Use of the Alaska Geotechnical Asset Management Database in Evaluation of Retaining Structures After an Earthquake

AKDOT&PF GAM Program Retaining Wall Database

Work on cataloging the Department’s retaining wall assets began in 2014 with the development of the Retaining Wall Inventory (RWI), a system-wide baseline inventory capturing general information into data fields describing location, classification (category, type, function), and dimension characteristics ( 1 ).

To enable application of asset management activities (e.g., monitoring of deterioration) to structures in the RWI, researchers later progressed to developing a system for condition assessment ( 2 ). RWI structures selected for condition rating in the Retaining Wall Management Program (RWMP) component of the GAM program are evaluated for characteristics in categories of hazard, risk, and appearance. Algorithms are applied to resulting scores in select characteristic data fields, to derive condition indices, which are ultimately condensed to basic Good, Fair, or Poor condition state determinations ( 2 ). The aspirational goal is that the RWMP will ultimately assess and track the performance of all retaining walls deemed capable of affecting the roadway in the event of failure, with the exception of walls associated with bridges already accounted for in the Department’s Bridge Management System.

The retaining structure data compiled in the State’s GAM program is housed within geodatabases hosted within AKDOT&PF’s Transportation GIS (TGIS) ArcGIS Online (AGOL) website ( 3 ). Within the platform, data concerning retaining structures (and other geotechnical assets) can be extracted in a variety of formats, such as shape files, file geodatabases, or tabular datasets. The data layers are available to view online, embed into other websites, take offline into the field using mobile devices, or use in desktop GIS software ( 4 ). Permitted users may access other tools for data filtering and analysis within the online platform ( 2 , 4 ).

Field assessments for condition state determinations in the statewide road network have been conducted on a subset numbering approximately 400 retaining structures among the total retaining structure population (approximately 1,353) cataloged in the RWI. Fortuitously, for the efforts highlighted in this paper, this subset includes nearly all of the cataloged retaining structures along National Highway System (NHS) routes in southcentral Alaska, and numbers 287 structures within a 100 mi radius of the November 30, 2018, M_W 7.1 Anchorage earthquake.

Earthquake Summary

On November 30, 2018, an M_W 7.1 earthquake struck southcentral Alaska, a region where more than half of the state’s population resides. The event was an intraplate earthquake located within the Pacific plate subducting below the North American plate. The epicenter of the earthquake was about 20 km north of Anchorage and was over 40 km deep ( 5 ). The recorded peak ground accelerations (PGA) in parts of Anchorage registered at greater than 0.5 g, with the majority of the region near the earthquake at approximately 0.3 g ( 6 , 7 ). The 5% to 95% significant duration of the event varied from strong-motion station to strong-motion station and was generally greater than 10 s but less than 20 s. The damage from the earthquake was significant enough to declare local, state, and federal emergency orders, opening the opportunity for Federal Highway Administration (FHWA) and Federal Emergency Management Agency (FEMA) funding to support repair efforts.

Damage resulting from the earthquake was observed throughout the region and included numerous structural and geotechnical failures. Those observations have been included in several publications, including those by West et al., EERI, and Franke et al. ( 5 , 7 , 8 ). The geotechnical failures included liquefaction and lateral spreading, slope failures, consolidation of poorly compacted fills, embankment settlement, and others. While a majority of the transportation infrastructure remained operable, there were key infrastructure that were affected within the AKDOT&PF right of way, including bridge approaches, embankments, and retaining structures. Figure 1 shows two of the more famous failures related to the earthquake. The response by AKDOT&PF was to quickly patch and repair those infrastructure to allow for use by the traveling public, followed by studies to evaluate the impacts of the damage and design repairs that address the long-term performance criteria (i.e., remainder of an anticipated 75-year design life).

OPEN IN VIEWER

Figure 1.

Damage to roadway embankments as a result of the M_W 7.1 Anchorage earthquake: (a) Vine Road embankment failure over a deep peat deposit north of the earthquake and (b) Minnesota Drive off ramp at International Airport Road in Anchorage, failure of foundation soil below the embankment.

