CCS and Power Systems
Carbon Storage - Monitoring, Verification, Accounting, and Assessment
In Situ MVA of CO2 Sequestration Using Smart Field Technology
Performer: West Virginia University
Project No: FE0001163
To date, researchers have selected two sites (one in Mattoon, Illinois and a second in Citronelle, Alabama) to be used as models in the development of the software.
The following provides summaries of the accomplishments to date for each site:
Mattoon, IL Site
Surrogate Reservoir Models (SRM) were constructed and datasets were generated. The SRMs were built with the purpose of pressure and water saturation prediction with a reasonable accuracy as well as to predict the CO2 mole fraction in the reservoir.
The dataset was divided into two sets due to the pressure and rate turbulence at the beginning. The first data set was generated in the time interval of the first to the seventh month while the time scale for the second dataset was the first to the seventh year. The first to seventh month dataset results showed a maximum error of ~6.5 percent for the water saturation and pressure. The first to seventh year dataset results showed a maximum error of ~3.4 percent for the water saturation and ~3.8 percent error for the pressure case.
Citronelle, AL Site
Heterogeneous porosity and permeability maps were generated for the model using values interpreted from resistivity and induction logs. Two different porosity values were interpreted based on true resistivity and induction logs for 51 simulation layers in 48 wells. Nine different models were built based on multiple geological realizations obtained from the two different porosity distribution and four porosity-permeability correlations.
Simulations were performed in order to generate data streams from the flow model. This was done by modeling an array of slim holes in the reservoir where Permanent Downhole Pressure Gauges (PDPG) might be installed. Then simulated carbon dioxide leakage data recoded during the simulation runs were collected. A method was developed for distorting the data from the pressure readings to emulate background noise inherent to field measurements using techniques such as including random pressure spikes and adding random white noise. Data cleansing routines were incorporated as a data pre-processing step for the high resolution data. These routines had the ability to remove outliers and reduce white noise using mathematical techniques such as moving average.
A study on leakage modeling and simulating in the reservoir was performed. A Leakage Detection System (LDS) was developed using pressure data received in high frequency streams from simulated Permanent Down-hole Gauges (PDG). A set of simulation runs were completed that provided simulated pressure behavior in the observations wells with respect to leakage rates and locations.