Design and Implementation of an Experimental Test Bench for the Study of Active Flow Control in Scale Wind Turbines

Understanding and simulating the wind energy conversion process for individual wind turbines and wind farms requires the ability to model multiple complex physical processes interacting at various spatial and temporal scales. Effective design of wind energy systems fundamentally relies on a solid understanding of physics and the accuracy of the mathematical models used in simulations [1]. After the development and modeling phases, testing and certification procedures are required to ensure the correct operation of the wind turbines. Test bench systems offer a controlled solution to achieve these requirements. These systems provide experimental reproducibility, advanced measurement capabilities, and adjustable operating conditions, integrating a comprehensive and detailed characterization of wind turbine performance. Various studies have presented advances in developing test benches to analyze the phenomena and variables involved in wind energy systems, such as the case of Averous et al. [2], who presented the development of a full-size 4 MW wind turbine tested following a multiphysics-in-the-loop power hardware concept. Milich et al. [3] presented the mathematical analysis of a wind turbine with NACA 4412 profile where various stages were developed, such as the development of a theoretical aerodynamic analysis through the design of the wind turbine model, the structural verification of the blades, the theoretical calculation of the torque, and an experimental evaluation of the generating device. This research contains the necessary elements of a functional aerodynamic model that allows physical construction validated by theory. Klein et al. [4] conducted a study about the relevance of different numerical and experimental methods to reproduce measurement models of wind turbines in wind tunnels, including a high blocking ratio. Also, Botasso et al. [1] presented an analysis of a wind tunnel at Milan Polytechnic, enabling experimental studies in aerodynamics, aeroelasticity, and control.

Reduced-scale models have emerged as a practical alternative for performing tests in controlled environments, offering significant benefits when full-scale testing is cost-prohibitive or logistically challenging. However, there are several difficulties and challenges to face when implementing reduced-scale models of wind turbines. The reduction in the models can modify critical aerodynamic parameters, such as Reynolds and Mach numbers, which affects the accuracy of the test results and makes it challenging to replicate the behavior on a full-scale wind turbine. These downscale models operate at lower Reynolds numbers than full-scale prototypes, resulting in changes in flow characteristics that affect the validity of the data when extrapolated to full-scale systems [5]. Another challenge that small-scale models face is the variability of response that may require different materials or structural redesigns to accurately mimic the stresses and deformations of the full-scale prototype [6–8]. Recent studies have demonstrated the effectiveness of tailored optimization algorithms in compensating for scaling effects and enhancing data reliability in reduced-scale testing; Li et al. [9] revealed a novel methodology with an optimization algorithm for designing a scale laboratory model to study the wake in a 2.5 MW turbine, thereby validating the possibility of obtaining scale models with a range of 1:300. Likewise, Canet et al. [6] formulated laws for the scaling of wind turbines, the work focused on designing a smaller model that mimics a full-scale machine. As a result, it was observed that with suitable options, models subjected to experimental testing with high scaling values can provide reasonably accurate performance results. Finally, Giahi and Jafarian Dehkordi [10] presented a study of the influence of dimensional scaling on simulation results through computational fluid dynamics tools, obtaining results entirely proportional to those obtained in the theory of aerodynamic similarities. With the growing interest in offshore wind turbines, numerous studies have focused on validating scale models to investigate the application of scaling laws and ensure model-prototype aerodynamic similarity. These studies have provided valuable comparative analyses with simulation results, enabling a deeper understanding of offshore turbine systems. Applying scaling laws has proven essential in addressing unique challenges posed by offshore environments, such as variable wind patterns and sea conditions [11,12].

Although power control is essential for all wind turbines, load alleviation and control have been established as the main challenge for large wind turbines. With the increase of wind turbine’s rotor diameter, the blades and their aerodynamic loads also increase, which demands either new blade materials or the implementation of fast response load relief mechanisms [13]. Various authors have studied these active control systems in recent years; however, in most cases, they have faced the obstacle of experimental validation in larger turbines due to the costs and accessibility of these turbines in wind farms [14–19].

