Wear mechanism of iron-base diamond-impregnated tool composites

Wear 303 (2013) 533–540

Contents lists available at SciVerse ScienceDirect

Wear journal homepage: www.elsevier.com/locate/wear

Wear mechanism of iron-base diamond-impregnated tool composites Janusz Stefan Konstanty n, Dorota Tyrala 1 AGH—University of Science and Technology, Faculty of Metals Engineering and Industrial Computer Science, Department of Physical Metallurgy and Powder Metallurgy, Mickiewicz Avenue 30, 30-059 Krakow, Poland

art ic l e i nf o

a b s t r a c t

Article history: Received 8 September 2012 Received in revised form 3 April 2013 Accepted 15 April 2013 Available online 22 April 2013

The main objective of the present work was to determine the effect of abrasion induced martensitic transformation occurring in the matrix of diamond-impregnated tool composites on their wear behaviour under quasi-industrial conditions. Various iron-base and cobalt-base powder mixtures were consolidated to a virtually pore-free condition by hot pressing at 850–900 1C. The specimens were subsequently checked for density and tested for resistance to both 3-body and 2-body abrasion. A series of diamond-impregnated specimens (segments) was also produced and tested for wear rate on abrasive sandstone using a special testing rig. The statistical analysis of wear data showed increased resistance to abrasion of alloys containing unstable austenite which could transform to hard martensite under tribological straining. The wear rate of diamond-impregnated composites was mainly affected by the diamond concentration, whereas statistically significant contribution of the matrix resistance to 3-body abrasion to the wear rate of the diamond containing material was exclusively found in iron-base composites containing austenite. & 2013 Elsevier B.V. All rights reserved.

Keywords: Diamond Metal-matrix composite Cutting tools Abrasion-induced martensite Abrasion

1. Introduction Nowadays diamond blades and wires are commonly used for sawing natural stone, concrete and ceramics. The cutting section of the tool consists of synthetic diamond crystals embedded in a metal, or metallic, matrix by various powder metallurgy (PM) fabrication routes [1]. While sawing the rigid diamond grits pass over the machined surface wearing away its mineral constituents which abrade and erode the matrix to expose fresh diamond crystals which take over the cutting action from mechanically degraded and dislodged ones. In order to attain the economically best sawing conditions, an ideal balance between the tool life and cutting rate has to be achieved. The tougher and more difficult to cut the workpiece the finer and stronger the diamond type to be selected is a general rule while the matrix has to wear at a rate corresponding to the rate of diamond breakdown. An incorrect choice of the matrix characteristics or diamond type, size and concentration yields a tool that wears away excessively or refuses to cut altogether. In practice, there are always several mechanisms of tool wear operating in concert. The interactions between the diamonds, workpiece, workpiece debris, and matrix occur in a variety of forms depending on size and amount of the abrasive swarf, its n

Corresponding author. Tel.: +48 12 617 2627; fax: +48 12 617 3190. E-mail addresses: [email protected] (J.S. Konstanty), [email protected] (D. Tyrala). 1 Tel.: +48 12 617 2586. 0043-1648/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.wear.2013.04.016

shape, hardness and cleavage properties, diamond strength and loading conditions, matrix resistance to wear, etc. Therefore it is convenient to confine the scope of theoretical analysis to two extreme cases such as circular sawing of difficult to cut granite and frame sawing of abrasive sandstone. 1.1. Circular sawing of granite with segmented blades The most difficult to cut types of granite are wet sawn in many passes with a reciprocating movement of the saw blade, which alternately operates in the up-cutting and down-cutting modes, as shown schematically in Fig. 1. The depth of cut (a) ranges between 0.4 and 20 mm whereas the feed rate (vf) is chosen to comply with the saw blade characteristics, machine rigidity and power, and to avoid overloading of the steel core. Typically, cutting rates (avf) of between 100 and 300 cm2/min are employed. In circular sawing of granite, the blade rotates in a constant direction at peripheral speeds (vs) of between 25 and 35 m/s. This leads to the development of a matrix tail behind each working diamond crystal, as shown in Fig. 2, which acts as a support during cutting. In down-cutting the working diamond particle penetrates into the stone to full depth while coming in contact with it. Thereafter as the diamond tracks across the abraded surface layer it emerges progressively and finally loses contact with the stone. In contrast to down-cutting, in the up-cutting mode diamonds gradually increase the depth of penetration to achieve the maximum values

