Minerals Engineering 23 (2010) 1237–1244

Contents lists available at ScienceDirect

Minerals Engineering

journal homepage: www.elsevier .com/locate /mineng

Improvement of energy efficiency of rock comminution throughreduction of thermal losses

Nenad Djordjevic ⇑Julius Kruttschnitt Mineral Research Centre, SMI, The University of Queensland, 40 Isles Road, Indooroopilly, Brisbane, Qld. 4068, Australia

a r t i c l e i n f o a b s t r a c t

Article history:Received 7 June 2010Accepted 18 August 2010

Keywords:ComminutionProcess optimizationThermalHPGR

0892-6875/$ - see front matter � 2010 Elsevier Ltd. Adoi:10.1016/j.mineng.2010.08.019

⇑ Tel.: + 61 7 3365 5888; fax: +61 7 3365 5999.E-mail address: n.djordjevic@uq.edu.au

The principal objective of this work was to develop a thermal imaging technique to measure the radiantheat coming from rock particles during or immediately after crushing, with the purpose of minimisingenergy losses while maintaining the efficiency of rock crushing. The main goal of the work was energyoptimization of crushing in High Pressure Grinding Rolls (HPGR). We were able to perform reproduciblemeasurements of the temperature increase that occurs during transient events such as dynamic rockbreakage and HPGR crushing. Results obtained show that with an increase of energy introduced, thereis an increase in the maximum temperature along the fractured surface as well as increases in the overallamount of thermal energy.

Results obtained during HPGR testing clearly indicate that there is an optimum intensity of pressure towhich rock needs to be exposed. Any further increase in pressure, results in only a marginal increase infragmentation and a significant increase in unproductive heating of rock. We were also concerned aboutthe effect of the size of particles coming into the HPGR. The fraction of new fine material (fines) producedduring HPGR crushing is much higher in the case of feed with a narrow size distribution, i.e. without pre-existing fines. Significantly, improved performance is achieved with a reduced amount of net comminu-tion energy.

Observed relative crushing inefficiency of feed with a wide fragment size distribution (containing finesand coarse particles), is due to a large amount of pre-exiting fines, which clog the pore space betweencoarser fragments. In the compressed zone of the HPGR this creates approximately hydrostatic compres-sive loading conditions, which require much higher pressure (i.e. energy) to cause breakage of coarserparticles. Experimental results indicate that up to 40% of energy can be saved through optimization ofthe applied pressure and modification of feed fragments size distribution.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Rock size reduction is an energy intensive process. Estimations ofenergy consumption by DOE (2004, 2005) during mining operationsshow that 39% of energy is used for beneficiation and processingoperations of which 75% is accounted for by comminution, indicat-ing that comminution consumes approximately �30% of the totalmining energy in the USA. It is likely that a similar ratio will be appli-cable in other countries. Tromans (2008) reports data compiled byNRC (2006) that show the total energy use in Canada for 2001 was7890PJ of which the industrial sector consumed 3000PJ and miningrequired 16.7% (501PJ) of this sector (NRC, 2004).

Similarly, in the 12 month period 2001–2002, total energy con-sumption in Australia was 5155PJ of which mining consumed260.9PJ (SOE, 2006), indicating comminution accounted for 1.48%

ll rights reserved.

of total energy usage. In the case of South Africa, total nationalenergy consumption in 2003 was 5159PJ (EIA, 2005) of which 6%was attributed to mining indicating 1.8% of the national consump-tion was utilized for comminution. Considering both the energyand the environmental costs it is now more than ever imperativeto use energy efficiently.

Based on the amount of mechanical energy introduced (in theform of rock deformation) relative to the area of new fractures gen-erated, it is known that efficiency of utilization of energy in com-minution is low. A comprehensive review by Fuerstenau andAbouzeid (2002), demonstrated that efficiency of energy utilizationduring rock crushing is in the range 0.1–1%.

More recently Tromans and Meech (2002, 2004) estimated thefracture surface energies of minerals based on ionic and covalentmodels of bonding and fracture and compared this with the energyrequired to generate a new fracture surface using the Bond WorkIndex. They concluded that the energy efficiency is only of theorder of <1–2%. Even when, as a reference, an empirical parameter

Table 1Stages of crack development in granite.

