Compressor Instability with Integral Methods Episode 1 Part 3 ppsx

The evaluation of an abrasive material for blast cleaning applications includesthe following important parameters: r particle size distribution; r average grain size.. The ship between a

Abrasive Materials

2.1 Classification and Properties of Abrasive Materials

A large number of different types of abrasive materials is available for blast ing applications Most frequently applied abrasive materials are listed in Table 2.1.Table 2.2 lists numerous physical, chemical and technical properties of commercial

clean-abrasive materials Basically, there can be distinguished between metallic clean-abrasive materials and non-metallic abrasive materials.

The evaluation of an abrasive material for blast cleaning applications includesthe following important parameters:

r particle size distribution;

r average grain size

2.2 Abrasive Material Structure and Hardness

2.2.1 Structural Aspects of Abrasive Materials

Structural aspects of abrasive materials include the following features:

r inclusions (water–gas inclusion and mineral inclusion)

C

 Springer 2008

Steel grit and steel shot 0.35

Table 2.3 lists typical values for some abrasive materials Table 2.4 displays

a commercial technical data and physical characteristics sheet for a typical blastcleaning abrasive material

Abrasive particles contain structural defects, such as microcracks, interfaces,inclusions or voids Very often, these defects are the result of the manufacturingprocess Strength and fracture parameters of materials can be characterised throughcertain distribution types A widely applied distribution is the Weibull distribution,and it was shown by Huang et al (1995) that this distribution type can be applied toabrasive materials The authors derived the following relationship between fractureprobability, particle strength and particle volume:

The strength parameterσ* is a constant, which is related to the defects

distri-bution The power exponent mW is the so-called Weibull modulus; it can be read

from a graphical representation of (2.1) Low values for m indicate a large intrinsic

variability in particle strength A Weibull plot for aluminium oxide abrasive ticles, based on the results of compressive crushing tests, is displayed in Fig 2.1.Values for the Weibull modulus estimated for different abrasive materials are listed

par-in Table 2.5 There is a notable trend par-in the values that both fracture strength andWeibull modulus drop with increasing particle size Therefore, scatter in strength

of abrasive particles can be assumed to be wider for larger particles The ship between abrasive particle size and fracture strength of the particles is shown inFig 2.2 This phenomenon can be explained through the higher absolute number ofdefects in larger particles The probability that a defect with a critical dimension (forexample, a critical crack length in a fracture mechanics approach) exists increaseswith an increasing number of defects

relation-This effect was also observed by Larssen-Basse (1993) relation-This author found alsothat the Weibull modulus of abrasive particles depended on the atmospheric humid-ity Larssen-Basse (1993) performed crushing tests with SiC-particles, and he foundthat, if humidity increased, the Weibull modulus and the number of fragments both

Andradite 80–90 12.091 (0.009) 1,767.61

increased This feature can be attributed to moisture-assisted sharpening of the tips

of surface defects present in the particles

The presence of defects, such as cracks and voids, affects the cleaning anddegradation performance of abrasive materials Number and size of defects are,

Table 2.4 Data sheet for a garnet blast cleaning abrasive material (Reference: GMA Garnet)

homo-Fig 2.1 Weibull plot for the strength of aluminium oxide particles (Verspui et al., 1997) Abrasive

particle size: 10–500 μ m

therefore, important assessment criteria Cast steel shot, for example, should notcontain cracked particles, as illustrated in Fig 2.3, in excess of 15% Cast steelgrit should not contain cracked particles, as shown in Fig 2.4, in excess of 40%(SFSA, 1980) Requirements for the defects of particles of metallic abrasive mate-rials are listed in Table 2.6

Table 2.5 Strength parameters for abrasive materials (Yashima et al., 1987; Huang et al., 1995)

Abrasive material Grain size

in mm

Fracture strength

in MPa

Weibull modulus

∗ a in MPa/mm 3

Fig 2.2 Relationship between abrasive particle size and particle fracture strength (values from

Huang et al., 1995)

