1 Introduction

Section:

ChooseTop of pageAbstractNotation1Introduction <<2Current design practice3Field monitoring of IBs4Previous laboratory expe...5Plexus pump-priming expe...6Conclusions

A large investment is continually made in infrastructure globally. For example, the UK infrastructure pipeline has committed £12.6 billion for roads and £46.2 billion for rail (IPA, 2016). Internationally, the global pipeline is estimated to be as high as US$97 trillion (GI Hub, 2020). In transportation networks, bridges represent a particularly vulnerable element to which the largest investment is linked (e.g. NIST, 2015; Nuti et al., 2009; Rasulo et al., 2004; Thoft-Christensen, 2012). In traditional bridges, joints and bearings have emerged as the main source of bridge maintenance problems and costs (Greimann and Wolde-Tinsae, 1986; Wolde-Tinsae and Greimann, 1988) due to the cyclic displacements caused by thermal gradients, traffic and dynamic loads, while both corrosion of and access to bearings provide particular maintenance challenges. Thus, integral bridges (IBs) are becoming increasingly attractive because of reduced maintenance issues at the bridge deck–abutment interface compared with traditional bridge construction.

IBs have received increasing attention by designers in the past few decades and are widely used in many countries for small- to medium-span highway bridges and overcrossings (Burke, 2009). They now constitute a significant part of the transportation infrastructure stock, with an estimated number in service of over 9000 in the USA alone (Fiorentino et al., 2021; Paraschos and Amde, 2011; White et al., 2010). IBs are likewise becoming more widely used in the UK, Europe and Asia (Bloodworth et al., 2011), while their design varies according to practices and requirements outlined by regional transportation authorities. In the USA, each state highway department has its integral abutment programme and has established guidelines concerning their design and construction. The specification of the American Association of State Highway and Transportation Officials (Aashto, 2012) is the most widely accepted IB design guideline in the USA, providing performance criteria for IB design. Parallel design guidance is provided in the Canadian Foundation Engineering Manual (CGS, 1978).

In the UK, PD 6694-1 (BSI, 2011) and Design Manual for Roads and Bridges (DMRB) by the Highways Agency (HA, 2003) are currently the reference guides, with these documents referring to European standards (BSI, 2004, 2007) and Ciria Report 760 on embedded retaining walls (Gaba et al., 2017) for relevant design parameters. Currently, the span lengths of IBs are limited due to the lack of an adequate evidence base on which to predict their performance. There is consequent conservatism in design guidance (Baptiste et al., 2011; Dicleli et al., 2003; Mitoulis et al., 2016; Zordan et al., 2011a, 2011b), which, in turn, reflects imperfect understanding of the behaviour of these structures under the imposed loads. In the UK, spans are limited to 60 m length and 30° skew (BSI, 2011; HA, 2003). Although design codes offer provision for the static design of IBs (BMVBS, 2013; Gaba et al., 2017), the use of IBs is limited mainly by the lack of explicit design guidelines coherent across different countries.

The rationale for the limitations in design codes lies mainly in the uncertainty related to the soil–structure interaction between the backfill and the abutment walls when IB decks expand due to seasonal thermal loads under ambient conditions (Gorini and Callisto, 2017, 2019; Huffman et al., 2015). When the bridge expands, substantial force is exerted on the abutment by the soil reaction, and this can significantly impact the integrity of the structure. Such inherently non-linear soil action is dependent on the magnitude and distribution (with height) of the wall displacement, which encompasses both translational and rotational displacements depending on the boundary conditions. In the longer term, as seasons of cyclic expansion and contraction of the bridge decks occur, there can be a build-up of significant lateral earth pressure behind the abutments (Figure 1(a)). This asymmetric cyclic stress–strain behaviour is known as ratcheting (Cui and Mitoulis, 2015; England et al., 2000; Horvath, 2004). However, soil conditions can vary from relatively loose to dense states with different compaction levels. Consequently, the pressure that builds up behind the abutment could significantly increase with time – by a factor of 4 or more. Accordingly, axial forces on the bridge deck may increase as well, by a factor of around 2 (part of the pressure being absorbed by the abutment foundation), while bending moments in the composite deck may increase by somewhat less than the axial forces, depending on soil stiffness and boundary conditions (Clayton et al., 2006; Fennema et al., 2005; Mahjoubi and Maleki, 2020; Shamsabadi et al., 2007). The thermal loading also causes ground settlement adjacent to abutments (which may be under approach slabs, if present), with gaps often observed at the surface between the abutment and backfill (Figure 1(b)). Moreover, subsidence behind the abutment wall can cause structural problems in approach slabs if the bending loads due to traffic are significant (Muttoni et al., 2013).

