1 Dept. of Civil and Environmental Engineering, University of Utah, 2 Dept. of Geology and Geophysics, University of Utah

Applications of EPS Geofoam in Design and Construction of Earthquake

Resilient Infrastructure

Bartlett, Steven F.1, Lawton, Evert C.1, Trandafir Aurelian C.2 and Lingwall, Bret N.1

Abstract

Expanded Polystyrene (EPS) geofoam has been used to construct earthquake resilient infrastructure in areas with high seismicity because of its extremely low mass density and relatively high compressibility when compared with traditional backfill materials. The primary applications of EPS in North America include construction of light-weight embankments over soft soils, construction/reconstruction of steepened slopes and placement of EPS as a light-weight backfill against retaining walls and buried structures. This paper summarizes recent research conducted at the University of Utah regarding the seismic design and construction of EPS geosystems to improve the resiliency of embankments, slopes and buried pipelines. This research includes: (1) assessment of the seismic stability of free standing EPS embankments subjected to large, nearby earthquakes and (2) development and construction of EPS cover/backfill systems to protect buried steel pipelines from potential rupture caused by seismically-induced permanent ground deformation (e.g., tectonic faulting, subsidence, liquefaction, land sliding, etc.). The assessment of (1) was performed for free-standing embankment similar to that used on the I-15 Reconstruction Project in Salt Lake City, Utah. The evaluations suggest that interlayer sliding may be initiated in some cases and the amount of sliding displacements depends on the amplitude and long-period characteristics of the inputted strong motion. Therefore, shear keys or other structural/mechanical restraints are recommended for free-stand EPS embankment systems where the seismically-induced sliding displacement is potentially damaging to the geosystem. Regarding (2), full-scale experiments show that a light-weight cover system consisting primarily of EPS block and an overlying flexible pavement cover offers significant benefits in protecting buried steel pipelines from the damaging effects of permanent ground displacement that often disrupts urban roadways in seismically active regions.

1

1. SEISMIC STABILITY ASSESSMENT OF FREE-STANDING EPS

EMBANKMENTS

The Utah Department of Transportation (UDOT) and Utah Transit Authority (UTA) have made extensive use of expanded polystyrene (EPS) geofoam embankment at numerous locations in Salt Lake County, Utah. Geofoam embankment has been used to avoid the triggering of primary consolidation settlement of the lacustrine clays that underlie much of the County. This paper presents analytical and numerical methods for evaluating the seismic response and stability of free-standing geofoam embankments (Figure 1) subjected to large amplitude strong motion generated by nearby, large earthquakes. The recommended approach consists of two levels of evaluation: (1) calculation of embankment sliding stability using pseudo static techniques, and (2) use of more advanced numerical analyses, such as finite difference method (FDM) or finite element method (FEM), to quantify the potential amount of sliding displacement and to evaluate the potential for rocking of the embankment. In the latter approach, a FDM method, as implemented in the commercially available software FLAC (Fast Lagrangian Analysis of Continua) (Itasca, 2005), was used to evaluate the sliding and rocking potential of the geofoam embankment (Bartlett and Lawton, 2006). For critical geofoam embankments, such as those at bridge approaches, we recommended that the numerical evaluations be performed in a coupled fashion using both the horizontal and vertical components of strong motion from a representative set of acceleration time histories valid for the seismic conditions of the design earthquake(s).

Fig. 1. Typical free standing EPS embankment used for the I-15 Reconstruction Project, Salt Lake City, Utah.

1.1. Pseudo Static Analyses

The potential for initiation of interlayer and basal sliding of a geofoam embankment can be evaluated using pseudo static techniques. This type of analysis is useful for evaluating the stability of simple systems when the embankment cross-section is a simple rectangle. In this approach, the inertial horizontal force acting on the geofoam embankment is applied at the

2

centroid of the mass, which is usually at the top of the embankment. To calculate the appropriate acceleration, the geofoam is treated as a single degree of freedom (SDOF) oscillator (Horvath, 1995) and its fundamental period, T0, is estimated using Eq. 1, Horvath (2004):

T0 = 2π{[(σ′v H)/(E*g)[4(H/B)2 + (12/5)(1+ν)]}0.5 (1)

where: σ′v is the vertical effective stress acting on the top of the geofoam from applied dead loads (i.e., pavement section), H is the geofoam embankment height, E is the initial Young’s modulus of the geofoam, g is the gravitational constant, B is the width of the geofoam

embankment and ν is Poisson’s ratio. When Eq. 1 is applied to an 8-m high by 20-m wide free-standing geofoam embankment using the properties given in Table 1, the value of T0 is 0.52 s. The horizontal inertial force, Fh, produced by the earthquake is applied to the centroid of the lumped mass, which is approximately located at the top of the embankment near the mid-point of the pavement section:

