Impact of the Employed Soil Model on the Predicted Behaviour of Integral Abutment Bridges

Transcription

1 Impact of the Employed Soil Model on the Predicted Behaviour of Integral Abutment Bridges Hany El Naggar & Ahmed Mahgoub Department of Civil and Resource Engineering Dalhousie University Nova Scotia, Canada, B3H 4R2 François Duguay & Arun J. Valsangkar Department of Civil Engineering University of New Brunswick New Brunswick, Canada, E3B 5A3 ABSTRACT Integral Abutment Bridges (IAB) are structures in which both of the superstructure and the substructure move together into and away from the backfill to accommodate translations and rotations. These bridges have no expansion joints and no bearings. Integral abutment bridges are characterized by their complex soil-structure interaction of the abutment walls and foundations as they are subjected to cyclic loading on a daily and seasonal basis. The Mohr-Coulomb (MC) soil model is an elastic-perfectly plastic model, which is often used to model soil behaviour in general and serves as a first-order model. Several researchers have used the MC soil model in the modelling of IAB due to its simplicity and minimum required information. However, as the backfill soil of such bridges is subjected to a loadingunloading-reloading process during which the stiffness of the soils keeps on changing, the soil stress-strain behavior is not linear and the linear elastic perfectly plastic Mohr-Coulomb model is not able to accurately describe the behavior of the soil during the unloading-reloading process. Consequently, a more advanced soil model such as the Hardening soil model is more appropriate. This paper investigates the impact of the employed soil model on the predicted stresses in the soil mass as well as on the predicted earth pressures on the abutment s wall. The computed results are compared with available field data for an integral abutment bridge in New Brunswick to assess the predictability of each of the used soil models. RÉSUMÉ Les ponts à culées intégrales sont des structures où la superstructure et la substructure bougent ensemble en s approchant et en s éloignant du remblai pour accommoder les mouvements latéraux et de rotations. Ces ponts ne comportent aucuns joints d expansion ou d appareils d appuis. Les ponts à culées intégrales sont caractérisés par l interaction sol-structure complexe de la culée et de la fondation lorsqu ils sont sujet à des charges cycle sur une base journalière et saisonnière. Le modèle Mohr-Coulomb (MC) est un modèle élastique parfaitement plastique qui est utilisé pour modéliser le comportement du sol et il sert de modèle de première-ordre. Plusieurs chercheurs ont utilisé le modèle MC pour la modélisation de PCI du à sa simplicité et au nombre restreint d informations requises. Par contre, lorsque le remblai d un PCI est sujet à un processus de chargement-déchargement-rechargement, la rigidité du sol varie. Le comportement contrainte-déformation n est pas linéaire et le modèle linéaire MC ne réussit pas à décrire adéquatement le comportement du sol lors du processus de chargement-déchargement-rechargement. Conséquemment, un modèle plus sophistiqué tel que le model d écrouissage du sol est nécessaire. Ce rapport examine l impact du modèle utilisé sur les contraintes prédites sur le massif et sur les poussées de terre contre la culée. Les résultats sont comparés avec des données de terrain d un pont à culées intégrales situé au Nouveau-Brunswick. 1 INTRODUCTION The use of integral abutments to accommodate thermal expansion of bridges is common practice in New Brunswick, and is also seen throughout Canada. The major advantage of using this construction method is the elimination of the expansion joint and bearing which are expensive to install and maintain during the lifetime of the bridge. The seamless transition between the roadway and the bridge that is created when using an integral abutment also eliminates the impact load at the ends of the superstructure that would normally be present due to the expansion joint. During daily and seasonal temperature change, the structure undergoes expansion and contraction, moving toward and away from the backfill. This movement in turn creates variations in stresses in the soil mass located behind the bridge, in earth pressure variations acting against the abutment, and in the stresses in the supporting piles. In order to better understand the effects of the thermal movement of the bridge on the soil-structure interaction, a 2-dimensional finite element analysis was done. This paper will look at the effect of the employed soil model on the predicted earth pressure within the soil mass, as well as the earth pressure acting against the abutment. The abutment will undergo cyclic loading, and the results over the cycles will be compared to field data from a monitored structure in New Brunswick. 2 INTEGRAL ABUTMENT BRIDGES In typical short and medium span bridges, bearings and expansion joints are used to accommodate the expansion

