2008:03

TECHNICAL REPORT

Conference in Minerals Engineering

5 6 February 2008 in Lule, Sweden

Lule University of Technology Technical Report Department of Chemical Engineering and Geosciences Division of Mineral Processing
2008:03 - ISSN: 1402-1536 - ISRN: LTU-TR--08/03--SE

PROGRAM
Konferens i Mineralteknik Conference in Minerals Engineering
5 6 februari 2008 i Lule

AVDELNINGEN FR MINERALTEKNIK DIVISION OF MINERAL PROCESSING

Freningen Mineralteknisk Forskning Stockholm

MinFo

Tisdag 5 februari
10.00 10.50 10.50 11.40 Registrering och kaffe Pr Weihed, Lule tekniska universitet Hlsar er vlkomna Plsson, Bertil, Lule tekniska universitet, Mineralteknik Aktuellt om utbildning och forskning Ordf. Pr Weihed Processteknisk utveckling och energibesparing i mineralindustrin Exempel frn MinBaS II och MinFo:s Energiprojektpaket Thomaeus, Marianne, freningen Mineralteknisk Forskning, MinFo Natural refinement of Solar Grade Silicon Raw Materials Larsen, Rune, NTNU, Norge Lunch p Olivers (tvrs ver gatan frn Stadshuset)

Session 1 11.40 12.10

12.10 12.40 12.45 13.45

Session 2 Ordf. Rolf Arne Kleiv 13.45 14.15 Character of gold within the Aitik ore body: Preliminary results from a geometallurgical study Wanhainen, Christina, Lule tekniska universitet & Johansson, Bjrn, Boliden Mineral AB 14.15 14.45 A characterising of the ore minerals due to mineralogical, chemical and textural properties in Malmberget Lund, Cecilia & Martinsson, Olof, Lule tekniska universitet Traceability by multivariate analysis on morphology data from grinding circuits Oghazi, Pejman & Plsson, Bertil, Lule tekniska universitet Paus

14.45 15.15 15.15-15.45

Session 3 Ordf. Bertil Plsson 15.45 16.15 Modeling and control of a zinc flotation circuit Lindvall, Henrik, Boliden Mineral AB & Pettersson, Jens, ABB Corporate Research 16.15 16.45 On Lifecycle Costs of Flotation Rinne, Antti, & Peltola, Aleksi, Outotec, Finland (presented by Harri Hokka, Outotec). Anvndning av processimulering fr att identifiera flaskhalsar i malmbehandlingssystem Alldn berg, Eva, LKAB & Plsson, Bertil, Lule tekniska universitet Middag p restaurang Lule, Kulturens Hus

16.45 17.15 1900

Onsdag 6 februari
Session 4 Ordf. Marianne Thomaeus 09.10 09.40 Systems aspects on in-pit crushing and coarse-particle slurry pumping Sellgren, Anders, Lule tekniska universitet 09.4010.10 10.10 10.40 Ciba RHEOMAXTM Density and Rheology Technology Flanagan, Ian, Ciba Specialty Chemicals, UK Rheological and depositional characterization of paste-like tailings slurries Wennberg, Thord, LKAB; Sellgren, Anders Lule tekniska universitet & Goldkuhl, Ingemar, MultiServe AB

10.40 11.10 Paus Session 5 11.10 -11.40 11.40 12.10 12.10-12.40

Ordf. Anders Sellgren Novel Fluidised-bed Tribo-electrostatic Separator for Dry Coal Preparation Dwari, Ranjan & Kota, Hanumantha Rao, Lule tekniska universitet A Predictive Chemistry Model for Cement Process Wilhelmsson-Hkfors, Bodil, Ume Universitet Surface thermodynamics and extended DLVO theory of Acidithiobacillus ferrooxidans and Leptospirillum ferrooxidans cells adhesion on pyrite and chalcopyrite Vilinska, Annamaria & Kota, Hanumantha Rao, Lule tekniska universitet Lunch p Olivers

12.45 13.45

Session 7 Ordf. Eric Forssberg 13.45 14.15 Optical sorting employing different analyzing configurations Kolacz, Jacek, Comex AS, Norge 14.15 14.45 14.45 15.07 Research needs in mineral processing a few directions and priorities Kota, Hanumantha Rao, Lule tekniska universitet Avslut/Closing Ordinarie flygbuss till Lule Kallax Airport frn hllplatsen vid Stadshuset

INNEHLLSFRTECKNING/CONTENTS
Welcome/Vlkommen Pr Weihed, Lule University of Technology

Technical Papers

Sida/Page

Anvndning av processimulering fr att identifiera flaskhalsar i malmbehandlingssystem. Alldn berg, Eva, LKAB & Plsson, Bertil, Lule University of Technology .................... 1 Novel Fluidised-bed Tribo-electrostatic Separator for Dry Coal Preparation. Dwari, Ranjan & Hanumantha Rao, Kota, Lule University of Technology..................... 15 Ciba RHEOMAXTM Density and Rheology Technology. Flanagan, Ian & Stocks, Paul, Ciba Specialty Chemicals .................................................... 31 Optical sorting employing different analyzing configurations. Kolacz, Jacek, Comex AS................................................................................................. 39 Natural refinement of Solar Grade Silicon Raw Materials. Larsen, Rune B., Srensen, Bjrn E. & Poitrasson, Franck, Norwegian University of Science and Technology & Ihlen, Peter M., Geological Survey of Norway ................................... 49 Modeling and control of a zinc flotation circuit. Lindvall, Henrik, Boliden Mineral AB & Pettersson, Jens, ABB Corporate Research ........ 59 A characterising of the ore minerals due to mineralogical, chemical and textural properties in Malmberget. Lund, Cecilia & Martinsson, Olof, Lule University of Technology.................................. 71 Traceability by multivariate analysis on morphology data from grinding circuits. Oghazi, Pejman & Plsson, Bertil, Lule University of Technology, Tano, Kent, LKAB & Bjrn Kvarnstrm, Lule University of Technology.......................................................... 81 Mineralteknik Aktuellt om utbildning och forskning. Plsson, Bertil, Lule University of Technology................................................................ 97 On Lifecycle Costs of Flotation. Rinne, Antti & Peltola, Aleksi, Outotec (presented by Harri Hokka, Outotec) ............... 107 Systems aspects on in-pit crushing and coarse-particle slurry pumping. Sellgren, Anders, Lule University of Technology .......................................................... 119 Processteknisk utveckling och energibesparing i mineralindustrin - Exempel frn MinBaS II och MinFo:s Energiprojektpaket. Thomaeus, Marianne, MinFo ......................................................................................... 127

II Character of gold within the Aitik ore body: Preliminary results from a geometallurgical study. Wanhainen, Christina, Lule University of Technology & Johansson, Bjrn, Boliden Mineral AB ........................................................................... 143 Rheological and depositional characterization of paste-like tailings slurries. Wennberg, Thord, LKAB, Sellgren, Anders, Lule University of Technology & Goldkuhl, Ingemar, MultiServ AB.................................................................................. 151 A Predictive Chemistry Model for Cement Process. Wilhelmsson Hkfors, Bodil, Ume Universitet, Viggh, Erik O., Cementa AB & Backman, Rainer, Ume Universitet .............................................................................. 161 Surface thermodynamics and extended DLVO theory of Acidithiobacillus ferrooxidans and Leptospirillum ferrooxidans cells adhesion on pyrite and chalcopyrite. Vilinska, Annamaria & Hanumantha Rao, Kota, Lule University of Technology........... 179 Research needs in mineral processing a few directions and priorities. Hanumantha Rao, Kota,Lule University of Technology ............................................... 195 - O - O - O -

Welcome!
Conference in Minerals Engineering
5-6 February 2008 Lule City Hall Division of Mineral Processing Department of Chemical Engineering and Geosciences

From Hard Rock To Heavy Metal

The hard rock-heavy metal chain at LTU

Find and characterize mineralizations and ores

Construction, extraction and maintenance

Process and refine minerals and metals

Remediation, environemental aspects of mining Applied geology Soil Mechanics and Foundation Engineering

Applied Geophysics Ore geology

Rock mechanics Rock engineering

Mineral processing Process metallurgy Chemistry

Operation and Maintenance Engineering Industrial production environment Sound and Vibration Fluid mechanics

The hard rock-heavy metal chain at LTU

Focus area Mining and Metallurgy Includes 4 departments and over 12 research divisions > 15 professors c. 50 tenured staff > 70 PhD students 610 dissertations/year Budget >100 MSEK/year

Reserach subjects
Applied geophysics Applied geology Ore geology Rock mechanics & rock engineering Soil mechanics and foundation engineering Mineral processing Process metallurgy Chemistry Operation and maintenance engineering Industrial production environment Sound and vibrations Fluid mechanics

Ambition
We want to be the most complete European mining university with specialization towards crystalline rocks Complete research environment from explorationextraction-processing-refining-remediation Strengths
Research Modern and applied education with theoretical depths

Vinnovas mining research programme

100 MSek 2007-2011

Research areas
Secured mineral resources through exploration Increased competitivness through developing production technology Particle technology in mining related processes Resource efficient extraction of base metals Reduced environmental footprint from mining

HLRC research centre

100 MSEK

Research identified by the need of LKAB

Sustainable mining Smart pellets (New products)

10

Centres of excellence
ARC MiMeR Swebrec JVTC (Romeo, CAOS)

11

Department of chemical engineering and geosciences

A leading educational and research partner for mining and exploration industry

12

Leading centre for research and education related to sustainable use of natural resources
Chemistry Geosciences Chemical engineering Mineral processing Metallurgy
http://www.ltu.se/tkg/

13

Staff
15 professors 25 senior staff (lecturers, researchers etc) 40 PhD-students 9 administration+technicians 4 adjunct professor

14

7 divisions
Biochemical and Chemical Engineering Chemical Technology Chemistry Ore Geology & Applied Geophysics Mineral Processing Process Metallurgy Applied geology

15

Division of Chemical Technology

specialisation towards process chemistry within the Chemical engineering design programme.

Material development, particularly synthetic particles of inorganic crystals and structured materials. Preparation, modification and characterisation of small particles and porous structures as well as the use of these materials in various applications. Application areas of interest are, among others, membranes, catalysts, sensors, adsorbents and model systems.

16

Division of Chemistry
fundamental and applied studies on various mineral surfaces and liquid/solid interfaces. Infrared (IR) and RAMAN spectroscopy, solid and liquid state nuclear magnetic resonance (NMR), potentiometric titrations. Examples of minerals are iron oxides and poly-metal sulphides in combination with clay minerals. Studies of active surface sites, adsorption/desorption, speciation and coordination of various ions and molecules of collectors and polymers on mineral surfaces Used in industrial processes such as froth flotation, agglomeration, flocculation as well as in environmental research.

