Instrumented Impact Testing: Influence of Machine Variables and Specimen Position

Transcription

1 OPEN REPORT SCK CENBLG158 Instrumented Impact Testing: Influence of Machine Variables and Specimen Position Work performed during a 3month secondment at NIST, Boulder CO (USA), JuneAugust 8 Enrico Lucon Institute for Nuclear Material Science, SCK CEN, Mol (Belgium) Chris N. McCowan and Ray A. Santoyo National Institute for Standards and Technology, Material Reliability Division, Boulder CO (USA) September, 8 SCK CEN Boeretang BE24 Mol Belgium

2 OPEN REPORT OF THE BELGIAN NUCLEAR RESEARCH CENTRE SCK CENBLG158 Instrumented Impact Testing: Influence of Machine Variables and Specimen Position Work performed during a 3month secondment at NIST, Boulder CO (USA), JuneAugust 8 Enrico Lucon Institute for Nuclear Material Science, SCK CEN, Mol (Belgium) Chris N. McCowan and Ray A. Santoyo National Institute for Standards and Technology, Material Reliability Division, Boulder CO (USA) September, 8 Status: Unclassified ISSN SCK CEN Boeretang BE24 Mol Belgium

3 Distribution List R. Chaouadi E. Lucon J.L. Puzzolante J. Schuurmans M. Scibetta S. Van Dyck R. Gérard NMS, SCK CEN NMS, SCK CEN NMS, SCK CEN NMS, SCK CEN NMS, SCK CEN NMS, SCK CEN SUEZTractebel Secretariaat NMS SCK CEN Studiecentrum voor Kernenergie Centre d étude de l énergie Nucléaire Boeretang BE24 Mol Belgium Phone Fax Contact: Knowledge Centre library@sckcen.be RESTRICTED All property rights and copyright are reserved. Any communication or reproduction of this document, and any communication or use of its content without explicit authorization is prohibited. Any infringement to this rule is illegal and entitles to claim damages from the infringer, without prejudice to any other right in case of granting a patent or registration in the field of intellectual property. SCK CEN, Studiecentrum voor Kernenergie/Centre d'etude de l'energie Nucléaire Stichting van Openbaar Nut Fondation d'utilité Publique Foundation of Public Utility Registered Office: Avenue Herrmann Debroux 4 BE116 BRUSSEL Operational Office: Boeretang BE24 MOL

4 SCK CEN Open Report BLG158 Page 1 of 45 Preface This work was performed during my secondment at NIST (National Institute of Standards and Technology) in Boulder, Colorado (USA), between June and August 8. An abridged version of this report was submitted to Journal of Testing and Evaluation for publication on 25 August 8. The report is also published as a NIST Internal Report (NISTIR). Abstract An investigation has been conducted on the influence of impact machine variables and specimen positioning on characteristic forces and absorbed energies from instrumented Charpy tests. Brittle and ductile fracture behavior has been investigated by testing NIST reference samples of low, high and superhigh energy levels. Test machine variables included tightness of foundation, anvil and striker bolts, and the position of the center of percussion with respect to the center of strike. For specimen positioning, we tested samples which had been moved away or sideways with respect to the anvils. In order to assess the influence of the various factors, we compared mean values in the reference ("unaltered") and "altered" conditions; for machine variables, ttest analyses were also performed in order to evaluate the statistical significance of the observed differences. Our results indicate that the only circumstance which resulted in variations larger than 5% for both brittle and ductile specimens is when the sample is not in contact with the anvils. These findings should be taken into account in future revisions of instrumented Charpy test standards. Keywords Instrumented Charpy tests, impact machine variables, specimen positioning, ttest analysis.

5 SCK CEN Open Report BLG158 Page 2 of 45 Table of Contents Preface... 1 Abstract... 1 Keywords Introduction Material and experimental Loosening machine bolts Foundation bolts Anvil bolts Striker bolts Changing the position of the Center of Percussion (COP) Specimen positioning Specimen not in contact with the anvils Lateral specimen offset Discussion Conclusions References... 44

