A miscibility study was conducted to investigate the effects of the C5 + content in injection gases on Minimum Miscibility Pressure (MMP). Currently available common MMP correlations are based on data from injection gases containing no C5 + fraction. However, laboratory analyses suggest that having a small (and yet relatively significant) C5 + fraction could have a large effect on reducing the MMP to considerably below the predicted values from existing correlations. This paper aims at investigating possible benefits for miscibility and CO2 EOR flooding with impurities, particularly with C5 + fraction. Previously, a PVT and miscibility study was conducted on a Cooper Basin (located in central Australia) reservoir fluid, including solubility-swelling and viscosity studies. Slim tube tests were performed using two injection gases - pure CO2 and a CO2-rich synthetic gas (80mol% CO2), which contained a small amount of C5 + fraction (0.3mol%). The MMP was expected to increase substantially due to the presence of a large amount of methane (15mol%). Correlations from literature predicted the MMP to be above 3433psia, but the MMP was measured to be 2880psia[1]. As most currently available correlations are based on injection gases with no C5 + fraction, a major contributing source to error with these correlations may have been the C5 + content of the gas. In order to investigate the effects of the C5 + fraction in the injection gas, a series of Rising Bubble Apparatus (RBA) experiments were conducted on the reservoir fluid with three injection gases at three different temperatures to investigate the effect of C5 + in the injection gas stream. Results indicate that the C5 + fraction does have a significant effect on the MMP that needs to be factored into current correlations when the injection gas has a C5 + content. Introduction A PVT and miscibility study was performed on a reservoir fluid from the Cooper Basin in central Australia (Oil A) with the purpose of evaluating the reservoir as a possible CO2 EOR candidate[1]. The composition of Oil A can be found on Table 1. The miscibility study with the reservoir fluid was performed with two injection gases; pure CO2 and a CO2-rich synthetic gas made up from a mix of pure CO2 and gas from a nearby high CO2 gas field (approximately 45 mol% CO2). The gas was sampled at the wellhead with a high flowing wellhead pressure (1705psia at 144°F). As a result, it retained in the vapour phase the heavier components present in the stream, up to and including C9. When the synthetic gas stream was prepared, this heavy fraction was carried over into the synthetic gas[1]. When the miscibility study was conducted using this gas the measured Minimum Miscibility Pressure (MMP) was much lower than published correlations had predicted. The measured MMP was 2880psia[1] whereas correlations predicted it to be above 3433psia[2–13]. This led to the concept that the C5 + fraction in the gas stream might have a greater effect on the MMP than what the correlations had predict. As a result, it was decided to further investigate the effect and try to evaluate the extent of the effect of the C5 + content by measuring MMP with a further enriched gas and at a range of temperatures. Most MMP correlations predict changes in injection gas composition through the critical temperature (Tc), and thus incorporate any component, including C5+. However, one possible reason why these correlations may not accurately predict the MMP is due to the fact that the injection gases in the laboratory data sets used in developing them contained no C5 + fraction. Emera and Sarma (2005)[3], presented a good summary of existing correlations for both pure and impure CO2 MMP and MMP data sets. The C5 + fraction in the synthetic gas used was 0.3mol%, and thus had little effect on the parameters that take the fraction into account in correlations, such as Tc. However, the effect produced by this small quantity of C5 + (yet large relative to common injection gases) may be greater than what current correlations predict, and hence, the motivation for this study.

Limited experimental work has been performed to study the effect of nC5 on Minimum Miscibility Pressure (MMP) when added to impure CO2 as an injection gas for flooding processes. Furthermore, current MMP correlations have been developed based on data where the injection gas contains no nC5 fractions. Consequently, their MMP prediction is not highly accurate when nC5 exists in the injection gas. In order to further investigate the effect of nC5 on impure CO2 MMP, this work presents the experimental investigation and subsequent development of a correlation to model this effect. Using Rising Bubble Apparatus (RBA), MMP data points were measured on three reservoir fluids and four injection gases at three temperatures. The three reservoir fluids were relatively light with stock tank gravities in the range of 44–52 °API and GOR's 100–1300 scf/stb. The base gas composition was 92mol% CO2 and 8mol% C1. To this nC5 was added to create gases with 1mol%, 3mol%, and 5mol% nC5. The nC5 was noted to decrease the MMP. Based on the experimentally measured MMP data set, a strong trend was found. The effect of addition of nC5 on the impure CO2 MMP has been correlated using Genetic Algorithm (GA). This correlation depends on the injected gas composition and critical properties. The GA-based correlation predicted the nC5 effect with greater accuracy than the currently available correlations, giving an absolute error of only 3.4%. It is envisaged that this work can be used to investigate the possible benefits in retaining the nC5 fraction in the injection gas and to provide a method of evaluating whether it is worthwhile to add or retain a small amount of nC5 to the injection gas to improve miscibility with the reservoir fluid. Thus, more oil can be recovered and the added injection gas expense may perhaps be justified. Introduction The Minimum Miscibility Pressure (MMP) of three reservoir fluid samples was measured at three temperatures (60, 80 and 100°C) using four injection gases by using the Rising Bubble Apparatus (RBA). The base gas, Gas#1, was created by mixing CO2 (92mol%) and methane (8mol%). To this, 1mol%, 3mol% and 5mol% nC5 was added to make Gas#2, Gas#3 and Gas#4 respectively. A more detailed summary of the gas compositions can be found in Table 1. The reservoir fluids were all relatively light, with gravities in the range of 44–52°API, however they ranged in the solution gas-oil ratio (GOR) from 100–1300scf/stb. A more detailed summary of the reservoir fluid composition and properties are provided in Table 2. Details of the project that led to the study on nC5 can be found in previously published articles reported in the appendix1,2. Based on the MMP data obtained, a correlation was developed by Genetic Algorithm (GA), one of the artificial intelligence techniques. Using a software developed by Emera and Sarma in 20053, a correlation was developed to model the effect of nC5 on the MMP. The flowchart of the software is presented in Fig. 1. Further to this, the software of @Risk from Palisade Group4 was used to develop the sensitivity analysis to study which of the input parameters of the correlation has a larger impact on the MMP. This is presented in Fig. 2. The results of the correlation were compared to those of commonly used correlations, namely the Alston et al.5 and Sebastian et al.6 correlations. Overall, the GA correlation presented in this paper showed improvement on the accuracy of the predicted MMP. Alston et al.5 correlation had an absolute error of 10.4%, Sebastian et al.6 correlation had an absolute error of 5.3%. We compared these correlations to the GA correlation developed in this work which had an absolute error of only 3.4%. Comparisons of the data are presented in the results.

No abstract

No abstract

TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractPVT fluid properties are an integral part of determining the ultimate oil recovery and characterization of a reservoir, and are a vital tool in our attempts to enhance the reservoir's productive capability. However, as the experimental procedures to obtain these are time consuming and expensive, they are often based on analyses of a few reservoir fluid samples, which are then applied to the entire reservoir.