Home » Project 4
Project 4
Project Outcomes
Hypotheses were built around known seismoelectric theory and subsurface petroleum systems. Boreholes were selected in key areas of Southern Africa, including Danhausse, Karoo, Majuba, Welkom, and Orkney, to test for live petroleum systems, conventional natural gas, coal bed methane, and false positives. Core data from the South African Department of Geosciences and Botswana National Survey was collected, digitized, and integrated into a specialized database to support algorithm testing. Velocity models were developed through field tests, correlating the borehole data with selected sites. The goal was to build models that accurately represent the subsurface velocity structures for better seismic interpretation. A database and software tools were developed to compare airborne tellurics data (measuring electromagnetic variations in the atmosphere) with ground tellurics collected using two different techniques. These tools enabled cross-validation of geophysical data and improved the detection of subsurface features.
Custom sensors were designed to measure fluid pressure gradients, dielectric permittivity, zeta potentials, and fluid conductivity of natural gas, liquid hydrocarbons, and water contact points. These sensors were tested for accuracy and reliability in borehole environments. Field tests were conducted using Chi-Dong bespoke equipment and PL14 in borehole samples. Comparative analysis helped refine calibration techniques. Beamforming techniques were developed to accentuate the seismoelectric signal, isolating electrical fields related to fluid interactions from noise caused by seismic activity. A new data-cleaning algorithm was created to reduce noise by localizing the seismic source and controlling heterogeneity in seismic velocity and electrical resistivity, ensuring more accurate data collection.
Data from the PL14 and Chi Dong sensors were synthesized and compared with airborne tellurics datasets over known borehole samples. This process aimed to validate the sensor outputs and check consistency with known subsurface features. Cross-hole seismoelectric data was analyzed using borehole data to improve regional calibration, particularly in complex formations like the Witwatersrand igneous and Ecca, Dwyka, and Beaufort sediment formations. A Markov chain Monte Carlo (MCMC) solver was developed and implemented based on real-world borehole data points collected during field trials. This helped improve the probabilistic models for predicting subsurface fluid behaviors. One issue faced was the computational intensity of the MCMC method, which slowed down processing time and required optimization. Deterministic inverse modeling, particularly 5D electric forward modeling, was used to process the collected data and to test the original hypotheses. The goal was to derive compact volumetric current source distributions, representing subsurface fluid movements. While this approach improved accuracy, a challenge arose in ensuring convergence in highly heterogeneous geological formations.