**4. Results**

The voltage dip was first applied on the actual Gamesa G52 WT, the technical specifications of which are shown in Table 2, and measurements were recorded, thus obtaining the positive-sequence values of active and reactive power. The measured voltage dip was then reproduced on both the detailed and the generic WT simulation models. Undoubtedly, the accuracy in the voltage dip's reproduction would affect the accuracy of the models' responses, particularly at the beginning and clearance of the fault. At simulation level, whereas older PVVC editions required the modeling of the physical test bench and the network to reproduce the voltage dip applied, the latest versions establish that a voltage-dependent source is required, setting the time series of data measured as the input signal to the voltage source, thus delivering exactly the same voltage values (see Table 1). This is currently the so-called play-back validation approach [21], which enables an accurate reproduction of the measured voltage dip [22], obtaining highly reliable results.


**Table 2.** Gamesa G52 WT technical specifications.

Both simulation WT models (manufacturer detailed and generic IEC) are therefore submitted to the measured voltage dip, shown in Figure 2, the residual voltage of which is: (i) 19.66% for phase A; (ii) 17.75% for phase B; (iii) 19.91% for phase C. The duration is 0.5705 s. The faults thus fully comply with the characteristics established by the PVVC in the case of three-phase voltage dips, as the residual voltage must be equal to or less than 20% plus the voltage tolerance (+3%), and the dip duration must be higher than, or equal to, 500 ms minus the time tolerance (50 ms) [5]. On the other hand, the time steps used for simulation are 10.024 μs in the case of the detailed model implemented in PSCAD/EMTDC and 1 ms for the generic Type 3 WT modeled in MATLAB/Simulink.

Furthermore, according to Figure 2, which represents the field measurements of the voltage dip and the fault data once applied in both the detailed and generic WT simulation models, it can be stated that the three data series match well and that, therefore, reproduction of the measured voltage in the dynamic simulation models has been performed adequately.

**Figure 2.** RMS voltage in each phase: measured and simulated.

The duration of the comparison window for the application of the validation criteria is 1000 ms, starting 100 ms before the beginning of the fault. On the basis of the information depicted in Figure 3, this 1000 ms time interval starts at [*T*1 − 100 ms] and lasts until [*T*1 + 900 ms]. *T*1 indicates the time when the fault occurs at one of the phases for the first time, as the voltage drops below the specific threshold (0.85 pu, also shown in Figure 3). *T*2 indicates the time when the deepest part of the voltage dip starts (taking into consideration both the reference and residual voltage parameters, *Uref* 1 and *Ures*1, see [5]). *T*3 indicates the end of the voltage dip's deepest part, and depends on *Uref* 2 and *Ures*2. Finally, *T*4 is the instant in which the voltage at the three phases is recovered, already above the threshold (0.85 pu).

**Figure 3.** Characterizations of zones during voltage dips according to the PVVC [5].

The validation criteria are therefore applied to the data series obtained over that time interval. As mentioned in Section 2.2, the difference in both three-phase active and reactive powers between measured and simulated data should not exceed 10% for at least 85% of the data series considered.

However, to apply this criterion correctly, it is necessary to know the test conditions. In other words, it is necessary to know which measurement and testing points have been considered.

In the case of WT model validation processes, the measurement point may coincide with the testing point, i.e., with the point at which voltage dips are applied in the actual WT by the voltage dip generator. It is important to note that the dynamic WT model to be validated will comprise all the elements downstream from the measurement point. For instance, in the case that measurement and testing points coincide, both located upstream from the transformer, the transformer and the WT model itself will be considered as validated. If, on the other hand, the testing point is located at the high voltage side of the transformer and the WT model's measurement point is at the low voltage side, only the WT model will be considered as validated. This is explained graphically in Figure 4.

In the present case, the testing point, i.e., the point in which the voltage dip generator is connected, is located at the high voltage side of the transformer, while two different measurement points were considered and therefore two verification procedures were performed, both in the high and low voltage sides. The WT transformer has a transformation ratio of 20 kV/690 V, as indicated in Figure 4.

**Figure 4.** Measurement and testing points for WT validation according to the PVVC.

Based on the information given in Section 2.2, 1 pu is considered as the rated value to apply the PVVC criteria. Hence, 0.1 pu is the maximum deviation allowed for at least 85% of the points within the data series to be compared. Tables 3–5 in Sections 4.1 and 4.2 respectively, estimate these validation parameters, in addition to the percentage of points that comply with this criterion and have a maximum deviation below 0.1 pu.

