eed

aneerin, Jap

Keywords:Roller test machineParticle bedBreakageGrindingSize reduction

s are op

variables involved in a rolling compression can be considered. The monitoring and data acquisitionsystems allow real-time monitoring of the pulverizing induced roller movements. Through parametric

f brittg is of

Single particle and particle bed breakage tests are commonlyused for investigating the comminution mechanism in industrialoperations (Liu and Schnert, 1996, Eksi et al., 2011; Barrioset al., 2011). Various compression testing methods have beenproposed and applied to measure the breakage characteristics ofsingle particle, which can be classied into single impact, doubleimpact and slow compression according to the mode of application

icroscopic effectsture and adhesioned for testsummariz

Bemrose and Bridgwater (1987).Vertical roller mills have been well accepted for grind

cement raw materials, clinker and slag, coal particles for ckilns and power plants for several decades. Although muchprogress has been made in understanding the grinding process ofvarious types of mill feeds (Schaefer, 2001), to the best knowledgeof the authors, a systematic study of the grinding mechanism andcomminution efciency of particle beds by a pure roller compres-sion cannot be found in literature. As one step toward such a goal, apilot roller test machine has been developed in this study, which

Corresponding author. Tel.: +86 10 62794588; fax: +86 10 62782159.E-mail address: zhouyd@mail.tsinghua.edu.cn (Y.D. Zhou).

Minerals Engineering 72 (2015) 6572

Contents lists availab

n

elsmachines, is widely accepted as an energy intensive process andmany efforts have been devoted to better understanding of theparticle breakage behaviour and improvement of energy efciencyof comminution (Tavares, 1999; Schaefer, 2001; Bourgeois, 1993).

particles which form the bulk bed, whilst the minclude particle size, particle shape, particle strucbehaviour. A list of techniques commonly employsusceptibility of particles to attrition werehttp://dx.doi.org/10.1016/j.mineng.2014.12.0140892-6875/ 2014 Elsevier Ltd. All rights reserved.ing theed by

ing ofementand industrial interest, particularly within the disciplines ofmining, civil, and chemical engineering. Typical processes includemineral processing, thermal power generation, cement manufac-turing, and pharmaceutical processing. In mining industry applica-tions, comminution of mineral blocks or particles for a nerpowder, such as grinding of ores and mineral particles in pulverizer

nearly defect-free materials, such as single crystals, to complexindustrial granules. Similar to the basic mechanism of particulateattrition by Paramanathan and Bridgwater (1983), macroscopiceffects of grinding induced size reduction may include particleimpact on containing walls, particle to particle abrasion andfracture due to external stresses and internal stresses within the1. Introduction

Fragmentation or size reduction ovarious types of compression loadinnumerical analyses on an elastic feed bed of 530 mm in thickness and 5001000 MPa in elastic modulus,it is found that the machine is capable of providing a maximum contact pressure stress in a range of 4.517.5 MPa. A series of fundamental tests have been conducted by the developed machine using a type ofbituminous coal and typical bound values of roller weight and speed. The size reduction results as well asthe measurements of roller movement demonstrate the capability of the machine as a suitable tool fortesting grinding performance. Some discussions of the potential extension of the machine are also givenin the nal part.

2014 Elsevier Ltd. All rights reserved.

le material particles byconsiderable scientic

of stresses and the number of contact points (Gotoh et al., 1997). Acomprehensive review of single-particle breakage tests was givenby Sankara (1985). He compared the single impact, double impactand slow compression tests. The materials investigated varied frommay inuence the pulverizing efciency, this study presents the development of a pilot roller testmachine, which can signicantly simplify the grinding conditions in actual pulverizers whilst the keyDevelopment of a pilot roller test machinthe pulverizing performance of particle b

