Even with all the advances in computer modeling, actual aerodynamics tests in wind tunnels remain indispensable for industries such as air and space travel or automotive design and construction. The measurements that wind tunnels allow are extremely cost-intensive; only a rare few companies can afford their own wind tunnel. Renting one, including the necessary engineering teams, for a single day can run up to €80,000 (approximately $107,000), on top of which come the costs for preparing the model.
For all these reasons, time in a wind tunnel should always be used in the most efficient manner possible. The measurement itself often lasts only a fraction of the total usage time — the majority is used for the preparation of the experiment and the subsequent processing of the data. Data processing times are particularly inefficient. Depending on the data volume and available computing power, it can take several hours, which translates to holding time, during which the installation and highly qualified personnel are paid to wait. To avoid holding time, image data are often first stored and analyzed later, overnight for example. Should it become evident that the data are not evaluable for any reason, the entire measurement must be repeated.
The FIBUS Research Institute for Image Analysis, Environmental Control & Flow Mechanics in Hamburg develops measurement methods and software solutions to accelerate wind tunnel experiments with an equal level of precision, thereby saving costs. Primary application areas are the analysis of vibration, torsion, and oscillation of the test object – such as a wing – under the influence of flow velocity during flight or the measurement of airflow and turbulence surrounding the object. In both instances, high-speed cameras are used to detect even the tiniest changes within the shortest time intervals, as well as to capture high-frequency oscillations.
Air current analysis with Particle Image Velocimetry
For classic aerodynamic measurement, such as the analysis of airflow and turbulence, the so-called “Particle Image Velocimetry” (PIV) method is used. Airflow is made visible by adding microparticles to the flow. In most cases, finely atomized oil or similar special fluids are used whose microparticles will remain suspended in air. Special imaging systems capture images of these particles within the airflow and calculate their trajectory using correlation analysis (cross-correlation between two consecutive images). From this process, a precise model arises of the flows and turbulence, such as those along an aircraft wing or a helicopter rotor.
During Particle Image Velocimetry of turbulent or whirling flows, the key is to capture the flow with the highest image rate possible, keeping the interval between images as short as possible. Only in this manner can the position of correlated particles in consecutive images be easily detected, and the exactness of the measurement can be maintained, even at high flow velocities. In this manner, both the particle direction and its velocity as a function of the interval between positions and of the time between images can be calculated.
Two rapidly consecutive images of an airflow marked with oil particles as raw material for PIV analysis
PIV can be measured either two- or three-dimensionally by using one or two cameras respectively. In the case of three-dimensional measurement, the particle’s position in space is determined using triangulation.
The FIBUS Institute developed special software for Particle Image Velocimetry, picCOLOR, designed for the analysis of high-resolution images in real time. It can be run on up to 32 processors at a time.
picCOLOR software’s underlying principle is not to follow each individual particle but rather to measure the changes in the image patterns between images 1 and 2. In so doing, computing power is optimally utilized and the computing process is substantially accelerated. To still achieve sufficient precision, the entire image is divided into small square analysis windows measuring 32x32 pixels by default, though these dimensions can be adjusted as needed for an application. The image analysis occurs in two steps:
- In the first step, pixel displacement between image 1 and image 2 within the same window is measured. However, particles that were located along the edge of the window in the first image may have exited the window in the second image.
- Thus, in the second step, the analysis window is displaced along a vector that arises from the general displacement trend of step 1. The pixel values within the displaced window are once again compared between images 1 and 2.
From both steps, picCOLOR computes a vectorial representation of the flows for all analysis windows in the image. The vectors represent the path and velocity of the particles. If the analysis is complete, the particles’ coordinates and the computed vectors can be stored without image data, thereby substantially reducing the data volume.
Previously, most PIV measurements were conducted on quasi-stationary flows whereby the course of the flow remained nearly unchanged over time, since the commonly used cooled, low-noise PIV cameras had very low frame rates. More recently, however, more and more measurement tasks are occurring in which non-steady flows that change unpredictably over time, shall be examined. In such cases it is important to generate PIV images in much faster sequence. This is known as “time-resolved PIV”.
For such PIV applications, FIBUS recommends the Bonito CL-400 Kamera from Allied Vision Technologies. The Bonito is equipped with a 4-megapixel Global Shutter CMOS sensor and delivers nearly 400 images per second at full resolution, thanks to its double 10-tap Camera Link Full+ interface.
Furthermore, Bonito makes use of a special PIV mode in which the camera generates not just one, but two consecutive images for PIV analysis. To minimize the time interval between the two images, they are generated without a shutter. The image recording is determined using an illumination pulse that is synchronized with the camera. The illumination is usually a laser light-section through which the particles flow. By eliminating the electronic shutter, the time interval between images can be reduced to approximately 550 ns, providing the advantage that the particles’ path and velocity can be detected even more exactly.
Time is money: streaming and image processing in real time saves hard cash
One further decisive advantage of AVT’s Bonito is the real-time data transmission thanks to the double 10-tap Camera Link Full+ interface, explained Dr. Reinert Müller, Managing Director of the FIBUS Institute: “Oftentimes in wind tunnels, very cost-intensive high-speed cameras are installed that reach high image rates over 1000 fps at full resolution, but they put out such an enormous data volume that their interface can’t transfer it in real time to the host computer. So the image data have to first be stored in the limited internal camera storage and then transferred to the computer after the fact at a lower data rate. Depending upon the data volume and bandwidth of the interface, this can last several minutes. Only then can the image processing even begin.”
This sequential process generates an enormous waste of time that, in wind tunnels, can be expensive. The Bonito from Allied Vision Technologies delivers “only” 400 images per second, but in many cases, the full 4 megapixel resolution is not even necessary. With corresponding ROI, very high image rates of over 1,000 fps can be achieved. The greatest advantage to this solution, however, is that image rates with the Bonito can be transferred and processed in real time – a substantial difference. “The use time saved on the wind tunnel and the well paid engineers’ wait time is worth a lot of hard cash! Add to that the fact that the Bonito is markedly more affordable in comparison to the usual high-speed cameras,” Dr. Müller said.
In the wind tunnel, a lot of time and money can be saved using real-time imaging. The clever use of ROIs, the FIBUS Institute’s intelligent image analysis algorithms und AVT’s Bonito cameras with integrated PIV mode make it possible.