Assessment of Retaining Wall Inventories for Geotechnical Asset Management

Transcription

1 TRB -0 Assessment of Retaining Wall Inventories for Geotechnical Asset Management Robert E. Kimmerling, P.E., L.E.G. (Corresponding Author) Principal PanGEO, Inc. Eastlake Avenue E, Suite B Seattle, WA - Phone: --00 FAX: rkimmerling@pangeoinc.com Paul D. Thompson 0 NE th Place Bellevue, WA 00 Phone: -- pdt@pdth.com Keywords: Geotechnical Asset Management, Retaining Walls, Transportation Asset Management Re-submitted: November, Word Count: Words + (x0) Figures + (x0) Tables = Word Equivalent

2 Kimmerling & Thompson Abstract With encouragement coming from federal transportation funding legislation on asset management, state departments of transportation are expanding the scope of assets that are included within their Transportation Asset Management Plans. Retaining walls represent a significant portion of the overall assets that comprise an agency s system of transportation corridors. However, use of conventional physical condition surveys in order to develop inventories of retaining walls presents a daunting task in terms of human resources and time required to follow the conventional bridge inspection program model for retaining wall asset management. An alternative to physical inventory and condition assessment is outlined for the purpose of system-wide retaining wall asset management based on the density of assets within specified subsets of transportation corridors. The alternative concept presented in this paper estimates system-wide quantities of retaining wall assets based on corridor subsets with similar density of walls per unit length of the corridor. The system-wide condition of retaining wall assets is then evaluated using a deterioration function based on age (i.e., life expectancy) of the asset. The resulting conclusions provide program managers with a tool for predicting system-wide rehabilitation, maintenance and replacement needs. The concept offers the flexibility of integrating with physical inventories and condition assessments with time, as well as incorporation of improved asset deterioration functions, or forecast curves.

3 Kimmerling & Thompson 0 INTRODUCTION Background The national surface transportation legislation under the Moving Ahead for Progress in the st Century Act (MAP-) emphasizes risk-based asset management for state departments of transportation. Earth retaining structures represent a significant portion of a department s infrastructure, but unlike bridges, statewide inventories of retaining structures have typically not been kept or maintained. Due to the large number of walls that exist as part of the surface transportation system, creating an inventory by physical, or even virtual (i.e., Google Earth Street View TM ) methods represents a task that requires significant allocation of human resources to accomplish. A further impediment to asset management for earth retaining structures is the difficulty of condition assessment. Unlike bridges, which permit visual inspection of at least exterior surfaces of interest, retaining walls effectively hide from view those components that are of most interest and value when evaluating the physical and structural condition of the structure. Condition assessment through non-destructive testing methods, such as used for asset management of pavements (e.g., falling weight deflectometer testing), is, at best, in its infancy for retaining walls. This paper presents an alternative approach to conventional asset management practices that use physical condition surveys and inventories. First, system-wide estimates are made of the quantity of retaining wall asset based on assumptions regarding the quantity, or density, of asset per mile within various corridor subsets within the system. Second, a system-wide evaluation of the overall condition, or health, of the assets is made using a deterioration, or forecast, curve. The result is a relatively simplistic and inexpensive conclusion that may be used for management of retaining wall assets at a programmatic, system-wide level. Purpose and Need for Retaining Wall Asset Management The MAP- Act calls on state transportation departments to prepare risk-based Transportation Asset Management Plans (TAMPs) for the National Highway System to improve or preserve the condition of the assets and the performance of the system. The legislation mandates the establishment of condition and performance targets for at least pavements and bridges, and requires the TAMP to include strategies leading to a program of projects that would make progress toward achievement of the targets. Although only pavements and bridges are mandatory in the TAMPs, states are encouraged to include all infrastructure assets within the right-of-way corridor in such plan ( USC (e)). Funding is also apportioned for the inspection and evaluation of other highway infrastructure assets on the National Highway System, including signs and sign structures, earth retaining walls, and drainage structures. ( USC (d)()(d)).