Property Stress threshold (MPa)

Crack closure 47.5 ± 2.9Crack initiation 81.5 ± 3.7Crack coalescence 104.0 ± 3.8Crack damage 157.3 ± 9.9Peak strength 206.5 ± 10.0

1238 N. Djordjevic / Minerals Engineering 23 (2010) 1237–1244

is used, such as energy consumed during single particle impactbreakage, the efficiency of SAG mill is in the order of about 15%(Fuerstenau and Abouzeid (2002)). What is not entirely clear ishow it is possible that such an important process, with a very longhistory of improvements, such as rock size reduction is so ineffi-cient; i.e. what kind of process consumes the remaining �99% ofthe introduced energy.

The accumulation in thermal energy of rock was investigated,using an advanced thermal imaging camera, to quantify how muchenergy is lost in the form of accumulated thermal energy of rock,and to explore potential relationships between this heat accumula-tion, the physical and/or mineralogical characteristics of the rock,the produced fragment size distribution and operational parame-ters of the comminution process. Ultimately it is hoped that a bet-ter understanding of heat losses, will generate options for themodification of operational and equipment parameters of commi-nution to reduce heat losses and improve the efficiency of rockcrushing. Several rock and ore types were used in testing: basalt,two types of copper ore (from North and South America) and nickelore from Australia. In order to be able to design more efficient com-minution process it is essential firstly to quantify modes of energyconsumption that occur during rock size reduction.

2. Rock breakage process and thermo-elasticity

When rock is exposed to a compressive load, micro-cracks startto propagate when the axial stress reaches about 40% of the ulti-mate failure stress (Peterson, 1978). In the case of uni-axial radialcompression, without lateral confinement, only a few cracks willpropagate. However, in the case of loading with some compressiveconfining stress, the confining stress prevents the large extensionof macro cracks, but failure occurs by the interaction of cracks toproduce macroscopic shear failure. The general consensus has beenthat the failure process can be broken down into a number ofstages, based on the stress–strain characteristics observed duringcompressive laboratory testing of rock samples. Brace and Bombo-lakis (1963) defined these stages as being: crack closure; linearelastic deformations; crack initiation and stable crack growth; crit-ical energy release and unstable crack growth; failure and post-peak behaviour.

An important feature from the point of view of energy utiliza-tion is crack initiation stress. At that point of loading, rock startsto suffer irreversible damage. Crack initiation represents the stresslevel where micro-fracturing begins. Martin and Durham (1975)notes the peak strength of granite is not a unique material propertybut is dependent on the loading rate. However, the crack initiationand crack damage stresses were found to be essentially indepen-dent of the loading rate. Results presented by Eberhardt et al.(1999) for granite show that crack initiation stress occurs at about40% of the peak strength, Table 1.

In the context of fracture mechanics failure due to compressiveload comes as a consequence of activation, propagation and coales-cence of numerous micro-flaws within the material. According tothe modified Griffith theory for the case of closed cracks the shearstress required for micro-crack initiation is proportional to thecoefficient of friction. However, in the case of tensile loading, thestrength of the material is described with parameters, which donot include friction.

This indicates that activation of material friction is one of thekey factors that contribute to the low energy efficiency of rock sizereduction. The best way to avoid frictional losses during size reduc-tion is to create conditions conducive to the tensile failure of rocks.However it is difficult to create machines that will induce tensilestress in the rock consistently and on the scale applicable forindustrial use. The alternative is to create rock which, when loaded

in a conventional compressive mode, will fail in tension. This willoccur only if rock particles are of an approximately spherical shape.The efficiency of the generation of tensile failure can be further in-creased through manipulation of the size distribution of particlesinvolved in the breakage process.

The theory of thermo-elasticity reconciles energy utilised in thedeformation of rock, up to the point of crack initiation (loading andunloading stress–strain curves do not overlap even in nominallyelastic range of deformation). The thermo-elastic stress analysisis based on the principle, that the temperature of a solid changeswhen the stress state changes. As long as adiabatic conditionsare maintained and the body is loaded within the elastic range,temperature change is proportional to the change of the sum ofthe principal stresses. Although the discovery of the effect wasmade in 1830 (W. Weber) and theoretically explained in 1853(Lord Kelvin) it was not possible until relatively recent times, tomeasure small changes in the temperature of the solid objectand use the thermo-elastic effect to indicate state of stress. Solvingthe differential equation of thermo-elasticity produces a solutionin the form:

DT ¼ T0 aDrqCp ð1Þ

where DT is the temperature change, T0 is the absolute (initial) tem-perature of the body, Dr is the stress change, a thermal expansioncoefficient, Cp is the specific heat capacity, q density of the material.

