Design of Piles for Downdrag (2024)

SECTION 1

Literature Review

A review of the previous report prepared for NCHRP Project 12-116, “Proposed AASHTO Specifications for Design of Piles for Downdrag” (Rollins 2019), and the broader literature was completed by the project team for other projects related to drag load. Research conducted through NCHRP, FHWA, and other national, state, and pooled-fund-sponsored research is included. The literature review contains information about projects and investigators who have performed and published the findings from full-scale, centrifuge, or modeling tests.

1.1 NCHRP Project 12-116, Literature and Synthesis Report

The previously produced report (Rollins 2019) details many of the investigations on the effects of downdrag on piles that have been conducted over the past 50 years. The review included investigations on single piles and pile groups for compressible clays and liquefiable soils. The investigations discussed in the most detail in the report are summarized in Table 1.1.

Rollins (2019) carefully explores the concept of downdrag and relevant field investigations in the literature. Additionally, as discussed in the synthesis of the literature review, there is (1) a disagreement between some field investigations and case histories regarding the development of a static neutral plane (specifically, several case histories have not observed this phenomenon), and (2) the case histories that do exist are often deficient in critical data (e.g., lack of load test or soil settlement relative to the pile settlement).

Although Rollins (2019) mentions other design methodologies, the focus is solely on the unified pile design method (Fellenius 1984, 1988, 2006, 2017, 2018, 2019; Fellenius et al. 2004). It does not appear to examine the mechanics of other design methodologies or extensions of the unified pile design method. The investigations discussed in Rollins (2019) have been included in the following literature review with an expanded context in view of these observations.

1.2 Additional Investigations

Investigations not discussed in the Rollins (2019) report but important to the literature are presented in Table 1.2. These investigations are referenced in the literature review that follows.

1.3 Literature Review

1.3.1 Downdrag Design of Piles

Design of deep foundation elements for geotechnical “capacity” or “resistance”—for instance, the AASHTO strength limit states [also known as ultimate limit state (ULS) design]—is considered a relatively mature process with ample, well-described procedures based on the considerable and

Table 1.1. Investigations detailed in NCHRP Project 12-116 Literature and Synthesis Report [adapted from Rollins (2019)].

Table 1.2. Investigations not included in NCHRP Project 12-116, Literature and Synthesis Report (Rollins 2019).

collective experience of the profession (e.g., Hannigan et al. 2016; AASHTO 2020). Conversely, design of deep foundations for the serviceability limit state (SLS), used to estimate the pile head deflection for a given service load, relies on load transfer analyses. Ample procedures exist for load transfer analyses (e.g., Coyle and Reese 1966; Coyle and Sulaiman 1967; Vijayvergiya 1977; Kraft et al. 1981; Adami and Stuedlein 2015; Li et al. 2017); however, there is limited familiarity with the procedure, and the typical bridge designer may not have sufficiently high confidence to adopt load transfer analysis. For SLS design, an analysis of load transfer, prescribing the rate of load reduction with depth as a function of axial pile compression, is necessary to evaluate the relationship between global pile head displacement and applied loading. In this design setting, it is acknowledged that axial pile compression results in the downward movement of the pile section relative to the soil surrounding the pile, resulting in positive side resistance (positive skin friction). The analysis of load transfer requires the prescription of a relationship between the relative axial displacement and the unit side resistance and toe-bearing resistance, which depends on factors such as pile interface type (e.g., steel, concrete), installation procedure (e.g., cast-in-place, driven), and sequence of installation [predrill and casing or casing then excavation (Li et al. 2017)].

Despite limited adoption, analysis of load transfer has been shown over the last several decades to be the key to understanding the development of negative side resistance (negative skin friction) that results in drag load. Drag load is understood to be the axial load induced within a driven pile or drilled shaft when side resistance is mobilized by soil moving downward relative to the shaft of the foundation element (Fellenius and Siegel 2008), also known as downdrag movement. Johannessen and Bjerrum (1965) recognized early the existence of downdrag and the resulting drag load. Subsequent investigations, such as Bozozuk and Labrecque (1969) and Endo et al. (1969), provided additional confirmation regarding the mechanics of downdrag. These early researchers suggested downdrag was a consolidation phenomenon governed by changes in the effective stresses, and the resulting drag load should be treated as a geotechnical design load. Crawford (1969) synthesized the early research efforts and concluded

• Negative side resistance (negative skin friction) can develop after small relative movements between the soil and the pile;

• Negative side resistance (negative skin friction) will develop in the upper portion of the pile, and positive side resistance (positive skin friction) will develop in the lower portion of the pile owing to the toe-bearing resistance developed under mobilized toe displacement;
• There must be an equilibrium between the sum of positive side resistance (positive skin friction) and end bearing and negative side resistance (negative skin friction) and the applied axial load;
• Location of the equilibrium changes with time due to loading or changes in effective stress; and
• Cyclic or transient loads may contribute to reversals in the direction of side resistance.

