Uncertainty Quantification by Monte Carlo Simulation of Lab-Derived Saturation Data from Sponge Cores

Abstract Fluid saturation data obtained from core analysis are used as control points for log calibration, saturation modeling and sweep evaluation. These lab-derived data are often viewed as ground-truth values without fundamentally understanding the key limitations of experimental procedures or scrutinizing the accuracy of measured lab data. This paper presents a unique assessment of sponge core data through parameterization, uncertainty analysis and Monte-Carlo modeling of critical variables influencing lab-derived saturation results. This work examines typical lab data and reservoir information that could impact final saturation results in sponge coring. We dissected and analyzed ranges of standard raw data from Dean-Stark and spectrometric analysis (including, gravimetric weights, distilled water volumes, pore volumes and sponge's absorbance), input variables of fluid and rock properties (such as, water salinity, formation volume factors, plug's dimension and stress corrections), governing equations (including, salt correction factors, water density correlations and lab mass balance equations) and other factors (for instance, sources of water salinity, filtrate invasion, bleeding by gas liberation and water evaporation). Based on our investigation, we have identified and statistically parameterized 11 key variables to quantify the uncertainty in lab-derived fluid saturation data in sponge cores. The variables' uncertainties were mapped into continuous distributions and randomly sampled by Monte-Carlo simulation to generate probabilistic saturation models for sponge cores. Simulation results indicate the significance of the water salinity parameter in mixed salinity environments, ranging between 20,000 to 150,000 ppm. This varied range of water salinity produces a wide uncertainty spectrum of core oil saturation in the range of +/- 3 to 10% saturation unit. Consequently, we developed two unique salinity variance models to capture the water salinity effect and minimize the uncertainty in the calculation of core saturation. The first model uses a material balance to solve for the salinity given the distilled water volume and gravimetric weight difference of the sample before and after leaching. The second model iteratively estimates the salinity required to achieve 100% of total fluids saturation at reservoir condition after correcting for the bleeding, stress and water evaporation effects. Our work shows that these derived models of water salinity are consistent with water salinity data from surface and bottomhole samples. Despite the prominence of applications of core saturation data in many aspects of the industry, thorough investigation into its quality and accuracy is usually overlooked. To the best of our knowledge, this is the first paper to present a novel analysis of the uncertainty coupled with Monte-Carlo simulation of lab-derived saturation's data from sponge cores. The modeling approach and results highlighted in this work provide the fundamental framework for modern uncertainty assessment of core data.