514.8-33 information required

514.8-3 VEHICLE PREPARATION FOR DATA ACQUISITION - INSTALLED EQUIPMENT

Accelerometers must be mounted on the vehicle walls, deck, and roof, as well as on brackets and shelves that are integral parts of the vehicle, as close as possible to the point(s) of attachment of the existing/planned installed equipment. The purpose is to measure the vibration environment of the vehicle at the input location(s) of the installed equipment which has the same configuration as the equipment subsequently used as a test item during laboratory testing. For instance, if a piece of equipment is mounted on a bracket in the vehicle and that bracket will not appear as part of the equipment during subsequent laboratory testing, the environment should be measured on the bracket as the input to the equipment.

If a mounting platform exists and the equipment to be installed thereon is not available, use a model of the equipment with the same mass and center of gravity. This ensures that the reaction of the installed equipment will be included in the data recorded at the input to the mounting platform.

In the situation in which the test item is instrumented and may be integrated into the control scheme as such as response control or possibly as a limit location, it is critical that any surrogate hardware employed must have strong dynamic similarity to the tactical hardware. It is recommended that a comprehensive modal test of the surrogate hardware be conducted to ensure proper dynamic response is maintained. In some instances, similarity acceptance criteria may be called out, to which the modal parameters of the surrogate hardware must comply.

The difference between the mechanical impedances of mountings for field-installed and laboratory-installed equipment should be considered, particularly for relatively massive equipment. A comparison of the field and

MIL-STD-810H METHOD 514.8, ANNEX F, APPENDIX A

laboratory frequency response functions is one method of evaluating this difference, and the use of average, extreme, or response laboratory vibration control techniques is considered a valid approach to minimizing any impedance mismatch.

514.8-3 COMMON ANALYSIS FUNCTIONS AND STATISTICAL DEFINITIONS OF VSD DATA ENSEMBLES

In preparation for a VSD effort, vibration time histories are recorded for a series of test conditions (defined as “events” in Annex F, paragraph 3) also referred to as "runs". An event is defined (for ground vehicles) as operation over a specific uniform terrain, for a specific test item configuration (load, tire pressure, etc.) at a constant speed. For an aircraft, one may have a list of events defined in terms of various modes of flight (level flight, rolling maneuvers, etc.) conducted at various airspeeds. A common form of analysis involves converting the complete time history (of a particular channel) into the compressed frequency domain format of the ASD function by dividing the time history into equal length data blocks and computing the ASD for each of the data blocks independently. When combining spectra, it is assumed that the spectra being combined represent a homogeneous set. (i.e., overall spectral levels are comparable and the spectra have the same general shape). If this is not the case, combining spectra should be avoided and another "spectra category" should be added to represent the test condition. When computing estimates of these ASD functions, it is desirable to compute the linear average (assuming the number of samples is sufficiently large $\left(n_{d}>30\right)$ ), the standard deviation and the peak, all as a function of frequency, over the length of a test run. The standard deviation represents the variation in the spectral data, as a function of frequency, at a given location on the vehicle due to randomness of the test process. Although the data are stationary, excursions about the mean occur in both the time and frequency domains.

In addition to computing the mean ASD of an individual event, the standard deviation, and peak versions of the ASD are often of interest. For ease of illustration, the following symbolic structure will be employed:

G₁(f) = ASD(instantaneous) ASD computed over a single time interval T

$G_{d}(f)=\operatorname{ASD}(\text { Standard Deviation })=G_{d}(f)=\left.\frac{1}{n_{d}-1} \sum_{i=1}^{n_{d}}\left(G_{i}(f)-G_{m}(f)\right)\right)^{2}$

Note that, statistically speaking, the normalized error for the (Peak) spectrum could be very high because of the limited number of degrees of freedom available from the peak values of $G_{i}(f)$. This spectrum is a maxi-max auto spectral density estimate and should be used with caution (if at all). An example generation of the spectra discussed above as computed from data acquired from a typical wheeled vehicle data is shown in Figure 514.8F-B.3.

MIL-STD-810H METHOD 514.8, ANNEX F, APPENDIX B

Figure description:

Typical Wheeled Vehicle Vibration Data 1 Test Course, 1 Lap, 1 Location Average, Average+Std Dev and Peak Spectra

Typical Wheeled Vehicle Vibration Data

Legend

• Average — color: red; symbol: Solid line
• Ave + Std Dev — color: dark blue; symbol: Solid line
• Peak — color: green; symbol: Solid line

Annotations

• Equation B.4 (text_label - position: Pointing to 'Ave + Std Dev' curve at approximately 2 Hz)
• Equation B.5 (text_label - position: Pointing to 'Peak' curve at approximately 5.5 Hz)
• Equation B.3 (text_label - position: Pointing to 'Average' curve peak at approximately 7.5 Hz)

Figure 514.8F-B.3. Average, average plus standard deviation, and peak spectra.

