Ground Vibration Testing Master Class: modern ...

40 downloads 266 Views 1MB Size Report
Through a Small Business Innovative Research (SBIR) project, CSA Engineering worked with AFFTC to evaluate a new way of instrumentation using ...
Ground Vibration Testing Master Class: modern testing and analysis concepts applied to an F-16 aircraft

Jenny Lau1, Bart Peeters1, Jan Debille1, Quentin Guzek1, Willam Flynn1, Donald S. Lange2, Timo Kahlmann3 1

LMS international, Interleuvenlaan, 68, B-3001 Leuven, Belgium, [email protected]

2

Air Force Flight Test Center, 307 East Popson Ave., Edward AFB CA 93524, U.S.A.

3

AICON 3D Systems GmbH, Biberweg 30 C, 38114 Braunschweig, Germany

ABSTRACT Ground Vibration Testing (GVT) of aircraft is performed very late in the development process. The main purpose of the test is to obtain experimental vibration data of the whole aircraft structure for validating and improving its structural dynamic models. Among other things, these models are used to predict the flutter behavior and carefully plan the safety-critical in-flight flutter tests. Due to the limited availability of the aircraft for a GVT and the fact that multiple configurations need to be tested, an extreme time pressure exists in getting the test results efficiently. The aim of the paper is to discuss modern testing and analysis concepts for performing a GVT that are able to help realize an important testing and analysis time reduction without compromising the accuracy of the results. For the past several years, LMS has organized so-called Master Classes on the GVT topic. The aim of the class is to introduce an integrated approach to handle the test preparation, modal testing, modal analysis, numerical model correlation, model updating, and model exploitation, to the industry by means of using a full-scale aircraft. This paper illustrates this approach by presenting testing and analysis results of an F-16 aircraft.

Introduction Over more than 80 years [1, 4, 5, 6, 7], methodology and technology used to understand and provide insight into the dynamic characteristics of an aircraft have significantly evolved. Ground vibration testing on a prototype of a new aircraft is often considered to be one important step used to describe the structural dynamics of the aircraft and for updating the finite element model. As a result, greater confidence on flutter behavior can be achieved, leading to mitigation of the hazards related to this inflight testing. Testing methods have evolved from normal mode testing (also referred to as sine dwell, or phase resonance method), which identifies one resonance frequency at a time, to modern multiple input multiple output random excitation methods, where measured transfer functions are analyzed with experimental modal analysis. The challenges behind a ground vibration test are not only providing a qualified model based on high quality test data, but also providing these results within a tight schedule and with limited budgets. Various directions and improvements have already been tested and/or applied such as different excitation methods for the optimization of quality of data and test time. Instrumentation is one of the time consuming tasks during a ground vibration testing campaign since the number of locations to measure increases in proportion with the size of aircraft. For example, nearly 900 sensors are used to test the A380 [7]. Alternative techniques like digital sensors, wireless sensors, and optical measurement systems have been investigated throughout the last 5 years in order to reduce the instrumentation time [6]. At the same time, in order to easily identify sensor locations, photogrammetry is studied and tested on a A737 and considered to be satisfactory in both accuracy and cost expectation [2].

2

LMS Master Class For the past several years, LMS international has organized worldwide, several Ground Vibration Testing Master Classes. The classes were conducted by our experienced engineers in the field. Lectures are given by professors and professionals in the aircraft development field. Topics ranged from testing to simulation, from equipment to instrumentation, from process explanation, to onsite test execution. Each time, we are challenged by availability of equipment, test object, test facility, etc. The test article ranged from a wind tunnel scaled aircraft model (Fig. a) up to an F-16 (Fig. b and Fig. c). The test facilities started from military base hangar to a testing hall of an aircraft manufacturer. It was, of course, not as stressed as a real ground vibration test campaign, since we didn’t have the objective to deliver precise test data as input for finite element models and with the inherent tight time constraints. Nevertheless, due to availability of the test facility, test article, test equipment and sometimes the engineers themselves, we typically only had 2-3 days for instrumentation. Besides, test articles are different for each master class and our engineers have to be well prepared and as efficient as possible during a ground vibration test campaign, in order to extract the maximum information from the test article in the short time allowed. During the ground vibration testing master class, our engineer involved the participants on encountered difficulties and tried out solutions to quickly solve the problems. In parallel, we also get use of these fabulous opportunities to validate new theories, new concepts and new testing capabilities in order to improve the whole procedure with our research engineers on a real case.

