Experimental Investigation of a Standalone Wind Energy System with a Battery-Assisted Quasi-Z-Source Inverter

Round 1

Reviewer 1 Report

The paper presents the control of a SCIG based wind turbine with energy storage incorporated with a Quasi-Z-Source Inverter. Also the paper includes experiments. I have the following comments and suggestions for the authors:

• The introduction is too general. I understand that part of the contribution of the paper is the integration of the battery in the wind turbine but no review of other references, including other energy storage and/or energy storage converter topologies, are presented. This is not a new topic and should be discussed in the introduction.
• The authors claim that the paper's main contributions are the extensive simulations of the system and the inclusion of the start operation or low wind conditions, but this is not clear in section 2. I will suggest adding a technical discussion about these points in section 2.
• The description of the controllers is too general and makes it impossible to reproduce the results. There is not a single equation in the whole paper. I suggest the authors add the exact equations of the models used to tune the controllers, the controllers' equations and mention the used controllers' tunning techniques.
• The experimental results are well presented and the description of each experiment is accurate.

Author Response

We would like to thank the reviewer for the valuable comments. The time and effort spent are much appreciated. Our responses to the comments are given below, whereas all the applied corrections are marked yellow in the revised version of the paper.

1. The introduction is too general. I understand that part of the contribution of the paper is the integration of the battery in the wind turbine but no review of other references, including other energy storage and/or energy storage converter topologies, are presented. This is not a new topic and should be discussed in the introduction.

A brief review of review of the energy storage technologies used in WECSs is now provided in the first paragraph of the Introduction. Due to the newly added references, the reference numeration in the text has been updated accordingly.

2. The authors claim that the paper's main contributions are the extensive simulations of the system and the inclusion of the start operation or low wind conditions, but this is not clear in section 2. I will suggest adding a technical discussion about these points in section 2.

A technical discussion about the start operation and operation at low wind conditions is added to Section 2, as requested. The new subsection “2.5 System Startup and Low-Wind Operation” discusses certain technical aspects related to the system startup and low-wind operation. This, in turn, allowed the comments in Sections 4.1 and 4.5 to be more focused on the obtained results.

3. The description of the controllers is too general and makes it impossible to reproduce the results. There is not a single equation in the whole paper. I suggest the authors add the exact equations of the models used to tune the controllers, the controllers' equations and mention the used controllers' tunning techniques.

The equations of the utilized controllers were not provided because they are all standard proportional-integral or integral controllers, whose equations are well known from the literature. The respective parameters (i.e., proportional and integral gains) were optimized through trial-and-error procedures performed as part of the simulation studies reported in [39] – this is now mentioned in the revised manuscript. The rule base and the scaling factors for the FLCs in Figures 4 and 5 are now provided in Section 2, whereas the hysteresis band width is defined in Section 3 as being equal to 2% of the reference phase current amplitude, as proposed in [18].

The equations of the utilized IRFOC algorithm were derived based on the conventional induction machine model, whose equations are now defined in the newly introduced subsection “2.1 SCIG and qZSI Equations” (the numeration of other subsections was updated accordingly). This subsection also contains the qZSI equations defining the relationship between the qZSI input and output voltage through D₀ and M.

Reviewer 2 Report

Well-structured paper, with consistent results. Below are the comments:

Regarding the WT characteristic, why not implementing one that has decent behavior in the low wind speed area (i.e. 3-6m/s)?  As starting speed, 6m/s seems too high in this regard.

Why the authors use rather old/low efficiency components? Here I mean the IE1 class generator and the lead acid batteries? Why not an IE3 generator and Li-Ion batteries? This would solve (partially) the problem of losses minimization and SOC excursion (down to 20% maybe).

The time constants seem rather high “(i.e., at t ≈ 95 s)”; besides being effective, the control algorithm should offer a quick response.

The authors stated that “the only study to consider the qZSI as part of a SCIG-based WECS is that reported in [24]” but they refer to one of their papers. In this case, they should mention that it was their work.

Author Response

We would like to thank the reviewer for the valuable comments. The time and effort spent are much appreciated. Our responses to the comments are given below, whereas all the applied corrections are marked yellow in the revised version of the paper.

1. Regarding the WT characteristic, why not implementing one that has decent behavior in the low wind speed area (i.e. 3-6m/s)?  As starting speed, 6m/s seems too high in this regard.

The utilized WT emulator was based on the generic WT model reported in the following papers:

Zaragoza, J.; Pou, J.; Arias, A.; Spiteri, C.; Robles, E.; Ceballos, S. Study and experimental verification of control tuning strategies in a variable speed wind energy conversion system. Renewable Energy, 2011, 36 (5), 1421-1430.

Martinez, F.; Herrero, L.C.; de Pablo, S. Open loop wind turbine emulator, Renewable Energy, 2014, 63, 212-221.

The guiding principle for selecting the values of the WT model coefficients was to obtain a typical small WT with the maximum power of 1.5 kW at the wind speed of 11 m/s, and with the tip-speed ratio equal to 6. The value of the WT blade radius was set to R = 1.5 m, which may be considered typical for such a small WT, whereas the air density was assumed to be equal to 1.225 kg/m³. The step-up gearbox with the 3:1 ratio was used to adjust the WT speed and torque to the coupled SCIG, that is to ensure that the mechanical torque and the rotor speed delivered to the SEIG shaft at the rated wind speed conditions closely correspond to the rated torque and speed values of the used induction machine (about 10 Nm and 1400 rpm, respectively). For the selected WT parameters, the maximum power delivered to the SCIG shaft at v_w = 6 m/s amounts to 244 W. Now, to extract approximately the same amount of power from the WT at v_w = 3 m/s (i.e., at the actual cut-in speed of the utilized WT), the blade radius would have to be about 2.8 times larger, i.e., R > 4 m, which would be highly untypical for this power range. Note also that in this case the maximum WT power at v_w = 11 m/s would be almost 12 kW.