6 SCK CEN Open Report BLG158 Page 3 of 45 1 Introduction Impact testing has a history extending back to the 185's [1], when engineers realized that loading rate can significantly affect material properties. For structural steels, the main result of increasing loading rate is to raise the yield and tensile strength, which in turn causes a decrease in impact energy required for cleavage fracture, an increase in energy required for ductile fracture and a shift in the ductiletobrittle transition temperature. Loading rate must therefore be considered when designing structures subject to dynamic loading, such as buildings in earthquakes, ships and overland vehicles in collisions, aircraft landing gear and, at the highest rates, structures subjected to ballistic projectiles and explosive shocks. The conventional, noninstrumented Charpy test, which has recently celebrated its official centennial [2,3], can provide useful comparative information, but generally cannot offer any detailed insight into the failure mechanisms, or yield quantifiable material properties. This is also true for some other impact test methods, such as the Pellini dropweight and the Izod pendulum test. Instrumenting the machine striker in order to measure and record the force applied to the notched sample during impact provides a much clearer picture of the events occurring, providing additional insight on the behavior of material under impact loading. Instrumented Charpy tests are currently regulated by test standard ISO [4], while standardization is also in progress within ASTM [5]. For fracture mechanicsbased structural integrity evaluations, using fatigueprecracked Charpy (PCC) specimens is particularly useful. At the time of writing, ISO and ASTM standardization of fracture toughness tests at impact loading rates is ongoing [6,7]. In the nuclear field, the effects of neutron exposure on the fracture properties of the reactor pressure vessel can be assessed more reliably by using the results of instrumented Charpy tests than by applying the current regulatory practice, which relies on an empirical correlation between fracture toughness and the temperature corresponding to a fixed Charpy energy level (41 J). This is done in the framework of the socalled Enhanced Surveillance Approach [811], which includes the determination of characteristic force values from instrumented Charpy tests and the derivation of alternative index temperatures [12,13]. Numerous studies have been conducted in the past [1417] in order to investigate the experimental factors, related to both the machine and the specimen, which can affect the results of a Charpy test in terms of absorbed energy, as provided by the machine dial or the optical encoder (dial energy, KV). In this study, we have examined the influence of some of these machine/specimen variables on the results of instrumented Charpy tests, namely in terms of characteristic forces at general yield (F gy ), maximum forces (F m ) and absorbed energies calculated from the instrumented test record (W t ). KV values have been also included in the analyses, as well as the ratio between W t and KV. 1 From the test machine point of view, we focused our attention on the bolts that connect the pendulum to its foundation and those that hold the anvils and the instrumented striker in place. Tests were run with the bolts partially loose (anvil, striker) or completely loose (foundation), and the results were compared to data obtained with all the bolts tightened to the manufacturer s specifications. Additionally, the center of percussion (COP) was moved away from the center of strike (COS) by adding weights to the pendulum hammer. As far as specimen positioning is concerned, we investigated the effects of a lateral offset with respect to the centered position of the notch between the anvils and of a gap between the 1 Note that F gy values were not considered in the case of lowenergy specimens, since general yield does not occur or cannot be unambiguously defined in case of fully brittle behavior.

7 SCK CEN Open Report BLG158 Page 4 of 45 sample and the anvils. Again, results were compared to those obtained under ideal conditions (i.e. specimen centered and flush against the anvils). The ultimate goal of this investigation was to find out whether, with respect to the "conventional" dial energy KV, instrumented Charpy forces and energies are less, equally or more sensitive to experimental variations that could eventually occur during Charpy tests. 2 Material and experimental NISTcertified impactverification specimens of different energy levels were used in this study. For all investigated variables, low (~ 15 J) and high (~ 1 J) energy specimens were used; superhigh energy (~ 24 J) samples were tested only when looking at the effect of foundation bolts. Tests were performed on an instrumented TiniusOlsen Model 84 pendulum with 47.6 J capacity, located at NIST in Boulder, Colorado. The instrumented striker conforms to the requirements of ASTM E 23 (i.e. 8 mm radius of the striking edge). Tests were conducted at room temperature (RT, ~ 21 C), in order to avoid issues related to temperature control. However, when investigating the effect of loosening the foundation bolts, an additional set of lowenergy specimens was tested at 4 C. This is the prescribed temperature for both low and highenergy specimens, when tested for the verification of impact test machines. The overall test matrix is given in Table 1. In the reference condition (i.e. with the machine fully compliant to ASTM E 23), sets of 5 specimens were tested for each energy level, except for the lowenergy samples, where two sets were tested (one at RT and one at 4 C). Five to seven specimens per energy level and per condition were tested when investigating bolt loosening and center of percussion. Eight or nine tests per energy level were conducted to evaluate specimen positioning effects. Table 1 Overall test matrix. Variable considered Low energy High Superhigh RT 4 C energy energy Reference tests Loose foundation bolts Loose anvil bolts 5 6 Loose striker bolts 5 5 COP position changed 7 5 Lateral specimen offset 9 9 Specimen/anvils distance 8 8 TOTAL To provide additional benchmarking, results are compared to the outcome of a recent instrumented Charpy roundrobin exercise conducted by NIST among several European and US laboratories (6 participants) using the same low and highenergy specimens tested here (1 tests per material and laboratory) [18]. For the comparisons at room temperature, we used grand mean values for force from [18] and absorbed energy values. For tests at 4 C, and with superhigh energy specimens, no force data were available, so absorbed energy values were considered.

8 SCK CEN Open Report BLG158 Page 5 of Loosening machine bolts ASTM E 23 requires the impact machine to be securely bolted to a concrete floor at least 15 mm thick or to a foundation having a mass at least 4 times that of the pendulum. The direct verification for parts requiring annual inspection includes ensuring that the foundation bolts and the bolts that attach the anvils and striker to the machine are tightened to the manufacturer's specifications Foundation bolts The machine used in this study is bolted to a concrete foundation having a mass of approximately 15 kg, over 5 times that of the pendulum. The Tbolts securing the machine to the foundation are tightened to a torque value of 515 J (38 ftlb), which is higher than the manufacture specification. Charpy tests on low, high, and superhighenergy specimens were performed with the foundation bolts completely loosened. This is obviously an extreme situation which is unlikely to occur in real life. The results (instrumented forces, instrumented absorbed energy, dial energy and ratio) and the comparison with reference data are shown in Table 2 and Figure 1 through Figure 7. In the Figures, solid lines connect average values and error bars (often hardly visible) correspond to ± 2σ. Note that, in the case of superhighenergy specimens, the acquisition time used was too short and the tails of the instrumented curves were truncated. As a consequence, W t and values cannot be considered reliable and are, therefore, reported in italics; however, the comparison with the reference condition remains meaningful. Table 2 Effect of loosening the foundation bolts. NOTE: only NIST test results are given. Baseline Foundation unbolted Material F gy F m W t KV F gy F m W t KV (kn) (kn) (J) (J) (kn) (kn) (J) (J) Low Energy RT Average Standard dev Low Energy 4 C Average Standard dev

