 or CS 3 + <40mph CS 2 + <40mph CS 1 H 20 # 325 (Oregon Coding Guide Page 22) C.21 and C.13 Rebar Corrosion Condition Rust/Black/Low Cover No stains, Epoxy/high cover H 20 Concrete Elements(Oregon Coding Guide Page 38-41) L.5 De-icing Loading High (Regions like Portland, Bend, Salem, La Grand ) Low (All Other Regions) M 15 Items 3, 4, and 5 NBI, or Geographical map Bridge/Deck/Rutting Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data L.1 ADT Loading >15000 vpd 1000- 14999 vpd <1000vpd H 20 Item 29 NBI D.10 Wearing Surface Type Design - AC Bare Concrete/S TR overlay/Ep oxy Open Grid M 15 Item 108A NBI (Also page 120 Oregon Coding Guide) L.3 (potential to be exposed to high ADT with studded tires) Location Loading - - I-5 highway Portland to Salem and I-84 Portland All other locations - H 20 Items 3, 4, and 5 NBI, or Geographical map C.2 Current Condition (amount of rutting) Condition Present (>0.5") None (<0.5") 4 H & 4 M (M) + 15 (Oregon Coding Guide Pages 22 & 23)
200 Bridge/Deck/Cracking (Non-Corrosion Induced) Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data C.9 Cracking Condition Unsealed cracks exist in the deck that are of severe size (>0.060 in. wide) and/or density (<3â apart) Unsealed cracks exist in the deck that are of moderate size (0.025 to 0.060 in. wide) and density (3â to 10â apart). Unsealed cracks exist in the deck that are of moderate size (0.025 to 0.060 in. wide) or density (3â to 10â apart). The surface of the deck is cracked, but the cracks are either filled/sealed or insignificant in size and density to warrant repair activities. H 20 #358 Deck Cracking Smart Flag (Oregon Coding Guide Page 79) D.18 Skew Design >30 o <30o M 15 Item 34 NBI L.1 ADTT Loading >5000 501-4999 <500 H 20 Item 109 NBI D.20 Thickness Design <7" >7" H 20 Bridge File L.2 Profile/ Dynamic Loading Loading >40 mph + CS 3 <40mph + CS 2 or CS 3 + <40mph CS 2 + <40mph CS 1 H 20 Item 325 (Oregon Coding Guide Page 22) S.10 Span Type Screening Continuous or Non Continuous Bridge/Superstructure/Cracking (Shear) Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data L.4 and D.2 Overload Loading If it has already posted for less than legal load or exposed to overload Other H 20 Item 41 NBI (See also Oregon Coding Guide on page 95 ) D.18 Skew Design >30 <30 L 10 Item 34 NBI D.6 (Year of Construction) Age Design <2000 >2000 L 10 Item 27 NBI (Year Built) D.20 AASHTO Shear Design Screening AASHTO requirements were not considered in design AASHTO requirements were considered in design
201 Bridge/Superstructure/Fire Damage Reason(s) for Attribute Incidences of fire on or below a highway bridge are not uncommon. This type of damage is most frequently caused by vehicular accidents that result in fire, but secondary causes such as vandalism, terrorism, or other damage initiators should not be discounted. If fire does occur on or below a bridge, an appropriate follow-up assessment should be con- ducted to determine how the fire has affected the load car- rying capacity and the durability characteristics of the main structural members and the deck. This assessment is typically performed during a damage inspection immediately follow- ing the incident. Damage to bridge components resulting from a fire is either immediately apparent during the damage inspection, Bridge/Superstructure/Strand Corrosion Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data L.3 (Exposure Environment) ENV Loading Coastal and Mountain Valley (general environment) Desert H 20 Geographical Map C.8 (Corrosion Induced Cracking) Existing Damage Condition CS 4 CS 3 CS 2 CS 1 H 20 Prestressed/Post Tensioned Concrete Elements (Oregon Coding Guide Page 40) C.1 Current Condition Condition 5 and less 6 7 or greater H 20 Item 59 NBI (See also page 42 and 104 Oregon Coding Guide) D.11 (Minimum Concrete Cover) Cover Design 1.5" or Less, Unknown between 1.5" and 2.5" Greater than or equal 