Fig 2.3 Cracks in cast steel shot particles; magnification: 10× (SFSA, 1980)

Fig 2.4 Cracks in cast steel grit particles; magnification: 10× (SFSA, 1980)

2.2.2 Hardness of Abrasive Materials

The hardness of abrasive materials is usually estimated by two types of tests:

a scratching test for non-metallic abrasive materials, which delivers the Mohs hardness, and indentation tests for metallic materials, which deliver either the Knoop hardness or the Vickers hardness Respective values for commercial abrasive

materials are listed in Table 2.2

Mohs hardness is based on a scale of ten minerals, which is provided in Table 2.7.The hardness of a material is measured against the scale by finding the hardest

Table 2.6 Particle defect requirements for metallic abrasive materials (ISO 11124/2-4)

Property Chilled iron grit High-carbon cast

steel shot

High-carbon cast steel grit

Low-carbon cast steel shot Particle shape Max 10% shot

max 5% for grit above 700 HV

Max 5% non-round

Shrinkage

defects

Apatite 5 Orthoclase (Feldspar) 6

If they do not scratch the plate, their hardness is<Mohs 7 It is because of this

pro-cedure that data sheets for mineral abrasive materials often list the Mohs hardness

as>7 only.

The principles of two frequently applied indentation hardness tests are trated in Table 2.8 In laboratory practice, an abrasive particle is embedded in

illus-a speciillus-al resin millus-atrix, illus-and it is then being polished in order to obtillus-ain illus-an even

Table 2.8 Indentation hardness measurement methods (Images: TWI, Cambridge, UK)

(b)

Fig 2.5 Vickers hardness distributions of two cut wire samples (Gesell, 1979) (a) Laboratory sample; (b) Work sample

The hardness of metallic abrasive particles is a probabilistic parameter, and thehardness values mentioned in data sheets are mainly mean values only Two typ-ical abrasive hardness distribution diagrams of cut wire samples are provided inFig 2.5 Figure 2.5a shows the distribution of a laboratory sample, whereas Fig 2.5billustrates the distribution of a working sample Although both materials had equalhardness designations of 420 HV, the distributions differed widely The laboratorysample had a unimodal distribution with a maximum at a Vickers hardness of about

430 HV, whereas the working sample featured a multimodal distribution The ness distribution of the laboratory sample can be expressed through a Normal dis-tribution – this is shown in Fig 2.6 This result points to a rather homogeneousresponse of the wire material to the indentation with the Vickers pyramid, which isnot always the case (Lange and Schimm¨oller, 1967) Such a distribution was alsoreported by Flavenot and Lu (1990) for steel wire shot

hard-Fig 2.6 Normal distribution function for the laboratory cut wire sample plotted in hard-Fig 2.5a

2.3 Abrasive Particle Shape Parameters

2.3.1 Basic Shape Definitions

The following three basic shape definitions are provided for abrasive particles usedfor blast cleaning applications:

r shot;

r grit;

r cylindrical

The corresponding designations are listed in Table 2.9 Examples for two shape

definition are displayed in Fig 2.7 The term shot characterises grains with a

pre-dominantly spherical shape Their length-to-diameter ratio is<2, and they do not

exhibit sharp edges or broken sections The term grit characterises grains with a

predominantly angular shape These grains exhibit sharp edges and broken sections

The term cylindrical denotes grains that are manufactured by a cutting process Their

length-to-diameter ratio is∼1 Thisshape canonly be found withcut steel wire pellets

2.3.2 Relative Proportions of Particles

Shape parameters characterise the shape of individual particles Wadell (1933) andHeywood (1933) were probably the first who gave rigorous analyses of shapeparameters Heywood (1933) considered the shape of a particle to have the followingtwo distinct characteristics:

r the relative proportions of length, breadth and thickness;

r the geometrical form

The relative proportion includes two parameters: (1) the elongation ratio (rE) and

