The behavior of a foundation transversally loaded at the top over highly porous and collapsible soil

Comportamiento de cimentación cargada transversalmente en su cabeza en suelo altamente poroso y colapsable

Roberto Kassouf ^a, David de Carvalho ^b & Paulo José Rocha-de Albuquerque ^b

^a Kassouf Engineering, Itatiba, Brasil. kassouf.engenharia@kassouf.com.br
^b Universidade Estadual de Campinas, Campinas, Brasil. d33c@uol.com.br, pjra@fec.unicamp.br

Received: August 17^th, 2015. Received in revised form: February 15^th, 2016. Accepted: October 3^rd, 2016.

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Abstract
In geotechnical engineering, the use of caisson foundations transversally loaded at the top is common. This type of foundation is frequently used on high-porosity and collapsible soils, which are common in many regions in Brazil. Because of the limited information available in the literature, two loading tests were performed with transversal loading in a caisson built at full scale (0.8 m diameter, 9.0 m depth, 1.6 m base diameter). A load test was performed with the soil in its natural humidity condition, and another load test was performed after the previous soil flooding. Load vs horizontal displacement curves and the horizontal direction coefficient of the soil were obtained. The results permitted the verification of the applicability of behavior-predicting theoretical formulae in addition to the proposal of parameters for use on the studied soil. Before executing the load tests, laboratory and in-situ tests were performed to investigate the local subsoil.

Keywords: transversal loading, caisson, porous soil, collapsible soil.

Resumen
En la práctica de la ingeniería geotécnica es común el empleo de fundaciones por pilas cargadas transversalmente en la parte superior. Frecuentemente este tipo de fundación es ejecutada en suelos de alta porosidad y colapsables, comunes en varias regiones de Brasil. Debido a la poca información disponible en la literatura, se realizaron dos pruebas de carga con cargamento transversal en una pila (diám. 0,8 m; prof. 9,0m; diám. base 1,60m). Se realizó una prueba de carga estando el suelo en su condición de humedad natural y otra después de la previa inundación. Fueron obtenidas las curvas carga x desplazamiento horizontal y los coeficientes de reacción horizontal. Los resultados permitieron verificar la aplicabilidad de fórmulas teóricas de previsión de comportamiento, así como proponer parámetros a ser utilizados para el suelo. Previo a la realización de las pruebas de carga fueron realizados ensayos de laboratorio y campo para la investigación del subsuelo.

Palabras clave: carga transversal, pila, suelo poroso, suelo colapsable.

1. Introduction

In the dimensioning of foundations to resist horizontal loads, the project criteria involve not the capacity of the ultimate horizontal loading but the maximum or pre-established displacement that can be reached. Currently, there are many mathematical methods for the prediction of the horizontal displacement of a pile. The common difficulty among these methods concerns the adoption of the geotechnical parameters to be used in the calculations.

The main parameter used is the modulus of the horizontal reaction (n_h), which is defined as the soil resistance along the foundation divided by its deflection at a point. Simplified mathematical models have been created for the analysis because the modeling of the horizontal action problem is tridimensional and extremely complex for routine solutions by the project planners. The most well-known and diffused theory for the evaluation of these actions is the "Theory of the Horizontal Reaction of the Soil", where the n_h factor represents the proportionality between the reaction and displacement acting on the soil mass. However, this factor is difficult to theoretically estimate. Nevertheless, this factor can be "measured" using load tests and obtaining a reliable value for the horizontal resistance of the soil where the construction will occur.

The soil's reaction is a function of many factors, such as pile properties, soil stress vs. strain, depth of the analyzed point, foundation displacement level, etc. Because of the difficulty of establishing a function including all of these factors, the simplified Winkler's hypothesis is normally used, where the soil's reaction is proportional to the pile's displacement.

For the horizontal loading, in the first few meters, the superficial soil has great influence over the load vs. horizontal displacement of the foundation's behavior. To predict the behavior of horizontally loaded foundations, theoretical approaches are available in the literature [1, 2, 3, 4]; however, parameters determined for the local soil are necessary for its application.

Superficial soils with porosities higher than 50% cover great area extensions in Midwest Brazil. Because of their large void volumes, these soils undergo great deformations under loads. In addition, many of these soils are collapsible, i.e., when the soils are under loads and when a significant increase in the moisture content or soil saturation occurs, the structure collapses, which results in unacceptable displacement values for the edifices [5].