2.5" H 20 Bridge File D.12 (Reinforcement Type) Strand Type Design Uncoated Epoxy coated L 10 Bridge File D.20 and S.10 Bad End Detail Design Has Strand Exposure to outside environment Unknown Do not have Exposure to outside environment L 10 Bridge File or may manifest within the first 12-to-24 month interval fol- lowing the fire. Based on this observation, bridges that have experienced a fire may be screened from the reliability assess- ment until an inspection, which has been conducted approxi- mately 12 months or more after the fire, confirms that the fire has not affected the typical durability characteristics of the bridge components. The purpose of this screening is to ensure that damage from the fire has not manifested after the damage inspection. Assessment Procedure This attribute is scored based only on the occurrence of a fire on or below the structure being assessed. It is assumed that an appropriate assessment immediately following the fire incident (i.e., damage inspection) has been performed. Fire incident has occurred and an inspection 12 months after the fire has not occurred Bridge is not eligible for reliability assessment until inspection confirms that the bridge is undamaged There have been no incidences of fire on or below the bridge, or inspections conducted approximately 12 months or more after the fire have confirmed that the bridge is undamaged Continue with procedure
202 Bridge/Superstructure/Impact Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data D.3 Clearance Design <15' 15-16 >17' L 10 Item 10 NBI (Minimum vertical under clearance) L.8 High Water Screening Look at item 71 in NBI database- if the code is 3 the chance of over top is occasional Item 71 in NBI database (See also page 117 Oregon Coding Guide) Bridge/Superstructure/Rebar Corrosion within the Span Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data L.3 (Exposure Environment) ENV Loading Coastal and Mountain Valley (general environme nt) Desert H 20 Items 3, 4, and 5 NBI, or Geographical map C.6 and C.21 Previously Impacted Active Corrosion Existing Damage Condition #362 CS 2 Prestressed/ Post Tensioned Concrete Elements CS 4 Prestressed /Post Tensioned Concrete Elements CS 3 #362 CS 1 Prestressed /Post Tensioned Concrete Elements CS 2 Prestressed /Post Tensioned Concrete Elements CS 1 CS 3 H 20 #362-Traffic Impact Smart Flag (page 83 Oregon Coding Guide) Prestressed/Post Tensioned Concrete Elements (Oregon Coding Guide Page 40) D.11 (Minimum Concrete Cover) Cover Design 1.5" or Less, Unknown Between 1.5" and 2.5" Greater than or equal 2.5" H 20 Bridge File or Cover meter D.12 (Reinforcement Type) Strand Type Design Uncoated Epoxy Coated H 20 Bridge File Bridge/Superstructure/Bearing Seat Problem(s) Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data C.21 Corrosion Condition CS 4 CS 3 CS 2 CS 1 H+ 20 Prestressed/Post Tensioned Concrete Elements (Oregon Coding Guide Page 40) D.18 Skew Design >30 o <30 o L 10 Item 34 NBI C.22 Debris Condition - Flood region Debris INS.RPT Not Susceptible L 10 Item 113 NBI (See also Oregon Coding Guide on page 121) L.4 Overload Loading If it has already posted for less than legal load or exposed to overload other L+ 10 Item 41 NBI (See also Oregon Coding Guide on page 95) S.10 Design Details Design Simple Support Continuous Support Integral Abutments M 15 Bridge File C.4 Failed Joints Condition CS 3 CS 2 CS 1 Joint-less H 20 Deck JointsâOregon Coding Guide Page 54-60 C.2 Existing Damage Condition CS 3 CS 2 CS 1 H 20 Bridge Bearing ElementsâOregon Coding Guide Page 61-66
