TWO NEW METHODS FOR DETERMINING FRICTIONAL SENSITIVITY O F EXPLOSIVES: OBLIQUE IMPACT

AND SHEAR FRICTION TESTS*

I. B. Akst Mason dr Hanger-Silos Mason Co., 1nc.i

Amarillo, Ter.

INTRODUCTION

Like social scientists and poets, people who study high explosives think a lot about sensitivity. They wonder about and explore the stimuli necessary to evoke responses, and are concerned with the nature of those responses. And although interest in sensitivity is as old as explosives, there are still new thoughts about it, some of which are proving to be more accurate if less poetic, with mysticism and necromancy gradually giving way to prediction and control. In the last number of years, it has become clearer that sensitivity of explosives

can not be expressed by a single number or ordered list, but should be separated into categories, each of which broadly defines the kind of stimulus to which the high explosive (HE) is sensitive. For example, there is sensitivity to shock, where the essence of the stimulus is the fast rise of a wave front involving mechanical energy. Other categories are thermal sensitivity, electrostatic sensitivity, and sensitivity to electromagnetic radiation.

But our principal concern will be with mechanical sensitivity. The best known measure of this is drop-hammer or impact sensitivity, characterized by a weight of a few kilograms dropping vertically onto, typically, a few milligrams of explo- sives. Other mechanical sensitivity tests include responses to friction, such as in the pendulum test. We shall endeavor to show that when one combines impact with friction and increases the scale, new information may be elicited which is different from or goes beyond that which is easily derived or derivable at all from either classical test alone.

The work to be described has been on two tests. One is nicknamed the Skid Test. It specifically measures the oblique impact sensitivity of large bare solid charges. The other we call the Shear Friction Test. It specifically measures the reactive response of moderately small samples of bare explosives subjected sud- denly to sliding friction while under high normal forces.

The results to be presented will show the responses and some of the effects of the kind of explosive, its mechanical strength, particle size and some other fac- tors; and we shall take a short exploration into the mechanism of initiation.

DESCRIPTION OF SKID TEST

First it is necessary to describe how these sensitivity measurements are made. The skid test requires an outdoor site well removed from population, because it often involves the detonation of 25 pounds or sometimes even 50 pounds of high explosives, along with some debris scattered by this detonation. A trifilar pendu- lum is arranged from long poles so that a hemispherical sample freely supported in a wooden ring can be dropped in an arc, the equator staying parallel to the ground, with the curved surface free to impact the target. The spherical surface

Work performed under the auspices of the U.S. Atomic Energy Commission. t A prime contractor of the U.S. Atomic Energy Commission.

636

Akst : New Ways of Finding Frictional Sensitivity 637

localizes energy reproducibly without the need for high precision in the drop. By appropriate length of cable and placement of the target, one can select various angles of impact. By means of another pole, the wooden ring and sample are hoisted remotely and then allowed to drop by means of a solenoid release.

The target is a hard, gritty surface, made to resemble rough concrete but standardized for reproducibility. Graded white silica sand is very thinly silver plated by chemical deposition for timing reasons. This sand is bonded by an electrically conductive epoxy to a steel plate, which is in turn bonded by a sand- filled epoxy to a large concrete pad placed on the ground.

The sample itself usually consists of a solid hemisphere of about 25 pounds ( 1 1.3 kg), which comes out to be 5.5 inches (14 centimeters) in radius with high-density explosives. The size is kept constant in order to fit the ring, etc., allowing the weight to vary with density. Sometimes a 50-pound charge (7" or 17.8 cm radius) is used. Three fine wires (.004" or 0.1 mm diameter) are stretched across the polar region of the hemisphere to make contact with the silver-plated sand and thus obtain an accurate measure of time from first touch to reaction. The onset of reaction is determined by photodiode, pressure gages or by photography. When light is produced, the accuracy of the timing is on the order of a few microseconds.

SKID TEST RESULTS

The principal data obtained are the height from which reaction occurs, and the kind or type of reaction. The latter is derived principally from the high-speed films, supported by auxiliary measurements made of blast pressure, and by visual evidence regarding crater, amount of explosive left, etc. The height of reaction is the lowest height from which reaction occurs at least once, there being no reaction in at least one test from the next lower height. Each height is related to the one on either side by d2, e.g., 2.5, 3.5, 5 , 7.1 and 10 feet. As mentioned above, the time from first touch to reaction is measured. The time is also obtained, less accurately, by counting frames from the flash of a fiducial light to light or other evidence of a reaction in the high-speed pictures. The fiducial light, which signals the start of the event, is a small photoflash bulb fired by a high-energy capacitor discharge unit triggered by the first touch of the fine polar wires on the silver sand. (The first flash of the bulb is very rapid, being followed by the burning photoflash wire fill.) The time from first touch to reaction averages about a millisecond and is rarely less than half a millisecond or more than 1 % milli- seconds.