C.J. Chi a, Y.D. Zhou a,, S.Q. Cao a, K. Arima b, Y. Chuma State Key Laboratory of Hydroscience and Engineering, Department of Hydraulic EnginbNagasaki Research & Development Centre, Mitsubishi Heavy Industries, Ltd., Nagasaki

a r t i c l e i n f o

Article history:Received 4 October 2014Accepted 8 December 2014

a b s t r a c t

Various types of pulverizerinto ne powders to achiev

Minerals E

journal homepage: www.for investigatingsb, T. Okafuji b

g, Tsinghua University, Beijing 100084, PR Chinaan

e commonly used in power plants for the purpose of breaking coal particlestimum combustion for the boilers. To investigate the effects of factors that

le at ScienceDirect

gineering

evier .com/locate /mineng

can signicantly simplify the grinding conditions in actual pulver-izers whilst the key variables involved in a rolling compression canbe considered. Firstly, main design issues about the test machine,including the mechanical realization, measurement and dataacquisition systems are introduced in the main portion of thispaper. A rough estimate of the range of maximum contact pressurestress is also given based on some fundamental nite elementanalyses. Secondly, elementary pulverizing tests have been con-ducted using the machine considering variable roller weights andspeeds, from which typical results of size reduction and rollermovement are provided. Some discussions on the coming exten-sion of the machine and concluding remarks are given in the nalpart.

was constrained along the longitudinal direction throughtwo liners installed on both lateral sides of the base frame(Fig. 3). A 2875 mm long ball screw was installed withinthe base frame and connected to the support table, whichserves a purpose of driving the support table along the lon-gitudinal direction.

(2) A horizontal groove was purposely built into the supporttable for containing the particle bed during pulverizing tests(Figs. 2 and 3). A small width of 32 mm was chosen for thegroove in order to obtain a higher pressure stress acting onthe feed bed by the overburdened roller weight. The grooveis 35 mm in depth, which poses a limit on the maximumlayer thickness and the largest vertical motion of the rollerpart.

Groove

Counterweight blocks

Roller

Base frame

Liners

Support table

66 C.J. Chi et al. /Minerals Engineering 72 (2015) 65722. Design and setup of a roller test machine

The working principle of the roller test machine developed inthis study is schematically shown in Fig. 1. The motions of theroller and the support table components are controlled by an ACservo motor, of which the rotation movement follows the ordergiven from a control computer through the controller unit. Themobilized roller up-and-down displacements and its rotationspeed, shall be monitored using photoelectric sensors and berecorded by a data acquisition system, which digitizes the analogsensor signals and transfers to the control computer for visualiza-tion and post-processing purpose. Main components for themechanical realization of the test machine, the control and mea-surement system, as well as parametric numerical analyses foran estimate of the maximum contact pressure stress are describedbelow.

2.1. Mechanical realization of rolling compression

As mentioned above, three main parameters that were deemedto be mostly inuential in the pulverizing process were focused inthe design of the test machine, including the roller weight, thekinematics of the roller (especially its rotation speed), and thethickness of the feed bed. The test machine was purposelydesigned such that each of these three factors can be controlledindividually for simulating various grinding conditions in actualpractice. As shown schematically in Fig. 1 and in photo in Fig. 2,the machine consists of three main parts: the base frame, the sup-port table and the loading roller. The features of these main partsare described as follows (from bottom to top):

(1) A 5000 mm long steel-made base frame of sufciently largestiffness was xed onto the ground, which comprised a150 mm thick concrete layer on a solid soil foundation and

Controller unit

Order

AC servo motor

Controls on the table speed

Test machine

Monitoring roller motion Photoelectric

sensors

Voltage outputDigital output

Command

ComputerMotion monitoring system

Motion control systemData acquisition hardware

Fig. 1. The working diagram of the developed roller test machine.was deemed strong enough for supporting the overallweight of the machine as well as vibration loads mobilizedduring the rolling tests. A straight and horizontal supporttable, which was also made of steel material and 2400 mmin length, was arranged on the base frame and its movement

Fig. 2. A front view of the roller test machine.Servo motor

Fig. 3. An iso-metric view of the test machine (3D design model).

