- Email: [email protected]
Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness
Minerals Engineering xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Minerals Engineering journal homepage: www.elsevier.com/locate/mineng
Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness Amar Sabih a, Peter Radziszewski b,⇑, Ivan Mullany c a
Mechanical Engineering, McGill University, 817 Sherbrooke St. W., Montreal H3A 0C3, Canada Metso, 795 ave George-V, Lachine, Québec H8S 2R9, Canada c Hatch, 2800 Speakman Drive, Sheridan Science & Technology Park, Mississauga, Ontario L5K 2R7, Canada b
a r t i c l e
i n f o
Article history: Received 30 May 2016 Revised 16 August 2016 Accepted 19 August 2016 Available online xxxx Keywords: Grinding media Hardness Abrasion Toughness Wear
a b s t r a c t It is recognised that grinding media wear can represent up to 50% of the operating costs of a given tumbling mill. Over the years, a number of works have explored the development of different ways and means to both understand grinding media wear as well as model and predict it. The focus of the present work is to examine the differences in microstructure, hardness, abrasion and impact toughness of grinding media from eight different manufacturing sources. Results will be presented for 125 mm diameter media typically used for SAG mills. A discussion will address any issues highlighted by the results than may contribute to predictive wear model development as well as indicate possible directions for future research. Ó 2016 Published by Elsevier Ltd.
1. Introduction In mineral processing, the two main operating costs are related to energy consumption and wear. Energy consumption in comminution processes especially as it related to efficiency has been the focus of much research and application. Current efforts aim at developing guidelines and potentially standards for energy based grinding performance metrics (McIvor, 2015) and benchmarks (Ballantyne and Powell, 2014; Nadolski et al., 2015). On the other hand, with respect to wear and particularly grinding media wear, there are still a few different schools of thought (Bond, 1963; Benavente, 2007; Gates et al., 2008; Chenje et al., 2009) all of which converge on the general notion that wear in comminution processes is a function of three main components which are the energy involved in wear, the chemical and mechanical properties of the media as well as the chemical and mechanical properties of the ore or slurry. The focus of the present paper will be to examine the components related to the mechanical properties of the media. Specifically, the focus will be on 125 mm diameter grinding media and will investigate the differences in microstructure, impact toughness, abrasion and hardness of grinding media from eight different manufacturing sources. Following a presentation of the results, a
⇑ Corresponding author. E-mail address: [email protected] (P. Radziszewski).
discussion will explore any possible relationships as well as possible avenues for future research.
2. Sample preparation Ten different SAG mill 125 mm diameter media samples from eight different manufacturing sites was collected for this investigation. The different media sampled are randomly listed as illustrated in Table 1. With respect to chemical composition, it should be noted that the standards ASTM E1479 (2011) and ASTM E1019 (2011) were followed. The chemical analysis indicated the percent weight results for aluminium, arsenic, carbon, cobalt, chromium, copper, manganese, molybdenum, niobium, nickel, phosphorus, silicon, sulfur, tin, tantalum, titanium, vanadium, zirconium and tungsten. Table 2 provides the percent weight results for chemical components common to all media sampled, while Table 3 indicates that number of media samples that contained the chemical components not common to all samples tested along with their range. Subsequently, the media samples were then prepared in order to accomplish the different tests for microstructure, impact fracture toughness, abrasion testing and hardness. All samples were cut under wet conditions such that a number of samples were obtained without microstructure modification including three from different radial positions in the ball as illustrated in Fig. 1. In the case of impact toughness, samples were prepared according
http://dx.doi.org/10.1016/j.mineng.2016.08.014 0892-6875/Ó 2016 Published by Elsevier Ltd.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
2
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
Table 1 Selected grinding media with the corresponding codes (125 mm media).
Table 3 The chemical composition results common to all media samples (%wt).
Coded provider
Coded provider
A5 B5 C5 D5 E5
F5 G5 H5 I5 J5
to the requirements of the Charpy V-notch specifications defined by the ASTM E23 (2002). The hardness profile was achieved by conducting a series of tests along the center line of the 125 mm media slice from which Charpy test specimens were machined from. Samples for abrasion wear test did not have any special preparation other than meeting the required sample size and shape for use in steel wheel abrasion test as described by the ASTM G65 (2015) and ASTM B611 (2013) standards. Chemical analysis was completed on the fractured halves produced by the Charpy impact test. Surfaces were prepared in the typical fashion required for metallographic analysis. This includes sample mounting in Bakelite, mechanical grinding, abrasion and polishing, etching followed by microscopic analysis (Modin and Modin, 1973; ASM, 1985; Vander Voort, 2007). 3. Results The results are presented in the following order: microstructure, impact toughness, abrasive wear and hardness. 3.1. Microstructure The microstructures of all grinding media investigated in this work are shown in Figs. 2–15. All appear to be quenched and tempered martensite, with two exceptions: (i) C5, the high C and high Cr alloys, which reveals a microstructure that is not readily identifiable (Fig. 15); (ii) J5 sample reveals the cast structure of dendrites with what appears to be a matrix of ledeburite i.e. the eutectic formed at the final stage of liquid solidification, which should be a mixture of white cementite and transformed austenite (Figs. 14 and 15). Apart from the white cementite matrix, the darker areas could be tempered martensite. Higher magnifications are required to clarify this. At optical microscopy magnifications, it is extremely difficult to differentiate between the characteristics of quenched and tempered martensite. Only the martensite laths can really be resolved, but it is still difficult to differentiate between martensite morphologies. Carbides and retained austenite cannot be resolved. These characteristics need to be defined by electron microscopy and X-ray analysis, if necessary. A literature review is required to ascertain whether the degree of tempering, or changes in martensite morphology, affect wear. Certainly the degree of tempering affects toughness, and is well documented. There appears to be inclusions (black particles) in all the grinding media samples.
% Aluminium % Arsenic % Cobalt %Molybdenum % Niobium % Tin % Tantalum % Titanium % Vanadium % Zirconium % Tungsten
No. of samples
Range (wt.%)
8 0 1 6 1 3 1 2 1 0 0
0.02 <0.01 0.01 0.01–0.06 0.08 0.01–0.08 0.08 0.01 0.03 <0.01 <0.01
These need to be quantified at some point since these may affect the toughness of the grinding media. A5 media: The microstructure of A5 media consists of tempered martensite but retained austenite was difficult to find in the microstructure (Fig. 2). Although this is a hypo-eutectoid structure, no pro-eutectoid ferrite was observed at grain boundaries. This may explain the high toughness for this media sample as compared to other samples tested. As the hardness is comparable with other, this could be as a result of good heat treatment or performing forging treatment. B5 media: The microstructure is composed of lath martensite and small islands (5–10 lm) inclusions of irregular shape. With regards to the ball application and measured hardness, the martensite seems to be slightly tempered in order to increase toughness and decrease retained austenite. The brighter area in Figs. 3 and 4 could be tempered retained austenite. C5 media: Based on the composition and in terms of the carbon content it is a hyper-eutectoid structure. The microstructure shows chromium carbides in a matrix of untempered martensite, which should lower dramatically the toughness (Fig. 5). There should be relatively high amount of retained austenite (not resolved with Nital etchant). D5 media: This grinding media sample is composed of tempered martensite, retained austenite and inclusions. The martensite looks to be a mixture of lath and needles due to the carbon content of this structure. The lamellar style of microstructure implies deformed structure due to forging treatment (Fig. 6). The inclusions were also elongated as a proof of forging. Precipitated ferrite was observed in this steel due to its pro-eutectoid characteristics. The relatively good toughness can be attributed to the forging treatment applied on this media sample despite of the presence of detrimental ferrite at primary austenite grain boundaries (Fig. 7). E5 media: The composition is almost the same as D5 media (Fig. 8), however, this sample showed lower hardness and higher toughness. Ferrite at the grain boundaries was observed in limited areas as compared to D5. The presence of this ferrite may contribute to lowering the impact toughness as compared to the D5 sample.