As a solution to this problem, this paper presents the design and implementation of an experimental test bench to study active flow control in small-scale wind turbines. The design methodological proposal is presented based on maximum critical requirements: test rotor size, blockage ratio, thrust force, and wind tunnel velocity. This section includes a finite element method (FEM) analysis to evaluate different configurations. As a result, the design and implementation of the general functional structure, dynamic sensors, and complementary elements are revealed. Finally, a scaling methodological proposal for the test bench is discussed. This approach is based on a 500-kW turbine prototype scaled to a laboratory turbine model of 0.9 m diameter, which achieves the maximum rotor diameter that will be tested in the test bench presented in this work, according to the maximum range of blockage ratio < =0.4 which allows the adequate correction of the performance parameters of wind turbines (C_P, C_T) through the mathematical models explored in the literature [20–23].

The main contributions are summarized below:

• A novel test bench design based on maximum critical requirements: test rotor size, lock ratio, thrust force, and wind tunnel speed.

• A comprehensive design methodology incorporating FEM analysis to evaluate different configurations that ensure a robust structure, functionality, and versatility of the test bench.

• A scaling methodology approach based on a review of the critical parameters that must be analyzed to carry out the reduced scaling of wind turbines for wind tunnel testing.

Figure 1 shows the outline of the test bench design methodology with which the various stages were established according to the design methodology described in Ref. [24].

The case study of this paper is limited to integrating the test bench system in the wind tunnel of the National School of Higher Studies Juriquilla of the National Autonomous University of Mexico (UNAM) (Fig. 2). The main characteristics of the tunnel are a maximum wind velocity of 25 m/s and an area in the test section of 1.6 m².

The model of one-dimensional theory and Betz’s law were applied to establish the axial force T as a design variable for the definition of the technical requirements of the test bench. One-dimensional theory examines a simple one-dimensional model for an ideal rotor. The rotor in this model is considered an actuator disk. This disk reduces the wind speed from V₀ upstream of the rotor, through u in the rotor plane, to u₁ in the wake. This behavior can be seen in Fig. 3 [24].

where A is the rotor area, and $Δp$ is the pressure difference in the rotor section.

where ρ is the air density.

The maximum C_p = 16/27 = 0.59 occurs when the value of the axial induction factor a = 1/3.

The aerodynamic theory of similarity indicates that to find similarity between model-prototype of wind turbines, the conditions of geometry, kinematic, and dynamic similarity must be met [9].

where D is the diameter of the rotor, and the subscripts denote m for the model and p for the prototype.

Dynamic similarity: According to Gasch and Twele [25], if the nondimensional curves of C_P, C_T, and moment coefficient C_M are similar, the flux conditions in a wind turbine must be the same.

Therefore, the scaling methodology used for the test bench proposed in this work will seek to achieve similar performances concerning their dimensionless dynamic numbers while maintaining geometric and kinematic similarity. The test bench aims to test novel active control systems, so it is crucial to achieve a high level of dynamic similarity that allows measuring variations in (C_P) and main loads and moments (C_T and C_M).

Table 1 shows the specifications and technical requirements of the system. The data in the table were derived from the operational parameters of the wind tunnel where the test bench is implemented and the functional characteristics integrated by the system. The instrumentation and data acquisition system is based on a National Instruments CompactDAQ-9178 device with I/O analog and digital modules (Fig. 4). The CompactDAQ system has a timing resolution of 12.4 ns, and the main parameters and characteristics of each module are described in Table 2.

The capabilities of the instrumentation and control system of the test bench integrate control and measurement devices for critical variables for wind turbines. The acquisition system incorporates high-precision torque, rotor speed, wind speed, and pressure measurements. Moreover, the system outputs control signals for pitch, flap, and brake control actuators. Finally, the test bench requires flexibility to adapt to several experiments. The modular architecture of NI CompactDAQ makes the test bench instrumentation and control system scalable and adaptable.

