Multi-Axial Failure Behaviour in Advanced Steels at Elevated Temperatures

Transcription

1 Multi-Axial Failure Behaviour in Advanced Steels at Elevated Temperatures K. Wasmer, F. Biglari 2 and K. M. Nikbin Imperial College, London, SW7 2AZ, UK. 2 Amir Kabir University, Tehran, Iran ABSTACT: Three different high chrome steels, identified as E9, P9 and P92 were tested at 625, 600 and 625 C respectively, using uniaxial and notched bar specimens of parent (PM), heat-affected zone (HAZ), weld (WM) and cross-weld (XW) specimens. The tests were part of a joint Brite/Euram European collaborative project called LICON in which each partner carried out only a limited number of relatively short term ( hours) creep rupture experiments. The variability in data was quantified using statistical methods deriving appropriate standard deviations. The results indicate a higher notch strengthening factor in parent compared to weld and cross-weld samples in all the steels. The comparison of creep rupture life of notched bar tests with plain bar creep tests are presented in terms of an equivalent stress criteria. The stress rupture behaviour in the notched bars is determined by a combination of the von Mises stress and the maximum principal stress. It has also been found that stress to rupture sensitivity is slightly higher in plain bars compared to the multi-axial notch tests. However the standard deviations for the equivalent stresses are found to be relatively insensitive to these differences. INTODUCTION Generally, most structural components experience a three-dimensional state of stress resulting from the type of loading or from other forms of constraint. For example, in the case of components operating at elevated temperature, these could be subjected to multiaxial stress states caused by thermal gradients as well as by holes and sharp changes in section. Multiaxial stress states are more complex and various than uniaxial stress states. Components under a high state of multiaxial tension can fail prematurely due to the inhibition of deformation and enhancement in fracture processes. However the actual void growth failure mechanism could be the same in all cases. This work, undertaken under a Brite/Euram programme [see Acknowledgements], covers the high temperature failure properties of three high chrome advanced steels, identified as E9, P9 and P92 and whose properties are shown in Table. These were tested at 625, 600 and 625 C respectively. Tests were carried out by different partners using uniaxial and notched bar specimens on parent, HAZ, weld and cross-weld specimens.

2 This work was necessary to establish the materials uniaxial and multiaxial behaviour under relatively short term test times. The results have been previously applied in to the LICON methodology [] to predict long term uniaxial test behaviour from short term multi-axial times. This paper describes a limited part of this European collaborative project and presents the investigation to determine the creep rupture properties of the materials. Given the relatively limited number of experiments that were performed for each material condition a statistical analysis was carried out to determine the means and the standard deviations. The comparison of creep rupture life of notched bar tests with plain bar creep tests are presented in terms of an equivalent stress criteria. The stress rupture behaviour in the notched bars is determined by a combination of the von Mises stress and the maximum principal stress. MATEIALS AND SPECIMENS The chemical compositions for each of the steels designated as E9, P9 and P92, are shown in Table. Large section welds [] were introduced in each material and subsequently uniaxial and notched bar specimens were extracted and machined from them as shown schematically in Figure. The geometry and the dimensions of the double notched bar, which correspond to the uniaxial specimen, are shown in Figure 2. The dimensions of the specimens correspond to those recommended in the code of practice for notched bar testing [2]. Two notch root radii were chosen to give the notch acuities a/ = 5 and 5 to correspond, respectively, with medium and sharp notch profiles. TABLE : Chemical analyses of E9, P9 and P92 (Weight in %) Mat. C Si Mn P Cr Cu Mo N Ni V W E P P WM/P Notched bars HAZ XW Weld Uniaxial Parent Figure : Specimen extraction for PM, HAZ, WM and XW materials

3 Figure 2: Dimensions (mm) for sharp and medium notched bar specimens. EXPEIMENTS Creep rupture tests were performed at 625 C on E9 and P92 and at 600 C on P9 on uniaxial and double notched bar specimens over a range of stresses. The applied net section stress σ net in the specimens were chosen to give lifetimes of 00-0,000 hours. With respect to the applied tensile load, a gross section stress was measured for the uniaxial specimens and a net section stress measured across the minimum notch root throat diameter to represent as an applied stress. In the event of failure in the notched bar specimens, fracture occurred across one of the two notches. Given four different material conditions (PM, HAZ, WM and XW) for the three steels, only a limited number of specimens (ranging from 2-4) were used for each condition except for some uniaxial tests on which more tests were carried out. A statistical analysis was performed to quantify the variability in the data. DATA ANALYSIS Modelling A convenient method of introducing a state of multiaxial stress into laboratory specimens is to subject circumferentially notched bars to an axial tensile load [2]. Changing the notch profile can simulate various states of stress. For a notched bar specimen with a gross radius b, notch throat radius a, and notch root radius, it is found that there is a point called a skeletal point [3] in the notch throat where stress is approximately constant independent of stress index n in the Norton creep law. The position of the