(3) A cylinder-shaped steel roller, 200 mm in thickness and400 mm in diameter, was arranged on the support table withthe shaft in a horizontal plane and orthogonal to the advanc-ing direction of the support table (Fig. 3). The roller wasdesigned to be xed horizontally, and can move up anddown freely in the vertical direction by slider (guide) rails(Fig. 4(a)) that were installed on both lateral sides of thebase frame. The desired rotation with reference to the cen-tral shaft can be achieved when a sufcient large frictionaction is triggered along the roller-table (or feed bed) inter-face, which reasonably comes from the advancing or reced-ing of the support table component. Specically, the originalthickness of the roller (200 mm) was narrowed to a muchsmaller value of 30 mm near the outer surface, as given inFig. 3, which ts the groove width value, and can reasonablyfavour the increase of the compression stress on the feed bedto a great extent. The total weight of the roller and theadjunct components, including the shaft, the bearing andthe slider rail for controlling the roller motion, is 67 kg.

(4) In order to simulate various rolling conditions of differentpressure levels in practice, it was designed that counter-weight blocks could be added on both sides of the rollershaft by using dual tension rods (Figs. 3 and 4). The counter-weight blocks were made of lead material of a big density11.3 g/mm3, and the maximum surcharge can be as largeas 350 kg considering the requirements of a proper opera-tion of the shaft and bearing components. It can be seen thatan equivalent roller weight in a range of 67417 Kg can bechosen for the grinding tests on this machine. According to

C.J. Chi et al. /Minerals Enginthe simplied analysis results given in the following Table 1,the lower and upper limit of contact pressure stress for ahorizontal elastic layer of 530 mm in thickness and 5001000 MPa in elastic modulus are 4.5 MPa and 17.5 MParespectively.

Roller

Slider rails

Counterweight blocks

Print mark sensorDisplacement sensor

Tension rods

(a) A schematic view of the sensors arrangement

(b) The print mark poster on the lateral surface of the roller

Fig. 4. The arrangement of photoelectric sensors for monitoring roller motionduring pulverizing tests.2.2. Control, measurement and data acquisition

In addition to the above main mechanical parts of the testmachine, two systems, including the motion control system andthe motion monitoring system (Fig. 1), were designed for thepulverizing test study. The motion control system can preciselycontrol the horizontal movement of the support table, leading toa rolling compression action by the roller on the table as well asthe feed layer on it. By the motion monitoring system it becomespossible to measure the real-time response of the rollers rotationspeed and its vertical displacement during the testing process.

The motion control system consists of three main parts, an ACservo motor (Type SGDV-120A01A002000 by Yaskawa Electric(China) Co., Ltd), a system controller unit, and a motion controlcard (Type MPC08SP by Leetro automation corporation, China).The main supply is three-phase 380 V that is reduced to 220 V bya step-down transformer to power the AC servo motor. For a givenmotion of the support table, the inputs for the motion control cardplaced in the computer can be determined and the correspondingcontrol commands are given to the servo motor through the con-troller unit. As mentioned above, a ball screw links the servo motorand the support table. Reasonably a proper control of the motion ofthe ball screw can drive the support table according to a prescribedway.

The motion monitoring system consists of two types of photo-electric sensors, a data acquisition system, electrical accessoriesand other inputoutput devices. Through the introduction of twotypes of high-performance photoelectric sensors, non-contactmeasurements of the roller motion become feasible, which isimportant for investigating the interaction between the roller partand the feed bed. A high resolution (

speed is schematically shown in Fig. 5, where a high voltage(10 V) indicates the recognition of black print marks, while alow voltage (0 V) means the light spot is on the white back-ground. From the response curve monitored during pulveriz-ing tests, the time variation of the roller angular speed canbe simply determined as the speed that the sensor light spotpasses the alternating print mark and underground section,that is, by the equation below:

stress level allowed for the test machine, a series of numerical anal-yseswere conducted in this section using the nite elementmethod.Fig. 6 shows the nite element model for the roller-feed bed-tablesystem. A linear elastic assumption was made for all components,and the elastic parameters for the steel-made roller and table areEs = 210 103 MPa and vs = 0.3. The interaction along the interfacesbetween the roller and the feed bed as well as the feed bed and thetable was modelled by a masterslave contact algorithm, in whicha small friction coefcient of 0.2 was adopted. A static equilibriumsolution of the roller compressing on the particle bed was investi-

ness, a higher contact pressure stress can be mobilized, and the

Table 1Numerical results of the maximum contact pressure stress for a roller statically arranged on a horizontal coal layer.