A DECENT DYN-200 dynamic torque sensor was integrated for torque and power measurements, allowing synchronous RPM and dynamic torque measurement. Additionally, BMP180 barometric pressure and temperature sensors were integrated.

In order to achieve the technical requirements expressed in Table 1, a first design was developed, which integrates a base with a degree-of-freedom for yaw positioning, a tower at the average height of the test section of the wind tunnel, a nacelle base for the installation of the main shaft, two bearings, measuring instruments, and electric generator (Fig. 5).

Design iterations were based on the von Mises safety factor criterion to obtain the final design. The objective parameter of this optimization methodology was the reduction of material and, consequently, cost. The FEM simulations were carried out with solidworks simulation with the selection of the AISI 304 material, and the elements for the analysis were restricted to the use of the two bearings where the radial and axial forces are concentrated.

Boundary conditions: A fixed geometry was applied to the underside of the tower base (Fig. 6). The thrust force T was applied in the test bench’s main shaft, and two forces of 5 kg were applied to the torque sensor and the generator position to simulate the weight of each element (Fig. 7).

Figure 8 shows the three design iterations that were analyzed with a static FEM analysis. The modifications from designs (a) to (c) mainly reduced structural elements and material thickness to obtain a lighter prototype.

Table 3 shows the static simulation results for the three design iterations, where the first variable to compare is the maximum stress (Pa), maximum displacement (mm), and minimum factor of safety (von Mises) presented under the same boundary conditions; it is possible to observe that design (b) obtained better performance for stress and a smaller displacement; however, the weight is higher to design (a) and (c). The maximum displacement of 2 mm appears on the tilt angle and is a reasonable value, considering the operating conditions of the wind turbine. This displacement is achieved only under the maximum operating wind velocity of the wind tunnel, an infrequent but critical scenario for the system’s safety. In these cases, the wind turbine is designed to automatically activate the breaking system to reduce structural loads and prevent possible damage to components.

The primary consideration for selecting the final design was the weight and simplicity of the manufacturing process since elements for structural support at the base of the nacelle were removed. This resulted in a lower manufacturing cost, a safety factor of 4.8, and a lower weight than the two design iterations (a) and (b).

The test bench hub was designed with a flexible flange system to exchange three-blade rotors, and the hub also integrates a pitch angle control mechanism. The pitch system in small-scale wind turbines plays a vital role in controlling the angle of the rotor blades to maximize energy capture and prevent mechanical damage in extreme wind conditions. The design process was based on aerodynamic principles, mechanical engineering, and integration with control systems.

Figure 9 shows the pitch angle system configuration. This design incorporates a steel bevel gear system mechanism where the gear ratio is 1:2 (20–40 T). The bevel gear system was coupled to a servomotor MG996 with a torque capacity of 9.4 kgf·cm, a movement range of 0–180 deg, and a response time of 0.17 s/60 deg controlled by pulse width modulation. Furthermore, the hub integrates a manual pitch setting mechanism in order to compare different responses in static and dynamic tests on the test bench.

The test bench was manufactured and assembled with its components and instrument devices. The main structure was manufactured with AISI 304 stainless steel, and the hub rotor was made of ONIX 3D filament, which is nylon reinforced with carbon microfibers. In addition, the blades were manufactured using three-dimensional (3D) printing PLA material and were subjected to a surface treatment with polyurethane coating (Fig. 10). The PLA material was selected based on the fact that in previous studies, it has been presented as an excellent solution for the micro wind energy applications due to its low deformation values, and it has also demonstrated less stress and strain during axial loading [26].

Figure 11 shows the CAD model of the final design and the test bench setup integrated into the wind tunnel, indicating the location of its main components.

According to the geometric, kinetic, and dynamic similarity conditions above, the conversion relationship that each physical quantity should satisfy for the model turbine can be deduced as follows in Table 4. The mathematical deduction of the parameters shown in the table was presented in Ref. [25].