Table 2 The chemical composition results common to all media samples (wt.%).
% % % % % % % %
Carbon Chromium Copper Manganese Nickel Phosphorus Silicon Sulfur
A5
B5
C5
D5
E5
F5
G5
H5
I5
J5
0.551 0.78 0.10 0.69 0.05 0.012 1.75 0.003
0.623 0.98 0.15 0.72 0.08 0.011 0.73 0.025
2.10 18.61 0.05 0.31 0.11 0.020 0.38 0.041
0.577 0.64 0.05 0.97 0.03 0.016 0.42 0.010
0.524 0.63 0.03 0.96 0.03 0.015 0.43 0.004
0.669 0.33 0.25 0.85 0.14 0.012 0.19 0.028
0.764 0.75 0.21 0.85 0.14 0.016 0.16 0.027
0.572 0.68 0.02 0.99 0.02 0.015 0.42 0.005
0.608 0.96 0.15 0.72 0.13 0.013 0.77 0.021
1.97 12.68 0.06 1.04 0.14 0.036 0.72 0.031
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
3
Fig. 1. Schematics drawings of a slice of the 125 mm media showing the locations where Charpy test specimens were cut from using EDM wire cutting and subsequent grinding.
Fig. 4. The microstructure of B5 media (50). Fig. 2. Microstructure of A5 media (1000).
Fig. 5. Microstructure of C5 media (500). Fig. 3. B5 media (1000), the brighter area could be tempered retained austenite.
F5 media: The microstructure for F5 steel shows tempered martensite and discontinued ferrite at the boundaries indicated by arrows in Fig. 9. As compared to B5 media sample, the F5 sample has less Chromium, which results in lower hardness, however this maybe elaborated as a result of tempering at higher tem-
perature or for tempering for longer time. As explained in the above general notes, Chromium lowers eutectoid carbon, therefore B5 media can be considered as eutectoid steel with no primary ferrite but F5 has a pre-eutectic structure. During hardening, austenitizing temperature has probably resulted in precipitation of pro-eutectoid ferrite at grain boundaries.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
4
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
Fig. 6. D5 media (100) is composed of tempered martensite, retained austenite and inclusion.
Fig. 9. F5 media (500) shows tempered martensite and discontinued ferrite at the boundaries - see arrows.
Fig. 7. Arrows refer to the grain boundaries (100).
Fig. 10. The microstructure of F5 media (50) shows the inclusions content.
inclusions or stringers was noticed. Such inclusions and stringers can further contribute to lower toughness. H5 media: Microstructure of H5 media sample is composed of tempered lath martensite and precipitated ferrite at primary austenite grain boundaries. This is a hypo-eutectoid steel in which uncontrolled austenitizing treatment can result in precipitation of ferrite at grain boundaries (Fig. 12). Furthermore, generally as tem-
Fig. 8. Microstructure of E5 media (200).
G5 media: This steel microstructure consists of lath martensite, retained austenite (brighter area) and ‘‘possible” inclusion islands. No ferrite was observed in the microstructure (Fig. 11). The martensite seems to have a classical morphology which implies insufficient tempering. This can possibly contribute to lower impact toughness. Further, the presence of continuous, elongated
Fig. 11. The microstructure of G5 media (500) shows the inclusions content.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
Fig. 12. Microstructure of H5 media (200). The arrows refer to the precipitated ferrite at primary austenite grain boundaries.
Fig. 14. Microstructure of J5 media (50).
Fig. 13. Microstructure of I5 media (200).
Fig. 15. Microstructure of J5 media (200).
pering process performs at higher temperature, the resulting hardness can be lower and associated impact toughness can be higher. For this media sample, it seems that it has been more tempered in comparison to the some of the other media samples tested (G5, F5, B5) as the martensite plates cannot be seen in the microstructure. This might have overruled the detrimental effect of ferrite at grain boundaries on impact toughness. I5 media: Microstructure is composed of tempered lath martensite and retained austenite. No ferrite was observed at primary austenite grain boundaries (Fig. 13). J5 media: The microstructure of J5 media sample seems to be a cast iron (Figs. 14 and 15). The microstructure seems to consist of primary hard, brittle cementite network at grain boundaries and tempered martensite in the middle of grains. Obviously, the brittle network contributed in lowering the toughness of this material. A summary of the microstructure results can be found in Table 4.
5
3.2. Impact toughness The results of Charpy tests are presented in the form of impact energies or the toughness of three locations within the steel media (Fig. 16, Table 5). The three locations as indicated in Fig. 1. The photos of the halves of broken CVN specimens in Fig. 17 show that a V-shaped ‘‘chevron” is clearly present in almost all
Table 4 Microstructure summary.
Tempered martensite Untampered martensite Retained austenite Ferrite present Cementite
A5
B5
X
X
C5 X X
D5
E5
F5
G5
H5
I5
J5
X
X
X
X
X
X
X
X
X X
X
X
X
X X
Fig. 16. The impact energies of the Charpy tests according to the specimen location within the steel media.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
6
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
all Charpy test specimen. According to the brittle fracture criteria, all the tested specimen fracture in a brittle manner because the impact energies (toughness) of all specimens are below the brittle fracture impact energy of 20 J. It is important to note that the Charpy impact energy is very sensitive to the carbon content (% C) in steels and carbon content above approximately 0.2% would result in a brittle fracture at room temperature (Rinebolt and Harris, 1951). It is interesting to note that for the two high carbon/high chromium content media, the fracture toughness is the lowest.
fracture surfaces of the tested Charpy specimens. This V-shaped chevron marks present on the fracture surface are well known characteristic of brittle fracture (Hertzberg, 1989). The impact energies of all the steel media (excluding A5) are independent of location. These results reveal that the toughness is the same all over the steel media. In the case of A5 toughness, the results showed that the toughness is higher close to the media center and decreases when getting closer to the surface. In general, the study of the fracture surface of the halves of the broken Charpy V-Notch impact specimens showed clearly that no ductile fracture evidence was found on the fracture surfaces of