Figure 1 (a) Build-up of lateral earth pressures behind the abutment under thermal loading; (b) gapping between abutment wall and backfill under cyclic thermal loading

A number of mitigation measures are available to reduce the excessive lateral pressures on the abutment walls. These include limiting the bridge length, skew and vertical penetration of abutments into embankments; using selected granular backfill (Al-Ani et al., 2018); providing approach slabs to prevent vehicular compaction of the backfill (Muttoni et al., 2013); using embankment benches to shorten wingwalls; and using suspended turn back wingwalls (Paraschos, 2016).

Moreover, semi-integral abutment designs (Figure 2) are used to remove passive pressures under bridge seats. In such designs, the end screen wall and deck beams are integral with each other, but the end screen wall does not provide support to the deck beams. Instead, a structure with bearings, which can accommodate horizontal displacement, is provided as support to the deck beams.

Figure 2 Semi-integral abutment design (BSI, 2011)

Compressible inclusions between the abutments and the backfill, such as expanded polystyrene geofoam, have also been proposed to mitigate the build-up of earth pressures and uncouple the response of the bridge from that of the backfill (Fiorentino et al., 2021; Horvath, 2000, 2005; Mitoulis et al., 2016; Mylonakis et al., 2007a). However, extra design and construction work should be allowed for at the design stage. Compressible inclusions between the abutment and the backfill allow dissipation of the lateral earth pressures and the control of displacements in the backfill in performance-based design (AbdelSalam and Azzam, 2016; Karpurapu and Bathurst, 1992).

It has also been observed that even if the superstructure responds linearly elastically under thermal loads (which is anticipated in IBs), the local non-linear material behaviour of the backfill could result in triggering a non-linear response in the entire soil–bridge system (McCallen and Romstad, 1994). Recurrent cyclic traffic loads during IB operation (assuming that there is no bridging slab) further compact the backfill and may also contribute to increases in lateral earth pressures. These effects can be replicated through mechanistic models that have been proposed for the numerical modelling of soil–structure interaction effects on IB abutments (Kappos and Sextos, 2009; Kloukinas et al., 2012; Kotsoglou and Pantazopoulou, 2007, 2009; Zhang and Makris, 2002). The influence of soil–wall separation and gapping on static wall pressures has been investigated in a recent paper by Efeoglu et al. (2021).

A field study of an IB equipped with an elastic inclusion (i.e. a layer of elastic material between the abutment and the retained soil) showed significantly reduced lateral earth pressures and tolerable settlements of the approach fill (Hoppe, 2005). This isolated system exhibited a mirrored behaviour, with increasing pressure effects occurring at each consecutive thermal cycle, while the backfill soil displacements showed a settling effect with a decreasing magnitude with an increasing number of cycles. An important finding was that both the developed pressures and the associated displacements were smaller than those in the conventional system: the peak pressures were seven times smaller and the settlement around four times smaller. Due to the lower absolute pressures and an approximately linear pressure distribution behind the abutment, the overall bending moments induced on the abutment walls were also greatly reduced. This approach may, therefore, lead a more sustainable solution to span longer distances (Caristo et al., 2018). However, without explicit adoption in codes, elastic inclusions are unlikely to achieve widespread use and there remain maintenance implications that can make this solution less appealing.

Reducing or removing uncertainties/barriers and improving the functionality of IBs, throughout their design, construction, operation and maintenance phases, provide a means of reducing infrastructure costs and increasing their value. This can be achieved by better diagnosis (i.e. developing know-how on the problem to reduce epistemic uncertainty) and feeding research findings from laboratory experiments, modelling and field-monitoring campaigns into national and international design code development (Dhar and Dasgupta, 2019). In support of this and similar goals, the UK Collaboratorium for Research on Infrastructure and Cities (UKCRIC, 2022) has recently created a suite of world-leading laboratory facilities combining multidisciplinary research teams with systems thinking and practice approaches to enhance the value of, and de-risk investments in, infrastructure and urban system interventions.

