EMPIRICAL COMPRESSION INDEX EQUATIONS
The majority of available empirical equations for estimating the compression index of clay in terms of its index properties are linear in form. Over the past few decades, a significant number of these formulas have been published based on high correlation coefficients and limited data. Significant measures such as standard errors and number of data points are often not reported. Several empirical formulas are published indicating that such empirical equations are applicable to all clay soils. Some published expressions are limited to few types of fine-grained and organic soils. Most available data exhibit a behavior in which significant scatter is observed when high values of liquid limit, water content, and void ratios are used. These are known to be associated with organic soils. A new method is being proposed that permits the assessment of available empirical formulas in terms of accuracy and applicability to different soil types. The new model relates regression coefficients expected between compression index and Liquid limit. Examination of some well known empirical formulas indicate that such equations may not be suitable for compression index prediction. Furthermore, several obscure empirical equations are shown to be more reliable than previously believed. A comparison was made of available and newly derived expressions based on combined data published independently by several authors. The proposed model provides useful insights into empirical formula development. Many questions can be raised regarding the validity of available regression equations when one data point could significantly alter the coefficient of correlation. The author utilized a systematic approach in developing expressions for different ranges of compression indices. This procedure eliminates any author biases on whether to include or to exclude one or more data points from a given data set. The author concluded that linear models relating the compression index of clay to its liquid limit seem to reveal that the slopes and intercepts of such models are interdependent.