In this paper, the differences between a performance analysis of a wind turbine by computational simulations and by experimental methods such as tests in the wind tunnel ”GroWiKa” belonging to the TU Berlin are analysed. The qualitative and quantitative aspects of both the rotor of the turbine and its electric generator are analyzed under specific experimental methods whose resulting graphs are compared with simulation software such as Qblade for rotor performance and Open AFPM for electric generator performance. Finally, from this comparison, an analysis is derived by which the accuracy of the information obtained by the used software is validated.
Due to the recent climate crisis and new trends regarding the development of energy production, wind energy has become one of the most used solutions in the field of renewable energies. This technology offers efficiencies and performance even beyond energy production systems such as solar energy and even internal combustion engines based on biodiesel. Another convenience of wind energy is the fact that it has a wide potential for places even hard to reach by other technologies. But, just as it has a great number of advantages, it also has disadvantages because it is a technology that is still under development and its mode of implementation depends on a great number of variables such as mechanical, electrical and climate factors that must be taken into account when developing the different types of projects.
Abstract
In this paper the differences between a performance analysis of a wind turbine by computational simulations and by experimental methods such as tests in the wind tunnel ”GroWiKa” belonging to the TU Berlin are analysed. The qualitative and quantitative aspects of both the rotor of the turbine and its electric generator are analyzed under specific experimental methods whose resulting graphs are compared with simulation software such as Qblade for rotor performance and Open AFPM for electric generator performance. Finally, from this comparison an analysis is derived by which the accuracy of the information obtained by the used software is validated.
Keywords: Wind Energy, Wind tunnel, Simulation, Measurements
1. Introduction
Due to the recent climate crisis and new trends regarding the development of energy production, wind energy has become one of the most used solutions in the field of renewable energies. This technology offers efficiencies and performance even beyond energy production systems such as solar energy 1 and even internal combustion engines based on biodiesel 2. Another convenience of wind energy is the fact that it has a wide potential for places even hard to reach by other technologies 3. But, just as it has a great number of advantages, it also has disadvantages because it is a technology that is still under development and its mode of implementation depends on a great number of variables such as mechanical, electrical and climate factors that must be taken into account when developing the different types of projects.
For this reason, this field of energy production has started to be helped a lot by different technological tools such as simulation software. Different computer programs have been developed hand in hand with this technology in order to help the development of projects by reducing costs in both machinery and time. The simulation capacity of the different software that is offered nowadays in the market depends a lot on needs of the users. It should be noted that all these tools have a high cost. Therefore, open license software has been developed by different creators such as Open AFPM for the simulation of electric generators and even TU Berlin has developed its own software for the simulation of wind turbines and rotors. These open-license software programs allow students and interested people a faster introduction and incursion into this topic, thus achieving a faster development of knowledge and collaboration also in the development of this type of technology. It is important to clarify that these software do not have such a high accuracy compared to commercial solutions used by large companies in the market, but they offer a solid base for its understanding.
The aim of this paper is to compare the results obtained from the measurement and simulation of the rotor characteristics of handcrafted small wind turbine. On one hand, the simulations were executed with the software Qblade and Open AFPM, and count on various theoretical assumptions that might affect our final result. On the other hand, the measurements were conducted at the GroWiKa, short form for ’großer Windkanal” (big windtunnel). They were made under certain ambiental conditions and using measurement equipment that will be described in the following sections.
2. Simulation
To carry out the simulation, a rotor developed by the TU of Berlin was taken as a model. This rotor has been constructed manually and its characteristics are presented in the following table:
Abbildung in dieser Leseprobe nicht enthalten
Table 1. Rotor Characteristics.