2 and contraction of the bridge due to the varying daily and seasonal temperature changes. Bridge expansion joints require costly maintenance and replacement operations during the lifetime of the bridge. They can become full of dirt or debris, and might even start leaking; creating another maintenance problem as salt and de-icing chemicals run down against the end of the girders. Bearings also need to be frequently maintained, as they generally do not last the lifespan of the bridge. Bearing replacement requires jacking up of the bridge as well as possible restriction on the structure or in some cases complete closure. Integral abutment bridges (IAB) on the other hand present numerous advantages compared to typical bridges. The elimination of the expansion joints and bearings reduce significantly the cost of maintenance of the bridge through its life, as well as reducing the construction cost (Clayton et al. 2006). The construction is accelerated as only a single row of straight piles is need for each abutment, instead of multiple rows of battered piles. The simple geometry of the abutment typically results in simpler formwork and faster construction. The beams are also better protected as the ends are encapsulated in concrete and hidden behind the concrete deck. (Clayton et al. 2006). One of the challenges that IABs present, compared to a typical bridge sitting on bearings, is that the expansion and contraction of the whole structure against the soil mass is a complex soil-structure interaction problem. The cyclic nature of the thermal load, mixed with the different design approach for the abutment backfill makes it hard to estimate the earth pressures and behavior of the structure over time (Springman et al. 1996; Wood and Nash 2000; Cosgrove and Lehane 2003; Clayton et al. 2006). Many theories and design guidelines have been brought forward by different engineers and researchers (Kunin and Alampalli 2000; Hassiotis et al. 2005), but a consensus on the proper analysis method is yet to be achieved. 2.1 Integral Abutment Bridges in New Brunswick The New Brunswick Department of Transportation and Infrastructure (NBDTI) has been building integral abutments bridges since NBDTI reports that Harper Road Overpass was the first IAB built in New Brunswick Located in the parish of Chatham, the bridge is a one span reinforced concrete structure spanning 33 meters. Since 1995, there have been 52 Integral and Semi-Integral bridges constructed in New Brunswick. The bridges length varies between 24 and 133 meters. Steel and concrete superstructures have been used with success. The bridges skews range from 0 degree to 26.5 degree. For the design of IABs, the NBDTI refers to the Ministry of Transportation of Ontario (MTO) handbook (Husain and Bagnariol 1996) as well as a variety of research articles (A. Thorne, personal communication, 2007, and S.Mayo, personal communication, 2013). The 2015 NBDTI Structural Design Standard now limits the maximum length for an IAB to 100 meters. The bridge skew is also limited to an angle of 20 degrees if the foundation conditions are suitable for an IAB. The NBDTI Standard detail for IAB backfill is shown in Figure 1. Compression foam and geocomposite systems are installed directly behind the concrete abutment. Reinforced earth is then installed using geogrid installed at every 600mm. Borrow D, which consists of 75mm crushed rock, is used to backfill between the layers of geogrid. Finally, the approach slab is installed on top of a layer of 31.5mm crushed gravel with 2 layers of 6 mil polyethylene between the slab and the crushed gravel (NBDTI standards and specifications 2015). Borrow A Materials Geocomposite drainage sys. Compression Foam Borrow D Materials Geogrid- Layer/ 600mm Figure 1. IAB Standard Backfill Detail as mentioned NBDTI Industrial Park-Route 7 Underpass The field data used in the comparison with the obtained 2D finite element model (FE-Model) results were taken from a bridge constructed in The Industrial Park-Route 7 Underpass was monitored by Huntley (2009) as part of her Doctoral work. The structure is a two-span bridge using prestressed reinforced concrete beams for the super structure. As reported in Huntley (2009), the superstructure consists of height New England Bulb-Tee 1800 girders (NEBT1800) spaced at 2200mm. A 225mm reinforced concrete deck is cast-compositely with the girders. The total width of the superstructure is mm of the height. East and West abutment respectively are 4405mm and 4593mm and both were 1200mm thick. A single row of 12 HP310x132 piles oriented along their weak axis was used to support each abutment. The top 3000mm of the piles were installed in a 1200mm wide trench backfilled with loosely placed sand material to allow easier movement of the structure during thermal movement. A simple layer of 1500mm of free draining granular backfill was used behind the abutment. The backfilling of Borrow A materials as in Figure 2 was used behind the drainage layer. Borrow A has a dust content not exceeding 25% and does not consist of mudstone, claystone and (or) siltstone or of any rocks mixed with these materials or with clayey or silty soil. In addition, lab tests like direct shear box and sieve analysis were performed in order to define the materials characteristics. The backfilling Borrow A materials were compacted in the site in 200 mm lifts to a minimum of 95% of the maximum dry density, which averaged 2200 and 2170 kg/m 3. No geofoam, geocomposite drainage system

