Upscaling Error and Connectivity

Part 8, Chapter 8: Volumetrics, Uncertainty, and Upscaling

Coarsening Has a Cost

Coarsening shrinks the cell count so the simulation runs in reasonable time, but it can erase the features that actually carry the flow. The classic case is a thin, connected high-permeability channel: it can conduct most of the flux even though it occupies little volume.

Upscaling error and connectivityfine field (channel)coarse (block-averaged)Once the channel is thinner than a coarse cell, averaging washes it out and the model loses flow.

Losing the Conduit

Once the channel is thinner than a coarse cell, block-averaging blends it into the surrounding shale. The coarse cell takes a middling permeability, the connected high-permeability path is broken, and the upscaled effective permeability falls below the true fine-scale value. The coarser the grid, the more flow the model quietly loses.

Resolve the Flow Paths

The rule is to coarsen only as far as the flow-controlling features can still be resolved, and to use flow-based upscaling, which honors the fine pressure field, where connectivity matters. Comparing the fine and upscaled flow response is the QC: if the upscaled model conducts noticeably less, the coarsening has gone too far.

This page is prerendered for SEO and accessibility. The interactive widgets above hydrate on JavaScript load.