These early works have proven to be insightful; unfortunately, some of the conclusions included in Crawford’s (1969) synthesis led to oversimplifications by later researchers, and some of Crawford’s (1969) conclusions have been widely ignored. Specifically, most design approaches assume negative side resistance (negative skin friction) is always fully mobilized and do not account for changes in the location of the neutral plane as a function of time.

1.3.2 Consolidation-Induced Downdrag

Fellenius (1984, 1988) and Fellenius et al. (2004) proposed the now widely accepted design methodology to determine the magnitude of the drag load via the unified pile design method. The unified pile design method implements the concept of a neutral plane at which the pile is in both force and displacement equilibrium with the surrounding soil. The unified pile design methodology states that

• The geotechnical axial resistance (i.e., for the strength or ULS) is governed by positive side resistance (positive skin friction) along the entire length of the pile;
• The maximum structural load in the pile occurs at the neutral plane and is a function of drag load and sustained (or dead) load on the pile only (i.e., it excludes live or otherwise transient compressive loads); and
• The downdrag of the pile (i.e., axial compression) is equal to the settlement of the soil at the location of the neutral zone.

The greatest benefit of the neutral plane method is that it specifically addresses the need, or more precisely, the lack of need for the consideration of the drag load in the assessment of geotechnical ULS (or strength limit state) resistance. If the combined drag load and sustained loading are greater than the existing geotechnical resistance, the pile will begin to compress elastically and downward relative to the soil. As the pile compresses, additional geotechnical resistance will be developed from the toe and/or newly mobilized positive side resistance (positive skin friction). Specifically, the neutral plane will begin to shift toward the ground surface until sufficient positive side resistance (positive skin friction) has been developed, leading to a new and higher elevation of the neutral plane location. Therefore, drag load cannot lead to plunging (Poulos 1997) and should not be considered a “load” in the traditional sense (i.e., applied to the pile head). Recently, practitioners have begun using the term “drag force” to describe the drag load (Siegel et al. 2013). Unfortunately, drag force is still used as a load in some standards, including the AASHTO LRFD [Load and Resistance Factor Design] Bridge Design Specifications (AASHTO LRFD BDS) (AASHTO 2020). This study aimed to improve these specifications.

The “traditional method” for the design of deep foundations in the presence of settling soils recommended in the AASHTO LRFD BDS (AASHTO 2020) specifies a double counting of the applied drag load (Abu-Hejleh et al. 2010). Specifically, AASHTO (2020) states that

• The geotechnical resistance above the lowest soil layers expected to contribute to downdrag should be neglected, and
• The calculated drag load should be treated as an additional axial load.

This process is overly conservative, costly, and not consistent with the known physics controlling load transfer (Fellenius 1988; Poulos 1997). The consensus of the geotechnical profession, as documented in the deep foundation literature, is that downdrag (soil compression) and drag load [arising from negative side resistance (negative skin friction)] are best treated as an SLS design check, as recommended in the basic tenets of the unified pile design method (Fellenius 1988). Although AASHTO LRFD BDS allows for the use of the neutral plane method, little to no guidance is given in the standard for the implementation of this approach (AASHTO 2020).

Fellenius (1988) demonstrated that the location of the neutral plane could be determined by balancing the dead load and drag load with positive side resistance (positive skin friction) and the end bearing, assuming the end bearing was fully mobilized. The soil settlement at the neutral plane was then calculated based on an added stress increment within the soil near the neutral plane. The relative displacement difference between the soil and the pile at the neutral plane is zero; therefore, the soil settlement at the neutral plane equals the pile settlement. The neutral plane method has since been adapted by Poulos (1997, 2008), Siegel et al. (2013), Wang and Brandenberg (2013), and Sun et al. (2013), as described herein.

Poulos (1997, 2008) proposed a simple analytical method to determine pile “stability” based on embedment in stable soils. However, the method assumes full mobilization of side resistance and overestimates the drag load and resulting pile settlement. Despite being overly conservative, the proposed methodology may have some merit for further investigation as a potential suitable design methodology, upon necessary modification. Additionally, Poulos (2008) concluded “normal” live load is not sufficient to reverse the direction of negative side resistance (negative skin friction). Therefore, the long-held perception that live load is offset by a reduction in drag load may not be suitable. Specifically, the results of a field investigation concluded that “drag load, dead load, and live load were observed to exist in repeated tests and be additive in nature” (Dasenbrock et al. 2012). Unfortunately, most standards including procedures summarized by Hannigan et al. (2016) and AASHTO (2020) allow for the discounting of live loads in downdrag calculations.