3 or reinforced earth is present. Finally, a 300mm reinforced concrete approach slab was cast-in-place, before asphalting the bridge deck and the roadway (Huntley (2009)). The East abutment dimensions in Figure 3 were used to create the FE-Models. 2.3 Instrumentation Over a hundred different sensors were installed at various locations on the east abutment. The data from deformation meters, tilt meters, contact earth pressure cells, and strain gauges were used. This paper will primarily focus on the development of a comprehensive finite element model of the east abutment simulating the variation of the earth pressure due to the thermal cyclic loading/unloading displacements (Huntley and Valsangkar 2013). Three vibrating-wire contact pressure cells were installed along the abutment centerline at elevations of approximately 1/4, 1/2, and 3/4 of the abutment height (Figure 3) in order to measure the lateral earth pressure acting on the east abutment as a result of the walls movement due to the thermal expansion and contraction of the superstructure. Additionally, the deformation/ transition and rotation for the abutment were measured using two vibrating-wire deformation meters and two vibrating-wire tilt meters which were bolted to the back face of the abutment and positioned horizontally and vertically at approximately 1/3 and 2/3 of the abutment height with the tilt meters positioned similarly (Figure 3) (Huntley and Valsangkar 2013). Additional details on the deformation and tilt meter instrumentation has been reported in (Huntley (2009)). 2.4 Summary of field observation The field observed data was measured during the research period of about three years. Figure 4 shows the transitions of the two monitored points along the east abutment height by upper and lower deformation meters. Figure 5 similarly shows the earth pressure for the three monitored points along the abutment height by the pressure cells. Each data point represents the average measurement during a 1- week period with changing of the temperature. Figure 3. The instrumentation layout for east abutment (Huntley and Valsangkar 2013) Away from backfill Toward backfill Aug-2004 Feb-2005 Sep-2005 Mar-2006 Oct-2006 Apr-2007 Nov-2007 Figure 4. The change of abutment transition with time (Huntley and Valsangkar 2013) Figure 2. Profile and dimensions of the east abutment and backfilling materials in millimeters (Huntley and Valsangkar 2013). Aug-2004 Feb-2005 Sep-2005 Mar-2006 Oct-2006 Apr-2007 Nov-2007 Figure 5. The change of earth pressure with time due to temperature variation (Huntley and Valsangkar 2013).

4 3 DEVELOPMENT OF FINITE ELEMENT MODEL Numerical analyses were conducted to examine the fashion in which earth pressures are generated behind an integral abutment as the structure was subjected to cyclic thermal loading/unloading displacements. The developed models were verified against the measured field data. Additionally, the developed FE models were used to study the effect of the employed soil model on the predicted earth pressures and its ability to simulate the applied cyclic loading and the accompanying soil structure interaction. 3.1 Description of FE-Model Two numerical models using PLAXIS 2D were developed. The first model was utilized to study the earth pressure generated on the east abutment due to the cyclic thermal displacements using the hardening soil model. The hyperbolic Hardening Soil model (Schanz et al. 1999) from the PLAXIS 2D library was used to model the stressdependent variation of stiffness of the soil materials due to the changing of the loading directions on the abutment. The adopted hyperbolic hardening soil model is from the double-stiffness models family, which is a refinement of the Duncan and Chang (1970) hyperbolic model. The hyperbolic hardening soil model used in this study supersedes the Duncan and Chang model because it uses the theory of plasticity rather than the theory of elasticity. This Hardening Soil model accounts for the soils dilatancy, and utilizes a yield cap. This first model followed the same geometry, layout, and construction procedure of the full-scale abutment and the results were compared with the field-measured earth pressures due to the changing of the temperature. The second model aimed to study the effects of using Mohr coulomb soil model on the results of the earth pressure distributions. Therefore, the second model is similar to the first one in geometry, layout and construction procedures with using a linear elastic-perfectly plastic Mohr-Coulomb constitutive model in defining the different types of soils. Crushed Stone East Abutment Loose sand Row of H-Piles Approach slab Drainage materials Borrow A Clayey Sand Mudstone Figure 6. The used