Managing sand production is becoming an increasingly important issue for petroleum production wells, particularly in challenging environments such as deep water, or gas fields with high flow rate production where sand tolerance is very low and changing the completion strategy could be costly. Sand production prediction is an integral part of overall field development planning. Consequently, it is important that the risk of sanding is accurately evaluated and appropriate measures are taken for well completion. The critical part in a sanding study is to define a proper rock deformation threshold based on laboratory tests or field observations. Due to the lack of sufficient field sanding data or for the sake of modelling simplicity, the rock deformation threshold is frequently defined either inappropriately or too simplistically because of the failure to recognize the intrinsic heterogeneities of reservoir rocks.
In this paper, two cases are presented with the deformation threshold "critical plastic strain limit" defined by a numerical simulation approach. One case represents a weak to medium-strength sandstone reservoir and the other a stronger, clean sandstone reservoir. In both cases, a comprehensive core testing dataset and a field-validated geomechanical model are available. In the weak-medium strength case, there were also several years of production history with sanding observations available from a few wells.
The numerical method uses an elasto-plastic material model where the rock behaviour is numerically defined from systematic triaxial compressive core tests and a rock failure criterion based on the plastic strain limits modelled from Advanced Thick-Walled Cylinder (ATWC) core tests with calibration against field sanding observations. From the modelling of a range of high-quality ATWC tests and field sand production experiences, strong correlations are found between the critical strain limit and the rock compressive strength for both cases indicating different critical strain limit thresholds for different rock strengths and rock types.
Although high-porosity and weaker rocks may still have a higher risk of sand production than low-porosity and stronger rocks, the risk of sanding may not be as high as predicted if a single threshold is used for the entire rock strength and porosity range. The use of a single threshold for the whole field could over- or under- estimate the risk of sand production, especially for reservoirs with a moderate risk of sand production.
Sanding evaluations of the two cases revealed that the predicted sanding risk is potentially misleading if a simple sanding threshold is defined without adequately capturing the heterogeneity in reservoir rocks. This approach may prohibit the selection of an appropriate completion strategy for the entire field. This paper is concluded by presenting a rock testing program procedure for sanding evaluation studies to better capture the heterogeneity in reservoir rocks and, hence, to predict the sanding risk more reliably.