In this case, the WT is operating under full load conditions of approximately 0.93 pu, also complying with the criteria established by the PVVC (full load test, ≥80% *Pn*, see Table 1).

#### *4.1. PVVC Criteria Applied at the Testing Point: 20 kV, Measurement Point 1*

On the one hand, Table 3 presents the IEC model's results for the PVVC criteria applied at the testing point, i.e., at the 20 kV voltage side, which is the so-called *measurement point 1* according to Figure 4. As testing and measurement points coincide in this case, validation results will affect both the transformer and the WT models.

**Table 3.** Verification of the PVVC validation criteria applied to the IEC generic WT model at the testing point, operating at full load conditions: 20 kV, measurement point 1.


On the other hand, Figure 5 shows the behavior of the active and reactive power during the voltage dip in the three cases analyzed: field test, detailed WT simulation model and IEC generic WT model.

**Figure 5.** Active and reactive power at measurement point 1 (20 kV). WT operating at full load conditions.

Regarding the IEC model results, considering the pertinent time window for evaluation of the data series (see Table 3), only 80% of the points subjected to the analysis comply with the validation criteria for active power. In the case of reactive power, this value is still lower, 56%. Therefore, the validation criteria established by the PVVC are not complied with in either case. However, as will be illustrated in Section 4.2, this non-compliance situation is reversed in one of the cases when the low-voltage measurement point is considered. It can thus be stated that, in this case, neither of the magnitudes subjected to the validation analysis comply with the PVVC criteria when it is applied at the testing point and that, therefore, the transformer model implemented along with the IEC WT model must be considered as non-validated.

Nevertheless, considering the graphic results in Figure 5a, it is clear there exists a reasonably good correlation between the IEC model and the field measurements in the case of active power. The differences are mainly because the generic model is not able to represent the transient periods of the actual WT with grea<sup>t</sup> precision. To improve the representation of these transient periods, two modifications should be carried out in the generator system defined by standard IEC 61400-27-1, the modular structure of which is available in [6]:


Regarding the reactive power behavior of the IEC WT model (Figure 5b), although there is also non-compliance with the PVVC, transient periods are better represented by the generic WT model in this case. However, the level of accuracy is still low if compared to the transient periods that appear in the detailed model's responses and field measurements. This is also because simplified models such as that developed by the IEC do not represent the fundamental component of the transformer inrush current. This can be understood more easily if we look at the graphic results shown in Section 4.2, specifically at the reactive power graph obtained for the low-voltage side measurement point (Figure 6b), in which reactive power fits much better than in the present case.

The differences between the field measurements and the IEC model, both in the active and the reactive power responses, which are mainly due to the inability of the generic model to accurately represent the transient periods, are described with grea<sup>t</sup> precision in [23]. Indeed, the study summarizes the simplifications implemented in the Type 3A WT model, which is the one modeled in the present work, as explained in Section 3.

The validation results for the detailed WT model, also operating at full load conditions, are shown in Table 4. As can be observed, both the active and the reactive power responses fulfill the PVVC criteria when applied at the testing point, since 91% and 90% of the data series analyzed, respectively, are below the maximum deviation allowed. This implies that the detailed transformer and WT models simulated were validated according to the Spanish Grid Code.


**Table 4.** Verification of the PVVC validation criteria applied to the detailed WT model at the testing point, operating at full load conditions: 20 kV, measurement point 1.

Compliance of the detailed model with PVVC validation criteria is also supported by Figure 5. On the one hand, Figure 5a shows the detailed model's active power response facing the voltage dip. It can be seen that there is an excellent correlation between this signal and the one provided by the field measurements. The same applies to the reactive power (Figure 5b), the signal of which is also very similar to the field measurements. In this case, transient periods are accurately represented since the detailed WT model considered the actual mechanical, electrical and electronic systems of the Gamesa G52 WT, including their parameters and algorithms.

> **Table 5.**

side, operating at full load conditions: 690 V,

Verification of the PVVC validation criteria applied to the IEC generic WT model at the low voltage

#### *4.2. PVVC Criteria Applied at the Low Voltage Side: 690 V, Measurement Point 2*

The detailed WT model representing the Gamesa G52 WT broadly complies with the validation criteria established by the Spanish Grid Code when analyzed at the testing point, and hence no additional validation analyses are required. However, the generic WT model developed by standard IEC 61400-27-1 does not meet the minimum requirements to be considered a validated WT model according to the same criteria when applied at the testing point. As listed in Section 4.1, there are two main reasons for this situation: the inability of generic WT models to accurately represent, on the one hand, transient responses of actual WTs, and, on the other hand, the fundamental component of transformer inrush current. Indeed, modifications in the structure of the original generator system model developed by the standard have been proposed in order to improve transient responses of the generic Type 3 WT.