9 SCK CEN Open Report BLG158 Page 6 of 45 High Energy RT Average Standard dev Super High Energy RT Average Standard dev Force at general yield (kn) F gy grand mean from NIST roundrobin ± 95% repr. limit (high energy) 13 High energy RT Super high energy RT 1 Reference Foundation unbolted Figure 1 Effect of loosening the foundation bolts on forces at general yield. NIST roundrobin information only refers to highenergy specimens.

10 SCK CEN Open Report BLG158 Page 7 of Maximum force (kn) Low energy RT Low energy 4 C High energy RT Super high energy RT F m reference value from NIST roundrobin ± exp. uncert. 23 Reference Foundation unbolted Figure 2 Effect of loosening the foundation bolts on maximum forces. NIST roundrobin information refers to low and highenergy specimens tested at RT Low energy RT Low energy 4 C W t data from NIST round robin ± 1.4 J 19 W t (J) Reference Foundation unbolted Certified value ± 1.4 J (ASTM E23) Figure 3 Effect of loosening the foundation bolts on instrumented absorbed energies for lowenergy specimens. NOTES: tests at 4 C were not performed in the NIST round robin; certified values only refer to tests at 4 C.

11 SCK CEN Open Report BLG158 Page 8 of W t (J) 175 High energy RT Super high energy RT W t data from NIST round robin ± 5% 1 Reference Foundation unbolted Figure 4 Effect of loosening the foundation bolts on instrumented absorbed energies for high and superhighenergy specimens Low energy RT Low energy 4 C KV data from NIST round robin ± 1.4 J 19 KV (J) Certified value ± 1.4 J (ASTM E23) 1 Reference Foundation unbolted Figure 5 Effect of loosening the foundation bolts on dial energies for lowenergy specimens.

12 SCK CEN Open Report BLG158 Page 9 of Certified value ± 5% (ASTM E23) KV (J) 18 Super high energy RT High energy RT KV data from NIST round robin ± 5% 1 1 Reference Foundation unbolted Figure 6 Effect of loosening the foundation bolts on dial energies for high and superhighenergy specimens Low energy 4 C Low energy RT High energy RT 1.75 Reference Foundation unbolted Figure 7 Effect of loosening the foundation bolts on the ratio between instrumented and dial energy. NOTE: superhigh energy specimens aren't shown since test records are incomplete.

13 SCK CEN Open Report BLG158 Page 1 of 45 None of observed variations appears particularly substantial. However, we used a simple statistical test (unpaired ttest) to determine whether the differences between mean values are statistically significant or not at the 95% confidence level. The degree of statistical significance depends on the value of the twotailed probability P 2 using a threshold value.5, as follows: P >.5 not significant < P <.5 significant < P <.1 very significant P <.1 extremely significant The results of the ttest for the loose foundation bolts are shown in Table 3. No influence of the loosening of the foundation bolts is detected, except for the maximum forces of the highenergy specimens. Table 3 Results of the statistical ttest for the loose foundation bolts. Specimens / Energy level Parameter P Result (difference is ) F m.671 Low energy (RT) W t KV Low energy (4 C) F m W t KV High energy Superhigh energy F gy F m W t KV F gy F m W t KV.8789 < Extremely significant It is also interesting to observe that all test results (reference and unbolted) are in excellent agreement with the NIST roundrobin results. Note that, for the superhighenergy specimens, the test machine would pass the indirect verification even with the foundation completely unbolted (Figure 6). Coincidentally, the low energy level at 4 C (Figure 5) would fail the indirect verification in the reference condition but remarkably would pass with the foundation bolts completely loose! In this particular case, the absorbed energy is slightly decreased for the unbolted condition. This is contrary to what might be expected, but the scatter for the unbolted result is much higher than for the bolted condition. The ratio between dial and instrumented energy (Figure 7) is fairly consistent between lowand high energy samples, and it is very close to unity for the tests conducted at 4 C. A direct comparison of the instrumented test records (Figure 8 through Figure 11, using only one curve per condition (for the sake of clarity) does not offer any additional insight. 2 In statistical hypothesis testing, P is the probability of obtaining a value of the test statistic at least as extreme as the one that was actually observed, given that the null hypothesis (i.e. no difference between the means) is true.

14 SCK CEN Open Report BLG158 Page 11 of Force (kn) Baseline Unbolted Displacement (m) Figure 8 Effect of loosening the foundation bolts on the instrumented test records of the lowenergy specimens (RT) Force (kn) Baseline Unbolted Displacement (m) Figure 9 Effect of loosening the foundation bolts on the instrumented test records of the lowenergy specimens (4 C).