The UKCRIC–Priming Laboratory EXperiments on Infrastructure and Urban Systems (PLEXUS) programme included a project that combined three of the new national facilities and a variety of research approaches to establish a comprehensive picture of the soil–structure interaction behaviour of IB abutments under lateral loading. This included a small-scale experimental campaign to complement the evidence base available in the literature, which is reported herein following a brief review of current international design practices (Section 2), an introduction to previous field monitoring techniques (Section 3) and a review of previous research on IBs (Section 4). The results from the experimental campaign (Section 5) are then presented and discussed in relation to Sections 1–4, along with conclusions and plans for large-scale experimental tests (Section 6).

2 Current design practice

Section:

ChooseTop of pageAbstractNotation1Introduction2Current design practice <<3Field monitoring of IBs4Previous laboratory expe...5Plexus pump-priming expe...6Conclusions

The design of IBs varies according to practices and requirements stipulated by local transportation agencies; a brief summary is presented in Table 1. The US (Aashto, 2012), Canadian (CGS, 1978) and UK (BSI, 2011; HA, 2003) guidance are perhaps the most authoritative.

Table 1 Summary information of design guidance

For abutment design, the earth pressure distribution behind the abutment is determined using a depth-dependent lateral earth pressure coefficient, K (Mei et al., 2017; Vahedifard et al., 2015), defined as the ratio of the effective lateral (horizontal) effective stress to the effective vertical stress at a specific depth. The value of K depends on many parameters, notably the nature of the soil (coarse grained against fine grained), its density and its loading history (overconsolidation ratio). There are three categories of horizontal earth pressure coefficient: at rest (K ₀), corresponding to zero horizontal wall movement and zero normal horizontal soil strain; active (K _a), representing a theoretical minimum value requiring sufficient outward wall displacement (i.e. away from the backfill); and passive (K _p), representing a theoretical maximum value requiring sufficient inward wall displacement (towards the backfill). There are several, theoretical and empirical, theories for establishing the lateral earth pressure coefficient. Coulomb (1773) first proposed a heuristic limit analysis framework (known today as the limit equilibrium method) associated with shear failure of a soil wedge within the backfill, using an optimisation procedure to identify stationary values among an infinite set of candidate lateral thrusts. Mayniel in 1808 extended Coulomb’s equations to include wall friction, and Müller-Breslau in 1906 further generalised Mayniel’s equations to incorporate an inclined backfill and wall. Coulomb’s solution provides the most useful tool for establishing earth thrusts by hand calculations, yet the method works solely with forces (not stresses) and thus cannot establish the point of application (elevation) of the overall soil thrust. Subsequently, the theory by Rankine (1857), based on limit stresses to predict active and passive pressure coefficients, produced exact stresses (hence predicting the point of application of soil thrust). However, its applicability is limited – notably to vertical walls with roughness equal to the inclination of the backfill under plane strain conditions – while the kinematics of the problem (i.e. soil and wall displacements) and the compatibility of deformations are essentially ignored. More advanced stress solutions encompassing inclined backfill and wall are available in the paper by Mylonakis et al. (2007b).

Importantly, the use of full passive pressures without regard to displacements and compatibility of deformations is not conservative, as it invariably suppresses the flexural effects of dead and live loads on the bridge girders. Modified coefficients based on Rankine’s solution have been proposed (Hanna and Diab, 2016; Kloukinas et al., 2015; Pain et al., 2017; Rajesh and Choudhury, 2017). For relatively short single-span IBs, the passive earth pressure coefficients were reduced by multiplying the relevant Rankine coefficients with modification factors. The displacement at the top of the bridge abutment due to thermal loading, d, is calculated using the following equation (BSI, 2011):

$$d=\alpha L_X\left(T_{\text{e};\text{max}}- T_{\text{e};\text{min}}\right)$$

1

where α is the coefficient of thermal expansion of the deck (on the order of 10⁻⁵ for concrete decks); L _X is the expansion length measured from the end of the bridge to the position on the deck that remains stationary when the bridge expands; and T _e;max and T _e;min are the characteristic maximum and minimum uniform bridge temperature components for a 50-year return period in the UK (National Annex to BS EN 1991-1-5 (BSI, 2003)), respectively.

