Correlation of SPT–CPT data from the United Arab Emirates

Abstract

The standard penetration test (SPT) is the most commonly used in situ test. However, cone penetration test (CPT) is becoming increasingly popular as an in situ test for site investigation and geotechnical design. Geotechnical engineers have gained considerable experience in design based on local SPT correlations. In the near future, the CPT design correlations will also be developed based on local experience and field observation. However, with the initial introduction of CPT data, there is a need for reliable SPT–CPT correlation so that CPT data can be used in the existence of SPT design correlations. Moreover, in those cases where only SPT results are available, engineers, who are more familiar with CPT interpretations, will translate the SPT blow counts (N-values) into CPT cone resistances (q_c-values).

Published SPT–CPT correlations have been reviewed and concluded that many of the correlations do not provide information on the statistical procedures used. The information on the geological variability of the test sites is also missing. It may be important if these parameters have an effect on these tests and correlation.

This study proposes a method to select and proceed the data for SPT–CPT comparison. The traditional and statistical methods were used for the correlation of SPT–CPT test results from the United Arab Emirates (UAE). Traditional (arithmetic average) method gave a higher ratio than literature values. Statistical approaches gave almost similar results and less high results than arithmetic average method.

The soils in this area are recent deposits, consisting mainly of sand with relatively high content of calcium carbonate. Although carbonitic soils are weaker than silicate soils, a high ratio may be explained by cementation, densification and shelly structure or gravel layers in the United Arab Emirates soils.

Introduction

Many geotechnical design parameters of the soil are associated with the standard penetration test (SPT) and the SPT is widely used around the world. On the other hand, static cone penetration test (commonly referred to as cone penetration test, CPT) is becoming increasingly more popular for site investigation and geotechnical design, especially in deltaic areas, based on the soil type and testing method. For many construction projects, it is common to find that the preliminary design is based on soil parameters obtained from standard penetration tests, whereas the construction quality control or the final design is based on static cone penetration test results, or vice versa.

It is very valuable to correlate the static cone tip resistance, q_c, to SPT N-value so that the available database of the field performances and property correlations with N-value could be effectively utilized. Hence, many empirical relations have been established between the SPT N-values and CPT cone bearing resistances, q_c (Robertson et al., 1983).

Sanglerat (1972) cites Meyerhof (1965) who suggested a relationship n=(q_c/N)=0.4 (q_c in MPa); but further, Meigh and Nixon (1961) showed that this simple relationship did not take into account the effect of grain size and made comparative tests in sand and gravel. They have recommended n-values as 0.2 MPa for coarse sand and 0.3–0.4 MPa for gravelly sand. Engineers of Franki Piles have found the n-values as 1.0 MPa for sand, 0.6 MPa for clayey sand, 0.5 MPa for silty sand, 0.4 MPa for sandy clay, 0.3 MPa for silty clay and 0.2 MPa for clays. De Alencar Velloso (1959) gave n ratios for different types of soils and found as 0.35 MPa for clay and silty clay, 0.2 MPa for sandy clay and silty sand, 0.35 MPa for sandy silt, 0.6 MPa for fine sand and 1.0 MPa for sand (De Alencar Velloso, 1959). Schmertmann (1970) pointed out that if only SPT data are available, these can be converted to cone penetrometer values. He suggested 0.2 MPa for silt, sandy silt and slightly cohesive silt–sand mixture, 0.3–0.4 MPa for clean, fine to medium sand and slightly silty sand, 0.5–0.6 MPa for coarse sand and sand with little gravel and 0.8–1.0 MPa for sandy gravel and gravel. He also suggested that instead of correlation of q_c with N-value, better correlation with f_s (sleeve friction) and N-value can be established, especially for cohesive soils. Robertson et al. (1983) presented the q_c/N ratio as a function of mean grain size, ‘D₅₀’. They proposed a soil behaviour-type classification, giving q_c/N ratio for each soil classification zone based on cone penetration test with pore pressure measurement tests (CPTU, piezocone). Lunne et al. (1997) cite Jefferies and Davies (1993) who presented a soil classification chart estimating N-values. This new development considers q_c by taking into account pore water pressure (u) and overburden stress (σ_vo′), using piezocone.

Although many authors proposed different correlations, it is quite recognisable that authors did not indicate the geology and geomorphology in their correlative works. The only indication of geology was given by Robertson et al. (1983), where they mentioned overconsolidation.

Ismael and Jeragh (1986) made a correlation on calcareous desert sands in Kuwait and compared it with the value of Schmertmann (1970) for clean, fine to medium sands and slightly silty sands. Their test value was higher than what Schmertmann suggested for clean, fine to medium sands and slightly silty sands. By comparing test results with the historical data of Robertson et al. (1983), they also found close agreement in the form of q_c/N versus mean grain size ‘D₅₀’.

Danziger and de Velloso (1995) made a correlation between CPT and SPT for some Brazilian soils. Values found were in the same range obtained by Schmertmann (1970). Different types of correlation were tested, and a linear correlation was found better suited for practical applications. A general trend was obtained in a similar pattern of Robertson's curve (increasing n-values with increasing grain size).

Most authors were apparently satisfied with an approximate relationship, indicating more or less an average. None of them mentioned a standard deviation or correlation coefficient. Much recent work has demonstrated that use of statistical methods would greatly assist in not only improving correlations but also indicate the dependability of the correlations derived. It is significant that such techniques have not been tried in improving correlation between SPT and CPT. A possible reason may be that along a vertical line, each 0.05 m, a reading of an electrical CPT is available, whereas for only each 0.5 m, an SPT result is determined. So over a depth of 1 m, there are 20 readings of q_c and at maximum two N-values.

This research study was carried out using an existing data set of the United Arab Emirates (UAE). The data set used in this study consists of cone penetration test (CPT) and standard penetration test (SPT) logs. Some of these logs were made available by the Technical University of Delft (TUD), the Netherlands and some of them were sent by Fugro Middle East (FME), the United Arab Emirates.