Quantitative integration of flow simulation and time-lapse seismic in a continuous CO2 EOR project, Delhi field, Louisiana

Abstract

Interdisciplinary dynamic reservoir characterization is the key to reveal the fluid and rock behaviors of a productive petroleum reservoir under CO2 enhanced oil recovery (EOR) process. Integrating 4D seismic data into flow simulation is a relatively new approach to perform dynamic reservoir characterization and assist reservoir model updating. In this research, time-lapse seismic data are utilized qualitatively and quantitatively to guide the history matching process of CO2 flood in Delhi Field. The non-uniqueness of history matching is largely reduced by effectively incorporating 4D seismic responses along with production data. The objective of this integration is to investigate an efficient way to develop a predictable model with time-lapse seismic data. Delhi Field is a Cretaceous-age oil reservoir in northeastern Louisiana. Discovered in 1944, the original oil in place (OOIP) of Delhi Field is estimated to be 357 MMBBLS. After primary and secondary water flood, 54% of the OOIP had been produced. The current EOR process, initialized since 2009, is aimed to produce an additional 15% of the OOIP by continuous CO2 injection. Three seismic surveys were acquired over an area of the Delhi Field to monitor the behavior of CO2 in the subsurface. Many papers in the literature have discussed the application of 4D seismic assisted reservoir simulation in water flood or EOR reservoirs. However, few of them have touched on a stratigraphically complex reservoir with continuous CO2 injection strategy like Delhi Field. In this research, a time-lapse seismic assisted reservoir simulation workflow is conducted to improve the CO2 flood history match in Delhi Field. Three levels of the history match process can be identified from this work. The first level is a preliminary model with a poor production rate and pressure match. After recognized the problems, a careful history match was performed by qualitatively incorporating time-lapse seismic interpretations. As a result, the well production history match is greatly improved. The final stage is to quantitatively compare synthetic and observed seismic attributes and use the objective function to further optimize the simulation model. A Petro-Elastic Model (PEM) was constructed to connect simulation outputs with seismic attributes. In this process, dynamic simulation results (pressure, saturation and CO2 density) along with elastic moduli were assembled into synthetic seismic attributes. The quantitative comparison of seismic inversion attributes and synthetic seismic attributes is an effective tool to evaluate previous history match processes and to guide the subsequent model updating. The values of this research mainly lie in the following two aspects. First, a seismic -guided updated model can be delivered to the subsequent reservoir engineering work in Reservoir Characterization Project (RCP). Secondly, this work provides an effective integration procedure of interdisciplinary dynamic reservoir characterization process for studying Delhi Field. This procedure is applied to the baseline and 1st monitor seismic surveys in this work, future works could expand it to the 2nd monitor to better understand the dynamic behaviors of subsurface fluids and rocks as the CO2 injection continues.