Roller mass (kg) Maximum contact pressure stress in the contact region (MPa)

h = 5 mm h = 10 mm h = 30 mm

Ec = 500 MPa Ec = 1000 MPa Ec = 500 MPa Ec = 1000 MPa Ec = 500 MPa Ec = 1000 MPa

67 4.8 6.6 4.6 6.4 4.5 6.3210 9.2 12.3 8.9 12.1 8.5 12.0417 13.6 17.5 12.5 16.8 11.6 16.0

68 C.J. Chi et al. /Minerals Engineering 72 (2015) 6572xti p90Dti 1

in which x(ti) denotes the roller angular speed at the middle pointof each zone of constant voltage response lasting for a time intervalDti (i = 1, 2, . . . N). From a series of data points of roller speed versuspulverizing time, the whole development history of roller rotationduring grinding tests can be determined.

The data acquisition system for the roller test machine includesthe data acquisition module and the corresponding software. Wechose DAM-3058 by DonghuaTest Ltd. (China) for our machinesdata acquisition module, which can record the analog voltage orcurrent output signal, and the data acquisition sampling frequencycan be as high as 12,000 readings per second. The correspondingsoftware allows a real-time visualization of the time variation ofsensor signal on a computer displayer.

2.3. Numerical analysis of the contact pressure stress by rollercompression

Onemay note the above testmachines incapability ofmeasuringthe contact pressure stress distribution along the roller-bedinterface mobilized by the rolling compression. The difculty liesin the fact that the contact area shall changewith the rolling processand an installation of membrane sensor(s) on the contact surfacemay disturb the interaction and in turn inuence the pulverizingbehaviour. In order to give a rough estimate of the contact pressure

(V)

8

10

12

tiTime (s)0 2 4 6 8 10

Vol

tage

0

2

4

6

Fig. 5. Time variation of voltage output from the print mark sensor (a particularcase of constant roller rotation speed as an instance).increase of its overall stiffness is also benecial for the build-upof pressure stress. A range of maximum contact pressure stressbetween 4.5 and 17.5 MPa can be expected for the grinding testsusing the machine on a feed bed of 530 mm in thickness and5001000 MPa in elastic modulus. The peak contact pressurestress, 17.5 MPa, is deemed to be representative of the grindingconditions in actual pulverizers.

3. Roller test study

3.1. Test cases

To verify the effectiveness of the above developed test machine,a series of fundamental tests were conducted. The particle material

Rollergated. The effects of different test conditions were investigated,including the roller weight, thickness of feed bed and the elasticmodulus of the bedmaterial. A range of themodulus of bedmaterial(Ec) of 5001000 MPa was chosen based on the compression testresults of prismatic coal samples. Table 1 lists a summaryof themax-imum contact pressure stress from the numerical results.

The results in Table 1 show that for a feed bed of smaller thick-Table

feed bed

Fig. 6. Finite element modelling of the roller-feed bed-table interaction in grindingtests (coal layer thickness h = 20 mm as an instance).

(a) v=0.1 m/s

0

0.2

0.4

0.6

0.05 0.5

Perc

enta

ge of p

ass

ing

p

Particle size (mm)

W=200kgW=400kgFeed

1

nginTable 2The basic properties of selected bituminous coal particle material.

Type HGI Particle density (kg/m3)

Initial surface moisture content(%)

Bituminous 60 1388 7.3

Table 3A summary of elementary test cases by the roller machine.

Test case Layer thickness (mm) Roller weight (kg) Table speed (m/s)

Case I 20 100 0.1, 1.0Case II 20 200 0.1, 1.0Case III 20 400 0.1, 1.0

W

Particles in this range are chosen for size reduction assessment

C.J. Chi et al. /Minerals Eselected for the tests is a type of bituminous coal from Jiamusi, Hei-longjiang Province, China. Table 2 lists the basic properties of thecoal particle material. A type of jaw crusher was applied to preparethe feed particles and the output coal particles was scalped by1.25 mm. The particle size distribution of the feed coal was deter-mined by sieve analyses according to the Chinese standard GB/T25651998. The particle size distribution in mass was found tobe 4% up to 0.5 mm, 5% between 0.5 and 0.63 mm, 63% between0.63 and 1.0 mm, and 28% between 1.0 and 1.25 mm. One can notethat the feed samples for the tests contain particles mostly within anarrow range between 0.63 and 1.25 mm, which are similar to thesampling requirements for the common Hardgrove grindabilityindex (HGI) tests (ISO 5074:1994).