15 SCK CEN Open Report BLG158 Page 12 of Force (kn) Baseline Unbolted Displacement (m) Figure 1 Effect of loosening the foundation bolts on the instrumented test records of the highenergy specimens Baseline Unbolted Force (kn) Displacement (m) Figure 11 Effect of loosening the foundation bolts on the instrumented test records of the superhigh energy specimens. Note the curves truncated at.14 m.

16 SCK CEN Open Report BLG158 Page 13 of Anvil bolts The torque applied to the anvil bolts for the baseline condition is 36.6 J (27 ftlb), as specified by the manufacturer. The applied torque on the bolts was reduced to 2.8 J (25 inlb), a condition which might be described as "fingertight". Five tests on lowenergy specimens and five tests on highenergy specimens were performed at RT with loosened anvils bolts and the results compared to the baseline data (Table 4 and Figure 12 through Figure 16). Table 4 Effect of loosening the anvil bolts. Baseline Anvil bolts loosened Material F gy F m W t KV F gy F m W t KV (kn) (kn) (J) (J) (kn) (kn) (J) (J) Low Energy Average Standard dev High Energy Average Standard dev

17 SCK CEN Open Report BLG158 Page 14 of F gy grand mean from NIST roundrobin ± 95% repr. limit 15 F gy (kn) 1 5 Reference Anvil bolts loosened Figure 12 Effect of loosening the anvil bolts on forces at general yield for the highenergy specimens F m (kn) Low energy High energy F m reference value from NIST roundrobin ± exp. uncert. 23 Reference Anvil bolts loosened Figure 13 Effect of loosening the anvil bolts on maximum forces.

18 SCK CEN Open Report BLG158 Page 15 of W t (J) 6 High energy Low energy W t data from NIST round robin ± 1.4 J or 5% 4 Reference Anvil bolts loosened Figure 14 Effect of loosening the anvil bolts on instrumented absorbed energies KV (J) 6 High energy Low energy KV data from NIST round robin ± 1.4 J or 5% 4 Reference Anvil bolts loosened Figure 15 Effect of loosening the anvil bolts on dial energies.

19 SCK CEN Open Report BLG158 Page 16 of Low energy High energy 1.75 Reference Anvil bolts loosened Figure 16 Effect of loosening the anvil bolts on the ratio between instrumented and dial energy. The results of the ttest to assess the statistical significance of the observed differences between mean values are provided in Table 5. Table 5 Results of the statistical ttest for the loosened anvil bolts. Specimens / Energy level Parameter P Result (difference is ) F m.144 Significant Low energy W t KV High energy F gy F m W t KV Extremely significant The effect of the loosened anvil bolts appears significant only on maximum force data. However, it is noted that F m increases for the lowenergy specimens and decreases for the highenergy specimens (Figure 13). ratios remain substantially constant and the difference between dial and instrumented energies does not exceed 5% (Figure 16). The direct comparison of instrumented curves (Figure 17 and Figure 18) does not offer additional insight.

20 SCK CEN Open Report BLG158 Page 17 of Force (kn) Baseline Anvil bolts loosened Displacement (m) Figure 17 Effect of loosening the anvil bolts on the instrumented test records of the lowenergy specimens Force (kn) Baseline Anvil bolts loosened Displacement (m) Figure 18 Effect of loosening the anvil bolts on the instrumented test records of the highenergy specimens.

21 SCK CEN Open Report BLG158 Page 18 of Striker bolts The bolts holding the instrumented striker in place were loosened from manufacture specification of 74.6 J (55 ftlb) to 4.5 J (4 inlb), which again might be described as "fingertight". Five tests on lowenergy specimens and five tests on highenergy specimens were performed at RT with loosened striker bolts and the results compared to the baseline data (Table 6 and Figure 19 through Figure 23). Table 6 Effect of loosening the striker bolts. Baseline Striker bolts loosened Material F gy F m W t KV F gy F m W t KV (kn) (kn) (J) (J) (kn) (kn) (J) (J) Low Energy Average Standard dev High Energy Average Standard dev

22 SCK CEN Open Report BLG158 Page 19 of F gy grand mean from NIST roundrobin ± 95% repr. limit 15 F gy (kn) 1 5 Reference Striker bolts loosened Figure 19 Effect of loosening the striker bolts on forces at general yield for the highenergy specimens F m (kn) Low energy High energy F m reference value from NIST roundrobin ± exp. uncert. 23 Reference Striker bolts loosened Figure Effect of loosening the striker bolts on maximum forces.

23 SCK CEN Open Report BLG158 Page of W t (J) 6 High energy Low energy W t data from NIST round robin ± 1.4 J or 5% 4 Reference Striker bolts loosened Figure 21 Effect of loosening the striker bolts on instrumented absorbed energies KV (J) 6 High energy Low energy KV data from NIST round robin ± 1.4 J or 5% 4 Reference Striker bolts loosened Figure 22 Effect of loosening the striker bolts on dial energies.

