Tools › Structures › Loads and Dynamics › Pyroshock Attenuation › Method

Pyroshock Attenuation: Method

Open the Pyroshock Attenuation tool

Pyroshock Attenuation: Distance and Joint Attenuation of a Shock Response Spectrum, Maximum Expected Flight Environment, Test Margins and Tolerances

The problem

Explosive bolts, separation nuts, pin pullers, cable cutters, pyrovalves and linear shaped charges release their energy in microseconds. The structure around them rings at thousands of hertz with accelerations of thousands of g, for a few milliseconds. Major structure seldom fails under such a shock, but relays, crystals, solder joints and ceramic parts do, so every component near a device needs a shock test specification. Early in a program there are no measurements on the vehicle: the level at a component has to be estimated from the level near the device, the length of the shock path and what lies in that path. This tool applies the empirical procedures printed in NASA-STD-7003A (Pyroshock Test Criteria) and NASA-HDBK-7005 (Dynamic Environmental Criteria), most of which go back to the Martin Marietta study of Kacena, McGrath and Rader (1970) and to Barrett's work for the Viking lander (1975), and then adds the test margins and tolerances of the standard. MIL-STD-810H Method 517.3 repeats the same prediction material for defense materiel and is the second source; where it differs (test tolerances, margins) the difference is stated below.

The source spectrum, its ramp and its plateau

A pyroshock is described by its maximax shock response spectrum (SRS) at Q = 10: the peak acceleration of a single-degree-of-freedom oscillator of each natural frequency, mounted at the point in question. A pyroshock SRS rises with frequency along a ramp (the standard gives 9 to 12 dB per octave as typical) up to a knee frequency and then runs roughly level, the plateau. The tool takes the spectrum as breakpoints joined by straight lines on log-log axes. The knee is the breakpoint of the highest level unless a knee frequency is entered.

The two parts are treated apart because the published data treat them apart. The handbook's procedure (section 5.3.2) and the standard (A.3.2) both say that joints reduce the peak SRS value and leave the up-ramp alone; Figure 5.7 of the handbook has one distance curve for the spectrum peak and another for the spectrum ramp; Kacena's design manual attenuates the peak of the spectrum by a structure curve and the low-frequency constant velocity line by its own curve, and takes the region below both lines as the prediction.

With rp the product of the factors on the plateau and rr the product of the factors on the ramp, the predicted spectrum at natural frequency f is

SRSpred(f) = min[ rr R(f), rp P(f) ],

where R(f) is the source spectrum up to the knee, carried on above the knee as a straight line at the slope of its last segment, and P(f) is the source spectrum from the knee upward, held at the knee level below the knee. When the ramp is attenuated less than the plateau, which is the usual case at long distances, the two lines meet below the original knee and the knee of the predicted spectrum moves down in frequency; that is the construction of Figure 3.9 of the design manual. With equal factors the source shape is kept.

Attenuation with distance

Structure type (NASA-STD-7003A Figure 3). Seven curves of the percentage of the source value remaining against distance along the shock path, to about 3 m: honeycomb structure, longeron or stringer, primary truss members, cylindrical shell, ring frame, complex equipment mounting structure and complex airframe. They come from Kacena and others (1970), whose Table 3.1 describes the structures. The standard notes that these results apply to the peak value of the pyroshock response; the tool applies the chosen curve to the plateau. For the ramp it offers the constant velocity line curve of the design manual (Figure 3.8, which levels off near 15 %; the manual says the curve came from a truss and is representative of many structures), the spectrum ramp curve of the handbook's Figure 5.7, the same factor as the plateau, or none.

Spectrum ramp and spectrum peak (NASA-HDBK-7005 Figure 5.7). Two curves for point sources, attributed by the handbook to Barrett (1975), to 2.5 m.

Equation (A.2) of the standard (equation 5.78 of the handbook), for point sources on complex structure:

SRS(D2) = SRS(D1) exp[ −8 × 10−4 fn(2.4 fn−0.105) (D2 − D1) ],

with the distances in meters and fn in hertz. It attenuates high frequencies far more than low ones, so it needs no division into ramp and plateau. The handbook's example (section 5.3.4.2: 10,000 g at 1 kHz measured 0.3 m from a device, a device of twice the explosive weight, a component at 0.8 m) gives about 14,000 g after the energy scaling and about 4,100 g after the distance correction; the tool returns 14,142 g and 4,155 g.

Entered percentages for the plateau and the ramp, for program data.

When the source spectrum is known at a distance D1 from the device and the component is at D2, the chart methods use the ratio of the curve values at the two distances. Each curve holds 100 % from the source to its first point (0.12 to 0.15 m, in line with the handbook's remark that source spectra are measured 0.10 to 0.13 m from the device). Beyond the end of a curve the last value is held and a note is printed: no attenuation is credited where no data are charted.

