«Please cite this article as: Mendes, R., Ribeiro, J.B., Loureiro, A., Effect of explosive characteristics on the explosive welding of stainless steel ...»
The angle of collision and the impact velocity are known to depend on parameters such as flyer plate thickness, stand-off distance (initial distance between the flyer and the parent or base plate), flyer geometry, explosive confinement (if any), the explosive to flyer mass ratio and the explosive energy. The collision point velocity is essentially determined by the properties of the explosive used to accelerate the flyer plate. So, once one of the limits of the weldability window determines that, for jet formation, the velocity of the collision point should be lower than the sound velocity, the use of high-explosives (such as plastic explosives, presenting detonation velocities of 6000m/s) is not possible in explosive welding.
Low velocity explosives are therefore needed for explosive welding. Moreover, fine tuning the welding conditions requires a proper control of its detonation velocity. Ammonium nitrated-based emulsion explosives, which are based on a mixture of emulsion matrix with a sensitizer agent, are among the explosives which are suitable given this requirement. Several authors have shown ;
 that the emulsion explosives can meet that requirement when their density is reduced. This is achieved by increasing the amount of sensitizers, such as hollow glass or polymer microballoons, in order to reduce the detonation velocity. Furthermore, emulsion explosives ensure low cost, water-resistance and a high level of safety and reliability. However, no studies on how the characteristics of the explosives and of the detonation process affect the weld quality are extant.
The aim of this research is to study the influence of the explosive characteristics, in particular the type, amount and size of sensitizers on the interfaces features of a stainless steel to a low alloy steel weld in a cylindrical configuration. Moreover, the weldability window for the stainless steel to low alloy steel system was tested for the cylindrical configuration.
In this study, an ammonium nitrate-based emulsion, sensitized with different amounts and types of sensitizing agents, and Ammonium Nitrate Fuel Oil (ANFO) explosives were used to perform stainless steel AISI 304L tube to low alloy steel 51CrV4 (spring steel) rod welds.
2. Experimental procedure
2.1. Materials Conventional explosive welding was conducted in cylindrical arrangement between TP 304L (ASTM A312) stainless steel tubes and 51CrV4  spring steel rods using ammonium nitrate-based explosives. The chemical composition of steels can be seen in Table 1. The experimental setup is illustrated in Fig. 3.
Table 1 – Chemical composition of the materials to be welded.
Stainless steel tubes (18 mm diameter, 1.5 mm thick and 215 mm long) and rods (11.0 mm diameter and 210 mm long) were used. The internal surface of the tubes and the external surface of the rods were polished with abrasive paper of 500 grits, in order to reduce roughness of the surfaces. Rods were centered inside the tubes in order to create a uniform annular standoff distance of 2 mm. Concentric with both tube and rods a polyvinyl chloride tube (internal diameter 48 mm and length 290 mm) was used to contain the explosive. Tabs were used to maintain the distance between the components of the experimental apparatus. Four ionization probes, positioned at fixed distances, were used in each setup to measure the detonation velocity. Five different ammonium nitrate-based emulsion explosives (ammonium nitrate (84%), water (10%) and oil plus emulsifier (6%)) were considered for this study. Three of those compositions were sensitized with different amounts (15%, 11% and 6%, w/w) of hollow glass microballons [HGMB], with an average diameter of 70 µm. The fourth and fifth were sensitized with different amounts (2% and 1.5% w/w) of expanded polystyrene spheres [EPS], with an average diameter of 1 mm. A sixth explosive composition, resulting from the mixture of ammonium nitrate (94%, w/w) and fuel oil (6%, w/w), known as ANFO, was also used. The weld experimental series and the density of the explosive used in each test are shown in Table 2.
Table 2 – Weld series and density of explosives used.
Fig. 3 – Schematic representation of the welding setup used in this study.
2.2. Detonation velocity measurements Detonation velocity, defined as the velocity at which a detonation front propagates through a given explosive, is one of the most important detonation parameters. The detonation velocity is, by far, the easiest to measure. The simplest way to determine its value is to measure the time that the detonation wave [DW] takes to travel between two given points. To achieve this objective the experimental technique should be able to detect the DW arrival at those predefined points and measure the interval between successive events of that kind with a time resolution of several nanoseconds. In this case, we have taken advantage of the highly ionized detonation products at the detonation front to close, at each ionization probe, an open electrical circuit branch.