Typical factors which may inuence the breakage efciency ofparticle bed by rolling compression were chosen in this studyand a summary of the test conditions is given in Table 3. Three dif-ferent roller weights, 100 kg, 200 kg and 400 kg, were chosen,which respectively represent the minimum, intermediate andmaximum bound values. Similarly, two velocities of the supporttable were chosen, including a lower one of 0.1 m/s and a higherof 1.0 m/s that is normally adopted in pulverizers. A feed bed of20 mm thickness containing the coal particles was adopted in eachtest, of which the length of a constant thickness section is 330 mm,and an inclined slope of 30 was arranged at both ends (Fig. 7) forsupporting the main particles into a horizontal bed. Only the cen-tral portion of the particle bed, as indicated by a red rectangle inFig. 7, is to be collected for size reduction assessment by sievetools. The purpose of such a treatment is to exclude the side effectnear both ends caused by varying thickness, and the designed test

Coal bed

Table

=30

10 mm >500 mm

vh

Fig. 7. A schematic illustration of the arrangement of feed bed in pulverizing tests.0.8

art

icle

W=100kg1

eering 72 (2015) 6572 69conditions can be representative of the circularly arranged longcoal bed under continuous grinding in actual pulverizers. Threetests were conducted for each case as given in Table 3.

3.2. Typical test results

3.2.1. On the particle size reductionFirstly, the roller weight (W) was taken as an independent var-

iable, and Fig. 8 presents the mean test curves of particle size dis-tribution before and after one-time pulverizing process. It can be

(b) v=1.0 m/s

0

0.2

0.4

0.6

0.8

0.05 0.5

Perc

enta

ge of p

ass

ing

part

icle

Particle size (mm)

W=100kgW=200kgW=400kgFeed

Fig. 8. Effect of roller weight (W) on the particle size reduction by one-time rollingcompression (ab).

ngin0.8

1tic

le70 C.J. Chi et al. /Minerals Eseen that reasonably an increase of the roller weight from 100 kgto 400 kg can signicantly improve the disintegration intensity,for either a lower or a higher table speed allowable for themachine. For the W = 400 kg case, nearly 30% of the feed particlesof 0.631.25 mm in size were disintegrated by one-time rollingcompression, and the generated products mostly fell in a size rangebetween 0.25 and 0.5 mm. Moreover, the gure shows that for theconsidered range of roller weight, the products for the ner parti-cles (

ngin(a) v=0.1 m/s

0

0.1

0.2

0.3

0.4

0.5

0.6

0 1 2 3 4 5

rolle

r ang

ular

spee

d (ra

d/s)

Time (s)

W=100kgW=200kgW=400kg

4

5

6

7

8

gula

r spe

ed (r

ad/s)

C.J. Chi et al. /Minerals E3.2.2. On the roller motion measurementThe monitoring system in the test machine allows the measure-

ment of roller movement during pulverizing tests on the coal beds.Fig. 10 presents typical results of the roller vertical displacementwith time, and the response during the compression on the feedlayer was focused. For each case of variable roller weights andtable speeds, a feed coal bed of the same geometry (Fig. 7) wascarefully prepared before the test, ensuring all the coal beds wereapproximately of the same compaction degree. The results inFig. 10 demonstrate that the development of the roller vertical dis-placement followed a similar variation pattern for all the cases.Each curve is roughly symmetric with respect to the midpoint,which can be attributed to the symmetry of the geometry of thefeed layer. It is also shown that the slope of the descending portionis a bit smaller than that of the ascending portion, which can bereasonably explained by the difference in mechanical interactionbetween the roller and coal bed particles at the two stages. Theascending portion in each curve indicates the climbing of the rolleronto the coal bed, and the compression from the roller has a trendof compacting the particle towards the central portion. For thedescending portion, differently the roller passed most portion ofthe coal bed and its compression on the particles arranged on thesloping portion may tend to carry away particles and lengthenthe bed to some extent.