During the VSD process, classical statistics are often employed in somewhat of an ad hoc manner to address unknowns and sample size limitations. The data from many locations and many test events can be combined (by axis) using different techniques to produce representative composite spectra. The first technique is a simple linear average of all average spectra from all channels and all events to produce an overall average spectrum. If the mean and the median of the distribution are the same (which is the case for Gaussian data), this represents approximately the 50th percentile of the spectral data. A second technique is a "standard" conservative approach often integrated into the VSD process in which the ASD(Sum) spectra from each channel and each event are combined by using the average of these spectra with the addition of one standard deviation. The standard deviation computed during this process represents considerations such as the spectral variance due to location, test course differences, and courses not considered and is not the same as $G_{d}(f)$. Mathematically this spectral average is shown as:

$G_{a}(f)=\frac{1}{M} \sum_{i=1}^{M} G_{s_{i}}(f)$

Where, $M$ represents the number of events considered in the computation of the average spectra.

The standard deviation of ASD values due to variations in test courses and instrumentation locations as a function of frequency is defined as:

$G_{e}(f)=\left[\frac{1}{M-1} \sum_{i=1}^{M}\left[G_{s_{i}}(f)-G_{a}(f)\right]^{2}\right]^{1 / 2}$

MIL-STD-810H

METHOD 514.8, ANNEX F, APPENDIX B

The final spectral measurement defined is: $G_f(f) = G_a(f) + G_e(f)$ B.6

Computation of $G_f(f)$ is a key component used in the VSD techniques defined in Annex F, Appendix D.

Consider the same ensemble of data as discussed in paragraph 2.7 (typical wheeled vehicle data (4 test courses, 3 data runs, 2 locations)). Figure 514.8F-B.4 includes an overlay of all 24 spectra with the average of the ensemble of $G_{s}(f)$ spectra shown in yellow and the final representative spectrum, $G_{f}(f)$, shown in green.

Figure description:

Typical Wheeled Vehicle Vibration Data 4 Test Courses, 3 Laps, 2 Locations; All Location 1 and 2 ASD's are ASD(Sum Gs(f

Typical Wheeled Vehicle Vibration Data

Legend

• All Loc 1 — color: red; symbol: Thin solid lines
• All Loc 2 — color: blue; symbol: Thin solid lines
• Avg [Ga(f — color: yellow; symbol: Thick solid line
• Sum [Gf(f — color: green; symbol: Thick solid line

Figure 514.8F-B.4. Combination of 24 individual spectra.

It is desirable for a vibration schedule to be a conservative estimate of the true environment within some credible bounds. A conservative estimate is required for two reasons. First, the test sample size is limited (usually one vehicle) and some allowance must be made for differences within a class of vehicle. Second, vibration control systems take the test specification in ASD form and create a time history (to drive the shaker) by assuming a Gaussian distribution of amplitudes in the time domain. Wheeled vehicle vibration data is nearly Gaussian, but generally has larger "tails" (i.e., more data at high levels and higher peak levels) than a Gaussian distribution of the same mean and standard deviation. Since damage occurs at the tails of the distribution (and not around the mean), it is necessary to amplify the overall level so that the Gaussian process will produce the higher levels commensurate with the measured data. Merely enveloping the peak spectra provides conservatism but results in an over-test since the test rms level is generally much greater than the highest individual level measured. To ensure that the final spectral estimate is at least as large as the actual measured data, this spectrum can be adjusted (amplified or attenuated) so that its rms value is in reasonable agreement with the largest rms value measured at any location during any data run provided the corresponding spectra are similar in shape.

MIL-STD-810H METHOD 514.8. ANNEX F. APPENDIX B

The resultant spectrum shown in Figure 514.8F-B.5 is compared to an envelope of all the average spectra from the data set (in this case, 24 spectra). Note that the resultant test spectrum is approximately an envelope of the individual spectra.

