Most sand production prediction models to date have the capability to indicate whether initial sand production may take place during the lifetime of a reservoir but they are unable to predict whether the sand production will be ‘problematic’ (excessive erosion, plugging, well sand-up, separator fill, etc.), particularly for systems that have some tolerance towards sand production.
In order to predict whether sand production will be ‘problematic’, it is important to estimate sand volumes / rates as a function of bottom-hole load conditions, drawdown, time, etc. This paper presents a model to predict sand volumes / rates for any type of clastic oil or gas reservoir. This model captures both the geomechanical aspects (rock deformation and failure) and the transport aspects (e.g. role of drawdown and watercut) of the sand production problem. It was extensively validated comparing predicted sand volumes and rates with field observations for a variety of oil and gas fields under various stages of depletion. For all these cases, the model predictions and field observations are in reasonable to good agreement.
The main focus of the current paper is the discussion of bean-up procedures, i.e. the question is addressed as to whether an appropriate bean-up strategy (e.g. small bean-up steps) may reduce the cumulative volume of produced sand.
We show that for wells without sand control, the cumulative amount of sand produced to surface is independent of bean-up procedure. If a large sand production volume is expected in such a well, beaning it up in small steps not only increases the risk of sand-up in itself, but also the risk of repeated sand-ups before the well can finally be produced. By contrast, for wells with sand control, the risk of plugging by fines mobilization is reduced by beaning up in small steps.
Conventionally, sand production prediction models have the capability to indicate whether initial sand production may take place somewehere during the lifetime of an oil or gas field 1–6. However, these models are unable to predict whether the sand production will be ‘problematic’ (e.g. in terms of erosion, plugging, well sand-up, separator fill, etc.), in particular for systems that have some tolerance towards sand production.
In order to predict whether sand production will be ‘problematic’, one needs to be able to estimate sand volumes and rates as a function of, amongst others, bottom-hole load conditions, drawdown, and time. In recent years, a number of sand rate prediction models have been published in the literature 7–14. However, owing to the scarcity of "dedicated" sand production field data (for example, Sand Influx tests), the apparent inconsistency that is often exhibited by production-related field data, and the fact that laboratory sand production experiments to date were mostly done on relatively small samples, there is still a wide variety in the understanding of the fundamental physical processes underlying sand production.
Previous sand failure prediction research has shown that the primary role of fluid flow (drawdown) in sand production is the transport of loose sand (debris) resulting from compressive rock failure, rather than failure of the intact sandstone itself (see, for example, refs. 5,15). This has led to the general consensus that sand failure is a necessary, but not sufficient condition for sand production. However, once the sand has failed, what are the exact circumstances that lead to its production? And what are the mechanisms by which the sand cavity re-stabilizes after initial failure? It is here that the opinions start to diverge.
In Refs. 7–11, sand production is directly coupled to an increase in porosity around the wellbore (perforation) as a result of volumetric plastic strain. Post failure stabilization (PFS) of the cavity after initial failure is assumed to result from local permeability increase as a result of sand production. This permeability increase in turn reduces the flow-induced drag forces on the sand grains, and therefore reduces sand production. In Refs. 7–9, sand production is directly coupled to the volumetric plastic strain, whereas in 10,11 it is coupled to the product of plastic strain and fluid flow via a semi-empirical "erosion criterion". None of these models addresses the impact of watercut on sand production.