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 13 Phase II Final Report 2. GENERAL APPROACH TO RISK MANAGEMENT This section describes a general approach for using risk to manage bridges with unknown foundations with respect to hazard-induced failure. The following discussion will introduce the steps required to quantify the risk of failure associated with a specific hazard, and then outline the steps needed to use risk to select an appropriate management plan. 2.1. Probability of Failure One of the first requirements for assessing the risk of bridge failure associated with a specific hazard is to quantify the probability of failure. The main idea of studying the occurrence of failure is to study both the occurrence of hazardous events and a bridge’s vulnerability to these occurrences. However, it should be apparent at this stage that the vulnerability of a bridge with an unknown foundation will be difficult to predict – hence the notion of probability. Probability of failure in this context refers to the likelihood of hazard- induced bridge failure within a specific range of time. For example, an annual probability of failure is the likelihood that a hazard-induced bridge failure will occur in any given year. The basic approach for quantifying a probability of failure involves the following steps: 1. Describe the uncertainty in the frequency and severity of a hazard. 2. Describe the uncertainty in a bridge’s vulnerability to this hazard. 3. Correlate these uncertainties with observed failures. 4. Calculate the probability of failure for multiple failure modes. 2.1.1. Hazardous Potential Natural hazards (e.g. scour, earthquakes, violent storms, etc.) are often difficult to forecast, but there are at least two factors used to describe the potential for hazards. The first factor is the relative severity of a possible hazardous event, or the magnitude of a hazardous event (e.g. an eight meter tsunami wave height). The second factor is the local frequency of occurrence, which describes how often an event of a specific magnitude occurs

NCHRP 24-25 Page 14 Phase II Final Report at a particular bridge location. These two factors are usually lumped in any given measure of the likelihood of a hazardous event. Thus, it is common to see geographic maps that show spatial contours of the likelihood of a hazard of a specific magnitude. For example, flood magnitudes are usually specified in reference to their expected return period (e.g. a 100-yr flood). Note that the magnitude of such a flood is site-specific. In other words, the 100-yr flood somewhere along a creek might entail a five meter rise in channel stage, whereas a 100-yr flood somewhere along a river downstream might entail a two meter rise in channel stage. Note also that the expected return period (e.g. 100-yr) is the inverse of the probability (i.e. P100 = 1 / [100 yr] = 0.01) that an event of this magnitude will occur in any given year. In other words, the probability of a 100-yr flood is actually an annual probability of occurrence that is site-specific and magnitude-specific. Thus, flood maps usually show contours representing different expected frequencies (e.g. 10-yr, 100-yr, etc.) overlaid on contours of elevation in the floodplain. 2.1.2. Vulnerability to Failure As noted earlier, bridges are generally designed to withstand the most common and frequent natural hazards, but budget and technology issues ultimately limit a bridge’s ability to withstand more severe (and usually infrequent) events. In other cases, vulnerability to failure may relate to an unforeseen change in wear or stress that was used to estimate the design life of the bridge. In either case, a bridge’s vulnerability to failure is divisible into two basic factors: the degree of stress or degradation that a bridge can safely withstand, and the corresponding severity of the hazardous event required to induce this degree of stress or degradation. In other words, the first factor is a description of a bridge’s specific mode of failure, while the second is a description of the type and degree of hazard required to induce this failure mode. Thus, it should be evident that any uncertainty about the foundation of a bridge also makes the bridge’s failure mode uncertain.