Figure description:

Typical Wheeled Vehicle Vibration Data 4 Test Courses, 3 Laps, 2 Locations; Envelope ASD's are Envelope of ASD(Avg; ASD(Sum is Derived From Ave+SD PSD's

Typical Wheeled Vehicle Vibration Data

Legend

• Envelope — color: red; symbol: thick solid line
• ASD(Sum — color: green; symbol: thick solid line

Figure 514.8F-B.5. Comparison of resultant test spectrum (green line) to envelope of individual spectra (red line).

It is important to note that the spectrum described above may be composed of different operating conditions which are not present for the duration assigned to the total environment. For example, if certain frequency components were contributed by operation over a particular test course, they would be applied to the test as if those components occurred for the entire duration, not just the segment represented by operation over that particular test course. In this case, test duration would also apply conservatism to the process, which may or may not be desirable. Problems of this nature are addressed further in the Annex F, Appendix D.

MIL-STD-810H METHOD 514.8, ANNEX F, APPENDIX C METHOD 514.8, ANNEX F, APPENDIX C Combination of Spectra (Fatigue Damage Spectra)

514.8-3 NARROWBAND RANDOM SPECIFICATION DEVELOPMENT PROCEDURE

To allow greater flexibility in adapting to project specific requirements, the narrowband or sine tones removed from the broadband (see paragraph 2.5) can be processed in a spreadsheet. A sample narrowband random spreadsheet is provided in Table 514.8F-D.VII. The sample is of a wheeled vehicle on the 2-inch Washboard Course. Three tones were removed from each of three events (5, 7.5 and 10 mph). Note that sinusoidal processing, rather than narrowband random, would generally be used for a wheeled vehicle on the 2-Inch Washboard Course. The example is used for ease of discussion only, and will not affect the description of the process.

The following definitions provide a column-by-column description of the narrowband random processing procedures.

514.8-3 SWEPT SINE EXCITATION

The MRS is extrapolated from the MRS of fixed sinusoidal signals at frequencies corresponding to the limits of the domain of sweeping.

Example: MRS for a swept sine from 300 Hz to 1200 Hz (Figure 514.8F-E.2).

Figure description:

System Resonance Response Spectrum

MRS vs. Frequency

Legend

• MRS Envelope (Left — color: Blue; symbol: Thick solid line
• MRS Envelope (Top — color: Black; symbol: Thick solid line
• MRS Envelope (Right — color: Dark Red/Brown; symbol: Thick solid line
• Excitation — color: Black; symbol: Thick horizontal line
• Resonance Curves — color: Black; symbol: Thin solid line

Annotations

• MRS (text_label - position: Pointing to the envelope of resonance peaks and trailing slopes)
• Excitation (text_label - position: y = 1.0, x from 200 to 1250 Hz)

Figure 514.8F-E.2. MRS for a swept sine from 300 Hz to 1200 Hz.

MIL-STD-810H

METHOD 514.8

(CTIS) settings), suspension settings (as applicable) etc.

NOTE: Modal surveys of both test fixtures and test items can be extremely valuable. Large test items on large complex fixtures are almost certain to have fixture resonances within the test range. These resonances may result in significant overtests or undertests at specific frequencies and locations within a test item. Where fixture and test item resonances couple, the result can be catastrophic. Similar problems often occur with small test items, even when the shaker/fixture system is well designed because it is very difficult and often impractical to achieve a lowest fixture resonant frequency above 2000 Hz. In cases where the fixture/item resonance coupling cannot be eliminated, consider special vibration control techniques such as acceleration or force limit control.

514.8-3.1

Event (Col. A) – This column lists the events and tones (or harmonics) removed from the broadband.

514.8-3.2

Speed on 2” WB (Col. B) – This column contains the ground speed of the given event.

514.8-3.3

Center Frequency Selected (Col C.) – The center frequencies of the narrowbands of interest may be set to the center frequencies of the narrowbands removed during spectral spike removal. However, it might be desirable to calculate the center frequency. For example, the center frequencies of a wheeled vehicle on a periodic course may be affected by slight changes in vehicle speed or the frequency resolution of ASD calculations. For proper control on an exciter table the tones must be harmonically related and are best set to the frequencies that would have resulted given ideal conditions. For a wheeled vehicle the frequencies can be calculated using the vehicle speed and the displacement of the periodic input (i.e., washboards spaced 0.6 meters apart). For a helicopter the frequencies can be set to the known blade passing frequency. Care should be taken to assure that the center frequency calculated does not vary significantly from the actual center frequency measured.

MIL-STD-810H METHOD 514.8, ANNEX F, APPENDIX D Table 514.8F-D.VII. Narrowband Random Calculations

MIL-STD-810H