Sei sulla pagina 1di 5

2011 ECTC Proceedings ASME Early Career Technical Conference Hosted by ASME District E and University of Arkansas Support

Provided by the ASME Old Guard and the Committee on Early Career Development March 31-April 02, 2011, Fayetteville, AR

VIBRATION SHAKER CONTROL SYSTEM


L. Killingsworth 1 , P.W. Whaley, Ph.D., P.E. 2 Department of Mechanical Engineering Oklahoma Christian University

NOMENCLATURE NS = Samples per buffer f = Frequency NC = Cycles per Buffer Ai = Iteration of Beam Acceleration =increment in frequency ABSTRACT A LabVIEW controller for an electrodynamic shaker capable of vibrating a double-cantilever beam at resonance is described. Since maximum response occurs at the natural frequency of the beam, accelerated high cycle fatigue testing is possible by driving the beam at a high resonant frequency that can yield 106 cycles within approximately an hour. A test system is described that is capable of providing sufficient fatigue to characterize uncertainty in high cycle fatigue. INTRODUCTION A LabVIEW controller is described that replaces an obsolete controller and data collector for a small, inexpensive 50 lb shaker. The LabVIEW program can drive the shaker with a sine wave, square wave, triangle wave or random noise, as well as measure up to four input signals. For high cycle fatigue testing, base acceleration, beam acceleration, maximum strain and temperature are needed. Since the resonant frequency changes as the beam undergoes fatigue, the LabVIEW program constantly tracks the resonant frequency. A high cycle fatigue test at 106 cycles can be completed in about an hour by driving the beam at a relatively high resonant frequency (Scott-Emuankpor, et.al., 2007). An experimental method for testing the hypothesis that the entropy gained during fatigue failure is a material constant was described using old analog technology (Whaley, et.al., 1984). A double-cantilever beam (which is more balanced than a single-cantilever beam) was base excited and instrumented on only one side. Accelerometers were used to maintain the desired strain amplitude instead of strain gages, which tend to fail early in the fatigue life. Maintaining resonance using the analog system presented a challenge since
1 2

the resonant frequency changes as a result of the fatigue damage. It was very difficult to maintain resonance near the end of the fatigue test when the resonant frequency changed so rapidly. It is apparent that a digital system that can automatically control resonant frequency is required. There are two objectives for this project: a LabVIEW program to maintain resonance during high cycle fatigue testing and sufficient fatigue data to characterize uncertainty. The first objective has been accomplished. The second objective is underway. The LABVIEW program is being used to perform resonant-dwell high cycle fatigue tests on numerous specimens, and the data will be compared to predictions developed from tensile testing (Whaley, 1998).

EXPERIMENTAL PROCEDURE Figure 1 shows a schematic of the electrodynamic shaker system used in resonant-dwell high cycle fatigue testing. A double cantilevered beam (the test specimen) is attached to the shaker (LDS Model V408) and instrumented with an accelerometer (Abeam) attached at some distance from the base (center) of the double cantilevered beam. A strain gage is installed on the beam where the maximum strain occurs. This location is easily identified after a few tests. It is the location of the crack that forms a consistent distance from the base of the beam. The accelerometer and strain gage are connected to a National Instruments SCB-68 DAQmx device. A computer equipped with LabVIEW can then communicate with shaker system through the SCB-68 device. A signal to drive the shaker is generated and measurement signals are passed through the SCB-68. The LabVIEW Virtual Instrument (VI) can be split into three main components: one VI to drive the shaker, one to measure the inputs, and one to track the resonant frequency. All three of these functions can be run simultaneously within a single LabVIEW VI. The driver portion of the program is to output a signal to drive the shaker. This process can be better understood by following along with the LabVIEW block diagrams in Figure 2.