203 Bridge/Substructure/Settlement Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data D.21 Footing Type Design Spread FTG on soil/unknown Foundation - Drill Shaft friction Pile /ETC If foundation was based on Rock/Piles we do not need to deal with other following attributes H 20 Bridge File D.22 Subsurface Condition Condition Slide zone, clay, silt, shale, gravel Limestone solid, Rock H 20 Bridge File C.3 Existing Settlement Condition Active (No monitor data) Occurred but arrested None H 20 Item #360 on page 81 Oregon Coding Guide S.10 Scour Rating Screening 4-6 (Oregon Scour Code) - >7 <3 Item 113 NBI (See also Oregon Coding Guide on page 124) Bridge/Substructure/Corrosion Damages (Spalling/Delamination/Cracking/Rust) Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data L.3 (Exposure Environment) ENV Loading Coastal and Mountain Valley (general environment) Desert H 20 Geographical Map C.8 (Corrosion Induced Cracking) Existing Damage Condition CS 4 CS 3 CS 2 CS 1 H 20 Prestressed/Post Tensioned Concrete Elements (Oregon Coding Guide Page 40) C.1 Current Condition Condition 5 and less 6 7 or greater H 20 Item 59 NBI (See also page 42 and 104 Oregon Coding Guide) D.11 (Minimum Concrete Cover) Cover Design 1.5" or Less, Unknown Between 1.5" and 2.5" Greater than or equal 2.5" H 20 Bridge File D.12 (Reinforcement Type) Rebar Type Design Uncoated Epoxy coated L 10 Bridge File C.4 Failed Joints Condition CS 3 CS 2 CS 1 Joint less H 20 Deck JointsâOregon Coding Guide Page 54-60 L.5 De-icing Loading High (Regions like Portland, Bend, Salem, La Grand) Low (All Other Regions) M 15 Items 3, 4, and 5 NBI, or Geographical map
204 Texas Steel Bridge Attributes Summary 205 Bridge/Deck/Spalling 205 Bridge/Deck/Punch Through 206 Bridge/Deck/Cracking 206 Bridge/Deck/Delamination 207 Bridge/Superstructures/Sectionless 207 Bridge/Superstructures/Impact 208 Bridge/Superstructures/Fatigue Cracking 208 Bridge/Superstructures/Fire Damage 209 Bridge/Superstructures/Deflection Overload 209 Bridge/Substructures/Corrosion Damages (Spalling/Delamination/Cracking/Rust) 209 Bridge/Substructures/Settlement A P P E N D I X B
205 Bridge/Deck/Spalling Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low R em ot e Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data D.11 Clear Cover Design <1" 1"-2" >2" H 20 Bridge file or Covermeter D.10 Overlay Design Yes No L 10 Item 108A NBI (See also pages 9 and 10 Texas Coding Guide) C.10 Delamination Condition Yes No H 20 Pages 5 and 8 Texas Coding Guide D.8 Mixed design (Water) Design Poor Mix/Poor H2O All Else M 15 Bridge file L.1 ADTT Loading >5000 <5000 L 10 Item 29 & 109 NBI D.20 Thickness Design <7" 7"-8" >8" M 15 Bridge file D.19 Cold Joints Design Yes No M 15 Bridge file (or observation) C.9 Cracking (mapdense) Condition Yes No M 15 Pages 30 and 31 Texas Coding Guide L.3 Environment Loading Above I-20 All Else H 20 Bridge file D.6 Age Years ofServices Condition 50+ Other M 15 Item 27 NBI (Year Built) Bridge/Deck/Punch Through Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low R em ot e Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data D.20 Thickness Design <7" 7"-8" >8" H 20 Bridge file C.9 Map Cracking Condition Yes No H 20 Pages 30 Texas Coding Guide (Deck Cracking) C.10 and C.12 Delamination / spall to rebar Condition or Screening if more than 10% Delamination and spalling >6% Delamination and spalling 2%-5% Delamination and spalling <1% M 15 Pages 5 and 8 Texas Coding Guide D.8 Poor Concrete Mix(Poor Water) Screening M 15 Bridge file L.1 ADTT Loading >5000 <5000 H 20 Item 29 NBI L.3 Environment Loading Above I-20 All Else L 10 Bridge file (PONTIS Report) Previous Punch outs /rep Screening/Yes or No Pages 5 and 8 Texas Coding Guide
