Wake model comparisons

wcomp runs the same windIO wind-farm cases through FLORIS, FOXES, and PyWake so the same wake model can be compared directly across independent implementations. Every chart on this site reads from a versioned dataset generated with wcomp's solver interfaces.

Generated 10/31/2026 against FLORIS 4.6.6, FOXES v1.9.3, and PyWake2.6.21.dev30+g7b7ffbca8.

Framework details

• Rotor averaging is the largest source of framework differences. On the quadratic radial deficit field the rotor-average error ranges from 0.4% (FOXES) to 6.2% (PyWake).
• Power and thrust tables are reproduced essentially exactly (power within 9e-12 kW, Ct within 1e-15 in range). The one visible difference is FLORIS's Ct, which is clipped by 1e-4 outside the cut-in/cut-out range.

Case comparisons

• The largest streamwise disagreement is TurbOPark with four turbines, FLORIS vs PyWake: 1.56 m/s RMSE (16.7% NRMSE).
• Differences grow with turbine count. Bastankhah 2016 deflection goes from 0.66 m/s RMSE (max over software pairs) with one turbine to 1.32 m/s with four.

Implementation matrix

Which software implements which wake model. Models with a linked name have a full comparison page (charts + quantitative metrics) generated from the dataset.