A Comparison of the Response of a Captive Carried Store to Both Reverberant Wave Acoustic Excitation and the Field Environment*

Transcription

1 A Comparison of the Response of a Captive Carried Store to Both Reverberant Wave Acoustic Excitation and the Field Environment* Mechanical Jerome S. Cap and Thermal Environments Department Sandia National Laboratories PO Box 5800 Albuquerque, NM (505) L( Thomas C. Togami Manufacturing and Rapid Prototyping Department Sandia National Laboratories p,yo 3 d -4 - "7 Eb.,." d L', J. Ron Hollingshead Sandia National Laboratories (ret.) Stores that are carried on high performance military aircraft are exposed to severe vibroacoustic environments from several different sources. Sandia National Laboratories conducted a test program to determine the viability of reproducing these field environments with a combined vibroacoustic test. This paper will present the results of that test series emphasizing the methods used to derive the laboratory inputs that produce the "best" possible match for the field response. NTRODUCTON Sandia currently certifies its gravity bombs to the captive carriage environment by exciting the unit with a single electrodynamic shaker driving through both lugs simultaneously. The test unit is vibrated separately in the three principal axes in order to simulate the three-dimensional field environment. Postulating that a- single combined vibroacoustic test could provide a superior match for the field environment, a series of vibroacoustic tests were conducted in Sandia's Acoustic Test Facility (ATF) [l]. Those tests were designed to establish a unique vibroacoustic test input for the maximum dynamic pressure (max q) conditions for external carriage on an F-15E and bomb bay carriage on an F-111. The ultimate goal of this project was to achieve an adequate simulation of the field environment, thereby providing a cost effective alternative to field testing. A more modest objective, the results of which are beyond the scope of this paper, was to demonstrate that a single vibroacoustic test could provide a better simulation of the system environment than the base excitation, single axis random vibration tests currently being used to certify Sandia's gravity bombs for flight. "This work was supported by the U.S. Department of Energy under Contract DE-ACO494AL

2 DSCLAMER Portions of this document may be illegible in electronic image products. mages are produced from the best available original document.

3 This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

4 PROPOSED TEST PLAN The test setup (as shown in Figure 1) consisted of a realistic mass mock-up version of one of Sandia's gravity bombs (designated a Vibration Flyaround or VFA unit) mounted upside down on two Unholtz Dickie TlOOO electrodynamic shakers located in the ATF reverberant chamber. Each shaker was run by a dedicated controller. The VFA was attached to the shakers using a simulated ejection rack and two pairs of flex webs. The individual pairs of flex webs were oriented at 90" relative to each other to permit bending about both the pitch and roll axes. This was done primarily to limit the bending moments applied to the shakers, but also served to increase the off-axis coupling of the single axis shaker inputs. The ejection rack was attached to the VFA in a manner that approximated the actual restraint conditions. The weight of the VFA was suspended from the overhead crane using- a strong - back attached to the ejectj.on on rack with nylon straps. Mic 1 0 Mic Nose Limit Gage 0 Mic 3 0 Tail Limit Gage r Figure 1: Vibroacoustic Test Setup and Control nstrumentation Locations The rst step in the test development process was to define the acoustic and vi,ration test inputs. The physical limitations of. the ATF will roll-off the acoustic input below 60 Hz. Therefore, vibration was the only input below that frequency. For frequencies above 00 Hz acoustics were the only input used for this test. For frequencies between 60 Hz and 00 Hz the vibration and acoustics are superimposed. Table and Figures 1 and show the gages used for the test. Since the acoustic system was controlled in 1/3 octave bandwidths, the results for the acoustic tests were analyzed using Power Spectral Densities (PSDs) with similar spectral resolution. Ratios of the laboratory PSDs to the flight PSDs were created for the VFA gages. The overall realism of the laboratory simulation was evaluated using equally weighted averages of those ratios. Averages were computed for several different groupings of

5 gages in order to study the problem from different perspectives (all VFA gages, vertical vs. horizontal, internal vs. external components). Table : Gage List for the Combined Vibroacoustic Test Series TEL,Mic# Z Tail Limit Control Microphones #, #, #3 Nose External Component Z N/A N/A /Aft nternal Comoonent / Forward nternal Component Figure : nstrumentation List for the VFA Response Gages DERVATON OF THE ACOUSTC NPUTS The initial choices for the acoustic environments were taken from the appropriate ML-STD-8 1OE [] profiles tailored to the aircraft/store parameters (dynamic pressure, length, etc.). Table 1 presents the Overall Sound Pressure Levels (OASPLs) and the SPL frequency break points by bomb station for the F-15E external carriage acoustic profile. The variation in acoustic level predicted in Table 1 for the nose and tail were ignored for the initial acoustic profile. The resulting initial acoustic profile presented in Figure 3 is a composite of the remaining station predictions. Table 11: ML-STD-810E Acoustic Profile for External Carriage on an F-15E (Conformal Fuel Tank Station, q=1450 lb/ft) Location Nose Center Section Tail Distance from OASPL (db) SPL Break Nose (inches) Points (Hz) / / / / /601.3