3.3. Abrasive wear Table 5 Average impact toughness results.
Charpy [J]
A5
B5
C5
D5
E5
F5
G5
H5
I5
J5
8.29
4.07
1.53
4.09
5.56
3.62
2.6
4.54
4.68
1.58
Abrasive wear was determined using a steel wheel abrasion test (SWAT) as described in the ASTM-G65 Standard (2015). Ottawa foundry sand was used as the abrasive medium in all SWAT tests to provide similar testing conditions for better comparison of the
A5
B5
C5
D5
E5
F5
G5
H5
I5
J5 Fig. 17. Halves of broken Charpy V-Notch impact specimens.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
7
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
abrasion wear characterization of all materials under study. The range of applied forces on the wheel in the tests was between 100.86 N to 504.3 N and the abrasion wheel rotation speed tested was 155 rpm. The abrasion wheel speed variations were determined to be negligible under the different load conditions met in the abrasion test. These results can be found in Figs. 18 and 19. As is typically the case with SWAT dry abrasion results (Radziszewski, 2009), wear increased with the increase of the applied force (Fig. 19) until some maximum at which point the wear tended to be independent of the applied forces. The applied force at which this maximum occurs tends to be around 200 N although as illustrated in Fig. 18 exceptionally the maximum wear rate can also be found at higher applied forces. On the other hand, the wear per unit energy results illustrated in Fig. 19 is typical for dry testing of steel media with the SWAT (Radziszewski, 2009). It tends to start high and decreases to some constant value as is typical of this test. Table 6 illustrates the tabulated average abrasion mass loss values at 200 N applied force. 3.4. Hardness profile Hardness tests for all the media included in this study were conducted using Rockwell Hardness type C test to draw the hardness profile from the center to the surface of all the media selected for this study. The hardness profile was achieved by conducting series tests along the center line of the steel slice used to machine Charpy test specimens (Fig. 1). Fig. 20 presents the HRC hardness
profile of 125 mm media while Table 7 provides the average hardness of each media tested. Despite that at the surface all media have approximately the same hardness, there appears to be 2 general populations of steels. Population #1 defined by media D5, E5, F5, H5 show a hardness increase from center to surface and a lower average hardness. On the other hand, population #2 defined by media A5, B5, C5, G5, I5, and J5 which shows little variation from center to surface. In general, the population #1 behaviour indicates a reduced hardenability (i.e. ability of the steel to form martensite across the section of the steel media by quenching). Hardenability is a result mainly of the Cr addition, although C can be influential as well. However, it is notable that the Cr levels of population #2 are all higher than those of population #1. In population #1, the highest Cr is 0.68 in H5 while in population #2, the lowest Cr is 0.75 in G5. Note that the lowest hardenability, as indicated by the lowest center hardness, is E5, which has the next to the lowest Cr level (0.63). The lowest Cr level is in F5 (0.33) but, within population #1; F5 has the highest C level, which may be offsetting the low Cr level. Based on the chemical composition it may be expected that the media with the highest C and Cr content (C5) has the potential to have the highest hardness and abrasive resistance values. However, as mentioned earlier, it does not belong to the same compositional family as the other steels, and the properties cannot really be compared to the other steels. However, C5 does not have the highest hardness. 4. Analysis The following analysis will explore possible relationships between: (i) Hardness and abrasion. (ii) Hardness and impact toughness.
Fig. 18. Dry wear versus applied force.
Fig. 20. Hardness profile for all the steel media.
Table 7 Average hardness results.
HRC
Fig. 19. Dry wear per unit energy versus energy.
A5
B5
C5
D5
E5
F5
G5
H5
I5
J5
55
57
55
50
40
47
55
45
55
55
Table 6 Abrasion wear loss at 200 N load.
Mass loss [mg/s]
A5
B5
C5
D5
E5
F5
G5
H5
I5
J5
8.59
6.81
3.86
3.76
7.49
4.85
6.12
6.26
10.10
7.56
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
8
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
(iii) (iv) (v) (vi)
Abrasion and impact toughness. Hardness and Cr, Mo, Ni content. Impact toughness and C content. Chemical composition and performance.
(i) Hardness and abrasion According to Archard (1953), abrasion wear is inversely proportional to hardness. As a result, one would expect that abrasion wear would decrease with increasing hardness. However, plotting the abrasion wear results as found in
Population #2 Population #1
Fig. 21. Abrasion versus average hardness.
Population #2 Population #1
Fig. 22. Charpy impact energy versus average hardness.
Table 6 against media average hardness as found in Table 7, indicates that this is only true for a portion of the samples tested as illustrated in Fig. 21. Essentially, the previously mentioned population #1 shows a clear relationship between increasing average hardness and decreasing abrasive wear while the remaining population #2 indicates no such relationship. Based on these results and assuming that these grinding media would not break in impact, one could expect to see most of this grinding media will show good abrasive resistance over the diameter of the balls. However, for D5, E5, F5, and H5 media samples, one might expect abrasive wear to increase as the media wears down. The correlation between hardness and abrasive wear has already been investigated by a number researchers (Archard, 1953; Rabinowicz, 1983; Gates et al., 2008; Radziszewski, 2009). In abrasive wear, it is generally considered that the hardest material resists abrasion best (Sundström et al., 2001). However some research has drawn contradicting conclusions and shown that the correlation between hardness and abrasive wear is not so clear. This contradiction is due to the different operating conditions, configuration parameters and types of test used (Subramanian, 1992). Moreover, this contradiction can be attributed to the methods of hardening the material. Khrushchov and Babichev (1964) found that the type of hardening (heat treatment or work hardening) governs any improvement of abrasive wear resistance. (ii) Hardness and impact toughness It is commonly accepted that hardness is generally inversely proportional to impact toughness; an assertion which is supported by some literature (El Fawkhry et al., 2014). Essentially, increasing hardness increases brittleness of a material which reduces impact toughness. However, plotting the impact toughness as illustrated by the Charpy V notch test results found in Table 5 against media average hardness as found in Table 7, indicates that this is somewhat true for a portion of the samples tested as illustrated in Fig. 22 while for the remainder impact toughness seems to be independent of media hardness. (iii) Abrasive wear and impact toughness Generally, impact toughness should decrease with decreasing abrasion wear as in the work illustrated by Emamian (2012). In the present case, the results found in Fig. 23 indicate that this is in general true for the media samples found in population #1 and somewhat less apparent for the samples found in population #2.
Population #2
Population #1
Fig. 23. Abrasion wear vs Charpy impact energy.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
(iv) Hardness and Cr, Mo, Ni content From the literature (Bell, 2016), chemical additives such as chromium, molybdenum, nickel can be used to increase metal hardness. The results presented in Fig. 24 tend to reflect the expected impact of these additives on metal hardness. However, the correlations as illustrated by the R2 values are quite weak. (v) Impact toughness and C content It is understood that impact toughness tends to reduce as carbon content increases (Johnson and Storey, 2008). This tendency is confirmed by the results found for these 10 media samples as illustrated in Fig. 25. (vi) Chemical composition and performance Examining how one chemical component at a time affects media toughness, hardness or abrasion values does not provide an understanding as to how the different steel media components work together to affect media performance. In order to underline this point, consider the noise in the data presented in Figs. 26–28 where toughness, hardness and abrasion is plotted against the chemical composition of the different media. These results indicate either that there is no explicit relationship between media performance and chemical composition or there is one but it is not a function of one component over another but rather a function of a number of components and their interactions and interdependencies. A review of the literature indicates that the use of chemical composition to predict material behaviour has been investigated in a few works (Steven and Haynes, 1956; Andrews, 1965; Kung and Rayment, 1982; Vanderschueren et al., 1990; Trzaska, 2013). In these works, the authors demonstrate, how an empirical combination of chemical composition can be used to predict some steel characteristics such as hardness. A similar model development for grinding media perfor-
9
Fig. 26. Toughness versus chemical composition.
Fig. 27. Average hardness versus chemical composition.
Fig. 24. Abrasion wear vs Charpy impact energy.
Fig. 28. Abrasion wear versus chemical composition.
mance could be possible by leveraging the observation that there seems to be a strong relationship between carbon content and impact toughness (Fig. 25) described in power function format such as:
y ¼ Axb
Fig. 25. Charpy impact energy vs carbon content.
ð1Þ
In the case of impact toughness, ‘‘y” would be toughness in joules, ‘‘x” would be the weight percent of carbon while A and b would be constant determined from the trendline correlation (2.8341 and 0.933 respectively).
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
10
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
One could further propose that the carbon content (C) could be corrected by the content of the other chemical components (xi) present in a given grinding media sample. However, as not all of the different media samples use the same chemical components, the form of the correction factors would include a reference value (xmaxi) as defined:
n Y xmaxi xi ci x¼C : xmaxi i¼1
ð2Þ
In this proposed form of the carbon correction, zero chemical content of a particular component would reduce to ‘‘1” (xmax/ xmax = 1). The reference value (xmaxi) for each chemical component was approximated using data from the literature (Michael, 2006) as well as the maximum values found in Tables 2 and 3. These reference values can be found in Table 8. Using the reference values and exponents found in Table 8 along with the chemical compositions illustrated in Tables 2 and 3, it was possible to produce the correlations found in Figs. 29–31. Using the resulting model, it was possible to compare the model prediction with the measured media performance as illustrated in Fig. 32.
Fig. 30. Average hardness versus corrected carbon.
Table 8 Reference values, constants and exponents used for carbon content correction.
