Full scale retaining walls

Materials Science & Engineering

By:

Professor R. J. Bathurst, Royal Military College of Canada

Co-authors:

M. P. Burgess, Prof M. Tétreault (Royal Military College of Canada) and M. Tony Allen (Washington State, Department of Transportation)

1. Introduction

The Geotechnical Research Group of the Civil Engineering Department at the Royal Military College of Canada (RMC) has been engaged in a long-term research project related to geosynthetic-reinforced soil structures over many years. A major component of this research has been the construction, surcharge loading and monitoring of instrumented full-scale geosynthetic-reinforced soil retaining walls. A total of 16 full-scale test structures had been built over a period of 12 years. The principal objectives of the experimental work have been to: (1) Develop a better understanding of the mechanical behavior of geosynthetic-reinforced soil walls at working load levels and at collapse under surcharging; (2) Create a database of results from carefully instrumented full-scale model retaining walls that can be used to calibrate numerical models, and; (3) Use the lessons learned from full-scale tests to improve current design methodologies for reinforced soil retaining walls.

2. Current Research Program

Within the long-term objectives described above an experimental program is currently underway that is focussed on the performance of geosynthetic-reinforced segmental (modular block) retaining walls (Jarrett et al. 1997, Simac et al. 1993)). At the time of writing, three walls have been constructed, one is currently under test, components of a fourth are being instrumented and, two walls are in the planning stages. A photograph of the first wall in this research program is shown in Figure 1. The walls were constructed and surcharge loaded in the RMC Retaining Wall Test Facility. This unique facility allows full-scale models to be constructed, surcharged, excavated and monitored in a controlled indoor laboratory environment. Staged uniformly distributed surcharge loads up to 115-kPa pressure can be applied across the entire soil backfill surface using a system of airbags.

3. General Description Of Test Walls

Each of the completed walls was 3.6 m in height, 3.3 m in width and retained a sand fill extending 5.5 m beyond the modular block facing column. The soil was reinforced with layers of polypropylene geogrid. The sand is contained within two parallel reinforced concrete walls that are the main structural components of the RMC Retaining Wall Test Facility. The inside surfaces of the test facility (sidewalls) are comprised of Plexiglas and multiple layers of lubricated polyethylene sheeting to ensure that wall performance corresponds to a plane strain condition. In addition, the 3.3 m wide facing column is built in three panels- two 1.15 m wide outside sections and a central 1 m wide panel. The central panel section and its footing support are instrumented and isolated from the outside panels to further decouple the instrumented section from the test facility sidewalls.

3.1 Current walls

Wall 1:

The first wall was constructed with a polypropylene geogrid with low strength and low stiffness properties in order to encourage large strains and large wall deformations under uniform surcharge loading. The wall was designed to satisfy current NCMA guidelines (Simac et al. 1993) with the added constraint that the reinforcement layer spacing not exceed a distance equal to twice the modular block toe to heel dimension (AASHTO 1997). The resulting design is illustrated in Figure 2. This wall is the control or reference case for the remaining wall structures in the current research program.

Wall 2:

The second wall was constructed in an identical manner to the first wall except that the polypropylene geogrid was modified by removing every second longitudinal member. Hence the reinforcement in this wall has 50% of the strength and 50% of the stiffness of the reinforcement used in the control structure. This was done to isolate the influence of reduced reinforcement strength and stiffness on wall performance.

Wall 3:

The third wall was constructed in an identical manner to the first wall except that four(4) layers of reinforcement were used (Figure 2). This wall allows the influence of reinforcement spacing to be isolated.

Wall 4:

The current wall that is currently under surcharge loading is identical to Wall 1 except that a wrapped-face construction was used. This wall will allow the influence of the facing type on wall performance to be isolated (i.e. compare the performance of a reinforced soil wall with a relatively stiff concrete facing to the performance of a nominal identical wall constructed with a very flexible wrapped-face construction).

3.2 Future Walls

Wall 5:

This wall will be constructed with a specially designed steel mesh reinforcement with strength similar to that of the control structure. This reinforcement will allow the influence of high stiffness reinforcement to be examined. This reinforcement will also allow complete collapse of the structure to be achieved since, unlike the polymeric reinforcement used in the preceding structures, the reinforcement will rupture rather than strain excessively under staged surcharging.

Wall 6:

This wall will be constructed with a polyester reinforcement material. Data from this structure will increase the database of results that is focussed on the influence of reinforcement stiffness, strength and strain at rupture on wall performance.

3.3 Wall Details

Facing units:

The facing units for all walls with the exception of Wall 4 (wrapped-face wall) are a solid masonry block unit with a continuous concrete shear key. The blocks are 300 mm long (toe to heel), 150 mm high, 200 mm long and have a mass of 20 kg. The blocks were modified to accept specially manufactured load rings that allow connection loads to be recorded continuously during construction, staged surcharging and excavation. To simplify interpretation of test results, the connections were designed so that there is no slip in the connections. Hence, the connection strength is equal to the strength of the reinforcement in the soil.

Reinforcement:

A weak and very extensible polypropylene geogrid was selected for the reinforcement. This material was purposely selected because it will not rupture but rather strain excessively under large surcharge loads. In this way the wall can be loaded to very large surcharge loads while generating large measurable strains and displacements in the wall components without catastrophic collapse of the wall. A careful set of in-isolation creep tests was performed on geogrid specimens to determine isochronous load-strain data and stress rupture curves for analysis purposes.