(2) the flatness ratio (rF) Both parameters are defined and illustrated in Table 2.10.Bahadur and Badruddin (1990) applied the elongation ratio to investigate the in-fluence of the abrasive particle shape on particle impact erosion processes Theyfound notable relationships between abrasive type, abrasive particle diameter andabrasive particle shape Some results of their study are provided in Fig 2.8 Silicacarbide particles became more elongated and less circular with an increase in theparticle size, while the opposite was the case with aluminium oxide particles Thegeneral variation of silica oxide was similar to that of silica carbide particles, thoughnot as systematic The elongation ratios for the silica carbide particles and for the

Table 2.9 Grain shape designations

Designation Grain shape Symbol

parti-ratios than silica carbide particles For a particle diameter of dP = 300μm, as

an example, the elongation ratio was rE = 0.53 for silica carbide, and rE = 0.7

for silica oxide A relationship between particle abrasive size and shape was alsonoted by Djurovic et al (1999) For starch media, these authors found that smallerparticles were less elongated than larger particles These results clearly show thatparticle shape may be considered an abrasive material characteristic

2.3.3 Geometrical Forms of Particles

The geometrical form is a volumetric shape factor, representing the degree to which

a particle approximates an ideal geometric form (cube, sphere or tetrahedron) Thefollowing two parameters can describe the geometrical form of particles: (1) the

sphericity (S ) and (2) the roundness (S )

Table 2.10 Shape parameters for abrasive particles

Parameter and definition Graphical expression

Fig 2.8 Relationships between abrasive material, particle size and particle shape (Bahadur and

Badruddin, 1990)

The sphericity, introduced by Wadell (1933), is defined and illustrated inTable 2.10 In two dimensions, the sphericity is related to the projection area ofthe sphere yielding the roundness, which is defined and illustrated in Table 2.10 aswell Both sphericity and roundness range from “0” for very angular particles to “1”for ideally round particles Hansink (1998) defined an alternative roundness scale,which is illustrated in Fig 2.9, for the assessment of the shapes of abrasive particles

This scale defines and quantifies the often used qualitative terms angular or rounded.

Several references used roundness–sphericity diagrams in order to characterise theshape of abrasive particles Such a roundness–sphericity diagram is illustrated inFig 2.10

Vasek et al (1993) and Martinec (1994) suggested a circularity factor, whichwas originally developed by Cox (1927), and a shape factor in order to characterise

abrasive particles The circularity factor (F0) is defined and illustrated in Table 2.10.For a perfectly round particle, circularity factor will be unity Gillespie (1996) andGillespie and Fowler (1991) applied image analysis in order to estimate circularityfactors (which they called “shape factors” in their papers) for shot peening media,

and they defined any value for the circularity parameter F0> 0.83 as acceptable for

shot peening applications Some of their results, featuring circularity factors for anumber of real abrasive particles, are illustrated in Fig 2.11, and it can be seen that anotable number of particles would not meet the critical circularity factor Figure 2.12shows a histogram of circularity factors based on an automatic image analysing

Designation Very angular Angular Sub-angular

Fig 2.9 Designations for angular and rounded particle shapes (Hansink, 1998)

Fig 2.10 Roundness-sphericity diagram for a garnet abrasive material (Reference: Bohemia

Garnet)

by distributions with certain statistical parameters Typical statistical parameters for

an assessment procedure are listed in Table 2.11; the listing very well illustrates thehigh number of assessment parameters delivered by an automatic image analysisprocedure

The shape factor (Fshape) is also defined and illustrated in Table 2.10 For circles,the shape factor is unity Table 2.12 lists some typical values for circularity andshape factors for a number of different abrasive materials

Fig 2.12 Frequency distribution functions of shape factors (Gillespie and Fowler, 1991)

Table 2.11 Statistics of circularity factors of cast steel shot S-280, based on automatic image

analysis (Gillespie and Fowler, 1991)

in terms of screen size, very fine particles are measured in micrometer or ter A number of “diameter” definitions are known The diameter is defined either

manome-in terms of some real property of the particle, such as its volume or surface area, or manome-interms of behaviour of the particle in some specific circumstances, such as settling inwater under defined conditions (Kelly and Spottiswood, 1982) In the area of blastcleaning, the particle size is usually given in mesh designation (according to theTyler-Standard-Screen sieve series), which barely mentions the related particle sizedistributions or the shape of the particles A regression study made to link the Tylersieve series to the corresponding average particle diameter delivers the followingrelationship:

Table 2.13 Sieve analyses results for two abrasive mixtures (Metabrasive Ltd.)

