Abstract
The New York State Department of Transportation (NYSDOT) started collecting AASHTO element-based inspection data as part of bridge inspections in 2016 after retiring its legacy condition inspection system that was in use for more than four decades. A research project was initiated to develop a methodology to integrate AASHTO element-based inspection data into NYSDOT bridge management decision-making processes, guidelines, and bridge performance measures. A logic was developed, based primarily on AASHTO element inspection data, for identifying the work needs for each bridge and predicting the change in element condition states should the resulting work recommendations be implemented. Based on a thorough evaluation of the developed logic, appropriate changes to NYSDOT’s bridge management system are recommended. This paper briefly describes the current study and findings.
Keywords
For the past several decades, the New York State Department of Transportation (NYSDOT) has been well-served by a sophisticated bridge management system that uses detailed inspection data to monitor and report on bridge and network performance and to guide planning for bridge maintenance, rehabilitation, and replacement. The recent (2016) transition from a state-specific inspection and recording system (“legacy system”) to one based on AASHTO elements requires substantial reconfiguration of the NYSDOT bridge management system. Although the bridge members that are rated are not substantially different under the two systems, the rating schemes are so conceptually different that it is not possible to simply map one rating scheme onto the other and fit the new rating data into the existing management system. Therefore, a research project was undertaken to develop and evaluate potential changes to the NYSDOT bridge management system that would enable the system to use AASHTO element inspection data directly in bridge and network performance measurement and in identifying bridge treatment needs ( 1 ).
A major objective of the research project was to develop decision rules that apply AASHTO element condition data ( 2 ) in selecting treatment levels for each element and in making project type recommendations for each structure. These decision rules are foundational to creating a bridge management system that makes effective use of the AASHTO element inspection data. The objectives for the new logic are to be more transparent, to provide useful information to bridge managers without being overly complex and detailed, to align well with actual bridge management and planning practice, and to be readily adjustable as experience is gained in the new rating system. This paper describes the treatment logic that resulted from the research project.
Using AASHTO element inspection data in treatment decision rules is not a novel concept. For example, AASHTOWare’s Bridge Management (BrM) software provides for the use of element health indices and element condition state percentages in setting up action triggers. However, to the best of the authors’ knowledge, BrM users continue to rely on National Bridge Inventory (NBI) major component ratings as the primary basis for action triggers. There is a dearth of reporting in the published or gray literature on the use of element inspection data in treatment selections. The recently published AASHTO Guide to Bridge Preservation Actions ( 3 ) is an exception. This guide provides suggested condition state thresholds to apply as maximum defect percentages to deem a bridge as in sufficiently good condition to merit preservation actions.
Background on NYSDOT Inspection Systems
Under the NYSDOT legacy system, inspectors rated individual bridge components on a 1 to 7 scale applied to the element overall. The AASHTO element rating system provides the simultaneous recording of both severity and extent of distress. NYSDOT bridge inspectors rate all the elements, including agency-defined elements, by recording the total quantity of the element (area, length, or number) and the quantity in each of the five possible condition states (Table 1).
New York State Department of Transportation Element Condition State Definitions ( 4 )
CS-5 is particular to the New York State rating system; the AASHTO manual ( 2 ) specifies only the first four condition states.
Severe = Red; Poor = Orange; Fair = yellow; Good = Green.
The elements are evaluated based on the defects that are present and their severity. Defects are specific indications of distress on an element. Examples of typical defects include concrete efflorescence, corrosion, settlement, and distortion. NYSDOT inspectors do not record quantities by defect. Inspectors do note some of the defects present in the narrative commentary on the inspection report.
An important distinction of the inspection rating systems in New York State, both the legacy and new systems, is that elements are rated on a span basis. For example, a three-span bridge will have three sets of deck element condition state quantities, one for each span. This is helpful because it indicates whether the defects are concentrated in a particular span or are dispersed throughout the bridge.