534

J.S. Konstanty, D. Tyrala / Wear 303 (2013) 533–540

Fig. 1. Kinematic diagram of diamond crystals that come into contact with stone in down-cutting (left) and up-cutting (right) modes.

Fig. 2. Schematic representation of the cutting zone in circular sawing.

while leaving the kerf. Consequently, the diamond breakdown is facilitated by downward rotation of the blade that, by its nature, creates additional, pulsing impact forces acting on crystals which enter the cutting zone. Their magnitude is proportional to the maximum chip thickness (hmax) approximated as [2] vf a hmax ≈ vs Cw

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 1 − aD D2

ð1Þ

where C—number of diamond crystals per unit surface and w—average width of diamond crystals.

The failure of each diamond crystal is governed in a complex manner by its shape, orientation and ability to withstand impact forces, the magnitude of mechanical and thermal loading, and the matrix retention properties. With irregular and friable grits the tool wear is mainly governed by chipping of the diamond crystals which decreases their height of protrusion thus reducing the space for swarf accumulation and removal (see clearance in Fig. 2), and consequently, creating harsh 3-body and 2-body abrasion conditions. On the other hand, if blocky and robust grits are used the sawing conditions must be carefully chosen in order to facilitate

their controlled micro-chipping. Otherwise, due to insufficient diamond penetration depths (hmax) the excessive friction leads to grit blunting as shown in Fig. 3 (steps I and II). This results in high forces oriented 80–851 normal to the stone surface [3], which being transmitted to the matrix may cause its plastic deformation (step III in Fig. 3). As each working grit is subjected to cyclic loading, applied at 3–35 Hz intervals depending on the blade diameter (D) and peripheral speed (vs), each time the yield strength of the matrix is exceeded the seat of the diamond crystal slightly opens and thereby the hold on the diamond is being gradually destroyed. Finally when pullout occurs the matrix tail is immediately removed by harsh 2-body abrasion (step IV in Fig. 3). The wear rate then decays over time as clearance increases, and other forms of wear, such as 3-body abrasion and erosion, take over. 1.2. Frame sawing of abrasive sandstone In the frame sawing operation, schematically shown in Fig. 4, the blade moves forward and backward alternatively at a slow sinusoidal speed with a maximum of around 2 m/s. Under such conditions, the removal of coarse and abrasive sandstone debris from the kerf is difficult. This creates severe wear conditions for the matrix and facilitates diamond pullout since forces act on diamond grits in alternate directions and build-up of the matrix tail becomes impossible. Graphic illustration of the cutting zone characteristic of frame sawing is presented in Fig. 5. In frame sawing the maximum chip thickness is attained while altering the movement direction and may be approximated as [2]   vf ls −2L3 hmax ≈ arc cos ð2Þ ns ls where ns—flywheel rotational speed, ls—length of stroke and L3—spacing of cutting edges.

J.S. Konstanty, D. Tyrala / Wear 303 (2013) 533–540

535

Fig. 3. Diamond pullout-controlled wear progression in circular sawing of granite.

1.3. Metallic matrix materials

Fig. 4. Kinematic diagram of frame sawing operation.