Fh = Sa * m (2)

where: Sa is the spectral acceleration corresponding to T0 obtained from the design basis earthquake acceleration response spectrum and m is the lumped mass of the system (combined mass of the pavement, road base and concrete load distribution slab). In the U.S., geofoam embankment is often considered to be a “retaining” structure/wall and as such, it is designed for a 5 percent damped Sa value that has a 10 percent probability of being exceeded in 50 years (i.e., average return period of 475 years) as specified by the American Association of Highway and Transportation Officials (AASHTO, 2010). An example 5 percent damped AASHTO spectrum for such an event is shown in Fig. 2.

Table 1. Initial elastic moduli and properties of EPS19, I-15 Reconstruction Project

Material Type Layer No. Thickness

(m)

ρ 4

(kg/m3)

E 5

(MPa)

ν6

K 7

(MPa)

G 8

(MPa)

Foundation Soil 1-10 varies 1840 174 0.4 290.0 62.1

Geofoam 11-18 8 18 10 0.103 4.2 4.5

UTBC1 19 0.610 2240 570 0.35 633 211

LDS2 & PCCP3 19 0.508 2400 30000 0.18 15625 12712

1 Untreated base course, 2 Load distribution slab, 3 Portland concrete cement pavement, 4 Mass density, 5 Initial Young’s modulus, 6 Poisson’s ratio, 7 Bulk modulus, 8 Shear modulus

Previous work by Bartlett and Lawton (2008) suggests that interlayer and/or basal sliding is the controlling failure mode for free-standing EPS embankments subjected to large earthquakes. Other behavior modes, such as horizontal sway and rocking (Raid and Horvath, 2004), are more

3

difficult to excite in free-standing EPS embankments due to their relatively large base to height aspect ratios. In applying pseudo static techniques to interlayer and basal sliding evaluations, values of horizontal acceleration at various heights within the embankment are linearly interpolated, starting at the top of the EPS embankment and continuing to its base (NCHRP 529) (Table 2). The horizontal acceleration acting at the top interface of the embankment is the Sa value from the design spectrum at T=0.52s, which is of 0.848 g for the example case (Fig. 2); the horizontal acceleration at the basal EPS/foundation soil interface is peak horizontal ground acceleration (pga), which is 0.339 g for the example case (Figure 2) and corresponds to the spectral acceleration at T=0 s (Fig. 2).

Fig. 2. Five percent damped acceleration response spectrum for Salt Lake Valley, Utah for deep soil site (Site Class D).

Subsequently, Eq. 2 is applied to the interpolated acceleration values at each interface elevation to estimate the inertial sliding force acting at that interface (Table 2). The frictional sliding resistance of the interface is calculated using the normal stress (i.e., vertical stress) acting at the interface multiplied by interface coefficient of friction and by the percentage of area available to resist sliding (expressed in decimal fraction). (The weight of the EPS is usually neglected in calculating the normal stress.) In this calculation, the coefficient of friction for geofoam-to-geofoam and geofoam-to-soil interfaces was estimated to be 0.8 and 0.6, respectively, based on direct shear testing from the I-15 Reconstruction Project (Bartlett et al. 2000). In addition, any potential bonding that develops between the EPS and the overlying concrete load distribution slab was ignored at interface #9 in this example (Table 2); but such a bond shear strength could be include if: (1) it can be reasonably obtained from experimental data, and (2) such a bond can be shown to persist throughout the design life of the embankment.

The recommended factor of safety against interlayer and basal sliding is 1.2 to 1.3, which may not be achieved at all interfaces relying on frictional resistance solely. For interfaces where unacceptably low safety factors are calculated, shear keys can be constructed during the placement of the geofoam block to reduce the potential for interlayer sliding. Such keys disrupt the development of horizontal sliding planes during earthquake shaking and are constructed by

0,8476072

00,10,20,30,40,50,60,70,80,9

0 0,5 1 1,5 2

Sp

ectr

al a

ccel

erat

ion

(g)

Period (s)

Design Spectrum - SiteClass D

4

periodically placing half-height blocks in the geofoam mass followed by placing full-height block in the successive layer (Fig. 3). The full-height block placed in the key acts as a barrier to sliding and the shear resistance of the block is mobilized to resist sliding. Therefore, the key greatly improves the factor of safety against interlayer sliding due to the relatively high shear strength of the EPS block (Table 2). The resisting force provided by the key is calculated by multiplying the shear strength of the block by the percentage of area occupied by the key (Table 2). We note that if a shear key is used at a particular interface, the area available for frictional contact must be reduced correspondingly when calculating the resisting sliding force (Table 2). Note also that shear keys were not required at interfaces 0 to 4, because the inertial forces in the embankment decrease at these lower elevations, increasing the factor of safety at lower elevations within the embankment.