Fig. a Wind tunnel model used during LMS Master class in Russia, 2008

Fig. b F-16 used during LMS Master Class in Belgium, 2008

Fig. c F-16 used during LMS Master class in U.S.A., 2010, source Edwards AFB No. AFFTC-PA-10268

3

Master class in the past In the summer of 2009, at one of the military bases of Belgium, Campus Saffrannberg at Sint-Truiden, we had the opportunity to perform a ground vibration test on an F-16 with a wingspand around 15m. A ground vibration test was conducted with an LMS SCADAS III frontend with 132 channels. As explained in [2, 3], identification of sensors location is one of the key challenges of a ground vibration testing campaign. During the master class, we had an opportunity to validate the use of photogrammetry techniques for identification of geometry coordinates. The photogrammetry technique, as explained in [2], implemented in AICON DPAPro solution, and is fully integrated in LMS Test.Lab Geometry. We decided to take the opportunity to validate the real-life use case of photogrammetry on a 15m by 10m aircraft.

Fig. d A picture used to process the coordinate of coded markers

Fig. e This picture shows some of the identified targets

The procedure consists of putting coded markers on the structure. By means of using coded markers the data processing is automated to a maximum degree. After this, pictures were taken in various angles and from different positions. Fig. d is one of the pictures taken with some coded markers (in black and white). One important remark is that these coded markers are standard available without using printer to print in correct format, thus speeding up the processs and more reliable in scale. Information such as camera charactersitics and scale information are introduced into the software, and pictures are imported in the AICON DPA-Pro software for processing. During the measurement process in situ, the camera was calibrated by the software. For the F-16, we placed around 300 coded markers and took 320 pictures with a Nikon D300. It took us 40 minutes to place the markers, and 30 minutes to take the pictures. We used 5 minutes for processing the pictures and 15 minutes to analyse the data. In total, less than 1 hour and 30 minutes, we obtained the geometry of cloud points directly in the Test.Lab software without any file importing or export manipulations. The accuracy is better than 2 mm for an object of 11m x 10m x 3m. In this case, we are satisfied with the result. In the future, we can also use automatic blunder detection which will guarantee even better results. Unfortunately, we instrumented the aircraft a few days prior to testing. Thus the locations that we measured are not those of the sensors. Nevertheless, we expected that the result can give us an overall image of the aircraft. Fig. e shows some coded markers (identified by yellow number labels) identified by one of the pictures. Fig. f shows the calculated coordindate imported into LMS Test.lab Geometry. These cloud points are named with the number of the coded markers. With less than 2 hours work, we obtained very good data accuracy with a relatively low cost solution. As usual, in the modal analysis context, we use the geometry to animate the mode shape in order to check the validity of mode shape. First of all, we performed random excitation and measured the transfer functions of 132 degrees of freedom, experimental modal analysis was performed with LMS PolyMAX solver. Although, these points didn’t correspond to

4

sensors location, we have superimposed the test geometry with the cloud points in Test.Lab Geometry workbook such that we can use the automatic mode shape expansion to visualize more easily mode shapes in a denser geometry. The advantage of automatic mode shape expansion is that there is no need to describe the relationship between the degree of freedom of the test data and the cloud points. LMS Test.lab will determine, based on a geometrical based algorithm, the closer test points to the corresponding cloud points in order to animate the unmeasured cloud points, thus generating “slave degrees of freedom”. Fig. g shows the first symmetrical mode animated with automatic mode shape expansion. The lines are connected between the measured degree of freedom and the cooresponding points resulting from photogrammetry. The automatic mode shape expansion technique can also be used on geometry defined by CAD or finite element model.