17

Div. of Ore geology & appl. geophysics

under graduate and graduate courses in hard rock geology such as mineralogy petrology structural geology, and economic geology, geophysical and geodesy. The focus of research concerns gold deposits, massive sulphide deposits, Fe-deposits and Fe-ox-CuAu. Geophysical research includes : ore- and water prospecting, construction-, and enviromental applications, paleomagnetism. New fields developes are processmineralogy/ore characterization 4D geolocial modelling of mining belts

18

Division of Mineral processing

Teaching mineral processing within the engineering programmes in chemical engineering and natural resources technology

Electrostatic beneficiation of Indian thermal coal Utilisation of mechanical activation, Ultra-fine grinding of quartz, Traceability of raw materials in process streams Flow and treatment of thick pulps among other things.

19

Division of Process Metallurgy

education within the disciplines of mineral processing and metallurgy. Research within Pyrometallurgy and Hydrometallurgy Subsolidus High Temperature Chemistry Slag Chemistry Leaching operations Thermodynamic modelling and CFD and kinetic modelling Mineralogical and chemical characterization of materials Biotechnical metal extraction Recycling and treatment of scrap and residues, reduction metallurgy as well as oxidation metallurgy during sintering.

20

Division of Applied Geology

Teaching in geology, geochemistry, ecology in several engineering programmes. Responsibility for biology and ecology classes in the teaching programme

Environmental Geochemistry, remediation Analytical Geochemistry, Hydrology and Ecology. Landscape Geochemistry. Basic research includes isotope fractionation of a variety of stable, relatively heavy isotopes of elements such as Fe, Si, Cu, Mo etc.

21

Undergraduate and graduate education Arena Earth Resources 3 or 5 years Chemical Engineering 3 or 5 years Natural Resources Engineering 3 or 5 years Masters programmes in english
- Chemical and Biochemical Engineering - Exploration and Environmental Geosciences - Minerals and Metallurgical Engineering

Teacher Education Freestanding cources Gemstone Technology 2 years

22

Welcome!

Anvndning av processimulering fr att identifiera flaskhalsar i malmbehandlingssystem Alldn berg, Eva 1 & Plsson, Bertil I. 2 LKAB, SE-983 81 Malmberget 1, Lule tekniska universitet, Avdelningen fr mineralteknik, SE-971 87 Lule 2

SAMMANFATTING LKAB beslt 2004 att starta en grundlggande studie som underlag fr framtida expansion av anlggningarna i Malmberget. En del processer och produktionslinjer riskerade att bli flaskhalsar i framtiden. I den hr rapporterade studien r det sovringsverket som underskts och resultatet visade att flera enhetsoperationer kommer att vara begrnsande med nuvarande maskinpark. Drfr var ett syfte med studien att underska hur byte av maskiner skulle pverka kapacitet och drift samt om de framtida produktionsplanerna kunde innehllas. Sovringsprocessen har genomgende simulerats med Modsim fr ett antal olika fall, som har skilt sig t i jrnhalt och partikelstorleksfrdelningar hos ingende malm. Modelleringsparametrarna har hmtats, till en del frn existerande underskningar i pilotskala, men i strsta mjliga utstrckning frn mtningar i existerande kretsar. Simuleringarna visade att installation av nya siktar skulle ge en vsentligt frbttrad drift, men ocks att nya krossar skulle vara en frbttring eftersom de skulle kunna kras fullmatade med snvare ansttning, vilket skulle ge processtekniska frdelar. Under 2007 har de flesta av de freslagna tgrderna genomfrts och resultatet visar att flaskhalsarna r borta och att halter och massbalanser r till och med bttre n vad som frutsgs i simuleringarna. 1. Inledning LKAB beslt 2004 att starta en grundlggande studie som underlag fr framtida expansion av anlggningarna i Malmberget. Resultat rrande anrikningsverket har redovisats tidigare (Tano, et.al., 2006a, 2006b). I den hr rapporterade studien r det sovringsverket som underskts och resultaten visar att flera delprocesser och produktionslinjer riskerade att bli begrnsande med nuvarande maskinpark. Drfr var ett syfte med studien att underska hur byte av maskiner skulle pverka kapacitet och drift samt om de framtida produktionsplanerna skulle kunna innehllas. Underskningarna har alla fljt ett gemensamt mnster: ett ftal tester i pilotskala eller som planerade frsk i fullstor skala, bygge och kalibrering av simuleringsmodeller fr enstaka enhetsoperationer eller processavsnitt, validering av simuleringar mot knda fall, storskalig simulering med varierande indata. Drivande fr den hr ansatsen har varit ndvndigheten av att minska 1

tiden mellan id och verklighet, underska potentialen i befintliga anlggningar samt att snabbt underska/utesluta alternativa layouter och krstt. Ett givet ml r d att fr sovringsverket (Fig. 1) kunna gra flaskhalsanalyser, som visar hur begrnsningar ndras med varierande ingende material. 2. 2.1. Sovringsverket Processen Sovringsverket i Malmberget behandlar malm frn bde vstra fltet, uppfordrat via Alliansens schakt, och frn stra fltet via Vitfors schakt som mynnar i den s kallade Lappkyrkan. Simuleringarna grs endast fr magnetitmalm, vilket innebr att Alliansens hematitmalmer och deras sovring inte tas med. Magnetitmalm frn Alliansen samt H-PAR, tas in i det simulerade fldesschemat som externa flden.

Figur 1.

Frenklat fldesschema ver sovringsverket fre 2006.

Processen simuleras med fyra ingenden: I1. magnetitmalm frn Lappkyrkan som i dag innehller en del grberg, I2. en separat strm med tillrednings- och anlggningsberg eftersom dess inflytande skall underskas, I3. magnetitmalm frn Alliansen (Vstra fltet), I4. H-PAR r en magnetit-fines som skiljs ut i hematitsovringen. Den har tre olika avfall: A1. grbergsavfall (30 200 mm) som sls ut i GRAM,

A2. A3.

mullavfall (0 4 mm) som kommer frn omseparering av avfall frn FAR-linjen, kross- och grovavfall (5 30 mm) som uppstr i krossning och PARlinjen.

Det finns tv huvudprodukter: P1. FAR som bestr av den naturligt uppkomna finandelen av malmen och som gr till fines-produktion, P2. PAR tas fram frn malmens grovfraktion och utgr rvara fr pelletsframstllningen. Den frsta sikten och separatorn bildar tillsammans GRAM-sektionen. Sikten var en stor vibrationssikt, med effektiv bredd 2400 mm, uppdelad i tre sektioner med effektiv lngd och lutning: 1770 mm 32, 2360 mm 16 samt 1770 mm 8. Siktmaskan r i genomsnitt 30 mm. Sovringsseparatorn (5) har trumdiametern 1200 mm och dess upp gift r att sl ut ett grovt grbergsavfall. Fint frn GRAM-sikt och magnetiskt frn GRAM-separatorn bildar tillsammans utgr tillsammans ingendet till fyra parallella primrsektioner. Dessa var utrustade med 5-dcks Mogen-siktar, dr de tre versta var divergatot-dck, fljt av ett fjrde dck med siktmaska 40 mm och ett femte dck med maska 18 mm. Siktarnas totalbredd var 1900 mm och lngd 1400 mm. Fint frn primrsiktarna upparbetas p torra svagmagnetiska trumseparatorer med 900 mm diamter. Grovt frn primrsiktarna gr till tv parallella krosssektioner, som vardera var utrustad med en kross typ Symons 7 Short Head och uppgavs kras med ansttningen 11 mm. Krossiktarna hade samma dimensioner som primrsiktarna, men hade siktmaskorna: 40, 35, 30, 25, 25 och 18 mm. Krosseparatorerna r av samma typ som primrseparatorerna. P grund av malmens snderfallsegenskaper kommer fint frn primrsiktarna till stor del best av renkrossade magnetitkorn p upp till ca 3 mm. Dessa anrikas ltt och gr via en autogen kross (Barmac Duopactor) till produktion av sinterfines och specialprodukter. Fint frn krossiktarna har strre andel halvkorn och gr via en Duopactor till anrikningsverkets kvarnsektioner fr produktion av ingende till pelletsverken. 2.2. Ingende material Det finns som tidigare nmnts fyra ingende material till sovringsverket: I1. Magnetitmalm frn Vitfors r den dominerande produkten, 11.4 Mton/r r 2003 vilket motsvarade en matning med i genomsnitt 1425 ton/h. Den har kraftiga variationer i partikelstorleksfrdelning och halter. I2. Grberget bestr av tillrednings- och anlggningsberg och uppfordras med malmen, men r ur simuleringssynpunkt srhllet. Mycket lite var vid det frsta tillfllet knt om partikelstorlekar och ev jrninnehll. 3

I3.

I4.

Magnetitmalm Alliansen (Vstra fltet) r en liten mngd, 0.5 Mton/r, eller 62.5 ton/h. Den r ngot grvre och har strre grbergsinblandning n den andra magnetitmalmen. H-PAR r magnetit i en fines-fraktion som kommer frn sovringen av hematitmalmerna. Mngden r ca 180 kton/r eller i snitt 23 ton/h.

Av de hr materialen r det frsts magnetitmalmen frn Vitfors som har strst genomslag p processen, bde partikelstorleksmssigt och i halt. Det senare ocks kopplat till partikelstorleken. Anlggningsberget ger mest en utspdning av jrnhalten in p GRAM-sektionen och momentana laststtar p GRAMsikten. I viss mn ger det ocks kad belastning p krossektionerna. De andra materialen kan verka strande eftersom de inte matas kontinuerligt. Variationerna i partikelstorlek pverkar ingende halt p grund av att malmerna snderfaller p ett sdant stt att renkrossade magnetitkorn terfinns i stor utstrckning i fraktionerna under 10 mm, se tabell 1 (Sundvall, 2004).
Tabell 1. Fraktion (mm) > 60 30 60 20 30 10 20 < 10 Sammansttning av ing GRAM, (matare 8) (72 % Fe i magnetit). % Fe 29.41 28.02 29.96 34.01 50.28 Magnetit (%) 40.8 38.9 41.6 47.2 69.8 Grberg (%) 59.2 61.1 58.4 52.8 30.2

ven om halterna i fraktionerna kan variera ngot upp och ner s r det frmst skillnaderna i ingende partikelstorleksfrdelning som kommer att ndra ingende totalhalter. Exempel p ingende partikelstorleksfrdelningar visas i Figur 2. Hr syns ocks att utslagningen av grberg p GRAM-sikten resulterar i ett finare ingende till egentliga sovringen, dvs till primrsikten. Magnetitmalm frn Alliansen motsvarar i partikelstorlek ungefr materialet frn matare 8 medan PAR frn hematitsektionen r mycket finare. Tillsammans gr variationer i ingende partikelstorleksfrdelning orsakade av segregeringseffekter i fickor och varierande andel anlggningsberg att frdelningen mellan primrseparatorer och krosseparatorer kommer att variera, men ocks att krossbelastningen blir ojmn. Detta frstrks av att vid ett grvre och drmed fattigare ingende, s kas matningen fr att hlla uppe produktionen av FAR och PAR. Detta leder till extra hg belastning p krossar och krossiktar.

