Some projects almost design themselves. Dense sand? Very stiff clayey soils? Deep groundwater? Well, projects within the collar counties surrounding Cook County and the city of Chicago occasionally encounter a different sort of soil beast: peat and muck. Largely plant-based deposits settled in kettle bogs formed following the Wisconsin Glaciation. These deposits are often revealed during investigations in support of roadways and infrastructure projects and the Burlington Northern and Santa Fe (BNSF) Railroad Bridge replacement over Interstate 294 was one such project.
Traditionally, the typical local response to encountering these materials has been either large-scale removal of the material or construction of land bridges consisting of tightly spaced, buried concrete pile bents with reinforced-concrete slab decks and deep, steel pilings. These traditional design and construction solutions, however, are costly to state and county agencies in both immediate construction dollars, as well as long-term maintenance.
The BNSF rail bridge replacement, part of the major reconstruction of Interstate 294, also known as the Central Tri-State Tollway, was required to maintain three active tracks at all times. A 3-track shoo-fly bridge would be constructed immediately to the south of the existing bridge and would require the eastern approach embankment to widen and push south by almost 90 feet (27.4 meters) at the toe. Beginning approximately 1,000 feet (305 meters) east of the bridge, the approach embankment runs parallel to Burlington Avenue, a municipal roadway and the only access point to a major public park; this roadway was also required to stay open. A 90-meter (300-foot) long temporary modular block retaining wall (Figure 1) was proposed immediately adjacent to Burlington Avenue to support the shoo-fly track widening.
The subsurface investigation of the southeast approach embankment revealed peat and soft, organic silts with thicknesses of greater than 27 feet (8.2 meters). Initial concepts revolved around designing a temporary bridge structure. However, after discussion with the Illinois Tollway and BNSF, the modular wall option was preferred and a geogrid-reinforced, timber pile-supported wall foundation system was developed as an alternative.
Site conditions and geology
The site lies approximately 15 miles southwest of the Chicago Loop at the limit between the Valparaiso and Tinley Moraines in a slight depression or basin fill area. The youngest natural soils within the basin fill consist of weak, compressible clay and silt of moderate to high organic content and muck. The compressible soils reach 27 feet (8.2 meters) in thickness and extend approximately 1,500 meters (5,000 feet) along the rail alignment on each side of Flag Creek, which represents the major drainage feature in the area. The basin fill has been removed in a number of select areas during development of the I-294 corridor; however, weak and compressible soils remain largely in areas along the length of the BNSF corridor beneath and adjacent to the embankment that is close to 100 years old.
The base of the rail embankment along Burlington Avenue sits at an elevation of about 645 to 647 feet (196.6 to 197.3 meters) and is relatively flat. The embankment is about 20 to 25 feet (6.1 to 7.6 meters) tall with a pre-construction rail tie elevation of about 670 feet (204.2 meters).
Soil and groundwater conditions
The subsurface conditions along the proposed temporary modular wall were investigated by advancing seven SPT borings to depths up to 50 feet (15.2 meters) below grade, pushing undisturbed Shelby tubes, performing in-situ pressuremeter testing, and installing a groundwater monitoring piezometer. Below the existing pavement structure along Burlington Avenue, the investigation encountered (A) man-made ground (fill); (B) upper organic silt loam and peat; (C) lower organic silt loam and clay; (D) soft to medium stiff silty clay loam; and (E) medium dense to dense sand and sandy gravel.
The units showed lateral continuity across the length of the proposed wall. The clayey fill material, placed to support the pavement structure of Burlington Road, is a consistent 5 to 6 feet (1.5 to 1.8 meters) thick, while the upper organic material and peat are present with consistent thickness in all but two of the borings.
The upper organic unit presents as black and dark brown silty loam and generally non-fibrous peat with thickness between 5 and 10 feet (1.5 and 3.0 meters). The material is comprised heavily of silt and sand-sized fractions with natural water contents measuring 110% to as high as 270%, and organic content of 22 to 30%. Liquidity indices (LI) of the upper organic are well in excess of 1.0.
A one-dimensional consolidation test was performed on a Shelby tube sample of this unit from boring SAB-18. The compressibility coefficient (cc) measures 2.32 with an initial void ratio (eo) of 3.6 for a sample with a moisture content of 151%, which is about the average moisture measured within this unit. The coefficient of consolidation (cv) at loading levels for 13.9 to 27.8 psi (95.8 to 191.7 kPa) were very slow, about 0.01 square feet per day (9.3 sq cm per day) or less.
