I. INTRODUCTION
The current precarious environmental situation, result of global pollution, shows that there will be an energy-environmental crisis due to high levels of pollution and the depletion of most non-renewable natural resources. Consequently, it is essential to exploit renewable energy, one of these being wind power [1].
The demand for energy generation through alternative sources has been growing globally, approaching 25% per year. This growth has been driven by an increase in global demand for electricity and the rising cost of non-renewable fuels, among others [2].
Electricity generation from renewable energy sources managed to reach 21% and 22.8% of the global generation in 2012 and 2014 respectively. From these figures, 3.1% corresponds to energy produced from wind power [3]. Moreover, the production of electricity from this resource is an excellent alternative for overpopulated areas. Countries such as Canada and China, among others, have established ambitious development goals, seeking to continue the exploitation of this energy source due to the challenges of acceptance in local communities [4], [5]. Results such as the 8.75 MW of wind power installed offshore in 2014 are the result of major technological advances and the time dedicated to developing this type of energy production [4].
In Colombia, wind energy is also one of the alternatives the country has to supply the increase on energy demand. According to the UPME expansion plan, it is estimated that in the next few years, 474 MW of wind power will be installed in the northern part of the Guajira department, which would replace 250 MW of coal-based thermal energy [6]. Given the importance that this type of alternative energy has for the country, it is necessary to generate spaces for researching on this energy source. Thus, the prototype developed seeks to serve as a tool for the study of systems based on wind power that can be interconnected to the electrical network and at the same time, can work in isolation as well. This can be achieved through the emulation and manipulation of wind profiles from different areas, done in a controlled way.
This paper presents the design, implementation and experimental validation of a low power wind turbine emulator and is structured as follows: In section 1, the introduction is presented; section II describes the materials and methods. Furthermore, it presents the main characteristics that a wind emulator must have to establish the parameters and objective functions for the design and implementation of the emulator in question. Section III presents the experimental validation of the implemented emulator operating not only manually (data provided by the manufacturer of the Eolos turbine) but also automatically (loading the wind profile of the Uribía-Guajira region in Colombia). Additionally, these results are discussed. Finally, the last section presents conclusions.
II. MATERIALS AND METHODS
An experimental methodology and a quantative approach is used in this research The main characteristics that wind turbine emulators should have are investigated and design is proposed. Such design is tailored to the requirements of the problem to be addressed (low power emulators with the possible incidence in electrical microgrids). Finally, the development is validated through the implementation and a set of experimental tests.
A. Features of Wind Turbine Emulators
Wind emulators aim to reproduce the static and dynamic behavior in a controlled environment of wind turbines without the need to rely on the existence of the actual wind turbine or the actual wind resource [7].
Fig. 1 shows the conversion process of a conventional wind turbine as well as some of its components, in which the wind energy is used to move a generator (transformation of mechanical energy into electrical energy). In this figure, a gearbox is displayed, which can be ignored at the time of the emulator implementation by using a permanent magnet generator [8]. This provides, in turn, the mechanical torque to the electric generator [9].
Generally speaking, wind emulators seek to replace wind speeds with a profile entered into a computer or micro-processing system, whereas wind turbines do so with a DC motor, thus establishing two constituent sections of the system. The first one corresponds to the mechanical section (motor-generator coupling), whilst the second one is related to the software developed for interface purposes, which allows interaction with the user and the corresponding motor speed control. Such sections are described below.
1) Electromechanical Section. Most of the wind emulators based on motor-generator have the particularity of using a DC motor that emulates the wind turbine, being the input current one of the parameters to be manipulated to control the wind speed. On the other hand, the generator used is characterized by being an induction generator, also called a permanent magnet generator, as can be seen in the literature consulted [10-11]. The selection of such a generator is directly related to the wind turbine -or turbines- to be emulated. For example, when the electric generator is induction or permanent magnet motor and the wind turbine to be emulated is of low speed, it is necessary to either have a generator with a high number of poles or not to exceed it above the rated revolutions per minute. Taking these aspects into consideration, section 3 presents the generator used as well as the reason for its use, which will allow the definition of the prototype developed.
2) Software section. In wind emulators, the associated software seeks -above all- to allow the compilation of the characteristic parameters of the turbine’s mathematical model so that it can be simulated and allow a particular output power value to be obtained as a result. Such a result takes into account the wind speed recorded in the wind profiles, the maximum power curve of the generator, the turbine model and any variable that affects the value of such power [12].
Wind turbine emulators are considered highly dependent on the equipment developed for their implementation. Hence, limits and restrictions are established from power and type of turbine to be emulated. Consequently, this generates the use of a certain speed control strategy. Finally, the type of computer tool used will determine the ease of use and scope of action of the application, as well as its interaction with the user from an appropriate interface.