A b
Charpy toughness
Hardness
Abrasion wear
1.0605 2.194
142.58 0.672
0.0454 -3.413
Chemical component
Xmax i
ci
ci
ci
Al C Co Cr Cu Mn Mo Nb Ni P Si S Sn Ta Ti V
1.3 – 0.1 30 0.5 14 0.11 0.1 3.5 0.05 2 0.15 0.1 0.1 0.1 0.1
2.4949 – 0.1000 0.6958 0.0500 0.7015 0.0100 0.0100 2.6908 0.0998 0.0300 0.2993 0.0200 0.0500 1.8962 0.2001
0.9005 – 0.6001 1.0061 0.1996 15.4788 0.0400 0.1802 7.5502 0.0500 0.1496 0.0200 0.2001 0.0300 2.0030 0.8008
2.9986 – 10.0098 4.9578 0.0100 0.5001 0.1999 1.0015 4.9940 0.1000 0.0100 0.0300 0.1000 1.0018 0.3000 2.8026
Fig. 31. Abrasion wear versus corrected carbon.
Fig. 32. Comparing measured with predicted media performance.
5. Discussion
Fig. 29. Charpy impact toughness versus corrected carbon.
Notwithstanding these interesting results, it is important to take a moment to consider the use of hardness, abrasion and toughness as adequate metrics to predict the wear performance of media in an industrial context. With respect to hardness, it is a metric frequently used to compare one media type to another. However, the results presented in Fig. 21, indicate that care should be taken as two different media may have different composition and hardness yet provide similar wear rates at least at the lab scale. With respect to wear by abrasion, the SWAT is one of many abrasion type tests that could be used to quantify abrasive wear. In addition, the SWAT has also been used to quantify abrasive wear in steel media prediction (Chenje et al., 2009). As a result, it is an
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
appropriate test to use to determine the abrasive wear characteristics of steel media. On the other hand, it would be interesting to see the correlation for abrasion as described by mass worn per unit of energy. Furthermore, it should be noted that abrasion wear characteristics do not include corrosion which in many cases has a very important effect on wear in an industrial system. With respect to impact toughness, the use of the Charpy test may be controversial in the context where all samples are brittle. However, the results and associated analysis indicates that maybe there is value in the results a Charpy V-notch test produces. Further analysis would be required especially in comparison with drop ball test results coupled to results from industrial applications. It is important to underline that Eqs. (1) and (2) coupled with the constants and exponents in Table 8 are empirical relationships and really only valid for the media used and the conditions under which the tests were executed. Furthermore, these relationships do not include factors that describe heat treatment which may or may not explain some of the scatter around the trend lines. On the other hand, these equations and particularly the correlation with media performance metrics indicate that there may indeed be a relationship between initial chemical composition and the subsequent media performance. 6. Conclusions The focus of this paper was to examine 125 mm diameter grinding media and investigate the differences in microstructure, hardness, abrasion and impact toughness of steel media from eight different manufacturing sources. Tests results were produced for chemical composition, microstructure, impact toughness, abrasive wear and hardness along with some explanations related to the results. The main observations based on the test results and associated analysis indicate that: (i) Abrasive wear tends to be a function of media hardness for media with an chromium content of less than 0.7% and independent of hardness for media with chromium content greater than 0.75%. (ii) Charpy impact toughness results are strongly correlated to carbon content that can be corrected with the chemical composition of the media. (iii) Hardness and abrasion wear are somewhat correlated to the chemical composition. It is important to underline that heat treatment was not considered in this analysis. Furthermore, it is well understood that these empirical models are only valid for the media used and the testing conditions used. And despite the shortcomings associated with empiricism, the media performance models do indicate how such performance is a function of the interaction and interdependencies between chemical components. Lastly, it is important to underline that lab scale abrasion testing and impact toughness testing provide some insight into the wear performance of grinding media. However, corrosion was not included and no effort was made to correlate the results presented here with industrial results. Acknowledgements This project was initiated and completed between the years 2007 and 2012. It was supported by an undisclosed company interested in understanding better the factors affecting steel media wear. Consequently, the authors would like to thank this company for the support of this work.
11
Furthermore, the authors would like to thank both Metso and Hatch for allowing us to publish and present this paper. References Andrews, K.W., 1965. Empirical formula for the calculation of some transformation temperatures. J. Iron Steel Inst. 203, 721–727. Archard, J.F., 1953. Contact and rubbing of flat surface. J. Appl. Phys. 24 (8), 981– 988. ASM, 1985. Metals Handbook: Metallography and Microstructures, vol. 9. Matal Park, Ohio. ASTM B611, 2013. Standard Test Method for Determining the High Stress Abrasion Resistance of Hard Materials. ASTM Int’l, West Conshohocken, PA. ASTM E1019, 2011. Standard Test Methods for Determination of Carbon, Sulfur, Nitrogen, and Oxygen in Steel and in Iron, Nickel, and Cobalt Alloys. ASTM Int’l, West Conshohocken, PA. ASTM E1479, 2011. Standard Practice for Describing and Specifying InductivelyCoupled Plasma Atomic Emission Spectrometers. ASTM Int’l, West Conshohocken, PA. ASTM E23, 2002. Standard Test Method for Notched Bar Impact Testing of Metallic Materials. ASTM Int’l, West Conshohocken, PA. ASTM G65, 2015. Standard Test Method for Measuring Abrasion Using Dry Sand/ Rubber Wheel Apparatus. ASTM Int’l, West Conshohocken, PA. Ballantyne, G.R., Powell, M.S., 2014. Benchmarking comminution energy consumption for the processing of copper and gold ores. Miner. Eng. 65, 109–114. Bell, T., 2016. Common stell alloying agents, properties and effects (accessed July 18 2016). Benavente, H., 2007. Correlacion empirica para estimar consumos de medios demolienda, X Simposium de Molienda Moly-Cop. Bond, F.C., 1963. Metal Wear in Crushing and Grinding. Allis-Chalmers Publication 07P1701. Chenje, T., Radziszewski, P., Hewitt, D., 2009. Steel media wear: experimentation, simulation and validation. CMP Conf., 557–572 El Fawkhry, M.K., Fathy, A.M., Eissa, M.M., 2014. New energy saving technology for producing Hadfield steel to high gouging applications. Steel Res. Int. 85 (9999), 1–8. Emamian, A., 2012. A study on wear resistance, hardness and impact behaviour of carburised Fe-based powder metallurgy part for automotive applications. Mater. Sci. Appl. 3, 519–522. Gates, J., Dargusch, M.S., Walsh, J.J., Saad, J.R., 2008. Effect of abrasive mineral on alloy performance in the ball mill abrasion test. Wear 5–6, 865–870. Hertzberg, W., 1989. Deformation and Fracture Mechanics of Engineering Materials, third ed. John Wiley & Sons, New York. Johnson, A.A., Storey, R.J., 2008. The effect of carbon on the Charpy V-notch ductilebrittle transition curve. Metall. Anal. 28 (2), 1552-16. Khrushchov, M.M., Babichev, M.A., 1964. The effect of heat treatment and work hardening on the abrasive wear of some alloys steel, friction and wear in machinery. Am. Soc. Mech. Eng. 19, 1–15. Kung, C.Y., Rayment, J.J., 1982. An examination of the validity of existing empirical formulae for the calculation of Ms temperature. Metall. Trans. A Phys. Metall. Mater. Sci. 13A, 328–331. McIvor, R., 2015. Efficiency Metrics for Identifying and Remediating Plant Grinding Performance Issues, SAG2015, Vancouver, 12 pages. Michael, R.J., 2006. Chapter 14 Carbon and alloy steels. (accessed August 7 2016). Modin, H., Modin, S., 1973. Metallurgical Microscopy. Butterworth & Co Ltd., London, UK. Nadolski, S., Klein, B., Gong, D., Davaanyam, Z., Cooper, A., 2015. Development and application of an energy benchmarking model for mineral comminution, SAG2015, Vancouver, 15 pages. Rabinowicz, E., 1983. The Wear of Hard Surfaces by Soft Abrasive, Wear of Materials. ASME, pp. 12–18. Radziszewski, P., 2009. The steel wheel abrasion test (SWAT): a tool to study wear, friction and ore breakage in the mining industry. In: Int’l J. on Wear, Friction and Lub. 17th International Conference on Wear of Materials, 267, 1–4, 15 June 2009, 92–98. Rinebolt, J.A., Harris, W.J., 1951. Effect of alloying elements on notch toughnessof pearlitic steels. Trans. ASM 43, 1175–1214. Steven, W., Haynes, A.G., 1956. The temperature of formation of martensite and bainite in low-alloy steels. J. Iron Steel Inst. 183, 349–359. Subramanian, C., 1992. Some considerations towards the design of a wear resistant aluminum alloy. Wear 155, 193–205. Sundström, A., Rendón, J., Olsson, M., 2001. Wear behaviour of some low alloyed steels under combined impact/abrasion contact conditions. Wear 250 (1–12), 744–754. Trzaska, J., 2013. Calculation of steel hardness after continuous cooling. Arch. Mater. Sci. Eng. 61 (2), 87–92. Vanderschueren, D., Kestens, L., Van Houtte, P., Aernoudt, E., Dilewijns, J., Meers, U., 1990. Influence of transformation induced recrystallization on hot rolling textures of low carbon steel sheet. Mater. Sci. Technol. 