Table 2. Example interlayer sliding calculation

H = 8 m Block thickness = 0.81 m number of interfaces 9 normal stress 25.36 kPa interface friction 0.8 (geofoam - geofoam) interface friction 0.6 (geofoam - soil) geofoam shear strength 23 psi (EPS19 used in shear key) geofoam shear strength 157.3 kPa

Horiz. mass inertial resisting shear resisting FS

interface Accel. (kg/m3) force sliding key force sliding

# (g) (N/m3) force coverage from key (w / key)

(N/m3) (%) (N/m3)

9 0.848 2585 21497 19073 6 9439 1.33 8 0.791 2585 20064 19478 4 6293 1.28 7 0.735 2585 18631 19681 3 4720 1.31 6 0.678 2585 17198 19884 2 3146 1.34 5 0.622 2585 15765 20087 1 1573 1.37 4 0.565 2585 14332 20290 0 0 1.42 3 0.509 2585 12898 20290 0 0 1.57 2 0.452 2585 11465 20290 0 0 1.77 1 0.396 2585 10032 20290 0 0 2.02 0 0.339 2585 8599 15217 0 0 1.77

Any sliding resistance provided by the fascia panel wall (Fig. 1) was omitted from the sliding calculations (Table 2). In the I-15 Reconstruction Project design, the fascia panel wall is not designed to resist horizontal seismic force imparted by the EPS backfill. Typically, a 0.2-m gap was left between the face of the geofoam block and the back of the wall to prevent interaction. However, this wall could be designed to resist horizontal inertial forces, if needed, for alternative situations.

5

0

0,2

0,4

0,6

0,8

1

1,2

0,0001 0,01 1

G/G

max

Shear strain (%)

Clay (Sun et al.1988)

Geofoam(Anthanasopouloset al. 1999)

1.2. Numerical Analyses

Despite their popularity and ease of use, pseudo static methods do not capture the complex interactions that occur in EPS embankments; hence more advanced methods are required for more realistic analyses. Advanced numerical techniques are warranted for cases where: (1) the factor of safety is below unity and estimates of the magnitude of the sliding displacement are required, or (2) analyses of more complex modes of failure are required (e.g., horizontal sway and rocking), or (3) potential interaction with adjoining structures (i.e., walls), earthen embankment or natural slopes is expected, or (4) complex EPS embankment geometries are involved.

For our evaluations of free-standing EPS embankment, we chose to model the geofoam embankment behavior using an explicit finite difference program called FLACTM (Fast Lagrangian Analysis of Continua) (Itasca, 2005). FLAC has the capability to model the 2D nonlinear dynamic response of the embankment and evaluate the potential for interlayer sliding, horizontal sway and rocking (Bartlett and Lawton, 2008). The interface nodes used in FLAC allow for sliding and separation, which allows for exploration of these more complex behaviors.

The elastic properties of the EPS19 used in the FLAC modeling (Table 1) are from large block testing performed by Elragi (2000) and are considered to be representative of the geofoam used on the UDOT and UTA projects in Salt Lake City, Utah. The nonlinear behavior of the EPS was modeled using FLAC’s hysteretic damping option, which allows for nonlinear, shear strain-dependent modulus and damping in the geofoam and foundation soil. The potential for yielding can be explored by inputting Mohr-Coulomb failure criteria, which

Fig. 3. Shear key installed in a geofoam embankment.

Fig. 4. Shear modulus degradation curves used in FLAC model.

6

was done for the horizontal sway and rocking evaluations. However, such criteria were not used for the sliding evaluations in order to accentuate the potential for interface sliding. (In addition, modeling results showed that interface sliding is initiated before the development of shearing of the block for cases where shear keys were not present in the EPS mass.)

To implement FLAC’s hysteretic damping option, shear modulus degradation and damping curves appropriate for EPS19 were obtained from Athanasopoulos et al. (1999) and for clay from Sun et al. (1998) (Figure 4). We used the three-parameter sigmoidal model (i.e., sig3 model) (Itasca, 2005) to fit the shear modulus degradation curves (Fig. 4). The sigmoidal model parameters used to match these curves are: a = 1, b = -0.45, and x0 = 0.3 for EPS19 and a = 1.017, b = -0.587, and x0 = -0.633 for the foundation clay.