If an assumption is made that the roller keep in touch with thebed during pulverizing on it, the displacement curves in the aboveFig. 10 also equivalently indicate the surface prole of the coal bed

of v = 1.0 m/s. From these perturbations it can be inferred that dur-

(b) v=1.0 m/s

0

1

2

3

0 0.1 0.2 0.3 0.4 0.5

Rol

ler a

n

Time (s)

W=100kgW=200kgW=400kg

Fig. 11. Comparison of monitored roller angular speed during pulverizing tests(h = 20 mm cases were taken as instances).ing the roller rolling on the feed bed, momentary slippages weretriggered along the roller-bed interface for each test, which canbe attributed to the complex interaction along the dual interfacesand the nonlinear behaviour of the coal particles forming the feedbed.

4. Discussions and conclusions

The typical test results described above clearly demonstrate thecapability of the developed machine for conducting pulverizingtests on particle beds under various conditions. In addition to theabove elementary functions, further developments shall be madeby the authors on the test machine to extend its capabilities,including but not limited to the following:

(1) A photoelectric sensor is to be installed onto the side framefor a non-contact measurement of the surface prole of thefeed bed. The aim is to determine the geometric feature ofthe feed bed before and after the rolling compression, whichis essential for the evaluation of the energy spent on theinternal deformation within the coal layer and the assess-ment of pulverizing efciency.

(2) A particular working condition of the roller, that is, a puresliding on the feed bedwith rotation around the shaft prohib-ited, is to be achieved by using caliper brakes for the roller. Aninitial vertical gap between the roller and the feed bed canalso be obtained by lifting the roller with two jacks on bothsides. Such a designed condition can somewhat simulatethe actual operating condition in vertical-spindle pulverizers(https://www.mhps.com/en/products/detail/coal_pulverizer.html), as the discrepancy between the velocity at the surfaceof the conical roller and that at the table surface, particulardue to the inconsistency near the inner and outer sidesurfaces, would cause slippage at both lateral sides. Theafter one-time rolling compression along the longitudinal section.It can be expected that the rolling compression in each test woulddecrease the original thickness of the feed bed (h = 20 mm), andthe adoption of different roller weights would mobilize differentsurface settlements of the particle layers. The monitoring resultsof the roller displacement in Fig. 10 approve the expectations well.From a series of tests using different roller speeds in a range of 0.11.0 m/s, it was found that one-time rolling compression can inducethe surface settlement of the feed bed by 3.13.9 mm (W = 100 kg),4.85.6 mm (W = 200 kg), and 6.07.0 mm (W = 400 kg). Obvi-ously, an increase of the roller weight from 100 kg to 400 kg, whichin turn enlarged the contact pressure stress along the interface, canmobilize an extra surface settlement of the coal bed of 3.04.0 mm.It was also found that the variation of the roller speed can inducesome slight difference in the settlement response.

Besides the roller vertical displacement, its rotation speedresponse with time simultaneously monitored during pulverizingon the feed bed is also shown in Fig. 11. If an assumption is madethat no slippage occurs along the two interfaces, namely betweenthe roller and the feed bed, and between the support table and thelayer, a linear relationship can be formulated between the giventable speed v and the roller rotation speed w as w = v/r, where rindicates the roller radius (200 mm). The curves in Fig. 11 demon-strate that averagely the roller rotation speed follows the linearrelationship, which are 0.5 rad/s and 5.0 rad/s for v = 0.1 m/s andv = 1.0 m/s case respectively. Moreover, notable perturbations ofthe roller speed were detected during pulverizing on the bed,which became more prominent for the case of a larger table speed

eering 72 (2015) 6572 71breakage efciency contributed by the roller sliding actioncan be investigated by such a test arrangement.

(3) In actual vertical-spindle pulverizers, it is common to usespring frame hydraulic preload or a spring canister con-necting to the journal assembly of each grinding elementfor extra pressure on the feed bed. The stiffness propertyof these springs shall inuence the grinding performance.Similarly, a spring frame can be added to the above testmachine for imposing extra pressure load on the rollershaft, which can provide an opportunity for better investi-gation into the inuence of spring frames of various stiff-ness properties.

