«Please cite this article as: Mendes, R., Ribeiro, J.B., Loureiro, A., Effect of explosive characteristics on the explosive welding of stainless steel ...»
Effect of explosive characteristics on the explosive welding of stainless steel to
carbon steel in cylindrical configuration
R. Mendes, J.B. Ribeiro, A. Loureiro
Reference: JMAD 5298
To appear in: Materials and Design
Received Date: 19 November 2012
Accepted Date: 22 March 2013
Please cite this article as: Mendes, R., Ribeiro, J.B., Loureiro, A., Effect of explosive characteristics on the explosive welding of stainless steel to carbon steel in cylindrical configuration, Materials and Design (2013), doi: http:// dx.doi.org/10.1016/j.matdes.2013.03.069 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Effect of explosive characteristics on the explosive welding of stainless steel to carbon steel in cylindrical configuration R. Mendes1, J.B. Ribeiro1, A. Loureiro2 ADAI - Assoc. for Dev. of Ind. Aerodynamics / LEDAP - Lab. Energetics and Detonics Department of Mechanical Engineering, University of Coimbra, Rua Luis Reis Santos, Polo II, 3030Coimbra, Portugal CEMUC – Department of Mechanical Engineering, University of Coimbra, Rua Luis Reis Santos, 3030-788 Coimbra, Portugal *Email address: email@example.com, tel. + (351) 239 790 700, fax. + (351) 239 790 771 Abstract The aim of this research is to study the influence of explosive characteristics on the weld interfaces of stainless steel AISI 304L to low alloy steel 51CrV4 in a cylindrical configuration. The effect of ammonium nitrate-based emulsion, sensitized with different quantities and types of sensitizing agents (hollow glass microballoons or expanded polystyrene spheres) and Ammonium Nitrate Fuel Oil (ANFO) explosives on the interface characteristics is analyzed. Research showed that the type of explosive and the type and proportion of explosive sensitizers affect the main welding parameters, particularly collision point velocity. The morphology of the wavy weld interfaces, chiefly the amplitude and length of the waves, is affected both by the impact velocity and the type and particle size of the explosive sensitizers, and increases with particle size. All the weld interfaces, except welds done with ANFO, displayed localized melted and solidified regions, whose chemical composition resulted from the contribution of both flyer and base metal.
Keywords: explosive welding; emulsions explosive; low alloy and stainless steels.
1. Introduction Explosive cladding or welding was first observed during World War I when soldiers noticed shrapnel casing bonding with steel or metallic surfaces. However, it was only in the 1940s that the first experiments were carried out by L. R. Carl in 1944. Later, in the 1950s, more thorough investigations were conducted by Rinehart and Pearson [1; 2] and in the 1960s by Philipchuck , Davenport and Duvall , Holtzman and Ruderhausen , Deribas , Zernov et al. , Bahrani and Crossland . This is when the commercialization of explosive welded plates began at DuPont .
Explosive cladding/welding is usually considered a solid state process [9; 10] in which the detonation of a certain amount of an explosive composition accelerates one of the materials to be welded against the other in order to promote a high-velocity oblique collision (see Fig. 1) that causes severe, but localized, plastic flow at the interacting surfaces. The high speed of the collision, between 0.6 and 3 mm/μs, brings about pressures considerably greater than the strengths of any known material. This result in significant plastic deformation and the mixing of the surface layers of the materials to be bonded which can result in a linear or a wavy interface.
Such extreme, localized, plastic deformations dissipate a considerable amount of kinetic energy of the colliding plates, and reduce considerably the amplitude of the reflecting tensile stress which tends to separates the plates. While the process is normally described as a cold technique, whereby no external heat is used to promote the bonding, localized high-temperatures are normally generated at the weld interface due the dynamics of the process. Temperatures of 700 ºC, at 1 mm from the surfaces, have been measured but it seems obvious that much higher temperatures (enough to melt, for example, molybdenum and niobium), are attained, at least for a few microseconds, at the interface . The nature of the oblique collision is such that parts of this molten layer are ejected from the system as a monolithic jet or as a dispersed cloud of particles that breaks up the oxide and/or other contaminant layers that may exist at the surface of the material. The estimated cooling rate of the remaining molten material is approximately 105 K/s, due to contact with the relatively large bulk of the materials to be welded. Such high cooling rates lead to the formation of ultrafine, nanometer, grains with a random orientation.