NCHRP 24-25 Page 15 Phase II Final Report For example, scour refers to sediment erosion that occurs in a stream or river that flows under a bridge. Scour may cause a bridge to fail when enough sediment erodes to undermine and collapse a pier, footing, or abutment. The amount of scour that will occur around a pier, footing, or abutment during a flow of a given magnitude is usually predicted using the methods described in the FHWA HEC-18 manual (2). This manual describes how to predict the depth of scour that may occur at a bridge site. There are three basic types of scour – long-term aggradation or degradation, contraction scour, and local scour – which may be induced by floods, droughts, or other phenomena that alter the fluvial sediment load. It is also worth noting that some scour holes are eventually refilled by natural sediment transport processes, which illustrates that some forms of degradation – and thus vulnerability – are temporary. Characterizing a bridge’s vulnerability to failure in this context implies that the mode of failure can be predicted – with reasonable certainty – using measured or inferred data. The FHWA HEC-18 manual for scour is perhaps the best example of using measured data to predict (and ultimately prevent) a site-specific mode of failure. Bridges with unknown foundations, however, lack information about the construction (i.e. form, depth, or geotechnical setting) of piers, footings, or abutments, which makes predicting their vulnerability to hazard-induced bridge failure more difficult. Furthermore, determining the substructure of an unknown foundation may be expensive; thus, it may be useful to relate what is known about the bridge and its setting to similar bridges in order to model and thereby predict the substructure. This predicted substructure could then be used to estimate the bridge’s vulnerability to failure and ultimately the bridge’s probability of failure. For example, if in any given year there are on average 2 scour failures out of 1,000 bridges with the same hazardous potential and vulnerability to failure, then the annual probability of failure for these bridges is 2/1000 or 0.002.

NCHRP 24-25 Page 16 Phase II Final Report 2.1.3. Correlations with Observed Failures If models already exist for predicting the hazardous potential of a site and the corresponding vulnerability of a bridge to fail, then these two quantities can be correlated with the number of observed hazard-induced bridge failures to generate a model of the probability of failure. The main reason for this correlation is to account for inherent uncertainties regarding the likelihood of a hazard and the corresponding vulnerability of failure for any bridge considered, which are often only qualitatively measured. The first step in this correlation is to collect the following information about each bridge that has failed: „ The nature and likelihood of hazard, and each site’s vulnerability to failure. „ The timing and magnitude of the event that ultimately caused bridge failure. Each known bridge failure then contributes to the estimate of the probability of failure for a given hazardous potential (HP) and vulnerability to failure (VF). Thus, for a given length of record the annual probability of failure for a given HP and VF is computed as the total number of failures for a given HP and VF divided by the total number of bridges that could have failed, divided by the number of years in the record. If this is done correctly, the sum of the annual probabilities of failure for all of the bridges that could have failed during the period of record should equal the total number of observed failures during the period of record divided by the number of years in the record. Note that it may also be prudent to scale these probabilities of failure up slightly to account for any uncertainties in the record. 2.1.4. Multiple Failure Modes The discussion thus far has focused on quantifying the probability of hazard-induced failure for a single mode or mechanism of inducing failure, which is the simplest case. Engineering experience, however, has demonstrated that multiple failure modes are often

NCHRP 24-25 Page 17 Phase II Final Report possible. If multiple failure modes are to be considered, the method for calculating the total probability of hazard-induced failure will depend on whether the individual failure modes are correlated or independent. The information in the following subsections is adapted from the National Institute of Standards and Technology’s Engineering Statistics Handbook (3). Independent or Competing Failure Modes The competing failure mode approach applies if the following three conditions apply. 1. Each failure mechanism leading to a particular type of failure (i.e. failure mode) proceeds independently of every other one, at least until a failure occurs. 2. The component fails when the first of all the competing failure mechanisms reaches a failure state. 3. Each of the k failure modes has a known life distribution model Fi(t). If these three conditions apply to the multiple failure modes, then the total probability of failure (Pt) can be calculated from each individual probability of failure (Pi) can be obtained from the following equation. ∏ = −−= k i it PP 1 )1(1 (1) An example of competing failure modes might be failure due to scour versus failure during an earthquake. Correlated Failure Modes There are several types of correlations between failure modes that are possible. The Engineering Statistics Handbook (3) offers three different ways to conceptualize correlated failure modes: modes acting in series, modes acting in parallel, or a combination of both. In the context of bridges with unknown foundations, an example of correlated failure modes might be an earthquake-induced collapse versus an earthquake-induced mass wasting event (i.e. a mud slide, avalanche, or dam failure).