Cobalt has traditionally been used as the matrix in diamondimpregnated tools [4]. Fine cobalt powders can be cost-effectively densified to virtually pore-free condition by typical 3-min hold at 800–900 1C and under a pressure of 30–35 MPa. In as-sintered condition the cobalt matrix shows excellent diamond retention properties which make it particularly suitable for sawing granite. As its resistance to abrasive wear can be increased by admixing tungsten carbide powders prior to consolidation, Co–WC matrices are commonly used for abrasive applications, such as sawing sandstone, fresh concrete, asphalt, etc. Unfortunately both cobalt and tungsten have a history of very high and changeable price, therefore concerns about price stability and growing demand for cheaper matrix materials have focused the toolmakers' attention on other metallic powders, preferably iron-base, which would combine excellent hot pressing characteristics with subsequent field performance similar to cobalt and Co–WC materials.

2. Experimental procedure and results Fig. 5. Schematic representation of the cutting zone in frame sawing.

By contrast to circular sawing, the frame saw acts at a slow speed and hence the working diamonds experience low impact loads. Therefore, the matrix resistance to wear takes priority over the strength of diamond grits. From relation (2) it becomes evident that extremely harsh wear conditions are created on so-called ‘slow frames’, which are characterised by low flywheel rotational speed.

2.1. Materials A number of wear resistant alloys manufactured from relatively inexpensive iron-based powders by the powder metallurgy (PM) hot press route were chosen as potential Co–WC substitutes in the manufacture of diamond-impregnated segments. Basic characteristics of the starting powders are shown in Table 1. Eleven PM alloy compositions were prepared by mixing the starting powders for 30 min in a chaotic motion Turbula T2C

536

J.S. Konstanty, D. Tyrala / Wear 303 (2013) 533–540

Table 1 Chemical composition and particle size of starting powders. Powder grade (producer)

Fe CN (BASF) Fe EN (BASF) M 166 (H.C. Starck) Fe:Cu 85/15 (H.C. Starck) Next 300 (Eurotungstene) DS 250 (H.C. Starck) T110 (Vale) 25GR 85/15–325 (ECKA Granules) 25GR 80/20–325 (ECKA Granules) Co EF (Umicore)

Chemical composition (wt%)

Mean particle size (μm)

Fe

Ni

Cu

Sn

Co

N

C

Others

Bal Bal Bal Bal Bal – – – – –

– – – – – – Bal – – –

– – 16.5 15 3 – – Bal Bal –

– – 3 – – – – 15 20 –

– – – – 25 – – – – 499.8

0.01 0.7 – – – – – – – –

0.005 0.8 – – – 0.02 0.6 – – –

– – 0.6 Sm2O3 – – Bal WC – – – –

7.3 4.7 1.6 2.3 4 2.5 1.6 10.5 11 1.3

Table 2 Powder composition, densities and phases present in the experimental PM alloys. Alloy no.

Designation

Powder composition (wt%)

Density (g/cm3)

Structure (main phases)

1 2 3 4 5 6 7 8 9 10 11

FeCuSn FeCuSnWC20 FeCuSnWC30 FeCoWC FeCuWC FeCuNi FeCuCN FeCuNi5CN FeCuNi10CN FeCuSnNi10CN CoWC

100% M166 80% M166–20% DS 250 70% M166–30% DS 250 80% Next 300–20% DS 250 80% Fe:Cu 85/15–20% DS 250 75% Fe CN–15% 25GR 85/15–325–10% T110 85% Fe EN–15% 25GR 85/15–325 80% Fe EN–15% 25GR 85/15–325–5% T110 75% Fe EN–15% 25GR 85/15–325–10% T110 82% Fe EN–8% 25GR 80/20–325–10% T110 80% Co EF–20% DS 250

