Slide2: Limit Equilibrium Method Analysis of a Reinforced Soil Structure Failure
On March 12th, 2015 a catastrophic failure of a 73-meter tall, reinforced soil structure (RSS) occurred at the end of Runway 5 of the Yeager Airport in Charleston, West Virginia, only 8 years after it was constructed.
Engineers were brought in for a post-failure investigation to determine the cause of this failure. The purpose was to not only to identify the cause of the failure but also to determine if the failure could be accurately predicted with Limit Equilibrium (LE) analysis using the original design plans, pre-failure analysis and the samples collected from the post-failure site.
Construction of the RSS
To comply with updated Federal Aviation Administration design requirements, Runway 5 of the Yeager Airport needed to be extended in length by approximately 160 meters. Due to the terrain and high elevation of the airport, a RSS was needed to be able to extend the runway. Construction started in August of 2005 and was finished by December of 2006.
29 soil boreholes were taken from the area of construction and 8 borings to the north east. The area to the north east was where the fill material would be gathered from. Prior to construction, LE analysis was conducted to analyze the design of the RSS.
During construction, the lower third of the RSS had vertical spacing of 46 cm and above that, vertical spacing increased to 91cm. Reinforced panel lengths ranged between 8m and 53m with a length to slope height ratio of 0.70 or greater and two woven polyester geogrids, Mirafi 10XT and 20XT were selected.
Three categories of fill material were used:
- Select Fill – Sandstone (max particle size of 0.15m). Used to mitigate installation damage.
- General Fill – Mainly clean sandstone shotrock. Clay was incorporated near face to provide a soil medium for vegetation to grow on the slope face.
- Retained Fill – Outside the perimeter of RSS. Mostly cobble and boulder-sized sandstone shotrock.
Failure of the RSS
From its completion in 2006, there were no observable changes until 2013 when minor surface sloughing occurred near the base of the RSS. The sloughing was determined to be located in the external rock-clay mixture which didn’t significantly change the slope’s geometry. A tension crack was also observed at the back of the RSS zone, suggesting a slip surface was developing. Initial investigation was later expanded due to the slopes progression and by 2015 borings were done to obtain soil and rock samples and to install piezometers and inclinometers.
By Feb 2015, the crest of the RSS had settled 23cm and appeared to be increasing in movement. By March 10th, engineers were on site to investigate. In the week prior to their arrival, a head scarp 28 m behind the crest of the slope had formed and by the 10th, the head scarp was approximately 1.8 m wide. There was also a period of heavy rainfall around this time where surface water drained into the RSS through the head scarp. Surface sloughing of 3.7m was found at the base of the slope and at this point properties below the slope were evacuated. The RSS failed two days later on March 12th, 2015. Samples were collected from the failed mass to be analyzed and the remaining structure was removed to stabilize the slope for further investigation.
Limit Equilibrium Analysis
After the failure, 2D LE analysis was conducted to evaluate the stability of the RSS both at its 2006 completion date and failure in 2015.
The information needed for analysis was determined based on initial data from construction and from tests done on collected samples from the failure. The parameters used for analysis are listed in table 1.
Table 1
It is important to note that the shear strengths were represented using two-parameter power functions because of the RSS fill material’s nonlinear failure envelope. The geosynthetic reinforcement strength was 117.8 kN/m based on ultimate tensile strength. Pullout capacity was based on a 0.8 coefficient of interaction and a rough friction angle of 41 degrees for select fill material.
Analysis was done in Slide2 using Spencer’s method and non-circular slip surfaces and the results were compared with those generated from UTEXAS as an internal check.
Starting Point for Analysis
In this analysis, the fill-rock interface was modeled as part of the underlying rock mass using undisturbed rock properties listed in table 1. The critical noncircular failure surface passed through multiple reinforcement layers near the RSS base which generated a factor of safety (FOS) of 2.14. If the assumption of disturbance was used for the fill-rock interface material, the FOS would have been 1.77. From this analysis, the global minimum slip surface was outside of the reinforced zone. The overall results from the analyses determined the slope was sufficiently stable after construction.
Analysis at the time of Failure
Post-failure investigation revealed material located in the fill-rock interface was found to have slickensides, suggesting some areas may have reached a residual state. Using the fully softened shear strength envelope which was used as the starting point for material strength, 0.8 was calculated as a FOS for the critical slip surface behind the RSS. This surface is designated as Surface 1 in figure 1 and corresponds with the location of the tension crack observed in 2013.
Figure 1: Critical failure surfaces for various analysis assumptions.
Back-analysis was then done using multiple non-linear failure envelopes to determine the rock-fill shear strength parameters that would generate a FOS of 1. Figure 1 outlines the results of these calculations. A FOS of 1 was calculated in Slide2 with a= 0.565 for Surface 1.
Figure 1: FSS and back-calculated shear strength envelopes.
Even though the location of Surface 1 agreed with the location of the tension crack discovered in 2013, the actual failure happened 28 meters from the crest of the head scarp, meaning that the failure surface passed through the upper layers of the RSS (Surface 2) as seen in Figure 1.
There could be a variety of reasons attributing to the difference between the calculated critical slip and the observed failure: characterization of the retained fill material, distribution of reinforcement forces or localized stress-strain relationships all could have been factors.
If the retained fill’s shear strength was 10-20% higher, then surface 2 would have had a higher FOS than surface 1. As the slope deformed over time, local stress concentrations developed creating uneven tensile forces. As the rock-fill region degraded, more stress was placed on the Mirafi 20XT geogrid. This was observed as there was significant tension in the upper reinforced area after the failure.
Critical Thinking
The Yeager Airport is an example of slope failure where a variety of different factors were at play and even after extensive numerical LE analysis, uncertainties remained about the failure. The post-failure analysis highlights how important it is to take these variables into account when conducting these analyses for future design purposes.
The use of Limit Equilibrium Method can provide engineers valuable insight for their projects. It is our hope that understanding past case studies like this will help engineers with their projects leading to safe designs in the future.
Read the full case study by Daniel VandenBerge et al.