NCHRP 24-25 Page 18 Phase II Final Report 2.2. Cost of Failure Regardless of methodology, certain general economic assumptions are necessary for computation of risk. These include commercial and non-commercial vehicle operating costs, passenger vehicle occupancy rates, the value of lost productivity and life, and bridge replacement costs. Thus, the total cost of a bridge failure is more than just the cost of constructing a new bridge. 2.2.1. Expenses per Mile for the Motor Carrier Industry The modified HYRISK methodology by Pearson et al. (4) contains an estimate of the average time that an average motorist might spend on a detour. Table 3 shows that the detour time varies according to ADT of the roadway. Table 3 Detour Duration versus ADT Average Daily Traffic (ADT) Detour Duration (days) ADT < 100 1095 100 ≤ ADT < 500 730 500 ≤ ADT < 1000 548 1000 ≤ ADT < 5000 365 ADT ≥ 5000 183 The cost of wear on the detour must be carefully weighed against the reduced wear on the closed roadway and bridge. If the wear on the detour will be significantly more than the comparative wear on the original road, then this may also be added to the cost of failure. Average expenses per mile for all types of for-hire truck transportation embracing truckload, less-than-truckload (LTL), and a wide range of specialized carriage were $1.78 in 2000 according to the Federal Highway Administration (5). Tables 4 through 6 provide addition metrics.

NCHRP 24-25 Page 19 Phase II Final Report Table 4 Occupancy per Vehicle Mile by Daily Trip Purpose Trip Purpose Mean Standard Error All personal vehicle trips 1.63 0.012 Work 1.14 0.007 Work-related 1.22 0.020 Family/personal 1.81 0.016 Church/school 1.76 0.084 Social/recreational 2.05 0.028 Other 2.02 0.130 1990 through 2000 and forecasts through 2005 Source: The 2001 National Household Travel Survey, daily trip file, U.S. Department of Transportation (www.bts.gov/publications/national household travel survey, accessed May 26, 2005) Table 5 Comparison of Total and Variable Costs per Mile Cost Category Automobiles Trucks Total per mile $0.45 $1.80 Driver costs -- $0.50 Total vehicle cost per mile $0.45 $1.30 Variable cost per mile $0.15 $0.43 Variable as % of total 33% 33% Source: Minnesota Department of Transportation (http://www.lrrb.org/pdf/200319.pdf, accessed May 26, 2005) Table 6 Values of Time Used in the Derivation of Road User Costs Vehicle Type Value of Time from MBC (1990 Dollars) Value of Time Adjusted (1998 Dollars using CPI) Small passenger car $9.75 $12.16 Medium/large passenger car $9.75 $12.16 Pickup/van $9.75 $12.16 Bus $10.64 $13.27 2-axle single unit truck $13.64 $17.01 3-axle single unit truck $16.28 $20.30 2-S2 semi truck $20.30 $25.32 3-S2 semi truck $22.53 $28.10 2-S1-2 semi truck $22.53 $28.10 3-S2-2 semi truck $22.53 $28.10 3-S2-4 semi truck $22.53 $28.10 Source: http://tti.tamu.edu/documents/407730.pdf, accessed on May 26, 2005, accessed May 26, 2005. The Highway Economic Requirements System (HERS) procedure for calculating travel time costs recognizes that the value of travel time differs between trips drivers take as part of their work (on-the-clock trips) and other trips. Time savings during on-the-clock trips are valued on the basis of savings to the employer. The savings include wages, fringe benefits, and for some types of trucks, vehicle cost and the inventory carrying costs of the cargo.