The free-standing embankment was modeled using a 1-m by 1-m grid spacing that consists of a 10-m thick clay foundation layer, 8-m high geofoam embankment, and 1-m thick lumped mass (Fig. 5). The 0.2-m thick bedding sand layer at the base of the EPS was ignored. The lumped mass represents the combined masses of the load distribution slab (LDS), road base, and concrete pavement (PCCP). The lumped mass was given elastic properties appropriate for concrete (Table 1), thus, as modeled, it acts as a coherent, very stiff system placed atop the geofoam that will essentially undergo little to no significant internal deformation.

Fig. 5. FLAC model used for dynamic simulations.

7

Interfacial nodes were used at interfaces 1 (base) through 9 (top) to allow sliding and separation in the embankment between the various layers (Fig. 5). Interface 1 is the basal foundation soil/geofoam contact surface, interfaces 2 through 8 are geofoam/geofoam contact surfaces (from bottom to top, respectively), and interface 9 is a geofoam/lump mass contact surface. FLAC requires Mohr-Coulomb properties and normal and shear stiffness at all interfaces. These include: interface friction angle, cohesion, tensile strength, normal stiffness, and shear stiffness (Table 3). However, no cohesion, dilation, or tensile bond strengths were applied at these interfaces. The lack of tensile bond strength allows for separation between the various layers. Thus, only friction was used at the interfaces to resist sliding. In addition, the interface friction angles given in Table 3 are intermediate values between the peak and residual friction angle. We also assumed that dilation due to sliding at the interfaces was negligible, so the dilation angle was set to zero. Also, the interface between the foundation soil and basal geofoam layer (i.e., interface 1) was “glued” so that slippage did not occur. This was done for two reasons: (1) sliding at this interface is constrained horizontally by the foundation of the panel wall (Fig. 1), and (2) it was desirable to have full seismic forces transferred to the geofoam embankment to maximize the potential for sliding at these interfaces. Lastly, only sliding friction was used at interface 9 at the top of geofoam/bottom of lumped mass contact surface. In reality, this interface is not solely a frictional contact, because the concrete load distribution slab is poured directly on the geofoam and some tensile and shear bonding undoubtedly occurs; this bonding was neglected our sliding evaluations.

Table 3. Interfacial properties used for sliding evaluation in the FLAC model

Contact Surface

Interface number (bottom to top)

Normal and Shear Stiffness (kn = ks) (MPa) Friction angle (degrees)

Geofoam-soil 1 102 311

Geofoam-Geofoam 2-8 102 38 Geofoam-Lump Mass 9 102 382

______________________________________________________________________

1 A glued interface was used for interface 1 in FLAC because the geofoam is abutted against the panel wall footing and cannot slide. 2 Neglects any tensile or shear bonding that may develop between the top of geofoam and base of the load distribution slab.

The seismic demand on the FLAC model was represented by 8 acceleration time histories taken from large, nearby earthquakes. For the Salt Lake Valley, the seismic hazard is dominated by a characteristic M7.0 to 7.5 event that ruptures on the Salt Lake City segment of the Wasatch fault with expected horizontal pga values of about 0.5 to 0.6 g. The horizontal and vertical acceleration time histories for this event were obtained from the Pacific Earthquake Engineering

8

Research (PEER) Center strong motion website and were unmodified (Table 4). The horizontal response spectra for these events are shown in Fig. 6. Note that the spectral accelerations at T=0.52 seconds (i.e., fundamental period of the embankment) vary from about 0.8 to 1.5 g, which implies that interlayer sliding will be initiated based on the pseudo static analyses discussed previously. The vertical components of these time histories were also included in the FLAC analyses (Bartlett and Lawton, 2008), but their response spectra are not shown herein.

Table 4. strong motion records selected for evaluations

Motion Earthquake M R (km) Component pga (g)

1 1989 Loma Prieta, CA 6.9 8.6 Capitola 000 0.52 2 1989 Loma Prieta, CA 6.9 8.6 Capitola 090 0.44 3 1999 Duzce, Turkey 7.1 8.2 Duzce 180 0.35 4 1999 Duzce, Turkey 7.1 8.2 Duzce 270 0.54 5 1992 Cape Mendocino, CA 7.1 9.5 Petrolia 000 0.59 6 1992 Cape Mendocino, CA 7.1 9.5 Petrolia 090 0.66 7 1994 Northridge, CA 6.7 6.2 Sylmar 052 0.61 8 1994 Northridge, CA 6.7 6.2 Sylmar 142 0.90

The acceleration records in Table 4 were selected because their earthquake magnitude, source distance, and soil conditions are similar to those expected for the central part of the Salt Lake Valley. These strong motion records were deconvolved to a depth equal to the base of the 2D numerical model (10 m below ground surface) using the 1D equivalent linear procedures described by Mejia and Dawson (2006).

