ABSTRACT
Residual stress measurements in a thick cylindrical steel electron beam (EB) welded sample are described. This is a particularly challenging problem because the residual stresses are distributed in a very narrow region, in and adjacent to the EB weld, when compared to more conventional welding methods. Several variations of the deep hole drilling (DHD) method were applied, including the original conventional method, an incremental approach and a newly developed over-coring method. It is demonstrated that the over-coring method can be used just as effectively as the more difficult incremental method.
KEY WORDS
Residual stress, Deep hole drilling, Plasticity, Over-coring
|
NOTATION |
|
|
DHD |
deep hole drilling |
|
OD |
outer diameter |
 |
diameter of the reference hole after gun-drill |
 |
diameter of the reference hole after 5 mm core extraction |
 |
diameter of the reference hole after 40mm core extraction |
 |
diameter of the reference hole after each trepan step |
 |
direct stress component in x direction |
 |
direct stress component in y direction |
 |
shear stress component |
 |
normalised distortion at angle 0 |
|
E |
Young’s Modulus |
|
EDM |
electric discharge machining |
INTRODUCTION
Electron beam welding (EBW), created in 1950′s, is a fusion welding method where a beam of high energy electrons are applied to heat the metal weld joint. No additional weld filler metal is introduced and the resulting fusion area is narrow relative to other welding processes. EB welding is preferred as a manufacturing process for high-value welds in the nuclear energy and aerospace industries [1] because there is very little distortion of the component. In order to create a narrow weld joint the process is often carried out in a vacuum to prevent dispersion of the electron beam. The requirement of the vacuum presents a practical issue for welding large components due to the operational cost and sizes of vacuum chamber available. Recent developments by the Welding Institute (TWI), in creating a moveable seal and vacuum system, have opened up the use of EBW on large, practical engineering components.
The determination of the residual stresses introduced by this novel welding technique remains important in determining the structural integrity of the component. However, the narrow weld region is highly constrained and consequently, it has been shown [2] that there are highly localised residual stresses in excess of the yield strength. Also the magnitude and distribution of residual stresses in thick cylindrical components only allows a few measurement methods to be applied and in this paper attention is directed to the development and application of the deep hole drilling method. First, the basic principles of the original method are described. This is followed an explanation of its application to an EB welded stainless steel cylinder. Finally, the results are reported and discussed.
MEASUREMENT TECHNIQUE
The deep hole drilling (DHD) is a semi-destructive, mechanical strain relaxation method that relies on the measurement of the distortion of a reference hole drilled through the component. The distortions between the stressed and unstressed states of the specimen are used to calculate the stress distribution, and initially the strain relaxation is assumed to be elastic. Leggatt et al [3] initially developed an analysis method for converting the measured distortions to residual stresses. Later, Granada-Garcia et al [4] and Kingston et al [5] advanced the method to include more precise measurement procedures and refined analyses.
Recent work showed that when residual stresses are close to the yield stress the assumption of entirely elastic relaxation of strains breaks down and plastic deformation occurs. To account for this an incremental deep hole drilling [6] method was developed. However, the technique currently provides only limited data and is time consuming. Alternatively, it has been shown [7] that by initially relaxing the residual stresses away from the yield surface it is possible to continue to use the conventional DHD method. In this paper, the initial step of reducing the residual stresses is undertaken by using a new over-coring method.
Irrespective of whether conventional, incremental or over-coring variations of the DHD technique are applied the conversion of the measured distortion to residual stresses remains the same.
The conventional DHD procedure involves 4 steps as illustrated in Figure 1. In the experiments, described later in the paper, the procedure was applied to a fully circumferential EB weld in a cylinder. For brevity the steps in the method were as follows.
1) Prior to gun-drilling two bushes were attached to the outer and inner surfaces of cylinder. A reference hole was gun-drilled through the component and reference bushes. A gun-drill is a special type of drill with a self-aligning tip. It is suitable for drilling long holes where a high degree of straightness is required. In these experiments a 1.5 mm diameter drill was used.