NCHRP 24-25 Page 20 Phase II Final Report Alternatively, off-the-clock time savings reflect the results of research examining choice situations (e.g., toll versus free route, speed, or housing location) that require choosing to save time versus money or safety. Table 7 shows estimated values of travel time. Table 7 Estimates of the Values of Travel Time Automobiles Trucks Travel Purpose Small Medium 4-Tire 6-Tire Business Travel Value per person* $21.20 $21.20 $21.20 $18.10 Average vehicle occupancy 1.43 1.43 1.43 1.05 Total business $31.55 $31.96 $32.47 $22.01 Personal Travel Value per person* $10.60 $10.60 $10.60 Average vehicle occupancy 1.67 1.67 1.67 Total personal $17.70 $17.70 $17.70 * 2000 Dollars Source: FHWA web site (http://isddc.dot.gov/olpfiles/fhwa/010617.pdf, accessed May 26, 2005) Statistics for Mean Hourly Wage Rate for each state are obtained from U.S. Department of Labor. It is also possible to obtain statistics for counties in each state from the same source. Value of time per individual for passenger car can also be calculated by multiplying mean hourly wage rate by 0.41. Table 8 shows a complete listing by State.

NCHRP 24-25 Page 21 Phase II Final Report Table 8 Values of Time State Mean Wage* ($/hour) Value of time† ($/hour) State Mean Wage* ($/hour) Value of time† ($/hour) Alabama 15.35 6.29 Montana 14.37 5.89 Alaska 20.27 8.31 Nebraska 15.89 6.51 Arizona 16.77 6.88 Nevada 16.49 6.76 Arkansas 14.21 5.83 New Hampshire 18.01 7.38 California 20.18 8.27 New Jersey 20.69 8.48 Colorado 19.14 7.85 New Mexico 15.87 6.51 Connecticut 21.35 8.75 New York 20.96 8.59 Delaware 18.77 7.70 North Carolina 16.40 6.72 District of Columbia 27.87 11.43 North Dakota 14.72 6.04 Florida 16.23 6.65 Ohio 17.26 7.08 Georgia 17.23 7.06 Oklahoma 14.97 6.14 Guam 13.20 5.41 Oregon 17.78 7.29 Hawaii 17.67 7.24 Pennsylvania 17.29 7.09 Idaho 15.76 6.46 Puerto Rico 10.61 4.35 Illinois 18.55 7.61 Rhode Island 18.38 7.54 Indiana 16.26 6.67 South Carolina 15.35 6.29 Iowa 15.38 6.31 South Dakota 13.98 5.73 Kansas 16.24 6.66 Tennessee 15.74 6.45 Kentucky 15.47 6.34 Texas 16.98 6.96 Louisiana 15.02 6.16 Utah 16.40 6.72 Maine 16.09 6.60 Vermont 16.66 6.83 Maryland 19.89 8.15 Virgin Islands 13.62 5.58 Massachusetts 21.78 8.93 Virginia 18.81 7.71 Michigan 19.03 7.80 Washington 19.65 8.06 Minnesota 19.15 7.85 West Virginia 14.65 6.01 Mississippi 13.77 5.65 Wisconsin 16.94 6.95 Missouri 16.57 6.79 Wyoming 15.63 6.41 * Source: http://www.bls.gov/oes/current/oessrcst.htm, accessed January 12, 2006. † Source: The value of time is assumed to be 41% of the mean wage as suggested by José A. Gómez-Ibáñez, William B. Tye, Clifford Winston, “Essays in Transportation Economics and Policy: A Handbook in Honor of John R. Meyer”, 1999. 2.2.2. Bridge Costs Table 9 provides estimates for bridge construction. These costs should be increased by about twenty percent for phased construction.