24 SCK CEN Open Report BLG158 Page 21 of Low energy High energy 1.75 Reference Striker bolts loosened Figure 23 Effect of loosening the striker bolts on the ratio between instrumented and dial energy. The statistical significance of the observed differences between mean values was assessed by means of the ttest at a confidence level of 95% (Table 7). Table 7 Results of the statistical ttest for the loosened striker bolts. Specimens / Energy level Parameter P Result (difference is ) F m.25 Very significant Low energy W t KV Significant High energy F gy F m W t KV Very significant In comparison with foundation and anvil bolts, loosening the striker bolts seems to have somewhat more detectable effects, particularly for the lowenergy specimens. The difference between dial and instrumented energies remains within 5% regardless of the torque applied to the striker bolts (Figure 23). Once again, no additional insight is provided by comparing instrumented test records for the two conditions (Figure 24 and Figure 25).

25 SCK CEN Open Report BLG158 Page 22 of Force (kn) Baseline Striker bolts loosened Displacement (m) Figure 24 Effect of loosening the striker bolts on the instrumented test records of the lowenergy specimens Force (kn) Baseline Striker bolts loosened Displacement (m) Figure 25 Effect of loosening the striker bolts on the instrumented test records of the highenergy specimens.

26 SCK CEN Open Report BLG158 Page 23 of Changing the position of the Center of Percussion (COP) The center of percussion (COP) of an object is defined as the point where a perpendicular impact will produce translational and rotational forces which perfectly cancel each other out at some given pivot point so that the pivot will not be moving momentarily after the impulse. Centers of percussion are often discussed in the context of a bat, racquet, sword or other long thin objects. For such objects, the COP may or may not be the "sweet spot" depending on the pivot point chosen. ASTM E 23 requires the COP of the pendulum hammer to be within 1% of the distance between the axis of rotation and the center of strike (COS) on the specimen (corresponding to the length of the pendulum). This requirement ensures that minimum force is transmitted to the point of rotation. Moving COP and COS away from each other is expected to cause additional vibrations and therefore increase vibrational energy losses, particularly when testing brittle materials. The location of the COP is determined from the equation: 2 gp L = (1) 2 4π where g is the local gravitational acceleration in m/s², p is the period of a complete oscillation measured with a suitable timemeasuring device in s, and L is the distance between the axis of rotation and the COP in m. The length of the pendulum used in this study is 9.65 mm, and the original position of the COP is mm from the axis of rotation. The COPCOS distance is therefore 1.86 mm, or.2% of the pendulum length. By adding 2 kg to the pendulum mass, we reduced the oscillation period and therefore shortened the distance between axis of rotation and COP to mm and increased the COPCOS distance to 6.95 mm (.77% of the pendulum length 3 ). Subsequently, tests were run on low and highenergy specimens and results were compared to data from reference tests and the NIST round robin (Table 8 and Figure 26 through Figure 3). Table 8 Effect of moving the COP away from the COS 4. Baseline COP moved Material F gy F m W t KV F gy F m W t KV (kn) (kn) (J) (J) (kn) (kn) (J) (J) Low Energy Average Standard dev It was not possible to add more weight to the pendulum mass, and therefore increase further the COPCOS distance. 4 For one of the low energy tests, instrumented data were lost. For two additional low energy tests, an incorrect value of pendulum length was entered and therefore KV is not reliable.

27 SCK CEN Open Report BLG158 Page 24 of 45 Baseline COP moved Material F gy F m W t KV F gy F m W t KV (kn) (kn) (J) (J) (kn) (kn) (J) (J) High Energy Average Standard dev F gy grand mean from NIST roundrobin ± 95% repr. limit 15 F gy (kn) 1 5 Reference COP moved away Figure 26 Effect of changing the COP position on forces at general yield for the highenergy specimens.

28 SCK CEN Open Report BLG158 Page 25 of F m (kn) Low energy High energy F m reference value from NIST roundrobin ± exp. uncert. 23 Reference COP moved away Figure 27 Effect of changing the COP position on maximum forces W t (J) 6 High energy Low energy W t data from NIST round robin ± 1.4 J or 5% 4 Reference COP moved away Figure 28 Effect of changing the COP position on instrumented absorbed energies.

29 SCK CEN Open Report BLG158 Page 26 of KV (J) 6 High energy Low energy KV data from NIST round robin ± 1.4 J or 5% 4 Reference COP moved away Figure 29 Effect of changing the COP position on dial energies Low energy High energy 1.75 Reference COP moved away Figure 3 Effect of changing the COP position on the ratio between instrumented and dial energy.

30 SCK CEN Open Report BLG158 Page 27 of 45 Table 9 shows the results of the statistical analyses conducted using the ttest. Table 9 Results of the statistical ttest for the modified COP location tests. Specimens / Energy level Parameter P Result (difference is ) F m.1453 Low energy W t KV Extremely significant High energy F gy F m W t KV Significant The most significant effect is observed on instrumented energies for the lowenergy specimens which, contrary to expectations, decreased when the COP was moved away from the COS. As a consequence, the ratio between instrumented and dial energies falls below.95 (Figure 3). Figure 31 and Figure 32 compare instrumented test records for low and high energy samples Force (kn) Baseline COP moved away Displacement (m) Figure 31 Effect of changing the COP position on the instrumented test records of the lowenergy specimens.