6, 1247–1259. Vander Voort, G.F., 2007. Metallography, principles and practice. In: Material Science and Engineering Series. ASM Int’l, McGraw-Hill, New York, 752 pages.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
Contents lists available at ScienceDirect
Minerals Engineering journal homepage: www.elsevier.com/locate/mineng
Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness Amar Sabih a, Peter Radziszewski b,⇑, Ivan Mullany c a
Mechanical Engineering, McGill University, 817 Sherbrooke St. W., Montreal H3A 0C3, Canada Metso, 795 ave George-V, Lachine, Québec H8S 2R9, Canada c Hatch, 2800 Speakman Drive, Sheridan Science & Technology Park, Mississauga, Ontario L5K 2R7, Canada b
a r t i c l e
i n f o
Article history: Received 30 May 2016 Revised 16 August 2016 Accepted 19 August 2016 Available online xxxx Keywords: Grinding media Hardness Abrasion Toughness Wear
a b s t r a c t It is recognised that grinding media wear can represent up to 50% of the operating costs of a given tumbling mill. Over the years, a number of works have explored the development of different ways and means to both understand grinding media wear as well as model and predict it. The focus of the present work is to examine the differences in microstructure, hardness, abrasion and impact toughness of grinding media from eight different manufacturing sources. Results will be presented for 125 mm diameter media typically used for SAG mills. A discussion will address any issues highlighted by the results than may contribute to predictive wear model development as well as indicate possible directions for future research. Ó 2016 Published by Elsevier Ltd.
1. Introduction In mineral processing, the two main operating costs are related to energy consumption and wear. Energy consumption in comminution processes especially as it related to efficiency has been the focus of much research and application. Current efforts aim at developing guidelines and potentially standards for energy based grinding performance metrics (McIvor, 2015) and benchmarks (Ballantyne and Powell, 2014; Nadolski et al., 2015). On the other hand, with respect to wear and particularly grinding media wear, there are still a few different schools of thought (Bond, 1963; Benavente, 2007; Gates et al., 2008; Chenje et al., 2009) all of which converge on the general notion that wear in comminution processes is a function of three main components which are the energy involved in wear, the chemical and mechanical properties of the media as well as the chemical and mechanical properties of the ore or slurry. The focus of the present paper will be to examine the components related to the mechanical properties of the media. Specifically, the focus will be on 125 mm diameter grinding media and will investigate the differences in microstructure, impact toughness, abrasion and hardness of grinding media from eight different manufacturing sources. Following a presentation of the results, a
⇑ Corresponding author. E-mail address: [email protected] (P. Radziszewski).
discussion will explore any possible relationships as well as possible avenues for future research.
2. Sample preparation Ten different SAG mill 125 mm diameter media samples from eight different manufacturing sites was collected for this investigation. The different media sampled are randomly listed as illustrated in Table 1. With respect to chemical composition, it should be noted that the standards ASTM E1479 (2011) and ASTM E1019 (2011) were followed. The chemical analysis indicated the percent weight results for aluminium, arsenic, carbon, cobalt, chromium, copper, manganese, molybdenum, niobium, nickel, phosphorus, silicon, sulfur, tin, tantalum, titanium, vanadium, zirconium and tungsten. Table 2 provides the percent weight results for chemical components common to all media sampled, while Table 3 indicates that number of media samples that contained the chemical components not common to all samples tested along with their range. Subsequently, the media samples were then prepared in order to accomplish the different tests for microstructure, impact fracture toughness, abrasion testing and hardness. All samples were cut under wet conditions such that a number of samples were obtained without microstructure modification including three from different radial positions in the ball as illustrated in Fig. 1. In the case of impact toughness, samples were prepared according
http://dx.doi.org/10.1016/j.mineng.2016.08.014 0892-6875/Ó 2016 Published by Elsevier Ltd.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
2
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
Table 1 Selected grinding media with the corresponding codes (125 mm media).
Table 3 The chemical composition results common to all media samples (%wt).
Coded provider
Coded provider
A5 B5 C5 D5 E5
F5 G5 H5 I5 J5
to the requirements of the Charpy V-notch specifications defined by the ASTM E23 (2002). The hardness profile was achieved by conducting a series of tests along the center line of the 125 mm media slice from which Charpy test specimens were machined from. Samples for abrasion wear test did not have any special preparation other than meeting the required sample size and shape for use in steel wheel abrasion test as described by the ASTM G65 (2015) and ASTM B611 (2013) standards. Chemical analysis was completed on the fractured halves produced by the Charpy impact test. Surfaces were prepared in the typical fashion required for metallographic analysis. This includes sample mounting in Bakelite, mechanical grinding, abrasion and polishing, etching followed by microscopic analysis (Modin and Modin, 1973; ASM, 1985; Vander Voort, 2007). 3. Results The results are presented in the following order: microstructure, impact toughness, abrasive wear and hardness. 3.1. Microstructure The microstructures of all grinding media investigated in this work are shown in Figs. 2–15. All appear to be quenched and tempered martensite, with two exceptions: (i) C5, the high C and high Cr alloys, which reveals a microstructure that is not readily identifiable (Fig. 15); (ii) J5 sample reveals the cast structure of dendrites with what appears to be a matrix of ledeburite i.e. the eutectic formed at the final stage of liquid solidification, which should be a mixture of white cementite and transformed austenite (Figs. 14 and 15). Apart from the white cementite matrix, the darker areas could be tempered martensite. Higher magnifications are required to clarify this. At optical microscopy magnifications, it is extremely difficult to differentiate between the characteristics of quenched and tempered martensite. Only the martensite laths can really be resolved, but it is still difficult to differentiate between martensite morphologies. Carbides and retained austenite cannot be resolved. These characteristics need to be defined by electron microscopy and X-ray analysis, if necessary. A literature review is required to ascertain whether the degree of tempering, or changes in martensite morphology, affect wear. Certainly the degree of tempering affects toughness, and is well documented. There appears to be inclusions (black particles) in all the grinding media samples.