100

80

Cumulative % finer

60

Anl.berg Ing_GRAM_mat6 Ing_GRAM_mat7 Ing_GRAM_mat8 Ing_sovr_mat6 Ing_sovr_mat7 Ing_sovr_mat8

40

20

0 100

1 000

10 000 Particle Size (m)

100 000

1 000 000

Figur 2.

Ingende material frn Vitfors.

3. 3.1.

Metodik Simuleringar Frutsttningar Simuleringarna r genomfrda i programmet MODSIM (King, 2003) ver 3.6.12. I dessa anvnds en frenklad beskrivning av ing material. Det antas best av endast tv komponenter och tv partikelklasser: magnetisk med 91 % magnetit (motsvarar 66 % Fe) densitet 4.7 g/cm, grberg med 12 % magnetit (ca 9 % Fe) densitet 3.0 g/cm. Anledningen till frenklingen r, dels att simuleringsmodellerna fr de svagmagnetiska magnetseparatorerna endast kan hantera tv komponenter (magnetiskt-omagnetiskt), och dels att det saknas data ver halvkornsfrdelningar i ingende material. Anlggningsberget antogs i den frsta simuleringsomgngen bara best av den omagnetiska komponenten. Kalibrering fr tre delkretsar Med ett s hr pass stort, men nd frenklat, fldesschema r det inte mjligt att kalibrera och validera hela schemat p en gng. I stllet r det uppdelat i tre delkretsar, som var fr sig har kalibrerats mot underskningar i driftskala.

3.2.

3.2.1. GRAM Malm med tillredningsberg frn Vitfors samt anlggningsberg som uppfordras med malmen kommer till det stora malmlagret (Lappkyrkan). Drifrn matas det till den stora sikten. Den skra magnetitmalmen har till stor del fallit snder under transporten, vilket gr att jrnhalten r vsentligt hgre i siktens finfraktion och att sovringsseparatorn har ett ganska ltt arbeta med att sl ut ett frhllandevis rent grberg. I figur 3 visas ett kalibrerat fldesschema fr 5

GRAM och i figur 4 dess partikelstorleksfrdelningar. Notera att observerade vrden (markrer) stmmer hyfsat med simulerade frdelningar (linjer). Det blir inte ngon riktigt storleksuppdelning ver separatorn. Detta r sannolikt inte korrekt. Malmen har finare styckefall n det rena grberget och det borde ge ett sep-konc som r finare.

Figur 3.

Fldesschema fr GRAM

Figur 4.

Partikelstorleksfrdelningar fr GRAM

3.2.2. Primrsikt-separation-omsep Primrsikten stllde till mycket bekymmer i kalibreringen. De frsta simuleringarna med en modell fr en probalistisk sikt gav alldeles fr bra siktresultat. Nsta frsk, dr Mogensen-sikten simulerades med en klasseringskurva gav en hyfsad anpassning till kalibreringsdata. Eftersom vi ville underska inverkan av siktbelastningen valdes till slut att simulera den, som en verbelastad enkeldckad vibrationssikt med fysiska dimensioner lika med nedersta dcket i Mogensen-sikten. I den andra simuleringsomgngen r dock alla siktdcken simulerade var fr sig. Fldesschemat visas i figur 5 och partikelstorleksfrdelningarna ver sikten i figur 6. Det r tydligt att finandelen i grovt primrsikt underskattas i simuleringen med vibrationssiktmodell. Siktens simulerade avskiljningsgrns p 12 mm stmde nd mycket bra med den designade. P grund av stora svngningar i partikelstorlek och flde utanfr designkapacitet anvndes vibrationssiktmodellen ven om den gav fr lite fint i grovprodukten.

Figur 5.

Fldesschema fr primrsikt separation omseparation

3.2.3. Kross-sikt-separering ven fr den hr delkretsen visade det sig att den bsta kompromissen var att simulera Mogensen-sikten som en verbelastad vibrationssikt med fysiska dimensioner lika med nedersta siktdcket. Vid det frsta simuleringstillfllet fanns det oskerheter om krossansttning, verklig spalt och uppmtta cirkulerande laster i krosskretsarna. Modellparametrarna fr krossen fick drfr anpassas p knn eftersom inget vldefinierat kalibreringsunderlag fanns att 7

tillg. Kretssimuleringen fick trots det hyfsad verensstmmelse med uppmtta vrden, se figur 7 fr fldesschema och figur 8 fr partikelstorleksfrdelningar.

Figur 6.

Partikelstorleksfrdelningar fr primrsikt separation omseparation

Figur 7.

Fldesschema fr kross sikt separation

Figur 8.

Partikelstorleksfrdelningar fr kross sikt separation

Partikelstorleksfrdelningarna visar p ett grundlggande problem fr den hr kretsen utg krossprodukt har i sin branta del ungefr samma partikelstorlek, som avskiljningsgrnsen p sikten. Det gr att: en liten ndring av krossens ansttning, lite mer fukt i malmen och drmed smre siktverkningsgrad, mer anlggningsberg in till sovringen, kommer alla att kunna ge kraftiga svngningar i cirkulerande last. 3.3. Underskta kapacitetsfall Nr de tre stora delkretsarna var ngot snr kalibrerade byggdes de tillsammans i en stor simuleringsmodell fr hela sovringen, se figur 1. I den ingick d ocks DuoPactor-krossarna och ficksystemet efter dessa. I den hr modellen krdes ett stort antal olika kapacitets- och haltalternativ med det oroande resultatet att nskad mngd slutprodukter kanske kunde innehllas det nrmsta ret, men inte p ngra rs sikt. Resultat - Diskussion Frsta simuleringsomgngen De fall som befanns vara mest intressanta att granska nrmre var den d prognostiserade produktionen fr r 2006 (12 Mton magnetitmalm + 0.9 Mton grberg) och 2009 (15.7 Mton magnetitmalm + 2 Mton grberg). Simulerade godsflden i olika delar av processen jmfrdes i ett kombinerat linje- och stapeldiagram med angivna/bedmda maximala kapaciteter hos enskilda maskiner och processavsnitt, se figur 9. 9

4. 4.1.

Malmberget sovring
1 000 1 000

Ton/h

100
S01SI0001 S01TR005 S01TR006 S01TR007 S01SE001 S05TR011 S31MA001 S31SI001 S31MA002 S31SE001 S31SE002 S32MA001 S32SI001 S32MA002 S32SE001 S32SE002 S30TR003 S30TR002 S30KR001 S30TR001 S30SI001 S30SE001 S30SE002 S41MA001 S41SI001 S41MA002 S41SE001 S41SE002 S42MA001 S42SI001 S42MA002 S42SE001 S42SE002 S40TR003 S40TR002 S40KR001 S40TR001 S40SI001 S40SE001 S40SE002 S04TR004 S04TR013 S04TR002 S04TR007 S09TR001 S09TR002 S09TR005 S04TR010 S12MA002 S12SE005 S04TR011 09TR007 S09TR008 S09TR009

100

Kapacitet

2006 N

2006 Rik

2009 N

2009 Rik

Figur 9.

Last och kapacitetsdiagram fr sovringen fre 2006.

Det visade sig att den nskade genomsttningen i krossektionerna inte skulle klaras fr den nskade produktionstakten r 2009. Den generella slutsatsen blev att de trnga sektorerna var i ordning. 1. Primrsiktarna genom att de vid fuktigt, mullande ingende gods skickade fr mycket fint i grovprodukten och samma sak hnde sedan p krossiktarna varvid en cirkulerande last med fint gods byggdes upp ver dem. 2. GRAM-sikt som inte skulle klara ing materialmngd fr r 2009 och dessutom var mekaniskt hrt sliten. 3. Konkrossarna som 2009 skulle f fr stora godsflden. 4. Omsepareringen verbelastades vid stora mngder grberg. 4.2. Andra simuleringsomgngen Fr de identifierade trnga sektionerna blev det nu aktuellt att tidigarelgga investeringar. I ett frsta skede byttes under 2006 GRAM-sikt och krossiktar ut mot nya med hgre kapacitet, se tabell 2. Samtidigt gjordes det fler detaljprovtagningar i knsliga processavsnitt och ingende material karakteriserades bttre till halt och partikelstorlek. Anlggningsberget befanns t ex innehlla en hel del fri magnetit! Med delvis ny maskinpark och bttre simuleringsunderlag gjordes nu en ny simuleringsomgng, dr de olika delkretsarna var bttre kalibrerade. Vid det hr tillfllet ingick ocks i uppdraget att bedma det tnkbara utfallet av det redan beslutade utbyte av krossar och primrsiktar som skulle ske 2007, se figur 10.

10

Tabell 2. Ny GRAM-sikt Nya primrsiktar Nya krossar Nya krossiktar

Nya maskiner i sovringsverket. 1 st Shenker Banana, bredd 3.15 m, lngd 9.6 m 4 st Mogensen Sizer SEL2026-D2, 2 divergator-dck, 2 par siktdukar med maskppning 35 och 16 mm 2 st Sandviks H6800, MC-yttermantel och D-innermantel. Slaglngd 32, 28 eller 24 mm, CSS 15- 22 mm 2 st Mogensen Sizer SEL2046, 4 par siktdukar med lngd 2400 mm och med maskppning 45-35-25-19 mm

Figur 10.

Normalproduktion 2007 med strmatade krossar.

I figur 10 visas en totalsimulering fr produktion enligt 2007 rs plan, dr varje primrsektion matas med 425 ton/h. Begrnsande i det hr fldesschemat r en hg cirkulerande last ver krossarna, som ligger mycket nra maximal kapacitet. Krossarnas CSS var hr antagen till 15 mm. I schemat syns ocks att den nya GRAM-sikten (typ Banana) simuleras som sammansatt av fem mindre siktar och att alla dck i Mogensen-siktarna har kommit med i simuleringarna. Fr att kunna g hgre i kapacitet mste den cirkulerande lasten tas ner och det kan gras p flera stt: minska krosssansttningen, ka siktmaskan p krossikten eller kra med fullmatad kross. Det senare r ett krav om kapaciteten skall upp till 500 ton/h och sektion. 4.3. Validering En sak av mycket stort intresse r, vad innebr bytet till nya krossar i sig sjlv? I ett frsk att bedma gjordes ngra jmfrande simuleringar som sammanfattas i figur 11. Hr betyder: 11

Kross2005 lget 2005 utan ngra modifieringar av krossar och siktar, Kross2006 lget sent 2006 med nya primrsiktar, NyFull lget 2007 med nya krossar och siktar samt fullmatad kross, NyStr lget 2007 med nya krossar och siktar samt strmatad kross.