B. Emulator Design
Taking into account the characteristics that wind turbine emulators based on a motor-generator must-have, an emulator was developed consisting of an induction motor which acts as a wind turbine, a permanent magnet generator with an encoder, and a 3 HP frequency converter to control the rotation speed of the motor. The data is entered through a program developed in LabView and the communication between the electromechanical part and the software is done through a National Instruments data acquisition card (NI USB-6008, 12 Bits), as can be seen in Fig. 2.
1) Equipment Selection. An Induction motor, used as a wind turbine, will have to act as a wind turbine that constantly changes speed when the wind speed increases or decreases. The selected motor was a WEG three-phase squirrel-cage induction motor, 2 HP (1.5 kW),1800 rpm, 220/240.
Variable speed drive, which is in charge of the motor speed control. The variable speed drive chosen was the WEG CFW 500. It also offers a vector or scalar control option.
Permanent magnet synchronous generator, 800 W. The great advantage of this type of generator is the elimination of brushes and reduction systems, which translates into savings in maintenance and space. The chosen generator has a speed encoder that plays an important role in designing the control system.
NI 6008 USB, National Instruments data acquisition card, which allows acquiring values of real variables to digitize them and generate control actions through a program developed in LabView.
2) Motor-generator Characterization. Once the electromagnetic components have been selected, their functioning must be verified when they are operating in order to recognize the parameters that allow establishing direct relationships with the operation of wind turbines and wind generators. To characterize the motor-generator group, the motor speed is controlled utilizing the CFW500 variable speed drive, changing the frequency in small proportions (0.5 Hz), from 3Hz to 65 Hz. Subsequently, current and voltage measurements are made to know the working power reached by the generator, as this depends specifically on the system frequency and the speed at which the induction motor rotates.
The data recorded in the test show the operation of the equipment when it works feeding a load of 10 Ω, as presented in the results in Fig. 3.
3) Turbine to be Emulated. Taking into account that the selected generator reaches a power of 800 W, the emulator will only be able to emulate turbines with powers equal to or less than the power defined above.
Consequently, within the turbines that were studied, and to select the one which is best adapted to the emulation process, the following were found: Air 30, Air 40, Wattnier 1kW, Bornay 600 and Eolos of 600 W and 1000 W respectively.
The chosen turbine was the Eolos 600 W, whose power vs wind speed curve is shown in Fig. 4.
Such a turbine to be emulated was selected, among other things, because of the speed range it offers power generation (between 3 m/s and 14 m/s), being this the range in which wind turbines with architectural integration as well as and wind turbines operate. In addition to this, it was selected because the generator of this equipment is very similar in its construction to that of the generator used for the emulator.
4) Relationship Between Frequency and Wind Speed Depending on Power. With the data obtained from the characterization of the motor-generator and the power data provided by the turbine manufacturer (Table 1), an equation that relates wind speed with the power generated is obtained through polynomial regression.
Power [W] | Wind Speed [m/s] | RPM Emulator | Frequency [Hz] |
25 | 4 | 341 | 12 |
40 | 5 | 439 | 15 |
90 | 6 | 673 | 23 |
110 | 7 | 722 | 25 |
145 | 8 | 840 | 29 |
190 | 9 | 982 | 33.5 |
280 | 10 | 1202 | 41 |
350 | 11 | 1361 | 46.5 |
470 | 12 | 1627 | 55.5 |
525 | 13 | 1730 | 59 |
600 | 14 | 1928 | 66 |
530 | 15 | 1730 | 59 |
490 | 16 | 1657 | 57 |
This is a sixth-order polynomial regression and is presented in equation (1).
Where X represents the wind speed and P the generated power.
As it will be explained later, equation (1) is then entered into the program developed in Labview so that it has - as input parameters - the wind speed values entered manually or preloaded as a wind profile. This has to be done in order to obtain a desired output power value and thus emulate the behaviour of the selected turbine. It should be noted that selecting another turbine implies the regeneration of an equation that allows the correlation stated above.
5) Control Implementation. The developed software aims to allow users to enter a wind profile to see what the behavior of the selected turbine maybe, working under the wind speed conditions described in such a profile.
The selected software, as mentioned above, was Labview by National Instruments, which manages a graphic programming environment and has facilities for handling different communication technologies. Another important feature that influenced the selection of LabView was the possibility of performing a data acquisition process using a National Instruments card.
Because the permanent magnet generator has an encoder whose signal is sent to the data acquisition card - together with the generated voltage and current values - it is possible, through mathematical processing, to establish the speed and therefore the revolutions per minute at which the wind emulator is rotating.