Methods
Element Treatments
Subject matter experts from NYSDOT and the project consultant team met during a 2-day workshop in June 2019 and participated in an expert elicitation process to define element treatments and treatment selection rules based on AASHTO element condition data.
Classification of Treatments
The expert elicitation produced a new, three-level categorization of element treatments and described in general terms what each treatment category means for each element. The three general categories of treatment are: light repairs, heavy repairs, and replacement. Reinforced concrete deck slab (Element 12) will be used as an example of treatment types and selection rules. For this element, repair levels are defined as follows:
Light repairs—light patching with a protective coating or sealer;
Heavy repairs—heavy patching (below the top mat of the rebar) with a thin overlay; and
Replacement—replacement of the entire deck slab in any given span.
Because of the large number of elements, the expert elicitation process did not elicit treatments and selection rules for every individual element. Treatments rules were grouped to cover similar elements that could have the same general treatments. Elements that were not specifically addressed were taken to be held to the same treatment rules as an analogous element with a different material composition, with certain exceptions permitted.
Treatment Selection Rules
The selection of treatment levels is determined by the percentage of the element in the poor (CS-3) and severe (CS-4) condition states. Treatment selection rules for reinforced concrete deck are shown in Table 2. The element replacement threshold of 5% shown for CS-4 means that deck replacement is triggered if at least 5% of the deck area is in CS-4. The 30% element replacement threshold value for CS-3 means that deck element replacement is triggered if the combined percentage of deck area in CS-3 and CS-4 is 30% or higher. If neither of these element replacement thresholds is met, then heavy repair is triggered if 1% (i.e., any minimal amount) of the deck area is in CS-4 or if 10% is in CS-3. Some elements originally had CS-2 thresholds established for treatments, but it was later determined that only the quantities in CS-3 and CS-4 were relevant to a treatment decision. A CS-2 threshold for reinforced concrete could lead to an infinite repair loop because the condition state definitions for reinforced concrete do not allow for a CS-1 rating on any patched areas. More generally, NYSDOT did not include CS-2 thresholds because it is not typical agency practice to make repairs on portions of an element that are in the conditions that are defined as CS-2.
Treatment Thresholds for Reinforced Concrete Deck (Element 12)
Note: A complete table of thresholds for all New York State Department of Transportation elements can be found in the research report.
Source: Leckrone et al. ( 1 ).
Severe = Red; Poor = Orange; Fair = yellow; Good = Green.
Project Type Selection Overview
This research project developed a logic to take element data from element treatment selection (described above) all the way to the identification of candidate bridge projects. An outline of the logic is as follows:
Stage 1A—Span work type identification
Step 1: Compute percentage in each condition state
Step 2: Select element treatments
Step 3: Identify span work types from element treatments
Stage 1B—Initial project type selection
Step 1: Apply multispan criteria
Step 2: Apply work type combination rules
Stage 2—Refined project type selection
Stage 3—Project selection alerts
The basic building blocks of this logic are the 11 work types. Work types for each span of a bridge are determined from the treatments that are selected for the elements on that span. For example, Work Type 5, deck replace, is triggered for a bridge’s span if that span’s deck element has met the treatment threshold for replacement. The work types and their triggers are presented in Table 3. Any span can have multiple work types triggered.
New York State Department of Transportation (NYSDOT) Work Types and Triggers ( 1 )
Vertical down is a term used in NYSDOT to refer to projects to replace failing joints and repair structure damage that occurred below the leaky joints. NYSDOT also has referred to this work type as “Bridge 5 to 7” in reference to repairs commonly applied to move a bridge rating from 5 to 7 (on the NYSDOT legacy system 1 to 7 scale).
After the triggered work types for each span are identified, they are compared with multispan criteria (described below) that determine whether the work type should be applied for the whole bridge.