The previous equation shows that for elastic, isotropic materi-als, under adiabatic conditions there is a proportionality betweenchange of stress and change of temperature of the material. Theheat generated from the thermo-elastic effect is small – giving atemperature increase of 0.2 �C for mild steel just below its yieldingpoint (Mason et al., 1994). In general the process of elastic rockdeformation, without fracturing or damage, will consume energy,however in the context of total energy delivered, the amount ofsuch energy will be negligible (i.e. much less than 1%). For instanceif we assume that failure of perfectly brittle material occurs atstress level of 100 MPa, with an average value of thermalexpansion coefficient (30 � 10�6 1/K), specific thermal capacityof 800 J/kg K and density of 2500 kg/m3, assuming an initial tem-perature of 20 �C, the increase of temperature due to an increaseof stress by 100 MPa, will be 0.03 K, which corresponds to a specificenergy of Q = 24 J/kg�0.007kw h/t.

In the context of fracture mechanics, failure due to compressiveload comes as a consequence of the activation and propagation ofnumerous micro-flaws within the material. According to the modi-fied Griffith theory in the case of closed cracks, shear stress requiredfor micro-crack initiation is proportional to the coefficient of friction.As a result of friction induced energy loss, compressive strength ofrock is approximately an order of magnitude higher than tensile rockstrength. In the case of inclined micro-flaw within solid rock, evenunder conditions of externally applied tensile stress, due to stresstransformation along inclined plane, there will be localised shearstress developments at the tip of such micro-cracks. Existence ofshear stress and friction will result in creation of heat.

According to Martin and Durham (1975), after initial yielding ofthe rock matrix due to loss of cohesion, the residual strength of therock is primarily controlled by friction. It is well known that post-

N. Djordjevic / Minerals Engineering 23 (2010) 1237–1244 1239

peak strength deformation in metallic and polymeric materials ismostly dissipated as heat. For metal it is commonly assumed that85–100% of plastic deformation results in heat. Detail experimentsperformed in the field of earthquake mechanics, demonstrated thatthe intensity of heat generated during friction results in localised,transient melting which effectively, temporarily, lubricates theearthquake focal region, reducing the coefficient of friction andincreasing velocity of slipping along the fault plane (Di Toro etal., 2004).

In the case of fractured rock, the bonds between mineral grainsare disrupted; cohesion along the fracture plane becomes zero, soshear strength becomes proportional to the normal stress acting onthe plane of fracture and parameters equivalent to the coefficientof friction. So with an increase of shear strength, due to the highernormal stress component acting perpendicular to crack surface,higher tangential shear stress is required to induce sliding alongthe fracture plane, and final disintegration of the rock. During slid-ing along the fracture plane the time rate of heat generation perunit of fracture area (Q) is taken as the product of shear stressand velocity of slip:

Q ¼ sV ð2Þ

This indicates that activation of material friction and shearstress acting along the fractured surface are the key factors thatcontribute heat generation during rock fragmentation. The solutionof the transient heat diffusion problem for the half space heatedfrom the external boundary by constant heat flux is given by Cars-lew and Jaeger (1959). The temperature on the surface is given byequation:

T ¼ 2Qðt=p/kcÞ0:5 ð3Þ

where Q is heat flux (J/s m2), T is the temperature (K), c is the spe-cific heat of rock (J/kg K), t is the time from beginning of heating (s),k is the thermal conductivity (W/m K), q is the density (kg/m3).

Eqs. (2) and (3) are used to model the temperature rise whichwill occur during high velocity crushing under confined conditionsor blasting.

The results indicate that under conditions of sliding with highnormal stress, as in the case of a compressed bed within HPGR,there is a propensity for generation of relatively high temperatures.Such conditions may also exist during blasting, where relativelyhigh shear stresses will be generated as well as high sliding veloc-ities Fig. 1.

Surface temperature due to sliding along fractured rock surface as function of shear

stress and velocity (sliding time 1ms)

0

50

100

150

200

250

300

350

400

450

0 1 2 3 4 5sliding velocity (m/s)

tem

pera

ture

(C)

10MPa

5MPa

1MPa

Fig. 1. Modelled temperature along the sliding surface as a function of shear stressand sliding velocity.