4 Kimmerling & Thompson 0 Retaining walls, like other physical features of the modern, constructed transportation system, have a finite life span. Over time, the asset experiences a loss of serviceability as a result of a deteriorated condition of its components or structural elements. As the condition of transportation infrastructure deteriorates, each agency responsible for the management of these assets will need tools to enable it to predict, at the programmatic and agency-wide level, the future needs for maintenance, rehabilitation and/or replacement. Conventional asset management, such as used for bridges, typically includes inventories or databases populated with information collected during physical condition surveys of individual assets. This approach has proven very effective for asset management of bridges and has led to a systematic approach to programming the maintenance, rehabilitation and/or replacement of bridge structures. However, the physical approach to condition assessment requires a sustained investment in the maintenance of the inventories combined with allocation of human resources to conduct the bridge inspections. Existing Approaches to Asset Management for Retaining Walls Asset management of earth retaining structures by means of physical inspection and inventory is well documented by Brutus and Tauber (). This report prepared for NCHRP Project -0 describes robust data collection methods and storage procedures for earth retaining structures that are similar to those used successfully for bridges for several decades. The process is time consuming for both data collection and data entry and management, with nearly 0 items of information recommended to be catalogued for each retaining wall included in the inventory. As such, the recommended process also relies on a significant allocation of human resources to implement the asset management strategy. An even more elaborate and comprehensive process is described in the Retaining Wall Inventory and Condition Assessment Program manual by FHWA (). SYSTEM-WIDE ASSET MANAGEMENT An Alternative to Physical Inventory The basic premise of the alternative approach to retaining wall asset management proposed here is that, for certain subsets of the highway network, there exists a relatively uniform density or number of retaining wall assets per unit length of the corridor. The number and type of subsets of the highway network could vary from state to state or agency to agency, depending on the predominant types of corridors under the entity s jurisdiction. For example, a predominantly rural state with a combination of both mountainous and flatland topography may choose to distinguish subsets of corridors on the basis that routes in the mountainous regions are likely to have more retaining wall assets per mile than the flatland

5 Kimmerling & Thompson 0 corridors, even though the routes may have otherwise similar characteristics in terms of average daily traffic (ADT) counts, corridor importance, etc. For explanation and example purposes, the following subsets of highway corridors are defined. For simplicity, a small number of subsets was chosen for this example. Topographical (i.e., mountainous versus flatland corridors) and other possible corridor subset distinctions have not been taken into consideration here. An agency employing this method should determine its own definitions and number of subsets based on the types and variability of corridors under its jurisdiction..) Urban, Limited Access Typically includes interstate freeways and other limited access corridors that run through congested urban or industrial areas where limited right of way has forced the use of retaining walls in lieu of side slopes for highway construction..) Rural, Limited Access Typically includes interstate and other limited access corridors in regions where the density of retaining wall assets is less as compared to the urban subset defined above..) Urban, Non-grade Separated Typically includes surface corridors in areas with limited right of way. Note that this subset may include corridors with varying numbers of lanes as the density of retaining wall assets is unlikely to be strongly a function of the width of the corridor (i.e., if walls were required on either side of the highway, the number of walls per mile is unlikely to be much different whether the channelization is lanes versus or more lanes)..) Rural, Non-grade Separated Typically distinguished from the urban subset based on the general ability to construct roadway with side slopes instead of retaining walls within the available right of way. Note that while the above highway corridor subset definitions do not distinguish subsets based on the number of lanes on the route, for the situation where the route is median and grade separated the density of retaining wall assets per mile will approximately double, since there exists the potential for walls on both the left and right sides of each of the travel directions. In this case the corridor should be split into two over the distance where the median and grade separation occurs for the purpose of estimating the density of retaining wall assets. Through GIS and other existing sources of data, transportation agencies already have a database of the route miles under their jurisdiction. Using this information, a database may be compiled that sorts all routes into the corridor subsets as defined above, or using agency defined subsets, as may be appropriate.

