Welcome to the new Energy Central — same great community, now with a smoother experience. To login, use your Energy Central email and reset your password.

Exemplar Modelling for Geothermal Reservoir Characterization. A case of Ensemble Machine Learning.

Geothermal development projects rely heavily on very disparate types of reservoir data that elaborate on subsurface geological environments. This data influences the understanding of the economic potential of the project. Geothermal reservoirs are unique in the wide range of rock types in which they are found. The geothermal exploration process captures this varying data through standard subsurface and petrologic analysis, surface geology and geophysical surveys (gravity, magnetic, and MT), and heat flow data to obtain a similarly varied analysis. The analysis expands upon the parameters such as rock permeability, porosity, resistivity, conductivity, seismicity, fluid volume, and heat flow within the reservoir. Efficient visualization of reservoir characteristics provides an accurate analysis of the geothermal subsurface on which operators can make informed decisions regarding project development while also minimizing their investment risk. The ensemble learning paradigm is a new Computational Intelligence (CI) algorithm in which diverse expert opinions are strategically incorporated and smartly integrated to solve a problem. This new learning paradigm has its root in human sociology where final decisions are made by considering various opinions of a "committee of experts" to obtain an overall "ensemble" decision. It has its origin in human psychology which suggested that the judgment of a committee is superior to those of individuals given that "the individuals have an acceptable level of competence". The ensemble learning methodology is a close emulation of the human socio-cultural behavior of seeking several people's opinions before making any important decision. This study demonstrates the application of ensemble machine learning in the characterization of the properties of a geothermal reservoir, based on previous successful applications in various industries including oil and gas. This is achieved by training the machine to select an overall best hypothesis to improve model performance, reducing the risk of selecting a poor model in implementing Machine Learning in geothermal reservoir characterization.