How the curves were obtained

The documents print these curves as charts only. Figure 3 of the standard and Figure 5.7 of the handbook are vector drawings in the published PDF files, and the curves were taken from the drawing coordinates, calibrated on the axis tick marks; the reading error is below about 2 % of the value, small against the scatter of the data behind the curves. Figure 3.8 of the design manual exists only as a scan marked "not reproducible"; it was traced from a 1,600-pixel image of the chart, calibrated on its grid lines, and is good to about 1.5 % of the value (3 % at distances under 15 in, where the printed line is faint). All seven curves of Figure 3 were compared with the original charts of the design manual (Figures 3.1 to 3.7, read by eye on their grids to about 4 %), which confirms which curve is which. At 20, 40, 60, 80 and 100 in the original charts read, and the digitized curves give: honeycomb 87, 74, 61, 53, 45 % against 88.4, 75.3, 64.1, 54.2, 46.3 %; longeron or stringer 78, 52, 44, 41, 39 % against 75.9, 52.5, 44.1, 42.5, 40.4 %; primary truss (the original line is legible only from 47 in) 21, 15 and 10.4 % at 60, 80 and 100 in against 20.5, 14.8 and 10.9 %; cylindrical shell 55, 30, 19, 13.3, 10.6 % against 54.3, 29.4, 18.1, 12.8, 10.3 %; ring frame 45.5, 22.7, 13.9, 9.7, 6.9 % against 44.3, 23.0, 14.1, 9.6, 6.9 %; complex equipment mounting structure 58, 30, 15.7, 7.9, 4.0 % against 59.8, 30.0, 15.2, 7.5, 3.8 %; complex airframe 50, 22, 10, 4, 1.8 % against 52.5, 22.6, 9.7, 4.2, 1.82 %. The largest differences, about 5 % of the value (honeycomb and cylindrical shell at 60 in, equipment mounting at 80 in), are of the size of the reading error and of the redrawing in the standard; the tool keeps the curves as the standard draws them. The original charts run a little farther than the standard's (ring frame and honeycomb to 130 in, longeron to 120 in). The dashed curves 3, 6 and 7 of Figure 3 are cut at 2.65 m, where the printed lines end.

Joints

The handbook (section 5.3.2, after Barrett): 40 % per joint, up to a maximum of three joints, for the peak SRS value, and no attenuation for the up-ramp portion. The standard (A.3.2): Kacena and others suggest 20 to 75 % depending on the type of joint and how it changes the shock path; "others suggest a 50 percent reduction for a major discontinuity and 30 percent per normal joint up to a maximum of three joints and no attenuation for the up-ramp portion of the SRS". The design manual's own table (section 3.2): solid joint 0 %, riveted butt joint 0 %, matched angle joint 30 to 60 %, solid joint with a layer of different material in the joint 0 to 30 %. The tool offers the two published rules and an entered reduction per joint, on the plateau only or on the whole spectrum. Under the 30 % and 50 % rule the major discontinuities are counted apart from the three normal joints; the sentence in the standard does not settle that, and this reading is the tool's.

Source energy

Equation (A.1) of the standard scales a reference spectrum to a device of different energy: SRSn = SRSr √(En/Er) at the same distance. The standard cautions that energy beyond what the separation needs tends to go into a blast wave and not into the structure, so the rule may over-predict for a larger device and under-predict for a smaller one.

Maximum expected flight environment

Test criteria rest on the maximum expected flight environment (MEFE): by section 4.2 of the standard, from a transient analysis, an envelope of measured data or, preferred when there are three or more measurements, a statistical limit. The statistical limit is the P95/50 normal tolerance limit of the SRS values of a zone, taken as log-normal (Appendix B): with y = log10 of the SRS value, Ly = mean of y + k sy and the limit is 10 raised to Ly. Table 2 of the standard lists k for n = 3, 4, 5, 6, 8, 10, 15, 20, 30, 50 and infinity as 1.94, 1.83, 1.78, 1.75, 1.72, 1.70, 1.68, 1.67, 1.66, 1.65 and 1.64. In decibels the limit lies k s above the mean, s being 20 times the standard deviation of y.

The tool takes the prediction as the MEFE, or adds an entered allowance, or treats the prediction as the mean and adds k s for an entered number of samples and standard deviation (for a number of samples between two table entries the factor of the smaller entry is used). An attenuated envelope of a source is not a statistical sample; the P95/50 choice is there for the case where the entered source is the mean of measured spectra.

Test levels and tolerances

For externally induced shock, simulated on the test article, Table 1 and section 4.3 of the standard require qualification at MEFE + 3 dB ("or 1.4 × MEFE") with two shock applications in each axis, protoflight at the qualification magnitude with one application in each axis, and flight acceptance, when performed, at MEFE with one application in each axis. For self-induced shock, where the item's own devices are fired, no magnitude margin is possible and the standard asks for two actuations for qualification and protoflight and one for acceptance. The tool uses 103/20 = 1.413 for 3 dB. Both margins are inputs because programs levy their own.

Section 4.8.1 sets the tolerance on the maximax SRS achieved in the test: ±6 dB for natural frequencies up to 3 kHz and +9/−6 dB above, with the SRS computed at one-sixth octave or finer and at least half of the SRS magnitudes above the nominal specification. The tool draws that band around the qualification level.

