You're reading an old version of this documentation. If you want up-to-date information, please have a look at v0.8.2.

Grid Convergence Study (Linux)

Summary

This is a grid convergence study of 5 cases. The case with the finest grid resolution, of 0.0625m, achieved an asymptotic ratio of 1.003 (asymptotic range is indicated by a value \(\approx 1\)). At zero grid resolution, the normalised velocity deficit measured 1.2 diameters downstream from the turbine was 43.04%, a 13.78% error against the measured value of 49.92%. For the centreline velocity (3% TI) transect, the root mean square error at the lowest grid resolution was 0.2004. For the centreline velocity (15% TI) transect, the root mean square error at the lowest grid resolution was 0.09075. For the axial velocity at \(x^*=5\) (3% TI) transect, the root mean square error at the lowest grid resolution was 0.1232. For the axial velocity at \(x^*=5\) (15% TI) transect, the root mean square error at the lowest grid resolution was 0.05688.

Grid Convergence Studies

Free Stream Velocity

This section presents the convergence study for the free stream velocity (\(U_\infty\)). For the final case, with grid resolution of 0.0625m, an asymptotic ratio of 1.63 was achieved (asymptotic range is indicated by a value \(\approx 1\)). The free stream velocity at zero grid resolution is 0.8046m/s. The grid resolution required for a fine-grid GCI of 1.0% is 0.03862m.

Free stream velocity (\(U_\infty\)) per grid resolution with computational cells and error against value at zero grid resolution

resolution (m)	# cells	\(U_\infty\)	error
1	144	0.756681	0.0595775
0.5	1152	0.790921	0.0170232
0.25	9216	0.793301	0.0140648
0.125	73728	0.794949	0.0120173
0.0625	589824	0.797055	0.00939936

Free stream velocity error against value at zero grid resolution per grid resolution

Wake Velocity

This section presents the convergence study for the wake centerline velocity measured 1.2 diameters downstream from the turbine (\(U_{1.2D}\)). For the final case, with grid resolution of 0.0625m, an asymptotic ratio of 1.003 was achieved (asymptotic range is indicated by a value \(\approx 1\)). The free stream velocity at zero grid resolution is 0.4583m/s. The grid resolution required for a fine-grid GCI of 1.0% is 0.1524m.

Wake centerline velocity 1.2 diameters downstream (\(U_{1.2D}\)) per grid resolution with computational cells and error against value at zero grid resolution

resolution (m)	# cells	\(U_{1.2D}\)	error
1	144	0.736009	0.605977
0.5	1152	0.644324	0.405919
0.25	9216	0.515944	0.125793
0.125	73728	0.45951	0.00265517
0.0625	589824	0.458319	5.60439e-05

Wake velocity error against value at zero grid resolution per grid resolution

Validation

At zero grid resolution, the normalised deficit of \(U_{1.2D}\), (\(\gamma_{0(1.2D)}\)) is 43.04%, a 13.78% error against the measured value of 49.92%.

Wake Transects

This section presents axial velocity transects along the turbine centreline and at cross-sections along the \(y\)-axis. Errors are reported relative to the experimental data given in (Mycek et al. 2014).

Centreline velocity (3% TI)

The root mean square error (RMSE) for this transect at the finest grid resolution of 0.0625m was 0.2004.

Root mean square error (RMSE) for the normalised velocity, \(u^*_0\), per grid resolution.

resolution (m)	RMSE
1	0.401335
0.5	0.271788
0.25	0.239064
0.125	0.190003
0.0625	0.200371

Normalised velocity, \(u^*_0\), (m/s) per grid resolution comparison. Experimental data reverse engineered from (Mycek et al. 2014, fig. 11a).

Normalised velocity deficit, \(\gamma_0\), (%) per grid resolution comparison. Experimental data reverse engineered from (Mycek et al. 2014, fig. 11a).

Centreline velocity (15% TI)

The root mean square error (RMSE) for this transect at the finest grid resolution of 0.0625m was 0.09075.

Root mean square error (RMSE) for the normalised velocity, \(u^*_0\), per grid resolution.

resolution (m)	RMSE
1	0.203884
0.5	0.136543
0.25	0.0818053
0.125	0.101083
0.0625	0.0907545

Normalised velocity, \(u^*_0\), (m/s) per grid resolution comparison. Experimental data reverse engineered from (Mycek et al. 2014, fig. 11b).

Normalised velocity deficit, \(\gamma_0\), (%) per grid resolution comparison. Experimental data reverse engineered from (Mycek et al. 2014, fig. 11b).

Axial velocity at \(x^*=5\) (3% TI)

The root mean square error (RMSE) for this transect at the finest grid resolution of 0.0625m was 0.1232.

Root mean square error (RMSE) for the normalised velocity, \(u^*_0\), per grid resolution.

resolution (m)	RMSE
1	0.201806
0.5	0.134912
0.25	0.135605
0.125	0.118328
0.0625	0.123211

Normalised velocity, \(u^*_0\), (m/s) per grid resolution comparison. Experimental data reverse engineered from (Mycek et al. 2014, fig. A12a).

Normalised velocity deficit, \(\gamma_0\), (%) per grid resolution comparison. Experimental data reverse engineered from (Mycek et al. 2014, fig. A12a).

Axial velocity at \(x^*=5\) (15% TI)

The root mean square error (RMSE) for this transect at the finest grid resolution of 0.0625m was 0.05688.

Root mean square error (RMSE) for the normalised velocity, \(u^*_0\), per grid resolution.

resolution (m)	RMSE
1	0.0474432
0.5	0.0520709
0.25	0.0470542
0.125	0.0609816
0.0625	0.056879

Normalised velocity, \(u^*_0\), (m/s) per grid resolution comparison. Experimental data reverse engineered from (Mycek et al. 2014, fig. A12a).

Normalised velocity deficit, \(\gamma_0\), (%) per grid resolution comparison. Experimental data reverse engineered from (Mycek et al. 2014, fig. A12a).

References

Mycek, Paul, Benoît Gaurier, Grégory Germain, Grégory Pinon, and Elie Rivoalen. 2014. “Experimental Study of the Turbulence Intensity Effects on Marine Current Turbines Behaviour. Part I: One Single Turbine.” Renewable Energy 66: 729–46. https://doi.org/10.1016/j.renene.2013.12.036.