Porosity Determination From NMR Log Data: The Effects of Acquisition Parameters, Noise, and Inversion

Abstract

Abstract An ongoing challenge for nuclear magnetic resonance (NMR) well logging is that the quality and utility of the data depend on the acquisition sequence, inversion parameters, and the logging environment. Some modern NMR logging sequences are intended to be applicable over a wide range of environments, and include measurements of transverse relaxation (T2), polarization (T1), and diffusion (D). But compromises are inevitable for any given NMR technique. For example the overall acquisition time is dictated by operating at "reasonable" logging speeds, so full polarization of long T1 fluids is rarely achieved. Additionally, the inversion process creates interdependencies between T1, T2 and D. These effects conspire to introduce inaccuracies in the reported porosity. We investigate the influence of acquisition parameters, inversion parameters, and noise on the determination of porosity from NMR data. A key focus is on fluids exhibiting the extremes of T1, T2 or D such as light hydrocarbons, gas, water at high temperature, or heavy oil. We make this investigation by three methods. The first method is to calculate a porosity sensitivity curve as a function of T1, T2 and D for multiple acquisition modes. The second method is to invert simulated NMR echo data. Lastly, we compare real NMR log data porosities with other porosity logs in fields exhibiting long T1 constants. In all three cases, we show how porosity is under-or over-estimated beyond the expected accuracy based on random noise alone. The results are discussed in terms of what measures can be taken to identify and minimize these errors.