NCHRP 24-25 Page 22 Phase II Final Report Table 9 Cost of Bridge Construction Bridge Superstructure Type, Demolition Total Cost ($/ft2) Reinforced concrete flat slab; simple span 50-65* Reinforced concrete flat slab; continuous span 60-80* Steel deck/girder; simple span 62-75* Steel deck/girder; continuous span 70-90* Pre-stressed concrete deck/girder; simple span 50-70* Pre-stressed concrete deck/girder; continuous span 65-110* Post-tensioned, cast-in-place, concrete box girder cast on scaffolding; span length <= 240 ft 75-110 Steel Box Deck/Girders: Span range from 150 ft to 280 ft 76-120 For curvature add a 15 percent premium segmental concrete box girders; span range from 150 ft to 280 ft 80-110 Movable bridges; bascule spans & piers 900-1500 Demolition of Existing Bridges: Typical 9-15 Bascule spans & piers 63 * Increase the cost by twenty percent for phased construction. Source: http://www.dot.state.fl.us/structures/Manuals/LRFDSDG2002AugChap11.pdf, accessed May 26, 2005. The modified HYRISK methodology (4) suggests that the ADT influences how quickly a bridge will be replaced, which increases the total construction cost. Table 10 shows the suggested cost multipliers for different ADT levels. Table 10 Cost Multiplier for Early Replacement Average Daily Traffic (ADT) Cost Multiplier for Early Replacement ADT < 100 1.0 100 ≤ ADT < 500 1.1 500 ≤ ADT < 1000 1.25 1000 ≤ ADT < 5000 1.5 ADT ≥ 5000 2.0 2.2.3. Price Elasticity of Demand Elasticity is defined as the percentage change in consumption of a good caused by a one-percent change in its price or other characteristics such as travel time, or road capacity. If prices decline, generally travel increases as lower-value trips become more affordable, conversely if price increases traveler may choose to forego trips, chain trips together or shift to different mode, route or destination.

NCHRP 24-25 Page 23 Phase II Final Report A detailed summary of demand elasticity is given Appendix B. However, demand elasticity was not incorporated in this study because research indicates that demand elasticity is very low (on the order of 3%), which is comparable to the uncertainty in the other elements of total cost of failure. Elasticity also does not account for loss of consumer surplus. While demand for use of a roadway (as measured in ADT) may go down as travel costs increase with a detour, the reduction in demand would not represent the net savings in travel costs due to the reduced ADT. Those who hypothetically choose not to travel because of the increased travel costs associated with a detour would experience a loss of consumer surplus (a cost to them) since they would now choose an alternative (travel route, place of business, domicile, etc.) that is not preferable to the original route. If the alternative were preferable, the user would have implemented it before the bridge failed and the ADT would already be reduced. 2.2.4. Loss of Life In the “Plan of Action for Scour Critical Bridges” published by Idaho DOT in 2004 (see Appendix B) the assumed cost per fatality is $500,000. This value assignment is obviously subjective and could vary considerably based on both economic and sociological factors. The number of lives lost is assumed to vary depending on the ADT and functional classification (see Table 11). High-ADT crossings, interstates and principal arterials are assumed to have more potential fatalities.

NCHRP 24-25 Page 24 Phase II Final Report Table 11 Assumed Number of Lives Lost in Bridge Failure Average Daily Traffic (ADT) Number of Lives Lost ADT < 100 0 100 ≤ ADT < 500 1 500 ≤ ADT < 1000 2 1000 ≤ ADT < 5000 2 ADT ≥ 5000 (Not an interstate or arterial) 5 ADT ≥ 5000 (interstate or arterial) 10 2.2.5. HYRISK Cost of Failure Equation The extension to the HYRISK equation developed by GKY & Associates, Inc. (see Appendix A) provides a simple equation for calculating the total cost of bridge failure. The only addition to the equation considered here is the cost of fatalities (see Section 2.4.4, entitled “Loss of Life”). Price elasticity of demand was not added due to the reasons already stated. Thus, when the previous considerations are implemented, the equation for calculating the cost of failure is given in Equation 2. XC S DAdTCTOCDAdTCTCeWLCCost 654321 100100 1 100100 1 +⎥⎦ ⎤⎢⎣ ⎡ +⎟⎠ ⎞⎜⎝ ⎛ −+⎥⎦ ⎤⎢⎣ ⎡ +⎟⎠ ⎞⎜⎝ ⎛ −+= (2) The terms in this equation are defined as follows. Cost = total cost of bridge failure ($), C1 = unit rebuilding cost from Table 9 or use local data ($/ft2), e = cost multiplier for early replacement based on average daily traffic from Table 10, W = bridge width from NBI item 52 (ft), L = bridge length from NBI item 49 (ft), C2 = cost of running automobile from Table 5 (i.e. $0.45/mi),or use local data C3 = cost of running truck from Table 5 or use local data ($1.30 /mi), D = detour length from NBI item 19 (mi),

