Economical Management of Sand Production by a Methodology Validated on an Extensive Database of Field Data

Abstract

From a scientific point of view, sand production prediction is one of the most complex problems that has been addressed by the oil industry over recent years. Yet, the economical importance of a sound prediction methodology leading to an optimized completion strategy is such that many corporations have dedicated vast resources towards the solution of this problem. De facto surveying the literature reveals that a large number of models have been proposed to predict sand production. Unfortunately none of them has been validated to such an extent that it can be considered as a standard by the industry. As a consequence a large gathering of sand production data was undertaken. After proper data analysis it is possible to demonstrate that, from the engineering point of view, sand production is a relatively simple phenomenon governed by a few key parameters. Available models are prescreened for their capacities to account for these parameters. The selected models are then calibrated and validated against sand production field data with a sufficient degree of accuracy for operative purposes. Introduction From an industrial point of view, sand production prediction is essentially an economical problem. When the sand production risk is under-evaluated, the problems which have to be confronted are numerous: safety, downhole and surface installation erosion, well cleaning which may require major work overs and impose production delays, etc. On the other hand, when the sand production risk is over-evaluated, the problem often becomes the unnecessary reduction of the productivity of the wells which may require the drilling of further wells to achieve the targeted production for a given reservoir (an average productivity index reduction of 30 % for internal gravel packed wells is a fair figure derived from Agip's experience). Consequently, the ideal strategy for an oil company would be to know right from the beginning of the production of a well for how long it can be produced without sand control and then to decide when to do a remedial work-over and if such a work-over is economical when compared to the immediate placement of a sand control completion. Finally, as the critical conditions of sand production may vary greatly for different wells from a given reservoir, such an analysis ought to be made on a well to well basis. These technico-economical objectives being set, one can then start to establish the elements of an efficient methodology of sand production management:–a sand production prediction model which is simple and versatile enough to be run on a well to well basis and which can be integrated into 3D reservoir simulations to allow the economical assessment of different completion choices;–a sand production prediction model which is reliable enough to avoid major errors;–modelling the Earth being a very delicate science, a site specific calibration technique for the model should also be available;–finally, even when calibration has been made, a sand monitoring system should also be available to detect the precursory signs of sand production and avoid major accidents. Knowing that many scientific studies had been performed over the years about sand production prediction. it was decided to avoid the duplication of efforts and to start from a completely different perspective. Keeping in mind the industrial objectives and prerequisites set above, the study therefore followed the five steps listed below:gather an important amount of sand production data from the field;analyse these data along with laboratory measurements to try to understand the physics of sand production as revealed by the field evidence;identify the key parameters governing this physics;calibrate available models on such data and evaluate which ones were the most reliable and the simplest; P. 227

Numerical Evaluation of Geomechanical Parameters Affecting Productivity Index in Weak Rock Formations-Part 1: Theory

“…The proposed numerical model can capture the episodic nature of sand production and quantify the resulting volume of sand and its concomitant impact on productivity (or change of skin) with drawdown. This is in contrast with other analytical (7)(8)(9)(10) and numerical (11)(12)(13)(14) sand production prediction models that basically identify the onset of sanding. However, this model cannot be applied to wormhole development in heavy oil sands where quite different sand production mechanisms are involved.…”

“…Assuming the trial stress increment is ∆σ T , the final stress becomes σ T = σ A + ∆σ T and the contact stress can be expressed as: The first estimation of the reduction factor β in the above equation is obtained using: (8) Due to the nonlinear characteristic of the yield surface with respect to stress, the first trial of the contact stress will not reach point E. The reduction factor is, therefore, revised iteratively by: ...................................................................................(9) with . The iteration can be halted when the departure of the stress from the yield surface reaches a certain tolerance, expressed as a relative number with respect to the equivalent uniaxial yield stress σ Y ; that is, the iteration halts when is less than a specified value, such as 10 -6 .…”

Prediction of Sand Production from a Saudi Sandstone Reservoir

“…This can be carried out by calculating the value of (t c ) from the following equation [16]: (1) where S(x) is given by (2) The value of (t c ) calculated from Eq. (1) is then compared with the value of (t t ) obained from statistical tables at a specified confidence level.…”

“…These amounts of produced sand vary from a few grams or less per ton of produced reservoir fluid to huge amounts possibly leading to complete filling of the borehole [1]. When oil or gas wells are drilled into unconsolidated or poorly cemented sandstone reservoirs, the decision about the need for sand control is almost clear.…”

Practical Approach to Achieve Accuracy in Sanding Prediction

“…5,6 Empirical Approaches Using Field Data. This approach probably remains the most widely used within oil companies.…”