4 skeletal point from the notch axis and the stress state of this point are sensitive to the ratios b/a and notch acuity a/. Table 2 indicates the influence of notch acuity a/ on the normalised stresses, σ * VM/σ net, σ * /σ net and σ * m/σ net where σ net is the load divided by initial cross-sectional minimum area at the notch root, σ * VM, σ * and σ * m are von Mises stress, maximum principal stress and mean stress developed at the skeletal point, respectively. As seen from Table 2, the von Mises stress at the skeletal point for the medium notch is larger than that for the sharp notches and that a higher state of triaxiality is generated in the sharp notched bar than in the medium notched specimens for the same net section stress. The equivalent stress σ eq is defined as the stress applied in uniaxial plain bar specimen which has the same lifetime as the notched bar specimen and is given [4] by σ eq * ( α ) σ = α σ + () * VM where α is determined from the experimental data. Thus for α = failure is controlled only by the principal stress, whilst for α = 0 it is controlled by the von Mises stress. The design lifetime of an engineering component is usually based on the time to reach a specific strain or rupture. The relationship between rupture time and stress is given by t = Hσ ν (2) where H and ν (the slope) are material constants. Notch strengthening or notch weakening can be expressed alternatively using a Notch Strengthening Factor (NSF) which is defined as the ratio of the net section stress to the equivalent stress. A specimen which requires a lower uniaxial stress than the net section stress σ net of the notched bar specimen is said to be notch strengthening (σ net / σ eq >) and, conversely, a specimen which requires a higher uniaxial stress than the net section stress σ net of the notched bar specimen is said to be notch weakening (σ net / σ eq <). Notch Acuity atio (a/) TABLE 2: Normalized skeletal stresses [3] Notch Type σ * VM σ net σ * m σ * m σ * VM σ * σ * VM σ net σ net σ * 5 Sharp Medium

5 ESULTS Figures 3-5 compare the creep rupture life of the notched bar and uniaxial tests on E9 and P92 material at 625 C and P9 at 600 C. The data is presented in terms of the net section stress. It can be seen that for all the steels the uniaxial plain bar specimens give the value of ν 0 which are similar to the average uniaxial creep index n for these materials, as shown in table 3. However, there is a difference in the value of ν between uniaxial plain and notched bar specimens. E9 and P92 show a difference of about a factor of 2 whereas P9 is closer to the slopes for the uniaxial tests. It is also noted that for all materials, the WM notched bar specimens fail fastest, followed by HAZ, XW and PM. Table 3 gives the average failure strain for the three materials under HAZ, PM, WM and XW conditions. Generally for notched bars the WM and XW, with ductilities of 5-0%, show the fastest rupture times and the parent with ductilities ranging between 8-25% shows the slowest. However the available data for HAZ which exhibits over 8-32% failure ductility fails faster than the parent. The complex nature of the HAZ failure needs further investigation. Figure 6 presents an example for the estimated NSF versus times to rupture for E9. It is clear that the parent material medium notch (PM - Medium) gives the largest NSF followed by the weld and the cross-weld. The experimentally measured and averaged NSF for the sharp and medium notch and each material as well as material condition are given in Tables 4-5 respectively. The NSF ranges between.-.7 and similar trends, to that of E9 shown in Figure 6, are present for the P92 and P9 materials. As shown in Tables 4-5 the highest value of NSF is observed in the medium notched bars made of parent material, whereas for the sharp notched bars of weld material the lowest value of NSF is obtained. Tables 4-5 contain two different value of NSF depending on the assumption made on the slope. The first NSF values assume the uniaxial slope for the notched bars, whereas the second NSF 2 values are based on the slope of the notched bars specimens. The variation in the NSF, due to the different methods of estimation, is within a range of scatter which do not affect the overall conclusions reached. The parameter α in Eq. for the materials E9, P9 and P92 were evaluated from the notch rupture data. The net-section stresses σ net of the notched bars and analytical values of the skeletal point stresses σ * VM and σ * were used to calculate α for each test. In the same way as for the NSF the values of α and α 2 are presented. The calculated α values

6 were averaged and shown in Tables 4-5 as ranging between 0.25 and 0.40 for HAZ, weld and cross-weld whereas for parent α is very close to 0. As mentioned in the analysis, having a value of α between 0.25 and 0.40 implies that failure is mainly controlled by von Mises stress whereas for the parent with a value of α 0, failure is controlled only by von Mises stress. The α values for E9 weld metal vary significantly due again to the data scatter and limited number of tests data available. In order to determine the variability of the data a statistical analysis was performed. The measure of the variability used is the standard deviation s, which is based on the deviation of individual observation about the mean value. Due to the small number of data available for each material condition, it was assumed that all uniaxial tests from the same material have the same slope and also that all notched bars have the same slope, regardless of the material conditions. Figures 3-5 suggest that this assumption is acceptable. Therefore, the slope was determined according to a best estimate line through the parent data-set for which there were the most number of tests. TABLE 3: Average failure strain and n values for E9, P9 and P92 for uniaxial Mat. E9 P9 P92 Cond. HAZ PM WM XW HAZ PM WM XW HAZ PM WM XW ε f % n ave TABLE 4: Average NSF and α values for E9, P9 and P92 for medium notch Mat. E9 P9 P92 Cond. HAZ PM WM XW HAZ PM WM XW HAZ PM WM XW NSF NSF α α TABLE 5: Average NSF and α values for E9, P9 and P92 for sharp notch Mat. E9 P9 P92 Cond. HAZ PM WM XW HAZ PM WM XW HAZ PM WM XW NSF NSF α α