7.81 8.73 9.27 8.47 8.75 8.06 7.85 8.01 8.12 8.03 9.26

(αFe) (αFe) (αFe) (αFe) (αFe) (αFe) (αFe) (γFe)+(αFe) (γFe)+(αFe) (γFe)+(αFe) (αCo)+(εCo)

mixer. The premixed powders were then poured into a graphite mould and consolidated to near-full density by passing an electric current through the mould under uniaxial compressive stress of 35 MPa. In each case the powder was held for 3 min at either 850 or 900 1C (for WC containing compositions) and subsequently cooled down at a rate of ∼6 K/s from the hot pressing temperature down to 550 1C. The detailed powder compositions and as-sintered densities of specimens prepared for wear tests are given in Table 2. In addition to diamond-free materials, a batch of cylindrical (ϕ7  8 mm) diamond-impregnated specimens was also produced in a similar way for testing wear by means of a purpose-built test rig designed to simulate the real application conditions. These specimens contained 35/40 US mesh (420–500 μm) medium quality synthetic diamond at 20 concentration (5 vol%). Prior to further experiments all diamond-free and diamondimpregnated specimens were tested for density by means of the water displacement technique, in order to check the degree of densification and to enable wear rate measurements. 2.2. Phase analysis The experimental iron-base alloys were designed to have either (αFe) or (γFe)+(αFe) matrix in the as-sintered condition. The phase stability was checked in order to assess the effect of strain-induced martensitic transformation of austenite on the alloy's resistance to abrasion. To do this, specimens representative of each alloy were examined by means of X-ray diffraction. Metallographic sections were first prepared by cutting out the central part of a specimen using an alumina cut-off wheel and mounting it in Bakelite. The resulting sections were then wet ground on #220 SiC paper and successively polished on cloths impregnated with 9, 6 and 1 μm diamond compound. Both grinding and polishing produce a plastically deformed layer. Its depth depends on sharpness and size of the abrasive.

Table 3 Phase compositions of the iron-base alloys. Alloy no

1 2 3 4 5 6 7 8 9 10 a

Designation

FeCuSn FeCuSnWC20 FeCuSnWC30 FeCoWC FeCuWC FeCuNi FeCuCN FeCuNi5CN FeCuNi10CN FeCuSnNi10CN

Volume fraction of austenite (vol%)a As-sintered (polished) condition

As-ground condition

o1 o1 o1 o1 o1 o1 o4 42 85 75

– – – – – – – 40 60 59

Assuming that V(αFe)+V(γFe) ¼100%.

Grinding and polishing experiments conducted on annealed polycrystalline brass indicate that the plastically deformed layer extends to 77 and 0.7 μm after plain grinding on #220 SiC paper and after fine polishing on 1 μm diamond respectively, whereas a significant plastic deformation occurs only after grinding and extends down to  8 μm beneath a 2 μm thick heavily scratched skin [5]. Therefore by using X-ray diffraction (XRD) it is possible to determine the proportion of (γFe) to (αFe) in the subsurface layer of each specimen both after grinding and after polishing in order to predict the amount of martensite which can be generated beneath the working face of the tool by abrasion. In the present work, Cu Kα1 radiation was used and two separate diffraction lines were measured, namely 111 and 110 for the fcc (γ) and bcc (α) structure, respectively. The volume fraction of austenite (V(γFe)) was calculated using the following equation: V ðγFeÞ ¼ ½1:44I 111γ =ðI 110α þ 1:44 I 111γ Þ100%

ð3Þ

J.S. Konstanty, D. Tyrala / Wear 303 (2013) 533–540

where I111γ and I110α are the integrated intensities of the 111 and 110 diffraction lines for the (γFe) and (αFe) phase, respectively. The results are given in Table 3. The cobalt-base Alloy 11 was left out from the XRD tests although massive strain-induced martensitic transformation had earlier been detected in sintered cobalt [6,7]. In contrast to ironbase alloys, however, the slight volume contraction attributed to martensitic reaction in cobalt generates tensile stress in the abraded layer which may adversely affect its wear resistance. As the Co–20%WC matrix is commonly used in the diamond tooling industry for abrasive applications, Alloy 11 was used in wear tests as a reference material.