31 SCK CEN Open Report BLG158 Page 28 of Force (kn) COP moved away Baseline Displacement (m) Figure 32 Effect of changing the COP position on the instrumented test records of the highenergy specimens. 2.3 Specimen positioning The influence of specimen impact position was investigated by moving the samples away from the correct location (i.e. flush against the anvils and with the notch centered between the anvils). Conditions with specimens moved away from flush, and specimen notch positions moved off center were evaluated. For every selected distance, multiple impact tests on low and highenergy specimens were performed. Due to the limited number of data available for each condition (from 2 to 4), we decided not to perform any statistical analysis (ttest) Specimen not in contact with the anvils In everyday laboratory practice, it is not uncommon that a Charpy specimen is not fully in contact with the anvils, particularly in the case of manual positioning and when the tests are performed at low or high temperatures, since the operator must fulfill the < 5 s transfer time requirement (ASTM E 23). A sample which was not against the anvils can be easily detected from its instrumented test record, which shows significantly amplified dynamic oscillations up to general yield and a characteristic delay in the rise of the force signal with respect to the onset of data acquisition. For our study, we selected three values of the distance between specimen and anvils:.5 mm, 1 mm and 2 mm. Up to 5 tests per condition were performed, although instrumented data were lost for several tests as will be discussed later. The results are summarized and compared with the reference data and NIST roundrobin results in Table 1 and Figure 33 through Figure 37.

32 SCK CEN Open Report BLG158 Page 29 of 45 Table 1 Effect of the distance between specimen and anvils. Low Energy High Energy Distance F m W t KV F gy F m W t KV from anvils (kn) (J) (J) (kn) (kn) (J) (J) mm (reference) Average Standard dev Distance from F m W t KV F gy F m W t KV W anvils t /KV (kn) (J) (J) (kn) (kn) (J) (J) mm Average Standard dev Distance from F m W t KV F gy F m W t KV W anvils t /KV (kn) (J) (J) (kn) (kn) (J) (J) 1 mm Average Standard dev Distance from F m W t KV F gy F m W t KV W anvils t /KV (kn) (J) (J) (kn) (kn) (J) (J) mm Average Standard dev

33 SCK CEN Open Report BLG158 Page 3 of F gy (kn) 15 1 F gy grand mean from NIST roundrobin ± 95% repr. limit Specimen distance from anvils (mm) Figure 33 Effect of the distance between specimen and anvils on forces at general yield for the highenergy specimens. 4 Low energy 35 F m (kn) 3 High energy F m reference value from NIST roundrobin ± exp. uncert Specimen distance from anvils (mm) Figure 34 Effect of the distance between specimen and anvils on maximum forces.

34 SCK CEN Open Report BLG158 Page 31 of High energy 1 W t (J) 8 6 W t data from NIST round robin ± 1.4 J or 5% 4 Low energy Specimen distance from anvils (mm) Figure 35 Effect of the distance between specimen and anvils on instrumented absorbed energies High energy 1 KV (J) 8 6 KV data from NIST round robin ± 1.4 J or 5% 4 Low energy Specimen distance from anvils (mm) Figure 36 Effect of the distance between specimen and anvils on dial energies.

35 SCK CEN Open Report BLG158 Page 32 of High energy Low energy Specimen distance from the anvils (mm) Figure 37 Effect of the distance between specimen and anvils on the ratio between instrumented and dial energies. Increasing the distance between specimen and anvils results in a generalized increase of the characteristic values of force and energy, as well as the ratio. The increment is particularly significant for maximum force values measured from lowenergy specimens (Figure 34), where the enhanced dynamic oscillations lead to a significant overestimation of F m. For the same reason, the uncertainty associated with F gy values from highenergy specimens increases, while ringing effects are sufficiently dampened once maximum force is reached. In general, brittle specimens are more affected than ductile specimens. As previously mentioned, the occurrence of instrumented data loss was problematic for the low energy samples at 1 mm and 2 mm distance: data were not acquired in 5 out of 9 tests, whereas in only one instance (out of 6 tests performed) high energy data were lost. No acquisition failure was recorded for specimens placed.5 mm away from the anvils. It is therefore clear that distances of 1 mm or more increase the likelihood of acquisition failure, probably due to false triggering. Two interesting features emerge from the comparison of instrumented test records (Figure 38 and Figure 39): the traces of the offset lowenergy specimens are clipped at a force level of 35.6 kn, which corresponds to the threshold of the acquisition system (Figure 38); curves for the reference condition do not reach this force level; on the displacement axis, the rise of the force signal roughly corresponds to the initial distance between specimen and anvils. In all cases, large inertia peaks are observed at the beginning of the instrumented traces, caused by the specimen rebounding between anvils and striker before full contact is achieved.

36 SCK CEN Open Report BLG158 Page 33 of Force (kn) Reference ( mm).5 mm 1 mm 2 mm Displacement (m) Figure 38 Effect of the distance between specimen and anvils on the instrumented test records of lowenergy specimens. Note the clipped force signals above 35.6 kn Force (kn) Reference ( mm).5 mm 1 mm 2 mm Displacement (m) Figure 39 Effect of the distance between specimen and anvils on the instrumented test records of highenergy specimens.