The specified dimensionless displacement (‘drift’, which is the horizontal movement applied on the top of the abutment wall from span over the abutment wall height) for full passive pressure development is equal to approximately 4 × 10⁻² for loose sand and equal to 1 × 10⁻² for dense sand (Clough and Duncan, 1991). Widely used curves for determining the lateral soil pressure for loose, medium and dense granular materials (Figure 3) are presented in NCHRP Report 343 (Barker et al., 1991) and Naval Facilities Engineering Systems Command (Navfac, 1982), while design curves are provided in the Canadian Foundation Engineering Manual (CGS, 1978) and the US Section of the Navy (Cole and Rollins, 2006).

Figure 3 Relationship between wall displacement and lateral earth pressure in sand: (a) wall moving towards the backfill (passive-like conditions) and (b) wall moving away from the backfill (active-like conditions) according to Barker et al. (1991) and Navfac (1982)

Barker et al. (1991) and Navfac (1982) recommend applying limit equilibrium solutions based on log spiral failure mechanisms for standard backfill configurations, since Coulomb’s failure wedge methodology is notoriously non-conservative for determining passive pressures (Keykhosropour and Lemnitzer, 2019; Xu et al., 2018). Aashto (2012) determines horizontal soil pressures on bridge abutments according to Rankine’s active soil pressures, based on variations in the earth pressure coefficient as a function of structural displacement from experimental data and finite-element analyses, leading to a practically linear relationship as shown in the following equation (Bal et al., 2018; Capilleri et al., 2020):

$$K=K_0+\varnothing d\le K_{\text{p}}$$

2

where d is the displacement of the IB towards the backfill and Ø is the slope of the earth pressure variation with horizontal displacement (which varies with the backfill type).

There is reasonable agreement between the predicted average passive earth pressure of standard compacted gravel backfill with the results from full-scale wall tests performed at the University of Massachusetts (Bonczar et al., 2005). According to the tests, the pressure coefficient K (MassDOT, 1999) can be estimated using the following empirical equation:

$$K=0.43+5.7\left[1-\exp \left(-190\frac{d}{H}\right)\right]$$

3

where d is the displacement of the IB towards the backfill soil and H is the height of the abutment. The first term in Equation 3 (0.43) can be interpreted as a K ₀ coefficient, while the multiplier (5.7) on the second term can be interpreted as the difference between passive and at-rest pressures (K _p − K ₀), being zero for zero displacement and maximum (5.7) for infinite displacement. To achieve a pressure equal to 99% of the active pressure requires a dimensionless displacement (drift) of approximately 2.4%. These values correspond to the case of a rough wall and a medium-dense granular backfill.

From experiments investigating the cyclic stresses in backfill soil on a concrete wall with a pinned connection to a strip footing (Firoozi et al., 2016; Yazdandoust et al., 2019) and according to PD 6694-1 (BSI, 2011) for full-height abutments on spread footings, which accommodate thermal movements by rotation or flexure, the earth pressure on the retained face for an integral abutment wall is dependent on (a) the thermal movement range (based on a 50-year return period), (b) the direction of movement (expansion or contraction) and (c) the magnitude of expansion or contraction for the combination of actions for the design situation under consideration (Figure 4). The design value of the earth pressure coefficient for expansion

$K_{\text{d}}^{*}$

can be estimated from Equation 4 but should not be taken as greater than K _p;t:

$$K_{\text{d}}^{*}=K_0+{\left(\frac{C{d}_{\text{d}}^{\text{'}}}{H}\right)}^{0.6}{K}_{\text{p};\text{t}}$$

4

where K ₀ is the coefficient of earth pressure at rest; H is the vertical distance from the ground level to the level at which the abutment is assumed to rotate;

$d_{\text{d}}^{\prime }$

is the wall deflection at depth H/2 below ground level; C is a dimensionless coefficient equal to 20 for foundations on loose soils with Young’s modulus E ≤ 100 MPa and 66 for foundations on rock or soils with E ≥ 100 MPa and which may be determined by linear interpolation for values of between 100 and 1000 MPa; and K _p;t is the coefficient of passive earth pressure used in the calculation of

$K_{\text{d}}^{*}$

.