3. Measurement thermal energy during rock breakage viainfrared imaging

Any object will emit electromagnetic energy due to its temper-ature, as long as the temperature of the object is above absolutezero. Thermal radiation as a form of electromagnetic radiationcan be transferred between two objects without the need for amedium of transport between them (as it is required for example,for acoustic energy). The thermal radiation emitted from the sur-face is not equally distributed over all wavelengths. When dealingwith thermal radiation from a real surface it is useful to use theconcept of ideal surface as a baseline (blackbody). The spectralemissive power of the ideal surface, for all temperatures and wave-lengths is described in Plank’s law. Using the Stefan–Boltzmannlaw it is possible to calculate the amount of energy emitted in alldirections and at all wavelengths at a given temperature. All ofthese laws are strictly valid in the case of an ideal emitter(blackbody).

Real materials are described as gray body, where the magnitudeof radiation from a surface is a function of both temperature aswell as surface properties. The surface property limiting the quan-tity of radiation is called the emissivity. Emissivity of real materialsis always less than unity, and for the rocks is typically in the range0.85–0.95. The quantity of energy leaving a surface as radiant heatis proportional to its emissivity and the fourth power of its abso-lute temperature given by equation:

Q ¼ reT4 ð4Þ

where Q is the hemispherical total emissive power (radiated powerper unit area, W/m2), r is the Stefan–Boltzmann constant(5.67051 � 10–8 W/m2 K), e is the total hemispherical emissivityof the surface, T is the surface absolute temperature (K).

From the measured radiation energy and emissivity constant ofthe surface, it is possible to calculate surface temperature of thematerials. Assuming that surface temperature represents averagetemperature of the materials, it is then possible to calculate ther-mal energy stored within materials, from the equation.

Q ¼ DT�m�Cp ð5Þ

where DT is the temperature difference (K), m is the mass of mate-rial (kg), Cp is the specific heat of material (J/Kg K).

Previous work (Wu et al., 2002, 2004) indicated that moderninfrared imaging technology may be suitable for the detection ofheat that is created during a process of rock deformation andbreakage. We performed a range of rock crushing experimentswhich were monitored using infrared camera (Model Silver420 M, produced by CEDIP/FLIR), operating in spectral range of1.5–5.1 lm. The main purpose of testing was to determine the sen-sitivity of the instruments to detect heat from the fracture surfaceimmediately after fracture creation. Tests were performed usingbasalt with introduced energy in the range of 0.25–1 kw h/t. Thetemperature was measured before and immediately (�5 s) afterbreakage.

The rock used is of a dark appearance and was fragmented inthe drop test machine using 25 kg of mass dropped from a heightof up to 1 m. From the infrared (IR) images we can notice that alongthe fresh fractures surface there is an increase in temperature, rel-ative to the rock surface area which was not fractured. Consideringthat both surfaces are of the same material (basalt), an error in as-sumed emissivity constant, does not influence the value of thetemperature difference between fractured and unfractured rock.The temperature was measured along the line, transecting theheated area, Figs. 2 and 3.

In the case of the low energy test the increase of temperaturealong the single fracture surface was �3 �C while in the case of

Fig. 2. IR image and temperature profile of fragments after 0.25 kw h/t drop weighttest (average temperature along fractured surface �34 �C).

Fig. 3. IR image and temperature profile of fragments after 1.0 kw h/t drop weighttest (average temperature along fractured surface �35.5 �C).

1240 N. Djordjevic / Minerals Engineering 23 (2010) 1237–1244

the high energy test the increase of temperature along the multiplefractured surfaces was about 6.7 �C. Increase intensity of the im-pact stress, increase temperature along the fractured surface. Con-sidering that in the case of the drop weight test the mostsignificant change occurs due to the increase of mass of weight,the resulting increase in shear stress (due to increased normalforce acting along the fracture plane), will result in a temperatureincrease along the fractured surface, as well as an increase of frac-ture area.

4. High Pressure Grinding Roll Testing

We performed measurements of thermal losses that occurredduring High Pressure Grinding Roll (HPGR) crushing of local basalt.In the case of HPGR rock is force-fed into the machine via a verticalhopper. Napier-Munn et al., 1996. Two counter-rotating rolls allowthe compression breakage to occur in a continuous fashion.

One of the rollers in the HPGR rotates on a fixed axis while theother is allowed to move linearly with a pressing force applied tothe moving roll. The moveable roller is forced up against the mate-rial in the gap between the rollers by a hydraulic oil cylinder sys-

tem. This oil pressure acts through four or two cylinders(depending on the manufacturer) and transmits the grinding forceover the cross-section of the diameter of the rolls where the bedhas formed (Napier-Munn et al., 1996; Daniel and Morrell, 2004).Several aspects of HPGR crushing were investigated, Fig. 4.