Then, after the application of multispan criteria confirms the work types for the whole bridge, a bridge project type is composed as a combination of the work types. The bridge project types are combinations that make sense to package in Bridge Management System (BMS) modeling based on considerations of capital planning, high-level cost estimation, and work mobilization and execution. It bears emphasizing that the identified project types are simply project candidates that are output by the selection logic in the BMS. Bridge managers will apply subject matter expertise and detailed information about the bridges and network-level needs when applying budget constraints to identify projects for advancement in the capital program.
Table 4 lists the set of possible projects, identified as combinations of work types. The project type number starts with the primary work type “group” and adds decimal numbers afterwards to show other work types included with the primary one in a given project. For example, Project Type 3.2.2.1 would be output by the model with the label “Superstructure Rehab with Deck Rehab and Repainting.” If no work type is triggered, the bridge project type is simply “Cyclical Maintenance.” If “Wearing Surface Replacement and Zone Paint” are the only work types triggered, then the project type is “Wearing Surface Replacement.” If “Zone Paint” is the only work type triggered, then the project type is “Zone Painting.” The project type numbering scheme assigns progressively higher numbers to progressively lighter treatments.
Bridge Project Types
The following discussion presents the detailed steps in project type selection. Selection proceeds in three general stages: (1) work type selection and initial project type selection, (2) refined bridge project type selection, and (3) project selection alerts.
Project Type Selection Stage 1—Work Type Selection and Initial Project Type Selection
Stage 1A—Span Work Triggered
Step 1: From element inspections, calculate the percentage of total assessed element quantity in each condition state.
Step 2: Compare these condition state percentages to element treatment thresholds to determine what treatments are triggered for each element.
Step 3: For each span, compare the treatments triggered with the list of treatments triggering each work type. Step 3 determines the work type selections (there can be multiple work types) for the span. See Table 3 for the list of work types and generalizations of identified treatments. A list of all the specific element treatment triggers for each work type can be found in the research report ( 1 ).
Stage 1B—Stage 1 Bridge Project Selection
Stage 1B includes two steps to process the work types that were triggered for the spans of a bridge into a “first cut” project type selection for the entire bridge. The 30 project types that will result from Stage 1B are identified in Table 4.
The Stage 1 logic is exclusively concerned with AASHTO element condition data. A second project selection stage (Stage 2, described later) applies a few additional criteria that further refines the project type selections for a small number of bridges.
Stage 1B, Step 1: Apply Multispan Criteria to the Span Work Types Triggered
For each work type triggered on the bridge, Step 1 applies multispan criteria to the work type to determine work types for the whole bridge. The aim of the multispan criteria is to avoid anomalous project selections for large, multispan bridges, in which a deteriorated condition on a very small portion of the bridge drives the model to select a very high-level replacement or rehabilitation project. Accordingly, the multispan criteria are very generous, in that they allow for most span work types to be confirmed as work types for the entire bridge. Test application of the criteria to the April 2019 bridge inspection dataset found the criteria to affect work type selections in only a very small number of cases, as desired.
The four multispan criteria that are applied to each work type on each span are as follows:
Bridge size—If the bridge deck area is less than 100,000 ft2, the triggered span work type is applied to the entire bridge.
Span length—If the span is longer than 199 ft, any work type triggered on that span is applied to the bridge.
Percent of span count—If the work type is triggered for at least 5% of all spans, the triggered work type is applied to the bridge.
Percent of bridge length—If the work type is triggered for at least 5% of the total bridge length, the triggered work is applied to the bridge.
Only one criterion needs to be met for the work type to be applied to the bridge. As can be seen, work types for any bridge up to 100,000 ft2 of deck area, or any bridge with 20 or fewer spans, will automatically meet the multispan criteria. The latter condition suffices because any one span on a 20-span bridge will account for 5% of the number of spans.
Step 1 of the Stage 1B logic can produce multiple work types for a bridge. For example, one span may have “Superstructure Replacement” triggered and confirmed for the bridge whereas another span may have “Superstructure Rehab and Deck Rehab” triggered and confirmed for the bridge. These multiples will be processed into a single bridge project type in Stage 1B, Step 2, of the project selection logic.