The interface allows the user to enter the wind speed they want to associate. Therefore, the programme needs to be able to enter different wind speeds in a period. This is achieved through a matrix linked to a case cycle that defines the working time of the entered wind profile simulation. With the average signal and the signal entered by the user, the error is calculated and entered into a PID block. The constants of such a controller were tuned in LabView from the experimental response of the emulator using the Ziegler-Nichols method. This allows the variable of interest (speed) to be controlled, which is used subsequently in equation (1) to determine a power output similar to that of the previously selected wind turbine.
The final scheme of the emulator designed and presented in this article can be seen in Fig. 5
III. RESULTS AND DISCUSSION
The wind turbine emulator developed is shown in Fig. 6 and can be separated into three structural parts, an electromechanical part, one corresponding to the data acquisition circuits and another one to the software.
Electromechanical parts of the emulator:
Permanent magnet generator
WEG three-phase induction motor
3HP WEG three-phase inverter
Motor generator coupling
Breaker 25 A
Fuses
Power connector
Data acquisition parts:
NI 6008 USB DAQ data acquisition card
ASC712 20 A Current sensor
USB DAQ-PC cable and non-inverting amplifier circuit
AC 120 / 14-5 V Transformer
Charge connection point
Emulator software:
The window for entering the wind profile.
The window for monitoring the behavior and activity of the emulator.
The emulator presents two operating options: The first one is one in which it is possible to load different speed data into the developed software. Such data are associated with a particular power that takes into account the selected wind profile. The second operation option is manual control of the wind speed, which depends on the frequency.
Fig. 7 and Fig. 8 show the two interface screens. In the former one, the wind profile is entered, whereas, in the latter one, the operation of the emulator can be seen.
When executing the program, the emulator starts operating at the minimum speed (3 m/s, approximately), which is obtained for the frequency of 3 Hz. At that particular moment, the graph of the loaded wind profile can be seen in the window presented in Fig. 7, as well as the option of the simulation time which can be of 1 minute, 10 minutes, 1 hour or in real-time.
Fig. 9 shows the variables of interest measured in the emulator. These are as follows: speed, operation frequency, voltage, current and generated power. Additionally, there is an option to enable or disable the manual operation to control the rotation speed depending on the frequency. Additionally, it also displays the time at which the simulation is being performed.
This part of the program also shows the power curve which is being generated. On the right-hand side of such graph, two curves can be seen when the emulator is in operation, one being the setpoint, whereas the other one is the curve showing how the emulator is working.
A. Manual test: Comparison with the Curve Provided by the Manufacturer of the Chosen Turbine Eolos 600 W
This first test consists of comparing the information provided by the manufacturer of the Eolos 600 W turbine with the data given by the emulator for the same wind speeds.
From the software, when enabling the manual option, the power values can be observed. Such values were obtained when entering the frequency corresponding to the wind values recorded in the power table given by the manufacturer of the Eolos 600 W turbine. Likewise, the power at the output of the prototype is measured. Fig. 9 shows an image of the graphic interface of the emulator whilst the previously described process was being conducted. Additionally, the power curves (black), the setpoint (blue) which changes as the selected frequency varies, as well as the behavior of the emulator (red) can also be seen.
Table 2 shows a comparison of the data provided by the manufacturer and those generated by the emulator.
Frequency (Hz) | Wind speed (m/s) | Turbine P. (W) | Emulator P. (W) |
12 | 4 | 25 | 25.8 |
15 | 5 | 40 | 35 |
23 | 6 | 90 | 94.3 |
25 | 7 | 110 | 112.5 |
29 | 8 | 145 | 149.35 |
33.5 | 9 | 190 | 195.69 |
41 | 10 | 280 | 291.1 |
46.5 | 11 | 350 | 349.97 |
55.5 | 12 | 470 | 492.9 |
59 | 13 | 525 | 548.89 |
66 | 14 | 600 | 619.5 |
59 | 15 | 530 | 544.5 |
57 | 16 | 490 | 508.25 |
With the obtained results in the tests reported in Table 2, the error was calculated and is presented in Table 3.
Turbine P. [W] | Emulator P. [W] | % Error |
25 | 25.8 | 3% |
40 | 41.3 | 3% |
90 | 94.3 | 5% |
110 | 112.5 | 2% |
145 | 149.35 | 3% |
190 | 195.69 | 3% |
280 | 291.1 | 4% |
350 | 349.97 | 0% |
470 | 492.9 | 5% |
525 | 548.89 | 5% |
600 | 619.5 | 3% |
530 | 544.5 | 3% |
490 | 508.25 | 4% |
Testing the emulator manually showed that, on average, it has an error margin of less than 6%, one of the lowest error rates obtained throughout the development of the emulator.
