Flow Assurance - Sand Production Basics

Introduction

Sand production is a critical objective in hydrocarbon systems as it was reported that more than 70% of wells produce from weakly consolidated reservoirs (Nouri et al. 2003). From the point of view of flow assurance, sand production could lead to severe issues with separation efficiency, material loss, and flow-path blockage due to sand fill. Erosion modeling should account for kinetic energy imparted by sand particles velocity, size, and the angle of impact they have with downhole tubular material and surface equipment such as chokes.

Reservoir depletion and changes in rock stresses could result in undesired sand production. Water breakthrough could also aggravate sand production as a consequence of increased drag forces, which make wells that exhibit water injection as EOR technique candidate for sand production.

.

Sand Production Control

Sand production could be stated as a systematic problem that consists of particle migration through different three systems (formation, sand-control media, and particle-carrying pipe flow within the borehole). Mechanical sand control methods besides chemical techniques (chemical sand consolidation (SCON) – sand agglomeration) are usually implemented to mitigate sand production downhole. ?Mechanical methods are usually primarily used in suspected wells and they include (sand screens, gravel pack, frac-and-pack, and expandable screens) while chemical treatments involve injection in the near wellbore area to improve strength of incompetent formation where SCON relies on the concept of fluid adhesion to sand grains while agglomerates enhance the attraction between sand particles through polymer bridging. ??

Selection of sand control methods depend on particle size distribution, sorting coefficient, uniformity coefficient, and reservoir fluid types. Inappropriate selection would lead to failure of the primary sand control method and would require a remedial action such as thru-tubing screens, gravel packs, or chemical injection. Key criteria for chemical treatments are: completion type, perforation interval length, placement method, and formation permeability and temperature.

Machine Learning Approaches for Sand Prediction

1. Ketmalee and Bandyopadhyay (2018) criticized the existing models for sand production as they require sonic and density logs, which are not usually available with most wells due to cost reduction. Hence, they introduced a model based on artificial neural networks. In this method, synthetic logs are generated to obtain values of the missing sonic and density logs from relevant logs such as (gamma rays, resistivity, and neutron-porosity logs). Those data are then used to predict potential sand production through the existing sand models. This technique was validated using three field cases. For each well, ANN model was trained using data points from nearby wells.
2. Alakbari et al. (2022) implemented machine learning approaches in order to predict the critical total drawdown (CTD) of sand production in gas wells to avoid limitations of existing correlations. They used response surface methodology (RSM) and support vector machine (SVM) which was trained and tested on 23 datasets collected from open literature. They used different statistical tests such as ANOVA and F-statistics test to prove robustness of RSM correlation which that CTD is a strong function of the total vertical depth, cohesive strength, effective overburden vertical stress, and transit time.
3. Abdelghany et al. (2022) conducted a comprehensive geomechanical modeling study on a full set of logs from two wells to infer the geomechanical elements and predict sand production. They demonstrated the challenge of key logs absence and that most engineers use empirical equations to predict the missing log intervals. Thereafter, they compared their model to the model driven by the Gardner equation and showed the robustness of their model in matching real field data.

References

Nouri, A., Vaziri, H., Belhaj, H. and Islam, R., 2003, September. Effect of volumetric failure on sand production in Oil-wellbores. In?SPE Asia Pacific Oil and Gas Conference and Exhibition. OnePetro.

Hassan, N.A., Yeap, W.J., Singh, R. and Nik Khansani, N.Z., 2020, November. Performance Review of Chemical Sand Consolidation and Agglomeration for Maximum Potential as Downhole Sand Control: An Operator’s Experience. In?SPE Asia Pacific Oil & Gas Conference and Exhibition. OnePetro.

Alakbari, F.S., Mohyaldinn, M.E., Ayoub, M.A., Muhsan, A.S., Abdulkadir, S.J., Hussein, I.A. and Salih, A.A., 2022. Prediction of critical total drawdown in sand production from gas wells: Machine learning approach.?The Canadian Journal of Chemical Engineering.

Ketmalee, T. and Bandyopadhyay, P., 2018, March. Application of Neural Network in Formation Failure Model to Predict Sand Production. In?Offshore Technology Conference Asia. OnePetro.

Abdelghany, W.K., Hammed, M.S. and Radwan, A.E., 2022. Implications of machine learning on geomechanical characterization and sand management: a case study from Hilal field, Gulf of Suez, Egypt.?Journal of Petroleum Exploration and Production Technology, pp.1-16.