37 SCK CEN Open Report BLG158 Page 34 of Lateral specimen offset Annex A1 of ASTM E 23 prescribes that the specimen be positioned against the anvils such that the center of the notch is located within.25 mm of the midpoint between the anvils. The use of selfcentering tongs is recommended to achieve this. In our study, we selected three values for the specimen lateral offset:.5 mm, 1 mm and 2 mm. All distances were outside the ASTM E 23 tolerance. For each condition, three tests on lowenergy and three test on highenergy specimens were conducted. No failed data acquisition was experienced. Results and comparison with reference values and NIST roundrobin data are provided in Table 11 and Figure 4 through Figure 44. Table 11 Effect of lateral specimen offset. Low Energy High Energy Lateral offset F m W t KV F gy F m W t KV (kn) (J) (J) (kn) (kn) (J) (J) mm (reference) Average Standard dev Lateral offset F m W t KV F gy F m W t KV (kn) (J) (J) (kn) (kn) (J) (J) mm Average Standard dev Lateral offset F m W t KV F gy F m W t KV (kn) (J) (J) (kn) (kn) (J) (J) mm Average Standard dev Lateral offset F m W t KV F gy F m W t KV (kn) (J) (J) (kn) (kn) (J) (J) mm Average Standard dev

38 SCK CEN Open Report BLG158 Page 35 of F gy (kn) 15 1 F gy grand mean from NIST roundrobin ± 95% repr. limit Lateral specimen offset (mm) Figure 4 Effect of lateral specimen offset on forces at general yield for the highenergy specimens Low energy F m (kn) 3 25 High energy F m reference value from NIST roundrobin ± exp. uncert Lateral specimen offset (mm) Figure 41 Effect of lateral specimen offset on maximum forces.

39 SCK CEN Open Report BLG158 Page 36 of High energy 1 W t (J) 8 6 W t data from NIST round robin ± 1.4 J or 5% 4 Low energy Lateral specimen offset (mm) Figure 42 Effect of lateral specimen offset on instrumented absorbed energies High energy 1 KV (J) 8 6 KV data from NIST round robin ± 1.4 J or 5% 4 Low energy Lateral specimen offset (mm) Figure 43 Effect of lateral specimen offset on dial energies.

40 SCK CEN Open Report BLG158 Page 37 of Low energy High energy Lateral specimen offset (mm) Figure 44 Effect of lateral specimen offset on the ratio between instrumented and dial energies. Moving the specimen offcenter appears to have a systematic, though moderate effect on instrumented forces. The force values steadily increase with the offset (Figure 4 and Figure 41). The effect is less pronounced and not systematic for energy values and ratio (Figure 42 to Figure 44), with both W t and KV being slightly lower for 2 mm than for 1 mm for the highenergy specimens. No substantial features emerge from the comparison of instrumented test records (Figure 45 and Figure 46). For low energy tests, the amplitude of dynamic oscillations seems to increase with respect to the baseline condition, but it is not directly correlated to the magnitude of the specimen offset.

41 SCK CEN Open Report BLG158 Page 38 of Force (kn) Reference ( mm).5 mm 1 mm 2 mm Displacement (m) Figure 45 Effect of lateral specimen offset on the instrumented test records of lowenergy specimens Force (kn) Reference ( mm).5 mm 1 mm 2 mm Displacement (m) Figure 46 Effect of lateral specimen offset on the instrumented test records of highenergy specimens.

42 SCK CEN Open Report BLG158 Page 39 of 45 3 Discussion Table 12 summarizes the results of the statistical ttest applied to the results of the tests conducted on low, high and superhighenergy specimens in order to investigate the effect of test machine variables. In the Table, the following convention is used to indicate the statistical significance (at the 95% confidence level) of the observed differences between mean values: Symbol Meaning * Significant ** Very significant *** Extremely significant Table 12 Statistical significance of the various machine parameters based on the ttest results. Test machine variable Foundation bolts Anvil bolts Striker bolts COP position Test parameter F gy F m W t KV F gy F m W t KV F gy F m W t KV F gy F m W t KV Low energy specimens (RT) * ** * *** Low energy specimens (4 C) High energy specimens *** *** ** * Superhigh energy specimens From the examination of Table 12, it is generally clear that the investigated effects are neither particularly substantial nor systematic, nor is the consistency between the different parameters and energy levels. Therefore, it is unlikely that a user could detect a given effect from a simple examination of an individual test record, without resorting to a more detailed statistical analysis. Force values (namely maximum forces) seem to be slightly more sensitive than energy values. As far as specimen position is concerned, our results show that moving the samples away from the anvils is more influential than displacing it sideways. The effects are clearly visible on the instrumented test record and are particularly significant when the distance between specimen and anvils exceeds.5 mm. Brittle specimens are more affected than ductile specimens, and for our test system clipping of the force signal due to saturation of the strain gage signal was observed. Moreover, the likelihood of losing the instrumented data due to mistriggering is greatly increased. A lateral offset of the sample affects forces more than energies, but the consequences are not very pronounced. Table 13 and Figure 47 through Figure 49 summarize the outcome of our investigation for brittle and ductile Charpy tests. Data are presented in terms of average forces and energies measured after "altering" one of the experimental parameters, as a function of the average values for the reference ("unaltered") condition.

