Rock physics changes due to CO2 injection : the CO2CRC Otway Project

Abstract

The CO2CRC Otway Project aims to demonstrate that CO2 can be safely stored in a depleted gas field and that an appropriate monitoring strategy can be deployed to verify its containment. The project commenced in 2005, with the baseline 3D seismic collected early in January 2008. CO2 was injected into depleted gas reservoir known as Waarre-C at Naylor field in April 2008. The first monitor survey was recorded in January 2009, shortly after the injection of 35,000 tonnes of CO2. Early predictions in the program suggested that the resulting time-lapse seismic effect will be very subtle because of the reservoir depth, small area, complexity, small amount of CO2/CH4 in 80/20 ratio injected and most of all partial saturation of the reservoir sand. The key challenge than presented to this research was how subtle exactly is the effect going to be? To answer that question I had to develop a workflow that will produce very accurate prediction of the elastic property changes in the reservoir caused by CO2 injection. Then the sensitivity of time-lapse seismic methodology in detecting subtle changes in the reservoir is investigated.The rock physics model I propose uses the “effective” grain bulk modulus (Kgrain) to represent the average mineralogy of the grains. The validity of this approach is confirmed by good agreement achieved between Vpsat core with Vpsat computed from the log data using the “effective” modulus. . The use of “effective” Kgrain was further justified by petrographic analysis. This has increased the modelling precision and changed the predicted time-lapse effect due to CO2 injection from 3% as an average over the reservoir sequence as previously computed to nearly 6%. The significance is that 6% change could be detected with high precision monitoring methodologies. The in-situ saturation type is homogeneous, according to the analysis path assumed in this thesis. If some patchiness exists in the reservoir it will be away from the wells and it would further elevate CO2 related seismic effect.The time-lapse seismic methodology at Otway site utilised very high survey density in order to increase sensitivity. On the negative side, weak sources and the change of the source type between the surveys resulted in non-repeatability greater or of the similar order as the time-lapse signal were expected to be. Hence the interpretation of the time-lapse P-wave seismic data assumed somewhat different path. I used the model-based post-stack seismic acoustic inversion in a similar way that history matching is used in reservoir simulation studies. I performed successive fluid substitutions, followed by the well ties and inversions. The objective was to look into the inversion error. Then the modelled fluid saturation case that result in minimal inversion error reflects the most likely state of the reservoir. Modelling using 35,000 tonnes of CO2/CH4 mix with 35% water saturation and 65% CO2/CH4 mix produced the smallest error when reinstating logs to the 2009 reservoir state.The time-lapse anomaly observed in the data exceeds predictions derived through the rock physics model, seismic modelling and simulation models. This is likely to be the case in general as the effect of CO2 onto a reservoir is difficult to predict. A “conservative” approach may result in an under-prediction of time-lapse seismic effects. Consequently, the predicted and measured seismic effects can be used as the lower and the upper bound of the time-lapse effects at Naylor field, respectively. The method presented here for analysis of a subtle time-lapse signal could be applied to the cases with similar challenges elsewhere.