At elevations of 633 feet (192.9 meters), the organic silty loam and peat is underlain by the lower organic deposit consisting of lower moisture organic silty clay loam. The silty clay loam has moisture content values from about 40 to 70% which are very close to the measured LL values of 57 to 74%. The plastic limit (PL) values were measured at 27 to 30% and the LI is about 0.7 to 1.0, or close to the liquid state.
At depths of about 20 feet (6.1 meters) and elevations of 630 to 620 feet (192 to 189 meters), the borings encountered a layer of soft to medium stiff, gray silty clay loam with higher strength and noticeably lower moisture content than the upper and lower organic deposits. Laboratory strength testing by unconfined compressive strength show undrained shear strength (Su) values closer to 4.1 psi (28.7 kPa) with corresponding moisture content values of 35 to 48%. Unlike the organic layers, this material is much more discontinuous across the site; where it was encountered it has a maximum thickness of only 5 to 7 feet (1.5 to 2.1 meters).
Deeper foundation soils across the site consisted of stiff to very stiff clays with Su values of 9.0 to 23.6 psi (62 to 163 kPa) and dense sand with Standard Penetration Test rates (N-values) generally greater than 40 blows per foot (0.3 meters).
Groundwater levels along the length of the proposed wall were monitored in an observation well. The well is screened within the upper organic deposit between about 6 to 10 feet (1.8 and 3.0 meters) below the Burlington Avenue pavement elevation. Hydrostatic groundwater was measured at an average elevation of 640 feet (195 meters), or only 3 to 4 feet (0.9 to 1.2 meters) below the existing grade. Considering the groundwater levels in the immediate vicinity of the site, precipitation and surface runoff are the only sources of groundwater recharge.
Pile and geosynthetic support design
The proposed temporary wall, slated to be a modular-type wall, was designed at a bit higher than 19 feet (5.8 meters) and would transfer an ultimate bearing pressure of around 31.9 psi (220 kPa) due to the required 3-track design. It was clear the in-situ soil profile would be incapable of providing adequate bearing capacity, even considering only the service dead load of 14.6 psi (100 kPa). Considering the required depth of embedment for the wall, excavation would take the leveling pad vertically through the fill and essentially to the top of the upper organic deposit; there would be no fill or ‘crust’ to provide load transfer from the soft and compressible deposits over to more rigid elements. In this case, the Stress Reduction Ratio (SRR) or ratio of the load transferred to the peat and muck versus the load transferred to the piles would need to be maximized. Furthermore, to provide meaningful cost savings, the foundation support would need to be provided via inexpensive timber piling.
The geometry of the piling was designed to support both the wall loading for 18-foot (5.5-meter) long wall stems, as well as to provide global stability for the embankment up and behind the wall. The final pile layout included two rows of large-diameter timber piles immediately beneath the wall, spaced at 6 feet (1.8 meters) on-center and capped with 3-foot (0.91-meter) square concrete caps for an (s-a)/a ratio, where s is defined as the spacing a defined as the distance between the caps, of 1. The loads in the piles were determined based on the tributary area and were calculated at nominal required bearing values of 232 kips (1032 kN) with corresponding factored loads of 151 kips (672 kN). The system then included four additional rows of smaller-diameter piles up into the embankment, designed for 108 kips (480 kN), to provide global stability support.
Previous research suggested an unreinforced piled-embankment system with this geometry and H/(s-a) ratio, with s and a previously defined and H as the height of the embankment, would provide a SRR of around 0.4, a stress recovery rate of about 0.3, and uniform settlement at the top of the embankment. Some evaluation methods for determination of arching in piled embankments tend to under-predict the loads transferred to the piles, however, so an SRR closer to 0.2 or 0.3 was anticipated, particularly considering the relatively large height of the fill.
The design converged on two layers of biaxial geogrid with a combined stiffness requirement of 30,000 lbs per foot (440 kN per meter) and allowable reinforcement strength of 800 to 1,200 lbs per foot (12 to 18 kN per meter) at 2% strain after reductions.