Given the lack of available information in the literature concerning horizontally loaded caissons on highly porous collapsible diabase soil, this study was developed to examine the performance of two load tests in a caisson, with soil at its natural and after-flooding.

Based on the horizontal loading tests, the effect of soil flooding on the load vs. horizontal displacement curve and the values for the horizontal reaction coefficient for the soil's conditions at its natural and pre-flooded were verified. The results obtained for the horizontal reaction coefficient were compared with the results for other types of foundations on similar soils.

The acting stresses and displacements on a pile under bending moments and horizontal loadings were determined using the theory of horizontal soil reaction, which is based on the model proposed by Winkler. The soil behavior under horizontal forces is simulated by a set of independent, identical and equally spaced springs. Thus, the soil's reaction is considered proportional to the displacement of the analyzed point. This supposition simplifies the problem, considering that the relation between the pressure of contact at the base of the foundation and its corresponding consolidation is the same for any support area. Using the model proposed by Winkler, the concept of the modulus of horizontal reaction, K, was introduced by [6]. It is defined as the relation between the soil's reaction (in units of applied force by the pile's length) and the corresponding displacement (eq. 1):

This notation presents the advantage of being independent of the diameter of the foundation. Therefore, the eq. 2 can be rewritten as:

where: k_h = the horizontal reaction coefficient (FL^-3) and D = the diameter of the foundation (L).

For pure sands, the elasticity modulus increases (approximately) linearly with depth. Therefore, the soil reaction to the load applied to the pile is assumed to also linearly increase with depth (eq. 3):

where: n_h = the modulus of the horizontal reaction of the soil (FL^-3) and z = the depth (L).

It is necessary to understand the variation of K throughout the foundation for the analysis of its behavior based on the theory of soil reaction. Refinements and sophistications in the reaction modulus function by depth are not justified because the errors in the results of the calculations are very small compared with the ones involved in the estimation of numerical values of the modulus of the soil reaction. [7] completely agree with this assessment because the results are satisfactory and can be obtained for the majority of practical cases as simple forms of variation of the reaction modulus with depth. Additionally, in practical problems, the uncertainty inherent to the soil's behavior estimate based on conventional tests is generally compatible with the small errors that can be introduced by the depth using a simple form of the soil reaction modulus function.

[8] were the precursors to the presentation of curves obtained for horizontal loads tests on sandy soils. They presented these curves in the form of n_h at the y-axis and the displacement y₀ at the x-axis. To create these curves, the cited authors used eq. (4) by [9] for the displacement for the application of only one horizontal load parallel to the ground surface, i.e.:

To calculate n_h, eq. 6 is used:

In the present study, the soil characteristics are analyzed using geotechnical laboratory tests and in-situ tests, with the aim of predicting its behavior in terms of deformability, resistance and collapsibility. To determine the horizontal reaction coefficient (n_h), the curves of the horizontal reaction coefficients(n_h) vs. the horizontal displacement at the surface (y_o) were obtained based on the performed load tests adopting a range of horizontal displacement values. [8] proposed the curves (n_h) vs. (y_o) using values in the 6.35 to 12.70 mm range. Based on these authors' studies but making a small modification in the range, the range between 6.0 mm and 12.0 mm was adopted, which was used by [10, 11, 12, 13]. The curves of the horizontal reaction coefficients vs. horizontal displacement were divided into two groups: load tests performed with natural soil moisture content and load tests performed with pre-flooding.

2. Materials and Methods

The aims of this study were achieved based on geotechnical analysis of the soil under investigation, execution of caissons, conduction of load tests in a caisson and analysis of the obtained data.

2.1. Geological and Geotechnical Characteristics

The study was conducted at the experimental site located at the School of Civil Engineering, Architecture and Urban Studies (Faculdade de Engenharia Civil, Arquitetura e Urbanismo) of the University of Campinas (Universidade Estadual de Campinas - Unicamp) in the municipality of Campinas, State of São Paulo, Brazil. The subsoil geotechnical properties were determined by the removal of undisturberd samples up to 8 m in depth and disturbed samples up to 9 m of depth (impenetrable). In-situ tests were performed: the standard penetration test (SPT) and the electric cone penetration test (CPT). The simple characterization tests are presented in Table 1, and the results for the triaxial tests, permeability tests, consolidation tests and the collapse index are presented in Table 2.

