| Abstract |
We seek to develop a robust geofluid flow modeling and data-interpretation process by which inter-well connectivity data acquired during routine flow monitoring can be quantitatively understood in terms of otherwise-unpredictable large-scale reservoir flow structure. To date flow models for both geothermal and hydrocarbon reservoirs have little predictive value, hence do little to reduce reservoir drilling and well-maintenance costs. The failure of reservoir modeling can in good part be attributed to ignoring the pervasive fracture-heterogeneity of crustal rock attested by well-log and well-core fluctuation systematics. These fluctuation systematics indicate that geofluids flow via spatially erratic fracture-percolation networks that cannot be predicted from traditional small-scale reservoir sampling. In geothermal reservoirs, spatially erratic in situ percolation flow networks can potentially be mapped using flow data observed at suitably large scale lengths. A possible means of flow-structure mapping is afforded by systematically recording and interpreting inter-well flow connectivity data. To investigate such flow-structure mapping of fracture-heterogeneous reservoirs, we compute percolation flow of geofluids in a numerical framework embodying well-log and well-core fluctuation systematics. In this heterogeneity framework, the spatially fluctuating numerical density ?(x,y,z) of grain-scale fractures attested by well-log and well-core data is modeled by power-lawscaling numerical fluctuations within the model volume. Model porosity (x,y,z) is proportional to fracture density, and model permeability ?(x,y,z) is derived from the well-core attested relation ? ? ?log(?). Single phase flow is simulated using the SUTRA finite-element solver for Darcy’s law in permeability-heterogeneous media, ?tP= • (?(x,y,z) P), P = geofluid pressure. We explore geofluid flow heterogeneity for three types of generic fracture heterogeneity that vary in their degrees of spatial fracture correlation parameterized by power-spectral exponent p, S(k) ~ 1/kp, k = spatial frequency. The generic heterogeneity types are: (i) p = 0, spatially uncorrelated or white noise fluctuations; (ii) p = 1, moderately spatially correlated ‘1/f’ noise fluctuations; (iii) p = 2, strongly spatially correlated Brownian noise fluctuations. Our simulations indicate that random spatial structures associated with ‘1/f’-noise and Brownian-noise degrees of in situ spatial fracture correlation can be deduced from studies of well-interconnectivity. |