Abstract
As surface-coated superparamagnetic nanoparticles are capable of flowing through micron-size pores across long distances in a reservoir, with modest retention in rock, they have novel use potential in subsurface applications. These particles change the magnetic permeability of the flooded region, and thus can be used to enhance images of the subsurface and characterize hydrocarbon reservoirs. We earlier demonstrated the feasibility of using magnetic nanoparticles to track flood-front in waterflood and EOR processes in a homogeneous reservoir. In this paper, we model the propagation of a “ferrofluid” slug in a heterogeneous reservoir and its response to a crosswell magnetic tomography system. Specifically, we highlight the magnetic response at a low frequency (10 Hz) to the magnetic excitations generated by a vertical magnetic dipole source positioned at the injection well. The “ferrofluid” alters only the magnetic permeability of the domain occupied by the fluid and is thus distinct from methods that rely on contrasts in electrical conductivity. The flow behavior of the magnetic nanoparticles is coupled with time-lapse magnetic measurements through applying appropriate mixing laws and effective medium theory. Fluid flow is computed with a reservoir simulator; the electromagnetic response is computed with an electromagnetic (EM) simulator developed at Duke University for the overburden/reservoir/underburden system.
The approach to monitoring fluid movement within a reservoir is built on established electromagnetic conductivity monitoring technology. Here we investigate the detectability of a contrast in magnetic permeabilities between injected and resident fluids. At the low frequency studied here, the induction effect is small, the casing effect is manageable, the crosswell response originates purely from the magnetic contrast in the formation, and changes in fluid conductivities are irrelevant. This approach thus offers a new and independent mechanism for tracking flood fronts.
Numerical simulations indicate that the influence of areal and vertical reservoir permeability heterogeneity on flood fronts can be detected. For areal permeability heterogeneity, we use a five-spot reservoir model (with injector in the center) and incorporate high- and low-permeability ellipsoidal features with two orientations. The most detectable heterogeneity is a low permeability feature perpendicular to the streamlines. For vertical heterogeneity, we devise a two-layer reservoir model with single-well radial injection with a variable thickness for the high-permeability layer and study the evolution of time-lapse magnetic tomography maps. The tomography maps are shown to be capable of detecting the vertical heterogeneity in different stages of the flood. This is particularly helpful for identifying thief zones. In all the cases, the magnetic response is sensitive to the pattern and distribution of streamlines; therefore, permeability heterogeneity could be deduced from time-lapse magnetic measurements.
By adding magnetic nanoparticles into the injection fluids for waterflood and EOR processes and utilizing the established EM crosswell tomography technique, we show the feasibility of inferring the major features of reservoir heterogeneity, as well as of tracking the injectant bank front, from the time-lapse magnetic responses. This can substantially improve the management and optimization of such floods.