Polymeric geogrid load behavior is both time and strain dependent. The calculated ultimate applied load on the wall included three tracks of live rail loading equivalent to about 35% of the total calculated load. The strain induced in the geogrid mobilizes primarily, however, from the longer-term service dead load of the wall and embankment, even though the tensile strength was required to meet the ultimate load. The loading was also only scheduled to be in service for approximately 14 to 16 months. Therefore, taking time and dead load into account the geogrid was designed to act with a stiffness at only 2% strain, as opposed to a higher stiffness at 5% strain. Post-construction settlement was estimated at about 1.2 inches (30 centimeters). The two layers of biaxial grid were installed 6 inches (15 cm) apart, with the bottom layer 6 inches (15 cm) above the top of the pile caps. The total load transfer platform aggregate thickness was 3 feet (90 cm) of Illinois Department of Transportation (IDOT) gradation CA-6 stone.
Monitoring and assessing the SRR
The timber support piles were to be driven through organic deposits and verified with the Washington State Department of Transportation (WSDOT) dynamic formula, in accordance with the IDOT standard. Due to the uncertainty of pile verification, both with the use of timber piling as well as driving through peat and organics, the capacity of the piles was verified by dynamic Pile Driving Analyzer (PDA) to assist in establishing site-specific driving criteria and ensure adequate axial capacity.
The overall effectiveness of the geogrid system, and a basic assessment of the applied SRR, was evaluated by inclusion of nested, 100 psi (690 kPa) maximum pressure piezometers, installed below and between the pile caps, to measure increases in pore pressure beneath the load transfer system, and an inclinometer installed within Burlington Avenue to monitor potential bulging and lateral load transfer within the organic materials. The upper piezometer was installed at a depth of 7 feet (2.1 meters) below the base of the leveling pad with upper organic soil unit 2 and the second piezometer was installed at 17 feet (5.2 meters) below pad elevation within the organic silty soil unit 3. The inclinometer was positioned about 8 feet (2.4 meter) from the face of the wall and extended from the surface of Burlington Avenue to a depth of 46 feet (14 meters), locked into the underlying sand unit 4. Schematics of the monitoring placements are illustrated in Figure 3.
Construction of the wall began in late May 2020 and took approximately one week to complete. The piezometers and inclinometer were installed two weeks prior to wall placement to allow for grout set-up and for initial, baseline pore pressure values to be established. Placement of the bottom row of modular block units on May 30 (Figure 1) resulted in a maximum pore pressure increase of only 0.2 psi (1.4 kPa), showing a significant shedding of load over the piles via the geogrid. The second row of blocks placed on June 1st (Figure 4) barely registered 0.1 psi (0.7 kPa) during placement and by the time the final row of blocks was placed on June 3 (Figure 4) the ground had actually achieved a state of pore pressure lower than the initial condition. For the remainder of construction, the maximum increase in pore pressure occurred during driving of the stability piles behind the wall, where pore pressures reached approximately 0.4 psi (2.7 kPa).
The inclinometer set approximately 8 feet (2.4 meters) from the front face of the wall deformed slowly, but consistently to a deformation of approximately 1 inch (2.5 cm) at a point 4 months after the completion of wall construction (Figure 5). The deformation had slowed considerably about 2 weeks following the completion of the wall, but again was accelerated in late June through late July by driving of the support piling behind the wall.
Pile- or column-supported walls reinforced with biaxial geogrids are a cost-effective alternative to deep foundation solutions on sites that contain deposits of peat and soft, organic soils. Even with relatively high expected loads, the load transfer platform system with multiple rows of geosynthetic over relatively cost-effective timber piling supported a 19-foot (5.8-m) wall while allowing less than 0.3 psi (2 kPa) of excess pore pressure within the underlying highly deformable deposits. Lateral movements of the wall over its service life were ultimately less than 2 inches (5 cm). Greater pore pressure and lateral movement were recorded as a result of additional pile driving behind the wall than that of the vertical wall load itself. As of this writing, the wall has been deconstructed, the shoo-fly tracks removed, and the new BNSF Bridge over I-294 is in full operation.
Mickey Snider, P.E. is a senior geotechnical engineer and technical services manager at Wang Engineering, Inc., a Terracon Company, since 2003. He has authored over 100 bridge and retaining wall structure geotechnical reports for the Illinois Department of Transportation (IDOT) and the Illinois Tollway and has over 15 years of experience performing various types of ground improvement studies including the use of geosynthetics.
All figures courtesy of Mickey Snider, Wang Engineering, Inc.
Maximizing stress reduction with biaxial georgics and timber piling
Location: Interstate 294, Western Springs, Cook County, Ill.
Owner: Illinois Tollway
Contractor: Walsh Construction
Prime engineering consultant: Gannett Fleming, Chicago, Ill.
Geotechnical subconsultant: Wang Engineering, Inc., Lombard, Ill.