HPGR crushing was performed on six samples of equal mass(18.86 kg). The initial pressure of the HPGR was set to be 20, 30(for two samples), 40, 50 and 60 bar. Infrared imaging (IR) of thecrushed product was performed immediately after crushing, withfragments in free fall, Fig. 5. To verify results collected using IRtechnology, average temperatures were also measured using athermo-probe. Crushing products were sized and comparedagainst feed size distribution and introduced net comminutionenergy.

An increase in the specific comminution energy, results in finerfragmentation. However, there is tendency that after a certainamount of energy is introduced, a further increase in consumedspecific energy, produces only minor or insignificant improve-ments in fragmentation. Due to the nature of force application,high energy concentration on surfaces creates a very fine fragmentwhich acts as a buffer/filter, effectively reducing the amount of en-ergy that can be transferred deeper into the rock.

This is observed in the diagrams of fragment size distributions.If two different HPGR tests, in terms of net specific energy, producethe same fragment size distribution, the obvious question is whatkind of process consumes energy the increment that exists be-tween these two tests. This is clearly illustrated in Figs. 6 and 7,which show asymptotic behaviour of the fragment size distribu-tion as applied pressure increases, or amount of net energy con-sumed increases.

Fragment size distributions produced at a set pressure in therange of 4–6 MPa are practically the same (i.e. within the limitsof experimental error). However, the energy delivered into the rockvaries within a wide range. To calculate thermal energy lost, besiderock temperature it is necessary to know specific heat capacity.

Specific heat capacity, also known simply as specific heat is themeasure of the heat energy required to raise the temperature of aunit mass of a substance by 1 K (or �C). The specific heat of basaltpreviously being measured, using a simple calorimeter, was foundto be �800 J/kg K, Fig. 8. From the known specific heat and temper-ature, thermal energy lost can be calculated. The amount of ther-mal energy stored in the fragmented rock is compared with netenergy introduced for rock crushing, Fig. 9.

Based on the previous graph, efficiency of energy utilization wascalculated as a function of the HPGR pressure setting, Fig. 10.

Results obtained during HPGR testing of basalt clearly indicatethat there is an optimum intensity of pressure to which given rockneeds to be exposed. Any further increase in pressure results inonly a marginal increase in fragmentation and a significant in-crease in unproductive heating of rock. Results indicate that about20% of net comminution energy introduced could be potentiallysaved through pressure optimization.

From the hydraulic pressure and projected area of the rolls, spe-cific grinding pressure used in HPGR is calculated to be in the orderof 5–15 MPa. It is of interest to consider how such a relatively lowpressure is able to induce fracturing of rock which has a compres-sive strength, frequently in the range 100–200 MPa and what arethe implications in terms of efficiency of energy utilization. This as-pect of HPGR crushing was modelled using the discrete elementmethod.

In the case of breakage under conditions such as the one whichexists in the HPGR crushing zone, applied grinding pressure doesnot interact with uniform solid material. It interacts with a numberof discrete particles. Compressive force is introduced into the rockfragments through the area which is a relatively small fraction ofthe projected area of the rolls. The reduced area, results in

Fig. 4. Cross-sectional representation of HPGR (Napier-Munn et al., 1996).

Fig. 5. Setting used during HPGR testing.

Fig. 6. Feed size distribution and fragment size distributions after HPGR crushingunder different pressure.

Fig. 7. Percent of crushed materials passing 0.3 and 1.18 mm sieve, as function ofnet specific energy (kw h/t) introduced.

Fig. 8. Measured specific thermal capacity of rocks.

N. Djordjevic / Minerals Engineering 23 (2010) 1237–1244 1241

increased pressure, to a level which is sufficient to overcome thestrength of individual rock fragments. So the external pressure of10–15 MPa becomes, due to a focusing of force at contact points,enhanced to an intensity which is higher the compressive rockstrength (Djordjevic and Morrison, 2006).

Highest contact stresses, due to smallest contact area, are pos-sible for spherical particles. The initial stage of crushing, althoughcharacterised with high contact stresses is also the most energyefficient. Assuming minimal confinement, failure of the particlesoccurs as the result of the development of a tensile stress fieldwithin the particle. Later stages of crushing, due to the develop-ment of debris, result in increased contact area of the rolls and rock

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0 2 4 6 8initial pressure (MPa)

spec

ific

ener

gy (k

wh/

t)

Consumed net energy(kWh/t)

Thermal Energy Loss(kWh/t)

Fig. 9. Consumed net energy by the HPGR and amount of energy lost as heat asfunction of the HPGR initial hydraulic pressure.