% Aluminium % Arsenic % Cobalt %Molybdenum % Niobium % Tin % Tantalum % Titanium % Vanadium % Zirconium % Tungsten
No. of samples
Range (wt.%)
8 0 1 6 1 3 1 2 1 0 0
0.02 <0.01 0.01 0.01–0.06 0.08 0.01–0.08 0.08 0.01 0.03 <0.01 <0.01
These need to be quantified at some point since these may affect the toughness of the grinding media. A5 media: The microstructure of A5 media consists of tempered martensite but retained austenite was difficult to find in the microstructure (Fig. 2). Although this is a hypo-eutectoid structure, no pro-eutectoid ferrite was observed at grain boundaries. This may explain the high toughness for this media sample as compared to other samples tested. As the hardness is comparable with other, this could be as a result of good heat treatment or performing forging treatment. B5 media: The microstructure is composed of lath martensite and small islands (5–10 lm) inclusions of irregular shape. With regards to the ball application and measured hardness, the martensite seems to be slightly tempered in order to increase toughness and decrease retained austenite. The brighter area in Figs. 3 and 4 could be tempered retained austenite. C5 media: Based on the composition and in terms of the carbon content it is a hyper-eutectoid structure. The microstructure shows chromium carbides in a matrix of untempered martensite, which should lower dramatically the toughness (Fig. 5). There should be relatively high amount of retained austenite (not resolved with Nital etchant). D5 media: This grinding media sample is composed of tempered martensite, retained austenite and inclusions. The martensite looks to be a mixture of lath and needles due to the carbon content of this structure. The lamellar style of microstructure implies deformed structure due to forging treatment (Fig. 6). The inclusions were also elongated as a proof of forging. Precipitated ferrite was observed in this steel due to its pro-eutectoid characteristics. The relatively good toughness can be attributed to the forging treatment applied on this media sample despite of the presence of detrimental ferrite at primary austenite grain boundaries (Fig. 7). E5 media: The composition is almost the same as D5 media (Fig. 8), however, this sample showed lower hardness and higher toughness. Ferrite at the grain boundaries was observed in limited areas as compared to D5. The presence of this ferrite may contribute to lowering the impact toughness as compared to the D5 sample.
Table 2 The chemical composition results common to all media samples (wt.%).
% % % % % % % %
Carbon Chromium Copper Manganese Nickel Phosphorus Silicon Sulfur
A5
B5
C5
D5
E5
F5
G5
H5
I5
J5
0.551 0.78 0.10 0.69 0.05 0.012 1.75 0.003
0.623 0.98 0.15 0.72 0.08 0.011 0.73 0.025
2.10 18.61 0.05 0.31 0.11 0.020 0.38 0.041
0.577 0.64 0.05 0.97 0.03 0.016 0.42 0.010
0.524 0.63 0.03 0.96 0.03 0.015 0.43 0.004
0.669 0.33 0.25 0.85 0.14 0.012 0.19 0.028
0.764 0.75 0.21 0.85 0.14 0.016 0.16 0.027
0.572 0.68 0.02 0.99 0.02 0.015 0.42 0.005
0.608 0.96 0.15 0.72 0.13 0.013 0.77 0.021
1.97 12.68 0.06 1.04 0.14 0.036 0.72 0.031
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
3
Fig. 1. Schematics drawings of a slice of the 125 mm media showing the locations where Charpy test specimens were cut from using EDM wire cutting and subsequent grinding.
Fig. 4. The microstructure of B5 media (50). Fig. 2. Microstructure of A5 media (1000).
Fig. 5. Microstructure of C5 media (500). Fig. 3. B5 media (1000), the brighter area could be tempered retained austenite.
F5 media: The microstructure for F5 steel shows tempered martensite and discontinued ferrite at the boundaries indicated by arrows in Fig. 9. As compared to B5 media sample, the F5 sample has less Chromium, which results in lower hardness, however this maybe elaborated as a result of tempering at higher tem-
perature or for tempering for longer time. As explained in the above general notes, Chromium lowers eutectoid carbon, therefore B5 media can be considered as eutectoid steel with no primary ferrite but F5 has a pre-eutectic structure. During hardening, austenitizing temperature has probably resulted in precipitation of pro-eutectoid ferrite at grain boundaries.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
4
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
Fig. 6. D5 media (100) is composed of tempered martensite, retained austenite and inclusion.
Fig. 9. F5 media (500) shows tempered martensite and discontinued ferrite at the boundaries - see arrows.
Fig. 7. Arrows refer to the grain boundaries (100).
Fig. 10. The microstructure of F5 media (50) shows the inclusions content.
inclusions or stringers was noticed. Such inclusions and stringers can further contribute to lower toughness. H5 media: Microstructure of H5 media sample is composed of tempered lath martensite and precipitated ferrite at primary austenite grain boundaries. This is a hypo-eutectoid steel in which uncontrolled austenitizing treatment can result in precipitation of ferrite at grain boundaries (Fig. 12). Furthermore, generally as tem-
Fig. 8. Microstructure of E5 media (200).
G5 media: This steel microstructure consists of lath martensite, retained austenite (brighter area) and ‘‘possible” inclusion islands. No ferrite was observed in the microstructure (Fig. 11). The martensite seems to have a classical morphology which implies insufficient tempering. This can possibly contribute to lower impact toughness. Further, the presence of continuous, elongated
Fig. 11. The microstructure of G5 media (500) shows the inclusions content.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
Fig. 12. Microstructure of H5 media (200). The arrows refer to the precipitated ferrite at primary austenite grain boundaries.
Fig. 14. Microstructure of J5 media (50).
Fig. 13. Microstructure of I5 media (200).
Fig. 15. Microstructure of J5 media (200).
pering process performs at higher temperature, the resulting hardness can be lower and associated impact toughness can be higher. For this media sample, it seems that it has been more tempered in comparison to the some of the other media samples tested (G5, F5, B5) as the martensite plates cannot be seen in the microstructure. This might have overruled the detrimental effect of ferrite at grain boundaries on impact toughness. I5 media: Microstructure is composed of tempered lath martensite and retained austenite. No ferrite was observed at primary austenite grain boundaries (Fig. 13). J5 media: The microstructure of J5 media sample seems to be a cast iron (Figs. 14 and 15). The microstructure seems to consist of primary hard, brittle cementite network at grain boundaries and tempered martensite in the middle of grains. Obviously, the brittle network contributed in lowering the toughness of this material. A summary of the microstructure results can be found in Table 4.
5
3.2. Impact toughness The results of Charpy tests are presented in the form of impact energies or the toughness of three locations within the steel media (Fig. 16, Table 5). The three locations as indicated in Fig. 1. The photos of the halves of broken CVN specimens in Fig. 17 show that a V-shaped ‘‘chevron” is clearly present in almost all
Table 4 Microstructure summary.
Tempered martensite Untampered martensite Retained austenite Ferrite present Cementite
A5
B5
X
X
C5 X X
D5
E5
F5
G5
H5
I5
J5
X
X
X
X
X
X
X
X
X X
X
X
X
X X
Fig. 16. The impact energies of the Charpy tests according to the specimen location within the steel media.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
6
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
all Charpy test specimen. According to the brittle fracture criteria, all the tested specimen fracture in a brittle manner because the impact energies (toughness) of all specimens are below the brittle fracture impact energy of 20 J. It is important to note that the Charpy impact energy is very sensitive to the carbon content (% C) in steels and carbon content above approximately 0.2% would result in a brittle fracture at room temperature (Rinebolt and Harris, 1951). It is interesting to note that for the two high carbon/high chromium content media, the fracture toughness is the lowest.
