ABSTRACT: Sand production has alluded production engineers over a century since oil production began. Starting from the '80s, rock mechanics specialists have attempted to explain, and predict the onset of sanding in a hydrocarbon well. Two common shortcomings in the traditional approaches are: (1) use of MohrCoulomb and similar failure criteria, which do not seem to explain the large, episodic manner of sand production from the reservoir rock; and (2) limited scope of laboratory rock mechanics tests e.g., triaxial tests at low confining pressures. Both aspects address the brittle failure of rock, which does not seem to account for the 'tons' of sand commonly produced fi-om a deep well. We have studied the development of a ductile failure zone around a well under high effective mean stress that leads to pore collapse and growth of a sand bank. This mechanism can explain the production of large amount of sand produced fi'om deep wells. Using the classical Kirsch equations for stresses around a circular hole, and a cap model to calibrate the constitutive behavior of rock, we have demonstrated that pore collapse indeed takes place around a borehole or perforation tunnel, leading to the growth of a sand bank. We have developed a graphical calibration procedure to validate the sanding prediction model. A case history of sand production prediction from a deep gas-condensate sandstone reservoir (Lower Devonian) is presented.
INTRODUCTION The industry considersand production a problem when the rate of sand production is high, for two reasons: (i) abrasion of pipes and control fittings in the flow system, which may lead to catastrophic situation, e.g., explosion in gas flow line; and (ii) closure of the well due to 'sand-up', i.e., high accumulation of sand particles in the wellbore bottom, causing severe decline in oil/gas production rate.
Certain amount of coarse sand particles can be tolerated, especially if the produced liquid is oil. Higher viscosity oil further lessens the blasting effect of sand particles. However, even a small amount of sand particles in a high flow rate gas well may not be tolerable because of the danger of intensive sand blasting effect leading to pipe explosion.
In an excellent review of the analytical models for predicting the onset of sanding, Sanfilippo et al. (1995) categorized them into two groups, based on the failure mode of rock: tensile or shear failure. The models belonging to the first group aim to evaluate a critical production rate, or more precisely, a critical pore pressure gradient around perforations, while the other ones deal with a critical effective stress state. Among the first group are the criteria proposed by Weingarten and Perkins (1992), applied also by Ramos et al. (1994). The second group includes the models proposed by Risnes et al. (1981), by Santarelli et al. (1991), and the extension of the latter one by Kessler et al. (1992). The model proposed by Risnes et al. (1981) was later modified by Sanfilippo et al. (1995).
Abass et al. (1993) have described the physical processes involved in four differentypes of mechanisms that cause large (grains) and/or small (fines) size solids generation (Fig. 1). These mechanisms are: (i) tensile failure, (ii) cohesional failure, (iii) shear failure, and (iv) pore collapse failure. As shown in Fig. 1, the tensile and/or shear failures create a locali?.ed shear fault/plane or shear bands. Around a wellbore or a perforation, these shear plane/bands are of small extent, implying that the amount of solid particles released from the rock matrix is also small. On the other hand, erosional sand production due to cohesive failure, and pore collapse failure can give rise to high rate of sand producti