The WT utilized in this study served primarily to demonstrate the validity of the proposed concept. Note that the respective power-speed characteristics could have been adjusted so as to achieve higher output power at lower wind speeds (resulting in oversized WT blades) or the whole WECS could have been scaled up with respect to the rated power, thus enabling higher power output at lower wind speeds. But, control-wise everything would basically stay the same, only the controller parameters would need to be further fine-tuned (of course, this would also require scaling up the ratings of the laboratory setup power components). We have previously used the same WT emulator as part of the laboratory test setup in several publications (in [18] and [39] among others), so we decided to use it in this study as well, especially since all the required equipment was already present in the laboratory. An additional argument in favor of this choice was the fact that we used the same WT model in the simulation study of the considered WECS, published in [39], which preceded the experimental investigation. This allows one to compare the obtained results, not only qualitatively, but also quantitatively. To sum up, in our opinion, the utilized WT served well for the purpose of proving the proposed control concept.

2. Why the authors use rather old/low efficiency components? Here I mean the IE1 class generator and the lead acid batteries? Why not an IE3 generator and Li-Ion batteries? This would solve (partially) the problem of losses minimization and SOC excursion (down to 20% maybe).

Selection of lower efficiency components has its pros and cons. For example, referring to the previous point, the efficiency class of the generator affects the minimum operational wind speed. An IE3 class machine generates more electrical power compared to an IE1 class machine for the same input power provided by the WT. This enables more efficient operation at low wind speeds. On the other hand, IE1 class machines are more suitable for testing of the loss optimization algorithms precisely because of the relatively large share of losses. In other words, the loss optimization makes greater impact in IE1 class machines. We definitely intend to test the developed control algorithms with IE3 class induction machines – both aluminum and copper rotor type – and we have made similar comparisons before in our papers (“further optimizations” mentioned in the last sentence of the Conclusion refer, in part, to this aspect). There is no reason to doubt that the proposed algorithms will prove successful on higher-efficiency machines, even probably with shorter optimization time. However, this type of analysis falls out of the scope of this study.

As for the choice of batteries, Li-Ion batteries would definitely allow greater depth of discharge and number of cycles, as well as the reduction of the required storage space, but their biggest drawback is their high price compared to lead-acid batteries, with the latter being several times cheaper for installation. Therefore, for small, stationary systems, as is the one considered in this study, the selection of Li-Ion batteries may not be considered economically viable. Lead-acid batteries represent the most mature battery technology today, which is important regarding their reliability, whereas the developments in the technology achieved in recent years have much prolonged their cycle lives. Finally, lead-acid batteries are the only battery energy storage system that is almost completely recycled (over 99% are recycled in Europe and USA). The reviewer is referred to the following paper for more details about the utilization of lead batteries for energy storage:

May, G.J.; Davidson, A.; Monahov, B. Lead batteries for utility energy storage: A review. Journal of Energy Storage. 2018, 15, 145-157.

Selected parts of the text above can now be found in the first paragraph of the Introduction, whereas the aforementioned review paper is added to the reference list.

In any case, one of the goals of this paper was to prove that a self-sustaining WECS can be assembled using low-cost components (e.g., generator and batteries), so in this regard we believe that, at this point of research at least, there is no need or scientific benefit in using more expensive components which would, most certainly, produce better results in terms of efficiency.

3. The time constants seem rather high “(i.e., at t ≈ 95 s)”; besides being effective, the control algorithm should offer a quick response.

The 95 s interval, which the reviewer refers to here, is related to the system startup. This is indeed a specific situation where the WT starting speed is rather far from the optimum. As seen in Figure 7a, the difference between these two speed values is almost 300 rpm at the beginning of the WT optimization. In addition, the WT optimization starts at t ≈ 30 s, so it actually lasts about 65 s and not 95 s. In normal operating circumstances, the optimization is much faster. For example, in Figure 11, the wind speed changes from 10 m/s to 11 m/s at t ≈ 20 s, causing the WT optimization to start over, but the optimization is this time over in less than 10 seconds, including the SCIG loss optimization.

In any case, the slowest dynamic response observed in the system is that of the WT-SCIG subsystem. Due to the comparatively large time constant of the corresponding mechanical subsystem, the WT optimization operates in time steps of 3 s. This is in order to give the PI speed controller enough time to force the WT – through torque control – to achieve the reference rotational speed. It has to be noted that the aforementioned time step of 3 s could potentially be reduced by using a more advanced speed controller, as suggested in the following paper:

Vukadinović, D.; Bašić, M.; Nguyen, C.H.; Nguyen, T.D.; Vu, N.L.; Bubalo, M. Optimization of a Hedge-Algebra-Based Speed Controller in a Stand-Alone WECS. 5th International Conference on Smart and Sustainable Technologies (SpliTech), Split, Croatia, 23-26 September 2020.

However, this type of analysis falls out of the scope of this study.

4. The authors stated that “the only study to consider the qZSI as part of a SCIG-based WECS is that reported in [24]” but they refer to one of their papers. In this case, they should mention that it was their work.

This fact is now mentioned in the penultimate paragraph of the Introduction (page 3, line 1 of the revised manuscript).

Round 2

Reviewer 1 Report

Thanks for addressing my comments. I am satisfied with the paper as it is.