fracture surfaces of the tested Charpy specimens. This V-shaped chevron marks present on the fracture surface are well known characteristic of brittle fracture (Hertzberg, 1989). The impact energies of all the steel media (excluding A5) are independent of location. These results reveal that the toughness is the same all over the steel media. In the case of A5 toughness, the results showed that the toughness is higher close to the media center and decreases when getting closer to the surface. In general, the study of the fracture surface of the halves of the broken Charpy V-Notch impact specimens showed clearly that no ductile fracture evidence was found on the fracture surfaces of
3.3. Abrasive wear Table 5 Average impact toughness results.
Charpy [J]
A5
B5
C5
D5
E5
F5
G5
H5
I5
J5
8.29
4.07
1.53
4.09
5.56
3.62
2.6
4.54
4.68
1.58
Abrasive wear was determined using a steel wheel abrasion test (SWAT) as described in the ASTM-G65 Standard (2015). Ottawa foundry sand was used as the abrasive medium in all SWAT tests to provide similar testing conditions for better comparison of the
A5
B5
C5
D5
E5
F5
G5
H5
I5
J5 Fig. 17. Halves of broken Charpy V-Notch impact specimens.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
7
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
abrasion wear characterization of all materials under study. The range of applied forces on the wheel in the tests was between 100.86 N to 504.3 N and the abrasion wheel rotation speed tested was 155 rpm. The abrasion wheel speed variations were determined to be negligible under the different load conditions met in the abrasion test. These results can be found in Figs. 18 and 19. As is typically the case with SWAT dry abrasion results (Radziszewski, 2009), wear increased with the increase of the applied force (Fig. 19) until some maximum at which point the wear tended to be independent of the applied forces. The applied force at which this maximum occurs tends to be around 200 N although as illustrated in Fig. 18 exceptionally the maximum wear rate can also be found at higher applied forces. On the other hand, the wear per unit energy results illustrated in Fig. 19 is typical for dry testing of steel media with the SWAT (Radziszewski, 2009). It tends to start high and decreases to some constant value as is typical of this test. Table 6 illustrates the tabulated average abrasion mass loss values at 200 N applied force. 3.4. Hardness profile Hardness tests for all the media included in this study were conducted using Rockwell Hardness type C test to draw the hardness profile from the center to the surface of all the media selected for this study. The hardness profile was achieved by conducting series tests along the center line of the steel slice used to machine Charpy test specimens (Fig. 1). Fig. 20 presents the HRC hardness
profile of 125 mm media while Table 7 provides the average hardness of each media tested. Despite that at the surface all media have approximately the same hardness, there appears to be 2 general populations of steels. Population #1 defined by media D5, E5, F5, H5 show a hardness increase from center to surface and a lower average hardness. On the other hand, population #2 defined by media A5, B5, C5, G5, I5, and J5 which shows little variation from center to surface. In general, the population #1 behaviour indicates a reduced hardenability (i.e. ability of the steel to form martensite across the section of the steel media by quenching). Hardenability is a result mainly of the Cr addition, although C can be influential as well. However, it is notable that the Cr levels of population #2 are all higher than those of population #1. In population #1, the highest Cr is 0.68 in H5 while in population #2, the lowest Cr is 0.75 in G5. Note that the lowest hardenability, as indicated by the lowest center hardness, is E5, which has the next to the lowest Cr level (0.63). The lowest Cr level is in F5 (0.33) but, within population #1; F5 has the highest C level, which may be offsetting the low Cr level. Based on the chemical composition it may be expected that the media with the highest C and Cr content (C5) has the potential to have the highest hardness and abrasive resistance values. However, as mentioned earlier, it does not belong to the same compositional family as the other steels, and the properties cannot really be compared to the other steels. However, C5 does not have the highest hardness. 4. Analysis The following analysis will explore possible relationships between: (i) Hardness and abrasion. (ii) Hardness and impact toughness.
Fig. 18. Dry wear versus applied force.
Fig. 20. Hardness profile for all the steel media.
Table 7 Average hardness results.
HRC
Fig. 19. Dry wear per unit energy versus energy.
A5
B5
C5
D5
E5
F5
G5
H5
I5
J5
55
57
55
50
40
47
55
45
55
55
Table 6 Abrasion wear loss at 200 N load.
Mass loss [mg/s]
A5
B5
C5
D5
E5
F5
G5
H5
I5
J5
8.59
6.81
3.86
3.76
7.49
4.85
6.12
6.26
10.10
7.56
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
8
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
(iii) (iv) (v) (vi)
Abrasion and impact toughness. Hardness and Cr, Mo, Ni content. Impact toughness and C content. Chemical composition and performance.
(i) Hardness and abrasion According to Archard (1953), abrasion wear is inversely proportional to hardness. As a result, one would expect that abrasion wear would decrease with increasing hardness. However, plotting the abrasion wear results as found in
Population #2 Population #1
Fig. 21. Abrasion versus average hardness.
Population #2 Population #1
Fig. 22. Charpy impact energy versus average hardness.
Table 6 against media average hardness as found in Table 7, indicates that this is only true for a portion of the samples tested as illustrated in Fig. 21. Essentially, the previously mentioned population #1 shows a clear relationship between increasing average hardness and decreasing abrasive wear while the remaining population #2 indicates no such relationship. Based on these results and assuming that these grinding media would not break in impact, one could expect to see most of this grinding media will show good abrasive resistance over the diameter of the balls. However, for D5, E5, F5, and H5 media samples, one might expect abrasive wear to increase as the media wears down. The correlation between hardness and abrasive wear has already been investigated by a number researchers (Archard, 1953; Rabinowicz, 1983; Gates et al., 2008; Radziszewski, 2009). In abrasive wear, it is generally considered that the hardest material resists abrasion best (Sundström et al., 2001). However some research has drawn contradicting conclusions and shown that the correlation between hardness and abrasive wear is not so clear. This contradiction is due to the different operating conditions, configuration parameters and types of test used (Subramanian, 1992). Moreover, this contradiction can be attributed to the methods of hardening the material. Khrushchov and Babichev (1964) found that the type of hardening (heat treatment or work hardening) governs any improvement of abrasive wear resistance. (ii) Hardness and impact toughness It is commonly accepted that hardness is generally inversely proportional to impact toughness; an assertion which is supported by some literature (El Fawkhry et al., 2014). Essentially, increasing hardness increases brittleness of a material which reduces impact toughness. However, plotting the impact toughness as illustrated by the Charpy V notch test results found in Table 5 against media average hardness as found in Table 7, indicates that this is somewhat true for a portion of the samples tested as illustrated in Fig. 22 while for the remainder impact toughness seems to be independent of media hardness. (iii) Abrasive wear and impact toughness Generally, impact toughness should decrease with decreasing abrasion wear as in the work illustrated by Emamian (2012). In the present case, the results found in Fig. 23 indicate that this is in general true for the media samples found in population #1 and somewhat less apparent for the samples found in population #2.
Population #2
Population #1
Fig. 23. Abrasion wear vs Charpy impact energy.
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
(iv) Hardness and Cr, Mo, Ni content From the literature (Bell, 2016), chemical additives such as chromium, molybdenum, nickel can be used to increase metal hardness. The results presented in Fig. 24 tend to reflect the expected impact of these additives on metal hardness. However, the correlations as illustrated by the R2 values are quite weak. (v) Impact toughness and C content It is understood that impact toughness tends to reduce as carbon content increases (Johnson and Storey, 2008). This tendency is confirmed by the results found for these 10 media samples as illustrated in Fig. 25. (vi) Chemical composition and performance Examining how one chemical component at a time affects media toughness, hardness or abrasion values does not provide an understanding as to how the different steel media components work together to affect media performance. In order to underline this point, consider the noise in the data presented in Figs. 26–28 where toughness, hardness and abrasion is plotted against the chemical composition of the different media. These results indicate either that there is no explicit relationship between media performance and chemical composition or there is one but it is not a function of one component over another but rather a function of a number of components and their interactions and interdependencies. A review of the literature indicates that the use of chemical composition to predict material behaviour has been investigated in a few works (Steven and Haynes, 1956; Andrews, 1965; Kung and Rayment, 1982; Vanderschueren et al., 1990; Trzaska, 2013). In these works, the authors demonstrate, how an empirical combination of chemical composition can be used to predict some steel characteristics such as hardness. A similar model development for grinding media perfor-
9
Fig. 26. Toughness versus chemical composition.
