Selection of Best Fit Correlation for Compression Index of Soil from District Bannu

Geotechnical investigation is one of the major steps in the safe and economic design of foundation for any structure. This requires some soil parameters to be established. CI (Compression Index) is one of these parameters. The stability of a foundation is not fully justified as long as the magnitude of settlement is unknown. The problem arises when the value of CI is needed badly and a consolidation test is set up for this purpose which consumes almost 3 days for its completion provided that the assembly would not be disturbed. Little amount of disturbance can cause serious error in further calculations and plotting of data. Undisturbed soil sample is the primary demand of consolidation test which is a tough job for a person having not been experienced very much in the fieldwork. In order to overcome these complexities, the researchers have established various correlations between compression index and some basic soil properties like Atterberg limits, void ratio, natural moisture content and OMC (Open Moisture Content), as they are simple to be determined and takes small amount of time. This research has been conducted on soil samples collected from a single spot at an interval of 5ft depth for the sake of validation of a suitable correlation for Bannu District, KPK (Khyber Pakhtunkhwa), Pakistan so that the determination of compression index becomes simple and convenient.


INTRODUCTION
which are easy to determine. Different researchers targeted different types of soil such as remolded, undisturbed, high plasticity and low plasticity clays tabulated as shown in Table 1 [2].
Terzaghi and Pecks [15] correlation is the most popular in this regard and used extensively in geotechnical problems but there are certain facts that must be considered while doing so. (1) Weather the type of soil is the same as was experimented upon by Terzaghi and Peck [15]. (2) The conditions of the sample recovery are consistent or not.
If the answer is yes then this correlation is OK but if the answer is no, then it might not yield appropriate results.
This may be witnessed from the work of other researchers such as Skempton [3] who worked on disturbed and remolded clay samples and arrived at the following correlation. This study was carried out by many other researchers with the difference of region and soil type [8][9][10][11][12].
The questions that arise are that how much reliable these results are? And why a particular model has been selected for this purpose? This is based upon subjectivity most of the times. Bayesian probabilistic approach was introduced for this purpose to carry out multivariate regression for CI. The initial void ratio and liquid limit possess highest degree of plausibility among various classes of formulas [13]. A more sophisticated approach is utilized in [14] in order to improve the accuracy, reliability and predicting power of the correlations. Which makes use of the AI (Artificial Intelligence) for resolving a particular problem or issue.

MATERIALS AND METHOD
The CI of undisturbed soil sample was determined using the conventional 1D-consolidation test carried out according to ASTM, D-2435. The sample was collected with the help of Shelby tube also known as thin walled sampler shown in (Fig. 1).
The technique of auger boring was utilized to probe into the ground up to a depth of 25ft with a diameter of 5 inches. The sample was recovered at each 5ft interval so that the variation of CI with the depth could be investigated. This is depicted in (Fig. 2). As the sample at one depth was obtained, the borehole was advanced further. In the lab the sample was tested in two ways. The consolidation test was conducted in order to get the realtime value of C c . While the index properties of soil like liquid limit, plastic limit, water content, specific gravity and OMC were determined and incorporated into the correlations to find the parallel set of values of C c . Three most suitable correlations were selected based upon the conditions of Bannu soil for this purpose that of Terzaghi and Peck [15], Nishida [16], Singh and Noor [6]. Both the calculated and actual values were compared using Microsoft excel application for all the three mentioned correlations.

Selection of Best Fit Correlation for Compression Index of Soil from District Bannu
The bar charts shown in Figs If carefully analyzed the value of SD, Singh and Noor [6] is the one correlation which bears minimum value.

CONCLUSION
The conclusion drawn after going through the entire procedure can be described as, "The results yielded by a correlation which was developed in 2012 are in very close resemblance with those came out from consolidation test.
Hence the suitability of this correlation for finding CI is hereby confirmed and recommended for all geotechnical engineers and consultants working out on projects in District Bannu, KPK. This study was carried out to a depth of 30 ft and with no water table.

RECOMMENDATIONS
It is strongly recommended that this study should be extended to a big area and more number of samples should be collected and tested in order to fulfill the demands of regression analysis. This research should be exercised in those regions where water table is high as this study does not cover this aspect.

ACKNOWLEDGEMENT
The cooperation and assistance offered by the Chairman, Department of Civil Engineering, and Technical staff of Soil Mechanics Lab., University of Engineering & Technology, Peshawar, Campus-III, should be highly acknowledged. It was their support and favor, which helped to complete this research study.