NCHRP 24-25 Page 25 Phase II Final Report A = average daily traffic (ADT) from NBI item 29, d = duration of detour based on ADT from Table 3 (days), C4 = value of time per adult in passenger car from Table 8 or use local data ($/hr), O = average occupancy rate from Table 4 or use local data (typically 1.63 adults), T = average daily truck traffic (ADTT) form NBI item 109 (% of ADT), C5 = value of time for truck from Table 6 or Table 7 or use local data ($22.01/hr), S = average detour speed (typically 40 mph), C6 = cost for each life lost (typically $500,000 or use local data), and X = number of deaths resulting from failure from Table 11 or use local data. Note that this equation is the sum of three basic concerns: the cost of reconstruction (i.e. the C1 term), two detour-related consumer costs (i.e. the C2, C3, C4, and C5 terms), and the potential cost of fatalities (the C6 term). Thus, this equation provides a template that is easily adjusted for local data and other concerns. 2.3. Risk of Failure Once the probability of failure and the cost of failure associated with a specific hazard are known (or estimated), the risk of failure is computed as the product of these two quantities. For example, if the annual probability of hazard-induced bridge failure is multiplied by the cost of bridge failure, then the risk of hazard-induced bridge failure will have the units: dollars per year. Thus, the annual risk of failure is only a fraction of the total cost of bridge failure because the occurrence of failure in any given year is uncertain.

NCHRP 24-25 Page 26 Phase II Final Report Estimating the risk of failure over longer periods of time (e.g. $ per decade) requires the use of probability theory. The proper way to adjust an annual risk to another length of time is to adjust the annual probability of failure and then multiply the adjusted probability of failure by the cost of failure. The following equation can be used to calculate the probability of failure over a specific period of time (PT, where T is the new period) from the annual probability of failure (PA). ( )TAT PP −−= 11 (3) Note that PT is the probability that at least one failure will occur in T years, which is greater than the annual probability of failure. Thus, the risk of at least one hazard-induced bridge failure in T years is computed as the product of PT and the total cost of bridge failure. This equation is useful for assessing the risk of failure over the remaining life of a bridge that has already been tentatively scheduled for replacement or retrofits (i.e. due to other operational concerns). 2.4. Mitigating Activities The former subsections dealt with quantifying the risk of a hazard-induced bridge failure. The main goal for developing these guidelines, however, is to reduce this risk in the most cost-effective manner. This necessarily entails listing the cost of any pertinent methods for mitigating a hazard-induced failure. Mitigating activities might include performing various forms of field reconnaissance to measure or infer any pertinent unknown foundation characteristics, increasing or changing the level or frequency of monitoring, or installing protective countermeasures or retrofits. Since field reconnaissance ultimately reduces the uncertainty in a bridge’s vulnerability to hazard-induced failure, the modeler should consider the cost of any geophysical method that can be used to determine the unknown foundation as well as the

