Light hydrocarbon solvent was used to produce bitumen from Cold Lake oil sand in three dimensional scaled physical modelling experiments. Injection of the gas was carried out through a horizontal well until the bottom hole pressure was slightly above the saturation pressure at model temperature. Production of diluted bitumen followed the injection cycle through the same horizontal well until pressure in the reservoir was depleted slightly below the saturation pressure at model temperature. At that pressure liquid production rates were very low. In some of the experiments with ethane gas, pressure in the model varied between 3.2 and 4.0 MPa, which is near the Cold Lake reservoir pressure. This favourable physical property of ethane was offset by a smaller improvement in product quality as compared to that from tests with propane. The results of these experiments show that the production rate of bitumen assuming a gravity drainage drive mechanism with a single horizontal well cyclic process was significantly higher than what could be expected from the molecular diffusion rate of solvent into the bitumen, indicating that other mechanisms, probably solvent dispersion or fingering, are important in the mass transfer of solvent into the bitumen. Reasonable efficiency and effectiveness of ethane as solvent were achieved, resulting in a small loss of solvent and low solvent to bitumen replacement ratio in the experiments. Based on the measurement of the residual oil saturation along the wellbore, full utilization of the horizontal well was not achieved in the model tests. Production enhancements would be expected if this technical deficiency could be overcome. Introduction Half of the original bitumen in place in Imperial Oil's leases at Cold Lake, Alberta is located either in bottom water reservoirs or water sensitive sands which are not amenable to exploitation by steam based processes. A potential alternative for these reservoirs is a solvent based process which has been the subject of laboratory investigations by a number of researchers(l-9) in recent years. The advantages of the solvent based processes are: little heat loss and limited water handling; the disadvantages are: high solvent cost and inherently low production rate limited by mass transfer of the solvent into the bitumen. The recovery process utilizing light hydrocarbon gas was first investigated by Dunn et al.(1,2) on Athabasca oil sands and later by R.M. Butler(3–9) on heavy oil and Peace River bitumen in a two dimensional scaled physical model. In their investigations, both researchers modelled solvent assisted gravity drainage with a pair of horizontal injector and producer wells in much the same way as the Steam Assisted Gravity Drainage (SAGD) process. The present study applied the concept of cyclic stimulation with a solvent gas through a single horizontal well. Compared to the two well SAGD type configuration, the advantage of this process concept is the halving of well costs. As illustrated in Figure I, solvent gas is injected to the horizontal well until the bottom hole pressure exceeds the saturation pressure of the solvent (3.6 MPa for ethane at 13 °C).

This paper entails the implementation and interpretation of the first successful interwell test ever reported in the industry to determine residual oil saturation in a gas-saturated reservoir. Two interwell tests were conducted in Golden Spike to measure the residual oil saturation to gas-flood at two different depths. This method, which involves the comparison of the whole partitioning and non-partitioning tracer curves to derive residual oil saturation values, is an improvement of the original method that compared only the breakthrough times. A chromatographic transformation technique was developed for curve comparison so as to avoid tedious simulation for data interpretation. For layers with different residual oil saturation, the transformation method works only for a pseudo single porosity reservoir with ordered layers, i.e. low residual oil saturation for a high permeability layer. The first test indicated that there were three layers with residual oil saturations of 7%, 15% and 20%. The results from the second test conducted in a lower production interval were masked by the presence of extensive fractures in the production zone. In spite of the interference from the fracture production, the residual oil saturation in some flow channels could still be estimated to be about 12%. Because it is unlikely that the tracers could enter the matrix during the test, the residual oil saturation measured is probably the oil saturation in some secondary channels. Sulphur hexafluoride, F13BI (brome-trifluoro-methane) and F12 (dichloro-difluoro-methane) were selected as the tracers from the previous lab tests. The tracers were pre-mixed and injected as a liquid. A Freon phase behavior program was developed to calculate the exact amount of the Freon's injected. Introduction Upon evaluation of various conventional methods, the interwell tracer test(1) has been identified as the most reliable means to determine residual oil saturation in Golden Spike, a low-pressure, low-porosity, gas-saturated carbonate reservoir. The original interwell method disclosed by Cooke(2) in 1971 involved the comparison of the relative breakthrough times of the partitioning and non-partitioning tracers for residual oil saturation calculation. Breakthrough time is not a well-defined quantity, as it is often obscured by dispersion, the detection limit, and most importantly, by the streamline and layer distributions. As a result, the interpretation technique, certainly inadequate, draws criticism(3,4). Consequently, because of the lack of suitable chemicals and interpretation technique, no single test has been tried in the field or, at least, published in the literature. To circumvent the problems anticipated in Cooke's method, our interwell method employed a whole curve comparison to derive a residual oil saturation value at any location on the curve using a simple "landmark" comparison technique. Under ideal conditions, residual oil saturation can be determined by layers. To demonstrate the feasibility of the method, extensive tests were performed in the lab(5). Slim tube displacement results indicated that residual oil saturation could be measured in an accuracy of± 1 % pore volume from the separation of tracers. This paper entails the design, implementation and interpretation of the two field tests conducted in Golden Spike in 1987.