The reaciions are classified as mild, moderate, or violent. They are numbered 1 through 6; 1 and 2 are mild, 3 and 4 are moderate, and 5 and 6 are violent, with 6 being detonation. Reaction 1 is merely a scorch mark on the target pad. Reaction 2 is a puff of smoke, with no light and very little or no blast pressure; it often does not break the explosive charge. Moderate reactions 3 and 4 produce light and consume a small to moderate proportion of the HE; blast pressures are typically less than one pound per square inch at 40 feet. Reaction 5 is quite violent, consuming essentially all the explosive, and producing intense light and several psi blast pressure.

Early in the development of the test, we decided to investigate the effect of impacting at different angles. Since it was' immediately clear that sensitivity was considerably greater at 4 5 O than at 90" from the horizontal-that is, in vertical impact-it was obvious that at still lower angles, sensitivity might be even higher. Thus 26", 1 4 O , and 7O, as well as the higher angles of 63' and 76' were chosen

63 8 Annals New York Academy of Sciences

to provide integral ratios of horizontal to vertical instantaneous force vectors, e.g., 4: 1 at 14O, 8: 1 at 7O, and 1 :2 at 63’. It was found that sensitivity was greatest at the 1 4 O angle, being less at 7”, 26’, 45O, and above. Sensitivity in this case means height at which reaction occurred, as may be seen in TABLE 1 . I t seems that as one approaches grazing incidence, which resembles most of the standard friction tests, sensitivity goes down, This may help to explain why the standard frictional tests fail to give any results on many of today’s secondary HE formula- tions. Note that the kind of reaction is not a function of the impact angle.

The angle effect was investigated on a few explosives with the same results. The test is now standardized at 14’ and 45’ from the horizontal.

Another study of oblique impact sensitivity had to do with some effects of composition. TABLE 2 shows the results from a number of explosives. These all have only the explosives known as HMX or RDX as the “solid” portion of the matrix, but they are formulated with different binders, different amounts of binders, and with various particle size distributions of the RDX or HMX. Note the differences in sensitivity in terms of both the height and the kind of reaction. For example, PBX 9404, which is a pressed plastic-bonded explosive (PBX) having 94% HMX with plasticized nitrocellulose, will detonate when dropped from a height of 5’ to impact at 45” or from 1.75’ when impacting at 14’. Com- pare 9404 with 9010, another pressed PBX, 90% RDX bonded by a chlorinated fluorocarbon: it will detonate in drops from 3.5’ at 45’ or 1.25’ at 14’. A third HMX-rich PBX (LX-04) containing approximately 85% HMX in a fluoro- carbon binder gives only mild to moderate reactions (2’s and 3’s ) . Still another

TABLE 1 EFFECT OF IMPACT ANGLE ON OBLIQUE IMPACT SENSITIVITY

HE Angle1 Vector Ratio’ Height ( f t ) Reaction

PBX 94049 I

14

26.5

45

76

3.5 5.0

1.25 1.75

8/1

4/ 1

2.5 3.5

2/ 1

3.5 5.0

1/1

3.5 5 .O

1/4

0 6 0 6 0 6

0 6

0 6

0.88 0 1.25 6

2.5 0 3.5 6

PBX 90104 14 4/ 1

45 1/1

3.5 5.0

Comp B-35 14 4/ 1 0 2

45 1/1 20.0 0

In degrees, from the horizontal.

94% HMX, 3 % tris-p-chlorethyl phosphate, 3% nitrocellulose.

60% RDX 7 5 ~ median particle diameter, 40% TNT.

* Instantaneous at impact, horizontal to vertical.

* 90% RDX, 10% chlorinated fluorocarbon (Kel-F 3700).

Akst: New Ways of Finding Frictional Sensitivity 639 TABLE 2

EFFECT OF COMPOSITION ON OBLIQUE IMPACT SENSITIVITY

HE Composition Height

Angle ( f t ) l Reaction

PBX 9404 94% HMX, 3% chlorethyl phosphate, 3% NC 14 1.75 6 45 5.0 6

PBX 9010 90% RDX, 10% Kel-F 37002 14 1.25 6 45 3.5 6

LX-04 85% HMX, 15% Viton A3 14 2.5 3 45 5.0 3

PBX 9011 90% HMX, 10% Estane 57404 14 20.0 2 45 20.0 0

Lowest height in feet from which any reaction is normally observed, there being at least one test at the next lower height.

a A chlorinated fluorocarbon by 3M.