Now, that the program test in manual mode has been completed, the next step is to verify the operation of the automatic mode. In such function, a 24-hour wind profile - which had been previously established - is loaded in one of the windows of the graphic interface implemented in LabView. Subsequently, a test is presented in which the wind characteristics consigned in a wind profile are used. Such characteristics are taken from a specific point of the Colombian geography.
B. Wind Profile: A Case Study in Colombia
One of the options to test the operation of the emulator is to load a wind profile from a specific place for watching its behavior for different wind speeds at different times. It should be noted that speeds below 3 m/s do not allow an adequate operation of the emulator. This is because the power generated is negligible.
Consequently, and to investigate new alternative energy sources that could help to expand the energy matrix in Colombia, the department of Guajira was chosen to select a geographic location whose environmental conditions were ideal for the exploitation of wind power. The selected location is the municipality of Uribía in the department of La Guajira, which is recognized for hosting the only existing wind farm in Colombia. The coordinates of this location are as follows: latitude 12.5, longitude -72.5.
To enter the wind profile into the program, data with the coordinates of the place of interest were extracted from NASA’s online database called Atmospheric Science Data Centre. The monthly average delivered by the atmospheric service is repeated so that it can be adjusted to the input times given in the Labview interface. The extracted data can be seen in Table 4. In addition to this, such data were entered as seen in Table 5.
Lat. 12.22 | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec | Annual Avg. |
Lon. 72.15 | |||||||||||||
Avg. 10 years | 8.94 | 9.27 | 9.02 | 7.46 | 6.43 | 7.56 | 7.86 | 6.67 | 5.36 | 5.11 | 6.13 | 8.08 | 7.31 |
Hour | Wind Speed [m/s] |
00:00 | 8.94 |
01:00 | 9.27 |
02:00 | 9.02 |
03:00 | 7.46 |
04:00 | 6.46 |
05:00 | 7.56 |
06:00 | 7.86 |
07:00 | 6.67 |
08:00 | 5.36 |
09:00 | 5.11 |
10:00 | 6.13 |
11:00 | 8.08 |
12:00 | 8.94 |
13:00 | 9.27 |
14:00 | 9.02 |
15:00 | 7.46 |
16:00 | 6.46 |
17:00 | 7.56 |
18:00 | 7.86 |
19:00 | 6.67 |
20:00 | 5.36 |
21:00 | 5.11 |
22:00 | 6.13 |
23:00 | 8.08 |
The profile is loaded into the program at the moment the simulation starts. Furthermore, the curve describing the profile can be seen in the first screen of the graphic interface, as shown in Fig. 10, in which the curve can be seen as well as the wind profile data located in the matrix placed on the left side.
As previously described, the simulation times can be changed. For instance, the wind profile with data from 24 hours of wind resource measurement can be reproduced in time of 1 minute, 10 minutes, one hour and real-time respectively. For this test, the obtained results in the simulations in one minute were recorded and such behavior can be seen in Fig. 11. Moreover, the 10-minute simulation time can be seen in Fig. 12.
It can be seen how, on average, 100 W was generated during the test when loading the wind profile originating from the Guajira region. Furthermore, it can also be seen how the emulator neither manages to stabilize nor to equal the behavior of the loaded wind profile when simulating in a time of one minute - at least in considerable increases of the wind speed -, due to the speed with which it changes.
On the other hand, in the test with a simulation time of 10 minutes, the program has more time to do all the processing, which causes the curve showing the behavior of the emulator to look similar to that of the loaded wind profile, as seen in Fig. 12.
IV. CONCLUSIONS
An emulator of a low power wind turbine was designed and implemented through the coupling of a three-phase induction motor and a permanent magnet generator. Such an emulator has a graphic interface developed in Labview which records its activity. Furthermore, it is in the graphic interface where the generated power can be visualized. Additionally, it allows the entry of wind profiles (wind speed vs time) which are processed together with the signals captured by the encoder of the permanent magnet generator to determine the error. Subsequently, such process allows its later entry to a PID block, whose constants were tuned through Ziegler-Nichols, in LabView - on the experimental response of the system - to control the variable of interest (speed) as well as determine the output power to be generated by the emulator, through its subsequent mathematical processing.
The system presents two operating options. One in which it is possible to load different speed data which are associated with a particular power taking into account the selected wind profile, whereas the second option of operation is manual control of the wind speed.
The test of the emulator in a manual way showed that on average, the emulator has an error margin less than 6%, compared to the data provided by the turbine manufacturer, which makes the emulator a reliable element to operate in a controlled laboratory environment.