Databases for Shallow Foundations

Overview

To assemble a large database for model uncertainty assessment, available field observations (e.g. settlement) or load tests (e.g. complete load-movement curve) on shallow foundations are reviewed next:

1. EPRl/Found/804: Since the late 1970s, research project RP1493 was conceived within the Electrical Power Research Institute (EPRI) in response to an industry request for improved analysis and calculation methods for transmission line structure foundations (Kulhawy and Trautmann 1995). This led to the compilation of a large database of 804 field load tests on shallow and deep foundations (Kulhawy et al. 1983b). This database is labelled EPRI/Found/804 in this book. It contains detailed information on site conditions (e.g. in situ and laboratory measurements of soil parameters), construction procedure, foundation geometry, test procedure (e.g. method of loading, load sequence and time intervals for the application of various levels of load) and load test data (load-movement plot for most cases and axial load distribution and load versus time plot for some cases). The load test records were supplied by 25 different electric utility companies and were obtained from technical literature. Table 4.7 lists the number of uplift and compression tests associated with the various foundation types (e.g. drilled shafts, driven and pre-augered piles, shallow foundations and anchors). As seen, uplift load test is dominant.

Table 4.7 Summary of the EPRI/Found/804 database in Kulhawy et al. (1983b)

• 2. Akbas/ShalFound/426: A large database of 426 case histories for a shallow foundation in cohesionless soil was compiled by Akbas
• (2007). This database is labelled Akbas/ShalFound/426 in this book. In the database, there are 167 case histories with complete load-movement data, as well as the location of test sites, general descriptions of prevailing soil type and friction angle either directly from a laboratory test or indirectly from in situ test results (i.e. SPT and/or CPT). The data was used to investigate the axial behaviour and bearing capacity of footings in cohesionless soil (Akbas and Kulhawy 2009a, b). The geometric and geotechnical properties are summarized in Table 4.8. The load tests were categorized into three subgroups based on the data quality: (1) ninety-seven cases with a load level high enough to interpret capacity, representing the best quality data; (2) twenty-eight cases where the load test was terminated in a non-linear transition region and it is possible to determine capacity with a small amount of curve extrapolation (e.g. Chin’s hyperbolic model); and (3) forty-two cases with small settlement values that were insufficient to determine capacity, even using curve extrapolation. This data group will be used to (1) evaluate the calculation methods for bearing capacity and settlement and (2) develop a probabilistic bivariate model of the load-movement response. For the other 259 cases, a single data point for a measured settlement is available for each case. The Akbas/

Table 4.8 Summary of geometric and geotechnical parameters for 167 field load tests with complete load-movement data in Akbas (2007)

ShalFound/426 database covers a wide range of foundation types from small plate 0.25 m wide to mat foundation up to 135 m wide (Table 4.9) and soil type from silt to gravel. The structures corresponding to these foundations include bridge, test footing, building, rank, embankment, chimney, nuclear reactor and silo. Figure 4.10 summarizes the soil exploration method data. The case histories were categorized into three subgroups based on the quality of data: (1) 253 cases for which N_SPT values and all the other geotechnical and geometric parameters are available for settlement calculation using the selected methods, (2) 66 cases in which N_SPT values are not available but could be estimated using other in situ test results (e.g. cone penetration, dila- tometer or PMTs) and (3) 107 cases where some necessary geotechnical or geometric information is missing or the only available test results are plate load or oedometer tests.

3. UML-GTR ShalFound07 and UML-GTR RockFound07: Two databases, UML-GTR ShalFound07 and UML-GTR RockFound07, were

Table 4.9 Summary of geometric parameters for the Akbas/ShallowFound/426 database in Akbas (2007)

Figure 4.10 Summary of soil exploration methods in the Akbas/ShalFound/426 database

compiled by Paikowsky er al. (2010) to develop the LRFD specifications for shallow foundations of highway structures used in the United States. The UML-GTR ShalFound07 database contains 549 load tests for shallow foundations in soil that were assembled from four sources. Table 4.10 shows that the predominant soil type is cohesionless (N = 463 for sand gravel), and the test type is PLT (N = 466 for В < 1 m). The majority of load tests were carried out in Germany (N = 254), the United States (N = 84), France (N = 60) and Italy (N = 56). The UML-GTR RockFound07 database in Table 4.11 is composed of