Figure 4 Earth pressure distributions for abutments that can accommodate thermal expansion by rotation and/or flexure (from BSI, 2011), where K ₀ is the coefficient of earth pressure at rest; H is the vertical distance from the ground level to the level at which the abutment is assumed to rotate; γ is unit soil weight; z is the soil depth; and K* is the design value of the earth pressure coefficient for expansion

Design guidance for IBs is developing worldwide with a particular focus on the effect of thermal loading. However, there are still unanswered questions related to longer-term effects due to cyclic loading, notably does the earth pressure distribution change after many years of thermal cycling loading, and should the cycling loading history be considered in the estimation of the lateral pressure distribution behind the abutment?

It is evident, therefore, that the design methods provided in guidelines are characterised by significant uncertainty on the degree of conservatism embedded in the methods and lack consistency between countries. This emphasises the need for further investigation both under controlled conditions in laboratories and through monitoring of IBs in the field, notably focusing on the abutment and associated backfill behaviour during a large number of repetitive cycles of displacement.

3 Field monitoring of IBs

Section:

ChooseTop of pageAbstractNotation1Introduction2Current design practice3Field monitoring of IBs <<4Previous laboratory expe...5Plexus pump-priming expe...6Conclusions

Field monitoring data from in-service IBs are significant both to inform and improve future design guidance and to refine experimental and numerical research. Many IBs globally are currently being monitored; see Table 2 for an overview of major monitoring studies. One of the monitored integral abutment bridges (IABs) is the Manchester Road Overbridge between Denton and Middleton: two side-by-side 40 m span IBs with no skew and 7 m high abutments carrying the A62. The strain in, and the earth pressure acting on, the abutments was monitored during construction and throughout the first 2 years of service. The first bridge was a conventional portal frame structure retaining granular backfill, while the second was constructed with contiguous bored pile abutments founded on glacial till. As the bridge deck expanded, lateral stresses increased, demonstrating a strong correlation between lateral stresses and bridge temperature (Barker and Carder, 2000).

Table 2 Summary information of monitored integral abutment bridges

Similarly monitored was a two-span, skewed IB of 50 m total length consisting of prestressed concrete beams and cast in situ deck structurally connected to full-height, 9 m high abutments founded in magnesian limestone over the M1–A1 Link Road at Bramham Crossroads, North Yorkshire. The field measurements (displacements of the abutment and deck, strains in the abutment and deck, earth pressures on the abutment) were recorded during construction and over the first 3 years of service. The measured lateral earth pressures after backfilling were consistent with predictions using the coefficient of earth pressure at rest (K ₀), calculated based on the estimated friction angle (ϕ′), while they increased slightly for each of the following summers (Barker and Carder, 2001).

The results from both monitoring campaigns were invaluable, yet they cover only a relatively short period after construction, whereas longer-term monitoring would be needed to determine the expected pattern of significant earth pressure escalation after more seasonal cycles. This would also have provided a better return on investment from the instruments installed (Barker and Carder, 2000, 2001).

A 15° skew, two-span IB of 91 m total length with 2.9 m high abutments in Trenton, NJ (USA), was monitored for a year. Abutment strains and soil pressures behind the abutment were monitored by the New Jersey Department of Transportation when revising the design specifications for IBs (Hassiotis et al., 2005). A steady build-up of soil pressures behind the abutment was observed.

A no-skew three-span IB of 82.3 m total length with 3.05 m high abutments spanning the Millers River between the towns of Orange and Wendell, USA, was monitored from 2002 to 2004, including longitudinal and transverse bridge displacements, backfill pressure distribution behind the abutment and abutment strains. The peak earth pressure at 2.5 m from the abutment top was observed to increase annually from 245 kPa (2002) to 280 kPa (2003) and 315 kPa (2004). Similar earth pressure increases were observed at other depths behind the abutment (Breña et al., 2007).

The Van Zylspruit River Bridge (a five-span IB of 90.45 m total length with no skew and 6.6 m high abutments, located on the N1 in South Africa) exhibited a maximum earth pressure significantly (∼1.75 times) higher than the at-rest pressure (Skorpen et al., 2018).