206 Bridge/Deck/Cracking Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low R em ot e Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data C.9 Existing Cracking Condition Yes No H 20 Page 30 Texas Coding Guide(Deck Cracking) D.20 Construction Tech/Spec Design Bad All Other M 15 Bridge file L.3 Environment Loading Above I-20 All Else H 20 Bridge file D.18 and D.19 Design Details (Cold Joints, Skew) Design Yes None H 20 Bridge file D.11 Cover Design <1" 1"-2" >2" H 20 Bridge file or covermeter Bridge/Deck/Delamination Similar Items in Guideline A ttr ib ut es T yp e of A ttr ib ut es High Medium Low R em ot e Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data D.11 Clear Cover Design <1" 1"-2" >2" H 20 Bridge file or covermeter D.10 Overlay Design Yes No L 10 Item 108A NBI (See also pages 9 and 10 Texas Coding Guide) C.12 Spalling Condition >6% 2%-5% <1% H 20 Pages 5 and 8 Texas CodingGuide C.10 Delamination Condition Yes No If more than 10% H 20 Pages 5 and 8 Texas Coding Guide D.8 Mixed design (Water) Design Poor Mix/Poor H2O All Else M 15 Bridge file L.1 ADTT Loading >5000 <5000 L 10 Item 29 NBI D.20 Thickness Design <7" 7"-8" >8" M 15 Bridge file D.19 Cold Joints Design Yes No M 15 Bridge file or Observation C.9 Cracking (mapdense) Condition Yes No M 15 Page 30 Texas Coding Guide (Deck Cracking) L.3 Environment Loading Above I-20 All Else H 20 Bridge file D.6 Age Years ofServices Condition 50+ Other M 15 Item 27 NBI (Year Built)
207 Bridge/Superstructures/Sectionless Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low R em ot e Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data L.3 Environment Loading Coast North of I-20 All Else H 20 Bridge file S.9 Existing SectionLoss Condition Yes No H 20 Pages 10, 11, and 15 Texas Coding Guide D.18 Deck Drainage onto Superstructure Design Yes No L 10 Bridge file C.22 Debris Condition Yes No L 10 Pages 23, 25, and 30 Texas Coding Guide (Pack Rust) C.4 Joint Leakage Condition Yes No L 10 Pages 23 and 24 Texas Coding Guide D.13 Built-Up Riveted Design Yes No H 20 Bridge file D.19 Deck Cold Joints Design Yes No M 15 Bridge file or Observation D.6 Age Exposure Design 50+ Other L 10 Item 27 NBI (Year Built) C.21 Corrosion Condition CR 3 or Greater/No Coating or Weather Steel Else L 10 Pages 10, 11, and 15 TexasCoding Guide Bridge/Superstructures/Impact Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low R em ot e Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data C.6 Existing Impacts Condition Yes No H 20 Pages 33 and 34 Texas Coding Guide D.3 Codes For Under clear (Vehicle) Design <=15'-6" 17'-6"<H<=15'-6" 17'-6"< Or No Highway under the bridge H 20 Item 54B NBI (Minimum Vertical Under Clearance)
208 Bridge/Superstructures/Fire Damage Reason(s) for Attribute Incidences of fire on or below a highway bridge are not uncommon. This type of damage is most frequently caused by vehicular accidents that result in fire, but secondary causes such as vandalism, terrorism, or other damage initiators should not be discounted. If fire does occur on or below a bridge, an appropriate follow-up assessment should be con- ducted to determine how the fire has affected the load carry- ing capacity and the durability characteristics of the main structural members and the deck. This assessment is typically performed during a damage inspection immediately follow- ing the incident. Damage to bridge components resulting from a fire is either immediately apparent during the damage inspection, or may manifest within the first 12- to 24-month interval follow- ing the fire. Based on this observation, bridges that have expe- rienced a fire may be screened from the reliability assessment until an inspection, which has been conducted approximately 12 months or more after the fire, confirms that the fire has not affected the typical durability characteristics of the bridge components. The purpose of this screening is to ensure that damage from the fire has not manifested after the damage inspection. Assessment Procedure This attribute is scored based only on the occurrence of a fire on or below the structure being assessed. It is assumed that an appropriate assessment immediately following the fire incident (i.e., damage inspection) has been performed. Fire incident has occurred and an inspection 12 months after the fire has not occurred Bridge is not eligible for reliability assessment until inspection