* A polyurethane rubber by Goodrich. A fluorocarbon by du Pont.

HMX-rich PBX, having 90% HMX in a resilient polyurethane rubber, gives no reactions at either angle up to 20f.

Thus it may be seen that differences are clearly established in sensitivity, and the test is able to differentiate various formulations and kinds of basic explosives in terms of both the height at which they will react and the reactions involved.

Having thus established that there are important differences in oblique impact sensitivity among explosives and explosive formulations, and that this test was able to distinguish them, it became of interest to discover how variables in the explosive itself might yield changes in reactivity or height sensitivity. TABLE 3 gives some of these studies. Formulation RX-04-AB and LX-04 differ, for ex- ample, only in the particle size, and one can see the difference in reaction type, whereas the height sensitivity stays fairly constant. One may also see the rela- tively smaller effect of changing the proportion of HMX over a moderate range, the compositions (RX-O4-P1, LX-04 and LX-07) otherwise being the same; here there are differences in height and in reaction also. Tests were also made of the effect of strength (compressional strength or compressional modulus of elasti- city). It may be seen that the stronger or more rigid the explosive, the more sensitive it is.

TABLE 4 compares the sensitivity derived in the skid test with other sensitivity measurements. Note in comparing height sensitivity, drop-hammer impact sensi- tivity, and shock sensitivity that the order or ranking is generally the same, but that there are important exceptions.

SKID TEST MECHANISMS

In efforts to elucidate the initiation mechanisms in oblique impact, a few meas- urements were made of the area of contact between charge and target as a function of time. This was done by measuring electrical resistance versus time, by the short-circuiting lines of a silver raster applied to the polar region of the hemispheres on the silvered sand, as crushing or abrasion took place. The curve obtained was something of a surprise. Instead of a relatively slow compressional increase in area followed by a sudden drop-off, as one might expect in a rebound,

640 Annals New York Academy of Sciences

TABLE 3

EFFECT OF SOME HE VARIABLES

HE Variable Magnitude Angle Height Reaction

RX-04-AR Type of HE RDX 14" 2.5' 6 45" 5 .O' 6

LX-04 HMX 14" 2.5' 3 45" 5 .O' 3

Rx-04-PI % HMX 80 14O 3.5' 2 45"

LX-04 85 14" 2.5' 3 45" 5 .Of 3

- -

LX-07 90 14" 2.5' 4 45" 3.5' 4

LX-04 Particle Size Small 14" 2.5' 3 45" 5.0' 3

RX-04-AB Large 14" 2.5' 6 45" 7.1' 6

RX-04-AH Strength Soft 14" 2.5' 2 45" 10.0' 3

RX-06AQ Rigid 14" 1.75' 4 45" 2.5' 5

-

more than three-quarters of the final area was in contact in about one-tenth of a millisecond. The area grew slowly after that, staying over three-quarters in con- tact for something over a half millisecond, then dropping off slowly for another half millisecond or so. The final area, as measured by this system (which used inert hemispheres) agreed with the area measured mechanically on explosive charges which did not react. The abraded area varied from 0.43 im2, in a drop from 0.6' at 14O, to 1.86 in.2 in a drop with a weaker material from 20' at 45'. Impacts from 14' tended to produce ellipses, with the major axis in the plane of fall, while the 45' impacts produced circular abrasions generally larger than those at 14' but changing less with changes in height. Differences in strength were also observed, producing differences in amount abraded.

TABLE 4 COMPARISON OF KINDS OF SENSITIVITY

HE

Oblique Impact

450 1 4 O Impact1 Shock2

50% Height, cm 50% Gap, mils

PBX 9404 5 .0' 1.75' 40 90 PBX 9010 LX-04 PBX 901 1

3.5' 5 .o'

)20.0'

1.25' 2.5'

20.0'

35 60 60

85 75 25

Comp B-3 >20.0' 5.0' 60 50 Cycloto175/253 >20.0' c2.5' 65 15

NOL type drop-hammer, 2.5 kg hammer, on sandpaper, approximate. a 41" LASL type, gap in mils of brass, approximate.

75% RDX, includes coarse CIass D; 25% TNT.