2) The internal diameter, 0o, of the reference hole was measured accurately at locations inside the reference hole through the entire thickness of the component and reference bushes. Diametral measurements were taken at 0.2mm increments along the reference hole and at angles of 22.5° increments using an air probe.
3) A cylinder, with a diameter of 5mm and with the reference hole as its axis, was trepanned free of the remainder of the component using electric discharge machining (EDM).
4) Finally, the internal diameter, 0, of the reference hole was re-measured at the same locations as in step 2.
Figure 1 An illustration of the procedures in standard DHD technique: step 1: drilling a reference hole; step 2: measurement of reference hole diameter; step 3: core trepanning and step 4: re-measurement of hole diameter
The diameter, 0O, of the reference hole measured in Stage 2 is the diameter when stresses are present. During Stage 3 the stresses are relieved, hence the diameter, 0, of the reference hole measured in Stage 4 is the diameter when stresses are not present. The differences between the measured diameters in Stages 2 and 4 enable the original residual stresses to be calculated.
In general, only the in-plane distortions (the plane normal to the reference hole axis) of the reference hole are measured throughout the DHD technique and only in-plane residual stresses are calculated. The normalised distortions of the reference hole are related to the residual stress components in the plane normal to the reference hole axis, axx, uyy and ct^,. To calculate residual stresses that vary with depth, it is assumed that the trepanned core is composed of a stack of annular slices, which act independently of one another and behave in a manner predicted by the constant remote stress analysis [4]. The measured distortions are related to the stresses using
where
where
is the normalised distortion at angle
E is Young’s Modulus,
and
are given by Bonner [8] and Garcia-Granada et al [4]. By determining the inverse of equation 1 the residual stresses
can be calculated.
If the component contains high magnitude, tri-axial residual stress, it is has been shown [6] that plastic relaxation occurs during trepanning. Like all mechanical strain relief techniques, the DHD technique is based on the assumption of elastic relaxation. Hence the distortion of the reference hole, on completion of trepanning through the thickness, can not represent the original residual stress field when plastic deformation occurs. Nevertheless, as shown by Mahmoudi et al [6], if the core is extracted in incremental steps and the diameter of the reference hole is measured between each increment then the residual stresses can be determined using the measured distortions and inverting equation 1. Since the incremental technique divides the trepan step into many stages, this technique can provide only limited depth resolution and is time consuming.
To avoid plasticity during trepanning, recent work by Hossain et al [7] demonstrated that removal of a section of material containing the region of interest allowed the initial high stresses to relax elastically away from the yield surface. In this paper, rather than only remove one cylinder (i.e. core) of material containing the reference hole along its axis, trepanning is carried out in two steps. First, a 40mm diameter core concentric to the reference hole is extracted followed by re-measurement of the reference hole. The difference between the diameters before (0o) and after (0) extraction permits us to determine the partially relaxed residual stresses. Finally, a 5mm core is removed from the 40mm core to completely relax the stresses in the 40mm core followed by re-measurement of the reference hole (0). The difference between the diameters before core extraction (0o) and after 5mm core extraction (0) allows us to calculate the initial residual stresses.
SPECIMEN AND MEASUREMENTS
The EB welded pipe investigated in this work was supplied through Rolls-Royce (RR), UK and manufactured by TWI Ltd. The specimen comprised of two cylinders, with an outer diameter of 356mm, inner diameter of 286mm and 235mm high. The cylinders were butt welded together through the thickness by using a reduced pressure electron beam welding method. A schematic of the cross-section of the sample is shown in Figure 2, together with photographs of the outside of the pipe at the locations of the measurements. Both cylinders were manufactured from type 304 stainless steel and, due to the nature of EB weld, no additional material was deposited. The axial length of the weld cap was approximately 5 mm.