It is clear from the published literature that there is no more than 10 years of reliable monitoring data available for IBs, whereas backfill stress measurements are required for a bridge that has been in service for more than a decade (Lock, 2002). It is currently unclear whether earth pressures would continue to increase, or increase asymptotically, or level off to a steady value towards the end of the service life of the bridge (implying a hypothetical need for a 120-year observation period (Yap, 2011)). Longer-term monitoring campaigns, notably focusing on the lateral soil pressure behind the abutment and incorporating redundancy to allow for instrumentation failures, are therefore essential. Linking the data feeds to digital twins would further improve understanding of IB behaviour, improve designs and inform maintenance strategies.

5 Plexus pump-priming experimental campaign

Section:

ChooseTop of pageAbstractNotation1Introduction2Current design practice3Field monitoring of IBs4Previous laboratory expe...5Plexus pump-priming expe... <<6Conclusions

To investigate the soil–structure interaction uncertainties related to the backfill behaviour behind IBs and establish the efficacy of different sensing technologies, the Plexus 1 g small-scale soil box experimental campaign was devised, thus also paving the way for experimentation at or near full scale in the Soil-Foundation-Structure Interaction Laboratory at the University of Bristol.

The Plexus rig was designed to simulate the effect on the backfill from abutment displacements due to seasonal expansion and contraction of the bridge deck. The monitoring regime included lateral stresses behind the abutment wall (pressure cells), the backfill surface displacement (linear voltage differential transformers (LVDTs)) and backfill soil deformation behind the abutment (particle image velocimetry (PIV)). Initial tests included the backfill material being loaded by a moving abutment wall having two different relative stiffnesses (i.e. flexible and rigid abutments), the displacements replicating horizontal thermal loading conditions associated with increasing cyclic displacements and multiple-cycle constant-displacement histories.

5.1 Experimental configuration

The 1525 × 1050 × 1150 mm test box accommodated the loading system (actuator) and a 1000 × 1000 × 1000 mm specimen of backfill. A 1000 mm high movable wall was hinged at the bottom of the soil box to simulate an IB abutment able to rotate about its base. The movable wall consisted of a 25 mm poly(methyl methacrylate) (PMMA) and 25 mm timber composite to simulate a flexible abutment wall, while the rigid movable wall consisted of a 25 mm PMMA, 25 mm timber composite, 50 mm aluminium frame and 25 mm timber composite producing a sandwich configuration. PMMA was used for the box wall to enable PIV observations of backfill displacements, while the remainder of the rig was designed without metal components to facilitate future trialling of a ground penetration radar as a monitoring tool (see Figure 5). The abutment wall, end wall and side wall were instrumented with pressure cells, while LVDTs were used to measure surface backfill displacements. The backfill consisted of uniform LBS fraction B (see Fiorentino et al., 2021).

Figure 5 (a) Annotated three-dimensional diagram of the test box; (b) test box filled with LBS

5.2 Thermal loading

Thermal loading from temperature-induced cyclic expansion and contraction of the bridge deck was simulated by push–pull pseudo-static motion of the movable wall, its displacement being controlled by the actuator mounted 870 mm above the wall base. In test 1, the flexible abutment wall was subjected to 12 loading cycles with a loading rate of 0.5 mm/s (to simulate static thermal loading (Lehane, 2011; Springman et al., 1996)), each cycle lasting at least 40 s. The cyclic displacements at the top of the movable wall started at ±5 mm, with increments of ±5 mm every two cycles, to reach ± 30 mm (drift ≈ 3.5 × 10⁻²; see Table 4). In test 2, the rigid abutment wall was subjected to 59 ‘loading’ cycles with a loading rate of 1.0 mm/s and cyclic displacements at the top of wall fixed at ±30 mm for each cycle (1 mm/s was considered slow enough to simulate static thermal loading). The 30 mm equals the seasonal deck movement (one end) of an IAB in London with a 131 m long concrete deck or a 91 m long steel deck (England et al., 2000).