The deconvolved motions were assigned at the base of the FLAC model after static force equilibrium was reached. For the dynamic simulations, a quiet (i.e., viscous) boundary was used at the base of the model in both the x and y directions. This boundary allows for perfect absorption of the incident wave and is the preferred boundary condition for modeling deep soil deposits (Mejia and Dawson, 2006). In addition, the sides of the soil model were changed to free-field boundaries for the dynamic simulations. This boundary type forces a 1D free-field boundary condition at the model’s edge, which essentially treats the boundary as if it were placed at an infinite distance.

The use of a quiet boundary required a stress time history to be input at the boundary; thus the velocity time histories obtained at the base of the soil column from the deconvolution analysis were converted to a stress wave for the subsequent FLAC analyses (Itasca, 2005; Mejia and Lawson, 2006).

9

Fig. 6. Horizontal acceleration response spectra used in dynamic simulations.

The FLAC model was used to estimate the amount of interlayer sliding displacement produced in the geofoam embankment for the candidate time histories. To do this, the relative sliding displacement between successive layers of geofoam was calculated and summed as a function of time for all layer interfaces to calculate the total relative sliding displacement (TRSD) time history shown in Figure 7. The maximum TRSD values for each time history case are tabulated in Table 5. The maximum TRSD values range from 0.01 m to greater than 1 m for the various cases listed in Table 5. We believe that TRSD values less than about 0.1 m are probably acceptable from an embankment performance standpoint; but larger values could have potentially damaging consequences to the adjacent panel wall, overlying approach slabs, and pavement section, etc).

The acceleration time histories used for case 4 produced unusually high amounts of interlayer sliding (1.3 m) (Table 5). Inspection of the original displacement time histories for this case (i.e., Duzce 270 record) revealed a maximum horizontal displacement of about 50 cm; whereas the Petrolia 000 record (cases 5 and 6, Table 5) had a considerably smaller maximum horizontal displacement of about 20 cm. This suggested that interlayer sliding is highly affected by the magnitude of the displacement pulses present in the input time history and their frequency and phasing (Bartlett and Lawton, 2008).

Response Spectra (5% Damping)

Motion 1 Motion 2 Motion 3 Motion 4 Motion 5 Motion 6 Motion 7 Motion 8

Spe

ctra

l Acc

eler

atio

n (g

)

Period (sec)

0.0

0.5

1.0

1.5

2.0

2.5

0 1 2 3 4 5

10

Fig. 7. Relative sliding displacement plot for various geofoam layers for case 1a

We noted that TRSD values are somewhat higher when the vertical component of strong motion is included in the analysis (Table 5). Some cases suggest that sliding displacement increases by a factor of 2 to 5; however, some cases produced essentially the same amount of sliding displacement (Table 5). This suggests that the magnitude of the vertical component of strong motion and its phasing relation to horizontal displacement pulses present in the horizontal record are important factors that affect the amount of sliding. In addition, the FLAC models showed that relative sliding is greater near the base of the geofoam embankment and becomes successively less in upper layers. This trend was consistent for all models. This result is opposite of that expected using pseudo static techniques, where the highest inertial forces and lowest factors of safety against sliding are found near the top of the embankment (Table 2). In addition, no sliding was observed in the numerical models between the top of the geofoam and base of the lumped mass, even though only a friction contact was used at this interface.

The increased sliding displacement near the base of the numerical models can be explained by wave propagation, phasing and damping. As the incident displacement wave in the foundation soil approaches the base of the embankment, it is most likely out-of-phase with the oscillation of the embankment. This increases the relative accelerations and forces causing the sliding displacement to be concentrated near the base. Once sliding has initiated at this location, it acts as a damper, thus significantly reducing the energy imparted to the overlying layers and isolating them from further displacement from the wave.

11

Lastly, we emphasize that the TRSD values reported in Table 5 are estimates from a numerical model that has not been fully calibrated with experimental or observation data; thus, some uncertainty exists in these sliding estimates. Because of this, we believe that it is important not to scrutinize solely the estimated displacements, but instead to see if their general magnitudes and trend suggest that a sliding stability threshold has been exceeded where the expected sliding may become unacceptably large. For example, two of the eight cases that included the vertical component of strong motion had TRSD values greater than 0.1 m (Table 5, cases 4b and 8b). Because this represents 25 percent of the vertical + horizontal component cases, it is probable that sliding displacement may be unacceptably large for the specific embankment configuration and earthquakes represented in these evaluations.