0

10

20

30

40

50

60

2 3 4 5 6initial pressure (MPa)

net e

nerg

y ef

ficie

ncy

(%)

Fig. 10. Efficiency of energy utilization as function of HPGR hydraulic pressure.

1242 N. Djordjevic / Minerals Engineering 23 (2010) 1237–1244

and between rock fragments, requiring higher external pressure toinduce same the stress within rock materials. A large part of addi-tional energy is consumed in shearing, sliding and re-crushing ofprogeny fragments. During this phase of rock disintegration a sig-nificant amount of thermal energy will be created.

Fig. 11. Wide feed size distribution – improvements in fragmentation.

Fig. 12. Narrow feed size distribution – improvements in fragmentation.

5. Effect of feed size distribution

From a practical operational point of view, the question is whatcan be done with existing equipment to reduce such thermallosses. Besides reducing the intensity of applied pressure we werealso interested in potential improvement of the efficiency of crush-ing that can be achieved through modification of the feed size dis-tribution, while keeping the same initial set pressure.

We analysed the response of low-grade copper ore originatingfrom the Bingham Canyon Mine operated by Rio Tinto’s KennecottCopper. Twelve tests were performed at different initial pressure(1–4 MPa). Two types of samples were considered: one is charac-terised with continuous fragment size distribution with a top sizeof about 9 mm and mean size of about 5 mm, while the second setof samples was characterised with fragment size distributionswhich started at about 3 mm with a top size of about 7 mm, while50% passing size for both sample being identical. The mass of sam-ples was 12 kg and 17.5 kg, for narrow and wide size distributionrespectively. From recording parameters of the HPGR, the net spe-cific comminution energy was calculated and correlated with ob-served temperatures and thermal energy.

Wide feed distribution, with a relatively large amount of finematerial, results in strong heating of the product and the creation

of very large flakes (compressed agglomerates composed fromcrushed rock fragments). Feed with reduced amount of fines, re-sulted in the creation of much cooler flakes, with the size of flakesbeing significantly smaller than in the case of feed with a widefragment size distribution. Based on the measured density and spe-cific thermal capacity of rocks, the amount of energy consumed toheat the rock was calculated and compared with consumed energy.

Results show that crushing using narrow fragment size distri-bution consistently converts a smaller fraction of introduced en-ergy into heat, compared with the same crushing setup using thesame type of rock, but with a wide feed fragment size distribution.This indicates that a simple measure such as bypassing (removing)fines from the feed may result in improved performance of theHPGR. Results of IR imaging of the crushed rock, show that a largefraction of net energy supplied to rock is transformed into heat,raising temperature of the rock, Fig. 11.

Results show that a more narrow fragment size distribution(without fines in feed) but with the same mean size, appears tobe conducive to more efficient crushing. This can be observed fromthe graph which shows the difference in cumulative product andfeed percent passing as a function of size, Figs. 11 and 12.

The fraction of new fine material produced during crushing ismuch higher in the case of feed with a narrow size distribution.Significantly this improved performance is achieved with a re-duced amount of net comminution energy. Narrow feed size distri-bution consistently is more efficient i.e., it produces more newfines for the same amount of specific energy introduced.

N. Djordjevic / Minerals Engineering 23 (2010) 1237–1244 1243

So if the objective is to create more fine particle (with size smal-ler than 0.85 mm), narrow fragment size distribution of the feedwill result in more energy efficient crushing. From the increasein average temperature of crushed rock relative to uncrushed ref-erence rock, specific thermal capacity of rock as well as rock mass,the amount of thermal energy can be calculated. Calculated spe-cific thermal energy was then compared with net energy trans-ferred to the rock via HPGR, Fig. 13. From these two values,efficiency of utilization of introduced net energy was calculated.

The graph indicates that on average the crushing of a wide feedfragment size distribution consumes about 15–20% more energythan crushing of rock with a narrow feed size distribution, whilecreating fewer fines. The observed relative inefficiency of wide feedsize distribution is probably due to the large amount of pre-exist-ing fines, which clog the pore space between coarser fragments. Inthe compressed zone of the HPGR this creates (approximately),hydrostatic compressive loading conditions, which require higherpressure (i.e. energy). This conclusion can be supported by themeasured time history of pressure for two cases, where the nomi-nal initial hydraulic pressure was the same (3 MPa), Fig. 14. Feedcharacterised with a relatively narrow size distribution (i.e., with-out large amount of very fine material) is prone to crushing at amuch lower pressure than a similar feed, but including a relativelylarge amount of fines.