The sample was measured in the as-welded condition. An angular position of 0° was marked on the sample and coincided with the EB weld start and stop positions. Two measurements were conducted through the weld centre line from the outer surface and made with respect to the 0° location. With reference to Figure 2, an incremental DHD measurement was performed at 180° while the new over-coring method was applied at 90°.
The incremental DHD technique was carried out with the extraction of 5 mm diameter core containing a 1.5mm diameter reference hole. In total 10 trepanning increments were conducted and the diameter of the reference hole was re-measured after each increment, ……^10). Then the diameters were compared against each other to calculate the original residual stresses by inverting equation 1.
In the over-coring method a 40mm diameter core, concentric to the reference hole, was first extracted. This was followed by re-measurement of the reference hole diameter. Then a 5 mm diameter core containing the reference hole was extracted and the hole was re-measured again. The residual stresses at each stage were then calculated by inverting equation 1.
RESULTS AND DISCUSSION
The residual stresses obtained at 180° are shown in Figure 3. Two sets of results are shown, one from the application of conventional DHD method and a second from the incremental technique. In both cases only the hoop and axial stresses are shown. The axial-hoop shear stresses were relatively low (less than 50MPa) compared to the principal stresses and have been omitted for clarity. Due to the depth resolution, the incremental DHD technique did not measure the first few millimetres through the weld from either side of the weld surface.
The conventional method initially measured tensile near surface residual stresses, but at greater depths the measured residual stresses remained relatively low. In contrast, the incremental technique revealed increasingly more tensile stresses in the hoop direction with increasing depth. These results are similar to those shown by Mahmoudi et al [6] for measurement of residual stresses in a quenched cylinder. Their results are reproduced in Figure 4 and illustrated that the significant plastic relaxation of the residual stresses took place during trepanning. In contrast, the incremental method was shown to provide results in the quenched cylinder that agreed well with neutron diffraction results.
There are no comparable finite element results available for the EB welded stainless steel cylinder but the results obtained from the incremental DHD measurements, shown in Figure 3, also revealed the presence of high tensile residual stresses. When the results obtained via the over-coring method are compared with incremental results (as shown in Figure 5) it is evident that similar trends were produced. The hoop and axial stresses followed the same trend, with the hoop stresses always larger than the axial stresses in the tensile stress region while the axial stresses were larger than the hoop stresses in the compressive region.
Although not shown here there was also an intermediate step in the over-coring method that was able to provide additional data. The results from these are being examined in further work but clearly show that relaxation of residual stresses in multiple steps provides supplementary information about the state of the residual stresses on the periphery of the over-core.
Figure 2 Electron beam weld pipe sample: (a) Schematic drawing of the EB weld pipe showing the top view with measurement locations; (b) Photograph of sample after incremental DHD measurement at 180°; (c) Photograph of sample after over-coring DHD at 90°
Figure 3 Residual stresses obtained from the conventional DHD and incremental DHD techniques at the 180° location of the EB welded stainless steel pipe
Figure 4. Measured residual stresses using conventional and modified DHD method and comparison with finite element predictions for quenching in a stainless steel cylinder
Figure 5 Comparison of the residual stresses obtained from the incremental and over-coring DHD methods
CONLUDING REMARKS
Several DHD techniques were carried out on a 304 stainless steel electron beam (EB) weld pipe. Two of the techniques, incremental and over-coring showed good repeatability, with a general trend of compressive residual stress near the outer diameter surface and increasing to high tensile residual stress within at mid-thickness of the cylinder wall. The conventional DHD method was not able to measure the residual stress because of plastic relaxation as found in earlier work [6]. The maximum tensile stress of 296 MPa was obtained by incremental DHD technique at a depth of 24mm from the outside diameter, while the maximum compressive stress of 177MPa was found via the over-coring method at the depth of 2.5mm. Finally, the newly developed over-coring method appears to improve significantly the depth resolution of residual stress measurement and retains the simplicity of the DHD technique compared to the incremental method.