However, we believe the potential for interlayer sliding in geofoam embankments can be easily addressed by systematically including shear keys during construction to disrupt any continuous horizontal slide planes. Shear key features were not included in the dynamic modeling due to difficulty of adequately representing the shear key in a 2D cross-sectional model. We believe a 3D model is required to fully represent their influence, but the pseudo static method suggests that they should be effective, as previously discussed.

Table 5. Summary of relative sliding displacement

Case Horizontal Motion Vertical Motion Max. total relative sliding displacement (m)

1a 1 Not applied 0.06 1b 1 1 0.06 2a 2 Not applied 0.01 2b 2 1 0.05 3a 3 Not applied 0.06 3b 3 2 0.06 4a 4 Not applied 1.3 4b 4 2 1.3 5a 5 Not applied 0.005 5b 5 3 0.01 6a 6 Not applied 0.05 6b 6 3 0.06 7a 7 Not applied 0.5 7b 7 4 0.6 8a 8 Not applied 0.6 8b 8 4 0.5

In addition to sliding, Raid and Horvath (2004) discuss horizontal sway and rigid body rocking as fundamental seismic behaviors that could affect embankment stability. Horizontal sway results from flexibility of the geofoam mass in the horizontal (x) direction, and rocking results from 2D rigid body rotation (Raid and Horvath, 2004). In reality, because the geofoam embankment is relatively flexible when subjected to strong motion, it will undergo both horizontal and vertical internal deformation as the mass attempts to sway and rock in response to

12

the earthquake strong motion. In attempting to analyze these behaviors, we found that horizontal sway and rocking modes were somewhat difficult to induce in the numerical model for the selected embankment geometry. This occurred because interlayer sliding was induced before these other behavior modes could fully develop, and once sliding was initiated these behaviors were muted. Thus, somewhat artificial constraints were imposed on the numerical model to stop interlayer and basal sliding and hence accentuate sway and rocking (Bartlett and Lawton, 2008). These preliminary results suggested that internal deformation caused by sway and localized uplift caused by rocking may cause localize yielding of some blocks usually near the base of the embankment. Thus, we recommend that consideration should be given to using geofoam blocks with higher strength than that of EPS19 in the basal layers of geofoam embankments that may undergo high levels of strong motion from nearby earthquakes.

2. LIGHT-WEIGHT COVER SYSTEMS FOR PROTECTION OF BURIED

PIPELINES

Full-scale experiments and numerical evaluations conducted at the University of Utah for a local natural gas utility show that a light-weight cover constructed with EPS block offers significant benefits in protecting buried pipelines from the damaging effects of vertical offset caused by permanent ground deformation (PGD) resulting from normal faulting, or mass movement processes (e.g., landsliding, liquefaction-induced lateral spread, etc.) (Lingwall, 2009; Bartlett et al. in press). The characteristics of EPS applicable to pipeline protection are its extremely low mass density (12 to 45 kg/m3) and its moderately high compressibility when compared with typical granular backfill materials. When used as a lightweight backfill in pipeline cover systems, EPS can minimize the bending, shear and axial stresses imposed on the buried pipeline as it displaces relative to the surrounding soil.

Geofoam has been used to reduce soil pressures on buried culverts (Sun et al. 2009) and to decrease static lateral earth pressure against buried walls (Negussey and Sun, 1996). Yoshizaki and Sakanoue (2003), Choo et al. (2007), Lingwall and Bartlett (2007) and Lingwall (2009) have evaluated the use of EPS as a lightweight cover/backfill system for buried pipelines. These studies show that EPS significantly reduces the vertical and horizontal soil pressures acting on the soil-pipeline system as the pipeline undergoes PGD. For example, Yoshizaki and Sakanoue (2003) have shown that the horizontal force on a 100-mm steel pipeline undergoing horizontal PGD is significantly reduced when compared to sand backfill. They conducted horizontal pipe displacements tests in a 2-m wide by 2-m high by 3.1-m long box using both a geofoam cover system and conventional sand cover. Force-displacement measurements were taken as the pipe was displaced a maximum distance of 150 mm. For the geofoam cover system, the peak horizontal force on the pipe was reduced by 40 to 60 percent compared with a typical sand cover. In addition, Yoshizake and Sakanoue (2003) found that the resistance of pipeline elbows subjected to PGD could be significantly improved by using EPS backfill placed around the pipe elbow. In addition, Choo et al. (2007) have shown that when geofoam is placed in the cover and

