Chapter 2: Experiments Carried out in 1989

2.1 General Remarks on the 1989 Experiments

The experiments were performed in the open, in a travertine quarry (IMEG) near Montemarano (Tuscany, Italy), on an area of flat and soft ground which had been chosen to minimise secondary fragmentation which might normally occur when primary fragments collide with the walls or floor of a chamber.

The Sussex research team, on site during the experiments, consisted of Professor Giuseppe Martelli, Dr. Pamela Rothwell, Miss Christina Tindall (then a 3rd year project student), Mr (now Dr.) Sam Coles and the group's technician, Mr. Brian Blackman. Both Dr. Mario Di Martino and Mr. Walter Ferreri, from the Turin group, also helped during the campaign period, as did Dr. Fabrizio Capaccioni of the CNR, Rome. These experiments were carried out in Italy primarily because of the collaboration between the Sussex and Turin groups - the Astronomical Observatory of Turin provided essential technical support for this project as well as the NAC fast-framing camera, the full-time help of Mario Di Martino and the technical knowledge of Walter Ferreri, principal photographic expert at the Turin Observatory. The targets were disrupted each on different days between the 9th and 13th of September 1989 - a fortunate period of good weather.

A long-lasting professional relationship between Professor Martelli and the management of the IMEG travertine quarry has guaranteed an excellent level of technical support on site in terms of access, electrical power supply and heavy lift equipment (when required for the construction of shelters and blast shields).

2.2 Choice and Construction of the Targets

Spherical targets of 205 mm diameter were used throughout this experimental campaign. This is the size that had been used in previous series of experiments by the Sussex/Turin group (see e.g. Capaccioni et al. 1984), and the choice was based upon the size of the vacuum tanks used in these earlier experiments, since the targets needed to be small compared to the size of the tanks used in these experiments (~1 m³), but still large enough to provide macroscopic fragments.

The targets were constructed from strong alumina cement mortar, since this mixture gives a good approximation to the mechanical parameters of many natural rocks. The artificial rock, which has a density of 1.77 g cm^-3, consisted of alumina cement, Carborundum (SiC) and water in the correct stechiometric ratio. While liquid, the cement mixture was shaken at frequencies ranging from 10 Hz to 100 Hz in order to release any air or gas trapped as bubbles. Once hardened, the targets were 'cured' by leaving them in water for a minimum of 27 days. The fully cured targets were then zonally painted in different colours to facilitate identification of surface fragments and reconstruction after the experiment. The cement of each target had been dyed with different colours to make fragments more clearly visible on the ground and avoid confusion in case the collection of a fragment had been overlooked from a previous experiment.

The targets used in shots 2 and 3 contained a denser and harder central core of diameter 105 mm. It was constructed separately and placed within the target by the following steps:

1. Three small indentations were made in the core using a centre point drill, at a mutual angular separation of 60° around a notional 'equator';

2. Three smooth, thin and rigid metal rods were used to support the core within the mould which was to be used for the main body of the target, accurately positioned to within 1mm. These rods were locked securely in order to prevent any movement while the target cement was being vibrated during the drying/curing process;

3. Mantle cement was poured in from the top of the mould and allowed to solidify and cure as outlined above;

4. The metal rods were removed and a small amount of liquid mantle cement was used to fill the remaining holes.

The density of this core, 2.05 g cm^-3, is the highest which could be achieved by increasing the proportion of SiC without upsetting the stechiometric ratio (i.e. keeping the nature of the rock constant).

Table 2.1 gives the physical characteristics of the two types of target materials - classified as 'mantle' (that used in the homogeneous targets and outer mantle of the cored targets) and 'core' (used in the cores of those targets used in shots 2 and 3).