Utfallskurvorna fr Kross2005 och Kross2006 kommer frn krossmodeller och parametrar, som anvndes i frsta simuleringen. Kurvan fr NyFull kommer frn krossmodell med anpassning till siktdata frn frsk med fullmatad kross i Kiruna. NyStr kommer frn den senare krossmodellen, men med andelen fint per krosslag snkt fr att ge en bsta anpassning till de gamla utfallskurvorna. Ingendet till krossen pverkas knappt av de olika modifieringarna. Det r frst nr kross och krossiktar byts ut som det blir ngon skillnad i utgendet frn krossen. Denna bestr i att den fullmatade krossen genererar betydligt mer fint material under 10 mm, vilket ger en lttare uppgift fr krossiktarna.
100

80

Cum % finer

60

40

Grovt_42SI001_1 Kross2005_Ing Kross2005_Utg Kross2006_Ing Kross2006_Utg NyFull_Ing NyFull_Utg NyStr_Ing NyStr_Utg Sek30_spalt19.3_Ing Sek30_spalt19.3_Utg

20

0
100 1 000 10 000 Particle size (m) 100 000 1 000 000

Figur 11.

Simulerade krosskurvor gamla och nya krossar.

Under hsten 2007 har provtagningar gjorts i de nya kretsarna fr att se om de gjorda frndringarna givet nskat resultat. I figur 11 r inlagt resultat frn en verifiering med stor spalt. Verifieringen r dock inte helt representativ eftersom den krdes d malmen innehll extra mycket av mullande magnetit. En frsiktig tolkning ger dock att simuleringarna har lite fel, men p skra sidan. Utfallet vid den aktuella provtagningen blev en cirklast runt krossarna p 360 ton/tim med CSS p 19 mm. Rgodsmatningen per sektion var 425 ton/tim, dvs samma som fr simuleringen i fig. 10. 5. Slutsatser Det hr bidraget har visat att det r mjligt att med ett systematiskt angreppsstt kunna identifiera potentiellt kritiska delkretsar och maskiner. Man kan ocks leka med olika scenarier fr att se hur belastningar frndras med ingende 12

partikelstorlek. Detta r speciellt viktigt i det hr fallet, dr en mullande malm i kombination med grovt anlggningsberg kan ge kraftiga svngningar i ingende partikelstorlek. Nr det gller det egentliga simuleringsarbetet, s mste vi konstatera att de flesta simuleringsmodeller fortfarande r mycket primitiva. Det r egentligen bara siktmodellerna, som har hyfsat sanna kapacitetsberoenden. Hr finns mycket att gra med att t ex frbttra modellerna fr magnetseparering. Simuleringarna hade bara varit akademiska vningar av tvivelaktigt vrde om inte anrikningsverkets personal hade stllt upp och kritiskt granskat resultaten under arbetets gng och kommit med berttigad kritik. Det hr gjorde ocks att simuleringarna gjordes s att det blev lite extra marginal i resultatet, vilket visar sig i figur 11 dr det r tydligt att simuleringarna har erred on the safe side. Tittar vi p utfallet fr produktionen efter installation av de nya krossarna och siktarna s blev grbergsbelastningen betydligt lgre n planerat, som mest 195 ton/tim jmfrt med planerade och simulerade 250 ton/tim. Det r en frklaring till att de simulerade cirklasterna runt krossarna blev mycket hgre jmfrt med utfallet vid samma CSS p krossarna. Byte till lngre siktar har medfrt bttre siktkapacitet och mindre knslighet fr igensttning orsakat av icke nskvrt material frn gruvan. En ovntad flaskhals som dk upp efter installation av de nya krossarna var att belastningarna p transportrerna efter krossarna har ftt begrnsas till 450 ton/tim jmfrt med dimensionerade 500 ton/tim. Godsnivn i H6800-krossarna r hgre n i de gamla Symons-krossarna vilket innebr att betydligt mer berg hamnar p dessa transportrer vid frreglingsstopp med fastkrning som fljd. En verraskning blev ocks att slitaget p krossmantlarna p de nya H6800krossarna blev s mycket strre n vntat. Nu mste vi mantla varje kross 5-6 gnger/r att jmfra med 1 gng/r fr de gamla Symonskrossarna. Orsaken till det stora slitaget r snedbelastning och att vi kr krossarna med omttade krosskammare (strmatning). Detta ska tgrdas under vren -08. 6. Referenser Tano, K., Plsson, B., Alatalo, J. and Lindqvist, L., 2006. The use of process simulation methodology in process design where time and performance is critical. In: S. Komar Kawatra (Ed.), Advances in comminution, SME Annual Meeting, St. Louis, MO, USA, 26-29 March 2006, Society for Mining, Metallurgy and Exploration, Littleton, CO, USA, pp. 481-493. Tano, K., Plsson, B. and Alatalo, J., 2006. Validation of a computer simulated circuit design at the LKAB Malmberget concentrator. Proc. 11th European Symposium on Comminution, Budapest, Hungary, 9-12 October 2006, Hungarian Chemical Society, Budapest, Hungary. 6 p.

13

King, R.P., 2003. Modular Simulator for Mineral Processing Plants. Mineral Technologies International, Inc.,1147 East 700 South, Salt Lake City, UT 84102, USA. Sundvall, ., 2004. SOVRING\ANLGGNINGSGRBERG/MK3, provtagning GRAM (in Swedish). Tables over chemical analyses by fractions. Internal LKAB-report.

14

Novel Fluidised-bed Tribo-electrostatic Separator for Dry Coal Preparation

R. K. Dwari and K. Hanumantha Rao* Division of Mineral Processing, Lule University of Technology SE-971 87 LULE, Sweden (*Author for correspondance, e-mail: hanumantha.rao@ltu.se) ABSTRACT Dry coal beneficiation has been examined by tribo-electrostatic method using Indian thermal coal sample from Ramagundam coal mines. The process of triboelectric coal/ash cleaning is carried out with a newly built cylindrical fluidised-bed tribocharger with internal baffles, all made up of copper metal. The charge transfer of coal maceral and mineral particles upon contacts with copper plate of tribocharger is measured. Separation of particles in an electrostatic separator according to the polarity of particle charge is discussed with respect to gas flow rate and residence time in fluidised-bed trobo-charger and the applied electric field. The coal and mineral particles charge with positive and negative polarities respectively. The magnitude of particles charge found to be relatively high illustrating greater efficiency of contact electrification in fluidised bed tribo-charger. The separation results with minus 300 m size fraction of coal containing 43% ash showed that the ash content is reduced to 18% and 33% with an yield of about 30% and 67%, respectively. These results are comparable to the maximum separation efficiency curve of washability studies on this coal sample. Since the ash percentage of coal particles collected in the bins close to positive and negative electrodes are about 70% and 20%, a better yield with low ash content can be accomplished on recycling the material.
Keywords: Coal preparation; Electrostatic separation; Fluidisation; Tribo-electrification; Particle charging

Introduction Coal is the single largest fossil energy source used world-wide and accounts for more than 60% of the total commercial energy consumed. Between 60-80% of this coal is used for electric power generation and most of which through a system of pulverised coal combustion (International Energy Outlook, 2005). The major portion of the coal used for such power generation is not clean enough to maintain the rigorous environmental standards required these days world over. The problem is the high sulphur content in coal used in most of the western countries or ash as is the case in countries like India. The sulphur exist both inorganic and organic forms and the SOX gas emissions lead to catastrophic environmental problems (Masuda et al., 1983). In the year 2005-06, the total production of coal in India was 344 million tones where 261 million tones of coal were used for the generation of electric power. The coal fired plants in India were among the most polluting sources as indicated by environmentalists and pollution control agencies. Substantial portion of ash is liberated as it enters the boiler from the mill. Any attempt to reduce the amount of ash going from the mill to the boilers in the already existing units without much investment would improve the performance of power generation significantly and increase the life of the boilers. Recently, the literature on dry beneficiation methods for coals with specific reference to high-ash Indian coals has been summarised by us (Dwari and Rao, 2007). Tribo15

electrostatic process is one of the key dry process techniques to separate the ash forming inorganic minerals from coal. Electrostatic separator with tribo-charging technique has great potential for coal preparation in fine sizes. There have been some investigations carried out in this direction but has not achieved commercial status in the coal beneficiation industry (Hower et al., 1997). Tribo electrostatic separation Tribo-electric separation involves charging of particles by contact or friction with other particles or with a third material, usually the walls of a container or pipe, followed by transport or free-fall through an electric field that deflects the particles according to the magnitude and sign of their charge. When two dissimilar particles are in contact or rub against each other, there is a transfer of electrons (charge) from the surface of one particle to the other until the energy of electrons in each material at the interface is equalised. When two metals of different work functions, M 1 and M 2 (eV), are brought into contact and then separated, the Fermi levels of the two metals coincide and a potential difference, Vc, is established across the interface. Harper (1951) suggested that they will exchange electrons by tunnelling so that thermodynamic equilibrium is maintained. The contact potential difference is given by: ( M 2 ) (1) VC = M 1 e The charge transfer Q during the contact is: ( ) (2) Q = CVC = C M 1 M 2 e where C is the capacitance between two adjacent bodies. The capacitance C is defined A as C = 0 , where A is the effective area of contact, z is the separation at contact, 0 is the z permittivity of free space = 8.85 x 10-12 Fm-1. Then, the charge transfer Q equals to A Q = 0 VC (3) z The surface charge density that can be generated during contact is Q ( ) (4) = = 0 M1 M 2 A z e In general, there are three maceral classes of coal with several sub-classes. The predominant maceral is vitrinite formed by carbonization of cellulose. The other two general classes are liptinite, formed from noncellulose materials such as resins and seeds, and inertinite materialized from charred plant remains. In the late 1970 and early 1980s, Inculet and co-workers (1982) were analysed maceral fractions of electrostatically beneficiated coal and found that different maceral types acquired different charge polarities. From petrographic analysis they found that a major portion of the vitrinite charged positively while inertinite charged negatively. Because of larger pores present in inertinite caused by the preservation of the original plant cell structure in this maceral type, they harbour negatively charged minerals not liberated by grinding. Several studies showed that clean coal generally charges positively and ash-forming minerals or high-ash coals charge negatively (Carta et al., 1976; Lockhart, 1984; Alfano et al., 1988). Many researchers have been contributed to understand particle tribo-charging characteristics and the separation of ash forming minerals from coal macerals (Bouchillon and Steele 1992, Ban et al. 1993, Tennal et al. 1999, Ahmadi et al. 2000, Trigwell et al. 2003)

