An accurate evaluation of injectivity is essential to the economics of any chemical EOR process. Most commercial simulators enable non-Newtonian behaviour modelling but it is often overlooked due to inadequate grid resolution. Indeed, in cases where shearthinning fluids are injected in a reservoir, shear rates and viscosities in the vicinity of the wellbore can be poorly estimated if the spatial resolution of the well grid-blocks is too coarse. This results in biases in injectivity and economics which we discuss here in the context of foam-based displacements.
We consider continuous foam injection in models of different spatial resolutions ranging from 1 to 100 m gridblock sizes and study the behaviour of injection wells obtained on the coarser grids compared with the results from a high resolution grid. This reference grid is sufficiently refined to account for near-wellbore large velocity gradients and render injectivity accurately. In this work we propose new formulations of the well index that capture shear-thinning behaviour that the conventional Peaceman calculation fails to address.
We first demonstrate that a poor evaluation of near-wellbore velocity leads to erroneously degraded injectivity on the coarser grids when compared to the reference grid. In order to correct these errors our modified well index is applied and validated in various scenarios of foam displacement simulation with radial grids. It captures a more accurate injectivity than the conventional Peaceman calculation once steady-state regime is reached. The modified well index we propose, used under a simplified form as direct input in reservoir simulation, significantly enhances injectivity estimates without resorting to grid refinements or modifying the shear-thinning model of the injected foam. In most cases it yields results that are closer to those obtained using grid refinements than the Peaceman formula at a much more attractive computational cost. Additional work remains to complete our understanding of injectivity in more complex settings, especially in the context of foam injection when effects such as foam dry-out and destruction in the presence of oil are as important on sweep efficiency as its shear-thinning behaviour.
Our workflow successfully corrects biases in the estimation of injectivity and yields more accurate results and avoids resorting to time-consuming methods such as grid refinements and physical input data alteration. Moreover it is simple to implement in most commercial simulators and does not require using empirical criteria. However, it bears some limitations which we also discuss.