6 s n v) 135 MS8lOE F15E OASPL=147. db PF15E Best Effort OASPL - MERT CFT (+sigma) 161.dB OASPL Octave Center Frequency (Hz) Figure 3: Vibroacoustic Test Series nitial and Best Effort Acoustic Profiles for the F-15E (q=1450 lb/ft) Figure 3 presents a comparison of the initial ML-STD-810E F-15E acoustic profile and the final tailored "best effort" acoustic profile. The initial ML-STD profile for the external carriage acoustic profile was remarkably close to the final profile. The +0 acoustic profile predicted by MERT [3] for the conformal fuel tank station is also shown for comparison (Reference [4] recommended the +0 levels for comparison with ML-STD-81OE). The shape of the MERT profile appears to be reasonable, but MERT seems to be over-predicting the acoustic levels. The F-111 bomb bay acoustic profiles derived from ML-STD-810E are dependent only on the dynamic pressure. Figure 4 presents the initial ML-STD-810E and final tailored "best effort" acoustic profiles for the F-111 data Best Efort ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; : ; ;Rf 113 Octave Center Frequency (Hz) Figure 4: Vibroacoustic Test Series nitial and Best Effort Acoustic Profiles for the F-111 (q=90 b/ft) The basic shape of the ML-STD-801E F-111 profile is considered to be valid, but the corresponding 177. db OASPL was inconsistent with the VFA flight vibration levels. Since the speed of the F-111 quickly bleeds off once the bomb bay doors are open, it

7 was decided to drop the ML-STD prediction and generate a reduced acoustic profile that produced a match for the measured flight data. This new acoustic profile was computed using a ratio of the flight vibration data from the F-111 and the F-15E. DERVATON OF THE VBRATON NPUTS A four gage configuration was chosen as the control scheme for the vibration test. The two primary control gages (one for each shaker) were located on the ejection rack near the forward and aft lugs. The secondary control gages (which were used as response limit channels) were located on the nose coincident with the Nose External Component (NEC) gage and near the tail on a hard point between the fins (each limit channel was applied to the nearest controller). All control gages were mounted externally so as to not interfere with the test unit. The input profiles at these four locations were tailored to produce the best average vibration response at the in-axis VFA gages. Deriving a set of input PSD profiles that produces the best response at five response locations was non-trivial (especially when a response limited test is being used). A solution to this problem was achieved using the assumption that the voltage drive signals for the individual shakers are uncorrelated, and the fact that the relationships between the drive signals and the various control, limit, and response signals are defined by the Cross Spectral Density (CSD) matrix. Appendix A presents the details of this derivation. TALORNG THE NPUT PROFLES All discussions pertaining to the optimization of the vibroacoustic input will focus on the F-15E simulation. The F-111 solution used the same process, although the task was simplified due to the understanding gained from the earlier F- 15E testing. A baseline acoustic test was performed using the initial ML-STD-810E acoustic profile. Revisions to this profile were made by adjusting the acoustic levels using an equally weighted average of the ratios of the desired and achieved PSD responses for all of the VFA gages. The vibration inputs were obtained using the steps in Appendix A. Once the individual vibration and acoustic profiles were obtained, a combined vibroacoustic test was performed using the tailored acoustic and vibration inputs. There were several issues that demanded special attention during the test series: 1) The controllable frequency range for the acoustic profile was between 63 Hz and 1000 Hz. Acoustic response outside of that frequency range is generated primarily through distortion and harmonics of the control spectrum. As a result of this phenomenon, a step occurs in the acoustic spectrum at the highest controlled frequency (the upper bandedge of the highest controlled 1/3 octave band). This step is as much as 3 db for OASPLs in the range of 145 db. However, the degree of distortion increases with increasing OASPL, which in turn decreases the size of the step. This behavior makes it difficult to predict just how the acoustic profile will turn out until a test is run. ) The vibration controllers had some difficulty with rigid body modes. This behavior is manifested in the form of a broadband amplification of the responses at the tail gage. While this behavior is primarily a nuisance, it had to be accounted for when defining vibration control spectrums.