Stage 1B, Step 2—Execute the Ordering and Combination Rules to Determine the Stage 1 Selected Project
Project types are selected by moving down through the work types shown in Table 3 and project types shown in Table 4. For example, if Work Type 1: Span Replace is selected, the project type will be 1: Bridge Replace, and the selector stops. If not, then the selector moves on to Work Type 2: Superstructure Replace and through its two combinations shown in Table 4 (i.e., with and without Substructure Rehab). This process is applied through all of the remaining work type combinations, stopping whenever one of the project types has been selected. If no work types are selected, then project type is 11: Cyclical Maintenance, meaning no element repairs or replacements were triggered for the bridge.
Project Types 6 Repaint through 10.2 Zone Paint are represented as standalone projects with the same title as the corresponding work type number. That is, they are not shown as combinations of multiple work types, as in the major rehabilitation project types in Groups 2 to 5. For example, for Project Type 6: Repaint, the project type selection logic will stop at the repaint work type and not register or indicate any other work types with higher numbers, such as Work Types 7: Deck Rehab or 8: Vertical Down. Because bridge repainting costs are typically much higher than the cost of repairs for Project Types 7 to 10, it is sufficient for cost estimating for the logic to capture just the repainting. That the selector stops at repainting does not mean that in practice a project with other repair work will not ultimately be developed. This interpretation also applies to Project Types 7 to 10.1.
Project Type Selection Stage 2—Refined Bridge Project Type Selection
Stage 1B Bridge Project Selection, which is based entirely on element condition data, provides useful information to bridge managers on the work needs of the system’s bridges. However, in some cases it can lead to selections that are not consistent with NYSDOT capital planning standards or not cost-effective, or simply not practical. For example, it is reasonable to expect that it will be more cost-effective to replace the entire structure rather than rehabilitate the substructure and replace the deck. The additional selection rules to account for these considerations are presented in Table 5 and explained in the text that follows. The table introduces a twelfth project type, “Supermaintenance,” which is based strictly on bridge size.
Substructure Rehab and Deck or Superstructure Replacement—Ideally, the BMS would calculate the cost of the project type and be able to compare that to a whole bridge replacement cost and select bridge replacement when the repair cost estimate exceeds some threshold percentage of bridge replacement. Cost research has not been done as part of this project. Until supporting unit cost data aer available, this criterion is used as a proxy for such a cost comparison.
Scour Critical NBI item—A bridge that is rated Scour Critical (rated 3 or less for data item 113 in the NBI coding system) is at serious structural risk from scour potential that requires replacement of the substructure to properly remedy this deficiency. Therefore, once any major rehabilitation project is needed on the bridge, it makes sense to replace the substructure, which effectively requires a full bridge replacement.
Deck age—In the unlikely event that a deck has not been replaced in over 50 years, it is not cost-effective to invest in a deck rehabilitation. The deck life expectancy has been reached and the deck should be replaced.
Extremely large bridges (over 270,000 ft2 deck area)—Bridges of this size are subject to their own unique maintenance regimen intended to keep them structurally sound and avoid the extremely high cost of replacement. Repairs are very costly as well as challenging to stage, so typical “rules of thumb” on treatments and costs may not apply. Moreover, work on these bridges is not entered in the funding pool with the rest of the state’s bridges. For these reasons, NYSDOT places these bridges in their own capital project category called “Supermaintenance.” Such bridges are subject to all of the selection logic and not assigned this special project type until Stage 2 of project type selection to provide the information on bridge conditions and repair needs that is generated by applying the treatment selection rules.
Recent capital project—A similar consideration applies to bridges that have recently had a major capital project completed. Such bridges would not be considered for a major project. Nevertheless, it can be useful to bridge managers to see the work needs of these bridges. This is particularly true if the project type selection logic is being used to rate the overall bridge condition and to assess bridge network performance.
