Flow-Based Upscaling

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

No Simple Average Fits

Real rock is not neatly layered, so neither the arithmetic nor the harmonic mean is exactly right. Flow-based upscaling sidesteps the question: it solves single-phase flow across the fine block, with a pressure drop end to end, and reports the single permeability that reproduces the same total flux. That is the effective permeability by definition.

Flow-based upscalingfine permeability fieldharmonicgeometricflow-basedarithmeticeffective k (log)Flow-based upscaling gives the true k_eff, near the geometric mean and between the bounds.

Inside the Bounds

The flow-based value always lies between the harmonic (the lower bound) and the arithmetic (the upper bound), and for a randomly heterogeneous field it sits near the geometric mean. At low heterogeneity all the estimates agree, so a quick average is fine; as the permeability contrast grows they spread apart and only the flow solve stays trustworthy.

The Price

Flow-based upscaling is more accurate but more expensive: it solves a flow problem for every coarse cell, and the result is directional, a different value along each axis, a permeability tensor. The practical choice is to average where the rock is mild and solve where it is not, spending the compute where heterogeneity actually controls the flow.

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