Abstract

Sanding onset prediction models are used to forecast the production condition, e.g. pressure drawdown or flow rate, at which sand production occurs. A sanding onset prediction model is critical for sand control decision-making in answering the questions whether or not sand control should be used, or when to use sand control. There are a variety of numerical and analytical sanding onset prediction models available in the literature. However most of those models either need one or another rock mechanics input parameter which is rarely available in field practice or demand extensive computations (e.g. Finite Element models) that is not practical in cases when quick sand control decision is needed. As a result, the usage of those models in field practice is limited.

In this paper, simple and easy-to-use analytical sanding onset prediction models are derived based on theory of poroelastoplasticity assuming shear failure or tensile stress induced sanding from open hole well or from the perforations of a cased well. To use those models, the only rock mechanics parameters needed are Biot's constant, Poisson's ratio, Uniaxial Compressive Strength (UCS), and in-situ stresses. With those data, a critical drawdown pressure for a well can be obtained for any given average reservoir pressure.

Using those models, published and field reported sanding problems are studied. Through this study, sanding mechanisms for those cases are explained and good agreement is reached between predicted and field measured critical drawdown pressure.

Introduction

Sanding onset prediction model predicts the production condition, e.g. pressure drawdown or flow rate, at which sand production occurs. Sanding onset prediction model is critical for sand control decision-making on whether or not sand control should be used, or when to use sand control. There are a variety of numerical and analytical sanding onset prediction models in the literature. Among those models, Morita et al. [1] proposed a very detailed and thorough mathematical model for sanding onset prediction. In their model, sanding is assumed to occur when the plastic strain at the perforation surface reaches a critical value. This value is presumabaly different from rock to rock and measurement should be made to determine its magnitude. Unfortunately, there are neither many published values for specific rocks nor published methodologies on how to measure this quantity in the field. As a result, sanding onset prediction models using critical plastic strain as a criterion are of limited practical usage. Besides, even though numerical sanding onset prtediction models are more general and flexible, they are computationally demanding that is not practical in cases when quick sand control decision is needed. As a result, analytical or semi-analytical models are more commonly used in engineering practice.

In petroleum engineering, a wellbore/perforation tunnel and their adjacent formation are often approximated as a thick-walled hollow cylinder. Using this approximation, analytical or semi-analytical solutions for stress state in the near wellbore/perforation tunnel can be obtained. Risnes et al.[2] studied the near wellbore stress state considering incompressible, steady state fluid flow into the wellbore in a bounded elastoplastic reservoir. Initial vertical strain is taken into account in their model. Bradford and Cook [3] studied non-steady state oil flow into a wellbore in an infinite boundary elastoplastic reservoir. Sanfilippo et al.[4] proposed a more general poroelastoplastic model for an infinite boundary reservoir by taking into account initial stresses before a well is drilled and produced. However, for simplicity, pressure in well drainage area is assumed to be uniform.

This content is only available via PDF.
You can access this article if you purchase or spend a download.