Jack arch spans—It is not feasible to rehabilitate the superstructure or replace the deck of a jack arch bridge span because removal of the deck is very difficult and would cause serious damage to the beams. The entire jack arch superstructure in each span would need to be replaced. This criterion can be somewhat complex in practice because the jack arch is a span type, not a bridge type.
Culvert replacements—A few structures are categorized as culvert type, but the main load carrying member is classified as an arch element. When this element is selected for replacement, the project type registers as “Superstructure Replacement” because of the arch element, when in reality it should be recorded as needing a complete structure replacement. Stage 2 project selection will correct this anomaly.
Other culvert work—NYSDOT does not typically regard culvert repairs as major rehabilitation projects. Any rehabilitation project types that are selected for culverts are changed to the “General Repairs” project type at Stage 2 of the project type selection logic.
Stage 2 Project Type Selection Logic
Project Type Selection Stage 3—Project Selection Alerts
A third stage is included in the project selection logic. At this stage, the BMS would not automatically change any project selections, but would simply generate an alert to the existence of one or more factors that may influence the bridge managers to consider some change to the project selected by the logic.
In some cases, the modeling logic will select a project type that might not be optimal for the bridge, given other considerations. For example, there may be some inadequacy or vulnerability that could be addressed when major work is being done. Some project types might not prove feasible or cost-effective given the structure type. At Stage 3 of project selection, the following considerations are applied to the selected bridge projects:
Inadequate underclearance,
Hydraulic vulnerability,
Fracture critical,
Railing condition,
Bridge, deck, or superstructure age over respective design life,
Adjacent box beam structure type,
Truss superstructure,
Timber construction,
Annual average daily traffic over 75,000,
Very large bridge (over 75,000 ft2 deck area), and
Historical significance listing.
Stage 3 of project selection affords the opportunity to apply other condition rating information assigned by the inspectors. In addition to the 11 considerations listed above, project selection alerts are generated when the overall condition rating assignments (e.g., NBI component general condition ratings) do not align with the level of work generated by the element-based selection logic.
Results
Bridge Project Type Selection Results
The Stage 1 project type selection logic was applied to the April 2019 National Bridge Inspection Standards (NBIS) bridge dataset (submitted to Federal Highway Administration as per NBIS) in an Excel Workbook model. The dataset includes all highway bridges in New York State. The number of bridges in each project type is presented in Table 6.
Project Types Assigned by Stage 1 Selection Logic: All National Bridge Inspection Standards Highway Bridges in New York State ( 1 )
Note: In project type nomenclature, Superstructure is shortened to “Super,” Substructure is shortened to “Sub,” and Rehabilitation is shortened to “Rehab.”
Use of Project Type Selection in Performance Measurement
NYSDOT uses a four-level scale for characterizing bridge condition and for measuring the performance of the bridge network ( 5 ). The scale is composed of four condition categories, considered as generally relating to the level of work needed to bring the bridge to a state of good repair. The “good” category includes bridges in good condition that generally require preventive and corrective maintenance actions such as bridge washing, deck sealing, and bearing lubrication. The “fair–protective” category includes bridges in fair condition that generally require relatively minor preventive and corrective maintenance actions such as bearing repairs, joint repairs, zone and spot painting, and girder end repairs. The “fair–corrective” category includes bridges in fair condition that generally require moderate preventive and corrective maintenance actions such as bearing replacement, deck replacement, and major substructure repairs. Bridges characterized as “poor” generally require major rehabilitation or replacement.
The bridge project types developed for this research project map readily onto the four-condition category scheme (Table 7). Project types beginning with the numbers 1 or 2 (Bridge Replacement and Superstructure Replacement) are on bridges that are “poor.” Project Types in Groups 3, 4, and 5 (all are Major Rehab projects) indicate bridges that are “fair–corrective.” Project Types 6 to 10 indicate bridges that are “fair–protective.” Bridges needing only Cyclical Maintenance (Project Type 11) are considered to be in “good” condition.
