**5. Secondary effects**

The computer code CARDAAV can evaluate several rotor shapes with a straight or curved blade and use a defined shape such as a parabola, catenary, or modified troposkien. The code also can include the so-called "secondary" Effects," such as the

*Aerodynamic Analysis and Performance Prediction of VAWT and HAWT Using CARDAAV… DOI: http://dx.doi.org/10.5772/intechopen.96343*

rotating tower, strut, and the spoiler. Another involved and unsteady phenomenon due to airfoil undergoing large and rapid variations of the angle of attack with time is the dynamic stall. During the rotor's rotation, the drag and lift coefficients present a different hysteresis than the case of static behavior. A dynamic delay of the stall to angles is substantially beyond the static stall angle, including massive recirculating regions moving downstream over the airfoil surface. In the case of VAWT, when the operational speed approaches its maximum, all the rotor blade sections go beyond the critical static stall angle of attack, the angle of attack changes substantially, and the entire blade operates under dynamic stall conditions, which will increase the unsteady blade loads and the wind machine structural fatigue. Different dynamic stall models with some modifications derived from static and dynamic airfoil tests have been incorporated into the CARDAAV computer code, such as the Gormont model and the indicial model [11, 34, 35]. Comparisons between aerodynamic performance predictions using dynamic-stall models show that the models provide a better prediction of the dynamic-stall regime characterized by a plateau oscillating near the experimental data of the rotor power function of wind speed [11].

### **6. Result and discussion**

This section presents a selection of results obtained by performing aerodynamic loads and performance prediction using CARDAAV, different variants, and QBlade computer codes, including dynamic stall. Results were achieved on Sandia 17-m and 34 wind turbine machine [36] and compared to available experimental data. CARDAAV is the original code using an ambient constant ambient atmospheric wind speed, while CARDAAS code incorporated a stochastic wind to account for the atmospheric turbulence. Both CARDAAV and CARDAAS are based on DMS methods.

The viscous code 3DVF uses numerical methods to simulate the flow field of VAWTs in cylindrical coordinates based on the solution of steady, incompressible, and laminar Navier–Stokes equations. Different dynamic stall models have been incorporated into the aerodynamic codes, namely the original Gormont model, the

**Figure 5.** *Normal force coefficient as a function of azimuthal angle for 17-m wind turbine machine at TSR = 2.86.*

**Figure 6.** *Comparison of aerodynamic torque with experimental data at TSR = 4.60 and 50.6 rpm.*

indicial model, and the MIT model [11, 34, 37]. **Figure 5** compares the normal force coefficient's distribution as a function of azimuthal angle for 17-m wind turbine machine at TSR = 2.86 using CARDAAS and 3DVF codes and experimental data. Using CARDAAS with low turbulence intensity induces a low variation in the ensemble-averaged values compared to the high turbulence intensity; obviously, for 0 turbulence intensity, both CRDAAV and CARDAAS give the same results. Comparing the two codes, one can conclude that CARDAAS and 3DVF predict quite well the distribution of the normal force coefficient. However, 3DVF is more time-consuming. **Figure 6** compares the normal force coefficient's distribution as a function of azimuthal angle for a 17-m wind turbine machine at TSR = 4.60 for a rotor rotation of 50.6 rpm. The distribution of the torque compares quite well with the experimental data. At the same time, it also gives the possibility to estimate the maximum and minimum torque encountered during the wind turbine rotation as a function of the azimuthal angle. **Figure 7** compares the wind turbine's theoretical performance with the experimental data for the Sandia 17-m turbine. The original CARDAA code does not include the variable interference factors used in CARDAAV code. As shown in this figure, CARDAA over predicts the power coefficient peak while CARDAAV shows a relatively good agreement with experimental data due to dynamic and secondary effects. Different dynamic stall models can be incorporated into the available aerodynamic codes. In **Figure 8**, the Gormont dynamic stall model and the improved Gormont model, along with the MIT dynamic stall models, have been used to predict the power generated by the wind turbine at different wind speeds.

All in all, the models forecast well the power generated by the wind turbine, but the early version of the Gormont over predict that power. In fact, the improved Gormont model by Paraschivoiu [11] takes into account the fact that high-level turbulence delays the onset of the dynamic stall. Based on that, the improved Gormont dynamic stall model is used at low turbulence zone only, namely between an azimuthal angle of 135 degrees and 15 degrees; the rest of the azimuthal angle will ignore the dynamic stall. In the MIT model case, the dynamic-stall regime is characterized by some variation compared to the improved Gormont model. In **Figure 9**, we present the case of the QBlade code used for HAWT and VAWT.

*Aerodynamic Analysis and Performance Prediction of VAWT and HAWT Using CARDAAV… DOI: http://dx.doi.org/10.5772/intechopen.96343*

**Figure 7.** *Performance comparison between theoretical and experimental data for the Sandia 17-m turbine.*

#### **Figure 8.**

*Comparison of rotor power with different dynamic stall models at 50.6 rpm.*

QBlade also uses the Blade Element Momentum (BEM) method and the DMS method to simulate HAWT and VAWT performance. Details on how to use QBlade can be found [31–33]. As an example, we generated the SG6043 airfoil pressure distribution generated using the XFOIL QBlade module. QBlade code can simulate the wake in the "Turbine Visualization" module, as shown in **Figure 10**. In **Figure 11**, we compare QBlade with CARDAAV code, including two dynamic stall models for the Sandia 34-m turbine, namely the indicial and improved Gormont models. The indicial model is used to simulate the effect of a dynamic stall at low

#### **Figure 9.**

*Example of the SG6043 airfoil pressure distribution generated by QBlade.*

#### **Figure 10.** *Example of the wake simulation generated by QBlade.*

*Aerodynamic Analysis and Performance Prediction of VAWT and HAWT Using CARDAAV… DOI: http://dx.doi.org/10.5772/intechopen.96343*

**Figure 11.** *Power distribution for the Sandia 34-m turbine using indicial and improved Gormont models at 34 rpm.*

tip-speed ratios; it is basically centered on the fact the dynamic stall is considered to be a superposition of several different effects that can be independently explored by using indicial functions [34]. As shown by **Figure 11**, all models predict quite well the power generated by the wind turbine but QBLADE slightly over predicts the power compared to CARDAAV. More validation of the QBLADE code in the case of VAWT can be found in [38].