537

index Ai(3) was calculated as: Aið3Þ ¼

∑ΔM i 104 V Aðρs þ 2:38Þ



μm=20 m



ð4Þ

where ΔM—mass loss of individual test piece per 20-s wear interval [g], V—sliding velocity [m/s], A—wear surface of test specimen [cm2] and ρ—specific density of tested specimen [g/cm3].

2.3. Wear testing 2.3.1. 3-Body abrasion 3-Body wear measurements were performed by means of the Micro Wear Test (MWT) [8,9]. Fig. 6 shows a schematic diagram of the tester. Each test specimen consisted of three test pieces (1 70.01 cm2 each) mounted in Epofix resin. The specimen was positioned in a special specimen holder and forced against an eccentrically grooved cast iron backing-wheel (  86 HRB) at a pressure of 800 kPa. Approximately 30 g portion of fine quartz abrasive (o 200 μm) was suspended in Butandiol 1.4 and spread over the backing-wheel by means of a specially designed spreader when the specimen holder and the backing-wheel were set into clockwise motion at the same rotational speed of 150 rpm. This resulted in the linear sliding velocity of 1.08 m/s. After each 20-s wearing interval the test pieces were cleaned ultrasonically, dried and weighed individually to the nearest 0.1 mg. The testing procedure required a few running-in intervals in order to produce a stable relief at the resin-specimen interface. As soon as the weight loss per wear interval stabilised an abrasive

Fig. 6. Schematic representation of the 3-body abrasive wear testing facility.

2.3.2. 2-Body abrasion The specimens subjected to the MWT were also tested for 2-body abrasion using the same wear testing machine. The test consisted of grinding the three test pieces representing each specimen (material) on #220 grit SiC abrasive paper for 20 s under a pressure of 300 kPa with water used as a coolant. The grinding paper was replaced with a fresh one after each test and, to ensure its possibly uniform wear, the specimen holder and turntable were set to a clockwise motion at 150 and 160 rpm, respectively, which yielded a sliding velocity altering sinusoidally between 1.29 and 1.39 m/s. Fig. 7 exemplifies the effect of unequal rotational speed of the sample holder and grinding disc on the path of an individual test piece versus the surface of the grinding paper. The loss of weight was measured for each specimen using the procedure described in the preceding section and an abrasive index Ai(2) for 2-body abrasion was calculated using Eq. (4).

2.3.3. Abrasion of diamond-impregnated specimens A set of 22 diamond-impregnated specimens (segments), representing the whole range of tested matrix materials, was used for wear tests carried out under quasi-industrial conditions. As shown schematically in Fig. 8 two specimens were tested simultaneously. Each of them was individually loaded and rubbed against a sandstone backing wheel, in the presence of water, to produce cutting action under conditions comparable with frame sawing. To ensure uniform wear of the specimen-machined part of the sandstone backing wheel the specimen holder and backing wheel were set to a clockwise motion at 150 and 160 rpm, respectively, which resulted in a spiral wear path (see Fig. 7) and sliding velocity of 1.37 m/s altering sinusoidally between 1.32 and 1.42 m/s. As the backing wheel did not wear in its central part and at the rim it was

Fig. 7. Predicted (left) and actual (right) path of a sharp metal pin on a grinding paper (sample holder: 150 rpm; and grinding paper: 160 rpm).

538

J.S. Konstanty, D. Tyrala / Wear 303 (2013) 533–540

necessary to grind it at regular intervals, usually after two testing cycles, in order to maintain its working surface flat. After a few segment sharpening cycles the measurements started. 20-s testing periods were maintained and the load of 20 N per specimen was found optimal. After each cycle the tested specimens were cleaned ultrasonically, dried and weighed individually to the nearest 0.1 mg to calculate the loss of volume. Testing cycles from 8 to 10 were performed on each specimen depending on its individual wear rate. Therefore the total loss of volume (ΔV) results were finally recalculated for an arbitrary wear path of 1000 m. The test specimens were also examined by means of a stereomicroscope for the number of diamonds present on their wear surfaces. It was evident from the experimental data that practically all specimens were prone to marked variations in the instantaneous wear rate. Similar variations were found in the number of exposed diamonds as exemplified in Fig. 9. As it was practically impossible to evaluate the actual number of cutting diamonds at any moment, an aggregate number of cutting diamonds, un-cutting (emerging) diamonds and pullouts was counted after each 20-s cycle to calculate the average number of diamonds (ANOD) seen on the wear surface of each diamondimpregnated specimen throughout the test. The results of all wear tests are summarised in Table 4.