Near field, mid field, far field

Section 3.2.3 of NASA-STD-7003A, given there as guidelines: the near-field environment is dominated by direct wave propagation, with peak accelerations above 10,000 g and substantial spectral content above 10,000 Hz; the mid-field combines wave propagation and structural resonance, 1,000 to 10,000 g and content between 3 and 10 kHz; the far-field is dominated by structural resonances, below 1,000 g with most content below 3 kHz. The category guides the choice of test method: pyrotechnic devices for the near field, mechanical impact or pyrotechnic devices for the mid field, and shakers, impact or pyrotechnic devices for the far field.

The tool reads spectral content from the knee frequency (the standard says a near-field spectrum has no knee below 10,000 Hz). The peak acceleration in the definition is the peak of the time history, which an SRS does not give, so that criterion is applied only when the peak is entered; it is then scaled by the plateau factor and the more severe of the two readings is reported. That combination is the tool's rule. MIL-STD-810H Method 517.3 paragraph 1.2.4 defines the three zones by what a test must control, with the same limits: frequency control up to and above 10,000 Hz for amplitudes greater than 10,000 g (near-field), from 3,000 to 10,000 Hz for amplitudes less than 10,000 g (mid-field), no higher than 3,000 Hz for amplitudes less than 1,000 g (far-field). It adds that distances from the device were left out of the definitions on purpose, because specific distances would imply particular structures and charges.

MIL-STD-810H Method 517.3 beside NASA-STD-7003A

Method 517.3 (Pyroshock) of MIL-STD-810H cites NASA-STD-7003A, NASA-HDBK-7005, Barrett (1975) and the Martin Marietta study as its sources for prediction, and agrees with the tool on every point of the attenuation: paragraph 2.3.3.1 is the source energy scaling of equation (A.1); paragraph 2.3.3.2 is the distance scaling of equation (A.2), with Figure 517.3-9 the same plot as Figure 4 of the standard; paragraph 2.3.3.4 gives the spectrum peak and spectrum ramp curves (Figure 517.3-11), the seven structure curves (Figure 517.3-12, described there as the attenuation of the peak amplitude of the time history response) and the joint rule of 40 percent per joint for up to three joints with no attenuation of the ramp. Its Figure 517.3-10 repeats the source spectra of four point-source devices, again as a chart only.

It differs in the test criteria, and the tool keeps NASA-STD-7003A as its default because the tool is a spacecraft and launch vehicle tool and the NASA standard states its margin as a requirement:

Margin. Method 517.3 prescribes no fixed qualification margin. With enough measured spectra it asks for a statistical limit (the 95th percentile with 50 percent confidence, as in the NASA standard); with too few it suggests enveloping the spectra and adding +6 dB to the envelope over the whole frequency range (paragraph 2.3.3.3 c). The allowance input of the tool can carry that 6 dB. Each of its five test procedures asks for at least three shocks, where the NASA standard asks for two per axis in qualification and one in protoflight and acceptance.

Tolerance. Paragraph 4.2.2 gives, as guidelines on the maximax SRS at one-twelfth octave: for Procedures I and II (near-field), −3 dB to +6 dB for at least 80 percent of the SRS values from 100 Hz to 20 kHz and −6 dB to +9 dB for the rest; for Procedures III and IV (mid-field and far-field with a mechanical test device) the same with 90 percent; for Procedure V (far-field on a shaker), −1.5 dB to +3 dB for at least 90 percent of the values from 10 Hz to 3 kHz and −3 dB to +6 dB for the rest; in every case at least half of the values above the nominal specification. That is tighter on the low side than the ±6 dB and +9/−6 dB of NASA-STD-7003A. The tolerance select of the tool draws either band.

Printing of equation (A.2). Method 517.3 typesets the exponent as "2.4fn −0.105" on one line; the NASA standard and the handbook print fn raised to the power 2.4 fn−0.105, and the handbook's worked example and Figure 4 confirm that form, which the tool uses.

What it assumes

That the shock travels along one path whose length and structure type can be named; that the attenuation of a spectrum can be expressed by one factor on its plateau and one on its ramp (or by a function of frequency in equation A.2); that the factors for distance, joints and source energy multiply; that the charted curves, drawn through the ground-test and flight data gathered up to 1970 by Kacena and others and up to 1975 by Barrett, carry over to the structure at hand (the standard adds that equation A.2 came from a point source on complex structure at sea level). The source spectrum is the user's: the tool ships no source data.

When it misleads

NASA-HDBK-7005 says of empirical models that the predictions may be in error by an order of magnitude at some frequencies, and that extrapolation is acceptably accurate only between similar structures and the same type of device. The curves stop near 3 m; composite structure, bonded joints, isolators and equipment mounted on soft panels are outside the data. Local resonances of the mounting structure can raise the response at a component above the level on primary structure, which no distance curve shows. The structure curves describe the peak response, and several of them (honeycomb, longeron) attenuate less at long distance than the velocity line curve does, so the choice for the ramp matters at low frequency: compare the options. The joint rules are round numbers from a few programs. The tool was built from Revision A of the standard (2011); a Revision B is cited in NASA papers and was not obtained, so the margins and tolerances should be checked against the revision a program invokes. A prediction made this way is a starting value for a test specification, to be replaced by measurements on the structure when they exist.