7 The standard deviations are given in Table 6 for the uniaxial and notched bars for the different steels using test data >200 hours. The values of the standard deviations between uniaxial and notched bars are found to be similar. In addition, the standard deviations for the combined uniaxial and the corresponding equivalent stresses, σ eq and σ eq 2, for notched bars are greater by a factor of approximately 2 than uniaxial and notched bars as shown in Table 6. Where σ eq and σ eq 2 have been derived from α and α 2 in Table 4. Figure 7 shows the example of the average slope for σ eq versus time to rupture plot for the parent E9 steel. Similar slopes of -/8 ± 0.3 were found for the other materials when using α or α 2 to estimate σ eq. TABLE 6: Standard deviations s for uniaxial, notched bars and the equivalent stress, σ eq Mat. E9 P9 P92 Cond. HAZ PM WM XW HAZ PM WM XW HAZ PM WM XW Uniax Notch σ eq σ eq Assuming the slope from uniaxial data and notched bars data ν σ net [MPa] HAZ - Uniaxial HAZ - Notch PM - Uniaxial PM - Notch WM - Uniaxial WM - Notch XW - Uniaxial XW - Notch t [h] Figure 3: Comparison of E9 uniaxial and notched bar stress/rupture test for HAZ, PM, WM and XW conditions at 625 C

8 ν σ net [MPa] PM - Uniaxial 0 2 PM - Notch XW - Uniaxial XW - Notch t [h] 0.7 Figure 4: Comparison of P9 uniaxial and notched bar stress/rupture test for PM and XW conditions at 600 C ν σ net [MPa] 0 2 HAZ - Uniaxial HAZ - Notch PM - Uniaxial PM - Notch WM - Uniaxial WM - Notch XW - Uniaxial XW - Notch t [h] Figure 5: Comparison of P92 uniaxial and notched bar stress/rupture test for HAZ, PM, WM and XW conditions at 625 C

9 N. S. F HAZ - Medium HAZ - Sharp PM - Medium PM - Sharp WM - Medium WM - Sharp XW - Medium XW - Sharp WM - Medium XW - Sharp PM - Medium PM - Sharp XW - Medium.2 WM - Sharp t [h] Figure 6: Comparison of E9 notched bar Notch Strengthening Factor (NSF) for HAZ, PM, WM and XW conditions at 625 C for two notch acuities and a range of rupture times ν Uniaxial Medium Sharp 8.0 σ eq [MPa] t [h] 0 4 Figure 7: Calculated average effective stress, σ eq, for parent E9 at 625 C notched bar versus t giving a slope, /-ν, of /-8 for the combined data.

10 CONCLUSIONS The comparison of creep rupture life of notched bar and plain bar creep tests on E9, P9 and P92 steels, in HAZ, PM, WM and XW conditions, are presented in terms of a equivalent stress criteria. The multiaxial stress rupture behaviour in the notched bars is determined by a combination of the von Mises stress and the maximum principal stress at a numerically determined skeletal point across the notch cross section. Notch strengthening, ranging between.-.7, in terms of net-section stress, over plain bar creep results, is presented for all notches and materials. Comparison between uniaxial tests for the steels show that the rupture index ν n. It has been found that for all material conditions, notched specimens show higher stress to rupture sensitivity than uniaxial specimens. Also notched weld specimens fail fastest followed by cross-weld and by parent material. The standard deviations between uniaxial and notched bar tests are similar but increase by about a factor of 2 if they are analysed together using an equivalent stress criterion. ACKNOWLEDGEMENTS This paper uses results from the Brite/Euram project LICON (BE95-309). The support from EEC and all the partners is gratefully acknowledged. EFEENCES. Draft LICON Code of Practice, (200), A Methodology for Life Prediction and Condition Assessment for Welds of efurbished and New Steam Cycle Plants, Brite/Euram Document LICON /CONS0/ T / 0, Oct Webster G.A, Holdsworth S., Loveday M.S, Perrin I.J & Purper H., (200), A Code of Practice for Conducting Notched Bar Creep upture Tests and for Interpreting the Data ESIS TC, Issue June Webster, G. A. and Nikbin, K. M. (200). Finite Element Analysis of Notched Bar Skeletal Point Stresses and Dimension Changes Due To Creep, to be Published in Int. J. of Fract. & Fatigue of Eng. Mat. and Struct., Hayhurst, D.., and Webster, G. A. (986). In: Techniques for Multiaxial Creep Testing. pp , Gooch, D. J. and How, I. (Ed.). Elsevier Applied Science, London.