Prediction of Sand Production in Gas Wells: Methods and Gulf of Mexico Case Studies
- J.S. Weingarten (Arco Alaska Inc.) | T.K. Perkins (Arco E&P Technology)
- Document ID
- Society of Petroleum Engineers
- Journal of Petroleum Technology
- Publication Date
- July 1995
- Document Type
- Journal Paper
- 596 - 600
- 1995. Society of Petroleum Engineers
- 2.4.3 Sand/Solids Control, 2.4.5 Gravel pack design & evaluation, 2.2.2 Perforating, 1.6.9 Coring, Fishing, 5.1.1 Exploration, Development, Structural Geology, 3.2.5 Produced Sand / Solids Management and Control, 1.2.3 Rock properties, 5.6.1 Open hole/cased hole log analysis, 1.14 Casing and Cementing
- 7 in the last 30 days
- 1,118 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 10.00|
|SPE Non-Member Price:||USD 30.00|
Sand production from weak, but competent, rock as a result of high production rates is a growing concern. In unconsolidated sand, the decision to gravel pack is usually clear; however, the decision is harder in weak rock because the need for sand control often depends on the desired drawdown or production rate. Also, wells that do not initially require sand control may later become sand producers. The ability to predict at what point sand problems will occur is useful. This paper presents a method for predicting sand production in gas wells and the results of applying that method to 13 fields in the U.S. gulf coast area. The method has since been applied extensively worldwide by Arco. In the fields addressed in this paper, rock strength was determined in one or more ways: core testing, log correlations by use of direct shear velocity measurements with the dipole sonic log, or log correlations with traditional sonic logs and calculated shear velocities. Rock strengths determined by core tests and log correlations are compared. The prediction method was incorporated into a simple log analysis program that facilitates quick identification and analysis of potential sand-producing zones, and this program is discussed. Field production data are presented and compared with theoretical predictions as well as with predictions based on traditional shear failure theories. The method presented differs from commonly used log-based sand-prediction models in two important ways. It models pressure gradients in the reservoir instead of assuming that all the pressure drop occurs at the perforation face. In addition, it allows for higher drawdowns than those permitted by the shear failure criteria, up to the drawdown that would produce tensile stresses at the perforation face. The method also addresses how allowable drawdown changes with reservoir depletion, which existing models do not consider.
Prediction of maximum sand-free production rate provides information for sand-control decisions and allows maximization of rate in those wells that are completed without sand control. In some fields, data, including both logs and core tests, are plentiful; but, in most, only minimal data are available for making this decision. The technique described in this paper was developed to work even with minimal data, while also taking advantage of extra data that might exist. The primary components of the method are prediction of rock strength, calculation of maximum drawdown for perforation stability, and calculation of reservoir failure. The tools are an analytical solution for hand calculation, a log analysis program for foot-by-foot analysis, and a spreadsheet for accelerating the hand calculation. We chose the analytical solution over a finite-element program because, in our experience, available data, timing, and resources generally do not justify the complexity of numerical simulation. Our goal was to have a method that would allow gravel-pack decisions to be made in the time period between logging and completion, typically a few days.
The minimum information required for the analysis are a log analysis, including sonic and density logs, gas properties (temperature, pressure, and gravity), and estimates of reservoir area, thickness, and depth. With this information, a synthetic shear velocity log is generated, rock strengths are estimated from correlations, and in situ stresses are estimated from rock properties. In addition, a complete set of data would include a dipole sonic log, confined compression and tension tests on core, and fracture gradient. The more data, the less uncertainty exists about the correlations.
The model predicts the onset of sand production and is not designed to apply to situations where some level of sand production may be allowable. Prediction of continuous or intermittent sand production requires a more sophisticated model, such as that described in Ref. 1. But for many situations, such as for the high-rate, high-pressure gas wells described in this paper, erosional and safety issues dictate that any sand production should be avoided.
The model also does not apply to most cases in which water is produced. In a water-wet rock, flow of water, which is the wetting phase, reduces the effective cohesive strength and leads to sand production at lower drawdowns than expected. This is addressed in Ref. 1 but is not incorporated into the simple model described in this paper. Chemical interactions may also occur between the water and the rock cementation that affect sand production, and these obviously are not taken into account.
All the fields analyzed are sandstone reservoirs in the Texas/Louisiana region of the Gulf of Mexico, either onshore or offshore. Depths ranged from ˜4,500 to 15,000 ft. Porosities varied from ˜20% to 37%. All the wells produced gas as a single phase, with minimal water production.
Rock Property Correlations
Rock strength was assumed to be defined by a Mohr-Coulomb failure criterion, characterized by an angle of internal friction and unit cohesive strength. From data presented in Ref. 1, the angle of internal friction was calculated from porosity with
From Ref. 2, the unit cohesive strength was calculated by
where Kb and E are in millions of pounds per square inch.
With these correlations, comparison of the predicted values with measured values are presented in the case studies below. In general, agreement was good. The angle of internal friction correlation is better for high stresses away from the perforation. At the low stresses around the perforation, the angle of internal friction may be higher than predicted owing to nonlinearity of the Mohr envelope.
|File Size||974 KB||Number of Pages||5|