Limitations for Compositional Modeling in Vaca Muerta Numerical Simulation
You have access to this full article to experience the outstanding content available to SPE members and JPT subscribers.
Vaca Muerta formation fluids run from medium black oils to very dry gas, according to source-rock maturity. High-shrinkage volatile oil and very rich gas condensate are found and compositional variation is areally continuous. Despite this huge difference in fluid nature, where saturation pressures are present, more than 200 kg/cm2 undersaturation is reported. Because of the complex dependence of fluid behavior with composition for near-critical gas condensate and volatile oil, the use of compositional modeling is mandatory. Cubic equations of state (CEOS) are the only types implemented until now in commercially available compositional numerical simulators. In this paper, the authors show the limitation of CEOS for modeling reservoir behavior of liquid-phase black and volatile oil in highly undersaturated reservoirs.
The Vaca Muerta shale, located in the Neuquén Basin, Argentina, ranks among the top shale oil and gas resources in the world. It is characterized by its areal extent, high gross thickness, and high organic content. The first well was drilled and completed in 2010 and despite more than 650 wells to date, Vaca Muerta shale history is relatively recent compared with other known shale plays.
Source-rock thermal maturity seems to be the main driver for fluid distribution, running from dry gas in very mature zones, progressing through gas condensate and volatile oil, to black oils of less than 100 m3/m3 gas/oil ratio (GOR) in immature zones.
Production forecasts on new wells are based primarily on the history of geographically close wells, but forecasts for new zones involve numerical simulation of regionally distributed properties, including reservoir-fluid characteristics. The original aim of the work described in this paper was the equation of state (EOS) description of the pressure/volume/temperature (PVT) properties of fluids in known zones, to use in extrapolation to those zones where information is sparse. The models and work flows described in papers SPE 177058 and URTeC 2436107 were the starting point for this paper; in those earlier works, the authors aimed to extend the work flow into more complex fluids.
An extensive EOS fitting program was launched. Surface and subsurface samples were taken for fluid characterization and PVT tests were performed for each. All presented, at least, 200 kg/cm2 of undersaturation (Fig. 1).
Equations of State
For decades, the PVT properties of reservoir fluids have been determined in specialized laboratories in which various tests to accurately assess thermodynamic behavior are carried out at initial conditions and through the fluid’s evolution as the reservoir is exploited. Alternatively, relatively simple correlations aim to obtain quick, reliable values of the PVT properties from certain easily measurable parameters. In general, these correlations have been obtained from experimental data collected by the authors, and in many cases, the range of validity of these correlations is limited severely by the data bank.
In the 1970s, the possibility of using EOS (complex analytical expressions between PVT and composition) to reproduce the thermodynamic behavior of a mixture of hydrocarbons began to be analyzed with growing interest.
Unlike correlations, use of EOS provides a rigorous theoretical base and a wide range of application, although originally, poor adjustments in the liquid densities were often the result. Next, various modifications, including CEOS, were proposed. Using EOS in technical software, it is possible to simulate the thermodynamic behavior of hydrocarbon mixtures. The thermodynamic simulation is carried out in two clearly differentiated stages: a regression or adjustment stage, and a prediction stage.
The first stage attempts to reproduce, in the most-accurate possible way, one or more parameters or properties measured experimentally in the PVT tests. This stage is critical. If a good adjustment is not achieved, it is difficult to make a good prediction of the evolution of the fluid thermodynamic behavior. The basic PVT experiments that usually are used in the regression are saturation pressure, constant-composition tests, differential-liberation or constant-volume tests, and separation tests.
Oil withdrawal, pressure, and compressibility are closely related. In fact, recovery for a given mean reservoir pressure drop is in direct proportion to compressibility. Even though shale oil plays are far from the possibility of such a simplistic description between production and pressure evolution, the main idea, according to the authors, is to make clear that compressibility, instead of volume, should be the target for the EOS description. Reservoir-pressure evolution with production and recovery calculations relies largely on compressibility, which itself is not a fitting parameter for usual CEOS. A very good fit for liquid volumes in CEOS—less than 5% maximum error in volume—leads to unacceptable differences in compressibility (over 30%).
To compare the consequences of this difference in terms of production, a black-oil model was run with three different fluid scenarios: one with experimental compressibility and two with compressibility obtained from CEOS. In the model, all formation, fracture, and other fluid properties are the same, as well as the bottomhole pressure (BHP) imposed.
Compressibility was used as a parameter to qualify the model obtained in each case. This was achieved by calculating the ratio in original oil volume that, keeping the same pressure drop, led to the same expansion, calculated from the observed and calculated change in monophasic volume.
Three fluids were selected from three wells: a low-GOR black oil from Well 11, a high-GOR black oil from Well 9, and a volatile oil from Well 3. For all the fluids, the Peng-Robinson CEOS with volume correction was selected. The fluid description was performed with 24 components. For each model, the main fitting parameters were saturation pressure and liquid volume as measured in a constant composition-expansion experiment, while the remaining parameters were checked for overall consistency. The decision was made to compare the behavior of monophasic volume change and its consequences in reservoir calculations. Details are presented in the complete paper.
Simulation Model and Results
To account for the lack of representativeness in compressibility for compositional models, it was decided to avoid any influence in phase change by running the models in black-oil mode. For this purpose, the volumetric behavior was exported in dead-oil mode from the fitted EOS, and an equivalent black-oil model was built from the experimental behavior. The reservoir-simulation model was the result of a simulation work flow aimed at integrating all available information and coupling explicitly hydraulic-fracture geometry and reservoir production with the following steps:
- Integrate all relevant information into a static model
- Simulate the hydraulic-fracture geometry
- Set up a dynamic model accounting for the stimulation treatment
- Simulate production and compare against observed history
The volumetric parameter used in numerical simulation was the formation volume factor, which depends on the stock-tank oil volume and the reservoir volume. Then another model was built, with the formation volume factor adjusted to the same experimental mean value, but with the calculated change in volume. The paper presents a summary of the work flow and the fluid models.
The results from numerical simulation showed that rates and cumulative production from CEOS compressibility scenarios were approximately one-half those obtained with experimental compressibility. Part of this difference may be explained from fluid behavior alone, but the trends in pore pressure also differed, adding to the total difference in production. The final field-pressure maps for each model are presented.
- CEOS is the only available tool for compositional modeling in commercial numerical reservoir simulators.
- CEOS is limited for representing oil-volume changes with pressure.
- Differences as low as 5% in relative volume can generate up to 30% difference in fluid compressibility. This difference is critical for production forecasting in highly undersaturated fluids.
- Vaca Muerta presents a high degree of subsaturation, and there are substantial resources with fluid characterized as volatile oil.
- Compositional simulation is the best approach for volatile oils, where mass transfer between the oil and gas phase is significant. On the other hand, compositional modeling is limited for adequately representing compressibility.
- Production results from simulation show that a 5% difference in relative volume for the fluid can result in up to 50% difference in the final recovery because of the lack of fit in compressibility.
Limitations for Compositional Modeling in Vaca Muerta Numerical Simulation
01 July 2019
Don't miss out on the latest technology delivered to your email weekly. Sign up for the JPT newsletter. If you are not logged in, you will receive a confirmation email that you will need to click on to confirm you want to receive the newsletter.
09 July 2019