Fig. 27. Average hardness versus chemical composition.
Fig. 24. Abrasion wear vs Charpy impact energy.
Fig. 28. Abrasion wear versus chemical composition.
mance could be possible by leveraging the observation that there seems to be a strong relationship between carbon content and impact toughness (Fig. 25) described in power function format such as:
y ¼ Axb
Fig. 25. Charpy impact energy vs carbon content.
ð1Þ
In the case of impact toughness, ‘‘y” would be toughness in joules, ‘‘x” would be the weight percent of carbon while A and b would be constant determined from the trendline correlation (2.8341 and 0.933 respectively).
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
10
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
One could further propose that the carbon content (C) could be corrected by the content of the other chemical components (xi) present in a given grinding media sample. However, as not all of the different media samples use the same chemical components, the form of the correction factors would include a reference value (xmaxi) as defined:
n Y xmaxi xi ci x¼C : xmaxi i¼1
ð2Þ
In this proposed form of the carbon correction, zero chemical content of a particular component would reduce to ‘‘1” (xmax/ xmax = 1). The reference value (xmaxi) for each chemical component was approximated using data from the literature (Michael, 2006) as well as the maximum values found in Tables 2 and 3. These reference values can be found in Table 8. Using the reference values and exponents found in Table 8 along with the chemical compositions illustrated in Tables 2 and 3, it was possible to produce the correlations found in Figs. 29–31. Using the resulting model, it was possible to compare the model prediction with the measured media performance as illustrated in Fig. 32.
Fig. 30. Average hardness versus corrected carbon.
Table 8 Reference values, constants and exponents used for carbon content correction.
A b
Charpy toughness
Hardness
Abrasion wear
1.0605 2.194
142.58 0.672
0.0454 -3.413
Chemical component
Xmax i
ci
ci
ci
Al C Co Cr Cu Mn Mo Nb Ni P Si S Sn Ta Ti V
1.3 – 0.1 30 0.5 14 0.11 0.1 3.5 0.05 2 0.15 0.1 0.1 0.1 0.1
2.4949 – 0.1000 0.6958 0.0500 0.7015 0.0100 0.0100 2.6908 0.0998 0.0300 0.2993 0.0200 0.0500 1.8962 0.2001
0.9005 – 0.6001 1.0061 0.1996 15.4788 0.0400 0.1802 7.5502 0.0500 0.1496 0.0200 0.2001 0.0300 2.0030 0.8008
2.9986 – 10.0098 4.9578 0.0100 0.5001 0.1999 1.0015 4.9940 0.1000 0.0100 0.0300 0.1000 1.0018 0.3000 2.8026
Fig. 31. Abrasion wear versus corrected carbon.
Fig. 32. Comparing measured with predicted media performance.
5. Discussion
Fig. 29. Charpy impact toughness versus corrected carbon.
Notwithstanding these interesting results, it is important to take a moment to consider the use of hardness, abrasion and toughness as adequate metrics to predict the wear performance of media in an industrial context. With respect to hardness, it is a metric frequently used to compare one media type to another. However, the results presented in Fig. 21, indicate that care should be taken as two different media may have different composition and hardness yet provide similar wear rates at least at the lab scale. With respect to wear by abrasion, the SWAT is one of many abrasion type tests that could be used to quantify abrasive wear. In addition, the SWAT has also been used to quantify abrasive wear in steel media prediction (Chenje et al., 2009). As a result, it is an
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014
A. Sabih et al. / Minerals Engineering xxx (2016) xxx–xxx
appropriate test to use to determine the abrasive wear characteristics of steel media. On the other hand, it would be interesting to see the correlation for abrasion as described by mass worn per unit of energy. Furthermore, it should be noted that abrasion wear characteristics do not include corrosion which in many cases has a very important effect on wear in an industrial system. With respect to impact toughness, the use of the Charpy test may be controversial in the context where all samples are brittle. However, the results and associated analysis indicates that maybe there is value in the results a Charpy V-notch test produces. Further analysis would be required especially in comparison with drop ball test results coupled to results from industrial applications. It is important to underline that Eqs. (1) and (2) coupled with the constants and exponents in Table 8 are empirical relationships and really only valid for the media used and the conditions under which the tests were executed. Furthermore, these relationships do not include factors that describe heat treatment which may or may not explain some of the scatter around the trend lines. On the other hand, these equations and particularly the correlation with media performance metrics indicate that there may indeed be a relationship between initial chemical composition and the subsequent media performance. 6. Conclusions The focus of this paper was to examine 125 mm diameter grinding media and investigate the differences in microstructure, hardness, abrasion and impact toughness of steel media from eight different manufacturing sources. Tests results were produced for chemical composition, microstructure, impact toughness, abrasive wear and hardness along with some explanations related to the results. The main observations based on the test results and associated analysis indicate that: (i) Abrasive wear tends to be a function of media hardness for media with an chromium content of less than 0.7% and independent of hardness for media with chromium content greater than 0.75%. (ii) Charpy impact toughness results are strongly correlated to carbon content that can be corrected with the chemical composition of the media. (iii) Hardness and abrasion wear are somewhat correlated to the chemical composition. It is important to underline that heat treatment was not considered in this analysis. Furthermore, it is well understood that these empirical models are only valid for the media used and the testing conditions used. And despite the shortcomings associated with empiricism, the media performance models do indicate how such performance is a function of the interaction and interdependencies between chemical components. Lastly, it is important to underline that lab scale abrasion testing and impact toughness testing provide some insight into the wear performance of grinding media. However, corrosion was not included and no effort was made to correlate the results presented here with industrial results. Acknowledgements This project was initiated and completed between the years 2007 and 2012. It was supported by an undisclosed company interested in understanding better the factors affecting steel media wear. Consequently, the authors would like to thank this company for the support of this work.
11
Furthermore, the authors would like to thank both Metso and Hatch for allowing us to publish and present this paper. References Andrews, K.W., 1965. Empirical formula for the calculation of some transformation temperatures. J. Iron Steel Inst. 203, 721–727. Archard, J.F., 1953. Contact and rubbing of flat surface. J. Appl. Phys. 24 (8), 981– 988. ASM, 1985. Metals Handbook: Metallography and Microstructures, vol. 9. Matal Park, Ohio. ASTM B611, 2013. Standard Test Method for Determining the High Stress Abrasion Resistance of Hard Materials. ASTM Int’l, West Conshohocken, PA. ASTM E1019, 2011. Standard Test Methods for Determination of Carbon, Sulfur, Nitrogen, and Oxygen in Steel and in Iron, Nickel, and Cobalt Alloys. ASTM Int’l, West Conshohocken, PA. ASTM E1479, 2011. Standard Practice for Describing and Specifying InductivelyCoupled Plasma Atomic Emission Spectrometers. ASTM Int’l, West Conshohocken, PA. ASTM E23, 2002. Standard Test Method for Notched Bar Impact Testing of Metallic Materials. ASTM Int’l, West Conshohocken, PA. ASTM G65, 2015. Standard Test Method for Measuring Abrasion Using Dry Sand/ Rubber Wheel Apparatus. ASTM Int’l, West Conshohocken, PA. Ballantyne, G.R., Powell, M.S., 2014. Benchmarking comminution energy consumption for the processing of copper and gold ores. Miner. Eng. 65, 109–114. Bell, T., 2016. Common stell alloying agents, properties and effects
Please cite this article in press as: Sabih, A., et al. Investigating grinding media differences in microstructure, hardness, abrasion and fracture toughness. Miner. Eng. (2016), http://dx.doi.org/10.1016/j.mineng.2016.08.014