confirms that the bridge is undamaged There have been no incidences of fire on or below the bridge, or inspections conducted approximately 12 months or more after the fire have confirmed that the bridge is undamaged Continue with procedure Bridge/Superstructures/Fatigue Cracking Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data D.17 Detail Category Design E0 or E' D orUnknown C or Better M 15 Bridge file or Observation C.18 History of Previous Cracking that was repaired Condition Yes No M 15 Page 30 Texas Coding Guide(Steel Fatigue) D.6 Year built Design Before 1975 orUnknown 1976-1984 After 1985 H 20 Item 27 NBI (Year Built) D.18 Skew Angle Design >30 <30 L 10 Item 34 NBI L.1 ADTT Loading >5000 <5000 H 20 Item 29 NBI S.7, C.19, and C.20 Active or unmitigated cracking due to any cause Screening Repair Must be shown to be working Pages 30 Texas Coding Guide Or Observation
209 Bridge/Superstructures/Deflection Overload Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data D.2 Load Posting Condition Cond Posting Des Post None H 20 Item 41 NBI -- Previous* Overload Damage Condition Yes No H 20 Bridge file -- Highway Ownership Condition Local State M 15 Item 22 NBI *Overload damages manifest in forms of settlement, rotation, and cracks. Bridge/Substructures/Corrosion Damages (Spalling/Delamination/Cracking/Rust) Similar Items in Guideline A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low Remote Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data L.3 ENV Loading Above I-20 All else H 20 Geographical Map C.8 (Corrosion Induced Cracking) Existing Damage Condition CS 4 CS 3 CS 2 CS 1 CS 4 H 20 (Texas Coding Guide Page 16â 20) C.1 Current Condition Condition 5 or less 6 7 or greater H 20 Item 60 NBI D.11 Cover Design 1.5" or Less, Unknown between 1.5" and 2.5" Greater than or equal 2.5" H 20 Bridge File D.12 Rebar Type Design Uncoated Epoxy coated L 10 Bridge File C.4 Joints Condition Condition 5 or less 6 7 or greater Joint less H 20 Joints Condition â Item 58 NBI details in bridge file or items #300 to #304 âTexas Coding Guide pages 23-24 Bridge/Substructures/Settlement A ttr ib ut es Ty pe o f A ttr ib ut es High Medium Low R em ot e Sc re en in g D eg re e of Se ve ri ty M ax Sc or e Source of data D.21 Footing Type Design Spread FTG on soil/unknown foundation - Drill shaft friction pile /etc If foundation was based on Rock/Piles we do not need to deal with other following attributes H 20 Bridge File D.22 SubsurfaceCondition Condition Slide zone, clay, silt, shale, gravel Limestone Solid, rock H 20 Bridge File C.3 Existing Settlement Condition Active (no monitor data) Occurred but arrested None H 20 Item #405 Texas Coding Guide on page 31 S.10 Scour Rating Screening 4-6 - >7 Or âNâ <3 Item 113 NBI (See also item #407 on Texas Coding Guide on page 32) Similar Items in Guideline
210 Controlling Damage Modes for Sample Bridges 211 Table C1. Controlling damage modes for RBI analysis of bridges in Oregon (CF Case 4). 212 Table C2. Controlling damage modes for RBI analysis of bridges in Texas (CF Case 3). A P P E N D I X C
211 Table C1. Controlling damage modes for RBI analysis of bridges in Oregon (CF Case 4). Bridge ID Inspection Interval Based on Case 4 Controlling Damage Mode 02376B 48 Rutting in deck and corrosion in substructure 07801A 24 Shear cracking, strand corrosion, and rebar corrosion within the span in superstructure 01741B 24 Corrosion related damage modes in superstructure and substructure 07935A 48 Rutting in deck, shear cracking, strand corrosion, and bearing seat problemsin superstructure 07935B 48 Rutting in deck, shear cracking, strand corrosion, and bearing seat problemsin superstructure 17451 24 Spalling in deck 16454 24 Rutting in deck, shear cracking, strand corrosion, and rebar corrosion withinthe span in superstructure 16453 48 Strand corrosion 9546 24 Strand corrosion and rebar corrosion within the