A SEISMIC FRAMEWORK FOR ESTIMATING HYDROCARBON RESERVOIR VOLUMES AND THEIR LIKELIHOOD
MetadataShow full item record
This thesis develops a framework to estimate the likelihood in fluid volumes in a hydrocarbon reservoir. It uses 3C-3D seismic data and well logs from the Blackfoot oilfield (Alberta-Canada). Results from cross validation techniques applied to distribution maps generated using the geostatistic method (thickness and percentage of sand) and the neural network method (porosity) are used to estimate the uncertainty related with the predicted distributions in the case of the Blackfoot oilfield (AB-Canada). These distribution maps as well as the estimated uncertainty associated with them are used as inputs in two different approaches for the application of uncertainty analysis in the estimation of hydrocarbon volumes (a Taylor expansion approach and a Monte Carlo approach). The results obtained using these two approaches give compatible hydrocarbon volume estimates for the Blackfoot pool, with P10~ 12 MMbbl, P50~ 8 MMbbl, and P90~ 5 MMbbl. Investigation about sources of uncertainty in seismic data revealed that the time picking error could explain, in the case of the Blackfoot reservoir, the uncertainty in the thickness parameter. In the second part of this project, well log data from the Gulf of Mexico are used together with fluid substitution method and uncertainty analysis to evaluate how the observed variability in rock properties of the Gulf of Mexico for each specific depth value affects the response of the attributes that respond to fluid discrimination. A larger concern for deeper reservoirs was identified in the predicted results. Nevertheless, the fluid substitution results were considered robust in most conditions investigated in this project, allowing discrimination of gas, fizz gas, and water saturated reservoirs in some of the attributes that respond to the fluid content. This last result could allow the estimate of the missing parameter in the HCPV estimation: the hydrocarbon saturation distribution map.