16

Mazumdar et al. (1995) suggested some relevant problems and felt the research needs on the electronic surface structure of coal and mineral particles, the effect of particle size distribution and the influence of surface contamination on tribo-charging and separation processes. They also outlined that several fundamental factors that influence the tribocharging and separation process are not yet fully understood and there are considerable uncertainties in the successful operation of tribo-electrostatic process and hence preventing its commercial implementation to date. Trigwell and Mazumder (2001) investigated the effects of surface composition on work function of materials by X-ray photoelectron spectroscopy and UV photoelectron spectroscopy in air and measured the work functions of copper, stainless steel, aluminium, nylon and polytetrafluorethylene. The studies showed that the work function varied considerably as a function of surface composition and the actual work functions of material surfaces differed from their expected values due to varied surface composition upon exposure to the environment. A slight change in environment and humidity changed the work function values leading to similar work function values for the coal macerals, pyrite, and copper, alternating between positive and negative, and inferring the possibility of bipolar charging over wide range. The polystyrene thought to be a good charging medium although questioned its tribo-charging efficiency due to soft nature of the surface and the charging surface is likely to get covered with a thin layer of fine coal powder. Based on the difference in the work functions of the contacting surfaces, an electrical imbalance between the two components develop and charge passes through the interface because of attaining charge equilibrium (Frass and Ralston, 1940; Zhou and Brown, 1988 ). Larger is the difference in work functions of contacting surfaces, greater will be the charge acquisition. Coal is a heterogeneous mixture of macerals and minerals with different work functions. If the difference between the mineral and maceral surface could be widened, there will be greater magnitude of charge acquisition on the particles and separation between maceral and mineral will be more efficient. The chemical pre-treatment of the raw material by regulating the ambient moisture by NH3 (Ralston, 1961) found to be beneficial for separation. Lawver (1958) improved potash separation by heat treatment. Zhou and Brown (1988) were reported the increase in the coal separation efficiency after chemical pre-treatment. Various inorganic and organic aliphatic, aromatic and hydrocarbon gaseous agents were used in dry nitrogen atmosphere in the fluidised-bed tribo-charging medium for the selective separation of the macerals. Turcaniova et al. (2004) studied the effect of microwave radiation on the triboelectrostatic separation of coal. Lowell and Brown (1988) suggested that polymers containing electronegative chemical groups (e.g. halogens) tend to charge negative, whereas polymers with electropositive groups (e.g. OH) are generally charge positive. Wiles et al. (2004) investigated the effects of surface modification and moisture on the rates of charge transfer between metals and organic material. There are two key aspects, first selective charging and then separation in an efficiently designed tribo-electrostatic separator. Selection of tribo-charger material is important in view of obtaining opposing charge polarities of mineral and macerals that greatly influence for a successful and efficient separation. The traditional methods for frictional electrification of particles are vibrating feeder, rotating cylinder, flowing through chutes, pipe and nozzles. In recent years, cyclones and fluidized bed are used. The separators used by researchers with different ways of achieving tribo-electrification are shown in Figure 1. Soong et al. (2001) investigated the effect of different designs in the separation zone of the tribo-electrostatic separator on the beneficiation of Slovakian coals. They worked on conventional parallel electrode plates and two new types of electrode plates (i.e., louvered

17

Fig. 1. Electrostatic separators based on contact or tribo-electrification (Manouchehri et al. 2000). plate and cylindrical concentric tube plate separators), and found that parallel plate separator is the best for coal separation. In the this paper, the results achieved on a high ash non-coking coal using a new fluidized-bed tribo-charger with internal baffles followed by a cyclone discharging system of the particles to the electric field are presented and discussed. Experimental Materials High ash non-coking coal sample from Ramagundam coal mines was used in the present tribo-electrostatic separation tests. The coal sample was ground, screened to 300 m size fraction and portions of representative samples are prepared for separation tests. The washability characteristics and petrography analysis of this coal sample were presented in an earlier publication (Dwari and Rao, 2006). Since quartz, kaolinite and illite found to be the major mineral phases in the Ramagundam coal, the charge behaviour of these minerals in tribo-electrification was analysed. Pure quartz sample (SiO2 > 99%) was obtained from Meviour S.A., Greece and the kaolinite (kaolinite 92%, quartz 4%, illite 3% and feldspar 3%) and illite (75% illite and 15% kaolinite) samples were supplied by Phybiosis, USA. Tribo-electrification and charge measurement Tribo-electrification was carried out with individual mineral phases of quartz, kaolinite and illite in fluidized bed tribo-charger and the magnitude of charge acquisition by the mineral particles was measured by the Keithley electrometer. A 20 g of mineral particles 18

were fed into the fluidised bed tribo-charger. The tribo-charging was carried out by following the step 1 procedure described in the next section. Different flow rates of nitrogen gas in the range of 3500-5500 l h-1 for fluidization and fixed interval of residence time (1, 2 and 3 min) were used for tribo-electrification. After tribo-charging, the particles were allowed to transport through the cyclone to the Faraday cup and the particles charge polarity and magnitude were determined by the Keithley electrometer. Tribo-electrostatic separation test procedure The experimental set-up for tribo-electrostatic separation of ash forming minerals from coal maceral is shown in Figure 2. The whole test unit mainly consists of a fluidised bed tribo-charger with internal baffles system (FTB), cyclone discharging system, two copper plate electrodes enclosed in a rectangular Perspex box, high DC voltage supply source (VS), collecting bins in the form of Faraday cups, nitrogen gas cylinder and Keithley electrometer with 6532 scanner card. The fluidised bed tribo-charger was made-up of copper metal of cylindrical structure of 0.065 m diameter and 0.7 m length. The top and bottom openings of tribo-charger were closed with funnel shape head which ends are connected with a pipe of 0.01 m diameter. The cylinder and funnel head was united with O-ring flange coupling. Filter cloths of 5 m pore size were sandwiched between the flanges at top and bottom to give support to the fine coal particles inside the fluidised bed chamber. An internal baffle system also made up of copper was integrated at the centre of the cylinder for efficient contact of fine coal particles with metal. The length of the baffle system is 0.64 m and placed at a distance 0.03 m from the top. The coal particles after tribo-charging were transported pneumatically through a pipe of diameter 0.01 m at a height 0.025 m from the base of tribo-charger and then through a cyclone to fall freely between the electrode plates. All the pipes and the cyclone were also made of copper. The length of cylindrical and conical sections of cyclone is 0.045 and 0.48 m respectively. The cyclone and vertex finder diameters are 0.04 and 0.03 m respectively. The apex diameter of the cyclone is 0.006 m. The cylindrical and conical sections were connected by O-ring flange coupling and therefore can be easily dismantled for cleaning. The bottom conical end of fluidised bed tribo-charger is connected to the nitrogen gas cylinder. The volumetric gas flow rate of nitrogen through the pipe can be regulated by the rotameter (RM) and gate valve V3 connected between the nitrogen cylinder and fluidised bed tribo-charger. The flow rate through the top of fluidised bed and cyclone were regulated by using the solenoid valves V1 and V2 respectively. The flow rate through the vertex finder can be regulated by valve V4, however this valve is closed all the time during the period of operation. The rectangular Perspex box has the dimensions of 1 x 0.52 x 0.52 m. Two electrode copper plates A and B of length 0.84 m and breadth 0.43 m are fixed within the box. The electrode plates are connected in such a way that top and bottom gaps between them can be adjusted. In the present tests, the top and bottom gaps of electrodes were maintained at 0.095 and 0.046 m respectively. The electrode A was connected to +ve supply source, while electrode B was connected to ve supply source. The electrodes were charged with a high DC voltage power supply. There were six collecting bins in the form of Faraday cups below the electrode plates to collect the material after passing through the electric field and the polarity and charge magnitude of the particles collected in each bin can be measured. All the Faraday cups placed on the Teflon surface. The dimensions of outer shield of Faraday cup is 0.365 x 0.07 x 0.04 m while the dimension of inner shield is 0.345 x 0.05 x 0.03 m. All the Faraday cups are connected to the 6532 scanner card of Keithley electrometer. To ensure stable voltage

19

Fig.2. Schematic representation of experimental set-up for electrostatic separation of coal. measurements the capacitance of Faraday cups were increased by connecting a 1500 nF capacitor in series between the shields of Faraday cups. The capacitance of Faraday cups, i.e., bins 1 to 6, are 1.519, 1.591, 1.569, 1.54, 1.53, 1.563 nF, respectively. By following the standard protocols supplied by Keithley, the potential difference V was measured using Keithley electrometer. The charge Q acquired by the coal particles at different bins were calculated by knowing the capacitance C of Faraday cup and voltage V by using the equation Q = CV. In the present beneficiation tests, a coal sample of 300 m size fraction was used. The D90 of the sample was 234 m and the mean diameter of particles was 86 m. A 25 g of coal sample was introduced from the top of the fluidized bed tribo-charger and then closed with top conical head sandwiched with a filter cloth. The test is a two step procedure. In the first step the valve V1 and V3 were opened while the valve V2 was closed so that gas will flow through the tribo-charger and the coal particles attain fluidization condition and get contact with wall and baffles of the tribo-charger. The mineral and maceral particles acquire opposite polarity based on their relative work functions with respect to copper. The minimum fluidization velocity calculated to fluidise a 100 m and 250 m particles was 0.01 and 0.06 m s-1, respectively. Accordingly, in order to achieve turbulent fluidisation in the tribo-charger, volumetric flow rates of nitrogen of 3500, 4500 and 5500 l h-1 were used. A residence time period of 30, 60 and 90 s was maintained for tribo-charging. In the second stage the valve V1 was closed while the valve V2 was opened and the charged particle were transported pneumatically to fall freely between the electrode plates through the cyclone by slowly increasing the velocity of nitrogen. The negatively charged particles were attracted by the positive electrode and the positively charged particles were attracted by the negative electrode

