Risk-Based Management Guidelines for Scour at Bridges with Unknown Foundations (2007)

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.

NCHRP 24-25 Page 34 Phase II Final Report 3. QUANTIFYING RISK OF SCOUR FAILURE This section gives an overview of the HYRISK methodology and then moves on to discuss the annual probability of scour failure assumptions, the scour risk equation, and the lifetime risk of scour failure that are used in the Scour Risk Management Guidelines section. 3.1. HYRISK Background The available literature contains several methods for quantifying the risk of scour- related bridge failure (see Appendix A), but none of them were deemed complete. HYRISK is a well known model that has simple data requirements, and ranks bridges according to their expected annual loss due to scour (i.e. scour that induces failure or heavy damage). This was deemed the most complete method available. The risk rankings produced by the model, however, were not intended to place exact monetary values on scour losses. In other words, the probabilities of failure in HYRISK were assigned qualitative values based on expert opinions for ranking purposes. Thus, the original HYRISK model was not intended to estimate how much money should be spent on scour countermeasures to protect a bridge that is approaching the end of its design life (i.e. its provisional schedule for replacement). A later extension of the model improved the cost of failure assumptions and permitted the modeler to adjust the cost of failure and the probability of failure in the risk equation, and calculate a cost-benefit ratio for scour countermeasures. Thus, the extended HYRISK model was selected as the base for the risk equation used in the proposed guidelines. 3.2. Annual Probability of Scour Failure One modification to the HYRISK method relates to the probability of failure assumptions. Interviews (see Appendix C) with State transportation officials lead to an estimate of approximately 33 failures per year for the 25 States interviewed (i.e. 33 out of about 161,000 bridges). This suggests that the annual average probability of failure is

NCHRP 24-25 Page 35 Phase II Final Report 33/161,000 = 0.000205, or about 1 in 5,000 per year. Scaling this to all bridges over water (i.e. 379,788) yields almost 80 scour failures per year. Applying the original HYRISK method to all of the bridges over water in the NBI database (i.e. 356,378 bridges, as of the end of 2005; see also Appendices A, D) yields about 60,511 failures per year (i.e. the sum of the individual probabilities of failure). This corresponds to an annual average probability of failure of 0.17, which implies that about 1 in 6 bridges fail per year due to scour. These assumptions clearly do not correspond with experience and result in exaggerated risk factors. Again, this was not a problem within the context of the original HYRISK methodology because HYRISK was primarily used to prioritize bridges. However, when using risk to select a course of action (guidelines), it is important that risk be as accurate as possible in order to properly account for the costs and benefits of various management activities. For this reason, all of the original HYRISK failure probabilities have been scaled down to a level corresponding to the approximate number of failures (nation-wide) obtained from the State interviews (see Appendix D). The new probability assumptions are given in Table 12, many of which are orders of magnitude lower than the original HYRISK assumptions (see Appendix A). This table lists the annual probability of failure (PA) for different scour vulnerability and overtopping frequency ratings. These scour vulnerability and overtopping frequency ratings are obtained from Tables 13 and 14 using common NBI data items.