Use of the GAM Database

The first step in developing the designs for long-term repairs involved combining observations from numerous stakeholders from AKDOT&PF and other agencies. With the list combined, several groups performed cursory field site visits to each of the areas where damage was observed, and initial documentation, including photographs, were collected. From that first inspection a variety of sites were removed from the list because existing damage/condition indications at those particular sites were deemed to be unrelated to the earthquake. However, several sites, including those with damaged retaining structures, remained on the list for further evaluation. The remainder of this paper will focus on the sites with retaining structures.

Several resources were available to aid in identifying retaining structure damage related to the earthquake. For example, Google Maps Street View was a very effective tool to look at structures and historic pavement conditions visible along the roadway. Another very useful tool was the GAM retaining structure database. The database captures condition survey results, photographs, other notes, and, in some cases, as-built plans and reports related to the retaining structures. The GAM database information was reviewed before visiting the selected sites where there was potential damage related to the Anchorage earthquake. This gave the field team a starting point to develop their plans for making observations and collecting data.

Once field data was collected, the team was able to revisit the GAM database and compare current observations with those collected in the baseline development of the database. In some cases, it was apparent that the damage observed in the field was related to the earthquake. In other cases, the GAM database yielded evidence one way or another that could not have been gathered by Google Maps Street View or other resources. The following sections document the data collected in the field after the earthquake and how the GAM database proved useful. Lessons learned from this experience are also presented to identify what worked well and where improvements can be made.

Retaining Wall Field Observations

In all, approximately 12 retaining structures belonging to AKDOT&PF were screened and evaluated by Golder for earthquake damage as a result of initial field observations of potential damage immediately following the earthquake by AKDOT&PF staff. One of these walls was not documented previously in the GAM database. After initial screening, the original list of 12 retaining structures was reduced to nine structures to target for a second round of evaluations, which included observations and measurements pertaining to plumbness of wall face, joint gap size, cracking, displacements and distortions, and other possible deficiencies. Recorded measurements and observations were then compared with the available as-built plans. Photographs of the walls were annotated to include locations and records of the observations during the post-earthquake evaluations (Figure 2). The meaning of the annotations is recorded in the reports that are included with the photographs in the GAM database, such that the developed nomenclature may be used in future evaluations. Seven of the 12 retaining structures consisted of mechanically stabilized earth (MSE) walls. There were also two cast-in-place reinforced concrete cantilever walls, one bin wall, and two gabion basket gravity walls as well. Examples of the observations and comparisons with the GAM database at four of the 12 retaining wall sites are presented below.

OPEN IN VIEWER

Figure 2.

Example of annotated photograph of Fire Lake gabion wall post-earthquake.

Spenard Road Wall

A cast-in-place concrete cantilever retaining wall was observed to have displacements at joints and potential rotation, as observed after the earthquake. The retaining wall supports residential yards and structures above Spenard Road. In this case, both Google Maps Street View and the GAM database were used to compare field observations with historic data. The field visit included measuring wall batter, joint displacement, and observations of the asphalt condition at the toe of the wall. The wall was found to have an average batter between 88 and 89.5 degrees.

Based on observations from historic imagery (Google Maps Street View) and the GAM database, it appears that the rotation and displacement had occurred before the earthquake. The retaining wall had a Fair rating and a batter of 89.7 degrees was noted in the inspection report dated August 5, 2016. A review of the design drawings called for a batter of 2 degrees. The recent field observations also did not appear to suggest that the retaining wall had experienced additional movement as a result of the earthquake. As seen in Figure 3, the joint indicating rotation of the wall had been painted as well, suggesting prior movement. This wall was removed from the list of walls on the post-earthquake repair roster.

OPEN IN VIEWER

Figure 3.

Wall rotation at expansion joint: (a) rotated panel seen at a distance and (b) in-line with the wall to see the rotation more clearly.

Fire Lake Gabion Wall

The Fire Lake gabion wall is a six-course gabion wall located on the east side of the Old Glenn Highway at approximately milepost 2.42. During the post-earthquake evaluation of the wall, broken and deformed baskets were observed along most of the length of the wall, which were also noted during the 2016 condition assessment for the GAM program. Comparisons between photos collected during the 2016 GAM program assessment and post-earthquake observations indicated that the bottom row of baskets had experienced additional deformation since the original assessment in 2016 (Figure 4).

OPEN IN VIEWER

Figure 4.

Broken and bulging gabion baskets in bottom row of Fire Lake gabion wall: (a) from 2016 geotechnical asset management (GAM) program review and (b) from 2019 post-earthquake evaluation.

