Formation Oil Characterization in Oil-Based Mud Environments Using Blind Deconvolution in NMR Data

Abstract

The low contrast between formation oil and oil-based mud (OBM) filtrate as well as noise associated in the signal can cause the T2 distribution for the different fluids to resemble each other, making it difficult to identify the formation fluid using nuclear magnetic resonance (NMR) data. One challenges is to quantify the remaining oil volume and estimate the formation fluid proprieties using the diffusion map (T2D) from two-dimensional (2D) inversion. Consequently, it is important to determine new methodologies that can properly enhance the evaluation of such environments. This paper presents an application of blind deconvolution with a maximum likelihood algorithm processing applied to enhance the NMR diffusion map (T2D), helping to identify, quantify, and characterize the remaining oil volume in an invaded zone. The blind deconvolution algorithm is effective even when no information about the noise is known, making it possible to enhance the T2D map, deconvoluting the point-spread function (PSF) from the signal. After the enhancement, a multiple asymmetric Gaussian fit is used to generate a modeled distribution of the formation oil to estimate the remaining formation oil volume and its T2Intrisic logarithmic mean. The methodology using the blind deconvolution over the T2 diffusion map was tested. Promising results provided a formation oil distribution consistent with expected fluid properties measured.