variables were first calculated in order to determine the adjusted coefficient of multiple determination and seek strong relationships among the potential predictors. Then the regression coefficients βi were calculated to establish the true regression function y ¼ β 0 þ β 1 x1 þ ⋯ þ β i xi

ð5Þ

The model utility test was subsequently performed by testing the null hypothesis H0: β1 ¼ ⋯ ¼ βi ¼ 0, according to which there was no useful relation between y and any of the predictors, against the alternative hypothesis Ha: at least one βi≠0. The test was based on f statistic having an F distribution R2a =k 2 ð1−Ra Þ=½n−ðk

f¼

þ 1Þ

where Ra2—adjusted coefficient of multiple determination, n—number of observations and k—number of predictors. Table 4 Mean abrasive wear indices, loss of volume and average number of diamonds. Alloy no

Designation

Ai(3)a (μm/ 20 m)

Ai(2)a (μm/20 m)

3.1. Methodology

1

FeCuSn

42.7 73.0

The wear test results were analysed statistically using the multiple regression approach [10]. The main objective was to build a probabilistic model that would relate the dependent variable y (ΔV) to all predictor variables xi (ANOD, Ai(3) and Ai(2)) or to a subset of these predictors. To this end correlation coefficients for all pairs of

2

FeCuSnWC20

39.1 73.7

3

FeCuSnWC30

33.671.4

4

FeCoWC

33.277.6

5

FeCuWC

37.9 74.8

6

FeCuNi

39.3 76.7

7

FeCuCN

28.6 74.3

8

FeCuNi5CN

24.1 75.1

9

FeCuNi10CN

23,3 74.7

10

FeCuSnNi10CN 20.8 75.1

11

CoWC

206.07 10.6 1a 1b 101.3 7 3.6 2a 2b 78.0 7 6.1 3a 3b 76.8 7 20.0 4a 4b 112.9 7 5.3 5a 5b 185.7 7 10.4 6a 6b 184.3 7 15.6 7a 7b 151.8 7 28.4 8a 8b 118.9 7 19.6 9a 9b 99.7 7 5.1 10a 10b 22.5 7 4.3 11a 11b

3. Statistical analysis

Fig. 8. Schematic representation of the purpose-built rig used to test diamondimpregnated specimens.

a

33.678.2

Specimen no.

ΔV (mm3/ 1000 m)

ANOD

185.1 204.5 216.3 84.2 150.9 81.1 47.6 286.3 119.8 105.1 133.3 79.4 44.7 50.0 53.2 55.3 77.4 80.4 32.1 68.4 52.2 68.1

11.0 10.3 10.8 20.3 13.6 20.2 22.8 9.1 12.5 20.1 12.4 18.8 19.7 19.4 19.8 20.5 13.0 13.2 19.3 12.1 19.4 19.9

Confidence intervals were estimated at 90% confidence level.

Fig. 9. Wear surfaces of two diamond-impregnated specimens having nominally the same diamond concentration.

J.S. Konstanty, D. Tyrala / Wear 303 (2013) 533–540

Table 5 Correlation coefficients (Alloys 1–11).

ΔV ANOD Ai(3) Ai(2)

Table 10 Backward elimination results (Alloys 8–10).