Fig. f Cloud points of the coded target directly imported in Test.Lab Fig. g Symmetric mode animated with automatic mode shape expansion

Master class in USA In May 2010, a ground vibration testing master class has been organized at the Air Force Flight Test Center (AFFTC) in Edwards, California, U.S.A. The test article was an F-16. Two current control electrodynamic shakers of MB Dynamics (Fig. h) are provided and instrumented by AFFTC personnel. The shakers are located at the wing tips and excited in the Zdirection (vertical direction). A 60 channel SCADAS III system was used to measure 2 force cells, 47 laser sensors in velocity, 10 acceleration sensors and send out 2 excitation signals to shakers. The laser sensors are measured in the Z-direction at the wing and horizontal tail and in the Y-direction at the vertical tail. Laser sensors in both the Z- and Y-directions measured the fuselage dynamics. The ten accelerometers are located at the left wing at the same locations as laser sensors in order to compare velocity data to acceleration data (Fig. i).

Velocity measurements Through a Small Business Innovative Research (SBIR) project, CSA Engineering worked with AFFTC to evaluate a new way of instrumentation using non-contact laser sensors (Fig. j). These small lasers are mounted to a frame and that can be

5

built well in advance of the test. When the test article is available, engineers simply position the sensors frame to the test article before the ground vibration test and do a quick check on the distance between the measured surface and the laser sensor. Fig. k shows the setup of a laser sensor on the frame. The main advantage is that most of the set up work is accomplished before the aircraft is available. Another advantage results when an aircraft configuration needs to be made such as fueling or defueling. Typically, the aircraft instrumentation is disconnected from the aircraft and the aircraft is towed outside the test hangar and is fueled. When the aircraft is back in the hangar, again the sensor stands are simply re-positioned and testing resumes with minimal time expended. A separate technical paper discussing these sensors has been written and will be presented at IMAC 2011.

Fig. h Shaker instrumented in the wing tip, source Edwards AFB No. AFFTC-PA-10268

Fig. i Laser sensor and the corresponding accelerometer instrumented in Left wing, source Edwards AFB No. AFFTC-PA-10268 For the GVT Master class, the setup of the test object was mostly coordinated by AFFTC since the test will mainly be conducted with these small lasers. Before the ground vibration testing master class, AFFTC had already evaluated the results measured with the CSA laser system with their own LMS Ground vibration testing system. The measured quantity of the laser is “velocity”, whereas quite often accelerometers are used instead. Nevertheless, experimental modal analysis is conducted without any problem in Test.Lab modal analysis even when acceleration and velocity data are combined in the data set.

Excitation methods and analysis tools During the master class, several excitation methods, such as multiple input multiple output random excitation, multiple input multiple output sweep sine excitation, multiple input multiple output stepped sine excitation and multiple input multiple output normal mode testing, are used to illustrate the difference between each other. Random excitation measured up to 64 Hz bandwidth and with 32 % overlap. It provides a quick method to identify on a wide frequency range the overall dynamic behavior of the structure. Fig. l and Fig. m illustrate the driving point FRFs and coherence, which shows that the reciprocity between driving points is achieved and the response at the driving points are noisy after 40 Hz. Fig. n compares the FRF measured at one of point of the left wing in Z direction; the red curve represents the measurement of laser in velocity/force format in -Z direction and green curve represents the measurement of accelerometerin acceleration/force format in +Z direction. The green curve is integrated in velocity/force for comparison. The amplitude of the measurement corresponds quite well whereas the velocity data seems to be a bit noisier than the

6

acceleration data. We also observed the phase difference which can easily explained by +/- Z direction of the sensors. Overall, below 40 Hz, velocity and acceleration data correspond quite well with each other. Stepped sine excitation are used to illustrate excitation around certain resonance frequency with different excitation levels in order to identify non-linearity. During the master class, sine excitation is conducted from 5 to 25 Hz with 0.125 Hz resolution. The test took 19 minutes to achieve 2 linear sweeps. Compared to random excitation, the measurement time is more important but this method allows higher excitation level. Another alternative is to use sweep sine excitation which will provide a similar effect but shorter transient time is applied between each sine spectrum. It’s quite often used today to replace random excitation to achieve higher excitation level and signal to noise ratio with less measurement time than stepped sine excitation.

