Introduction
Materials with designed graded properties are very attractive for the industry since they are capable of offering a substitute to homogenous materials—thus providing improved performance of machines and structures (Ref
1-
3). Specific engineering structures in nuclear and geothermal plants have to cope with high temperature and operate fault-free even under very corrosive environmental conditions (Ref
2,
4). In order to increase the reliability and decrease service cost, it is considered that some elements made of austenitic steels can be replaced by multilayer metallic materials with required properties (Ref
1). For example, titanium alloys offer excellent corrosive resistance, but have lower elasticity modulus compared to steel. The multilayer plate consisting of titanium alloy and steel layers could provide an adequate corrosion resistance, stiffness and endurance. Such plates could be obtained for example by explosive welding. The explosive welding technology was developed for industrial applications in the 1960s and offers multilayer materials with required physical properties (Ref
5,
6). Higher fabrication costs are acceptable if the service life of industrial equipment containing multilayer material is increased. The explosive welding is unique technology since it applies energy of detonation to cause an impact of one material to another one during which a bond is created. The process of joining is difficult to investigate because of the high velocity of detonation wave (around 2000 m/s). This high velocity results in a very high rate of all processes taking place during explosive welding such as plastic strain, phase transformation, heating and cooling.
Mechanisms leading to bond formation are still not fully understood. For the time being, explosion welding is defined as a solid-state welding process (Ref
6). However, recent studies have revealed that a thin diffusion layer (Ref
7-
11) or even larger melted areas (Ref
11,
12) are formed between the materials that are bonded. Moreover, the melted areas could include intermetallic phases whose volumes are increased during post-welded heat treatment leading to a decrease in fatigue lifetime (Ref
12). The welded area also includes highly deformed grains and even micro-cracks as results of very rapid cooling rate (Ref
13,
14). Such welding inhomogeneities occupy a relatively small volume of materials comparing to the size of heat-affected zone created during traditional welding. As a result, the fatigue properties of multilayer materials could not differ significantly from mechanical properties of the parent materials (Ref
15). The very important issue from many points of view is associated with an existence of residual stresses in explosively welded elements. The residual stresses originate as a result of high deformation of joined materials locked in by a bond creation. If bonded materials have different thermal expansion properties, the performed heat treatment can only change the state of residual stresses, but cannot release them (Ref
16). It is well known that the residual stress has an influence on important properties of the materials. For example, tensile residual stress can induce corrosion cracking (Ref
17,
18), while the compressive stress can increase endurance and fatigue properties (Ref
19,
20). Additionally, the knowledge of residual stress distribution in explosively welded materials provides important insights regarding the process of explosive welding. In a recent paper (Ref
16,
21), it was demonstrated that tensile residual stress exists in the layer of titanium Grade 1 explosively welded to steel and the performed heat treatment changes the tensile state to the compressive one. Due to its high cold formability (elongation around 50%), titanium Grade 1 is often used as a flyer layer (the layer accelerated by explosion). Features of the joining process such as lower stiffness (due to elasticity modulus and thickness) of flyer layer than stiffness of base plate made of steel and high cold formability of flyer layer enable one to create a simple model explaining the observed tensile residual stress in the titanium layer. During an explosion, the titanium layer is considerably compressed in one direction but elongated along the plane of impact—assuming no friction due to a thin melted zone. When the high pressure is released, the bond is already created and the previously induced elastic elongation is locked. As a result, the tensile residual stress is created. But in case of more than two welded layers with different physical properties, the above model could be invalid.
Sedighi and Honarpisheh (Ref
22) determined residual stress distribution in explosively welded Al-Cu-Al layers by using the incremental hole drilling method. It was found that the maximum value of the residual stress exists on the surface of the three-layer plate and the minimum compressive stress was identified in the Cu interlayer. The authors proposed a similar model as the one that was described above to explain such a distribution. However, the size of investigated samples and layers thickness is much smaller than studied in (Ref
16,
23).
Yasheng et al. (Ref
24) successively applied a milling technique to measure residual stresses in an explosively welded plate made of titanium, tantalum and steel layers. It was found that the highest tensile and compressive stresses occur at the interfaces and they are accompanied with the highest stress gradient. The tensile type of residual stress was found on the surface of the welded plate. The authors did not provide any explanation of the measured residual stresses.
Saksl et al. (Ref
25) applied micro-x-ray diffraction using synchrotron radiation to determine residual stresses in an explosively welded plate made of austenitic and pressure vessel steels. The applied technique gave the results of measurements of the principal residual strains at microscale in the vicinity of interface (± 200 μm from interface). It was found that the tensile type of maximum principal residual stress exists in the austenite phase (the flyer layer—austenitic steel), while minimum principal residual stress is compressive. In the base plate (pressure vessel steel), both principal stresses are compressive in the analyzed ferrite phase of steel.
Taran et al. (Ref
26) applied the neuron diffraction technique for residual stress measurement in an explosively welded cylindrical adaptor made of AISI 316L and titanium Grade 2. The compressive residual stresses were found in outer surface (steel), while tensile stresses occur in the inner surface (titanium).
In the present research, a three-layer plate composed of titanium Grade 12 and pressure vessel steel ASME SA516 Grade 70 with interlayer made of titanium Grade 1 is investigated. Titanium Grade 12 offers a better corrosion resistance and strength than titanium Grade 1 or titanium Grade 2 (Ref
27,
28), but its cold formability is low (i.e., elongation is no more than 20%) and, as a result, it is necessary to apply an interlayer with a higher cold formability (Ti Grade 1) for the successful welding process.