ΔV

ANOD

Ai(3)

Ai(2)

1

−0.79 1

0.56 −0.22 1

0.11 −0.20 0.18 1

Ra2

No of predictors

1 2

3 2

0.754 0.753

p-Value (H0: β1 ¼ ⋯ ¼ βi ¼ 0)

0.00001 0.000002

p-Value (H0: βi ¼0)

3 2 1

Ai(2)

0.018 0.001

0.273 0.005

0.523 –

0.31 –

The calculations were then repeated separately for the ironbase Alloys 1–7, having (αFe) matrix, and Alloys 8–10, which undergo abrasion-induced martensitic transformation (γFe)(αFe). The resulting correlation coefficients are given in Tables 7 and 8. The stepwise regression (variable selection) data is presented in Tables 9 and 10, and the best multiple regression models are described by means of Eqs. (7) and (8).

Ai(3)

Ai(2)

1

−0.90 1

0.43 −0.46 1

−0.09 −0.16 0.27 1

ΔV

ANOD

Ai(3)

Ai(2)

1

−0.80 1

0.31 0.31 1

−0.02 0.60 0.90 1

0.831 0.0003 0.841 0.00004 0.791 0.00002

Ai(3)

0.0014 0.0017

p-Value (H0: βi ¼ 0) ANOD

1 2 3

ANOD

0.00001 0.00001

ANOD

p-Value (H0: β1 ¼ ⋯ ¼ βi ¼ 0)

0.054 0.006

p-Value (H0: βi ¼ 0)

ΔV ¼ 154−10:7ANOD þ 3:8Aið3Þ

ΔV

Ra2

0.964 0.969

p-Value (H0: β1 ¼ ⋯ ¼ βi ¼ 0)

Ai(2)

Table 9 Backward elimination results (Alloys 1–7). Step No of predictors

3 2

R2

Ai(3)

Table 8 Correlation coefficients (Alloys 8–10).

ΔV ANOD Ai(3) Ai(2)

1 2

No. of predictors

ANOD

Table 7 Correlation coefficients (Alloys 1–7).

ΔV ANOD Ai(3) Ai(2)

Step

models built by the backward elimination of predictors corresponding to the largest p-values (40.05). Hence the best model with two independent variables has the following estimated regression equation:

Table 6 Backward elimination results for the wear tests data of Table 4. Step

539

Ai(3)

Ai(2)

0.00004 0.548 0.053 0.000004 – 0.053 0.00001 – –

f was used to determine the observed significance level (pvalue) being the probability calculated assuming H0 is true. For small p-value, i.e. significant test data, another null hypothesis H0: βi ¼0 was then tested against Ha: βi≠0 in order to indentify the best model by eliminating predictors having uncertain influence on y. In this case the test statistics for testing H0 was |βi/S(βi)| and pvalue was calculated basing on t distribution.

ð6Þ

ΔV ¼ 345−13:8ANOD

ð7Þ

ΔV ¼ −28−4:5ANOD þ 7:2Aið3Þ

ð8Þ

The calculations based on the wear test data show clearly that irrespective of the matrix properties there is an inverse relationship between the number of diamonds protruding from the wear surface of a diamond-impregnated specimen and its rate of wear. Such a correlation well accords with the industrial experience. The effect of matrix resistance to abrasion is a more complex issue to which there is no straightforward answer. Regression calculations carried out for all alloys suggest that Ai(3) has bearing on the specimen wear rate, although the data presented in Table 9 imply that for the (αFe) matrix specimens Ai(3) is statistically insignificant since its p-value is very high (0.55) and the highest of the three. It is also reasonable to remove Ai(2) from the model built for Alloys 1–7 and to represent it by a simple regression Eq. (7). The p-value of Ai(2) is markedly lower (0.053) but, more importantly, its regression coefficient is low and has a negative value (ΔV ¼400−14.4ANOD−0.3Ai(2)) which misleadingly implies that a diamond-impregnated composite wears faster as the matrix resistance to 2-body abrasion increases. In the case of partly austenitic Alloys 8–10 it is evident that the martensitic transformation occurring in the matrix during tribological straining decreases the wear rate of diamond-impregnated specimens. After dropping Ai(2) the p-values of ANOD and Ai (3) markedly decrease (Table 10). It is noteworthy that both for the whole set of results and for the two analysed subsets of data there is a very low correlation between ΔV and Ai(2) (Tables 5, 7 and 8). This confirms that the test parameters have been properly chosen to simulate the real saw blade wear conditions (mechanisms), where unacceptably short tool life due to excessive contribution of 2-body abrasion must be avoided at all cost.