Graduate Assistant Professor of Mechanical Engineering

Abeam ENDEVCO NI SCB-68 NI NI SCB-68 USB-6251 LabVIEW LabVIEW LabVIEW Box BOX

Counterweight Abase

Macbook Computer

Shaker

PCB Amp

LDS Power Amp. Figure 1. System schematic showing (left to right): user interface, data acquisition/control box, amplifiers, shaker instrumented with accelerometers. In Figure 2a, the VI starts out by using a DAQmx Create Channel (AO-Voltage-Basic).vi, which creates the channel to generate an analog output voltage. This is shown below number 1 on the LabVIEW block diagram in Figure 2a. Next, the Waveform Generation Buffer (multi).vi (number 2 on the diagram in Figure 2a) is used to generate the waveform signal sent to the shaker. This subVI can generate a sine, square, triangle, or saw tooth waveform when the desired frequency, amplitude, samples per buffer, and cycles per buffer are set by the user on the front panel. LabVIEW then uses the DAQmx Timing (Sample Clock).vi (number 3 in Figure 2a) to set the sample clock rate and the number of samples to generate. From the waveform generation, LabVIEW outputs a waveform (number 4 in the block diagram in Figure 2a) on the front panel. Connected to the output waveform and waveform generation will be the DAQmx Write (Analog 1D Wfm NChan NSamp).vi, which will write the waveform to a task containing an analog output channel. This is shown above the number 5 in Figure 2a. Above the number 6 in Figure 2a, the DAQmx Start Task.vi will be used to transition the task to a running state to begin the signal generation. Figure 2b displays the while loop that will be used after the start task so that the task will continue to run until the stop button on the front panel is pushed by the user. This is done by placing the DAQmx Is Task Done.vi (number 9 on the LabVIEW block diagram in Figure 2b) inside the while loop. This subVI allows the task to run until the task is complete, the user stops the task by hitting the stop button on the front panel, or an error occurs. The while loop also allows the software to update the frequency while the task is running. This is done by referencing the desired frequency (number 7 in Figure 2b) within the while loop, connecting the cycles per buffer and samples per buffer inputs into the while loop, and multiplying the desired frequency by the samples per buffer then dividing by the cycles per buffer to calculate the new sample clock rate (number 8 in Figure 2b). This is derived from the equation resources the task has reserved are released. This is shown above number 7 in Figure 2b. The input portion of the LabVIEW program accepts up to four input channels to account for beam acceleration, base acceleration, temperature, and strain measurements and then displays them graphically on the front panel. This program can be better understood by following along with the LabVIEW block diagrams in Figures 3a and 3b as the process is explained. The majority of this program will be within a while loop to ensure the readings are continuously updated. The program starts out in Figure 3a with a DAQmx Create Channel (AI-Voltage-Basic).vi (number 1), which creates one or more channels to measure analog voltage. Continuing along in Figure 3a to number 2, LabVIEW uses the DAQmx Timing (Sample Clock).vi to set the sample clock rate and the number of samples to acquire. Next, the DAQmx Start Task.vi (number 3 in Figure 3a) is used to transition the task to a running state to begin measurement. From there, the program runs the DAQmx Read (Analog 1D Wfm NChan NSamp).vi (number 4 in Figure 3a), which reads one or more waveforms from a task containing one or more analog input channels. The signal will then be sent into a splitter (number 6 in Figures 3a and 3b) that splits the combined signal into separate signals. From the splitter, each signal (base acceleration, beam acceleration, and strain) is displayed graphically (Abase, Abeam, and Strain in Figure 3b) on the front panel and sent through the express VI named Measure (number 8 in Figure 3b), which performs voltage measurements on the signal. This program also uses the DAQmx Stop Task.vi within the while loop (number 5 in Figures 3a and 3b). Once the user stops the task, the task will then be cleared using the DAQmx Clear Task.vi (number 7 in Figures 3a and 3b) so that any resources the task has reserved will be released. Within the while loop in Figure 3b is a Write to File express VI (number 9), which will write the incoming data to a text-based measurement file so the number of cycles is recorded.

SampleClockRate =

# samples N s * f = s Nc

(1)

Once the user stops the task, the task is then cleared using the DAQmx Clear Task.vi (number 10 in Figure 2b) so that any

a. b. Figure 2. Driver VI, which creates an analog output channel, generates an output function, writes the analog waveform to an output buffer, and updates the desired frequency.

a. b. Figure 3. Block Diagram of the Measurement VI, which creates analog input channels, reads the waveforms, measures the accelerations of the base and the beam, and writes the accelerations to a text file. The portion of the LabVIEW program that tracks the resonant frequency will utilize the fact that resonance occurs at the frequency where beam acceleration and strain are maximum. A series of true/false statements are used to determine if the beam acceleration increases or decreases as the frequency changes. The frequency is changed so that the beam acceleration maintains the maximum amplitude. The user sets the incremental change in frequency on the front panel. If the beam acceleration decreases, then the frequency is adjusted and the accelerations compared. A flow chart of this process is shown in Figure 4. This flow chart was implemented in LabVIEW by using a shift register within the while loop, which also stores the time, the beam excitation frequency, beam acceleration, base acceleration, and strain into a text file. The data in the text file can be opened with a suitable spreadsheet (MS Excel), the time and frequency at each increment multiplied together and summed for the entire test in order to calculate the number of cycles the fatigue specimen endured.

Figure 4. Flow chart for tracking the resonant frequency based on a ratio of the accelerations.