13

trench sidewalls, it significantly reduces soil-pipeline interaction forces for high-density polyethylene (HDPE) pipe. These conclusions were based on twelve experiments and subsequent analyses from centrifuge tests performed on various geofoam remediation strategies. The strategies were successful in reducing the peak transverse lateral force at the soil-pipe interface by as much as 80 to 90 percent, depending on the block placement configuration. The reduction in the lateral force led to a 45 to 60 percent reduction in the pipe bending strain when compared with systems without EPS remediation. In addition, the EPS systems reduced the axial strain in the pipe by approximately 15 to 30 percent. Lastly, Lingwall and Bartlett (2007) have evaluated the benefits of an EPS cover system for protecting a 0.6-m diameter steel pipe that crosses the Salt Lake City, Utah segment of the Wasatch fault. The numerical results suggested that when compared with a conventional sand cover, the EPS cover can undergo approximately 4 times greater vertical displacement before pipe yielding was predicted by the model (Lingwall and Bartlett, 2007). The modeling results also suggested that some benefit is gained by placing EPS under the pipe on the up-thrown side of the fault to provide a cushion in the trench bottom.

2.1. Uplift Tests

To verify their concept, Bartlett and Lingwall (2007) performed a set of full-scale uplift tests on a pipe system undergoing vertical ground displacement (Fig. 9) (Lingwall 2009). The design concept was to create a controlled “slot” failure of the cover system by using EPS block placed above the pipe and allowing the pipe to move upward; thereby compressing and ultimately displacing the block as would occur on the down-thrown side of a normal fault.

Two full-scale tests were performed: (1) prototype EPS cover system test, and (2) native soil backfill test and where the second test served as a baseline for comparison with first test. To construct this test, the trench was over-excavated and the 0.3-m diameter steel pipe was placed in bedding sand. Crane rigging, which would be eventually used to uplift the pipe, was positioned 1.07 m from each end of the 4.57-m long strand of steel pipe and maintained in a vertical orientation throughout backfill placement and compaction. The clayey native soil in the trench walls held a vertical cut and geofoam blocks were atop the pipe and against the trench sidewalls (Fig. 9). However in

Fig. 9. Photograph of full-scale uplifttest in progress (Lingwall, 2009).

Figure 8. EPS cover system tested by Lingwall (2009).

14

some places, the trench walls did not fit tightly against the geofoam blocks. In these locations, bedding sand was used to fill the gap between the blocks and trench side walls. To complete the EPS cover system, a pre-poured 152-mm reinforced concrete load distribution slab was placed atop the EPS29 and covered by 52.5-mm of granular compacted fill. The load distribution slab was included in the test because the planned pipeline alignment would be positioned under existing urban highway, a reinforced concrete slab is required by UDOT standard drawings.

Instrumentation for this test consisted of total pressure cells placed atop the EPS and string pod potentiometers attached to rebar tell-tales to measure displacement. The string pods were suspended by 51-mm x 152-mm lumber frames that were founded outside the zone of uplift so that absolute uplift measurements could be made. For each uplift test, a 996 kN truck mounted crane was used and the rigging was connected to the inline tension load cell to measure the total uplift force that developed in the cable. All instrumentation from the load and pressure cells and the potentiometers were connected to a datalogger for continuous data acquisition. During the test, the pipe was uplifted as slowly as possible to obtain a reasonable record of the force-displacement history. Uplift continued until the pipe had completely uplifted through the cover system.

A comparison trench-cover system was constructed and tested using compacted native soil in place of the EPS system. This was similarly instrumented and brought to failure in uplift. This comparison showed that the EPS cover system was reasonably efficient in reducing the total uplift force (Fig. 10). The native soil backfill system reached a peak force of 520 kN at a displacement of 0.06 m. In contrast, the EPS cover system reached a peak force of 136 kN at a displacement of 0.18 m. Thus, the total uplift force had been reduced by 73 percent by using the EPS cover system instead of native soil backfill. Additional numerical modeling of these and other tests are discussed in Lingwall, 2009.

Based on these tests and evaluations, we conclude that a light-weight cover system constructed of EPS geofoam offers significant benefits in protecting buried steep pipelines from the deleterious effects of vertical PGD such as that produced by normal and reverse faulting. Because of its extremely low mass density, an EPS cover system significantly reduces the vertical and/or uplift forces on a pipe system undergoing vertical PGD. Full-scale testing and numerical modeling have demonstrated that the total force on the pipe is reduced by a factor of about 3 to 4 when compared with a soil cover. Furthermore, we believe that an EPS cover system constructed atop a pipe that undergoes horizontal PGD is beneficial, as is the case for strike-slip faulting. Horizontal displacement tests performed by Lingwall (2009) suggest that the total force on the pipe can be reduced by a factor of about two, even if the pipe is pushed horizontally into a sand backfill. This benefit occurs because the light-weight cover system constructed of EPS significantly reduces the vertical stresses on the pipe and sand, which in turn reduces resistance of the soil to horizontal pipe displacement. Thus, we conclude that the application of light-weight cover systems constructed primarily of EPS block have beneficial effects for buried pipes undergoing both vertical and horizontal ground displacement.