Table 4.10 Summary of the UML-GTR ShalFound07 database

Source: data taken from Paikowsky et al. 2010

Source: data taken from Paikowsky et al. 2010

• 122 tests for shallow foundation (N = 61) and drilled shafts (N = 61) (for which the tip load-movement curves were measured) in soft to hard rocks from ten different countries. It covers various rock types, such as hardpan, fine-grained sedimentary and igneous volcanic rocks.
• 4. SHRP2/ShalFound80: As noted in Samtani et al. (2010), much of the data for spread footings are related to smaller-size footings that are typical for buildings. However, the footings for bridges are large compared to those for buildings where the design is generally controlled by tolerable settlement (i.e. SLS) rather than bearing failure (i.e. ULS). To advance the second strategic highway research programme into practice (SHRP2), Samtani and Allen (2018) developed a large database from the United States, as well as European sources. Serviceability limit state design for bridges is a SHRP2 solution, including the development of design and detailing guidance and a framework for SLS calibration to provide one-hundred-year bridge life (e.g. Samtani and Kulicki 2019,2020). The database in Table 4.12 is composed of eighty data points for measured settlement of spread footings used in highway bridges.

NUS/ShalFound/919

Among the databases reviewed in Section 4.5.1, two in Paikowsky et al. (2010) were not published, although they have a large number of load test data. To support the development of deterministic and probabilistic analyses and to hasten digitalization, it would be worthwhile to merge different databases (e.g. Samtani and Allen 2018; Lesny 2019). A large database containing 919 prototype model tests on shallow foundations was compiled by Tang et al. (2020) from the literature. This database is labelled NUS/ ShalFound/919 in this book. This database provides the test site location,

Table 4.12 Summary of the SHRP2/SpreadFound/80 database in Samtani and Allen (2018)

soil conditions, foundation dimensions and load test data. It is briefly introduced next.

The NUS/ShalFound/919 database contains 483 compression and 436 uplift load tests. For axial compression loading, it consists of 56 field load tests in cohesive soil (natural soil condition), 141 model tests in centrifuge (controlled soil condition), 101 full-scale model tests in a calibration chamber and 185 field load tests in cohesionless soil. For axial uplift loading, it consists of 123 field load tests in cohesive soil, 19 model tests in centrifuge, 143 full-scale model tests in a calibration chamber and 151 field load tests in cohesionless soil. The load type, test type, soil type, number of load tests, ranges of foundation width and embedment depth and geotechnical parameters in the NUS/ShalFound/919 database are summarized in Table 4.13. The geographical regions cover Australia, Brazil, Canada, Chile, China, France, India, Ireland, Israel, Italy, Japan, Kuwait, Norway, Portugal, Singapore, Saudi Arabia, South Africa, Sweden, Taiwan, Thailand, Turkey, United Kingdom and United States. In situ soil property measurement includes SPT, CPT, PMT and dilatometer test. Strength test type includes unconsolidated-undrained triaxial (UU), consolidated-und- rained triaxial (CU), unconfined compression (UC), direct shear (DS), and field vane (FV). The cohesive soil types are broad (e.g. marine clay, sandy/ silty clay, lean clay), covering a wide range of undrained shear strength

Source: data taken from in Tang et al. 2020

s_u = 10-300 kPa. The cohesionless soil types are also broad (e.g. well/ poorly graded sand/gravel, silty/clayey sand/gravel) covering a wide range of relative density D_r = 13%-100% and effective stress friction angle ф = 26°-53°. Load test data of foundations in rock will be presented and discussed in Chapter 6, as the majority of test data were performed on drilled shafts in rock. They are used to evaluate the calculation methods for tip resistance of rock sockets.

As mentioned in Section 3.6, it is reasonable to take measured capacity as a peak or asymptote of the load-movement curve. Except for scaled-model tests in dense soils in centrifuge (e.g. Gemperline 1984; Zhu et al. 2001), however, most measured load-movement curves in full-scale field load tests do not show clear indications of bearing capacity failure (Akbas 2007) because large foundation movements are required. The foundation capacity has to be interpreted using a failure criterion (Paikowsky et al. 2010). In the current work, four failure criteria are considered: (1) peak load (q_p) (e.g. Gemperline 1984; Zhu et al. 2001), (2) the tangent intersection method that determines the measured capacity as the load (q_TI) at the intersection of two tangents to the initial and final linear portions of load-movement curve (e.g. Stas and Kulhawy 1984; Consoli et al. 1998; Akbas and Kulhawy 2009a),

(3) the L|-L, method that defines the measured capacity as the load (q_L2) at the beginning of the final linear part of load-movement curve (e.g. Hirany and Kulhawy 1988; Akbas and Kulhawy 2009a) and (4) the load (q₁₀o_/oB) at the foundation movement of 10% of the foundation width (e.g. Lutenegger and Adams 2006; Cerato and Lutenegger 2007; Akbas and Kulhawy 2009a).

Paikowsky et al. (2010) stated that the L,-L, method is similar to the minimum slope failure criterion proposed by Vesic (1963).