Table 4 Summary information for tests 1 and 2

Table 4 Summary information for tests 1 and 2

Test ID	Abutment wall	Total cycles	Loading rate: mm/s	Displacement history: mm; and drifts
1	Flexible PMMA + timbera	12	0.5	2 {±5 mm; ±10 mm; ±15 mm; ±20 mm; ±25 mm; ±30 mm} 2 {(±5.7; ±11.5; ±17.2; ±23.0; ±28.7; ±34.5) × 10−3}
2	Stiff sandwich sectionb	59	1.0	59 {±30 mm} 59 {±34.5 × 10−3}

a 25 mm PMMA + 25 mm timber

b 25 mm PMMA + 25 mm timber + 50 mm aluminium frame + 25 mm timber

View larger version

5.3 Instrumentation layout

The instrumentation employed consisted of TPC-4000 series total earth pressure cells (TEPCs) to measure lateral stresses and LVDTs and a high-resolution camera on the side of the test rig to measure displacements. The TEPCs are designed to measure total pressure (combined effective stress and pore water pressure) in soils and at soil–structure interfaces, yet, as the LBS was dry, they directly provided effective stress measurements. The locations of the three end wall TEPCs (1–3), four movable wall TEPCs (6–9) and side wall TEPCs (4 and 5) are shown in Figure 6.

Figure 6 Layout of the pressure cells in the test box: (a) movable wall; (b) end wall; (c) side wall including exact position of the actuator

Figure 7 shows the positions of the nine LVDTs (1–9) placed in three rows on the backfill surface and four LVDTs (11–13) measuring the movable (abutment) wall displacement. The high-resolution camera (Canon 70D 5472 × 3648 pixels) focused on the PMMA side wall to record ‘full-field’ backfill deformation using the PIV method, regarded as slow ‘fluid motion’ (Stanier and White, 2013).

Figure 7 Layout of the LVDTs: (a) plan view of the test box; (b) front view of the movable wall

5.5 Results

The two tests have multiple differences (i.e. wall stiffness, ‘loading’ rate, displacement history increasing or kept constant), making a systematic comparison difficult. Nevertheless, in Figure 8(a), a preliminary comparison is proposed between the earth pressure distributions obtained using Rankine’s theory and PD 6694-1 (see Figure 4 and Equation 4) and the passive-like pressure recorded at three instants: (a) at cycle 12 of test 1 (i.e. the second cycle at +30 mm after the increasing displacement history), (b) at cycle 2 of test 2 (i.e. the second cycle at +30 mm) and (c) at cycle 12 of test 2 (i.e. the twelfth cycle at +30 mm). The vertical distance from the ground level to the assumed point of rotation of the abutment (H) is 0.96 m, while the wall deflection at depth H/2 below ground level (

$d_{\text{d}}^{\prime }$

) was 0.7 times the horizontal movement of the end of the bridge deck. C is 20 assuming that the Young’s modulus of the sand is, on average, less than 100 MPa. The critical friction angle was taken as ϕ _cs = 32° and the elastic modulus as 20 MPa (Ng et al., 1998). The coefficient of passive earth pressure (K _p;t) was obtained through linear interpolation for values of ϕ _cs between 30° (K _p;t = 4.29) and 35° (K _p;t = 5.88) for a non-smooth vertical wall with δ/ϕ′ = 0.5 retaining a horizontal backfill (see table 8 in BSI (2011)).

Figure 8 (a) Measured earth pressures and (b) ratio of horizontal to vertical stress when the wall is moving into the backfill soil (passive)

The lateral soil pressures measured in both tests were larger than the pressures calculated according to Equation 4 from PD 6694-1 (see Section 2) using the maximum and minimum LBS densities. This may be caused by the specific configuration of the experiments and the influence of the backwall (as the distance between the abutment and the backwall was 1 m when at least 1.7 m would be necessary to develop a complete passive failure wedge). However, the shape of the lateral pressure distribution on the abutments, resulting from the three experimental measurement points, mimics the shape of the envelope proposed by PD 6694-1. The lateral pressure on the abutment after the 12th cycle in test 2 was larger than that after the 12th cycle in test 1, attributed to the larger cyclic lateral displacement in test 2 (an expected result) and/or the lower stiffness of the abutment wall in test 1 (also an expected phenomenon).

The lateral soil pressures on the movable wall in Figure 8(a) were normalised separately by the vertical stress, γz, where γ is unit weight and z is soil depth (Figure 9(a)), and by the maximum vertical stress, γH, where H is the total depth of soil (Figure 9(b)), thereby making them scalable for general translation. The vertical axes of Figure 9 are normalised by the total height of the backfill. The ratios of horizontal to vertical stress measured by the bottom pressure cell at cycle 12 in test 1 and at cycle 2 in test 2 were very close to that calculated from PD 6694-1, while the pressure at cycle 12 in test 2 is larger.