Table 2.1: Physical properties of the target materials

2.3 The Contact Charge Compared to Other Impacting Techniques

A number of different impacting techniques are available for carrying out catastrophic disruption experiments, each more or less suitable in particular ways for the aims of various research programmes. These are listed below, with examples of their use and a brief description of their relative merits:

• Drop method (e.g. Hartmann 1978, Nakamura et al., 1983): This technique involves quite simply dropping a projectile onto a target. Maximum velocity achievable with this technique is approximately 50 m s^-1 (even this velocity requires a drop of over 100 m if gravity alone is used) but the performance is totally reproducible and safe.
• Electromagnetic gun (e.g. Kawashima et al., 1983): Using a conducting sabot or carrier (which, in more recent designs, soon vaporises to form a conducting plasma) and two parallel current-carrying rails, this system uses the Lorenz force in a particular geometry to launch projectiles of a few grams at up to 10 km s^-1. As well as power problems and the required track length, significant difficulties associated with this technique are (i) the problem of 'shaping' the very high current (10-100 kA) pulse, the initiation of the conducting plasma discharge behind the sabot and the problems connected with the varying resistance of the rails.
• Powder gun (e.g. Mizutani et al., 1981, Matsui et al., 1982): Based upon conventional weapons, this technique uses the just the combustion of propellant in a powder chamber to accelerate the projectile. The powder chamber is separated from the projectile by a diaphragm, which is set to fracture at some predetermined pressure. This low-cost approach does allow reasonably sized projectiles to be accelerated to some km s^-1 but the maximum velocity attainable is restricted by the local sound velocity in the propellant gases, given (for an ideal gas) by where K is the Boltzmann constant, T the gas temperature, g the ratio of specific heats for the gas and m the molecular mass. As the projectile velocity approaches the driving pressure drops rapidly and any subsequent increase in the amount of propellant results in only a minimal increase in projectile velocity. The maximum practical velocity achieved so far is approximately 2.8 km s^-1 (Cerroni 1985). The performance of this technique is unacceptable in the low velocity regime (below ~100 m s^-1) due to non-reproducible frictional losses, which cause erratic variations in projectile velocity.
• Two-stage light gas gun (e.g. Seigel 1965): By coupling a conventional powder gun and a 'light gas' compression section, some of the problems associated with the simple powder gun can be overcome. The detonation of the explosive in the first stage causes compression of the light gas section, containing typically hydrogen or helium. The low molecular weight of this gas and the high temperatures combine to yield high projectile velocities. The maximum theoretical velocity attainable with this technique, for a vanishingly small projectile, is given by where is the local sound velocity in the (light) gas and g the ratio of specific heats, as above. In practical conditions (e.g. finite temperature, realistically < 5000 K) the maximum velocity does not exceed ~11 km s^-1. As well as requiring considerable maintenance, light-gas guns are non-portable and difficult to arrange in any direction other than horizontally due to their considerable size (typically ~10 m length) - this would be a distinct disadvantage in our experiments, which benefit from symmetry about a vertical direction.
• Modified shaped charge (e.g. Martelli and Newton, 1977): This technique involves a moulded shaped charge of plastic explosive (see Fig. 2.1), which is used to rapidly collapse and accelerate a metal liner to velocities tending to , where is the detonation velocity of the explosive used. The stability and consistency of the collapsing liner depends strongly upon the angle a (normally 30 - 45°), but in most cases a jet of liquid metal is produced with a gradient of velocities. In the modified shaped charge method the fastest drops of liner are allowed to escape, while the remainder is deflected by a secondary charge (appropriately delayed using an inert spacer). Although attractive because it is capable of producing projectiles at velocities up to 20 km s^-1 (Martelli, 1982), this technique suffers from two main problems; firstly the extremely accurate machining required to produce metal liners which perform consistently shot after shot and, secondly, the fact that it is impossible to accurately fix a priori the shape and mass of the projectile, which must be determined retrospectively from X-ray flash photography of the projectile in flight.

  

  Figure 2.1: A schematic of the modified shaped charge, capable of consistently producing velocities of the order of 10 km s^-1. The second, smaller piece of explosive is detonated sympathetically when the initial shock wave reaches it.

• Contact explosive charge (e.g. Martelli et al., 1991, Housen et al., 1991): Consistently capable of producing impacts equivalent to projectiles travelling at velocities of the order of 6 - 8 km s^-1, this technique has been refined and rendered qualitatively reproducible in recent years. It is also very portable and does not require significant power sources on site, but requires some ad hoc assumptions to be made regarding parameters (mass, velocity, etc.) of the equivalent projectile. This is the technique used by the Sussex/Turin group for the experiments discussed in this thesis, and is described in more detail below.