Based on the work presented in the above sections, the follow-ing conclusions can be drawn in regard to the developed roller testmachine:

(1) The present roller test machine can consider key variablesthat may inuence grinding efciency by rolling compres-sion, including the thickness of feed bed, roller weight (orgrinding pressure), the velocity of the support table and inturn the roller rotation speed. A special sliding action byconstraining the rotation of the roller can also be simulated.The capabilities of the developed machine allows a deeperinvestigation into the grinding performance of various actualpulverizers through laboratory test study.

(2) The roller movements, including its vertical displacements

(4) A series of fundamental rolling tests have been conducted onparticle beds belonging to a type of bituminous coalmaterial. The effects of roller weights and the speed of thesupport table have been studied and reasonable test resultswere provided. Besides the purposely-designed one-timepulverizing test, the test machine is also capable of doingrolling tests on single particle, and repeated pulverizing testsconsidering various combinations of controlling factors.

Acknowledgements

The authors are grateful for the kind support from MitsubishiHeavy Industries Ltd., Japan. Additionally, the support from theNational Natural Science Foundation of China (No. 51479096)and from State key Laboratory of Hydroscience and Engineering(Nos. 2012-KY-04, 2013-KY-02, and 2014-KY-1) as well as the sup-port by Tsinghua University Initiative Scientic Research Programunder Grant 20131089285 are gratefully acknowledged.

References

Barrios, G.K.P., de Carvalho, R.M., Tavares, L.M., 2011. Modeling breakage ofmonodispersed particles in unconned beds. Miner. Eng. 24 (3), 308318.

Bemrose, C.R., Bridgwater, J., 1987. A review of attrition and attrition test methods.Powder Technol. 49 (2), 97126.

72 C.J. Chi et al. /Minerals Engineering 72 (2015) 6572and angular speed mobilized during pulverizing process,can be monitored by photoelectric sensors in a non-contact manner in the developed machine. These measure-ments can help determine the work for deforming the feedbed as well as that for particle disintegration, which is ofgreat signicance for the assessment of the grindingperformance.

(3) A series of 3-D nite element analyses have been conductedto study the maximum contact pressure stress on an elastichorizontal feed bed allowed by the test machine. For a feedbed of 530 mm in thickness and 5001000 MPa in themodulus, it is found that a maximum contact pressure stressof about 17.5 MPa is available for the test machine.Bourgeois, F.S., 1993. Single-particle fracture as a basis for microscale modeling ofcomminution processes. Department of Metallurgical Engineering, University ofUtah.

Eksi, D., Hakan Benzer, A., Sargin, A., et al., 2011. A newmethod for determination ofne particle breakage. Miner. Eng. 24 (3), 216220.

Gotoh, K., Masuda, H., Higashitani, K., 1997. Powder Technology Handbook, 2nd ed.Marcel Dekker, New York.

ISO 5074:1994 Hard coal determination of Hardgrove grindability index.Liu, J., Schnert, K., 1996. Modelling of interparticle breakage. Int. J. Miner. Process.

44, 101115.Paramanathan, B.K., Bridgwater, J., 1983. Attribution of solidsI: cell development.

Chem. Eng. Sci. 38 (2), 197206.Sankara NS. Australasian institute of mining and metallurgy bulletin and

proceedings 1985; 291(4).Schaefer, H.U., 2001. Loesche vertical roller mills for the comminution of ores and

minerals. Miner. Eng. 14 (10), 11551160.Tavares, L.M., 1999. Energy absorbed in breakage of single particles in drop weight

testing. Miner. Eng. 12 (1), 4350.

Development of a pilot roller test machine for investigating the pulverizing performance of particle beds1 Introduction2 Design and setup of a roller test machine2.1 Mechanical realization of rolling compression2.2 Control, measurement and data acquisition2.3 Numerical analysis of the contact pressure stress by roller compression

3 Roller test study3.1 Test cases3.2 Typical test results3.2.1 On the particle size reduction3.2.2 On the roller motion measurement

4 Discussions and conclusionsAcknowledgementsReferences