Fig. 1-Schematic representation of the explosive welding process.
One of the greatest attributes of this technique is that it allows very large surface areas of dissimilar metals to be welded, which is not possible using conventional fusion welding. The welded metals remain in their wrought states, and no continuous cast structures are created;
neither the microstructures, nor the mechanical and corrosive properties of the wrought parent components are altered; there are neither heat-affected zones nor continuous-melt bands exhibiting mixed chemistry; there is almost no diffusion of alloying elements between the components. Therefore, explosive welding is an effective joining method for virtually all combinations of metals. Its only limitation is the sufficient ductility (10%) and fracture toughness (30 J) of materials to undergo rapid deformation without fracture . Explosive welding is mainly used for the production of laminated metals in the form of sheets, rods or pipes in order to improve their corrosion or wear resistance, thermal conductivity or even antifriction properties.
A notable characteristic of explosive welding is the formation of waves at the interface between the welded materials. In comparison with the other occurring interface geometries, flat/smooth or continuous melted, the wavy interface is the one producing the best welding characteristics [13For this reason, the mechanism of wave formation in explosive welding has been studied for years [18; 19]. Despite the efforts there is not a unique explanation for that phenomenon, and four different explanations can be found in the literature: a) the Jet Indentation Mechanisms, used by Abrahamson  and later by Bahrani et al. , in which the indentation action created by the jet (salient jet according to Abrahamson and re-entrant jet according to Bahrani et al.) is responsible for the creation of a hump ahead of the collision point. According to Bahrani et al.
, the interaction between the hump and the re-entrant jet traps the ejected material and is responsible for the formation of vortices; b) the Flow Instability Mechanism: in this case the process of wave formation is considered a hydrodynamic phenomenon similar to what happens at the interface between two liquids with different horizontal velocities, which is known as KelvinHelmholz Instability. According to Hunt , the velocity discontinuity occurs between the parent plate and the re-entrant jet, and according to Robinson  between the parent plate and the salient jet. The instabilities that trigger the wave formation are either created as a result of wave interferences, as suggested by Ben-Artzy , or as a result of oscillations in the detonation wave process which are transmitted to the flyer plate and, through it, to the plate interface, as suggested by Plaksin et al. ; c) the Vortex Shedding Mechanism, according to several authors [19; 25] the waves are formed due to a vortex shedding mechanism similar to that responsible for the formation of the von Kármán Vortex Street which can be observed for the flow of a viscous fluid after an obstacle; d) the Stress Wave Mechanism, proposed by El-Sobky and Blazynski ; In this case, the interface waves are the result of successive interference from rarefaction waves in both plates. Knowing which of these mechanisms better describes the process of wave formation has been a long-standing problem in the field of explosive welding. It may be, as Carton tried to demonstrate , that some of these mechanisms may occur at the same time and dominate over the others at different welding conditions.
Attempts to relate the explosive welding parameters to the characteristics of the interfacial waves began in Russia by Deribas and his team  at the end of the 1960s and beginning of the 1970s.
