Loading...
Thumbnail Image
Publication

Smart Physics-Inspired Compositional Dimensionless Type Curves for Unconventional Enhanced Oil Recovery

Syed, Fahad Iqbal
Citations
Altmetric:
Abstract
Dimensionless Type curves have been developed and used in the Oil and Gas industry for primary production performance evaluation. To Date, there is no physics-based dimensionless performance type curve developed for Enhanced Oil Recovery (EOR) of even conventional hydrocarbon-producing reservoirs. Predicting the production performance of unconventional and tight hydrocarbon reservoirs is challenging. Each unconventional well drilling and completion normally cost a company $ 6-12 Million. Unconventional EOR (UEOR) is the next step in unlocking untapped unconventional and tight hydrocarbon reservoirs' full potential and helps in minimizing environmental footprints by targeting the remaining hydrocarbon left behind and consequently avoiding unnecessary drilling and minimizing carbon emission. To conduct a successful UEOR project, oil and gas companies perform comprehensive simulation studies to screen and select candidate wells (pilot) for UEOR, predict their response to the UEOR methods and agents, and forecast the performance of wells’ ongoing UEOR. This requires running thousands of simulation cases that might take several months to complete comprehensive techno-economic assessment and evaluation. AI-empowered Dimensionless Type Curves that honor physical laws can offer fast-track screening and accurate solutions. In this dissertation, Smart Physics-Inspired Compositional Dimensionless Type Curves (SPiC TCD) for UEOR are presented that aim to address the above-mentioned problem and save millions of dollars by optimizing the UEOR practice and consequently reducing the carbon emission and environmental footprints and using subsurface resources in an environmentally beneficial way, which is the current portfolio of the oil and gas industry. SPiC TCD respond to operators’ W3H questions (Where to inject, When to inject, What to inject, and How to inject an EOR solvent) while performing comprehensive field screening and designing unconventional EOR pilot(s). W3H methodology provides fast-track AI-aided physics-inspired solutions based on historical wells' performance with existing subsurface reservoir and fluid descriptions and hydraulic fracture geometries and flow properties. This technique enables operators to make quick decisions on unconventional EOR pilot candidates’ selection and design to optimize design criteria such as the choice of injection solvent type and volume estimation, the optimum start of injection and soaking time as well as the frequency of this cyclic process and estimation and the soaking duration for the optimum oil recovery. To generate Smart Physics-Inspired Compositional Dimensionless Type Curves, a manageable number of numerical simulation cases are defined through the Physics-Guided Design of Experiment workflow. The workflow covers a wide range of operational design parameters pertaining to W3H criteria, reservoir rock and fluid properties, hydraulic fracture design, and their corresponding flow-related parameters such as fracture conductivity, fracture half-length, fracture height, fracture spacing, and the number of hydraulic fracture clusters per stage. Conventional design of experiment workflows fails in case of dealing with a system that operates based on known governing physical laws. This affects the accuracy of the proxy models and probabilistic modeling. Therefore, a detailed workflow is developed which is a physics quality control module to evaluate the response of the generated cases using the design of experiment techniques. It ensures the generated multidimensional distribution of the input parameters creates physically meaningful responses when solving the fluid flow equations. The next step is to train a family of machine-learning algorithms. Deep neural network algorithms are employed to build the proxy models for Smart UEOR dimensionless type curve generation. Upon completion of the training, the physics-based blind hindcasting and model response evaluation according to the physical laws are conducted. The generated physics-based AI proxy models are capable of generating thousands of cases based on the different reservoir and fluid descriptions as well as hydraulic fracture properties and W3H operational design criteria within an hour instead of months. It enables fast and accurate decision-making for optimal UEOR practice in unconventional and tight oil reservoirs.
Description
Date
2022-12-31
Journal Title
Journal ISSN
Volume Title
Publisher
University of Kansas
Collections
Research Projects
Organizational Units
Journal Issue
Keywords
Petroleum engineering, Artificial intelligence, Engineering, CO2 Injection, Compositional Type Curves, Data Mining, Huge Data, Hydraulic Fracturing, Unconventional EOR
Citation
DOI
Embedded videos