Bridge Project Types and Corresponding Condition Categories
Poor = Red; Fair-Corrective = Orange; Fair-Protective = yellow; Good = Green.
Applying the condition category scheme to the selected project types for the 2019 NBIS bridge dataset gives the percentages of bridge counts and deck areas in each condition category, as shown in Figure 1.
Bridge network performance based on selected project type: all National Bridge Inspection Standards (NBIS) bridges in New York State ( 1 ).
Conclusions
During the research, NYSDOT project team members gained a better understanding of the analysis opportunities afforded by AASHTO element inspection data along with some of the particular data attributes that need to be accounted for when designing uses for the data.
NYSDOT has found that the granularity of the condition state rating system of AASHTO element inspection, which combines severity and extent (quantity) in a two-part measurement concept, proves to be most effective for treatment selection when the basis for selection is the quantities in poor and severe condition states rather than a condition index that is a weighted average of all of the condition state quantities. Relatively small portions of an element in Condition State 3 (poor) or 4 (severe) can suggest the need for repair or replacement, while the condition index can still be a high value.
The selection logic developed in the research project benefits from NYSDOT collecting data on a span-by-span basis. It adds an extra level of granularity to an already highly granular inspection rating system. Element treatment triggers, which are based on percentages in poor and severe condition states, take maximum advantage of the granularity of the AASHTO element rating system when applied to the more limited quantities in a span than to the quantities in an entire bridge.
NYSDOT found that the Stage 1 selected project types afford a promising concept for translating element inspection data into a bridge condition measure that can be used in expressing bridge network performance.
In addition to establishing logic to select element treatments and bridge project types, the research project produced a set of modeling assumptions (“action effectiveness models”) in relation to the effects of light and heavy treatments on the condition of each element after the treatment. These models will be used in the bridge management system to predict future system conditions resulting from applying a candidate project type. The models include transition probability matrices for light and heavy treatments that identify, for each element and condition state, the percentage change to lower-numbered (better) condition states. For example, a heavy repair of a steel closed box girder (Element 102) is modeled as transitioning 80% of the quantity in CS-4 into CS-1 and the remaining 20% into CS-2. A discussion of the treatment effect modeling and the full transition table of element effects can be found in the research report ( 1 )
Integration of AASHTO element data into the NYSDOT bridge management system also requires element-level deterioration modeling. At the time of writing this paper, a research project to create these models is in progress.
In summary, the research project has made considerable progress on integrating AASHTO element inspection data into the NYSDOT bridge management system. Models for element treatment and bridge project selection and for modeling the effects of treatments were developed. The research project also created logic (SQL queries) and detailed instructions for programmers to apply these models in the existing NYSDOT bridge database system. The database queries and instructions are contained in Appendix K of the research report ( 1 ).
Footnotes
Acknowledgements
James Flynn and Paul Campisi of NYSDOT and Colleen Domingo of Gannett Fleming made important contributions to the research, as did the following participants in an expert elicitation workshop held early in the study: Brian Kelly, Richard Marchione, Norman Duennebacke, Rick Hunkins, and Steve Wilcox of NYSDOT; and Peter Weykamp of Gannett Fleming.
Author Contributions
The authors confirm contribution to the paper as follows: study conception and design: S. Alampalli, K. Gager, T. Leckrone; data collection: K. Malarich, P. D. Thompson: analysis and interpretation of results: K. Gager, T. Leckrone, K. Malarich, P. D. Thompson; draft manuscript preparation: S. Alampalli, K. Gager, T. Leckrone, K. Malarich, P. D. Thompson. All authors reviewed the results and approved the final version of the manuscript.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research is funded in part by the State Planning and Research Program (SPR) funding from the Federal Highway Administration (SPR project C-15-04).
The opinions and findings reported in this paper are those of the authors and may not be that of the organization that the authors represent.