Rock mechanical properties may be derived from formation evaluation logs using empirical relations or using algorithms based on microscopic rock models, which relate the dynamic deformation behavior of the rock to the static deformation behavior. In this study, we compare both approaches. In a first step, the algorithm FORmation MEchanics Logging (FORMEL) is fed with synthetic data sets for sandstones and shales to determine the sensitivity of the algorithm for variations in the input data (porosity, compressional slowness, density). The UCS, Young's modulus and bulk modulus calculated from FORMEL are compared with results from empirical relations. In a second step, the FORMEL calculated rock mechanical parameters are compared with data from triaxial tests on different sandstones and the Woodford shale. This study shows that variations in petrophysical input parameters have a large influence on the FORMEL results and conclude that models relating on more than one input parameter are preferred over empirical relations which relate rock mechanical parameters to only one value. The differences between laboratory analyses and calculations using FORMEL indicate that also microscopic models are not universally applicable but need to be calibrated against laboratory test data for different formations.
Accurate knowledge of rock mechanical properties is essential for many geomechanical applications. Examples include the prediction of wellbore stability or the investigation of the reservoir and overburden compaction behavior during reservoir depletion. One precise way to obtain rock mechanical properties is to perform triaxial laboratory tests on specimens obtained from core samples. However, extensive coring is expensive and time-consuming and is therefore not carried out routinely. In addition, measurements made on core material can be biased due to relaxation and alteration of cores after recovery. As an alternative to core testing, rock mechanical properties may also be obtained from formation evaluation (FE) logs. Continuous information of mechanical properties is frequently provided by correlation-based methods. For a given well, log-derived petrophysical data such as: acoustic slowness, porosity, and bulk density are correlated with laboratory-derived rock mechanical properties. Examples are given by [1] and [2] for common oilfield rock types (sandstones, shales, carbonates). One drawback associated with empirical relations is the fact that they may be specific to the geological setting of the region from where they were developed. An alternative log-based approach is the use of microscopic rock models that calculate deformation behavior. While some empirical correlations between microscopic attributes (such as crack density) and mechanical parameters (such as stresses) are involved, the employment of micromechanical models has the potential of correctly reflecting the sensitivity of rock mechanical parameters to changes in measured petrophysical properties. Reference [3] proposed FORMEL (FORmation MEchanics Logging), a constitutive model describing the microscopic processes occurring in a rock sample during mechanical loading. So far, FORMEL is applicable to sandstones, shales and chalk. The model is calibrated against results from triaxial laboratory tests and considers the effect of the deformation behavior with respect to processes such as crushing at grain contacts, and shear sliding along cracks.