2.1. Rotor characteristic curve
To simulate the operational conditions and obtain the rotor characteristics the software Qblade was used. It is important to take into account following considerations when simulating with Qblade:
- Reynolds number iteration: the Reynolds number is an important dimensionless quantity in fluid mechanics used to help predict flow patterns in different fluid flow situations. It is the ratio of inertial forces to viscous forces within a fluid which is subjected to relative internal movement due to different fluid velocities 4:
Abbildung in dieser Leseprobe nicht enthalten
where p is the density of the fluid (m), u is the flow speed (mm), L is a characteristic linear dimension (m), p is the dynamic viscosity of the fluid (mg) and v is the kinematic viscosity of the fluid (g-). As we can see in the definition and considering our measurement data from ”GroWiKa” it is not possible to calculate accurately the Reynolds number. In this case, and starting with a supposed Reynolds Number of Re = 106, an iteration must be executed until acceptable realistic results of the parameter A Re, which represents the discrepancy between simulation and reality (A Re < 2 • 106), are obtained 5. As we observe in the Figure 1, after the first iteration, which give us already an acceptable result, we observe a decrease in the Power coefficient c P of the rotor:
Abbildung in dieser Leseprobe nicht enthalten
Figure 1. Power coefficient - Tip Speed Ratio diagramm for a wind speed of 13,5 ms for the initial and the iterated Reynolds number
We observe how the modification of the Reynolds number has an influence on the Power Coeficcient cP, and we conclude that as close as we get to the real Reynolds number our power coefficient will keep decreasing. - Aerodynamic airfoil analysis: Qblade uses the XFoil Analysis Software for the design and analysis of subsonic isolated airfoils. Given the coordinates specifying the shape of a 2D airfoil, Reynolds and Mach numbers, XFoil can calculate the pressure distribution on the airfoil and hence lift and drag characteristics. 6 XFoil Analysis provides acceptable results but has several limitations as well that can affect the characteristic curves obtained from simulation.
2.2. Generator characteristic curve
The generator simulation tool OpenAFPM was used to simulate the generator characteristics. For the simulation the generator is first drawn in Finite Element Method Magnetics (FEMM), flux density on magnet surface calculated then turned at rated RPM for no load and rated current. This is an iterative process, where variables are initialized and then updated every iteration. This process is continued until it converges to a single solution 7. Some of the data was obtained through the test stand manual, while the rest were either measured, assumed or even suggested by the simulation software itself. This was due to the difficulty in measuring some values such as winding fill factor without opening up the rotor disk. The input parameters that were used and its sources has been attached to the appendix. The influence that a few of these parameters had on the simulation output namely the generator power and mechanical torque can be found in Table 2.
Abbildung in dieser Leseprobe nicht enthalten
Figure 2. Input values for simulation and sources
Abbildung in dieser Leseprobe nicht enthalten
Figure 3. The influence of stator thickness on generated Power for rated current
Abbildung in dieser Leseprobe nicht enthalten
Figure 4. The influence of mechanical clearance on Mechanical Torque for rated current
As seen in Table 2 it was observed that the generator stator inputs had the biggest influence on the simulation results. A difference of 0,5mm in the mechanical clearance for example led to a 3,54% increase of Output Power for rated current. This had a huge influence on the outcome as these values were measured using a ruler in the test stand and had a large margin of error. Certain inputs e.g. stator nominal temperature, had little to no influence on the generator power or the mechanical torque
3. Measurements
The model used for the tests was developed by the fluid dynamics department of the TU Berlin. As described in Table 1, the rotor is made of European large wood and has been manufactured by handwork.
Abbildung in dieser Leseprobe nicht enthalten
Figure 5. Rotor of the Wind Turbine Model
3.1. Rotor
The measurements were conducted at the GroWiKa, short form for ’’großer Windkanal” (big wind-tunnel). An incremental encoder-sensor and a ”Bosch BMP085” sensor were placed along the tunnel to measure absolute pressure and enviromental temperature. To measure the pressure difference, a Baratron sensor with two pitot tubes are set up in each side of the wind tunnel’s wall facing the wind flow. One pitot tube is used to measure the static pressure and another is to measure the total pressure. The Power coefficient cP curves have been obtained from the measured data in order to be compared with the curves obtained with the simulation tools. It is important to mention few aspects from the ”GroWiKa” that might have prevented us from obtaining more accurate results:
- Sensor errors: the sensor ”Bosch BMP085” has an accuracy of ± 1 hPa for the measurement of the pressure and ± 1 ° 8.