The aim of the research reported in this paper is the determination of the residual stress distribution in the three-layer plate (Ti Grade 12/Ti Grade 1/Steel) obtained by the explosive welding process. Two methods of residual stress determination were applied: the sectioning method (Ref
29,
30) and the standard hole-drilling strain gage method (Ref
31).
Analysis of Results
The distribution of the relieved axial strain obtained by the sectioning method reveals a similar tendency in all
A,
B,
C samples (Fig.
5). The highest positive strain values are recorded in the middle of the titanium Grade 12 layer. From this location, the strains decrease along thickness-crossing zero value in the titanium Grade 1 layer and reaching a minimum negative value at around 16-17 mm from the plate surface. Samples
A and
B with perpendicular symmetry axes and not subjected to the heat treatment demonstrate an almost identical relieved strain distribution,
\(\varepsilon_{x}^{r} \approx \varepsilon_{y}^{r}\). The relieved strain distribution in the
C sample subjected to the heat treatment is characterized by more extreme values. The strain range in the
C sample is equal to 0.0031, while in the
A and
B samples is equal to 0.002.
The relieved strain distributions approximated to fourth-order polynomial functions in the A and B samples offer the calculation of the distribution of residual stress components \(\sigma_{x}^{r} \left( z \right)\) and \(\sigma_{y}^{r} \left( z \right)\). We can observe that residual stress relaxed by deflection f of the sheet is several times lower than residual stress relaxed by extension ΔL. The negative values of \(\sigma_{x}^{r} \left( z \right)\) and \(\sigma_{y}^{r} \left( z \right)\) occur on the plate surface and achieve minimum in the middle layer of titanium Grade 12. Approaching the interface with titanium Grade 1, the values of \(\sigma_{x}^{r} \left( z \right)\) and \(\sigma_{y}^{r} \left( z \right)\) grow, but still remain negative in the middle of titanium Grade 1 layer. Zero values are obtained very close to the second interface at the depth around 11-13 mm from the plate surface. In general, the positive values of \(\sigma_{x}^{r} \left( z \right)\) and \(\sigma_{y}^{r} \left( z \right)\) occur in the steel layer.
The principal residual stresses determined by the hole-drilling strain gage method in all four points give negative values in the range [−392, −252] MPa. These values represent averaged residual stresses along the depth of the hole (2 mm) drilled in titanium Grade 12. The resulting values correspond to negative values obtained by the sectioning method. However, they are lower by around 100-200 MPa from the ones that are provided using the sectioning method. This difference could be explained by a fact that residual stresses relaxed by extension Δ
L (the sectioning method) represent averaged values over the length
L. Values
\(\sigma_{x}^{r} \left( z \right)\) and
\(\sigma_{y}^{r} \left( z \right)\) are equal to zero at the ends of slices; thus, in the middle of slices the residual stress could approach the values obtained by the hole drilling method. We also have to note that the relatively low rotational speed applied for titanium alloys, i.e., 6000 rpm, could lead to the overestimation of residual stresses by around 19% (Ref
40,
41).
The existence of compressive residual stresses in titanium Grade 12 layer cannot be explained by the simple model proposed in Ref
16,
22. In the present case, the layer of titanium Grade 12 is explosively welded to steel through interlayer made of titanium Grade 1. It is a very unusual combination of welded materials in which one titanium alloy hard deformable (elongation
A = 20%—Ti Grade 12) must compose solid interface with another titanium alloy. The residual stresses could be a result of the phase transformation. For example, the martensite phase transformation from β to α’ (hexagonal martensite) leads to increase in volume and, as a result, compressive residual stresses could be induced. However, titanium Grade 12 (Ti-0.3Mo-0.8Ni) is classified to α or near-α alloy with a little amount of β phase (Ref
42). The more the volume changes from a body-centered cubic crystal structure (β) to hexagonal type (α’) is not significant (Ref
43,
44) with the small production of residual stresses.
The previous model (Ref
16,
22) does not consider the effects taking place at high strain rates. According to Weertman (Ref
45), the total strain in a direction parallel to a shock wave front must be equal to zero. It results from the summation of tensile plastic strain and elastic hydrostatic compression strain. In the present case, the source of the shock wave is an impact of two titanium layers and the front of the wave is parallel to the created interface. Thus, the propagating shock wave creates hydrostatic compressive stresses in both titanium layers which are locked by the formed solid bond. The interior force equilibrium requires the generation of tensile residual stress in steel layer. In accordance with the presented model, the tensile residual stress could not be locked during shock wave, but this phenomenon is observed in some explosively welded materials (Ref
16,
22). We must note that Weertman (Ref
45) did not consider many effects associated with the source of shock wave such as local and rapid temperature rising/cooling, and different thermal and mechanical properties of welded materials. It is concluded that the process of residual stress formation during explosive welding depends on many factors such as mechanical and thermal properties of joined materials, dimensions of bonded layers and welding parameters. Thus, we can also forecast that the values of residual stress components are not equally distributed along the welded plate due to some variation in detonation velocity.
The observed adiabatic shear bands (ASB in Fig.
10) result from the very high compressive stresses generated during the impact of titanium Grade 12 layer with titanium Grade 1. During the impact of two titanium layers, compressive stress is induced and, consequently, the material from two layers is ejected in the parallel direction to the interface (from left to right in Fig.
9 and
10). The ejection of titanium Grade 1 creates some irregular bulges, and the ejection of titanium Grade 12 initiates adiabatic shear bands in the shape of a curve. When the adiabatic shear band is initiated, the compressive stress is partially relaxed, but due to the propagation of detonation wave the processes of material ejection and stress increasing start again until formation of another adiabatic shear band.