20

and based on their magnitude of charge polarity they are deflected and collected in bins 1 to 6. The samples collected in each bin were subjected to ash analysis by following the ASTM standard procedure. Tests were conducted at three different 10, 15 and 20 KV applied voltage. In order to understand the effect of applied electric field on the charged particles and the extent of particles collected in different bins, the size distribution of particles collected in each bin was analysed by Cilas particle size analyzer. Results and Discussion Effect of flow rate and residence time on tribo electrification of quartz, kaolinie and illite The effect of gas flow rate and residence time on charge acquisition by the ash forming inorganic particles in tribo-charger was initially studied. The results obtained after tribo-charging of the individual mineral phases of quartz, kaolinite and illite in fluidised bed tribo-charger with internal baffle system are shown in Figures 3, 4 and 5 respectively. Since quartz had higher work function, 5.4 eV (Kim et al. 1997) than copper, 4.38 eV (Inculet, 1984), it accepts electrons from copper and becomes negatively charged and the results shown in Figure 3 are in good agreement with these reported work function values. The results also show that increasing the flow rate of nitrogen for fluidisation, the magnitude of charge acquisition increases. Similar effect of gas flow rate on charge acquisition was reported earlier by Nifiku et al. (1989). It can also be observed that with increasing tribo-charging time the magnitude of charge acquisition also increases. The increase in tribo-charging time allows all the particles to come in contact with copper tribo-charger and accordingly the particles acquired higher charge magnitude. Similarly, the results in Figures 4 and 5 reveal that kaolinite and illite also charged negatively upon contact with copper corroborating higher work functions of these minerals than copper. However, an increase in the gas flow rate for higher fluidisation intensity, the charge acquisition is decreased until 4500 l h-1 and then increases at 5500 l h-1. Since these mineral samples were very fine and in micron and submicron size range, the aggregation and dispersion of the particles in the fluidised bed affected the charge magnitude. The increase in fluidization velocity imparts higher force and numerous contacts with metal surface resulting higher electron transfer in the case of quartz particles. The increase in tribo-charging time invariably increased the magnitude of particles charge at all the gas flow rates studied. This is obviously related to the time needed for all the particles to come in contact with the metal. The magnitude of charge acquisition observed to be relatively very high in comparison to our previous studies of contact electrification with cylindrical rotating drum (Dwari and Rao 2006). In the fluidised bed, the particles effectively come into contact with baffles and walls of the tribo-charger and therefore higher magnitude of charge acquisition. This significantly suggests that the fluidised bed tribo-charger with internal baffle system has a higher efficiency of tribo-electrification. Effect of voltage and tribo-charging time on particles charge The effect of applied voltage on the particles charge collected at different bins at three different tribo-charging times of 30, 60 and 90 s is shown in Figures 6, 7 and 8 respectively. The coal maceral and minerals are charged in the fluidized bed and passing through the cyclone based on their relative work functions. The particles entering into the electric field get deflected towards +ve or ve electrode plate according to their charge polarity and magnitude, and fall in different bins in between the electrodes. As shown in the experimental set-up, bin 1 is adjacent to +ve electrode plate while bin 6 is nearer to ve electrode plate and accordingly the negatively and positively charged particles are expected to collect at bin 1 and bin 6

21

Flow rate, l/hr

0 3000 -0.2 3500 4000 4500 5000 5500 6000
0 3500 -0.5 4000

Flow rate, l/hr

4500 5000 5500 6000

-0.4

-1

Charge, C/g

Charge, C/g

-0.6

-1.5

-0.8

-2

-2.5

-1
-3

-1.2
1 min 2 min 3 min

1 min 2 min 3 min

-3.5

-1.4

Fig.3. Effect of gas flow rate and tribocharging time on charge acquisition by quartz particle.
Flow rate, l/hr
0 3500 -0.2 4000 4500 5000 5500 6000

Fig.5. Effect of gas flow rate and tribocharging time on charge acquisition by illite particle
0.18
10 KV, 30 sec

0.16 0.14 0.12 0.1 0.08 0.06 0.04

15 KV, 30 sec 20 KV, 30 sec

-0.4

Charge, C/g

-0.8

-1

-1.2
0.02

-1.4
0
1 min 2 min 3 min

Charge, C/g

-0.6

0 -0.02
+Ve

6
Ve

-1.6

Bin number

-1.8

Fig.4. Effect of gas flow rate and tribocharging time on charge acquisition by kaolinite particle.

Fig.6. Effect of voltage on charge acquired by coal particles collected at diffrent bins at 30 s tribo-charging time.

22

respectively. However, the particles collected in all the bins are found to be positively charged with a very low magnitude in bin 1 and successively increasing in the bins towards ve electrode. The +ve charge of particles in all the bins could be due to non-liberated particles or coating of fine coal maceral on mineral particles leading to an overall positive charge acquisition. It can be seen in Figure 6 that the particles collected in bin 6 have higher magnitude of positive charge and hence collected nearer to the ve electrode plate while particles collected at bin 1 and 2 are either negative or lower magnitude of positive charge. It can also be observed that the magnitude of particles charge increasing from bin 1 to 6 suggesting that carbon rich particles are collected in the bin close to the negative electrode. The magnitude of particles charge collected in bins 2 to 5 increases with increasing the applied voltage from 10 to 20 KV. With increasing applied voltage, the electrostatic force between the electrodes increases leading to increased deflection of charged particles towards positive and negative electrodes and therefore the particles collected in the bins are in a different distribution.. This explains the magnitude of charge alterations in the bins close to the electrodes with increasing the applied voltage. Figures 7 and 8 show the results of particles charge collected at different bins at 60 and 90 s tribo-charging time respectively. The results display that highly positively charged particles collected at bin 6 which is bordering to the negative electrode and the charge of the particles increased in the bins from positive plate to negative plate. At 10 KV applied voltage, the particles collected close to +ve electrode (bins 1 and 2) and ve electrode (bin 6) have lower and higher magnitude of charge than at 15 and 20 KV. It can also be seen from Figures 6 to 8 that increasing tribo-charging time, the magnitude of particles charge increases at 10 and 15 KV applied voltage. In the case of 20 KV applied voltage, the increase in tribocharging time increased the magnitude of particles charge in bins 1 and 6 that were adjacent to +ve and ve electrodes.
0.35
0.18
10 KV, 60 sec 15 KV, 60 sec

10 KV, 90 sec 15 KV, 90 sec 20 KV, 90 sec

0.16

0.3

20 KV, 60 sec

0.14

0.25

0.12

Charge, C/g

Charge, C/g

0.2

0.1

0.08

0.15

0.06

0.1

0.04

0.05
0.02

0 0
+Ve

0
1 2 3 4 5 6
Ve

0
+Ve

6
Ve

Bin Number

Bin number

Fig.7. Effect of voltage on charge acquired by coal particles collected at diffrent bins at 60 s tribo-charging time.

Fig.8. Effect of voltage on charge acquired by coal particles collected at diffrent bins at 90 s tribo-charging time.

23

The effect applied voltage on the weight percentage of material and size distribution of particles collected at different bins at 60 s tribo-charging time is shown in Figure 9. A typical size distribution of particles collected at different bins at 10 KV applied voltage and at 30 s tribo-charging time is presented in Figure 10. It can be observed from Figure 9 that the weight percentage of particles collected at positive plate is higher than particles collected at other bins except bin 4 and 30 percent of total coal sample is stacked to the electrode plates. The percentage of particles collected at bin 4 decreased with increasing applied voltage. At higher voltage, the electrostatic force between the electrodes is greater which influence on the deflection of particles. Oppositely charged particles are dragged in opposite direction with increased force at higher applied voltage. It can also be seen from the figure that the mean particle size increasing from positive plate to bin 6. However, the mean particle diameter of particles adhered to ve plate is lower. This is clear that finer particles are attracted more towards the plates and carbon rich particles having higher particle size and higher magnitude of positive charge are collected at bins 5 and 6.

160 wt% and particle size 140 120 100 80 60 40 20 0

+Ve Plate Bin1 Bin2 Bin3 Bin4 Bin5 Bin6 -Ve Plate

10 KV, wt% 15 KV, wt% 20 KV, wt% 10 KV, mean dia m 15 KV, mean dia m 20 KV, mean dia m

Plate and bin number

Fig. 9. Effect applied voltage on the weight percentage of material collected and size distribution of particles collected at different bins at 60 s tribo-charging time.

24

100 90 80

Cummulative wt%

70 60 50 40 30 20 10 0 0.01

Left plate Bin 1 Bin 2 Bin 3 Bin 4 Bin 5 Bin 6 Right plate

0.1

10

100

1000

Particle size, m

Fig.10. Size distribution of particles collected at different bins and plates at 10 KV applied voltage and at 30 s tribo-charging time. Effect of voltage and tribo-charging time on coal beneficiation The proximate analysis of Ramagundam bulk coal shows that it contains 5.8% moisture, 23.5% volatile matter, 43.2% ash and 33.3% fixed carbon. The washability studies carried out with this coal reveal that 25 % ash clean coal can be achieved with 65% yield. A typical ash percentage of coal collected at different bins at 10 KV applied voltage and 60 s tribo-charging time is shown in Figure 11. The ash analysis of coal particles collected at plates and different bins suggests that it is possible to reduce the ash content of coal from 43% to 18% with the present system. The ash percentage of coal collected at positive and negative plates are 70 and 21.7% which illustrate that mineral particles are charged negatively and attracted towards positive plate whereas carbon rich particles are positively charged and collected at negative plate. It can also be observed that the ash percentage of coal collected at bin 6 is lower than the ash content of coal collected at negative plate. This could be due to coating of fine coal particles on the mineral surface which drag the minerals to the negative plate and thus higher ash content than the particles in bin 6. In order to understand the performance of electrostatic separation, cumulative recovery of combustible matter and cumulative yield as a function of cumulative ash of the coal particles collected from negative plate to positive plate were examined. The beneficiation results with 300 m fine coals with fluidised bed electrostatic separator at different tribo-charging times of 30, 60 and 90 s are shown in Figures 12, 13 and 14 respectively. The feed coal ash content can be read from these figures corresponding to 100% yield or 100% combustibles. Figure 13 shows that it is possible to achieve 33% ash clean coal from 43% ash coal with 67% yield at 10 KV applied voltage and at 60 s tribo-charging time. The total combustibles recovered equal to 80% at these test conditions. The increase in applied voltage does not improve the separation and in fact the separation is lesser at higher voltage than at 10 KV. The separation results are nearly the same at 15 and 20 KV applied voltage. At 10 KV applied voltage, the separation results are similar at 30 and 60 s tribo-charging time (Figures 12 and 13), although the particles acquired higher magnitude of charge at higher tribo-charging time. The optimum parameters for best separation appear to be 60 s tribo-charging time and 10 KV applied voltage. 25

80
10KV, 60 sec

100 90 80 70

100 90 80 70 60 50 40 30
Cum wt%, 10 KV Cum wt%, 15 KV Cum wt%, 20 KV Combustible recovery%, 10 KV Combustible recovery%, 15KV Combustible recovery%, 20 KV

70

60

40

Cum. yield %

Ash%

60 50 40 30 20 10

30

20

10

20 10 0 50

0 +Ve Bin1 Bin2 Bin3 Bin4 Bin5 Bin6 -Ve Plate Plate

Plate and bin number

0 15
Ve

Fig.11. Ash percentage of coal collected at different plates and bins at 10 KV applied voltage and 60 s tribo-charging time.

