Prediction of Reservoir Compaction and Surface Subsidence: Field Application of a New Model

Abstract

Summary A new loading-rate-dependent compaction model for unconsolidated clastic reservoirs is presented that considerably improves the accuracy of predicting reservoir rock compaction and surface subsidence resulting from pressure depletion in oil and gas fields. The model has been developed on the basis of extensive laboratory studies and can be derived from a theory relating compaction to time-dependent intergranular friction. The procedure for calculating reservoir compaction from laboratory measurements with the new model is outlined. Both field and laboratory compaction behaviors appear to be described by one single normalized, nonlinear compaction curve. With the new model, the large discrepancies usually observed between predictions based on linear compaction models and actual (nonlinear) field behavior can be explained. Introduction Withdrawal of fluids from hydrocarbon reservoirs can result in reservoir compaction and surface subsidence. Well-known examples are Goose Creek, Wilmington, Inglewood, Nigata, Bolivar coast, and Groningen.1 A considerable amount of scientific effort has been spent on the subject, which has resulted in a framework of theoretical and laboratory procedures that are commonly applied.4–12 It has become apparent over the last decade, however, that the application of these procedures to a number of well-documented field cases results in strong discrepancies between predicted and observed field behavior. Two basic types of anomalies have been recognized.Apparent nonlinear field behavior that has not been confirmed in laboratory compaction experiments. Examples are the Bolivar coast heavy-oil fields,13,14 Wilmington,15–17 and the San Joaquin Valley.18Differences between laboratory-measured and field-derived reservoir rock compressibilities.3 As compaction and subsidence can have environmental, technical, and financial consequences, it was decided to start an investigation into the causes of these discrepancies. The following possible causes were put forward:pressure-lag effects in neighboring clay layers;plastic deformation, e.g., caused by grain breakage;effects of the in-situ stress state;previous deeper burial of reservoirs during their geologic history;core disturbance; andloading-rate effects. Point 1 has been dealt with in a separate study19 that demonstrated that the pressure-lag effect cannot explain the observed discrepancies. Therefore, Points 2 through 6 have been considered in more detail. Experimental Results Uniaxial compaction experiments were carried out in oedometer and triaxial compaction cells5 to study Points 2 through 6. The experiments were carried out on more than 200 sandstone samples covering a large range of atmospheric porosities (5 to 38%) and various degrees of consolidation (see Table 1). In these experiments, the axial deformation of the core samples was monitored continuously as a function of the applied axial effective stress. Results of typical laboratory experiments are shown in Figs. 1 and 2. More experimental details on a large subset of the experiments carried out are given in Ref. 20. The compaction behavior of all samples studied in the various experiments was very similar, as summarized in Fig. 3. The following observations were made.Compaction curves at different but constant loading rates form a fan of lines (dashed curves, further referred to as "virgin" compaction curves). The lower the loading rate, the more the sample will be compacted at a given stress level. The compressibility along the virgin compaction curves will be denoted by cm, o.The shifts, ?sz, between compaction curves for different constant loading rates, sz, defined in Fig. 4 are quite systematic and related to stress level, according toEquation 1where b is a material constant that depends on rock type (see Prediction of Field Compaction Behavior). Nonuniaxial compaction measurements showed that there is hardly any effect of the horizontal-to-vertical stress ratio on the measured b values.Every time the loading rate is suddenly increased within a loading cycle, a compaction curve like A'-B' or B-C will result (Fig. 3). A sudden decrease in loading rate will result in a compaction curve like A-B. In both cases, the resultant compaction curve will finally reach the virgin compaction curve corresponding to continuous loading at the new loading rate.Interruption of loading results in creep (C-D and B'-C'). When loading is resumed, the sample shows a compaction behavior similar to that after an increase in loading rate (D-E and C'-D').The compaction behavior during reloading after partial unloading (E'-F') strongly resembles that observed after creep or after an increase in loading rate. (For large amounts of unloading, a different compaction behavior results that is not important for the scope of this paper.)The behavior described above has a large influence on the uniaxial compressibility, which is directly related to the slope of the curve of ?h vs. sz. For example, when the loading rate is increased, the uniaxial compressibility initially becomes much lower. Conversely, when the loading rate is decreased, the uniaxial compressibility initially becomes much higher.No major differences with respect to these findings were found between the compaction behavior of artificially sedimented sandpacks and that of moderately consolidated and unconsolidated (original and remolded) reservoir samples.The compaction behavior after an arbitrary loading history is restored to first-cycle behavior by not too large a mechanical disturbance (e.g., as encountered during the coring process and sample-preparation procedures).