NCHRP 24-25 Page 36 Phase II Final Report Table 12 Annual Probability of Scour Failure Overtopping Frequency (from Table 13) Scour Vulnerability (from Table 14) Remote (R) Slight (S) Occasional (O) Frequent (F) (0) Failed 1 1 1 1 (1) Imminent failure 0.01 0.01 0.01 0.01 (2) Critical scour 0.005 0.006 0.008 0.009 (3) Serious scour 0.0011 0.0013 0.0016 0.002 (4) Advanced scour 0.0004 0.0005 0.0006 0.0007 (5) Minor scour 0.000007 0.000008 0.00004 0.00007 (6) Minor deterioration 0.00018 0.00025 0.0004 0.0005 (7) Good condition 0.00018 0.00025 0.0004 0.0005 (8) Very good condition 0.000004 0.000005 0.00002 0.00004 (9) Excellent condition 0.0000025 0.000003 0.000004 0.000007 Note that scour vulnerability is a surrogate for NBI item 113, and that overtopping frequency indicates how often this vulnerability is tested. The scour vulnerability is a function of substructure condition (NBI item 60) and channel protection (NBI item 61) ratings, while the overtopping frequency is an implied attribute of the waterway adequacy rating (NBI item 71). In other words, the overtopping frequency is a measure of a site’s likelihood of a scour event, and the HYRISK scour vulnerability is a measure of a bridge’s vulnerability to scour failure. Note also that small values for scour vulnerability (or NBI item 113) correspond to a high risk of scour-induced failure. Table 13 Bridge Overtopping Frequency versus NBI Items 26 and 71 Waterway Adequacy (NBI Item 71 Code) Functional Class: (NBI Item 26 Code) (0) (1) (2) (3) (4) (5) (6) (7) (8) (9) (N) Principal Arterials, Interstates (01, 11) O O O O S S S R N Freeways, Expressways (12) Other Principal Arterials (02, 14) Minor Arterials (06, 16) Major Collectors (07, 17) F O O O S S S R N Minor Collectors (08) Locals (09, 19) Br id ge C lo se d U nu se d F F O O O S S R N Key: N = Never; R = Remote (T > 100 yr); S = Slight (T = 11–100 yr); O = Occasional (T = 3–10 yr); F = Frequent (T < 3 yr)

NCHRP 24-25 Page 37 Phase II Final Report Table 14 Scour Vulnerability versus NBI Items 60 and 61 Substructure Condition (NBI Item 60 Code) Channel Protection (NBI Item 61 Code) (0 ) F ai le d (1 ) I m m in en t F ai lu re (2 ) C ri tic al C on di tio n (3 ) S er io us C on di tio n (4 ) P oo r C on di tio n (5 ) F ai r C on di tio n (6 ) S at is fa ct or y co nd iti on (7 ) G oo d C on di tio n (8 ) V er y G oo d C on di tio n (9 ) E xc el le nt C on di tio n (N ) N ot A pp lic ab le (0) Failure 0 0 0 0 0 0 0 0 0 0 0 (1) Failure 0 1 1 1 1 1 1 1 1 1 N (2) Near Collapse 0 1 2 2 2 2 2 2 2 2 N (3) Channel Migration 0 1 2 2 3 4 4 4 4 4 N (4) Undermined Bank 0 1 2 3 4 4 5 5 6 6 N (5) Eroded Bank 0 1 2 3 4 5 5 6 7 7 N (6) Bed Movement 0 1 2 3 4 5 6 6 7 7 N (7) Minor Drift 0 1 2 3 4 6 6 7 7 8 N (8) Stable Condition 0 1 2 3 4 6 7 7 8 8 N (9) No Deficiencies 0 1 2 3 4 7 7 8 8 9 N (N) Not Over Water 0 1 N N N N N N N N N The substructure condition code (NBI item 60) rates the general condition of a bridge’s foundation, which should include a qualitative evaluation of how much scour – if any – has been observed at the bridge. Likewise, the channel and channel protection condition code (NBI item 61) is a qualitative measure of the observed stability of the stream (related to long-term aggradation or degradation). In the HYRISK methodology these two codes were deemed the closest potential measures of a bridge’s vulnerability to scour failure. The NBI database at the end of 2005 has data for 297,796 bridges with known foundations. This selection excludes culverts and only includes bridges with known foundations that are over water (i.e. NBI item 113 ≠ “U” or “N” or “6”) and with no missing values for NBI items 26, 60, 61, 71, and 113. These bridges were selected for analysis

NCHRP 24-25 Page 38 Phase II Final Report because they have enough information to evaluate the relationship between the HYRISK scour vulnerability and NBI item 113. Figure 2 plots the relationship between the HYRISK scour vulnerability and the NBI item 113 in such a way that the size of the dot is directly proportional to the number of bridges that correspond to these integer values. This figure clearly shows that the relationship between the HYRISK scour vulnerability rating and the NBI scour evaluation is uncertain. This uncertainty results, in part, from using prediction variables (i.e. NBI items 60 and 61) that do not account for all the characteristics that influence a bridge site’s scour potential, and that do not explicitly predict the scour depth required to undermine the bridge’s foundation. A closer look at the selected NBI data, however, shows that there is a strong relationship between NBI item 113 and HYRISK’s scour vulnerability. Figure 3 shows the relationship between NBI item 113 and the average scour vulnerability value (i.e. the average scour vulnerability for each NBI item 113 rating) for bridges with known foundations. This figure shows that the HYRISK scour vulnerability for bridges with known foundations is consistent with NBI item 113, and thus is a useful predictor of a bridge’s annual probability of failure.