The edometric tests conducted on undisturbed samples with flooding at different stress levels indicate the collapsible characteristics of the soil. Tests performed with tensions of 100 kPa, 200 kPa and 400 kPa. According to [15], indicated that the representative sample at 1 m of depth was observed to be collapsible, with a very elevated index for all the flooding stresses. For the 2 m and 3 m deep layers, it was possible to observe the low collapsibility index for the 100 kPa and 200 kPa stress, which in turn had a high index for the 400 kPa stress. Based on the data obtained for the 100 kPa stress, only the 1 m, 4 m and 8 m depths were observed to be collapsible. However, for the 200 kPa stress, only the 2 m and 3 m depths were not collapsible. Finally, for the flooding stress of 400 kPa, all of the depths were observed to be collapsible, with the exception of the 8 m depth. With the conduction of triaxial tests (CU), numerical values were obtained for the angle of friction and the cohesion intercepts (Table 2) for each depth. [16] verified a significant reduction of these values with the soil saturation.

The results for the SPT and CPT tests performed at 0.50 m from the caisson axis are presented in Table 3.

The grain size analysis indicates that the soil is mostly composed of silt and clay. However, its permeability characteristics with coefficients of the order of 1.5 x 10^-3 cm/s and its characterization through results of the CPT test, when using classification graphs, indicate that its behavior is that of a sandy soil. Therefore, the analysis will be performed based on parameters proposed for the soil of granular behavior.

The information obtained based on the cone penetration test (CPT) and the standard penetration test (SPT) regarding the resistance parameters, soil type and unsaturated condition allow the prediction of a low resistance and most likely collapsible soil.

2.2. Preparation of Caissons

Three caissons were built to perform the load tests: one to undergo the transversal load and two to serve as a reaction to the applied load. All of the caissons had lengths of 9 m, shaft diameters of 0.8 m, base diameters of 1.6 m and base heights of 0.7 m. The caissons were set up along the depth with 15 bars with f=25.0 mm and stirrups with f=10.0 mm, spaced every 15 cm in steel with fyk=500 MPa. The concrete used exhibited fck=20 MPa, and the slump test was equal to 22 mm. The caissons were built using a manual excavation process. The caissons were constructed in a manner in which they would not undergo significant structural strains when subjected to loadings.

2.3. Load Tests

To determine the load applied at the top of the caisson, a 500 kN load cell was used, and three deflectometers, with 0.01 mm precision, fixed on a reference beam were used for the measurement of the displacements. In two of the deflectometers, the readings were taken at the same level of the load application (horizontal), and in the third, the reading was taken along the vertical axis of the load.

Two load tests were performed in the same caisson following the guidelines of the standard [18]. The first test was of the slow maintained load type, with the soil in its natural moisture content state. The second, of the quick maintained load type, was performed after the superficial soil flooding with a pit for 48 h. The slow test was conducted with loads applied in equally increasing values, with readings of the displacements taken at pre-determined time intervals until the stabilization of the displacements, when a new load was increased, never more than 20% of the predicted working load. The consolidation was considered stabilized when the value read between the two successive times did not surpass 5% of the total consolidation of the loading stage. To perform the quick test, the load was maintained for 10 min, reading the initial and final stage values. The total load with an increase not higher than 10% of the working load proceeded until reaching twice the predicted working load for the caisson. Unloading was performed soon after the last loading stage in four load levels.

3. Results and Analysis

Based on the performance of the load tests, the load vs. horizontal displacement curves were obtained. Plots of the horizontal reaction coefficient vs. horizontal displacement are presented in Figs. 3 and 4.

The analyzed soil was lateritic and highly porous and unsaturated; its sand particles were cemented by clay and silt lumps, where the main minerals present were kaolinite and gibbsite. Hence, this soil had matrix potential because of the suction, which was lost when the soil is flooded, leading to a reduction in the resistance. This phenomenon can be observed in the analysis of Fig. 3 and 4, where the loss of stability can be noted in the presence of water, thus reducing its resistance characteristics. Note that the n_h values in the pre-flooded situation practically do not vary, which demonstrates a resistance loss of the soil from the beginning of loading (Fig. 4). Table 4 presents the load and horizontal displacement values in addition to the results for the horizontal reaction coefficients as a function of the subsoil condition.

