Documented results from
inflow control completions

Troll West is a high-permeability thin oil zone in the North Sea. Before 1992, horizontal wells drilled into these zones suffered from uneven inflow: high-permeability intervals produced fast, watered out early, and left large portions of the reservoir undrained. Conventional completions could not balance the inflow along the wellbore.

Norsk Hydro engineers, including Flowpro Dynamics founder Kristian Brekke, designed an inflow control liner using spiral flow paths to create controlled pressure drop at each completion interval. The device was filed as European patent EP 0 588 421 and installed on a Troll West well in 1992, the first commercial ICD deployment anywhere.

Published results in SPE 24762 compared the inflow-control liner against stinger completion and reduced perforation density on the same field. The ICD completion delivered a 50% higher plateau rate and 14% more cumulative oil than the conventional alternative. Those numbers established inflow control as an industry-standard tool and triggered the adoption curve that has run for thirty years.

Standard reservoir simulators cannot resolve annulus phase segregation. That gap matters for autonomous inflow control: AICDs and AICVs respond to whichever fluid phase arrives at each device, and phase distribution in the annulus depends on gravity, well inclination, flow rate, and completion geometry. Ignoring it produces production forecasts that diverge from field behaviour.

Flowpro Dynamics, Aker BP, and Quickersim co-authored SPE-225617-MS at the SPE Europe Energy Conference in Vienna, June 2025. The paper introduces a CFD-based upscaling method that resolves annulus phase segregation at the single-joint scale, then upscales results into a form reservoir simulators can evaluate directly.

The method was benchmarked against full Ansys Fluent CFD across a representative range of well inclinations, flow rates, and phase ratios. Insight reproduced the segregation and valve-interaction behaviour with fidelity, while running approximately 3,000,000 times faster than Ansys Fluent on the same problem. The result: completion studies that were previously infeasible, screened in hours instead of weeks.