6 Kimmerling & Thompson A basis should be established for the quantity of retaining walls per mile in each inventory subset. This may be accomplished by collecting a statistical sampling of the number of retaining walls per mile for each of the various subsets using representative corridors and physical condition surveys. Next, the age of the assets should be established. This could be accomplished in a number of ways. One possibility would be to use an existing asset management system for the bridges within the same corridor and applying the assumption that the walls would be of the same vintage as the bridges within a given corridor. A more accurate, but also more labor intensive approach would be to use as-built plan information or construction history records for a sampling of individual retaining walls within a given corridor. An estimate of the quantity and age of all retaining wall assets within the agency s system may now be calculated and tabulated in a format similar to Table. The corridor subsets correspond to the corridor definitions provided above. TABLE Example Summary of System-wide Retaining Wall Assets In-service, by Decade Retaining Wall Assets in Service (ft of wall face) Corridor Subset Decade Total 0 0 0,000,000 0, ,000,000 0, ,000 0,000 0, ,000 0,000 0,000,000 0,000,000 00,000 0,000 0,000,00,000 0,000,000,000,000 00,000 0,000,00,000 0,000,000,00,000 0,000 0,000,0,000 0,000,000,000,000 00,000 0,000,00,000 00,000,000,00,000 00,000 0,000,0,000 Although Table is entirely hypothetical, it incorporates (in concept) some of the historical aspects of the way in which retaining wall assets have grown as part of the overall transportation asset management for an agency. First, it reflects the initial build-out of the interstate system during from the 0 s through the 0 s. Second, it captures the reduction in retaining wall assets on rural and non-grade separated corridors as these corridors are replaced with segments and spurs of limited access facilities. The values in Table are based on a number of assumptions by the authors and their own career experience. The values are likely more representative of a rural, western state in the U.S. with a transportation system constructed mostly within the twentieth century.

7 Kimmerling & Thompson Deterioration Curves for Retaining Walls With an inventory constructed as described in the preceding section, an estimate of age is available for all of the walls, but condition data may be very limited. But it is condition, rather than age, which drives maintenance, repair, and rehabilitation needs. If it is not possible to institute a routine comprehensive inspection program, an alternative might be to construct a relationship between age and condition, using a deterioration model. A simple deterioration curve based on a design life of 0 years and a logarithmic decay with time is shown in Figure. The condition axis is presented in percentage terms where 0% represents a new asset with perfect or nearly perfect serviceability and a full design life expectancy. A condition of 0% represents a completely failed asset with no useful service life remaining. In this application, a deterioration curve may also be thought of as a forecast curve when used as a predictor of asset condition at some point in the future. The condition of the retaining wall assets within a corridor, or for the entire system, may then be determined by applying the deterioration curve to the quantity of asset at a particular age. It should be noted that considerable opportunity exists for improving the accuracy of deterioration curves. Deterioration curves for retaining walls could potentially be made more reliable by consideration of the following factors. Recognition that certain wall types may be prevalent in corridor subsets based on the age of the corridor and the type(s) of wall systems employed during original construction. Actual mortality data may be available for such an occurrence, or, alternatively, expert opinion could be used to modify the deterioration function and / or life expectancy. Deterioration curves are also dependent on regional considerations, including factors such as quality of material used for wall backfill which can affect corrosion and wall component life expectancy. Similarly, cold weather and marine environments can potentially reduce the actual life expectancy for retaining walls as compared to design life assumptions used for engineering design purposes.

8 Kimmerling & Thompson Condition 0% 0% 0% 0% 0% 0% 0% 0% % % 0% Years in Service FIGURE Deterioration curve based on design life of 0-years. Applying Figure to the wall assets in Table, it can be seen that in this hypothetical example the oldest retaining wall assets on the system are approaching a loss of service which will require some action in order to keep all corridors on the system in full service. The action may include maintenance, rehabilitation or replacement of the assets. Of a more practical use for the program managers of an owner agency is the ability for the proposed process to be used as a forecasting tool. By applying the deterioration curve to future years, say, at the decade-by-decade level, a forecast of the system-wide needs for retaining wall maintenance, rehabilitation and / or replacement can be made. This approach accomplishes the goal of managing assets for the future rather than defaulting to a reactionary approach where retaining wall assets are dealt with only as they reach unacceptable levels of service (i.e, failure states or conditions). Table provides a summary of the system-wide condition of the hypothetical assets in Table for the year. Table breaks out the quantity of asset by age and then applies the deterioration function of Figure to determine the condition according to the age of the assets. The condition index multiplied by the square footage of asset indicates the quantity of asset that is expected to be in good health, according to the deterioration function (Figure ). Conversely, if the condition index (CI) subtracted from unity (i.e., CI) is multiplied by the square footage of asset a prediction is obtained of the quantity of asset expected to require some form of maintenance, rehabilitation or replacement in the year.