- Handcrafted rotor: as it was a rotor manufactured by manual procedures, its geometrie, which is given by the manufacturer [], might have small variations that will affect the final result. Also, as it was checked by the team members during the measurement campaign, the trailing edge of the three blades was slighly damaged.
3.2. Generator
For the generator test stand two sensors were used namely
- Lorenz DR-2212 Sensor : which measured the Rotational speed (RPM) and Torque at the shaft connected between the motor and the generator
- Danisense DS200UBSA-10V Fluxgate transducer : was used to measure the alternating current (AC) as well as the direct current (DC) after rectification
A voltage divider was also used to prevent overload of the data acquisition (DAQ) device. The frequency of the motor which corresponds to a certain motor RPM was set using a frequency inverter. Careful consideration was taken not to measure in the range of natural frequency of the stand to avoid excessive vibration of the test stand. First the open circuit voltage was measured followed by measurements with various resistors to represent the load. The test matrix can be found attached in the appendix. The measurements were recorded at short bursts of 1-2 seconds. The results were recorded with the LABView program, in voltage signals. These were then converted into the required values by using the post processing program prepared by the generator group.
4. Comparison
4.1. Rotor
The graphs obtained by the Qblade simulator were generated using the BEM (Blade Element method) procedure. For the development of this simulation, the following input parameters were taken into account, as shown in the following table
Abbildung in dieser Leseprobe nicht enthalten
Table 2. Input parameters.
On the other hand, to make a closer adjustment to reality, the software allows to simulate certain losses such as Prandtl tip losses, Prandtl root losses, 3D corrections, Reynolds Drag correction and Foil interpolation. These options were taken into account when simulating the rotor.
For the results of the simulation the computer program allows the use of different graphs according to the final result to be evaluated. In this case we obtained the coefficient power (Cp) graphs according to the number of revolutions per minute.
Abbildung in dieser Leseprobe nicht enthalten
Figure 6. Graphic of Coefficient power Vs RPM
With this graph it is possible to observe the performance of the rotor depending on the wind speed. It is clear that the best results are obtained for the nominal rotation speed. It can also be seen that despite being a hand-built wooden rotor, it shows acceptable performance results.
In other hand, the following figure shows the Power coefficient cP curves obtained from the measured data:
Abbildung in dieser Leseprobe nicht enthalten
Figure 7. Power coefficient - Fan rotational speed diagramm for different values of wind speed
Compared to the graph obtained by simulation, this graph 7 shows more dispersed measurement points by taking data in an experimental way. The similarity of performance in the two power coefficients and in the revolutions per minute can be observed.
4.2. Generator
During the post processing we generated lot of data with the help of LABVIEW software. Following are the remarkable graphs obtained from generated data.
Abbildung in dieser Leseprobe nicht enthalten
Figure 8. Mechanical Torque vs RPM (at fixed loads)
With the help of this graph we can observe variation of mechanical torque with respect to RPM at certain load conditions. Maximum torque accrued on generator shaft is nearly 9Nm at 500rpm and 0.8 ohm load. For the Open Circuit scenario a Torque of 0,77Nm at 700,40 RPM was measured. This was in close range to the simulated value of 0,80 Nm at 687,90 RPM. The form factor of the measurement results also closely resembled that of the simulation in Figure 9. It should be noted that the sharp decrease for the speeds lower than 100 RPM werent even measured in the test stand to begin with. With the exception of data points between 500 RPM and 600 RPM the data measured showed expected trends. Eventhough its difficult to notice in the simulation graph, the trend is a slight increase of Torque with increasing speeds. The causes for the difference in expected and measured values will be discussed further in detail in 4.2.1.
[...]
- Quote paper
- Shivaraj Patil (Author), 2019, Comparisons between simulation and measurements of a handcrafted small wind turbine, Munich, GRIN Verlag, https://www.hausarbeiten.de/document/981428