4. Conclusions 3.2. Results and discussion The whole set of observations obtained for all tested materials was first analysed. The correlation coefficients between the variables are presented in Table 5, whereas Table 6 shows the sequence of

In view of the presented results the following conclusions have been reached. 1. The wear tests which involve 3-body abrasion, 2-body abrasion and a combination of these two mechanisms in a quasi-industrial

540

2.

3.

4.

5.

J.S. Konstanty, D. Tyrala / Wear 303 (2013) 533–540

testing diamond-impregnated composites ranked the examined matrix alloys in different order. In the quasi-industrial wear test performed on a sandstone counter-body the exposed diamonds are undercut from each direction due to the forward motion of each test piece, at a rate comparable to frame sawing, and rotation on its central axis which prevents formation of a matrix tail. Therefore the test seems to well imitate conditions encountered in industrial frame sawing operations. Irrespective of the matrix composition, the increasing number of exposed diamonds (increasing diamond concentration) effectively protects the matrix against abrasion by decreasing load on each individual diamond crystal, thus reducing its depth of penetration and widening the clearance for chip removal. A statistically significant, inverse relationship between the matrix resistance to 3-body abrasion and the wear rate of diamond-impregnated composites has been found for ironbase alloys containing metastable austenite. In such matrices a carbon containing, abrasion-induced martensite markedly increases their resistance to 3-body abrasion and by generating compressive stress under the working face of the tool may potentially improve retention of working diamonds. In the case of (αFe) phase matrix alloys the effects of their resistance to both 3-body and 2-body abrasion on wear rate of diamond-impregnated composites made thereof have been found statistically insignificant.

Acknowledgements The authors gratefully acknowledge Professor W. Ratuszek for her able assistance with XRD. The work was supported by the Polish Ministry of Science & Higher Education through Contract 11.11.110.158.

References [1] J. Konstanty, Powder Metallurgy Diamond Tools, Elsevier, Oxford, 2005. [2] J. Konstanty, Theoretical analysis of stone sawing with diamonds, J. Mater. Process. Technol. 123 (2002) 146–154. [3] S.W. Webb, Diamond retention in sintered cobalt bonds for stone cutting and drilling, Diamond Relat. Mater. 8 (1999) 2043–2052. [4] J. Konstanty, Cobalt as a Matrix in Diamond Impregnated Tools for Stone Sawing Applications, second ed., AGH UWND, Krakow, 2003. [5] K. Geels, The true microstructure of materials, Structure 35 (2000) 5–13. [6] J. Konstanty, A Bunsch, Hot pressing of cobalt powders, Powder Metall. 34 (1991) 195–198. [7] J. Konstanty, A Bunsch, A. Cias, Factors affecting hardness and ductility of hotpressed cobalt powders, Powder Metall. Int. 23 (1991) 354–356. [8] K. Muller, E. Fundal, The Struers micro wear model and test, in: Proceedings of the 12th International Plansee Seminar 89, Wear Resistant Mater., vol. 2, Reutte, Tirol, Austria, May 8–12, 1989, pp. 481–495. [9] E. Fundal, The Struers micro wear test, Structure 3 (1989) 3–4. [10] J.L. DeVore, Probability and Statistics for Engineering and the Sciences, seventh ed., Cengage Learning, Belmont, 2008.