Backfill:

A uniform particle size beach sand was used for the soil in the reinforced and retained soil zones. This material was selected because it has no cohesion, is easily compacted and has quantifiable and repeatable mechanical properties. Direct shear, triaxial and plane strain laboratory tests have been carried out to quantify mechanical properties of the sand for limit-equilibrium based back-analyses and for numerical modelling.

Surcharging:

Each wall was subjected to staged uniformly distributed surcharge loading. Typical surcharge increments were in the order of 10 kPa. Each load level was left in place for 100 hrs or more to record creep deformations in the reinforcement and to measure other time dependent deformations of the structure. The surcharge loading was continued until very large strains in the reinforcement layers were recorded and/or wall deformations were excessive. In each wall a well-defined failure surface through the reinforced soil zone was detected during and after the surcharging program. The walls were unloaded in stages to record strain relaxation in the reinforcement layers.

Toe Release:

After the surcharge program was complete and the surface of the sand backfill unloaded, the toe restraint at the bottom of the facing column (Walls 1, 2 and 3) was released temporarily to investigate the effect of footing restraint on the magnitude and distribution of forces in the reinforcement layers.

Excavation:

Each wall was excavated in lifts to visually examine internal failure surfaces and the condition of the reinforcement layers.

3.4 Instrumentation

A typical instrumentation plan is illustrated in Figure 3. The walls were instrumented to record: horizontal and vertical toe loads; facing deflections; reinforcement displacements and strains; connection loads between the reinforcement layers and the wall facing column; earth pressures within and at the base of the soil mass; internal soil mass movement and; retained soil surface displacements. More than 250 instruments were monitored using a computer-controlled data acquisition system. The data acquisition program was designed so that all data from the beginning of wall construction can be plotted rapidly. Hence, a complete and current history of wall performance is available to the research group at any time during the experiment. In this way "real time" decisions can be made regarding whether to leave the surcharge load in place (i.e. continue monitoring creep deformations) or to increase the surcharge load level if creep deformations have stopped.

3.5 Example results

The data from the first three walls are currently being analysed but some preliminary results are presented here to illustrate the comprehensive monitoring program and to point out some interesting observations. Figure 4 illustrates wall deflections recorded for Wall 1. The horizontal deflections were recorded at reinforcement elevations on the outside of the facing column. Each jump in a deflection curve corresponds to the application of a new surcharge load. Creep of the structure during each load increment is clearly evident in the figure.

Figure 5 shows the distribution of reinforcement strains recorded in Wall 2 after the maximum surcharge load of 85 kPa was applied. These strain profiles are superimposed on the wall cross section. Also illustrated is the internal failure plane observed during wall excavation and the predicted curve using Coulomb theory. In fact, the directly observed failure plane was indistinguishable from a log-spiral approximation.

Figure 6 shows the distribution of strains in the reinforcement at the end of construction. The plot shows that the strains are very low but that they are, nevertheless, largest at the connections.

Figure 7 shows the distribution of strains in layer 5 of Wall 2 at different surcharge load levels. Only after the surcharge load reached 50 or 60 kPa did the peak reinforcement strain move from the connection to a location on the reinforcement corresponding to the internal failure plane in the reinforced soil zone.

A large amount of data for the first four walls in this test program is currently being analysed and the results compared for the four different configurations. Some preliminary observations can be made:

• Connection loads for the structures with a modular block facing construction are the largest loads in the reinforcement at the end-of- construction condition.
• The vertical normal load acting at the toe of the facing column is much greater than the sum of the block weights due to soil down drag forces acting at the back of the facing column. This has important implications to connection design and confirms that for the wall batter used in these experiments the current NCMA method (Simac et al. 1993) to compute normal forces at the block interfaces is conservatively safe.
• The toe of the facing column carries a significant portion of the horizontal earth forces acting on the facing column. This load capacity is not accounted for in current "tie-back wedge" methods of analysis and design and is a source of conservatism in current design practice.
• Current Coulomb earth pressure theory consistently over-estimates reinforcement and connection loads. Preliminary calculations suggest that the loads in the reinforcement are only 60 to 70% of the values calculated from Coulomb theory using the "actual" soil friction angle.
• Concurrent with the experimental program are numerical simulation studies that are being used to calibrate numerical models. Once these numerical models are available, parametric analyses will then be carried out to investigate a large range of wall types, soil materials and reinforcement arrangement and properties. The combined experimental and numerical work will guide the development of reinforced soil wall design guidelines that are currently under review in North America.

Acknowledgements

Funding for the research program completed to date was provided by the Washington State Department of Transportation, National Concrete Masonry Association, Natural Sciences and Engineering Research Council of Canada, Academic Research Program of the Department of National Defense (Canada) and a grant from the Department of Infrastructure and Environment (DND Canada).

References/Bibliographie

1. AASHTO, 1997, Standard Specifications for Highway Bridges , American Association of State Highway and Transportation Officials, Washington, D.C., USA.
2. Jarrett, P.M., Bathurst, R.J. and Tetreault, M. Récents développements dans les méthodes de dimesionnement de massifs renforcés a parement cellulaire, en amérique du nord . Recontres'97, Rheims, France, 8-9 October 1997, 10 p.
3. Simac, M.R., Bathurst, R.J., Lothspeich, S. and Berg, R.R., 1993, Design Manual for Segmental Retaining Walls (Modular Block Retaining Systems) , First Edition, (Collin, J.G., 1996, Second Edition, Editor), National Concrete Masonry Association, Herdon, Virginia, USA, 250 p. 1