The pre-flooding of the soil caused a pronounced increase in the displacements for the same load applied in the tests with natural soil. The results proved that the pre-flooding caused a reduction of the applied load by 2.5 to 3 fold to reach displacements of 6 and 12 mm. These results demonstrate the intensity of the effect of the variation of the soil moisture on the horizontal displacement characteristics of the pile.

The values for the horizontal reaction coefficients obtained in the literature for different types of deep foundations built on soil with similar behavior to the one from the present study (under natural and pre-flooded conditions) are presented in Fig. 5.

By analyzing Fig. 5, it is verified that in piles built on similar soils (sandy behavior, unsaturated and lateritic), the values for the horizontal reaction coefficient vary between 2 and 12 MN/m³. With the exception of the precast concrete pile with 0.18 m and the foundation in this paper (belled caisson) with 0.8 m, the other piles exhibited diameters of 0.4 m. The observations indicate that the execution process of the pile did not affect the results. On the same graph (Fig. 3), the pre-flooding effect on the soil can be evaluated, with the reduction of n_h values, in some cases obtaining values lower than 1 MN/m³. Generally, the values were reduced by at least 50%.

4. Conclusions

Based on the results obtained and the analysis performed, the following can be concluded:

]]> Because the first few meters of the soil have a great effect on the horizontal loading, their proper geotechnical characterization in terms of porosity, resistance and collapsibility is important.
• The laboratory and field tests indicated a highly porous, collapsible and low resistance soil, allowing the prediction of low n_h values and the great effect of the variation in the soil's moisture content.
• The horizontal reaction coefficient obtained for the soil at its natural moisture content was 12 MN/m³, a higher value when compared with the literature for soils of high porosity, which suggests values of 2 MN/m³, thus indicating the importance of its determination for each type of soil and therefore avoiding generalizations.
• The loads for pre-flooded soil to reach certain displacement levels were observed to be on average three times smaller than the ones for the soil at its natural moisture content. For the horizontal reaction coefficient of the soil, the pre-flooded values were approximately 50% lower than the ones for the soil at its natural moisture content.
• For the collapsible soils, the analysis of the possibility of saturation or great humidity variation of the superficial soil during its construction service life was of fundamental importance in the determination of project parameters.

References

[1] Décourt, L., Load-deflection prediction for laterally loaded piles based on N-SPT, Proceedings of International Conference on Piling and Deep Foundations, 4(1), pp. 549-555, 1991.         [ Links ]

[2] Carvalho, D., Albuquerque, P.J.R., Claro, A.T. e Ferreira, C.V., Análise de estaca carregada transversalmente no topo, em solo residual de diabásio, Seminário de Engenharia de Fundações Especiais (SEFE 3), 1, pp. 145-154. 1996.         [ Links ]

[3] Velloso, D.A. e Lopes, F.R., Fundações - Volume 2. Rio de Janeiro: COPPE-UFRJ, 2002        [ Links ]

[4] Reese, L.C. and Van Impe, W.F., Single piles and piles groups under lateral loading. London: CRC Press - Taylor and Francis Group, 2011.         [ Links ]

[5] Valencia, Y, Camapum, J. y Lara, L., Aplicaciones adicionales de los resultados de pruebas de carga estáticas en el diseño geotécnico de cimentaciones. Revista DYNA [Online], 79(175), 2012. [date of reference August 5^th of 2016] Avaliable at: http://www.revistas.unal.edu.co/index.php/dyna/article/view/28167/43527        [ Links ]

[6] Terzaghi, K., Evaluation of coefficients of subgrade reaction. Géotechnique, 5(4), pp. 297-326. 1955.         [ Links ]

[7] Matlock, H. and Reese, L.C. Generalized solutions for laterally loaded piles. Journal of Soil Mechanics and Foundation Engineering Division - ASCE, 86 (SM5), pp. 63-91, 1960.         [ Links ]

[8] Alizadeh, M. and Davisson, M.T., Lateral Load Tests on Piles - Arkansas River Project. Journal of the Soil Mechanics and Foundation Division - ASCE, 96 (SM5), pp. 1583-1604, 1970.         [ Links ]

[9] Matlock, H. and Reese, L.C., Foundation analysis of offshore pile supported structures. Proceedings of International Conference on Soil Mechanics and Geotechnical Engineering, 5(2), pp. 91-97, 1961.         [ Links ]