20

25

30

35

40

45

Cum. Ash%

+Ve

Fig.13. Effect of applied voltage on coal beneficiation at 60 s tribo-charging time

100 90 80 70 60 50 40 30 20 10 0 15 20 Ve 25 30 35 40 45
Cum. Ash%

100 90 80 70 60 50 40 30 20 10 0
15
Ve

100 90 80 70

100 90 80 70 60 50 40 30 20 10 0 50 +Ve

Combustible recovery%

Cum. yield%

50 40 30 20 10 0

Cum wt%, 10 KV Cum wt%, 15 KV Cum wt%, 20 KV Combustible recovery%, 10 KV Combustible recovery%, 15 KV Combustible recovery%, 20 KV

Cum. Yield %

60

Cum. wt%, 10 KV Cum wt%, 15 KV Cum wt%, 20 KV Combustible recover%, 10 KV Combustible recovery%, 15 KV Combustible recovery%, 20 KV

20

25

30

35

40

45

50
+Ve

Cum. Ash%

Fig.12. Effect of applied voltage on coal beneficiation at 30 s tribo-charging time.

Fig.14. Effect of applied voltage on coal beneficiation at 90 s tribo-charging time.

26

Combustible recovery %

Combustible recovery %

50

Conclusions High-ash thermal coal beneficiation was studied by using an electrostatic separation. The separation process involves tribo-electrification with a fluidised bed tribo-charger and separation with an electrostatic field. The following conclusions are derived: 1. A new cylindrical fluidised bed tribo-charger with internal baffles system made-up of copper metal has been assembled for efficient tribo-electrification of particles. The unit consists of several collecting bins below the electrode plates of separator, which acts as Faraday cups and they are connected to Keithley electrometer through the scanner card. It is thus possible to measure the charge polarity and magnitude of particles collected in different bins. 2. The build-up of charge on the ash-forming quartz, kaolinite and illite minerals in the present fluidised bed tribo-charger found to be greater when compared to other tribo-charging systems. The minerals acquired negative charge upon contacts with copper and the charge measurements of coal particles collected in the bins showed that the carbon rich particles are charged positively. 3. The effect of gas flow rate and residence time of fluidisation and applied voltage showed that these parameters have strong dependence on particles charging and therefore on coal beneficiation. The increase in gas flow rate and residence time increases the magnitude of charge acquisition by the particles. The optimum conditions for coal separation found to be 60 s tribo-charging time at a gas flow rate enough for particles fluidisation and 10 KV applied voltage. 4. At the optimum conditions, the minus 300 m size coal sample containing 43% ash is reduced to 18% ash with about 30% yield. A high yield of 67% can also be obtained but with a moderately reduced 33% ash. In this case, the combustible recovery found to be nearly 80%. 5. Since the coal particles collected in the bins adjacent to +ve and ve electrodes contain 70% and 20% ash, a better separation results can be accomplished by recycling the material. Thus, the electrostatic separation with fluidised bed tribo-charging system realised to be a potential technique for non-coking coal beneficiation. 6. It is possible to treat the fine coal material with the vapours of acidic and/or basic organic solvents in the fluidised bed tribo-charger for altering the surface energetic structure of particles and thereby enlarging the work function difference between the contacting surfaces for efficient tribo-electrification and separation. The results of these studies will be presented in a future publication. Acknowledgement The financial support for the project on Electrostatic Beneficiation of Indian Thermal Coals by the Swedish International Development Cooperation Agency (SIDA/SAREC), Stockholm, Sweden, is gratefully acknowledged. References Ahmadi, G., He, C., Ban, H., Stencel, J.M., Air flow and particle transport in a triboelectric coal/ash cleaning system counter flowing straight duct design. Particulate Science and Technology, 2000, 18, 213-256. Alfano, G., Carbini, P., Ciccu, R., Ghiani, M., Peretti, R., and Zucca, A., Progress in Triboelectric Separation of Minerals, In: Proceedings of XVI International Mineral Processing Congress, ed. E. Forssberg. Elsevier Pub B.V, Amsterdam, 1988, pp. 833-844. Ban, H., Schaefer, H.L., Saito, K. and Stencel, J.M. Particulate tribocharging characteristics relating to electrostatic dry coal cleaning. Fuel, 1993, 73, 1108-1113.

27

Bouchillon, C.W. and Steele, W.G., Electrostatic separation of ultrafine ground low rank coals. Particulate Science and Technology, 1992, 10, .73-89. Carta, M., Del Fa, C., Ciccu, R., Curreli, L. and Agus, M., Technical and economical problems connected with the dry cleaning of raw coal and in particular with pyrite removal by means of electric separation, In: Proceedings of 7th International Coal Preparation Congress, ed. A. Partridge. Adept Printing pvt. Ltd, Australia, 1976, K2, pp.1-35. Cross, J.A., Electrostatics: Principles, Problems and Applications, Adam Hilger, Bristol, England, 1987, pp. 17-90 and 248-273. Dwari, R. K., and Hanumantha Rao, K., Tribo-electrostatic behaviour of high ash non-coking Indian thermal coal. International Journal of Mineral Processing, 2006, 81, 93-104. Dwari, R. K., and Hanumantha Rao, K., Dry beneficiation of coal a review. Mineral Processing and Extractive Metallurgy Review, 2007, 28, 177-234. Frass, F. and Ralston, O.C., Contact potential in electrostatic separation. Industrial Engineering Chemistry, 1940, 32, 600-604. Fricke, G., The use of electrostatic separation processes in the benefication of crude potassium salts. Phosphorous & Potassium, 1977, 90, 42-45. Haga, K., Applications of the electrostatic separation technique. In Handbook of Electrostatic Processess, ed. J.S. Chang, A.J. Kelly and J.M. Crowly. Marcel Dekker, Inc., New York, 1995, pp. 365-386. Harper, W.R., The volta effect as a cause of static electrification. Proceeding Royal Society London A. Math. Phys. Sci., 1951, 83-103. Hower, J.C., Hang, B., Schaefer, J.L. and Stencel, J.M., Maceral/microlithotype partitioning through triboelectrostatic dry coal cleaning. International Journal of Coal Geology, 1997, 34, 277-286. Inculet, I.I., Bergougnou, M.A., Brown, J.D., Electrostatic beneficiation of coal. In: Physical Cleaning of Coal Present and Developing Methods, ed. Y.A. Liu. Marcel Dekker, New York, 1982, pp. 87-131. Inculet, I.I., Electrostatic Mineral Separation, Research Study Press Ltd., England, 1984, 153 pp. International Energy Outlook. http://www.eia.doe.gov/oiaf/ieo/coal.html, Retrieved on September 9, 2005. Kim, S.C., Son, N.W., Kim, D.H., and Oh, J.G., 1997, High efficient centrifugal and triboelectrostatic separation of unburned carbon from fly ash for ash recycling. In: The 12th Korea-US joint Workshop Energy & Environment, Taejon, Korea, 1997, pp 308-313. Lawver, J.E., Method of Separation Sylvite from Sylvinite ore. 1958, US Patent 2,839,190. Lockhart, N. C., Review paper. Beneficiation of coal. Powder Technology, 1984, 40, 17-42. Lowell, J. and Brown, A., Contact electrification of chemically modified surfaces. Journal of Electrostatics, 1988, 21, 69-79. Masuda, S., Toraguchi, M, Takahashi, T., and Haga, K., Electrostatic beneficiation of coal using cyclone-tribocharger. IEEE Transactions on Industry Applications, 1983, IA-19(5), 789-793. Mazumder, M.K., Tennal, K.B., and Lindquist, D.A., Triboelectric Separation of Coal from Mineral Impurities. In: Proceedings of Annual meeting Electrostatic society of America, Laplacian press, Morgan Hill, CA, 1995, pp. 59-70. Manouchehri, H.R., Hanumantha Rao, K., Forssberg, K.S.E., Review of elctrical separation methods Part 1: Fundamental aspects. Minerals and Metallurgical Processing, 2000, 17(1), 1-14. Nifuku, M., Ishikawa, T. and Sasaki, T., State electrification phenomena in pneumatic transport of coal. Journal of Electrostatics, 1989, 23, 45-54.

28

Ralston, O.C., Electrostatic separation of mixed granular solids. Elsevier pub., Netherlands, 1961, pp. 261. Tennal, K.B., Lindquist, D., Mazumder, M.K., Rajan, R. and Guo, W., Efficiency of electrostatic separation of minerals from coal as a function of size and charge distribution of coal particles. IEEE, 1999, pp. 2137-2142. Trigwell, S., and Mazumder, M.K., Tribocharging in electrostatic beneficiation of coal: Effects of surface composition on work function as measured by X-ray photoelectron spectroscopy and ultraviolet photoelectron spectroscopy in air. Journal of Vaccum Science and Technology A, 2001, 19(4), 1454-1459. Trigwell, S., Tennal, K.B., Mazumder, M.K., Lindquist, D.A., Precombustion cleaning of coal by triboelectric separation of minerals. Particulate Science and Technology, 2003, 21, 353-364. Turcaniova, L., Soong, Y., Lovas, M., Mockcvciakova, A., Orinak, A., Justinova, M., Znamenackova, I., Bezovska, M. and Marchant, S., The effect of microwave radiation on the triboelectrostatic separation of coal. Fuel, 2004, 83, 2075-2079. Wiles, J.A., Fialkowski, M., Radowski, M.R., Whitesides, G.M. and Grzybowski, B.A., Effects of surface modification and moisture on the rates of charge transfer between metals and organic materials. Journal of Physics and Chemistry B, 2004, 108, 2029620302. Zhou, G. and Brown, J.D., Coal surface conditioning for electrostatic separation. The Canadian Journal of Chemical Engineering, 1988, 66, 858-863. Soong, Y. Link, T.A., Schoffstall, M.R., Gray, M.L., Fauth, D.J., Knoer, J.P., Jones, J.R. and Gamwo, I.K., Dry beneficiation of Slovakian coal. Fuel Processing Technology, 2001, 72, 185-198.

29

Process and Performance Increases Using Ciba RHEOMAXTM DR Technology

Ian Flanagan Ciba UK PLC Paul Stocks Ciba UK PLC

ABSTRACT Overview
Flocculants manufactured using Ciba RHEOMAX DR Technology show superior flocculating properties to conventional chemistry flocculants for many different mineral ore types and offer a significant reduction in overall consumption at similar settling rates.

