Operational efficiency in formation evaluation has become critical considering current trends in the oilfield industry. The impact is especially felt by advanced measurement logging services, including fluid typing by nuclear magnetic resonance (NMR), for which logging speeds are limited by physics and large suites of multidimensional data. However, improving efficiency cannot come at the cost of the accuracy of the answers. Recent work shows that optimization of a single frequency of operation (that is, a single depth of investigation [DOI]) provides only modest efficiency gains based on careful acquisition mode design with tailored sensitivity to known formation properties and expected reservoir fluids. Significant improvements in speed require a multifrequency approach in which data across multiple DOIs are used simultaneously.
One focus of this work is to address the challenge of maintaining the necessary accuracy of NMR fluid-typing answers while still benefiting from a full suite of NMR diffusion and relaxation data. We expand on previously described optimization techniques of diffusion-relaxation acquisition modes wherein subsets of the NMR sequence are distributed across multiple frequencies or DOIs. We call this method split shell acquisition (SSA). The subsequent method of processing all data together into a single 3D inversion is called combined shell processing (CSP). The petrophysical applicability is for cases in which an invasion profile from independent multiple frequencies is not required.
The numerous ways to approach NMR sequence optimization include analytical sensitivity calculations, quick-look inversion computation for porosity accuracy, signal-to-noise ratio (SNR) expressions, and Monte Carlo methods. We used all four methods in combination with the SSA strategy of assigning different parts of the NMR sequence to different frequencies of operation. We defined our optimization criteria as obtaining the fastest possible logging speed without compromising answer quality for the expected fluids. Two key differentiators are that the method exploits the different NMR properties of each frequency of operation toward optimal sensitivity to fluids and that the inversion fitting function simultaneously fits data from all frequencies.
Modeling and acquired tool data were used to build and verify the operating modes. We demonstrate an achieved increase in logging speed of almost a factor of 3 compared with single-frequency logging speed. We also show how the efficiency gains can be used alternatively to provide enhanced polarization and sensitivity to long relaxation time constants for both carbonate and sandstone reservoirs.