TESTING The driver portion of the LabVIEW program was first tested by connecting the LabVIEW system to an oscilloscope to make sure the program was actually producing the signal and waveform that was selected on the front panel by the user. After the output waveforms were confirmed, the shaker was connected to the National Instruments SCB-68 box and the driver program was tested by running the program through LabVIEW to see if it could drive the shaker. It is very important to maintain a consistent ground connection for the instrumentation. Once the program was demonstrated to successfully drive the shaker, it was necessary to determine if the frequency could be changed while the program was running. When the frequency was changed on the front panel, the frequency of the shaker followed the change. The measurement program was tested in a manner similar to that of the driver program. The accelerometer on the base of the shaker, the accelerometer on the beam, and the strain gage were connected to the National Instruments SCB-68 box in a differential input configuration. The beam was then driven using a LabVIEW function generator and amplifier and the voltages from the accelerometers were measured and collected using the LabVIEW program. Both graphs on the

front panel of the measurement program displayed sine waves at the correct frequency and amplitude, so the measurement program was a success. The resonant frequency of the beam was tested by driving the beam at different frequencies and recording the beam acceleration, or strain, at each frequency. These acceleration and strain amplitudes were then plotted against frequency to generate a curve in order to find the resonant peak. Examples of typical Labview frequency sweeps plotting acceleration amplitude and strain amplitude versus frequency are shown in Figure 5. The peak on the curve in Figures 5a and 5b is the resonant frequency. These plots are important because they were performed in LabVIEW in a matter of minutes and provide the resonant frequency of the specimen to be set on the front panel without accumulating numerous samples before a fatigue test is performed. Once the resonant frequency of the beam was found, the tracking ability of the resonance program was tested on a test specimen with a resonant frequency of 82 Hz. The following plots in Figure 6 show that the program was successful in climbing (Figure 6a) and descending (Figure 6b) to the resonant frequency and maintain that frequency.

a. b. Figure 5. Examples of tests to find the resonant frequency of a fatigue specimen.

a. b. Figure 6. Plots of tests performed on resonance program within LabVIEW to determine whether resonance could be tracked and maintained.

Scatter in fatigue data is very common and makes design for fatigue resistance very challenging. The absence of a universally accepted theory of fatigue damage makes empirical approaches the only option. This makes design for fatigue resistance very expensive because lots of data are required to quantify the statistical uncertainty. This problem is especially challenging for high cycle fatigue because each data point requires such a long time.

several tests per day compared to several days per test using conventional testing. REFERENCES Scott-Emunakpor, O., Shen, M.-H., George, T, Cross, C.J. and Calcaterra, J., Development of an Improved High Cycle Fatigue Criterion, ASME Journal of Engineering for Gas Turbines and Power, 2007, Vol. 129, pp. 162- 169. Whaley, P.W., Chen, P.S., and Smith, G.M., 1984, Continuous Measurement of Material Damping During Fatigue Tests, Experimental Mechanics, 24(4), pp. 342-348.

Figure 7. Predicted Fatigue Life Tolerance Bands for 6061-T6 Aluminum A new theory of fatigue damage is being developed based on the entropy lost during failure. Figure 7 shows the preliminary results of that research for 6061-T6 Aluminum. The lines are theoretical predictions of the mean fatigue life with tolerance bands (95% with 95% confidence). The data were collected using a servo-hydraulic tensile tester. Significant scatter in the data is clear, although distorted by the logarithmic scale. The resonant-dwell high-cycle fatigue test system described in this paper is the cost-effective tool needed to provide the data for validating the tolerance bands indicated in Figure 7. The significance of a successful theory of fatigue is that significantly less data are required as compared to an empirical approach. Reduced testing is used to validate the model rather than replicate every possible service loading. RECOMMENDATIONS AND CONCLUSIONS A LABVIEW driver and data collection system for a small electrodynamic shaker has been developed and tested. Accelerated high cycle fatigue tests are underway. In the future, a stop function within LabVIEW needs to be developed to stop a fatigue test just as beam failure occurs. This could be done using a true/false statement to determine whether or not the resonant frequency has decreased by a certain percentage of the original resonant frequency. More testing is needed to determine the resonant frequency decrease corresponding to crack initiation. For a thin beam, most of the fatigue life will be crack initiation. A PID control should be included in LabVIEW to keep the strain amplitude constant. This will ensure the strain amplitude remains consistent from specimen to specimen. Hundreds of fatigue specimens are needed to characterize uncertainty in fatigue life predictions. Accelerated resonantdwell high cycle fatigue testing is very cost-effective, allowing

Potrebbero piacerti anche