8 4) The vibration control scheme was restricted by constraints on the allowable control spectrum shapes (at the time of the test the limit spectrums had to be either a straight line PSD or a scaled version of the control spectrum). The tail limit gage signal was lowpass filtered at 100 Hz prior to sending it to the controller to prevent the acoustics from feeding back into the vibration control system., RESULTS Figures 5 and 6 cover the range of agreement between the laboratory results and the field responses for the F-15E simulation. Figure 7 presents the averages for the ratios of the laboratory results to the flight data for all gages, vertical gages, and lateral gages respectively. Overall, the F-15E results are considered to be a good match. The low off-axis laboratory response below 10 Hz is due to only having the single axis vibration input available. The poorest matches at high frequency (>lo00 Hz) occur at the components buried deepest in the unit (see Figure 5), while the best matches corresponded to components near the surface of the unit (see Figure 6) c O-' lo+ -FELD --LABORATORY 10' Requency (HZ) 1 o3 Figure 5: F-15E Vibroacoustic Test Series Example of nternal Response (FCZ) l0 -FELD /4\,,,,,,,,,, --LABORATORY Requency (HZ) 1o3 Figure 6: F-15E Vibroacoustic Test Series Example of External Response (AECY) Figures 8-10 present the same comparisons f o r the F-111 simulation as were shown in Figures 5-7 for the F-15E. Even though the average response was matched fairly well, the results for this test show a greater gage to gage variation. -ALL AVERAGE This is believed to be due at --VERT AVERAGE, least in part to the non-.-lat AVERAGE p. uniform nature of the bomb J, \.; bay acoustic profile. The poor 1o3 10 k?equency (HZ) agreement at low frequency is believed to be due at least in Figure 7: F-15E Vibroacoustic Test Series part to the fact that the Response Ratio Averages for VFA Gage Combinations laboratory test used only 5,;,

9 vibration to simulate the response when in fact acoustics are contributor in this frequency range in the bomb bay environment. a significant U cn 1o -~ 10" -FELD --LABORATORY F~PQUENCY - HZ 10' 1o3 ' Figure 8: F-11 1 Vibroacoustic Test Series Example of nternal Response (FCZ) SUMMARY ' Figure 9: F-111 Vibroacoustic Test Series Example of External Response (AECY) A process has been developed to obtain a "best" fit to the response data from a single F ~ ~ Q U E N C(HZ.) Y 1o3 aircraftjstore configuration using a combined vibroacoustic test. Procedures were implemented to maximize the use of mathematical tools in order to obtain an optimal solution. Figure 10: F-111 Vibroacoustic Test Series n general, the results for the Response Ratio Averages for VFA Gage Combinations ~ - 1 5external ~ carriage test prod uc ed accept ab 1e res u 1t s (additional acoustic capacity can be used to "fill in" the low spots as deemed necessary). The fact that the spectral shapes of the VFA response PSDs were reproduced simultaneously for all three axes with minimal tailoring of the acoustic profile confirms that the acoustic input is the correct choice for simulating the high frequency portion of this environment (>00 Hz). The results for the F-111 showed more scatter in the responses, but further research into the nature of bomb bay acoustics has led to the conclusion that the complex nature of this environment will tend to produce such results. The concept of using two independent shaker profiles to produce a least squares fit to five response locations was validated for the low frequency portion of the response spectra (<00 Hz). This result will permit the captive carry environment to be approximated with a combined vibroacoustic test in Sandia's current acoustic facility

10 rather than having to use an acoustic c h a m b e r. with generating acoustic noise down to 10-0 Hz. dimensions suitable for The ML-STD-810E prediction for the F-15E external carry acoustic profile proved to be quite close to the final optimized solution. n light of the fact that MERT is being developed as a replacement for ML-STD-glOE, it is planned to research the apparent differences between the MERT predictions for the acoustic profile and the corresponding ML-STD-8 1OE prediction. ACKNOWLEDGMENTS The authors would like to acknowledge the assistance of Dave Smallwood in the derivation of the vibration control scheme and Michael Nusser for his work in designing the fixtures used for the test. REFERENCES 1 ) Rogers, J. D. & Hendrick, D. M., "Sandia National Laboratories' New High Level Acoustic Test Facility"; Proceedings of the ES, 1990, pp ) ML-STD-8 1OE; "Environmental Test Methods and Engineering Guidelines." 3 ) Leak, C. E., "How MERT Fits 810 or ML-STD-810, The Next Generation;" 63rd Shock and Vibration Symposium, Vol 11, pp ) Heaton, P. W., "Comparison of Captive Flight Acoustic Prediction Techniques," 66th Shock & Vibration Symposium, Vol, pp APPENDX A: METHODOLOGY FOR OPTMZNG THE NPUT SPECTRA FOR THE VBRATON CONTROL SCHEME t was desired to control the vibration portion of the vibroacoustic test in such a way as to achieve the best match for the flight responses in a least squares sense. Given the fact that there were two inputs and five vertical (Z) axis responses (the off-axis responses were not considered), this is not a trivial task. n addition, there is a complication arising from the fact that there is significant cross-coupling between the shaker control accelerometers. The solution to - t h i s problem was derived with technical guidance from Dave Smallwood. Figures 1 and present the layouts for the control and response accelerometers for the VFA. The solution to this control problem was based on the assumption that the voltage drive signals coming from each controller are uncorrelated and can therefore be treated as independent inputs. Equations 1-7 define the governing equations used for this derivation: A=HE Where A is the acceleration response, E is the diagonal matrix of drive signals in volts, and H is the matrix of transfer functions. &[AE']= &[El =? HE[EE'] ] Where E represents the expected value.