The previous results were obtained without changes of theHPGR gap settings. Purely due to the nature of the compressedbed, and lack of pre-existing fines, required pressure and grindingforce are radically lower. Considering that the genuine crushing of

Fig. 13. Thermal energy as fraction of net energy introduced for the case of copperore with narrow (N) or wide (W) feed size distribution.

0

2

4

6

8

10

12

0 20 40 60 80time (sec)

pres

sure

(MPa

)

wide distribution

narrow distribution

Fig. 14. Time history of applied pressure (initial set oil pressure 3 MPa).

rock matrix is possible only due to a small contact area of rock ma-trix and rolls, it is even more important to maintain that the rela-tively high contact pressure, is above the compressive strength ofrock.

As the results of DEM modelling of a piston and die test demon-strated, in the case of a compressed bed being composed of a mixof fine and coarse material, the size of contact area of rock and rollswill increase, reducing contact pressure below compressivestrength of rock (Djordjevic and Morrison, 2006). Applied energywill be then mainly transformed into heat. This is reason why feedcharacterised with a large fraction of fines, will not result in effi-cient crushing, due to the inability to amplify stresses at contactpoints.

The shape of particles and size distribution of particles within acompressed bed, determines the number of inter-particle contact.In the case of narrow feed size distribution, without fines, size dis-tribution is narrower, and therefore the number of contacts perparticle (average coordination number) are smaller. In the case offull size distribution, including fines, the number of contacts perparticle is larger. Due to the lower confinement, the strain experi-enced by the particle in the case of narrow size distribution, mustbe much higher, relative to the wide feed size distribution, for thesame specific grinding pressure. Larger strain will result in moreefficient fragmentation of the rock characterised with narrow feedsize distribution. So to achieve the desired level of strain of the par-ticle bed, in the case of wide feed size distribution, higher pressureis required.

6. Discussion and conclusions

We were able to perform reproducible measurements of thetemperature increase that occurs during highly transient eventssuch as dynamic rock breakage and High Pressure Grinding Roll(HPGR) crushing. Results obtained show that with the increase ofenergy introduced, there is an increase in the maximum tempera-ture along the fractured surface as well as increases in the overallamount of thermal energy.

Results obtained during HPGR testing clearly indicate that thereis an optimum intensity of pressure to which given rock needs tobe exposed. Any further increase in pressure results in only a mar-ginal increase in fragmentation and a significant increase in unpro-ductive heating of rock. Due to the nature of rock fracturing, evenunder optimal conditions, such as pure tensile fracturing, some en-ergy losses in the form of thermal energy are unavoidable.

The fraction of new fine material produced during HPGRcrushing is much higher in the case of feed with a narrow size dis-tribution, without fines. Significantly, improved performance isachieved with reduced amount of net comminution energy. Theobserved relative inefficiency of wide feed fragment size distribu-tion (fines and coarse particles), in our opinion is due to the largeamount of pre-exiting small particles (fines), which clog the spacebetween coarser fragments. In the compressed zone of the HPGRthis creates approximately hydrostatic compressive loading condi-tions, which require much higher pressure (i.e. energy) to causebreakage.

Feed characterised with a relatively narrow fragment size dis-tribution (i.e., without a large amount of very fine material) isprone to more efficient crushing at a significantly lower pressurethan feed with the same top size, but including a relatively largeamount of fines. Rock friction is an efficient mechanism for thegeneration of heat. During compressive loading, which results inshear rock failure, friction is the mechanism which at a micro-levelis responsible for the creation of a tensile stress field around themicro-flaws, under conditions of macroscopic compression. There-fore, micro-crack initiation and growth are characterised with high

1244 N. Djordjevic / Minerals Engineering 23 (2010) 1237–1244

frictional forces and heat generation. In the larger scale, frictionalong cracked surfaces in the post-peak-strength region of the rockfailure process is probably the source of a major part of thermal en-ergy released during rock size reduction process.

This suggests ways to minimise losses of energy due to transfor-mation into heat. If the failure mode which activates strong frictionis minimised then thermal losses will also be minimised. In partic-ular post-failure friction is a strong source of thermal energy. Thiscan be achieved when failure occurs not in the form of shear failurebut as tensile failure. In the case of macro-tensile failure, the ten-sile strength of the material is not influenced by the coefficientof friction, therefore, under favourable conditions; a much largerfraction of energy introduced will be consumed in the creation ofa new fracture surface. Experimental results indicate that up to40% of energy can be saved through optimization of the appliedpressure and modification of feed fragment size distribution.

