Frank Male

                RESUME

Frank Male

Postdoctoral Fellow

Email Address: frmale@utexas.edu

LinkedIn URL: https://www.linkedin.com/in/frank-male-a2504297/

 

RESEARCH HIGHLIGHT

My research focuses on data analysis for reservoir engineering problems. Problems I am trying to solve include: production and productivity analysis for unconventional oil and gas wells, better understanding of the factors influencing permeability, and optimizing water flood and CO2 Enhanced Oil Recovery projects.

Current approaches to these problems often either fail to take full advantage of the data or the appropriate physical models. With hybrid machine learning approaches that combine physical intuition and data, we can more more efficiently extract value from hydrocarbon resources.

Part of the reason behind current approaches not taking full advantage of the data is the lack of integration between geoscientists and petroleum engineers. I work on integrating geological concepts into machine learning, big data analytics, and statistical analyses of petroleum systems.

 RESEARCH PROJECTS

Project A: Tight oil production analysis

            For my PhD I developed an award-winning physics-based scaling approach to shale gas decline analysis, as part of an integrated study of major US gas-producing plays. Since then, I have successfully applied it to tight oil production in the Bakken, Eagle Ford, Permian, and Austin Chalk.

            Analysis of these results has led to increased understanding in the geological, petrophysical, and completions drivers of production from these plays. For instance, aggressive completions  improve the cash flow for wells drilled in mature basins, but do not necessarily increase the recovery. Being able to better balance cash flow and capital costs is a goal of this research.

 

Project B: Integrated reservoir characterization

            In spite of 90+ years since Kozeny’s paper, permeability in reservoir formations remains difficult to estimate. I work closely with geologists, using interpretable machine learning approaches trained on thin section analysis to improve permeability predictors and better understand what processes impact permeability.

            The results point to geological processes that can inform the distribution of permeability in reservoir models following sequence stratigraphic concepts.

 

Project C: Improved Oil Recovery analysis

            Capacitance Resistance models allow for fast, data-driven analysis of the connectivity between producers and injectors. I have developed an open-source Python library, pyCRM, for calculating well-to-well connectivity in water floods.

            With CRM, operators can more effectively balance injection volumes, identify candidates for conformance jobs or recompletion, and improve reservoir models for more detailed study. Also, CRM provides a process for identifying geologic features (such as sealed faults) that can later be included in 3D geologic models.