FE-Model (PLAXIS 2D) in the study 3.2 Materials and soil properties The different parameters of subsurface soil layers were extracted from Huntley (2009) as part of her Doctoral work. Borrow A backfilling material was simulated in FE-Model according to the laboratory tests results of direct shear box and sieve analysis. Table 1 shows the different soil properties in the first FE- Model (by Using Hardening Soil Model) and Table 2 shows the properties in the second model (by Mohr Coulomb Model). The reinforced concrete abutment was modeled as a volume element using the elastic model as shown in Table 3. Additionally, the H-steel piles rows are simulated using an equivalent plate elements assuming elastic behavior and using the axial and the flexural rigidities based on the used H-steel beam dimensions (I=93.7x10-6, A=16.7x10-3 ) and the internal spacing between the piles (1160mm) to mimic the same configuration as in the case study. Table 4 shows the used piles parameters in the finite element models. The process of calibration was performed by refining the interface properties in the FE-Models, especially, between the inner abutment surface and the drainage layer and between the approach concrete slab and the backfilling layers. This was done by adjusting the values of the interface reduction factor values and the virtual thickness values. Table 1. The soil parameters using Hardening soil model (Huntley 2009) Material Ø 0 C (Kpa) E 50 (Mpa) E ode (Mpa) E ur (Mpa) ɣ (KN/m 3 ) Crushed stone Borrow A Drainage- Layer Clayey sand Mud-stone Loose Sand Table 2. The soil parameters using Mohr Coulomb model (Huntley 2009) Material Ø 0 C (Kpa) E (Mpa) ɣ (KN/m 3 ) Crushed stone Borrow A Drainage Layer Clayey sand Mud-stone Loose Sand Table 3. The elastic model parameters for the concrete (Huntley 2009). Material E (Mpa) ɣ (KN/m 3 ) υ (poisons ratio) Concrete

5 Elevation (Y/H) Table 4. The used parameters of H-steel piles (Huntley 2009). Material EI (KN/m 3) EA (KN/m) υ (poisons ratio) Pile 16.20x x Construction sequence The staged construction technique was used to simulate the construction process for the case study as follows: 1. The initial in-situ stress was simulated using the Ko procedure in which the initial geostatic stresses were established assuming increasing vertical stress with depth (σv = ɣ Z) and horizontal stresses based on σh = Ko σv. The groundwater table was assumed to be below the considered profile. 2. The excavation works for the watercourse. 3. Activate the backfilling lifts behind the abutment until the abutment foundation level at Installation of steel pile works and the loose sand trench of 1200mm width. 5. Completing the construction works of the concrete abutment until the finishing level Completing the backfilling works behind the abutment until the bottom level of crushed stone at level Placement of the 450 mm crushed stone layer. 8. Pouring of 300mm reinforced concrete approach slab. 9. Apply the cyclic displacement due to the effect of the thermal change on the superstructure (Contraction or expansion) on the exact locations of upper and lower transition sensors in the inner face of the abutment. Table 5 shows the average cycle displacement per each year. Table 5. The average cyclic displacement for different sensors during the three years (the period of study) Displacement Time (mm) Upper Lower Displacement Direction sensor sensor Jan Away from the backfilling July Towards the backfilling Feb Away from the backfilling July Towards the backfilling Feb Away from the backfilling July Towards the backfilling During the first year, the earth pressure due to the contraction (during the winter) is ranging between 0.80 and 12.0 KPa at the locations of the three observed points. However, during the expansion mode (during the summer) the earth pressure increases to be ranging from 56 to 54 KPa with seasonal variation of 55.2Kpa in the upper observed point and 42 KPa at the lower one. Consecutively, the variation of the earth pressure in the upper and lower points in the second year increases to be 58.2 and 49.8 KPa respectively. Similarly, in the third year the variation increases to by 62.5 and 58.4 KPa. Figures 9 and 10 show the error percentage in estimating of the earth pressure for the two considered soil model respectively. Figure 9 indicates that the maximum values of the error (%) in estimating the earth pressure of the first FE- Model is 11.5% which appears due to the third cyclic displacement at the location of the third point (at elevation ). However, the maximum value of the error (%) in points 1 and 2 is 2% or less. On the other hand, the percentage of error (%) in estimating earth pressure in the second model by using Mohr Coulomb is varying from 14% to 23.7% for point 1, from 2% to 30.3% for point 2 and from 12% to 34% for point 3 as shown in Figure 10. Therefore, Using Hardening Soil model which was used to model the stress-dependent variation of stiffness of the soil materials is the most suitable way in simulating and predicting the soil structure interaction due to the cyclic loading