NCHRP 24-25 Page 27 Phase II Final Report cost of any other attending analyses that are used to assess the bridge’s vulnerability. The cost of different monitoring or protective countermeasures, likewise, should include both installation cost and maintenance cost. The provisional schedule for replacing or closing the bridge should also be considered. 2.5. General Guidelines for Risk Management Once the risk of failure has been quantified and the various mitigating actions are known, the main task is to use the estimated risk, which is an uncertain measure, to select an appropriate course of action. Figure 1 presents a structured decision tree that uses pertinent aspects of the risk of failure to justify one or more of the mitigating actions already identified for managing risk associated with unknown foundations. The initial decisions in this figure primarily involve identifying bridges with foundations that might be determined easily – and thus analyzed like all other bridges with known foundations – or identifying bridges that are too important or potentially too vulnerable to delay action. Thus, the first step (“Can the foundation be inferred?”) is to search harder for foundation records that could be used to adequately determine the foundation, and effectively remove these from the population of unknown foundations. The following summarizes the pertinent findings from a careful literature review and interviews (see Appendices B–C) regarding common assumptions for unknown foundations. „ Older structures (built before 1960) were usually built on timber piling. „ Depth of piles can be assumed as at least 10 feet for bridges with unknown foundations. „ If rock is near the surface, spread foundations can be assumed to support bridges with unknown foundations. „ The top of a typical spread footing can be assumed to be 3 feet below the top of the soil and the bottom 7 feet below the top of the soil.

NCHRP 24-25 Page 28 Phase II Final Report If the foundation cannot be inferred, then any bridges that are deemed critical for emergency services or national security (i.e. “Is it a high priority structure?” in Figure 1) should benefit from the most aggressive mitigating activities. High priority structures are bridges that are so important that every possible effort should be made to determine the foundation and protect it as necessary. In other words, the ramifications of failure are so devastating that investment is warranted even if a cost-benefit analysis doesn’t justify such action. Each State Transportation Agency can set its own definition for these high priority structures, with the following suggestion provided herein: „ Principal arterials „ Evacuation routes „ Bridges that provide access to local emergency services such as hospitals „ Bridges that are defined as critical by a local emergency plan (e.g., bridges that enable immediate emergency response to disasters) Principal arterials have importance beyond the simple measure of ADT. Oftentimes these are critical economic links that have national economic importance. On a regional level, principal arterials are the major (and in some rural cases, only) link between towns, cities, and other developed areas. Failure of a principal arterial will affect far more than just the traffic that normally travels across the bridge. As traffic is rerouted, the traffic that normally travels the minor arterials and collector roads may be caught in severe delays resulting from extreme overcapacity. Evacuation routes are also suggested in this category since these routes are oftentimes the only practical means of evading natural disasters (e.g., hurricanes). The risk of injury and death – not from the bridge failure, but from the natural disaster - may be too great to bear if such a route is not available due to failure. Any bridges that are not high priority may still be an unacceptable hazard if it is in poor condition. Thus, the next step should be to estimate the risk of failure (i.e. “Calculate

NCHRP 24-25 Page 29 Phase II Final Report the risk of failure” in Figure 1), and establish minimum performance levels (MPL; i.e. maximum probabilities of failure) for different functional classifications (i.e. NBI item 26). Any of the remaining bridges with unknown foundations with an estimated probability of failure greater than its pertinent MPL (i.e. “Does the bridge meet the minimum performance level?” in the figure) should also receive the most aggressive management plan. The most aggressive management plan for high priority bridges or bridges that don’t meet their MPL arguably involves, at a minimum, the following steps: „ Perform foundation reconnaissance and any standard failure analyses to determine the risk of failure and consider any pertinent mitigating actions (e.g. countermeasures, or bridge replacement or closure). „ Use sound engineering judgment to select a mitigating plan of action, which may include replacing or closing the bridge. Any bridges that are not removed from the population of unknown foundations via the first two decisions outlined in Figure 1 should then be subjected to a structured benefit- cost analysis similar to the one outlined in the figure to select a risk management plan. The estimated risk can be used as a potential benefit that may justify the cost of implementing certain mitigating actions. It should be evident that the safest management plan for a bridge with an unknown foundation is to use foundation reconnaissance to determine the foundation before considering other mitigating actions, and that increased monitoring or the installation of countermeasures or retrofits without sufficient analysis are less safe but potentially helpful alternatives. It should also be evident that increased monitoring may reduce the risk of death if the bridge’s imminent failure is detected early enough to stop traffic prior to structural failure. Furthermore, installing a countermeasure or retrofit using sound engineering judgment and monitoring its effectiveness during significant