9 Kimmerling & Thompson TABLE System-wide Summary of Retaining Wall Asset Condition in Age of Retaining Wall Assets In (years) Asset Requiring Maintenance, Rehabilitation or Replacement (ft ) Total Area (ft ) Condition Index by Age of Asset ( CI) 0 0, ,000 0, ,00 0, ,00 0 0, ,0 0,0, ,0 0,00, ,000 0,00, ,000 0,00, ,000,00, ,000 Total:,0,00 A composite condition index for the entire system of retaining wall assets may also be obtained by determining the percentage of asset predicted by the deterioration function to still be in good condition relative to the total quantity of wall in service. In this example the systemwide condition index is indicated to be slightly greater than 0.0, or less than one-tenth of the wall assets over the entire system expected to need maintenance, rehabilitation or replacement in the year. Condition State Deterioration and Age One drawback of the deterministic deterioration model presented in Figure is the assumption that all walls of the same age are in the same condition. For many purposes this may be an oversimplification which may not produce the desired accuracy of future needs estimates. One way to relax this assumption is to borrow a technique from bridge management, adapted to the situation where periodic systemwide inspections are not possible. Nearly all state transportation agencies inspect bridge abutments by classifying them into (typically) four condition states, each state describing the severity of deterioration noted in a visual inspection. Many states inspect bridge and culvert wingwalls in the same way. Table shows typical condition state descriptions adapted from AASHTO (). TABLE Example Condition States for Concrete Abutments State Description Little or no deterioration Minor cracks and spalls Exposure of reinforcing steel but no significant corrosion Significant section loss of concrete or steel

10 Kimmerling & Thompson A single abutment or wall can be characterized by dividing its length or surface area among these four states. Any group of walls on a corridor or other subset of a network can also be described in this way. Since condition state inspections are very common in asset management, a variety of tools have been developed to make use of this information. Markov and Weibull deterioration models are especially common, as described in NCHRP Report (). With these models, future condition can be predicted as a probability of each condition state any number of years into the future. The median time to transition from each condition state to the next one is enough information to forecast condition as a function of asset age. For example, research in Florida based on years of bridge inspection data found the transition times shown in Table for concrete bridge and culvert wingwalls (). TABLE Example Wingwall Transition Times Transition Median years From state to state 0 From state to state From state to state These transition times were developed by counting the average fraction of walls that remain in the same condition from one year to the next, as opposed to deteriorating to the next-worse condition state. This fraction can be denoted p jj, the probability of transitioning from a given condition state j to the same state j. The median transition time can be computed from (): log(0.) t = log( ) p jj The probability of making a transition to the next-worse state can be approximated as (-p jj ). Using these relationships, a transition probability matrix can be constructed, showing the probability of each possible transition over one year. For wingwalls in the Florida study, the transition probability matrix is shown in Table. Forecasting of condition state probabilities for any number of years into the future is done by multiplying this matrix by itself: y k = x j p jj j Where y k is the fraction in state k one year later, x j is the initial fraction in state j, and p jk is the transition probability in one year from j to k. This multiplication can be repeated as many times as necessary to project into the future. Table shows the first and final years of a 0-year projection of wingwall condition using the Florida results.