The comparison of a theoretical prototype wind turbine of 0.5 MW and a lab-scale model with a geometric scale factor of 1:45 was applied using the conditions of the Similarity Theory above. The scale conversion parameters below in Table 5 were implemented to perform the BEM simulations using the open-source software qblade [27]. qblade software performs an optimized unsteady polar-BEM [28], and the BEM parameters are shown in Table 6.

Figures 12–14 present the results of the steady BEM analysis of the power, thrust, and moment coefficients versus TSR, where the rotor blade corresponding to the prototype wind turbine is the line with the triangle marker, and the one corresponding to the model wind turbine is the line with the hexagram marker. As can be seen, the behavior of both coefficients is not sufficiently similar. As shown in Table 5, the Reynolds number difference between both is significant; there is no dynamic similarity anymore. As a proposal, this approach required an optimization process in future work, which would require considering the use of enhancing low Reynolds airfoils and a geometric variation of the twist angle and chord.

The technological development and innovation of wind turbines have allowed their energy contribution to be on the rise. Test benches are presented as an essential key tool for the experimental analysis of wind turbines at the laboratory level. This paper presents the methodology for designing a small-scale wind turbine test bench and its implementation in the ENES Juriquilla UNAM, Mexico wind tunnel. As a result, a comparative analysis of the design iterations based on FEM simulations was developed. The test bench was designed and manufactured for a maximum wind speed of 25 m/s and a safety factor 4.8. The system can measure rotational speed, torque, pressure, temperature, voltage, and current variables. The system can measure rotational speed, torque, pressure, temperature, voltage, and current variables through a National Instruments CompactDAQ data acquisition system. In addition, this study revealed a configuration for active pitch control through servomotor and bevel gears. The final design features functional structures that allow a 45 deg yaw angle adjustment, and the nacelle base design integrates slots to install auxiliary components that will allow future studies, such as diffusers, to increase wind turbine performance.

Although wind tunnel tests could not fully replicate full-scale conditions or replace field tests, they are essential for validating and adjusting models and new active control systems, such as pitch, camber morphing, and flap. The selection of scale values in reduced models for wind tunnels is usually based on a combination of aerodynamic, structural, and flow similarity criteria that attempt to approximate the behavior of full-scale prototypes as closely as possible.

Theoretically, adjusting the static pressure in a wind tunnel could modify the flow density and thus achieve a higher Reynolds number. However, the tunnel infrastructure used in this study has limitations regarding the adjustable static pressure range, which limits direct correspondence with the Reynolds values of the prototype. The methodology suggested in this work is based on the studies analyzed, where it has been proven that adapting the geometric similarity can achieve similarity in the critical aerodynamic parameters. These investigations have demonstrated that, although the exact Reynolds number is not equal, the wind tunnel data of reduced models remain representative [1–12]. As future work, developing methods based on intelligent algorithms for assisted scaling for large wind turbine rotors is crucial.

The authors would like to thank Universidad Nacional Autónoma de Mexico through PAPIIT TA100323 project²

• PAPIIT TA100323 Project through Universidad Nacional Autónoma de Mexico (Funder ID: 10.13039/501100006087).

The authors attest that all data for this study are included in the paper.

a =

axial induction factor

A =

rotor area, m²

C_M =

momentum coefficient

C_P =

power coefficient

C_T =

thrust coefficient

D_m =

diameter of the model rotor, m

D_p =

diameter of the prototype rotor, m

p =

pressure, Pa

P =

power, W

T =

axial force, N

u =

wind velocity in the rotor plane, m/s

u₁ =

wind velocity in the wake, m/s

V₀ =

wind velocity upstream of the rotor, m/s

Greek Symbols

λ_L =

geometric scale relationship

λ_V =

velocity scale relationship

ρ =

air density, kg/m³

σ =

stress, Pa

Acronyms

BEM =

blade element momentum

CAD =

computer-aided design

PLA =

polylactic acid

PMSG =

permanent magnet synchronous generator

RPM =

revolutions per minute

RTD =

resistance temperature detector

TTL =

transistor-transistor logic

TSR_m =

tip speed ratio of the model

TSR_P =

tip speed ratio of the prototype