This paper describes the results and interpretation of a large-scale laboratory sand production test. In the experiment, a real well with multiple perforations was simulated, using an outcrop rock selected to represent core material from a specific field. The objective of the experiment, which was the first of its kind, was to investigate the influence of both effective stress increase and drawdown on sand production behavior, taking into account the influence of the presence of casing and cement and of perforating. An additional objective of the test was to investigate the influence of a water cut on the sand produced. The laboratory well behaved very realistically, in terms of both oil and sand production.
Reliable predictions of sand production potential are required to make realistic sand production management and contingency planning possible. Unnecessary application of sand exclusion measures results in increased completion costs and considerable loss of well productivities. Further, sand prediction may assist in selecting the most attractive sand control techniques .
Although, over the years, a large number of conceptual, analytical and numerical models for sand production prediction have been developed, see e.g. [2–10], the value of these models may still be questioned, considering the discrepancy observed between the model predictions and field observations. To improve and validate the sand prediction models, reliable sand production data are indispensable. The sand production data available can be divided into laboratory sand production test data and field sand production data collected from real wells. Laboratory sand production tests typically concern small-scale simulation of flow through perforations or cylindrical cavities in stressed cylindrical samples (see e.g.). Advantages of such laboratory tests include the controlled stress and pressure boundary conditions, the extensive monitoring facilities available and the simple geometry used, which facilitates interpretation of the experiments. Field observations involve far more complex situations (e.g. perforation interaction), with many uncertainties concerning the actual downhole situation and only limited possibilities of imposing prescribed boundary conditions. Nevertheless, interpreting and predicting such field observations are the prime objective of sand prediction modeling.