The resulting immediate changes in the circuit created by the closure of the branches at each probe are then recorded in an ultra-fast signal recorder, like a digital oscilloscope (in our case, a Tektronix DSA 602, with a time resolution of 1 ns). Fig. 4 shows a typical oscillogram obtained with this technique.
Fig. 4 - Typical example of a result used to determine the detonation velocity.
2.3. Microstructure and hardness Specimens for microstructure analysis were extracted from the beginning and the end of welded coupons in planes parallel and perpendicular to the direction of propagation of the detonation wave. Specimens were ground and polished, according to ASTM E3-11, and electrolytic etched at 5 V for 45 s, using a solution of 15 g of oxalic acid in 150 ml of hydrogen peroxide. Microstructure examination was carried out using an optical microscope Zeiss Axiotec 100 HD and a Scanning Electron Microscope Philips XL30 with EDS. The elemental map distribution was achieved using an electron probe microanalyser Camebax SX50.
HV0.05 microhardness measurements, according to ASTM E384, were taken in longitudinal and transversal sections of the welds, in order to sample very small regions in the weld interfaces.
3. Results and discussion
3.1. Collision velocity and angle During the welding process, the points A, B and C (see Fig. 1) form an isosceles triangle since the linear distance along the flyer and along the base plate, between the collision point at time zero (C) and the collision point at time t (when A hits B) must be the same. This means that the distance from C to A is equal to the distance from C to B. Once all the distances are directly related with velocities, it is possible to write the Eq. (5). In this very common case, when β 10⁰ is possible to simplify the Eq. (5) to the Eq. (6) with a minor error.
From the three parameters of Eq (5), the collision angle β, is the independent variable, the detonation velocity Vd is measured as described in section 2.2 and the impact velocity Vp is estimated from the Gurney Theory  adapted for converging configurations, as shown in Fig. 5, by equation (7) .
Fig. 5 – Schematics of converging configuration considered for impact velocity evaluation. M – Flyer tube; C – Explosive; N-External confinement.
and where M, C and N are the mass per unit of length of the flyer tube, the explosive and the external confinement, respectively. In the equation (7) 2E, is the Gurney velocity, which according to Kennedy  is approximately 60% of the heat of explosion of the explosive composition.
Once the value of the impact velocity Vp given by the Eq. (7) corresponds to the terminal velocity, it is necessary to take into consideration its acceleration, in order to estimate the impact velocity at the point where the flyer tube collides with the inner cylinder, For this purpose, as suggested by Chou and Flis, , an exponential acceleration history, as given by the Eq. (10), was considered.
Where PCJ is the Chapman-Jouguet detonation pressure and c1 and c2 are empirical constants.
Those constants were calibrated so that, as is typical for explosive metal flyer acceleration (see for example [52; 53]), after eight shock wave reflections its velocity reaches 80% of the Gurney Value (Vpcal = 0.8 VpGurney). The time required by the flyer to achieve that velocity TVpcal was evaluated considering that, during the reflections, the shock wave propagates within the flyer at the longitudinal bulk sound speed of the steel Cb, (Cb = 4.496 mm/µs vd. , pag. 214), for a distance equivalent to eight times the flyer thickness and a time of 2.34 µs.
The calibration procedure is illustrated in Fig. 6 for the experiment “Weld 1”. For this particular case, the time required to achieve 80% of the Gurney Velocity (Vptarget = 0.8 x 0.671 mm/µs) was TVptarget = 2.34 µs. The empirical constants, c1 and c2 of the Eq. (10), were then chosen, so that flyer velocity matches that value at that time. Once calibrated, the constants c1 and c2, the application of the Eq. (7) and (9) and the use of a simple integration procedure of the velocity enable the results shown as open circles in the graph of Fig. 6. Given the initial gap between the flyer and the base rod of 2 mm, the collision velocity can be read from the same graph (now from the open diamond curve) for that travelling distance.