NCHRP 24-25 Page 39 Phase II Final Report 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 NBI Item 113 Code Sc ou r V ul ne ra bi lit y C od e Figure 2 HYRISK scour vulnerability versus NBI item 113

NCHRP 24-25 Page 40 Phase II Final Report y = 0.4472x + 3.403 R2 = 0.8582 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 0 1 2 3 4 5 6 7 8 9 NBI Item 113 Code A ve ra ge S co ur V ul ne ra bi lit y C od e Figure 3 Average HYRISK scour vulnerability versus NBI item 113 3.3. The Scour Risk Equation The HYRISK equation developed by GKY & Associates, Inc. (see Appendix A) for the annual risk of scour failure in these guidelines is given below. CostPKRisk AA ⋅⋅= (4) The terms in this equation are defined as follows.

NCHRP 24-25 Page 41 Phase II Final Report RiskA = annual risk of scour failure ($/year), K = risk adjustment factor based on foundation type and type of span based on NBI items and where available from more developed databases, foundation information, PA = annual probability of failure based on NBI items 26, 60, 61, 71, and 113 (see Table 12), Cost = total cost of failure ($, see Equation 2), The first thing to note about this equation is that it is the product of three main factors: the annual probability of failure (PA), the cost of failure (everything between the braces), and a risk adjustment factor (K). The risk adjustment factor permits downward risk adjustments based upon knowledge of the structural and/or foundation design. The equation for K is given below. 21KKK = (5) In this equation K1 is a bridge type factor based on NBI data, and K2 is a foundation type factor based on information, which may be obtained from State inventories but is not in the NBI database. The values presently recommended for K1 are 1.0 for spans less than 100 feet long and 0.67 for rigid continuous spans with lengths in excess of 100 feet. This factor adjusts to reflect the benefit of structural continuity which can compensate for loss of intermediate supports. The factors are subjective, based on a limited delpic survey and data developed in FHWA RD-85-107, Tolerable Movement Criteria for Highway Bridges (6). The influence of rigidity, type of structure, etc., has significant effects on the tolerable movement criteria, which may be defined as an increase in maximum stress to a point below yield, therefore precluding collapse.

NCHRP 24-25 Page 42 Phase II Final Report The values recommended for K2, given below, should be developed for both abutment and pier condition, selecting the largest value for the analysis. „ 1.0: unknown foundations or spread footings on erodible soil above scour depth with pier footing top visible or 1- to 2 ft below stream bed „ 0.8: pile foundations when length is unknown, are less than 19 ft, or are all-wood pile foundations „ 0.2: foundations on massive rock These factors are again subjective and should be revised or adjusted using local experience or further forensic studies. It should be noted that even structures supported by massive rock foundations may still incur damage due to inadequate waterway openings or other causes. Therefore, the risk adjustment factor cannot by definition be zero in a dollar- based risk analysis. 3.4. Lifetime Risk of Scour Failure The HYRISK extension (see Appendix A) demonstrates that the lifetime probability of failure (PL) can be related to the annual probability of failure (PA) and to the provisional remaining life of a bridge (L) as follows. ( )LAL PP −−= 11 (6) Once the lifetime probability of failure is known, the lifetime risk of scour failure (RiskL) can then be calculated by substituting PL for PA in the risk equation (Equation 4), as shown below. The lifetime risk of scour failure is an estimate of the monetary risk of failure during the provisional remaining life of the bridge. CostPKRisk LL ⋅⋅= (7)