2.4 Details of the use of the Contact Charge

Holsapple (1980) carried out an experimental study of the equivalence between high- and hypervelocity impact by projectile and the effect of an explosive charge placed in direct contact with the target at the desired impact site. Although the study applied primarily to nuclear explosions and their TNT equivalents, the results are applicable to the plastic explosive more commonly used these days. The conclusion of this study was that it is straightforward to simulate a simple projectile impact at a specified velocity using a contact (or more precisely, buried) explosive charge, if the ratio of charge depth d, to charge radius r, can be controlled. For a projectile velocity of 6.2 km s^-1, which is the detonation velocity of the explosive used in these experiments, the requirement is that this ratio d/r should be 1.8±10% and that r should be approximately of the target diameter.

The Sussex/Turin group used the contact charge as follows: A small cavity in the base of the artificial rock target was filled with plastic explosive, as shown in Fig. 2.2. The dimensions of the cylindrical cavity were 26 mm (depth) by 25 mm. The target, with contact charge in place, was placed at the centre of a specially constructed steel table of height 1m. This table had at its centre a small hole, though which the detonating cable was passed.

Figure 2.2: Contact charge technique used by the Sussex/Turin group in 1989 and 1992.

Figure 2.2 shows the experimental arrangement with steel table, artificial rock target and contact charge. The explosive used, Gelatin 2b, has a detonation velocity of 6.1 km s^-1 and an acoustic impedance similar to that of the target (»7×10⁵ g cm^-2 s^-1).

A ballistic pendulum was used to find the energy delivered by the explosive, and hence the momentum delivered by an equivalent projectile. This pendulum consisted of a square block of steel weighing 4.7 kg, which had been provided with a cavity identical to that in the base of the target. This block was suspended by four wires, each 6 m long. The cavity was filled with explosive and upon detonation of the explosive, the height gained by the pendulum was recorded by means of fast-framing cameras. Since the impact velocity of the equivalent projectile is fixed (at the detonation velocity of the explosive), the mass is uniquely determined, in this case as 1.85 g.

The contact charge method provides a fairly realistic simulation of a rock-rock impact (see e.g. Housen et al. 1991) because the acoustic impedance of the artificial rock target is well matched by that of the exploding charge located against the target itself. This conclusion has been reinforced by a study of the morphology of some of the fragments coming from the region in direct contact with the explosive charge; in some cases there were clearly evident conical impact structures, as shown in figure 2.3. These formations are very similar to those found in rock which has experienced impact cratering and fragmentation (see e.g. Fujiwara 1985), when the impact pressure exceeds approximately 5 MPs. This is a result of the shear induced by the forward compressional wave and the transverse stress applied to the edges of the transient crater.

Figure 2.3: Two fragments (with an end-on view of the second) originating near the contact point, clearly showing the conical impact structures often found following 'real' impact upon rock targets. The two fragments are both approximately 3 cm in length.

2.5 Fast-Framing and CCD Camera Arrangements

Target disruption was recorded using a single NAC E-10 16 mm fast-framing camera, together with two fast-shuttered CCD TV cameras for quick on-site analysis. The NAC cine camera, using Eastman/Kodak 7222 "Double-X" Negative film, was running at 700 frames per second in all the 1989 experiments. This frame rate, considered sufficient to catch the fastest fragments of observable size (~1 cm), was chosen based upon an estimate of the observed velocity of ejected fragments from other similar experiments (see e.g. Capaccioni et al., 1984) and taking into account the field of view available with the geometric arrangement used here. This choice was soon confirmed to have been well made - Martinsson (1990) indeed reports that the maximum speed observed in the fragment and dust envelope from these experiments is of the order of 20 m s^-1 in our experiments, and our own later analysis of the film suggested no fragments moving too fast to be detected. Figure 2.4 shows the arena arrangement used in 1989.