11 Kimmerling & Thompson TABLE Transition probability matrix for wingwalls in Florida () State this year State next year State State State State TABLE Portion of a 0-year forecast of condition Year Forecast fraction in each condition state State State State State Based on the transition times in Table, the life expectancy of a concrete wingwall was estimated at years using the same methods as in NCHRP Report (). When condition states are converted to a condition index, the profile of age vs condition was found to be as in Figure (). This more precise model can be used in the same way as in the previous example. If comprehensive retaining wall inspection data cannot be obtained for a given corridor, but age is known, then the data in Table and the curve in Figure can be used to estimate corridor condition using the methods described in NCHRP Report (). If a sample of walls on the corridor are inspected, and if all walls on the corridor are believed to be of roughly the same age and condition, then an equivalent age can be derived from what is known of the corridor s condition using Figure. NCHRP Report () also provides a more precise mathematical method to compute the equivalent age. If each condition state is associated with a corresponding maintenance, repair, or rehabilitation action, then the quantities and costs of such work can be estimated using the estimates of inventory quantity described above, and typical unit costs.

12 Kimmerling & Thompson Condition index Age of element (years) FIGURE Age versus condition of a bridge wingwall. Limitations Since actual data regarding the deterioration of retaining walls is limited a theoretical approach based on the design life of retaining walls or typical condition state transition times may be used for needs estimation. This has to be treated as an approximation, and it is important to supplement this type of analysis using maintenance supervisor reports of hazardous wall conditions that this type of methodology cannot reliably discover. Corridors that contain retaining wall assets that are approaching the end of expected service life, or where maintenance supervisors have observed serious or recurring problems, may be targeted for physical inventory and condition assessment. In time, as real-world data become available regarding the mortality of retaining walls, deterioration curves may be revised and refined. The Florida research on bridge deterioration () found that about 00 pairs of inspections are required in order to develop a valid model, a target that can be met over a period of - years if a suitable sample of retaining walls is observed. For the time being, however, it should be acknowledged that deterioration functions for retaining walls will likely have an inherent range of error that is relatively broad. Also, the proposed approach for estimating wall assets on a corridor and system-wide level does not necessarily allow for distinguishing among various wall types and materials, which could strongly influence deterioration rates.

13 Kimmerling & Thompson 0 Most wall systems that are routinely deployed on transportation corridors are the result of relatively new innovations in earth retention engineering and therefore little is known about the actual life span of such systems. Mechanically stabilized earth (MSE) and soil nailing technologies are two examples of earth retention systems that are relatively new. Routine use of MSE retaining wall systems on transportation corridors dates from the 0 s, while soil nailing did not reach routine use until the 0 s (). The maximum age of these systems is therefore only about 0 and years, respectively. However, the target design life of these structures is generally in the range of to 0 years (,). Although infrequent, case histories exist in the literature that document wall failure where the age of the asset is considerably less than the design life (). When selecting a deterioration function for use with the alternative method proposed in this paper, it is important to consider whether such premature loss of service is indicative of a systemic condition that may apply elsewhere within a corridor subset or even system-wide. Since premature loss of service is generally wall-type specific, over-emphasis on a single occurrence that is not the result of a systemic condition (i.e., a wall failure that is anomalous or unique to a particular wall at a particular location) would lead to deterioration functions that over predict the quantity of wall assets that are nearing the need for maintenance, rehabilitation or replacement. It may be necessary therefore to employ theoretical deterioration curves that are based on life expectancy rather than observed rates of service loss. These curves should be replaced with deterioration functions based on actual retaining wall mortality as such data become available through physical condition surveys and research. Conclusions and Future Needs An alternative approach to retaining wall asset management is presented that is based on the density of assets per mile within a corridor or subset of the transportation network. This approach provides a simple, low initial investment method of estimating the system-wide retaining wall assets owned by an agency, and for estimating system-wide preservation needs. A deterioration model may be applied to the assets according to asset age in order to estimate a condition index. What is known about condition from a sample-based inspection program may be used to estimate an equivalent age as a starting point for predicting further deterioration. A composite condition index for the entire system may also be predicted. The approach described herein is not intended to be a complete substitute for conventional physical inventory and condition evaluations of retaining wall assets. A certain amount of ground truthing is required to validate the input into the asset model, including verification of the quantity of retaining wall asset per mile of each network subset or corridor. This can readily be accomplished using conventional asset management processes and procedures as discussed above and documented elsewhere (, ). Using the corridor subset concept, large portions of an entire transportation system could be modeled using the approach