Acknowledgements

This work was carried out under the auspice of the Centre forSustainable Resource Processing, which is established and sup-ported under the Australian Government’s Cooperative ResearchCentres Program.

References

Brace, W.F., Bombolakis, E.G., 1963. A note on brittle crack growth in compression. J.Geophys. Res. 68, 3709–3713.

Carslew, H.S., Jaeger, J.C., 1959. Conduction of Heat in Solids, second ed. ClarendonPress, Oxford.

Daniel, M.J., Morrell, S., 2004. HPGR model verification and scale-up. Miner. Eng. 17(11–12), 1149–1161 (Communition ’04).

DOE, 2004. Profile of Total Energy Use for US Industry, Industrial TechnologiesProgram. US Department of Energy, Energy Efficiency and Renewable Energy,December.

DOE, 2005. Mining Industry of the Future Fiscal Year 2004 Annual Report, IndustrialTechnologies Program, US Department of Energy, Energy Efficiency andRenewable Energy, February.

Djordjevic, N., Morrison, R., 2006. Exploratory modelling of grinding pressurewithin a compressed particle bed. Miner. Eng. 19 (10), 995–1004.

Di Toro, G., Goldsby, D.L., Tullis, T.E., 2004. Friction falls towards zero in quartz rockas slip velocity approaches seismic rates. Nature 427, 436–439.

EIA, 2005. Energy Information Administration, International, South Africa. <http://www.eia.doe.gov/emeu/international/safrica.html>.

Eberhardt, E., Stimpson, B., Stead, D., 1999. Effects of grain size on the initiation andpropagation thresholds of stress-induced brittle fractures. Rock Mech. Rock Eng.32 (2), 81–99.

Fuerstenau, D.W., Abouzeid, A.-Z.M., 2002. The energy efficiency of ball milling incomminution. Int. J. Miner. Process. 67, 161–185.

Mason, J.J., Rosakis, A.J., Ravichandran, G., 1994. On the strain and strain ratedependence of the fraction of plastic work converted into heat: an experimentalstudy using high-speed infrared detectors and the Kolsky bar. Mech. Mater. 17,135–145.

Martin, R.J., Durham, W.B., 1975. Mechanisms of crack growth in quartz. J. Geophys.Res. 80, 4837–4844.

Napier-Munn, T.J., Morrell, S., Morrison, R.D., Kojovic, T., 1996. MineralComminution Circuits – Their Operation and Optimisation. Julius KruttschnittMineral Research Centre, Monograph, vol. 2. The University of Queensland,Brisbane, Australia.

RC, N., 2004. Improving Energy Performance in Canada – Report to Parliamentunder the Energy Efficiency Act, Energy Publications, Office of Energy Efficiency.Natural Resources Canada, Ottawa, ON. p. 31.

NRC, 2006. Energy Efficiency Trends Analysis Tables: Total End-Use Sector – EnergyUse Analysis. Office of Energy Efficiency, Natural Resources Canada. <http://www.oee.nrcan.gc.ca/corporate/statistics>.

Peterson, M.S., 1978. Experimental Rock Deformation – The Brittle Field. Springer-Verlag, Berlin.

SOE, 2006. Indicator: HS-33 Energy Use by Sector, State of the Environment,Australian Government, Department of the Environment and Water Resources.<http://www.environment.gov.au/soe/2006/publications>.

Tromans, D., Meech, J.A., 2004. Fracture toughness and surface energies of covalentminerals: theoretical estimates. Miner. Eng. 17, 1–15.

Tromans, D., 2008. Mineral comminution: energy efficiency considerations. Miner.Eng. 21 (8), 613–620.

Tromans, D., Meech, J.A., 2002. Fracture toughness and surface energies of minerals:estimates for oxides, sulphides, silicates and halides. Miner. Eng. 15, 1027–1041.

Wu, L., Liu, S., Wu, Y., Wu, H., 2002. Changes in infrared radiation with rockdeformation. Int. J. Rock Mech. Min. Sci. 39 (6), 825–831.

Wu, L., Wu, Y., Liu, S., Li, G., Li, Y., 2004. Infrared radiation of rock impacted at lowvelocity. Int. J. Rock Mech. Min. Sci. 41 (2), 321–327.