In a three dimensional scaled physical modeling experiment, Cold Lake oil sand was subjected to cyclic stimulation with supercritical ethane through a single horizontal injector/producer well located at the base of the reservoir. Hot ethane gas was injected into the bottom of the formation until the bottle hole pressure reached slightly above the ethane supercritical pressure of 4.9 MPa. Production of diluted bitumen followed the injection cycle through the same horizontal well until pressure in the reservoir was depleted slightly below 4 MPa. At that pressure liquid production rates were very low During the experiment, temperature in the model varied considerably with the distance from the wellbore as well as with the number of stimulation cycles. Supercritical ethane enhanced the cyclic solvent gas process by improving the early production rate. As well there was an increase of recovery of solvent in the blowdown at the end of the experiment. Both are important factors for the process economics. Time animation of the temperature profile revealed that solvent gas preferentially invaded the region in close proximity to the production end of the horizontal well. The observation was consistent with the post test measurement of residual oil saturation in the model which showed that bitumen was preferentially depleted from that region. Evidently, full utilization of the horizontal well was not achieved in the experiment. Production enhancements would be expected if this technical deficiency could be overcome. Introduction The current in situ recovery process practiced at Cold Lake Alberta by Imperial Oil Resources Limited (IORL) uses high pressure steam to reduce bitumen viscosity and mobilize the oil. The process has been applied successfully for many years to produce bitumen from good quality sands from the thick Clearwater formation. However, only a small fraction of IORL's reserve holding at Cold Lake is in the good quality reservoir, with the rest categorized as either thin, shaley or bottom water reservoirs. P. 521

The residual oil saturation to gas-flood is an important parameter to evaluate the potential of the Golden Spike D3 ‘A’ pool as a candidate for enhanced oil recovery and gas storage. Conventional methods, including logging, sponge coring and single well tracer testing, are not applicable to this low-pressure, low-porosity, gas filled carbonate reservoir. An interwell tracer method which works on the chromatographic separation of tracers with different Henry's law constants was therefore proposed and tested in the lab. Stringent selection criteria based on the detection limit and Henrys law constant were established to screen chemicals for application. According to the criteria, sulphur hexafluoride and halocarbons (Freon), which can be detected in the sub ppm range using a gas chromatograph equipped with an electron capture detector, were selected for this study. Henry's law constants were determined experimentally using a static equilibrium method and a dynamic slim tube displacement method. In the preliminary screening of chemicals for lab testing, the Henry's law constant could be estimated from vapor pressure after correcting for the non-ideal behavior using the regular solution theory. It was demonstrated in the slim tube tests that residual oil saturation could be determined within I pore volume % accuracy from tracer separation using the simple chromagraphic theory. A mixing cell model was also developed to simulate the slim tube test results. This model was also capable of handling dual porosities that are common to carbonates. Introduction The Golden Spike D3 ‘A’ pool(1,2), located 29 km southwest of Edmonton in central Alberta, is a carbonate reservoir. With the on-coming of the Beaufort Sea gas development, Golden Spike is being considered as a candidate for Beaufort gas storage and enhanced oil recovery. In as much as both projects have substantial economic incentive and are somewhat adversely affected by each other, the two projects need to be carefully evaluated using the best possible data. Residual oil saturation to gas-flood is the key parameter for the evaluation of either process. Various conventional methods for Sor determination, such as production history, sponge coring, laboratory gas floods, logging and single-well tracer testing, have been extensively reported and compared in the literature(3,4). Unfortunately, these methods are not satisfactory for low-pressure, low-porosity, gas-filled carbonate reservoirs such as Golden Spike. Therefore, an interwell method(5,6) using Freon's with different vapor pressures was proposed and studied in the lab. The interwell method works on the chromatographic separation of Freon in the reservoir. According to the chromatographic theory, the Freon with the highest Henry's law constant (or K value) or the lowest solubility in oil is produced first. Thus, by comparing the production profiles of various Freon's, the average residual oil saturation between wells can be determined. Although the principle of the interwell method was disclosed by Cooke(6) in 1971, not a single test has been reported in the literature due to a lack of suitable chemicals and the potential interpretation problems(3). The Golden Spike test is the first attempt ever in the industry to apply the theory to the field.

No abstract