Seismic attribute analysis and CO2 monitoring within the Mississippian reservoir, Wellington field, Sumner County, Kansas

The Mississippian reservoirs are prolific hydrocarbon producers in the mid-continent of the United States but are challenging to image seismically due to their high heterogeneity and thicknesses typically below seismic resolution. In 2016, the Kansas Geological Survey (KGS) conducted a pilot study in the Mississippian chert reservoir at Wellington field, Sumner County, Kansas to determine the feasibility of injecting CO2 for enhanced oil recovery (EOR) and geologic storage of CO2 in this geological environment. This study evaluates the use of seismic methods for imaging the injected CO2 in the Mississippian reservoir. Time-lapse seismic comparison of an arbitrary line extracted from the pre-injection 3-D seismic survey and a coincident post-injection 2-D line proved ineffective due to differences between a 3-D volume and 2-D profile imaging. Therefore, imaging the CO2 centered on analysis of the post-injection 2-D seismic line characteristics both near to and far from the injection well, KGS #2-32, where CO2 saturation varied from 50% to 0% respectively. Fluid substitution modeling was used to evaluate Mississippian reservoir acoustic property changes due to CO2 saturation changes. CO2 saturation levels of 10%, 25%, and 50% displayed a decrease (larger absolute value) in normal-incidence acoustic impedance of up to 2.4%, 6.1%, and 13.3% with increasing CO2 saturation, respectively. Amplitude Variation with Offset (AVO) analysis of the post-injection 2-D seismic line evaluated the amplitude response at seismic gathers near to and far from the injection well. The majority of the CDP’s (including CDP 203230 – location of KGS #2-32) exhibit significant scatter but an overall decrease in amplitude magnitude with offset is present indicating a Class I AVO response of the Mississippian reservoir. Noise identified in the gathers was suppressed using f-k filtering; however, intercept-gradient crossplots and gradient curves throughout the seismic line were too scattered to reliably identify changes in CO2 saturation. Three factors contributed to the negative result: 1) Class I AVO increasing impedance, high matrix incompressibility of carbonate reservoirs make fluid detection challenging, 2) a relatively small amount of CO2 (20,000 tons) was injected in the reservoir and 3) poor near surface conditions during data acquisition resulted in noisy seismic data. Although these challenges made it difficult to image the injected CO2 in the field, fluid substitution modeling suggests a decrease in normal-incidence acoustic impedance with increasing CO2 saturation, which may yield detectable changes in seismic data under favorable field conditions