Sieve size in μ m Weight in %

2.4.1.3 Particle Size Distribution Models

A number of models were developed to mathematically describe the size tions of fine-grained comminution products, which include abrasive particles Thesemodels have empirical relationships, which to a greater or lesser extent were foundcapable of describing comminution products size distributions Table 2.14 lists themost frequently used models These equations are all of the general type:

The higher the value for nM, the more homogeneous is the grain size structure of

the sample For nM → ∞, the sample consists of grains with equal diameters.Figure 2.14a, b shows fits for the sieve analysis from Table 2.13 by two commonparticle-size distribution functions The Rosin–Rammler–Sperling–Bennett (RRSB)distribution is of particular interest because its distribution parameters are utilised

by some authors as a measure of the ageing and reusability of metallic abrasiveparticles (Wellinger et al., 1962)

distri-Table 2.14 Particle-size distribution functions (Kelly and Spottiswood, 1982; Schubert, 1988)

Function Formula M0(dP ) Significance of d* Equation Logarithmic probability erf

According to regulations in ISO 1117, the particle diameter is defined according to

a particle “size class” A size class of “140”, for example, means a particle diameter

of 1.4 mm

If the particle size distribution is known from the sieve analysis, several “average”

diameter values of the particle sample can be estimated The median diameter, d50,

is the 50% point on any cumulative distribution curve (Fig 2.13b) For the examples

presented in Table 2.13 and Fig 2.13, this diameter is dP50= 510μm (Alumina 700)

and dP50 = 280μm (MG 65), respectively The geometric mean diameter, dPG, isbased on the assumption of an even graduation in size from maximum to minimum,and it assumes an equal number of particles in each size average:

dPG= dP max+ dP min

In the examples given in Table 2.13 and Fig 2.13, this diameter is dPG= 740μm

(Alumina 700) and dPG = 362μm (MG 65), respectively A third approach is the

definition of a statistical diameter, dPSt, which follows the equation:

dPSt =

n

i =1(mi· dPi)

For the examples in Table 2.13 and Fig 2.13, the statistical diameter is dPSt= 613μm

(Alumina 700) and dPSt= 345μm (MG 65), respectively

2.4.3 Alternative Abrasive Particle Size Assessment Methods

Particle sizes, but also particle size distributions, can be assessed also by plying image analysis methods This alternative approach is not a standard in

the blast cleaning industry, although promising results have been reported forthe image analysis of shot peening media Gillespie (1996) and Gillespie andFowler (1991) performed comparative size measurements by using conventionalsieve analysis, a digital micrometer and image analysis Some results are dis-played in Fig 2.15 The agreement between the three methods depended on thesieve size; it was very good (less than 2 wt.%) for the smaller sieve sizes Theaverage difference between sieve analysis and image analysis was 2.71 wt.% Im-age analysis is of definite interest because this method can deliver information

on particle size as well as on particle shape Promising experience is available

on the shape assessment of particles, either of shot peening media (Gillespie andFowler, 1991; Gillespie, 1996) or of erosion debris (Momber and Wong, 2005b),with image analysis methods Further details on this application are provided inSect 2.3

Optical methods for the assessment of particle sizes are very familiar in particletechnology Sparks and Hutchings (1993) have, however, shown that these methodsmust be applied with caution to broken abrasive particles Especially glass particlesshow different optical properties whether they are round (e.g glass beads) or broken(e.g glass grit) Broken glass particles would, in a correct orientation with respect

to a laser beam, diffract light in such a way so as to suggest that they were of largerdiameter

Fig 2.15 Comparison between abrasive size assessment methods (Gillespie, 1996)