Fig. j Small laser sensors are attached in a frame for velocity measurement, source Edwards AFB No. AFFTC-PA-10268

Fig. k Laser sensors setup, source Edwards AFB No. AFFTC-PA-10268

The result of these excitation methods are FRFs which were analyzed with experimental modal analysis. Another excitation method is the normal mode excitation which excite one mode at a time, meaning the structure is in resonance. The vibration response of the structure is measured directly and animated in a geometry display. This provides a very good feel for the dynamics of the structure. The operator modifies excitation frequency, excitation level and phase of the shakers in real time. The objective is to obtain the 90° phase difference between acceleration and force, or 0° phase difference between velocity and force. Another important aspect is the positioning of the shakers. For example, when exciting the first bending mode of the wing, we place the shakers in Z direction close to the wing tip where useful energy is injected in the wing tip in order to better excite the structure. In other words, for each mode, an optimum shaker position is necessary. Thus normal modes testing is used only whenever more understanding is needed at certain frequency such as non-linearity phenomena or very closely spaced modes. Fig. o and Fig. p show the symmetric bending wing mode animated on the test frame and on the CAD model. Animation on the CAD model is achieved with the automatic expansion of measured mode shape. The master/slave degree of freedom relationship is calculated by searching the closest geometrical point (or closest point belonging to the same topology definition such as on the same surface or connection line). The automatic mode shape expansion can easily be used for the test geometry and the CAD geometry (or finite element model). This capability provides better visualization and hence interpretation of the mode shapes. Fig. q and Fig. r show the anti-symmetric bending wing mode animated on the test frame and on the CAD model.

7

Conclusion In this paper, we have illustrated a few concepts on real-life tests. First of all, photogrammetry as a cost and time efficient solution for identification of sensor locations is applied on a large structure without compromising accuracy. Also, noncontact laser sensors are considered to be convincing in reducing setup time during a GVT, and the data is seamlessly integrated for modal curve fitting. Automatic mode shape expansion is used for better visualization of mode shape animation. Future work and experience on improvement of ground vibration testing can be conducted during these master classes.

Fig. l shows the driving points FRFs

Fig. m shows the coherence of the driving points

Fig. n show the velocity FRF (red) and acceleration FRF integrated in velocity format (green)

8

Fig. o Symmetric mode shown in test geometry

Fig. p Symmetric mode shown in CAD geometry

Fig. q Anti-symmetric mode shown in test geometry

Fig. r Anti-symmetric mode shown in CAD geometry

Acknowledgments

I would like to thank Donald S. Lange, Structural Technical expert of Edwards Air Force Base who helped us to organize the LMS Ground Vibration Testing Master Class in 2010 and especially on providing his advice and experience on ground vibration testing. I would like to thank Timo Kahlmann of AICON 3D Systems GmbH who showed and helped us with the photogrammetry measurement during the LMS Ground Vibration Testing Master Class in 2009. References [1] B. Peeters, Modern solution. In Proc. IMAC XXVI International Modal Analysis Conference, Orlando, Florida, USA, Feb 2008 [2] G. Foss, Photogrammetry of aircraft structures for sensors location. In Proc. IMAC XXIV International Modal Analysis Conference, 2004 [3] S. Pauwels, Experimental Modal Analysis: Efficient geometry model creation using optical techniques, In Proc. 75th Shock 1 vibration Symposium, 2004 [4] R. J. Dieckelman, The ground vibration Test - A Boeing IDS perspective, In Proc. IMAC XXIII International Modal Analysis Conference, 2005 [5] D. L. Hunt, A comparison of methods for aircraft ground vibration testing, In Proc. IMAC III International Modal Analysis Conference, 1985 [6] C.R. Pickrel, New concepts GVT, In Proc. IMAC XXIV International Modal Analysis Conference, 2006 [7] D. Göge, Ground Vibration Testing of Large Aircraft - State-of-the-Art and future perspectives, In Proc. IMAC XXV International Modal Analysis Conference, 2007