Reid and Sherif  in the UK and Cowan et al.  in the US, predicted an increase in the wavelength of the interfacial waves as a function of the collision angle. Later, Jaramillo et al. , following the works of Cowan and Holtzman  and Salem and Al-Hassani , tried to link the wave characteristics to the ratio of the thicknesses of the base and flyer plates. Jaramillov found that the wavelength and wave amplitude increase with that value. He also found what he considered to be a considerable discrepancy between the experimental results of the wavelength and the calculations made using equations developed by Godunov et al.  assuming an hydrodynamic behaviour of the materials at the weld interface. Recently, Manikandan et al.  found a direct relation between the kinetic energy lost at the collision of the plates and the amplitude and length of the waves at the welding interface. The subject continues to merit the attention of several authors [18; 34-36]. Thanks to their work, and in accordance with older results, it has been found that the wavelength and the wave amplitude increase with the stand-off distance, the explosive load and the base plate thickness. Nevertheless, according to a model developed by Balasubramanian et al.  the amplitude and wavelength increases with the flyer plate thickness, in contrast to observations made by Cowan et al., and Jaramillov et al. .
These discrepancies, as stated by Jaramillov et al. , suggest an insufficient understanding of the process of metal adhesion at high collision velocities.
The conditions that should be met in order to achieve good welds is called the weldability window or criteria. A weldability criterion based on the flyer plate velocity and flyer mechanical properties developed by Cowan et al.  is mentioned by Mousavi and Al-Hassani , and is considered to give poor results. Criteria based only on the collision point velocity, although allowing the development of empirical equations to establish the weldability limits, did not provide an overall picture of the process. At present, the most used and well-known weldability criterion is based on the collision point velocity Vc, and on the collision angle β, as defined in Fig. 1. In the β- Vc space, the weldability window is defined by four lines or limits (vd. Fig. 2), the first theoretical explanation for which was offered by Wittman . There are four conditions for the establishment of those limits. The first limit is linked to the formation of a jet at the collision point;
the rightmost line of the weldability window is a consequence of this condition. To meet this condition, as stated by Walsh at al. , the collision point velocity should be smaller than the sound speed of the materials. This limiting value for collision point velocity is a weak function of the collision angle β, so instead of a straight vertical line the rightmost limit of the weldability window is a slightly concave left vertical line. The second limit is the formation of a wavy interface;
the leftmost line of the weldability window is a consequence of this condition. Kuzmin and Lysac (vd.  cited in ) state that this line, which defines the transition collision velocity Vc,tr (above which we end up with a wavy interface), is a function of the collision angle; so, it should not be a straight vertical line. However, most authors consider it a vertical line, and therefore independent of the collision angle. Cowan et al.  have even proposed the equation Eq. (1) for the determination of Vc,tr (expressed in mm/µs) that appears as a function of the materials densities (ρp and ρf, for the density of the parent and flyer plate materials, expressed in kg/m3), Vickers hardness HV,p and HV,f (expressed in N/mm2) and a critical Reynolds Number (Rcr), that takes values between 8.0 and 13.0, for the asymmetric explosive welding configuration.
2Rcr (HV, p + HV, f ) Vc,tr = (1) (ρ p + ρ f ) The third limit is the achievement of an impact velocity Vp, where by the impact pressure at the collision point exceeds the yield stress of the materials, in order to promote plastic deformation.
The lower limit of the weldability window is a consequence of this condition. Deribas and Zakharenko, as mentioned by Zakharenko el at., developed an equation for this limit (vd. Eq.
2), in which the minimum collision point velocity Vc,min, related to the impact velocity through the Eq. (5) is determined as a function of the material Vickers Hardness Hv [N/m2], the material density ρ [kg/m3], the collision angle β, and a constant k1 of values between 0.6 (for clean surfaces), and 1.2 for imperfectly cleaned surfaces .
With all parameters referring to the flyer plate properties and where Tm is the melting temperature [⁰C], k the thermal conductivity [erg/cm ⁰C], Cp the thermal capacity [erg/g ⁰C], ρ the material density in [g/cm3], h the thickness [cm], C0 the bulk sound speed [cm/s], and N a constant that for several metals takes the value of 0.11 , with β expressed in radians.
Fig. 2 – Generic weldability window with the definition of limits .
As the results from the previous description show, successful welding depends on Vc, Vp and β.However, once Vc and Vp are related through β we are left with only two independent variables.