14 Kimmerling & Thompson 0 described in this paper with validation on representative corridor subsets using physical inventory and condition state evaluation. The power of the alternative approach presented in this paper is in providing program managers with a method of allocating limited funding to the task of retaining wall asset management based not on knowledge of which specific assets require maintenance, rehabilitation or replacement, but on a reliable model that estimates the quantity of asset that will require attention in the near future over the entire system. Improvement is also needed in the determination and application of deterioration function(s) used to predict system-wide condition indices and retaining wall condition states. It is the author s opinion that this may best be accomplished at the agency level due to the variability in the historic practice of retaining wall type selection and usage within various jurisdictions. An example of this was the extensive use of steel and concrete crib walls in the state of California as a result of the availability of standard plans for these systems (). Used in many applications during construction of the interstate system by Caltrans, these wall types were largely supplanted by the emerging mechanically stabilized earth systems in the latter decades of the twentieth century. However, not all states were using the same standard plan designs for such systems during the same period of construction. Therefore the deterioration curves may differ from jurisdiction to jurisdiction for wall types that otherwise appear essentially identical. The quality of materials used to backfill retaining walls also varies from agency to agency based on the occurrence, or lack thereof, of natural deposits of high quality, durable, granular aggregates. With all other factors being equal, the use of less than ideal backfill material will likely reduce the life expectancy of a given wall system. In the interim, while statistical data is still being collected and analyzed regarding deterioration functions for retaining walls, expert opinion will likely prove valuable in addressing these issues. Although the focus of this paper has been on retaining wall assets for transportation corridors and systems, a logical extension of the concept of estimating assets according to density within a particular network subset or corridor can be applied to many other types of transportation assets, such as pavement striping, signage, illumination, and signalization. The next logical step would be to apply the proposed method to a representative portion of a transportation system. The size of the portion should be sufficiently large to avoid skewing of the results by anomalously large or small amounts or retaining wall assets within the portion of the system selected. It is expected that using a realistic data set the methodology proposed herein would be validated and potential improvements would be identified. Acknowledgment This research was funded in part by Alaska Department of Transportation & Public Facilities. The authors remain exclusively responsible for the material presented in this paper.

15 Kimmerling & Thompson References () Brutus, O. and Tauber, G. Guide to Asset Management of Earth Retaining Structures, Final Report NCHRP Project -0, Gandhi Engineering, Inc., New York, NY, October 0. () DeMarco, M., Keough, D., and Lewis, S. Retaining Wall Inventory and Assessment Program (WIP), National Parks Service Procedures Manual, Report No. FHWA-CFL/TD--00, Federal Highway Administration, Central Federal Lands Highway Division, Lakewood, CO, August. () AASHTO. Guide for Commonly-Recognized (CoRe) Structural Elements. American Association of State Highway and Transportation Officials, Washington, D.C.,. () Thompson, Paul D., Kevin M. Ford, Mohammad H.R. Arman, Samuel Labi, Kumares Sinha, and Arun Shirolé. Estimating Life Expectancies of Highway Assets. National Cooperative Highway Research Program Report, Transportation Research Board of the National Academies, Washington D.C.,. () Sobanjo, John O., and Paul D. Thompson. Enhancement of the FDOT s Project Level and Network Level Bridge Management Analysis Tools: Final Report. Florida Department of Transportation Contract BDK -0,. () O Rourke, T.D., and C.J.F.P. Jones. Overview of Earth Retention Systems: 0-0. Design and Performance of Earth Retaining Structures. American Society of Civil Engineers, Geotechnical Special Publication No., 0, pp. -. () AASHTO. Standard Specifications for Highway Bridges, th Edition. American Association of State Highway and Transportation Officials, Washington D.C., 0. () AASHTO. LRFD Bridge Design Specifications, th Edition. American Association of State Highway and Transportation Officials, Washington D.C., () Caltrans. Standard Plans, Plans C and C, State of California Department of Transportation, Sacramento, CA, January.