Official US Government Icon

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure Site Icon

Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

The latest general information on the Coronavirus Disease 2019 (COVID-19) is available on Coronavirus.gov. For USDOT specific COVID-19 resources, please visit our page.

United States Department of Transportation United States Department of Transportation
OFFICE OF RESEARCH, DEVELOPMENT, AND TECHNOLOGY AT THE TURNER-FAIRBANK HIGHWAY RESEARCH CENTER

Bridge Deck Condition Rating Forecast—Base Models

Base models are deterministic statistical models that are easy to understand and implement. With base models, historical time duration for each deck condition rating of selected types of bridges is calculated from training subsets of the National Bridge Inventory (NBI) (Federal Highway Administration (FHWA), 1995) data. These historical time durations are applied to bridges of the same type to forecast their deck condition rating. To date, base models have been developed for cast-in-place (CIP) concrete decks of three types of bridge superstructures: steel girder, prestressed concrete girder, and concrete slab. In the future, more factors, such as climatic zones and route classification, will be incorporated when assembling the training datasets to improve the performance of the models. The following is a step-by-step description of the modeling approach.

Step 1, Bridge Grouping

Bridges are grouped based on two items in the NBI database: deck structure type (NBI item 107) and main structure type (NBI items 43A and 43B). The inventory data show that more than 75 percent of all bridges have CIP concrete decks. Steel girder, prestressed concrete girder, and concrete slab bridges are the top three main structure types, and models are available for these main structural types using CIP concrete decks. The NBI query criteria and the corresponding number of bridges for each group are shown in Table 1.

Table 1. Bridge Grouping Criteria and Number of Bridges.

Concrete CIP (NBI Item 107 = 1)

Number of Bridges

(Based on 2018 NBI Data)

 

NBI Item 43A

NBI Item 43B

Training

(Defined in Step 2)

Total

Steel Girder

3 or 4

2 or 3

11,892

118,851

Prestressed Concrete Girder

5 or 6

2 or 3

5,453

65,225

Concrete Slab

1 or 2

1

8,483

55,907

Prestressed Box

5 or 6

5 or 6

1,759

30,333

Prestressed Slab

5 or 6

1

405

2,582

Prestressed T-Channel

5 or 6

4 or 2

1,472

6,196

Concrete Girder

1 or 2

2 or 3

3,711

9,639

Concrete Box

1 or 2

5 or 6

914

6,778

Concrete T-Channel

1 or 2

4 or 2

4,646

25,996

Step 2, Training Data Preparation

Bridges that satisfy the following requirements are included in the model training datasets:

  1. At least 30 years of NBI data.
  2. No increase in deck condition rating.
  3. No historical data discontinuity.

The number of bridges in each training dataset is shown in Table 1.

Step 3, Computing the Time-in-Condition Statistics from the Training Datasets

Statistics, including lower bound, upper bound, mean, and median of the time-in-condition for each condition rating and bridge group are computed from that group’s training dataset. The lower bound and upper bound of time-in-condition are defined as the 25th percentile and the 75th percentile of the dataset, respectively. A percentile is a measure used in statistics indicating that the value or values fall lower than a given percentage of observations. For example, at the 25th percentile, the observed values fall lower than 25 percent.

Figure 1 depicts the histogram of the steel girder bridges training group when their deck condition rating is 7. The lower bound, upper bound, mean, and median values computed from the training dataset are shown as vertical lines in the figure. Statistics of time-in-condition are computed for each condition rating of each group generating the values that are needed for the next step.

Figure 1. shows the time-in-condition histogram for cast-in-place concrete decks of steel girder bridges when Condition Rating is 7. The horizontal axis shows the number of years of Time-in-Condition, which ranges from 0 to 40 and is divided into 20 bins. The vertical axis shows the number of bridges, which has the range between 0 and 800 with the step of 100. The Lower Bound and Upper Bound are shown as red vertical lines on the histogram plot. They are close to 6 and 26 respectively. The Median and Mean are presented as green vertical lines on the histogram too; their values are close to 15 and 16.

Source: FHWA.

Figure 1. The Histogram of Condition Rating 7 of CIP Concrete Decks of Steel Girder Bridges.

Step 4, Creating Condition Rating Life Cycles for Each Bridge Group

The statistics computed in the previous step are used to create deck life-cycle predictions for each bridge group. The deck life-cycle prediction stops at a condition rating of 3. The assumption is that reaching a condition rating of 3 will trigger rehabilitation or replacement projects, and consequently, forecasting beyond that level would be futile. In addition, data at a condition-rating data of 3 or lower are sparse and would not result in reliable forecasts.

Figure 2 presents the life cycles for CIP concrete decks of steel girder bridges. The lower bound, mean, median, and upper bound values of the life cycles are shown as stepped curves.

Figure 2. shows the life cycles for cast in place concrete decks of steel girder bridges. The Upper Bound, Lower Bound, Mean, and Media of the life cycle of the bridges are shown as four stepped curves on the plot. All starts are condition rating 9 and ends at condition rating 3. The horizontal axis shows the year from 0 to 120 with the step of 10. The vertical axis shows the Condition Ratings from 3 to 9 with the step of 1.

Source: FHWA.

Figure 2. The Life Cycles for CIP Concrete Decks of Steel Girder Bridges.

Step 5, Condition Forecast for Each Bridge Deck

The life-cycle curves developed in Step 4 are used in conjunction with the historical condition ratings of a deck within its bridge group to develop condition forecast curves for each bridge deck. Four condition forecast curves are created for each deck representing the lower bound, median, mean, and upper bound of condition ratings. Figure 3 presents an example of the developed condition forecasting curves and how they appear on the LTBP InfoBridge™ (FHWA, 2019) website.

Figure 3. presents an example of deck condition forecasting curves and how they appear on the LTBP InfoBridge™. The horizontal axis shows the year between 1992 and 2070. The vertical axis shows Condition Ratings between 3 and 9 with the step of 1. In the plot, historical data are shown in blue. Four stepped curves representing upper bound, lower bound, mean, and median of the condition forecast are shown in green. The forecast start year is depicted to be a vertical line on the plot.

Source: FHWA.

Figure 3. An Example of Deck Condition Forecasting Curves.

References

  1. Federal Highway Administration. (1995). Recording and Coding Guide for the Structure Inventory and Appraisal of the Nation’s Bridges, , Washington, DC, obtained from: https://www.fhwa.dot.gov/bridge/mtguide.pdf, last accessed March 18, 2020.
     
  2. FHWA. (2019). Deck Condition Forecasting Curves, InfoBridge, obtained from: https://infobridge.fhwa.dot.gov/, last accessed March 18, 2020.

 

 

Last updated: Tuesday, August 24, 2021