A sensitivity analysis was carried out during a course of study to develop a model for predicting sand production from Gulf Coast gas wells that produce free water. A multiple linear-regression analysis was used to incorporate data from producing gas wells and log-derived properties of reservoir rock in a useable model. This model fits the field data of water-producing gas wells. It will be a risky proposition to analyze each individual parameter related to sand production when designing sand-control measures. This study shows that the combined effects of these parameters are making a significant contribution in the process. The results indicate that the volume of water is not relatable, but the presence of free water (uncondensed from the gas phase) increases the sanding tendencies of most gas wells. As a reservoir depletes, its tendency to produce sand increases; efforts to reduce sanding should be directed toward reducing the drawdown across the completion; and many of the log-derived parameters that show merit in correlating the sanding tendencies of dry gas wells have no value in water producing gas wells.
Numerous solutions to halt sand production from oil and gas wells have been attempted1 with various degrees of success. The most prevalent remedy is the gravel-pack completion, which blocks the influx of loose sand with specially selected gravel held in place by screens. This method is particularly expensive but not nearly as costly as losing a producer. Therefore, it is vital to know whether a well will produce sand before it is placed on production. The economic implications of sand problems are critical, requiring continuous improvement in sand-control techniques and sand production prediction methods.
Production problems attributed to sand production are sand fill-up and bridging inside the wellbore that shut off production; erosion damage to downhole tubular goods, safety valves, and artificial-lift equipment; downhole casing and formation damage that sometimes causes premature abandonment of completions; sand accumulations in surface lines and equipment; and abrasive wear on surface chokes, valves, and pipes. One of the most prevalent problems associated with sand control is localized screen erosion.
There are, however, loosely consolidated sand reservoirs in known sand-producing areas from which economically attractive production rates might be obtained without the use of any form of sand control. This class of reservoir rocks, which are strong enough to produce without sand control, may have high-order permeabilities, but they are never enormous. This observation makes the high permeability of most Gulf Coast sands desirable from the standpoint of high deliverability but a nuisance in terms of sand-production problems. Therefore, the ability to predict which wells will require sand control has great economic value to gas producers.
Most models that this industry has used (with varying degrees of success) to predict sand production have had their origins in openhole electric-log analysis. The usual rock properties that are inferred from the logs are the shear modulus, Young's modulus, bulk compressibility, and Poisson's ratio. There is no certainty or strong consensus that rock properties can be accurately calculated from electric-log response. Independent determination of dynamic (from log response) and static (from cores) rock properties is routinely done in the laboratory, but the properties measured with both techniques often do not agree. From the practical standpoint of evaluating friable sands, several important considerations favor the use of dynamic measurements obtained from well logs. The measurements are first made in situ and, therefore, should be fairly representative of the confining stresses the formation will experience at completion. Conversely, the static measurement requires the recovery of an unaltered core. Second, the dynamic measurements obtained from well logs provide continuous curves that reveal changes and trends.1
A study of 16 U.S. Gulf Coast wells2 concluded that Mohr's stress-analysis technique with a 200-psi (1 .378-mPa) safety factor would have been a viable method for making sand-control decisions. Although Mohr's stress-analysis technique, referred to here as the dry model, is good when applied to wells with no water production, its limitations were quickly realized. Even though the experience to date has been predominantly with gas and oil production, it is believed that production with a high water cut may require high intrinsic shear strength. This threshold criterion for water-producing sands has yet to be established.
Weissenburger3 also realized the need for a system to predict sand production. An engineering system provided an iterative pathway to integrate rock mechanics, geology, logging, and reservoir- management information. Morita4–5 provided a numerical model and a parametric study of sand-production prediction without the effect of water production.
An analytical method for predicting sand production in gas wells that make water is not currently available. This study, however, uses field data and existing correlations to extend Mohr's stress-analysis method, as developed by Coates and Denoo,6 to include depletion and water production in gas wells. The result is presented in the form of a multiple linear-regression equation that is based on theoretical presentation by Ghalambor7 and Koliba.8 Now that the ever-present aspects of water production and the depletion of gas wells are more accurately modeled, a "life of the well" approach can be introduced to the completion decision rather than an evaluation of the well at only one point in time.