Figure 9 Normalised distributions of earth pressures with depth: (a) x-axis normalised by γz and (b) x-axis normalised by γH represented against the y-axis normalised by total height of the soil (H = 96 cm)

In test 2, the lateral pressure on the stiffer abutment after 12 cycles was larger than that after two cycles (all cycles consisting of ±30 mm displacement). Figure 8(b) shows the rapid increase in lateral soil pressure with increasing number of cycles and displacements (up to cycle 12) in test 1 and with constant displacement and increasing cycles in test 2, compared with the theoretical passive Rankine value and comparative values suggested by Barker et al. (1991) and Navfac (1982) shown in Figure 3(a).

Backfill surface settlement is significantly smaller for the stiffer abutment after two cycles than after 12 cycles (Figure 10). The settlements rapidly increased with the number of cycles. The settlement behind the stiff abutment after 12 cycles is slightly larger than that for the 12th cycle behind the flexible abutment, attributed to abutment stiffness and/or amplitude of cyclic loading.

Figure 10 Settlement behaviour of the backfill: (a) test 1 – 12th cycle; (b) test 1 – 12th cycle (zoomed-in); (c) test 2 – second cycle (zoomed-in); (d) test 2 – 12th cycle (zoomed-in). The black dashed line offers the same reference level for a better comparison

The densification of the backfill in the zoomed-in areas of Figures 10(b)–10(d) was analysed using the GeoRG PIV Matlab analysis package (Stanier et al., 2015). As shown in Figure 11 (where the volumetric strain value is shown with the colour of contour and the black arrows only present the deformation direction of the soil backfill), the deformation of the backfill after 12 cycles in test 1 (Figures 11(a) and 11(b)) was larger than that in test 2 when the actuator was at maximum extension. In contrast, the opposite behaviour was observed when the actuator was at its maximum contraction. The lower lateral soil pressure on the flexible abutment than that on the stiff abutment was attributed to the backfill becoming denser in the stiff abutment test. The densification of the backfill after two cycles in the stiff abutment configuration (Figures 11(c) and 11(d)) is much larger than after 12 cycles (Figures 11(e) and 11(f)), thus indicating that the amplitude of the backfill densification decreases with increasing loading cycles in the same test.

Figure 11 Percentage volumetric strain (ε) and deformation direction (black arrows) of the soil backfill: test 1 – 12th cycle (a) maximum extension position and (b) maximum contraction position; test 2 – second cycle (c) maximum extension position and (d) maximum contraction position; test 2 – 12th cycle (e) maximum extension position and (f) maximum contraction position

It is evident from these results that the abutment stiffness, number of cycles, backfill material state and magnitude of abutment horizontal displacement are key parameters in determining IB performance and warrant further investigation in controlled experiments. Pressure cells and PIV measurements are particularly suitable for comparing the performance of different test configurations.

6 Conclusions

Section:

ChooseTop of pageAbstractNotation1Introduction2Current design practice3Field monitoring of IBs4Previous laboratory expe...5Plexus pump-priming expe...6Conclusions <<

This paper discusses the behaviour of IBs under thermal loading by reviewing (a) current international design practices and (b) lessons learnt from previous experimental research including field monitoring. The soil pressure behind the abutment wall is a crucial factor for the design and performance of IBs, while the backfill behind the abutments significantly influences performance. The Plexus experimental campaign, described herein, demonstrated the efficacy of its monitoring processes in establishing the soil–structure interaction behaviour of IB abutments under thermal loading and identified key research needs. In particular, pressure cells and PIV provided useful data for performance comparisons in laboratory environments, while settlement measurements proved less informative. The research revealed a need for precise density monitoring of the backfill throughout the duration of the tests and all over the soil specimen. In a non-metallic Plexus test rig, this could be achieved by ground penetration radar, X-ray tomography and/or similar techniques.

The small-scale tests have created results that are suitable for analytical and numerical validation, although scaling effectiveness needs to be proven. Further, larger-scale physical model tests of IB abutments, using different types of backfill compacted to counter unwanted deformations due to ratcheting (e.g. incorporating rotation of principal stresses) and to cover the variety of materials likely in practice, are required to ensure improved sustainable and resilient designs of IBs.