A bulge was also observed in the upper half of the wall at approximately 85 ft from the north end of the wall. The peak of the bulge was visible in photos collected in 2016, indicating that the bulge in the wall was present before the earthquake (Figure 5). A 3 in. gap was also noted post-earthquake between the wall and the retained soil from approximately 70 to 100 ft from the north end of the wall. Longitudinal cracks and settlement were also observed upslope of the wall. The gap behind the wall and the cracks and settlement upslope of the wall were not documented in the GAM database. No photos were available in the GAM database of the slope behind the wall for comparison. Conversations with local residents indicated that the longitudinal cracks and visible settlement upslope of the wall resulted from the earthquake.

OPEN IN VIEWER

Figure 5.

Fire Lake gabion wall viewed east: (a) from 2016 geotechnical asset management GAM program condition rating assessment and (b) from 2019 post-earthquake evaluation.

The wall batter was also measured every 10 ft along the length of the wall and ranged from 89.6 degrees (0.7%) to 79.5 degrees (18.6%) with an average batter of 87.1 degrees (5%). The wall batter was not recorded in the GAM database; however, comparisons with as-built plans indicated that the wall is more vertical on average than the specified batter of 84.3 degrees (10%) in the as-builts. Because of limited information, it was not possible to determine if the steeper batter had resulted from the earthquake.

Elmore Road MSE Walls at Campbell Creek

Two MSE walls are located on the east and west side of Elmore road for the Campbell Creek Trail Pedestrian Tunnel passing under Elmore Road. The wall system consists of precast concrete panels with reinforcing strips. During the post-earthquake review, the columns of panels were numbered from north to south. Panels in each of the columns were then assigned letters from top to bottom, with the top panels assigned the letter A. Annotations to photos of the MSE wall post-earthquake included distance from the north end, measured spacing between panels, panel numbers, and plumbness (Figure 6).

OPEN IN VIEWER

Figure 6.

Annotated photograph of east mechanically stabilized earth (MSE) wall post-earthquake.

Post-earthquake observations included several instances of cracking of the coping at the top of the wall and irregular panel spacing, which is consistent with observations documented in the 2016 GAM condition rating assessment. Additionally, out-of-plumb panels had a maximum batter of 87.6 degrees post-earthquake, which was consistent with the batter recorded in the GAM database pre-earthquake of 87.3 degrees for out-of-plumb panels.

Panel spacing of up to 2.5 in. was measured along the west wall post-earthquake, which was greater than the typical spacing of approximately 1 in. at this site. Because of the larger spacing, some aggregate was observed to be leaking through the panels during the post-earthquake review. The larger spacing and leaking aggregate was not noted in the GAM database; however, the original construction shims were still in place during the post-earthquake evaluation indicating that similar spacing existed during construction. Based on comparisons with the GAM database, the irregularities observed during the post-earthquake evaluation were not related to the 2018 earthquake.

International Airport Road and Minnesota Interchange MSE Wall

The MSE wall is located on the north side of the International Airport Road and Minnesota Interchange Northwest Ramp. The wall is approximately 1,300 ft long with a maximum height of 25 ft, according to the GAM database. The wall system consists of precast concrete panels with galvanized steel reinforcing strips. The post-earthquake evaluation was only conducted along a portion of the wall, near the locations of the longitudinal cracks in the roadway that resulted from earthquake. A series of 40 columns of panels were numbered from east to west between 305 and 501 ft from the east end of the wall to facilitate recording the locations of observations. Panels in each of the panel columns were assigned letters from top to bottom, with the top panels assigned the letter A. Observation locations were then identified by panel (Figure 7).

OPEN IN VIEWER

Figure 7.

Annotated photograph of mechanically stabilized earth (MSE) wall post-earthquake.

The MSE wall panels appeared to be in the same condition post-earthquake as documented during the GAM program survey, with large gaps between panels, and chips and minor cracks in some of the panels. The batter was measured at each column of panels (approximately every 5 ft) post-earthquake and ranged from 88.2 degrees to 90 degrees. This range was consistent with the recorded batter (88.7 degrees) in the GAM database; however, the location of the batter measurement noted in the GAM database is unknown. Observations shortly after the earthquake indicated translation of the wall. Significant settlement and uplift were observed near the toe of the wall over the course of multiple site visits post-earthquake. Additionally, the settlement was visible in the coping at the top of the wall, and freshly broken pieces of coping were observed at the base of the wall within the same zone (Figure 7). There were little to no photos of the toe of the wall or the roadway available for comparison in the GAM database; however, the observed settlement and cracked coping were consistent with the locations of the longitudinal cracks that formed as a result of the earthquake.