[10] Miguel, M.G., Execução e análise de provas de carga horizontal em estacas em solo colapsível, MSc. Thesis, Departament of Geotechcnics, University of São Paulo, São Carlos, Brasil, 1996.         [ Links ]

[11] Menezes, S.M., Carvalho, D., Sampaio, F.M.T. e Ribeiro, K.D., Provas de carga horizontais em estacas pré-moldadas de concreto cravadas em solo de alta porosidade. Seminário de Engenharia de Fundações Especiais (SEFE 5), 2, pp. 500-506. 2004.         [ Links ]

[12] Miranda Júnior, G., Estacas submetidas a esforços horizontais em solos colapsíveis do interior de são paulo nas condições natural, melhorada e inundada. PhD. Dissertation, Faculty of Agricultural Engineering, University of Campinas, Campinas, Brasil, 2006.         [ Links ]

[13] Zammataro, B.B., Estudo de comportamento de estacas tipo escavadas e hélice contínua submetidas a esforços horizontais, MSc. Thesis, Departament of Geotechcnics and Transportation, University of Campinas, Campinas, Brasil, 2007.         [ Links ]

[14] Gon, F.S., Caracterização geotécnica através de ensaios de laboratório de um solo de diabásio da região de Campinas/ SP, M.S Thesis, Departament of Geotechcnics and Transportation, University of Campinas, Campinas, Brasil, 2011.         [ Links ]

[15] Vargas, M, Introdução à Mecânica dos Solos.São Paulo: McGraw -Hill do Brasil, 1978.         [ Links ]

[16] Paschoalin-Filho, J.A., Estudo do comportamento à tração axial de diferentes tipos de estacas executadas em solo de diabásio da região de Campinas-SP. PhD. Dissertation, Faculty of Agricultural Engineering, University of Campinas, Campinas, Brasil, 2008.         [ Links ]

[17] Rodriguez, T.G., Caracterização geotécnica de um solo de diabásio por meio de ensaios SPT e CPT M.S. Thesis, Departament of Geotechcnics and Transportation, University of Campinas, Campinas, Brasil, 2013.         [ Links ]

[18] Associação Brasileira de Normas Técnicas. NBR 12131: Estacas - Prova de Carga Estática. Rio de Janeiro: ABNT, 2006.         [ Links ]

R. Kassouf, received the BSc. In Civil Engineer in 1986, MSc. in Geothecnical Engineering in 2012, He is currently director - Roberto Kassouf Engenharia Ltd, since 1986. Professor of the Faculty of Integrated Metropolitan Campinas (Metrocamp) course in Foundations of Civil Engineering disciplines I and II and Soil Mechanics. Master in Civil Engineering from Unicamp - University of Campinas. Works in the area of Civil Engineering, with emphasis on Geotechnical, acting on the following topics: geotechnical investigation, deep foundations and foundation pathology. ORCID: 0000-0002-2963-2883

D. Carvalho, received the BSc. In Civil Engineer in 1978, MSc. in Engineering in 1985, PhD. in Geothecnical Engineering in 1991. He is Associate Professor at University of Campinas since 1998, developing researches in Foundations, Soil Mechanics and In-Situ tests. He guided several students in geotechnical researches and has published papers in conferences and technical journals. Currently teaches classes for Civil Engineering, Agricultural Engineering and Geology. ORCID: 0000-0002-5270-3627

P.J.R. Albuquerque, received the BSc. In Civil Engineer in 1993, MSc. in Engineering in 1996, PhD. in Soil and Foundation Engineering in 2001 and postdoctoral in 2010 at the Polytechnic University of Catalunya in Barcelona/Spain. He is Associate Professor at University of Campinas since 2001, teaching and researching, with emphasis in in-Situ tests, laboratory tests and foundations. He was the secretary of the Technical Commission of In-Situ Tests of the Brazilian Association of Soil Mechanics (ABMS) between 2007-2010 and the president for the Biennium 2011/2012. He is secretary of the Technical Commission of Foundation of the Brazilian Association of Soil Mechanics for the Biennium 2013-2016. In biennia 2013-2016 will act as Vice-President of the Regional Center of São Paulo (ABMS). Since 2015 is Brazilian representant of ABMS on TC102 Technical Commission of Foundations of ISSMGE. ORCID: 0000-0003-0726-7165