11 Gyx = (3) Gxx = E[EE] (4) Gyx = HGxx (5) H = GyxGxx-l (6) Gyy = HGxxH (7) G y x is a Cross Spectral Density (CSD) matrix relating the accelerometer responses to the drive signals G,, is a CSD matrix for the drive signals (the diagonal terms are the autospectrum or PSDs of the drive signals). Gyy is a CSD matrix for the accelerometer responses (the diagonal terms are the autospectrum of the accelerometer signals). n order to implement this solution, the terms in the CSD matrix were measured for the test setup in Figure 1. Responses were recorded with both shakers driven simultaneously. These results are the basis for G,,, G,, and Gxx. t was decided to perform the analysis using a set of 3 Gyy matrices corresponding to the control gages (x), the limit gages (x), and the VFA gages (5x5). The G,, matrices were constructed accordingly. The analysis code MatlabTM was used to implement the solution of these equations using the following steps. 1 ) Use the VFA flight PSDs and the system CSDs to derive the pseudo-inverse least squares solution for the best average drive signals. Figure 11 presents a comparison of the flight response and the predicted laboratory response PSDs computed for a typical gage location using the best average drive signals. -FLGHT --PREDCTED 1o-6 10' 1 LABORATORY Frequency ( H Z 7 Figure 1 1 : Flight vs. Laboratory (NECZ) Based on deal Drive Signals bring the qualitative response. the nose VFA responses back in line and success is dependent on Figures 1 and 13 present the limit channel and the forward ) Compute the corresponding control gage PSDs. Tailor these PSDs to produce practical straight line test specifications. Compute the associated drive signals. Compute the PSD profiles at the limit and VFA gages using the new drive signals. Due to the use of straight line PSDs at the control points, the system will be overdriven at its fixed base resonances. 3) Define straight line response limit PSDs to with the desired levels. This step is understanding the dynamics of the VFA realized and straight line PSD profiles for rack control channel respectively. The

12 notch in the realized PSD levels at the forward control gage around 70 Hz is a direct result of response limiting at the nose limit gage. 4) Compute drive signals based on the limit channel straight line PSDs. The minimum envelope of the drive signals derived from the control and limit channel straight line PSD profiles represents the desired response limited drive signal. Figure 14 presents the comparison of a typical VFA flight PSD with the predicted and achieved laboratory PSDs resulting from the final response limited drive signals. 10- E 10-1 i r -O- -LMT -- REALZED 10 FREQUENCY (?$ Figure 1: Desired vs. Realized Responses Nose Limit (NELZ) Response Limited Drive Signals O- 1 o-6 [ 10 \ \ \ LLCONTROL -- REALZED FREQUENCY (A9) Figure 13: Desired vs. Realized Responses Forward Rack Control (FRCZ) Response Limited Drive Signals --PREDCTED LABORATORY -.-ACHEVED LABORATORY l Frequency (Hz? Figure 14: Flight vs. Laboratory Responses (NECZ) Response Limited Drive Signals This procedure is susceptible to factors such as noise and numerical round-off errors which distort the results (especially near resonant frequencies). This is believed to be the source of the spike at 40 Hz in the predicted laboratory PSDs in Figures 11 and 14. However, the solution was believed to be stable, so it was left up to the controllers to overcome this anomaly (which they did successfully as seen from the achieved laboratory PSD profile in Figure 14). The single most important outcome of this process was the verification that the low frequency portion of the vertical response at the five in-axis VFA gage locations could be matched using two independent inputs as was theorized at the start of the project. From this result it is further postulated that if a good match is not achievable, then vibration input via the bomb s lugs is not the only input source.