Lessons from the Field

During the post-earthquake evaluations of the retaining walls, photographs were taken of the entire length of the wall face and annotated to include locations, measurements (e.g., height, plumbness, gaps), and observations of displacement or damage to the walls. The annotated photographs proved useful for evaluating the condition of the wall by comparing the observed damage (e.g., cracks, settlement) with previously collected measurements within the database (e.g., batter and gap width).

While post-event photographs and measurements were useful during the post-earthquake evaluation, they were difficult to compare with the baseline photographs, observations, and measurements contained in the GAM database in some instances because the locations at which the data was collected were not well-recorded in the database. For example, at the International Airport Road Interchange MSE wall, discussed previously, several observations have been made. One of the photos from the database is shown in Figure 8 and indicates that panel corner chipping had been observed before the earthquake. Proper and sufficient documentation attached to the photos in the GAM database would have facilitated easer identification of panels with prior damage.

OPEN IN VIEWER

Figure 8.

Geotechnical asset management (GAM) database photo (August 2016) of damaged mechanically stabilized earth (MSE) wall panel (AKDOT&PF).

Additionally, in the post-earthquake evaluations, photographs were taken of the overall sites, including the toes and tops of the walls, where possible. These photographs allowed for better comparison with Google Maps Street View and Google Earth. The earthquake inspection team also attempted to collect photographs of the retaining walls at similar angles to the photographs included in the GAM database to allow for further clarity during comparison. As mentioned previously, having the GAM database available before performing the field work allowed the team to focus efforts on evaluating earthquake damage, rather than developing an understanding of the site, wall type, extents, and other information. Photographs included in the database supported the development of this framework.

There have been some benefits to the GAM database because of the additional data collection related to the earthquake reconnaissance and repair work. Another snapshot in time can be drawn on and utilized for evaluation of changes over time. More detailed evaluation of several walls can be included in the database. Also, efforts to collect as-built and other time-of-construction data have been collected and can be included in the database for future use.

Some aspects that could be improved in the collection and presentation of information in the database include:

It is acknowledged that each of these elements that help the asset management approach move toward a more complete picture of how sites change over time require at least two elements: one being initial resources to collect and catalogue the information, and another being repeated measurements to understand how conditions are changing over time.

It has been attempted here to suggest tips that may be useful in routine GAM inspections, as well in potential evaluations related to performance assessed after an extreme event such as an earthquake.

Conclusions

The AKDOT&PF GAM database was a useful tool in the evaluation of retaining structures and the damage that occurred as a result of the November 2018 Anchorage earthquake. The GAM database was a timely source of specific and succinct information useful in planning response and assessment by AKDOT&PF in the wake of the event. The asset owner was able to quickly access and use concise information about the location and characteristics of retaining structures in the earthquake impact area as well as the pre-existing condition of those structures.

Using the GAM database facilitated the discernment of damage attributable to the earthquake. The database provided information (i.e., photographs, rated condition, batter and height measurements, recorded observations) that saved hours of evaluation and provided further assurance when engineers had to define whether damage was related to the earthquake or not. Thus, this dataset directly facilitated proper and defensible qualification for Federal relief funds for repairs.

Maintaining inventory and tracking the condition and performance of retaining structures and other geotechnical assets yields benefits outside of standard asset management purposes such as minimizing life cycle costs. Notably, the asset management process does not typically adequately account for unpredictable adverse events such as earthquakes. It is therefore acknowledged that, although the GAM database was not explicitly developed for emergency response application, the benefits extend further, as illustrated in the study described here.

The measures recommended in this study could enhance the use and value of the Alaska GAM database. These include, for example, enhancing photo record variety and documentation, numerous direct measurements at duplicative locations, and comprehensive inclusion of historical documentation (e.g., as-built records).

As a foundational step in the asset management process, asset inventory maintenance and condition assessment needs to be an ongoing activity. A one-time baseline assessment of condition loses value as time elapses during a scenario in which follow-up assessments are not undertaken.