The measured detonation velocity Vd the calculated (according to the aforementioned procedure) impact velocity Vpcal and the collision angle β as well as the parameters used in the calculations, are shown in Table 3.
Fig. 6 – Procedure for calibrating of the constants c1 and c2 of Eq. (11) and evaluating the impact velocity.
Table 3 – Calculated values of Vd, Vpcal and β, as well as the parameters used for the calculations.
* ΔH is the heat of explosion. Values taken from Sanchidrián and Lopez.
3.2. Analysis of the welding conditions As described in the introduction, a weldability window was built using the equations 1 to 3 and a limiting collision velocity equal to the lower bulk sound speed of the materials to be welded. In this case, that value was found to be the one for stainless steel 304 L (4496 m/s ). The values used in the calculations are shown in Table 4, and the weldability window in Fig. 7. The conditions used in each of the welds are also plotted on that graph. As seen in that Figure, five of the six performed welds are slightly over the upper weldability limit. According to the literature, (see Fig.
2), those welds may present excessive melting. It is also possible to observe (see also Table 3) that, despite sharing the same basic energetic material (ammonium nitrate), it is the nature of the explosive composition, rather than its density, that mainly determines one of the welding parameters - the collision point velocity. In fact, although the explosives used in welds 1 and 6 present similar densities (0.774 and 0.760 g/cm3, respectively), they show quite different collision point velocities (3667 and 2370m/s, respectively).
Table 4 – Values of the parameters of equations (1) to (3) used for the calculation of the weldability window. All values taken from references [55; 56] unless stated otherwise.
*In Eq. (2) the hardness of the base plate is expressed in [N/m2]. **In the Eq. (3) the density of the flyer plate is expressed in [g/cm3]. ***.
Fig. 7 – Weldability window for the stainless steel carbon steel combination and experimental conditions verified for the performed experiments.
3.3. Morphology of the outer surface of the welded samples Fig. 8 shows the surface appearance of four of the welded samples. Figs 8a and 8b illustrate welds made using explosive sensitized with 15% (weld 1) and 8% (weld 3) of HGMB, respectively. The images show that on one hand weld 3, carried out with a higher collision point velocity and a smaller amount of HGMB, has a better surface appearance than weld 1, performed with lower Vc and a higher proportion of HGMB. On the other hand, weld 4 (explosive sensitized with EPS (2 % w/w) and a Vc higher than that of weld 6) shows much greater surface roughness, as can be seen from the comparison of Figs 8c and 8d. Weld 5 (explosive sensitized with 1.5% of EPS) displayed a surface appearance (not shown) similar to weld 4, with significant roughness, whilst weld 2 (explosive sensitized with 11% of HGMB) has an intermediate surface appearance (not shown) between weld 1 and weld 3.
These results show that the morphology of the outer surface of the welds is more a result of the type of sensitizing agent (HGMB or EPS) added to the emulsion matrix than of the explosive’s density or collision point velocity. It seems that the superficial appearance of the stainless steel flyer is affected by some instability in the propagation of the detonation wave, induced by the sensitizer added to the explosive matrix, in this case HGMB or EPS. This disturbance appears to increase with the increase in diameter of the spheres of the sensitizer additive. The diameter of the polystyrene spheres (1 mm) is much higher than that of HGMB (70 μm), as previously mentioned. The surface appearance of each specimen described above is similar around the perimeter and along the length. Preliminary tests proved that the concentricity of the tube and rod is essential to obtain this result.
Fig. 8 – Surface appearance of the welded samples: a) Weld 1; b) Weld 3; c) Weld4; d) Weld 6.
3.4. Morphology of the weld interfaces All the welds displayed wavy interfaces, with small melted regions coloured dark, as shown in Fig.
9. These melted regions may be incipient or even absent in weld 6, which used ANFO.
Fig. 9 – Morphology of the interface between carbon steel and stainless steel: a) weld 1; b) weld 2;
c) weld 3; d) weld 4; e) weld 5 and f) weld 6. Definition of the amplitude (A) and wave length (λ).