Figure 2.4: 1989 Arena arrangement (not to scale) showing the three cameras, target table and the mirrors used to reflect views to the CCD camera equipment

2.6 Data Collection at the Test Site

The ground surrounding the experimental table, out to a distance of 30m, was divided into 24 sectors of 15° each - this allowed us to measure fragment positions on the ground, as they were collected, to within a few cm of radial distance and within 1 degree of angle. In order to study the distribution of fragment ejection angles, all fragments which could be found after the shot were collected in plastic bags and labelled with a letter and number code corresponding to their position on the ground.

2.7 Transcription of the NAC 16mm Cine Film to Video

The four 100 foot reels of film (one recorded for each shot) from the NAC fast-framing camera were processed in Italy by Mr. Walter Ferreri, photographic expert at the Turin Astronomical Observatory. On their arrival at the Sussex laboratory, the films were transcribed from the 16 mm cine format to PAL video tape for the digitising, using the apparatus shown in Fig. 2.5. By playing the 16mm film at standard speed and recording it on a video system, the film sequences were effectively slowed down from the original 700 frames per second to only 24. Given that PAL video has a standard 50 fields per second (where sequential 'odd' and 'even' fields form TV frames) there was no doubt that the cine film would be fully captured. The intention was to grab only one from each pair of fields, thus building a digitised set of all original frames.

Figure 2.5: Apparatus used to transcribe the 1989 cine films (hardware not to scale).

This work was all carried out in our laboratory at Sussex. The room lights were switched off during transcription, and the skylight was covered using black plastic. A standard Fumeo 16mm cine projector (item 1 in Fig. 2.5) was used to project the NAC images onto a matt white board, item 6 above. The CCD camera used (item 2 in Fig. 2.5) was a Philips LDH 0460/01 monochrome video camera incorporating a manual gain control (MGC). The lens (item 3) attached to the camera was a Sony "TV Zoom Lens" type 816358 with aperture at maximum (1.8) and ¦ = 20 mm, chosen to give a full-screen picture of the projected image. A Kodak Wratten-Gelatin ND 1.00 filter (item 4, giving an attenuation factor of 10^1.00 = 10) was placed in front of the zoom lens and in front of this was an HA3 infra-red filter (item 5) which served the purpose of minimising infra-red glare from the hot-filament projector.

Several 'dry runs' (i.e. without the video recorders running) were carried out with each reel of film, carefully monitoring the CCD output using the oscilloscope in order to find the optimum setting (of 1V peak-to-peak on the video signal) for the MGC. The output from the CCD video camera was then fed through a video timer and recorded separately on both VHS and Betamax machines. This had the advantage that, since both generate standard PAL video signals, either could have been used as input to the video digitiser. Two 'originals' of each resulting video cassette were made, with the total footage on each tape (i.e. all four shots) being about 15 minutes.

Although the system might seem unsuitable, with the projector running at 24 frames per second and the CCD camera recording 25 per second, this did not cause any problems. Clearly all frames of cine film would be caught, with one in 24 being caught twice. It was very clear from the computer-animated sequences just when this had happened - it did not occur within the early frames for any of the 4 shots (see Section 4.3), and thus was not a hindrance.

2.8 Direct Measurement of Fragment Size & Shape

Before any analysis was made of the NAC high-speed film, the size and mass of fragments from shots 1 and 3 were all measured by hand. The size was measured to an accuracy of 1 mm using a pair of digital callipers and the data entered, together with the mass, into a simple database program called ROX-INP which I myself wrote as part of a 3^rd year project at Sussex (Giblin, 1989). The fragment axial sizes were measured according to the top-down method, that is to say the largest dimension was found first and labelled a, then the largest dimension perpendicular to this was found and labelled b, and finally the fragment diameter was measured perpendicular to both of these and labelled c. Fragment masses were measured using a digital balance, to an accuracy of 0.01g. Although shots 1 and 3 were not covered by these measurements, this data provided an essential comparison to the measurements made from the digitised film records, as described in Chapter 4. Table 2.2 summarises the results of these measurements, including the total recovered mass as a fraction of the original target mass.

Table 2.2: Details of in-hand measurements made upon recovered fragments

The data files created by the ROX-INP program was readable by the software developed for this project, and the data has been used wherever possible. Appendix 6 contains a listing of all fragment data, both from the NAC film records and from the in-hand measurements.