43 SCK CEN Open Report BLG158 Page 4 of 45 Table 13 Average values of characteristic forces and absorbed energies obtained in this study. Fracture behavior Brittle Ductile Material/Tests Low energy (RT) Low energy (4 C) High energy Superhigh energy Variable considered F gy F m W t KV (kn) (kn) (J) (J) Reference condition Foundation bolts Anvil bolts Striker bolts Moving COP Not against anvils (.5 mm) Not against anvils (1 mm) Not against anvils (2 mm) Lateral offset (.5 mm) Lateral offset (1 mm) Lateral offset (2 mm) Reference condition Foundation bolts Reference condition Foundation bolts Anvil bolts Striker bolts Moving COP Not against anvils (.5 mm) Not against anvils (1 mm) Not against anvils (2 mm) Lateral offset (.5 mm) Lateral offset (1 mm) Lateral offset (2 mm) Reference condition Foundation bolts

44 SCK CEN Open Report BLG158 Page 41 of % +5% 1:1 F m 5% Altered values (Fm kn; Wt,KV J) W t, KV 1% Foundation unbolted Anvil bolts loosened Striker bolts loosened Moving COP away from COS Sample.5 mm from anvils Sample 1 mm from anvils Sample 2 mm from anvils Lateral offset.5 mm Lateral offset 1 mm Lateral offset 2 mm Reference values (Fm kn; Wt,KV J) Figure 47 Effect of testmachine variables and specimen positioning for brittle specimens (lowenergy level at RT and 4 C) % +5% 1:1 F m 5% 1% Altered force values (kn) 24 F gy Foundation unbolted Anvil bolts loosened Striker bolts loosened Moving COP away from COS Sample.5 mm from anvils Sample 1 mm from anvils Sample 2 mm from anvils Lateral offset.5 mm Lateral offset 1 mm Lateral offset 2 mm Reference force values (kn) Figure 48 Effect of testmachine variables and specimen positioning on characteristic forces for ductile specimens (high and superhigh energy).

45 SCK CEN Open Report BLG158 Page 42 of Superhigh energy +1% +5% 1:1 5% 1% Altered energy values (J) Foundation unbolted Anvil bolts loosened Striker bolts loosened Moving COP away from COS Sample.5 mm from anvils Sample 1 mm from anvils Sample 2 mm from anvils Lateral offset.5 mm Lateral offset 1 mm Lateral offset 2 mm 125 Low energy Reference energy values (J) Figure 49 Effect of testmachine variables and specimen positioning on absorbed energies for ductile specimens (high and superhigh energy). In the case of brittle behavior (Figure 47), most of the investigated variables tend to increase maximum forces and absorbed energies as a result of additional vibrational energy losses, except when the COP is moved away from the COS. In general, altering the specimen position before impact is more influential than loosening the test machine bolts. The variations observed with respect to the reference condition are relatively moderate, i.e. less than 5%, except when the specimen is moved away from the anvil by more than.5 mm. For specimen/anvil distances of 1 and 2 mm, absorbed energies increase by much more than 1%. Fully ductile specimens (Figure 48 and Figure 49) are generally less sensitive than brittle specimens to the investigated variables. In particular (Figure 48), maximum forces are essentially unaffected (variations well below 5%), whereas for general yield forces only the distance between specimen and anvil causes variations between 5% and 1%. In terms of absorbed energies (Figure 49), from both the instrumented test record and from the machine encoder, an increase is generally observed, but no larger than 56% of the reference value.

46 SCK CEN Open Report BLG158 Page 43 of 45 4 Conclusions Our investigation on the effect of testmachine variables (bolting of foundation, anvils and striker; position of the center of percussion) on instrumented Charpy test results leads to the general conclusion that most of the observed variations are lower than 1% and generally neither consistent nor systematic, particularly in case of fully ductile behavior. As a consequence, they are unlikely to be detected from the instrumented test record or from the value of absorbed energy yielded by the machine encoder. The most notable effect is when the specimen is not in contact with the anvils; particularly for brittle tests (such as the lowenergy NISTreference samples) and when the sample is more than.5 mm from the anvils, absorbed energies can be overestimated by considerably more than 1%. It would be advisable that instrumented Charpy test standards require data from Charpy tests in which the gap between specimen and anvils was more than.5 mm to be discarded. This requirement would be relatively easy to implement, since (a) the test record of such a test can be easily recognized (delay between test initiation and force signal rise; very pronounced dynamic oscillations prior to general yield), and (b) the displacement corresponding to the rise of the force signal provides a rough estimate of the initial distance between specimen and anvils. One the most significant outcomes of our study is that a "good" machine has a high probability of still delivering high quality results, including passing the indirect verification of ASTM E 23, even when it is unbolted from its foundation, the striker and anvils are just "finger tightened" and the specimens are offset from the midpoint between the anvils. However, we expect that these results are strongly influenced by machine design. So, the magnitude of the effects reported here may differ substantially from results obtained on a machine having a different design. Finally, we can say that instrumented forces are more sensitive than absorbed energies to the variables considered, whereas calculated (W t ) and dial (KV) energies are equally (in)sensitive. This implies that, if Charpy tests are performed using an instrumented striker, effects can be detected which would not be noticed if one looked at only energy values returned by the machine encoder.