![]()
Abstract The study area belongs to Qaidam Basin of Tibetan Plateau, which is the highest altitude petroliferous basin worldwide. It has extremely complex geological conditions and can be regarded as the "Mount Everest" in terms of the difficulty of performing effective reservoir characterization. We solved the problem to a great extent with 3D static modeling by integrating multi-disciplinary knowledge, data and workflows.
Firstly, building a 3D structural model as accurately as possible was taken as the core work for structurally โdouble complexโ study area because this was so important to determine success or failure of the whole study. So, in this step, the closely interactive combination of geophysical & geological modeling was adopted in an iterative way. To be specific, we first built our geophysical models of velocity, dip and anisotropy to produce the seismic image based on original seismic and logging data. Then we carried out seismic structural interpretation and generated the structural model for the first round to easily locate any mismatches between seismic image, seismic interpretation, image logging data (mainly dip azimuth and dip angle of strata), correlated well markers and reasonable structural patterns. If any mismatch existed, every discipline needed to double-check the potential problems, which could mean another round of seismic processing-seismic interpretation-structural modeling work, to make sure all these data were organically matched with each other finally to generate more robust structural model with more than 40 zones vertically. Secondly, stochastic simulations such as Sequential Indicator Simulation and Gaussian Random Function Simulation were chosen to build property models such as lithology model, mineral composition models, and quality models of porosity, permeability and water saturation. The most important data basis for these models were element logging interpretation results. Every property model was given a set of appropriate modeling parameters and constraints for its own. Thirdly, discrete fracture network (DFN) was built based on well image logging interpretation results, neural network analysis and optimized seismic/geometrical fracture drivers, and trend modeling, and upscaled fracture properties necessary for Dual poro/Dual perm dynamic simulation were obtained.
As results, structural accuracy was improved dramatically, and many geomodels with tens of millions of cells were generated to effectively characterize the saline lacustrine thin tight oil reservoirs of the thick study zone in 3D manner, for both matrix and fracture parts of the reservoirs. More importantly, based on these models and hundreds of resultant maps, many useful conclusions were reached geologically and timely support to client E&P activities in many aspects was provided, such as OOIP/reserve submission to the country, well placement design and deployment and field development plan optimization.