The determination of residual gas saturation in gas reservoirs from long spontaneous and forced-imbibition tests is addressed in this paper. It is customarily assumed that when a gas reservoir is overlaying an aquifer, water will imbibe into the gas-saturated zone with the onset of gas production. The process of gas displacement by water will lead to forced imbibition in areas of high drawdown and spontaneous imbibition in areas of low drawdown. It is further assumed that in the bulk of the reservoir, spontaneous imbibition will prevail and the reservoir will be water-wet. A final assumption is that the gas behaves as an incompressible fluid. All these assumptions are challenged in this paper. A series of experiments is presented in which it is demonstrated that the residual gas saturation obtained by a short imbibition test is not necessarily the correct residual gas saturation. Imbibition tests by different methods yield very different results, while saturation history and core cleaning also seem to have a strong effect on the determination of residual gas saturation. It was found, in some cases, that the residual gas by spontaneous imbibition was unreasonably high. This was attributed to weak wetting conditions of the core (no water pull by imbibition). It is expected that this work will shed some new light on an old, but not-so-well-understood, topic.
When a porous medium is partially or fully saturated with a nonwetting phase, and a wetting phase is allowed to invade the porous medium, the process is called imbibition. For the problem addressed in this work, the nonwetting phase is assumed to be gas, and the wetting phase is assumed to be the aquifer water. If the medium is dry and the water is imbibing, then the imbibition is primary (Swi=0). If the water is already in the medium, the imbibition is secondary (Swi>0). If there is no driving force other than the affinity to wet, the imbibition is spontaneous. If there is any other positive pressure gradient, the imbibition is called forced.
Numerous papers have been written on the subject of residual oil saturation from imbibition, but fewer have been prepared on the subject of residual gas saturation from imbibition. The common perception is that many of the principles that cover oil and gas reservoirs are the same.
Agarwal1 addressed the relationship between initial and final gas saturation from an empirical perspective. He worked with 320 imbibition experiments and segmented the database to develop curve fits for common rock classifications.
Land2 noted that available data seemed to fit very well to an empirical functional form given as
In this model, the only free parameter is the maximum observable trapped nonwetting phase saturation corresponding to Sgr (Sgi=1). This expression does not predict residual phase saturation, only how residual saturation scales with initial saturation.
Zhou et al.3 studied the effect of wettability, initial water saturation, and aging time on oil recovery by spontaneous imbibition and waterflooding. A correlation between water wetness and oil recovery by waterflooding and spontaneous imbibition was observed.
Geffen et al.4 investigated some factors that affect the residual gas saturation, such as flooding rate, static pressure, temperature, sample size, and saturation conditions before flooding. They found that water imbibition on dry-plug experiments was different from waterflooding experiments with connate water. However, they concluded that the residual gas saturation from the two types of experiments was essentially the same.
Keelan and Pugh5 concluded that trapped gas saturation existed after gas displacement by wetting-phase imbibition in carbonate reservoirs. Their experiments showed that the trapped gas varied with initial gas in place and that it was a function of rock type.
Fishlock et al.6 investigated the residual gas saturation as a function of pressure. They focused on the mobilization of residual gas by blowdown. Apparently, the trapped gas did not become mobile immediately as it expanded. The gas saturation had to increase appreciably to a critical value for gas remobilization.
Tang and Morrow7 introduced the effect of composition on the microscopic displacement efficiency of oil recovery by waterflooding and spontaneous imbibition. They concluded that the cation valency was important to crude/oil/rock interactions.
Chierici et al.8 tested whether a reliable value of reserves could be obtained from reservoir past-production performance by analyzing results from six gasfield experiments. They concluded that different gas reservoir aquifer systems could show the same pressure performance in response to a given production schedule.
Baldwin and Spinler9 investigated residual oil saturation starting from different initial water saturation using magnetic resonance imaging (MRI). They concluded that at low initial water saturation, the presence of a significant waterfront during spontaneous water imbibition indicated that the rate of water transport was less than that of oil. At high initial water saturation, the more uniform saturation change during spontaneous water imbibition indicated that the rate of water transport was greater than that of oil. The pattern of spontaneous imbibition depended on sample wettability, with less effect from frontal movement in less water-wet samples.
Pow et al.10 addressed the imbibition of water in fractured gas reservoirs. Field and laboratory information suggested that a large amount of gas was trapped through fast water imbibition through the fractures and premature water breakthrough. The postulation was made that such gas reservoirs would produce this gas if and when the bypassed gas was allowed to flow to the production intervals under capillary-controlled action. The question of whether the rate of imbibition could enhance the production of this trapped gas was raised. Preliminary experiments on full-diameter core pieces showed that the rates of imbibition were extremely slow and that if the different imbibition experiments were performed in full-diameter plugs, the duration of the experiments would be prohibitively long. These experiments formulated the experimental strategy presented in the following sections.