Field readings (1st Year) Plaxis (1st Year) Field readings (2nd Year) Plaxis (2nd Year) Field readings (3rd Year) Plaxis (3rd Year) Results and discussion In this section, the results of the FE- Model s including the earth pressure distribution along the inner face of the abutment were estimated due to the thermal cyclic displacements during the observation period (three years) and compared these results with the observed field data at the locations of the three pressure cells along the abutment. Figure 7 shows the comparison between the earth pressure at the locations of pressure cells and the results of the first FE-Model using the Hardening Soil model. Similarly, Figure 8 reports the comparison but using the Second FE-Model with the Mohr Coulomb soil model Seasonal Change of earth pressure (KPa) H= Total Height of abutment and Y= point depth Figure 7. The comparison between the observed Field earth pressure and the results of FE- Model with Hardening Soil Model. 3

6 Error% Error % Elevation (Y/H) Field readings (1st Year) Plaxis (1st Year) Field readings (2nd Year) Plaxis (2nd Year) Field readings (3rd Year) Plaxis (3rd Year) Seasonal Change of earth pressure (KPa) H= Total Height of abutment and Y= point depth Figure 8. The comparison between the observed Field earth pressure and the results of FE- Model with Mohr Coulomb Soil Model st Year 2nd Year 3rd Year Point 1 Point 2 Point 3 Figure 9. The error (%) in estimating earth pressure in the first FE- Model by using Hardening Soil model st Year 2nd Year 3rd Year Point 1 Point 2 Figure 10. The error (%) in estimating earth pressure in the second FE- Model by using Mohr Coulomb model SUMMARY AND CONCLUSION Detailed finite element analyses using PLAXIS 2D were conducted to examine the manner in which earth pressures are generated behind an integral abutment due to cyclic displacements. The analyses considered the same geometry, layout, and construction procedure of the fullscale Industrial Park-Route 7 Underpass bridge abutment, located in New Brunswick, to simulate the complex soilstructure interaction and to validate the results of the numerical model. This study showed that the elastic-perfectly plastic Mohr Coulomb soil model is not capable of simulating the soil behaviour during the cyclic loading conditions, as the soil stiffness is nonlinear and stress-dependent, whereas this model utilizes a constant stiffness. On the other hand, the Hardening Soil Model that accounts for the stressdependent variation of stiffness was capable of simulating effectively the non-linearity of the soil behaviour. The conclusions of the study can be summarized as follows: - The Hardening Soil model can simulate effectively the non-linearity of the soil behaviour due to the cyclic loading. - The maximum values of error % in estimating earth pressure by using the Hardening soil model are 11.5% at point 3 and 2% for points 1 and 2. - The maximum values of error % in estimating earth pressure by using the Mohr Coulomb soil model is varying from 14% to 23.7% for point 1, from 2% to 30.3% for point 2 and from 12% to 34% for point 3. REFERENCES Clayton, C.R.I., Xu, M., and Bloodworth, A A laboratory study of the development of earth pressure behind integral bridge abutments. Géotechnique, 56(8): doi: /geot Cosgrove, E.F., and Lehane, B.M Cyclic loading of loose backfill placed adjacent to integral bridge abutments. International Journal of Physical Modelling in Geotechnics, 3(3): doi: /ijpmg Hassiotis, S., Khodair, Y., and Wallace, L.F Data from full-scale testing of integral abutment bridge. In Proceedings of the Transportation Research Board 84th Annual Meeting, Washington, D.C., 9 13 January Transportation Research Board, Washington, D.C. Huntley, S. A Field Performance and Evaluation of an Integral Abutment Bridge. Ph.D. Thesis. University of New Brunswick, Canada. Huntley, S.A., and Valsangkar, A.J Field monitoring of earth pressures on integral bridge abutments. In Canadian Geotechnical Journal, 2013, 50(8): , /cgj Husain, I., and Bagnariol, D Integral abutment bridges. Structural Office Report SO-96-01, Structural Office, Ministry of Transportation, Ont.

7 Kunin, J., and Alampalli, S Integral abutment bridges: Current practice in United States and Canada. Journal of Performance of Constructed Facilities, ASCE, 14(3): doi: /(ASCE) (2000). NBDTI Standard specifications. The New Brunswick Department of Transportation and infrastructure, Fredericton, N.B. PLAXIS 2D [Computer software]. PLAXIS BV, Amsterdam, Netherlands. Schanz, T., Vermeer, P. A., and Bonnier, P. G. (1999). The hardening soil model: Formulation and verification. Beyond 2000 in computational geotechnics 10 years of Plaxis, Balkema, Rotterdam, Netherlands, Springman, S.M., Norrish, A.R.M., and Ng, C.W.W Cyclic loading of sand behind integral bridge abutment. Transportation Research Laboratory Report 146, Transportation Research Laboratory, Wokingham, Berkshire, UK. Wood, D.M., and Nash, D Earth pressures on an integral bridge abutment: a numerical case study. Soils and Foundations, 40(6): doi: /sandf. 40.6_23.