Abstract
Natural fractures have been widely found and thought to be an important factor in creating a complex hydraulic fractures network and improving hydrocarbon production in shale reservoirs. Nuclear magnetic resonance (NMR) is extensively applied to measure in-situ petrophysical properties, and magnetic nanoparticles provide good contrast agents to distinguish hydrogen relaxation time for NMR. The objective of this paper was to estimate spatial distribution of natural fractures in shale gas reservoirs by applying magnetic nanoparticles to change the NMR relaxation times. Firstly, a three-dimensional (3D) reservoir model including multiple natural fractures was built to investigate the flow of nanoparticles and the distribution of volume magnetic susceptibility (VMS) when injecting magnetic nanoparticles. The sensitivity analyses about nanoparticle concentration and nanoparticles size were investigated. Secondly, a forward model was introduced about how to obtain Carr-Purcell-Meiboom-Gill (CPMG) echo decay trains from given relaxation times. Thirdly, an inversion method was presented to convert the typical CPMG echo decay curve into the distribution of T-2 relaxation amplitudes and times. The inversion method was used to show the change of T-2 amplitude and time for the two synthetic cases with and without magnetic nanoparticles.
The dynamic distribution of nanoparticles concentration and VMS are graphically displayed along each time step in 3D mesh. The results show that magnetic nanoparticles bring much larger VMS while most nanoparticles only flow into these natural fractures (NF) directly connected with wellbore. The others' NF have somewhat higher VMS than matrix. Additionally, based on various sensitivity cases, a higher concentration of nanoparticles yields a stronger magnetic field, and larger nanoparticle size could lead to higher VMS, although the nanoparticles face stronger flow resistance and less diffusion movement. The CPMG decay curve is a multi-dimensional exponential function related to relaxation amplitudes and times, and the least squares minimization technique can be applied to obtain the T-2 amplitudes from CPMG curve. The comparison between two synthetic cases shows the amplitude for small T-2 time increases and the amplitude for large T-2 time reduces, which provides a clear indicator to detect the locations of natural fractures because their T-2 largely reduce. The nanoparticle model provides valuable guidance about choosing parameters for optimizing magnetic nanoparticle injection design to enhance VMS and NMR signals. The inversion model introduces an efficient path for estimating T-2 distribution and petrophysical properties from the acquired NMR CPMG signals. Additionally, magnetic Nanoparticles provide excellent relaxation contrasts to distinguish the magnetization signals of formation for estimating the spatial natural fractures distribution. Consequently, in-situ fracture characterization and the development of hydraulic fracture treatments could be beneficially improved.