Siegel et al. (2013) developed a neutral plane method within the LRFD framework and included a simplified method to account for differing degrees of end bearing (a simplified t-z methodology). As indicated by the Endo et al. (1969) investigation, the toe response is a contributing factor to the location of the neutral plane. The Siegel et al. (2013) method recognizes and accounts for the large impact that toe mobilization can have on the location of the neutral plane, the magnitude of drag load developed, and total pile settlement. The Siegel et al. (2013) method was adopted by Hannigan et al. (2016) and provides a well-structured design guide for the neutral plane. However, this procedure requires the use of fully mobilized side resistance, which can result in an overestimation of drag load.

Wang and Brandenberg (2013) identified some of the shortcomings associated with the neutral plane method and proposed a modified neutral plane model using an embedded vertical beam in a nonlinear Winkler foundation. Primarily, Wang and Brandenberg (2013) determined the assumption that the neutral plane is located where the pile and soil relative displacement is zero was inaccurate and that the neutral plane is located where the relative velocity between the pile and the soil is equal to zero. Additionally, Wang and Brandenberg (2013) demonstrated how the neutral plane method only considers the final location of the neutral plane, but because the location changes as excess pore pressures dissipate, the neutral plane can underpredict or overpredict the actual pile settlement. Wang and Brandenberg (2013) proposed a simplified design methodology based on the degree of consolidation. However, this approach is difficult to distill into everyday design calculations and will be particularly difficult to implement for post-liquefaction settlements in which the initial excess pore pressure regime would be required

to be estimated and for which three-dimensional drainage governs the dissipation regime (presenting an extremely difficult design challenge).

A major criticism of the neutral plane methodology is that it assumes side resistance can be represented by either fully mobilized positive side resistance (positive skin friction) or negative side resistance (negative skin friction) and that there is no smooth transition side resistance where its polarity reverses. However, near the neutral zone, partially mobilized side resistance will exist as the resistance changes from positive to negative. To address this shortcoming, Sun et al. (2013) proposed generalized neutral plane design profiles and implemented “transition zones” for the mobilization of side resistance. AASHTO (2020) and Hannigan et al. (2016) both use a “trigger value” of 0.4 inches (10 mm); however, the literature suggests even a few millimeters is sufficient to initiate downdrag (Crawford 1969; Fellenius 1988; and Siegel et al. 2013). Indeed, a common supposition of the neutral plane method is, due to the low threshold of mobilization, all deep foundations undergo downdrag in response to soil creep. Unfortunately, these suppositions are largely based on laboratory scale models of relatively rigid, small-scale piles reported by Hanna and Tan (1973). In the small-scale models, the extreme contrast in soil-pile stiffness required the very soft clay material serving as the compressible soil necessarily produced fully mobilized side resistances under just millimeters of movement. In reality, many soil-pile (or soil-shaft) interfaces harden to significant relative soil-pile interface movements (e.g., Adami and Stuedlein 2015; Li et al. 2017). The use of transition zones in the design profiles proposed by Sun et al. (2013) represents a more realistic load distribution. However, the generalized nature of the design profiles means the profiles should not be used to design critical structures or foundations prone to downdrag.

1.3.3 Liquefaction-Induced Downdrag

Boulanger and Brandenberg (2004) applied the neutral plane method to the problem of liquefaction-induced downdrag and produced an empirical approach to calculate the side resistance in liquefied zones while allowing for increased strength due to pore pressure dissipation. Fellenius and Siegel (2008) applied the unified pile design method to liquefaction and concluded liquefaction-induced downdrag was primarily a serviceability problem. Conversely, Vijayaruban (2014) considered several design cases not considered by Fellenius and Siegel (2008), specifically, multiple liquefiable zones and liquefaction at the pile toe. Most recently, liquefaction-induced downdrag on piles was investigated by Ishimwe et al. (2018), who used blast-induced liquefaction testing of full-scale instrumented piles. Significantly, Ishimwe et al. (2018) provided a closed-form method to determine the location of the neutral plane following liquefaction and determined the preliquefaction neutral plane was located at the ground surface even though the piles were not loaded to maximum resistance. Ishimwe et al. (2018) provided validation of findings presented in Vijayaruban (2014) and was contradictory of other existing understanding of the neutral plane. Specifically, prior to liquefaction, no negative side resistance (negative skin friction) developed, and the neutral plane was located at the ground surface for these axially loaded piles.