NCHRP 24-25 Page 30 Phase II Final Report events – but using methods short of using standard failure analyses to guide the installation – may be safer than relying on automated monitoring to predict imminent failure. Thus, Figure 1 first suggests that the cost of automated monitoring be compared to the risk of death (i.e. the product of the lifetime probability of failure and the estimated cost of death) to determine if automated monitoring is warranted (i.e. “Is automated monitoring warranted?” in the figure). If automated monitoring is warranted (e.g. risk > monitoring cost), then the risk of death can be neglected in the risk of failure that is used to determine if countermeasures or retrofits are warranted (i.e. “Are countermeasures/retrofits warranted?” in the figure). Countermeasures or retrofits are probably warranted if the risk of failure is greater than the estimated cost of a countermeasure or retrofit, in which case automated scour monitoring is probably not warranted. If countermeasures or retrofits are warranted, then the cost of foundation reconnaissance and standard failure analysis should be compared to the cost of the proposed countermeasure or retrofit to see if analyses are warranted (i.e. “Are foundation reconnaissance and standard analyses warranted?” in Figure 1). Foundation reconnaissance and standard failure analysis are probably warranted if the cost of foundation reconnaissance and standard failure analysis is less than half the cost of the countermeasure or retrofit (i.e. above “Are foundation reconnaissance and standard analyses warranted?” in the figure). Otherwise, it is probably most cost-effective to install countermeasures without the standard analysis and develop a bridge closure plan that includes monitoring the bridge’s vulnerability during several significant events (i.e. to the right of “Are foundation reconnaissance and standard analyses warranted?” in the figure).

NCHRP 24-25 Page 31 Phase II Final Report If monitoring is warranted instead of countermeasures or retrofits, then a bridge closure plan should be developed that involves monitoring the bridge for any signs of degradation or increased vulnerability (i.e. below and beside “Are countermeasures/retrofits warranted?” in Figure 1). If automated monitoring was warranted, then the vulnerability to failure should be monitored continuously. Otherwise, at a minimum, this monitoring (i.e. “Monitor failure mode(s).” in the figure) should be more intensive and perhaps more frequent than the standard biennial inspections. If this monitoring reveals a problem (i.e. “Is the vulnerability significantly increasing?” in the figure), then further mitigating activities are warranted.

NCHRP 24-25 Page 32 Phase II Final Report No Yes No No No Yes No Are retrofits/ countermeasures warranted? Are foundation reconnaissance and standard analyses warranted? Include risk of death in lifetime risk of failure calculation. Neglect risk of death in lifetime risk of failure calculation. Develop a bridge closure plan. 1. Install automated monitoring. 2. Develop a bridge closure plan. 1. Install countermeasures/retrofits without field reconnaissance or standard analysis, or close or replace the bridge. 2. Consider developing a bridge closure plan. 3. Monitor failure mode(s) during significant events. 1. Use field reconnaissance to determine the foundation. 2. Treat as a known foundation and perform standard analysis of failure mode(s). 3. Consider countermeasures/ retrofits, bridge replacement, or bridge closure. 1. Treat as a known foundation and perform standard analysis of failure mode(s). 2. Consider countermeasures, bridge replacement, or bridge closure. Is it a high priority structure? Does the bridge meet the minimum performance level? Is automated monitoring (AM) warranted? Look for foundation records (e.g. pile driving, test pile, or material quantity records). Can the foundation be inferred? Is the vulnerability significantly increasing? Yes Yes Yes Yes No Yes No Calculate risk of failure. Monitor failure mode(s). Yes No Was AM warranted? Figure 1 General risk management guidelines flow chart

NCHRP 24-25 Page 33 Phase II Final Report A detailed application of these guidelines is presented in Section 5, entitled “Scour Risk Management Guidelines. The next two sections, however, present the supporting analysis for the scour guidelines in a similar manner to the general approach to risk management.