Nevertheless, regarding the second reason, the IEC model's reactive power response does comply with the PVVC criteria when the transformer is not taken into account in the validation process, i.e., when the inrush current effects are no longer considered a problem. Therefore, when applying the criteria at the low voltage side, 690 V, *measurement point 2* according to Figure 4, the percentage of data series below the maximum deviation allowed increases to 87%, higher than the 85% set as the target (Table 5). Indeed, Figure 6b shows that the correlation in reactive power between the generic WT model and the field measurements is much better than in the previous case (Figure 5b). Active power, on the other hand, continues to be in non-compliance, with only 80% of data series within the margin established. The graphic results for active power (Figure 6a), are very similar to those obtained at the testing point (Figure 5a, Section 4.1).

It can therefore be stated that the non-compliance situation in the previous case for reactive power is now reversed, since the response of the generic IEC WT model does fulfill the PVCC validation criteria at the low voltage side.


measurement

 point 2.

**Figure 6.** *Cont*.

**Figure 6.** Active and reactive power at measurement point 2 (690 V). WT operating at full load conditions.

#### *4.3. PVVC Criteria: WT Operating at Partial Load Conditions*

The previous sections have presented the analysis of the IEC Type 3 WT model operating under full load conditions. Indeed, such a condition is presented as the worst-case scenario with respect to the compliance of the generic model with Spanish PO 12.3. Thus, the non-compliance cases have been extensively discussed, and the causes analyzed. The modeling modifications that should be implemented within the generic WT model to improve the compliance results have also been presented.

Furthermore, in order to enhance and enrich the current section, a different study case was also analyzed. In this case, the WT is operating at partial load conditions of 0.20 pu, which is in line with the requirements established by the PVVC (partial load test, 10–30% *Pn*, see Table 1). The voltage dip applied to the IEC WT model working at partial load conditions is the same as that applied to the previous compliance analyses (see Figure 2).

Tables 6 and 7 show the percentage of points that comply with the PVVC validation criteria when applied to the high and low voltage sides of the transformer, respectively. Figures 7 and 8 show the active and reactive power responses in those two measurement points.

A value of 0.1 pu must be the maximum error for at least 85% of points within the data series to be compared. As can be observed in Table 6, the active power response fulfills the PVVC validation criteria at the high voltage side of the power transformer, i.e., at the testing point (or measurement point 1, see Figure 4), since 88% of the data series analyzed are below the maximum deviation allowed. Indeed, if the graphic results in Figure 7a are considered, it can be observed that there is a good correlation between the IEC model response and the field measurements, which justifies the compliance of the active power response at measurement point 1 (see Figure 4) with the PVVC validation criteria. Moreover, the reactive power response fails to comply with PO 12.3, since only 58% of the data series are below 0.1 pu. The differences in this case may be clearly observed in Figure 7b. As in the case of the PVVC criteria applied to the high voltage side with the WT operating at full load conditions (Figure 5b), this non-compliance situation is also due to the inability of the generic WT model to represent the transformer inrush current.

However, in general, the compliance results improved in comparison to the WT operating at full load conditions.

**Table 6.** Verification of the PVVC validation criteria applied to the IEC generic WT model at the testing point, operating at partial load conditions: 20 kV, measurement point 1.

**Figure 7.** Active and reactive power at measurement point 1 (20 kV). WT operating at partial load conditions.

When the validation criteria are applied to the low voltage side (Table 7 and Figure 8), both the active and the reactive power responses of the WT operating at partial load conditions fulfill the PVVC criteria, since 91% and 88% of the data series analyzed, respectively, are below 0.1 pu. The good correlation that exists both in the active and the reactive power between the field measurements and the IEC model is observed in Figure 8a,b, improving as regards the PVVC applied to the low voltage side. The situation for

the reactive power has thus reversed, since it complies with the PVVC criteria in this case. This is for the same reasons as those given in Section 4.2: simplified or generic models do not represent the fundamental component of the transformer inrush current.

**Table 7.** Verification of the PVVC validation criteria applied to the IEC generic WT model at the low voltage side, operating at partial load conditions: 690 V, measurement point 2.


**Figure 8.** Active and reactive power at measurement point 2 (690 V). WT operating at partial load conditions.