1.3.4 Pile Group Downdrag

The neutral plane method is typically discussed in reference to single piles; however, piles are most often installed within groups for transportation infrastructure. There is a consensus within the literature that pile groups reduce downdrag in piles; however, there is little agreement with how the group effects are quantified. Fellenius (1988) applied the neutral plane method to small (narrow) pile groups and perimeter piles of large (wide) pile groups. In contrast, Fellenius (2019) concluded interior piles of wide pile groups are not affected by side resistance

and, by extension, are not affected by downdrag [this is attributed to the stiffer response of the perimeter piles relative to interior piles, a phenomenon observed in several field investigations, including Okabe (1977) and Russo and Viggiani (1995)]. Jeong and Briaud (1992) developed reduction factors for pile groups (both narrow and wide) based on a three-dimensional finite element model. For piles spaced at 2.5 diameters, Jeong and Briaud (1992) predicted reduction factors of 0.15 to 0.5. However, Lee et al. (2001) surveyed several investigations that considered the effect of downdrag on both narrow and wide pile groups. Lee et al. (2001) concluded (1) the reduction factor for downdrag in pile groups compared to single piles was typically overestimated by existing methods (which still represents most of the work in quantifying downdrag group effects), and (2) reduction factors from 10% to 40% and 30% to 70% were observed for 3 × 3 (narrow) pile groups and 5 × 5 (wide) pile groups, respectively. The question of magnitude of pile group downdrag is not resolved and requires further investigation. Unfortunately, AASHTO (2020) and Hannigan et al. (2016) do not include guidance to reduce downdrag in pile groups.

1.3.5 Downdrag Mitigation

Generally, evaluations of downdrag and the resulting drag loads have been investigated to determine whether mitigation is necessary, and if so, which methods should be used to reduce the drag load. Bjerrum et al. (1969) proposed the use of bitumen coatings to reduce the soil-pile interface side resistance, which was further expanded by Briaud and Tucker (1997). However, the use of bitumen has recently begun to lose favor among practitioners, citing a lack of effectiveness and the associated costs. Alternative friction-reducing coatings, like plastic sheeting, have also been investigated (Tawfiq and Caliendo 1995). However, Budge et al. (2015) demonstrated that the use of coatings and sleeves simply alters the load distribution rather than changes the magnitude of the drag load. Alternatively, Hannigan et al. (2016) identified preloading or increasing the size and number of piles as mitigation methods for consolidation-induced downdrag. In a similar vein, Briaud and Tucker (1997) recommended using “a curtain of dummy piles” for large pile groups because interior piles experience less downdrag.

1.3.6 Existing Design Standards

As previously mentioned, the traditional method for the design of downdrag and the associated drag load has been to neglect side resistance above the deepest consolidating soil layer and to consider the drag load as a factored permanent load. This is the method predominantly adopted by the AASHTO (2020) design standards. The traditional method is also presented in the Eurocode (2004) design procedure. While the AASHTO (2020) code permits the use of the neutral plane method as presented by Briaud and Tucker (1997), the standard offers limited commentary or consistency within the code to facilitate implementation. The majority of the AASHTO (2020) downdrag design procedures were influenced by Allen (2005), which developed geotechnical strength load factors for drag loads, a natural extension of the traditional method for an LRFD framework.

Several geotechnical design manuals have adopted some version of the neutral plane method including the Canadian Foundation Engineering Manual (Canadian Geotechnical Society 2006), Hong Kong’s Foundation Design and Construction (Hong Kong Geotechnical Office 2006), and the FHWA Design and Construction of Driven Pile Foundations (Hannigan et al. 2016). Some state DOT offices have adopted the neutral plane method; the method is included in the Minnesota DOT (MnDOT) 2017 Geotechnical Engineering Manual (MnDOT 2017). In contrast to the traditional method and the work done by Allen (2005), the neutral plane method [as proposed by Fellenius (1984, 1988)] recognizes drag load does not affect the geotechnical strength limit state.

The developing geotechnical consensus favors the neutral plane methodology over the traditional method.

1.4 Summary

Based on discussions with personnel in numerous state DOTs and based on the available literature, the knowledge gaps identified include an incomplete understanding of the following items: (1) soil settlement (as a function of depth within the soil profile); (2) pile toe movement following development of drag load; (3) side shear mobilization; (4) pile stiffness, especially for composite piles; (5) the location of the neutral plane at any time; (6) when to use the neutral plane method; (7) the quantity and quality of data required to achieve the project objectives; (8) appropriate use of load combinations in LRFD design; and (9) pile group response when subject to downdrag.