15

Fig. 10. Force-displacement plots for pipe uplift tests (Lingwall, 2009).

0

100

200

300

400

500

600

0 0.05 0.1 0.15 0.2 0.25 0.3

Displacement (m)

Fo

rce

(kN

)

Geofoam Section

Soil Section

16

References

AASHTO LRFD Bridge Design Specifications, Customary U.S. Units, 5th Edition, with 2010 Interim Revisions.

Bartlett, S., Negussey, D., Kimble, M., and Sheeley, M. (2000). “Use of Geofoam as Super-Lightweight Fill for I-15 Reconstruction.” Transportation Research Board 79th Annual Meeting, Washington, D.C.

Bartlett, S. F. and Lawton E. C. (2008). “Evaluating the Seismic Stability and Performance of Freestanding Geofoam Embankment,” 6th National Seismic Conference on Bridges and Highways, Charleston, S.C., July 27th – 30th 2008, 17 p.

Bartlett, S.F., Lingwall, B. N., Trandafir, A. C. and Lawton E. C. (in press). “Protection of Steel Pipelines from Permanent Ground Deformation Using EPS Geofoam,” in Increasing the Seismic Resilience of Natural Gas Systems - Select Topics of Interest, ASCE Technical Council and Lifelines and Earthquake Engineering, 33 p.

Choo, Y. W., Abdoun, T. H., O’Rourke, M. J. and Da, H. (2007). “Remediation for buried pipeline systems under permanent ground deformation,” Soil Dynamics and Earthquake Engineering 27 (2007) 1043-1055.

Elragi, A.F. (2000). “Selected Engineering Properties and Applications of EPS Geofoam.” PhD Dissertation. State University of New York College of Environmental Science and Forestry, Syracuse, NY.

Horvath, J. S. (1995). Geofoam Geosynthetic, published by Horvath Engineering, P.C., Scarsdale, New York, U.S.A.

Horvath, J. S. (2004). “A Technical Note Regarding Calculating the Fundamental Period of an EPS-Block-Geofoam Embankment,” Report No. CGT-2004-1, Manhattan College, School of Engineering, Center for Geotechnology, 10 p.

Itasca Consulting Group, Inc. (2005). “FLAC: Fast Lagrangian Analysis of Continua: Structural Elements, Version 5.” Minneapolis, Minnesota.

Lingwall, B. (2009). “Protection of Buried Pipelines from Permanent Ground Displacement Using EPS Geofoam,” Dissertation, Department of Civil and Environmental Engineering, University of Utah.

Lingwall, B. and Bartlett, S. F. (2007). “Conceptual Design and Modeling: EPS Geofoam Cover System for Buried Pipelines,” prepared for Questar Gas Corporation, Salt Lake City, Utah, Aug. 2007, 22 p.

Mejia, L. H. and Dawson, E. H. (2006). “Earthquake Deconvolution in FLAC,” 4th International FLAC Symposium in Numerical Modeling in Geomechanics, 2006 Itasca Consulting Group Inc., Minneapolis MN.

17

NCHRP Report 529, “Guideline and Recommended Standard for Geofoam Applications in Highway Embankments, Transportation Research Board, Washington, D.C., 2004, www.TRB.org

Negussey, D. and Sun, M.C. (1996). “Reducing Lateral Pressure by Geofoam (EPS) Substitution”. Proceedings of the 2nd International Conference on EPS, Tokyo, 1996.

Riad, H. L. and Horvath, J. S. (2004). “Analysis and Design of EPS-Geofoam Embankments for Seismic Design,” ASCE Geo-Trans 2004, July 27-31, Los Angeles, California.

Sun, J. I., Golesorkhi, R. and Seed, H. B. (1988). “Dynamic Moduli and Damping Rations for Cohesive Soils,” Earthquake Engineering Research Center, University of California, Berkeley, Report No. UCB/EERC-88/15, 42 p., 1988.

Sun L., Hopkins, T. C., and Beckham, T. L. (2009). “Reduction of Stresses on Buried Rigid Highway Structures Using the Imperfect Ditch Method and Expanded Polystyrene (Geofoam),” Research Report KTC-07-14/SPR-228-01-1F, Kentucky Transportation Center, University of Kentucky, Lexington Kentucky, 49 p.

Yoshizaki K. and Sakanoue, T. (2003). “Experimental Study on Soil-Pipeline Interaction Using EPS Backfill,” Proceedings of the Pipelines 2003 International Conference on Pipeline Engineering and Construction, Baltimore, MD, July 13-16, 2003, pp. 1126-1134.