A Fast-Track Method to Estimate Reservoir Porosities from Petrophysical and Post-stack Seismic Data

Hydrocarbon reservoirs depending on the discipline of focus, i.e. geology, geophysics, petrophysics or reservoir engineering, may be characterized from different perspectives. In the context of reservoir geology, characterization includes constructing a geological (static) model. Building such models is a timely process and requires access to interpreted data, i.e. petrophysical logs, processed seismic data, etc. In case of time limitation and/or inaccessibility to sufficient data, any workflow that is able to provide a first-pass insight into the properties of the reservoirs is of immense value. In this study we use data from the Lower Sarvak Formation of an Iranian oil field located within the Dezful Embayment of the Zagros Basin to propose a method that may be used to briefly characterize a reservoir, utilizing petrophysical and 3D post-stack seismic data. In the first step, several scenarios were devised with each including a specific type of logs and algorithm as part of Electrofacies analysis. As a result, for each scenario 25 facies (clusters) were defined in which each cluster covers a prescribed range of porosities. It is noted that the porosities within each cluster increase as the cluster number increases, accordingly. In the second step, for every scenario, Supervised Neural Network analysis was applied to the generated discrete facies logs and post-stack seismic data to produce a 3D seismic facies cube. Next, 6 continuous average apparent porosity maps were generated for the Lower Sarvak Formation. The generated maps were subsequently smoothed and normalized to a range between 0 and a maximum porosity of 25%. In the next step, the porosity maps of the 6 scenarios were compared with the porosity map derived from static model building (using acoustic impedance from post-stack seismic inversion, interpreted petrophysical logs and geostatistical methods). It was found that the scenario using Neutron (NPHI), density (RHOB) and sonic (DT) logs as input data and Multi Resolution Graphic-based Clustering as the clustering algorithm shows best correlation with the static model. This finding suggests that in the absence of sufficient time and/or data, particularly during the development of green fields, this method may be used as a fast-track approach to gain an overall understanding on the distribution of reservoir porosities within a hydrocarbon field.