span 00988A 48 Shear cracking in superstructure and settlement and corrosion in substructure 01056A 24 Most corrosion related damage modes 9358 24 Strand corrosion and rebar corrosion within the span 16873 72 All damage modes equal 18175 48 Most damage modes in superstructure 01895A 48 Rebar corrosion within the span for superstructure and settlement in substructure 9915 24 Strand corrosion and rebar corrosion within the span 8994 24 Rebar corrosion within the span 8896 48 Cracking in deck 20666 72 All damage modes equal 19739 48 Rutting in deck, corrosion related damage modes in superstructure, and settlement in substructure 19738 48 Rutting and spalling in deck, corrosion related damage modes in superstructure, and settlement in substructure 19284 48 Settlement in substructure
212 Table C2. Controlling damage modes for RBI analysis of bridges in Texas (CF Case 3). Bridge ID Inspection Interval Based on Case 3 Controlling Damage Mode 01-139-0-0769-01-007 24 Corrosion in substructure 02-127-0-0014-03-194 24 Fatigue cracking 02-127-0-0094-04-057 72 All damage modes equal 02-220-0-1068-02-058 24 Cracking in deckâsection loss, impact 05-152-0-0067-11-188 24 Fatigue cracking 08-030-0-AA01-31-001 48 Deck damages and substructure 12-085-0-1911-01-003 48 Section loss and corrosion in substructure 12-102-0-0027-13-195 48 All damage modes equal 12-102-0-0500-03-320 48 All damage modes equal 15-015-0-0025-02-162 48 All damage modes equal 15-015-0-B064-55-001 72 All damage modes equal 18-057-0-0092-14-210 24 All damage modes equal (screen because of pin and hanger connection) 18-061-0-0196-01-133 24 Punch through and cracking in deck, corrosion in substructure 19-019-0-0610-06-162 24 Impact 23-141-0-0251-05-020 48 Corrosion in substructure 23-215-0-0011-07-056 48 All damage modes equal 24-072-0-0167-01-059 24 All damage modes equal
Abbreviations and acronyms used without definitions in TRB publications: A4A Airlines for America AAAE American Association of Airport Executives AASHO American Association of State Highway Officials AASHTO American Association of State Highway and Transportation Officials ACIâNA Airports Council InternationalâNorth America ACRP Airport Cooperative Research Program ADA Americans with Disabilities Act APTA American Public Transportation Association ASCE American Society of Civil Engineers ASME American Society of Mechanical Engineers ASTM American Society for Testing and Materials ATA American Trucking Associations CTAA Community Transportation Association of America CTBSSP Commercial Truck and Bus Safety Synthesis Program DHS Department of Homeland Security DOE Department of Energy EPA Environmental Protection Agency FAA Federal Aviation Administration FHWA Federal Highway Administration FMCSA Federal Motor Carrier Safety Administration FRA Federal Railroad Administration FTA Federal Transit Administration HMCRP Hazardous Materials Cooperative Research Program IEEE Institute of Electrical and Electronics Engineers ISTEA Intermodal Surface Transportation Efficiency Act of 1991 ITE Institute of Transportation Engineers MAP-21 Moving Ahead for Progress in the 21st Century Act (2012) NASA National Aeronautics and Space Administration NASAO National Association of State Aviation Officials NCFRP National Cooperative Freight Research Program NCHRP National Cooperative Highway Research Program NHTSA National Highway Traffic Safety Administration NTSB National Transportation Safety Board PHMSA Pipeline and Hazardous Materials Safety Administration RITA Research and Innovative Technology Administration SAE Society of Automotive Engineers SAFETEA-LU Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (2005) TCRP Transit Cooperative Research Program TEA-21 Transportation Equity Act for the 21st Century (1998) TRB Transportation Research Board TSA Transportation Security Administration U.S.DOT United States Department of Transportation