Akst: New Ways of Finding Frictional Sensitivity 64 1

If from this measure one considers the volumes abraded from the hemisphere and considers various feasible lossless distributions of the input energy (calcu- lated from height of fall) through this volume, one can impute a temperature increase of 65" to 1300OC. The latter assumes 25-pound dropping 1.75 feet, with distribution of half the energy over one of the smaller areas and through a thick- ness of 0.001". Going to the work of Wenograd,* and making some guesses for RDX and HMX (not investigated in the reference), a temperature of some- where between 700" and 4OOOC should be necessary to cause detonation in the length of time observed in these events, namely about % to 1% millisecond. Therefore, one may tentatively conclude that the initiation process is a fairly straightforward thermal one. A small quantity of mechanical energy is trans- formed into heat energy and is distributed over a small enough volume in a short enough time to raise the temperature of a small amount of HE to its ignition point. In some cases, energy loss must then be so rapid that the reaction goes out, resulting in the grade 2 reaction. Or, if reaction energy production exceeds losses, deffagration can take place, making the transition to detonation in some cases. Obviously, the exact mechanisms of the energy transformation and dis- tribution-which must be greatly different in oblique impact from what they are in vertical impact-and the ignition and growth processes, are not at all simple. But understanding this much can perhaps start one on the path toward modifying the explosives or their surroundings to decrease sensitivity.

SHEAR FRICTION TEST Some aspects of the skid test have been very difficult to instrument (for ex-

ample, the contact area versus time measurement described above required a great deal of effort to get a few data points). Because of that and because we wish to explore the initiation mechanisms further, an apparatus has been designed and built to measure frictional sensitivity under normal forces in a more linear fashion. In this apparatus, small flat samples of explosive are subjected to pres- sures up to 7,000 psi; a falling weight then snatches a grit-surfaced steel strip from underneath the sample. The steel strip is backed by a strip of Teflon, and

* WENOGRAD, J. 1960. The behavior of explosives at very high temperatures. The 3rd Symposium on Detonation. Princeton, N. J.

TABLE 5 SHEAR FRICTION SENSITIVITY REACTIONS~

Normal Pressure (psi)

1,OOO 2,000 3,000 4,000 5.OOO HE

PBX 9404 2 5 5 5 5 PBX 9010 1 2-3 4 5 5 LX-04 0 2 3 3 3 PBX 901 1 0 2 3 3-4 4 Comp B-3 0 0 1 2 3

Reaction types are similar to the Skid Test: Mild Reaction 1 Scorch and

2 Smoke Moderate 3 Light (smoke, burning)

4 Deflagration Violent 5 Deflagration (all HE consumed)

6 Detonation

642 Annals New York Academy of Sciences

the surface against the explosive is normally coated with sand bonded to steel with an epoxy, very much as in the skid test.

Instrumentation includes time-resolved curves of normal force, steel strip pull force, and position and velocity of the strip. This test is still undergoing develop- ment, problems having been posed by the attainment of proper velocity and satisfactory reactions as well as by the instrumentation.

Some of the same explosives studied in the skid test have been tested, with the results shown in TABLE 5 . Note that in general they follow the same pattern as those in the skid test; but there are reversals. For example, the explosive having HMX and polyurethane, which had the lowest oblique impact sensitivity, reacts more easily than some other formulations in this test.

Some curves have been obtained showing the variation of normal force or pressure with time; there are also dynamic records of pull force. And of course, if one divides those two forces, one can get dynamic coefficients of friction, as shown in TABLE 6, which includes a comparison with the static or low-speed coefficients of friction.

TABLE 6 COEFFICIENT OF FRICTION

Normal Force Dynamic on Grit Low-Speed on 16 RMS Steel Material (lb) R 5 0 0 in/min ,400 in/min

G.02 sec 0-0.2 sec 0.8-1.0 sec PBX 9404 500 - .35 .6

1000 .3 .4 .6 LX-04 500 - .3 .4

1000 .5 .3 .4 Teflon 3000 .09, on smooth steel - -

We expect that the shear friction test will add to the understanding of frictional initiation. It has thus far demonstrated that good reactions can be obtained and that explosives can be differentiated. I t is not expected to replace the skid test, but it should be able to augment it at least by screening explosives early in devel- opment when only small quantities are available, and when data to aid decisions as to further development are particularly valuable.

In addition to information of potential academic value, these tests, especially the skid test, have provided information directly relatable to safety, by providing data which require little extrapolation to possible HE handling circumstances: moderate size charges, uncased; drops from ordinary working height and lower; concrete floors. This information can be and has been useful in developing new explosives formulations as well as reducing danger of reaction by making the target something softer than concrete.