Process Description
Feedwell turbulence can cause shear degradation of the flocculated particles resulting in higher flocculant usage and a reduced ability to control bed level in the thickener, when employing conventional flocculant chemistry. Ciba RHEOMAX DR products create structures, which are more robust than those formed by conventional flocculants and are therefore more tolerant of fluctuations in plant and feedwell conditions. In this case the customer is undertaking solid/liquid separation of the tailings stream, using a conventional thickener to recover water, an expensive and limited resource due to the location of the plant in an environment, which can be particularly dry. The reason behind the employment of Ciba RHEOMAX DR technology is to improve the compression and dewatering characteristics of the operation and give substantially higher underflow densities, without compromising Yield Stress (related to Rake torque).

Issue
The customer requirement is to produce sufficient clean water from their tailings thickener for re-use in the crushing and grinding circuits. During dry periods extra pressure is placed upon the tailings thickener to produce a higher underflow density and hence more return water.

Solution
The use of Ciba RHEOMAX 1050 at this site resulted in a reduction in overall flocculant usage of at least 34%. Higher underflow density solids were achieved in the order of 10% (ie 50% to 55% w/w). This also equated to an extra 4% of reclaim water. During the first stages of treatment with Ciba RHEOMAX 1050 drought conditions occurred. Under such conditions the plant is normally required to reduce throughput rates due to thickener return water limitations. With the improvements in thickener operation brought about through the treatment with Ciba RHEOMAX 1050 the plant continued to operate at full capacity resulting in no production losses.

1. Introduction
Ciba RHEOMAX DR products show superior flocculating properties over conventional flocculants, offering a significant reduction in overall consumption, more tolerance to process fluctuations and effectiveness over a wider dynamic operational range. Ciba RHEOMAX DR products also produce high density underflows at relatively low yield stress.

Process & Performance Increases using Ciba RHEOMAX DR Technology

31

Ciba RHEOMAX DR products are different to traditional flocculants in that they are produced from a different high molecular weight monomer species using different raw materials and requiring complex preparation. Polymerisation is therefore also more difficult making the products more expensive than traditional flocculants. This paper describes the effects and benefits obtained when these products were added to a copper processing plant tailings thickener.

2. Application to a Copper tailings thickener

The plant detailed processes over 100,000 tonnes per day of copper ore. The process encompasses crushing and grinding of the ore and then the removal of the copper via a number of flotation steps. Water is used extensively throughout the process and all waste streams are sent to the tailings thickeners at typically 27% -30% solids. Water is recovered and the thickened underflow at typically 50% solids is disposed of via a tailings dam. In this case the critical reagent employed in the circuit is water and given the arid operating climate of the plant sustainable water management is crucial to extending the life cycle of the plant and wherever possible recovery and recycling of water is considered best practice. There are also limits on the availability of fresh water supplies which carry a high cost and an associated environmental impact as they come from a desalination plant. Thickener operation is also challenged by limitations on rake torque and pump control which reduces the attainable capacity and threatens onward expansion of the overall plant operation. In this particular plant a shortage of water can reduce the number of operational hours the plant can process and hence greatly affect revenues generated from the copper production. Priorities are therefore twofold. 1. 2. To maintain / increase a clean and reliable source of recyclable water, thus negating the frequency for fresh water replenishment. To improve thickener operation dynamics and increase throughput capacity

Process & Performance Increases using Ciba RHEOMAX DR Technology

32

Figure 1: Copper processing flow sheet

3. Scope of the trial

Testing on a stream of thickener feed was done on site for a period of one month. The Ciba RHEOMAX DR was added to the thickener tailings feed in the same place as the traditional flocculant and progressive sampling took place to control settlement rate and Ciba RHEOMAX DR addition. Standard flocculant preparation and dosing equipment was used for the trial as detailed below.

Figure 2: Ciba RHEOMAX DR Preparation & Dosing

The scope of the on the following aspects

trial concentrated

Process & Performance Increases using Ciba RHEOMAX DR Technology

33

Density of Solids - Limitation on Rake Torque - Limitation on pumping - Thickening Capacity - Variation of minerals in the thickener Water Recovery - Water Limitations - Availability and Water Cost - Environmental Requirements Expansion of current Operations - Problem with Thickener Operation

The primary objective was defined as:

Increase Solids in Underflow from 50% to 55%

Table 1: Plant Variables / Objectives

Concentrator Plant Variables Capacity Plant trial Feed Well Solids Typical Underflow Solids Standard Flocculant Standard Dose Range Alternative Product Dose Range 130,000 TPD Thickening Tailings Process 10 % (after in line dilution) 50% Incumbent supplier 10 - 18 gpt Ciba RHEOMAX 1050 4 - 10 gpt

4. Results
The Ciba RHEOMAX DR was added into the tailings thickener feed launder and Immediately it could be seen that the flocc size had increased and that the flocculated aggregates were more resistant to the shear Imparted by the thickener launder. Ciba RHEOMAX DR dose was therefore reduced accordingly to maintain the required settlement rate and a dose reduction of 34% was immediately achieved. Figures 3 to 6 highlight the results obtained during the trial.

Process & Performance Increases using Ciba RHEOMAX DR Technology

34

Figure 3: Plant Throughput (Trial Data)

Plant Throughput & Dose

180,000 20.0

Std Flocculant
160,000 140,000

RHEOMAXTM 1050

18.0 16.0 14.0

Throughput (TPD)

120,000 100,000

Dose (g/tds)

12.0 10.0 80,000 8.0 60,000 40,000 20,000 6.0 4.0 2.0

0 0.0 Date 06/05/2007 13/05/2007 20/05/2007 27/05/2007 01/04/2007 08/04/2007 15/04/2007 22/04/2007 29/04/2007

Figure 4 : Solids Content Vs Dose

Solids Content vs. Dose
56.00

54.00

Solids Content (%)

52.00

50.00

48.00

46.00 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0

Dose (g/tds) Standard

RHEOMAX 1050

Process & Performance Increases using Ciba RHEOMAX DR Technology

35

Figure 5 : Solids Content Vs Rake Torque

Solids Content vs. Torque
60.0

55.0

50.0

Torque

45.0

40.0

35.0

30.0 46.00

48.00

50.00

52.00

54.00

56.00

Solids Content (%)

Standard

RHEOMAX 1050

Figure 6 : Solids Content Vs Amps

Solids Content vs. Amps
150

140

130

Amps (%)

120

110

100

90 46.00

48.00

50.00

52.00

54.00

56.00

Solids Content (%)

Standard

RHEOMAX 1050

Process & Performance Increases using Ciba RHEOMAX DR Technology

36

5. Conclusions
Testing with Ciba RHEOMAX DR has been successfully carried out on the tailings thickener and compared with the incumbent traditional flocculant. The employment of Ciba RHEOMAX DR on the tailings thickener resulted in the following process improvements.

Extra water recovery: 232 litres/second Fresh water replenishment saving: 20,023 USD per day Improved underflow density: 8.2% increase Increased plant capacity: 20,000 tons per day (+15%) Increase profitability through extra copper: 1.3 million USD per day Reduced dose rate: 48% decrease Tailings disposal benefits; Reduced torque; Improved pump control

Details of the exact process improvements and cost savings are detailed below in Table 3.

Table 2: Improvements and Savings

Description
Plant Capacity Thickener Feed U/F Density (Standard Flocculant) U/F Density (Ciba RHEOMAXTM 1050) Fresh Water Cost - Desalination Water Recovery with Standard Flocculant TPD x (UDCF - TFS)/(UDCF x TFS) x 100 Recovery with Ciba RHEOMAXTM 1050 TPD x (UDRH - TFS)/(UDRH x TFS) x 100 Extra Water Saving with Ciba RHEOMAXTM 1050 RWRH - RWCF Fresh Water Saving Per Year (360 days) Extra Water x CFW 20,023 232

Acronym
TPD TFS UDCF UDRH CFW RWCF RWRH

Value
130,000 27.00 49.00 53.00 1.00 216,175 236,198

Units tpd %wt/wt %wt/wt %wt/wt USD/m3 m3/day m3/day m3/day lt/sec USD/year

7,208,280

Process & Performance Increases using Ciba RHEOMAX DR Technology

37

Table 3: Benefits
Operational Benefits
Higher underflow density

Economic Benefits
Reduced requirement for fresh plant water from external sources Lower reagent consumption

Environmental Benefits
Reduced fresh water replenishment

Greater plant throughput Increased recovery of quality water / leach reagent More consistent dose response to fluctuations in feed solids Reduced shear degradation due to fluctuations in feed well turbulence

Less water volume to tailings

Improved leach efficiency

Reduced water loss to evaporation

More consistent plant performance

Lower energy consumption

Lower energy consumption of pumps (ie improved pumpability)

Reduced land requirement for tailings disposal

The trial has proved that Ciba RHEOMAX DR is a suitable, more cost effective, solution than traditional flocculants on the plant in question. Similar results have also been obtained on other mineral ore types making Ciba RHEOMAX DR an alternative choice in any Tailings Management Strategy or plant expansion.

6. References
Arbuthnot, I.M., Triglavacanin, R.A. (2005) Designing for Paste Thickening-Testwork and Sizing for Paste Thickeners. Paste 2005, Jewell, R. and Barrera, S. (eds), Australian Centre for Geomechanics, Perth, Bodley A. J., Cooling D. J. and Nikraz, H. N. 2003. Comparison of the Physical properties of Treated and Untreated Residue. Paper prepared for Environmental Science and Technology, published by the American Chemical Society. Cooling, D.J., Hay, P., Guilfoyle, L., (2002) Carbonation of Bauxite Residue. The 6th International Alumina Quality Workshop, 2002. Crozier, M., Jonhson, J. and Slottee, J.S. (2005) A systematic Approach to designing Paste thickener Projects. Paste 2005, Jewell, R. and Barrera, S. (eds), Australian Centre for Geomechanics, Perth,. Jewell, R.J., Fourie, A.B. and Lord, E.R. (2002) Paste and Thickened Tailings-A Guide, The University of Western Australia,.

Process & Performance Increases using Ciba RHEOMAX DR Technology

38

OPTICAL SORTING EMPLOYING DIFFERENT ANALYZING CONFIGURATIONS Jacek Kolacz Comex AS Box 53, 1309 Rud, Norway E-mail: jacek.kolacz@comex-group.com

ABSTRACT A new optical sorting system based on a complex image analysis has been developed at Comex AS - Norway. It makes it possible to separate or identify different mineral particles regarding their colour, shape, pattern, size and indirectly their composition. The optical system has successfully been tested for sorting and identifying metal and mineral particles providing high separation efficiency in the range of 90-99.9 %. The main advantage of this system is connected with its universality and at the same time very sophisticated image processing functions. The system can be operated in different configurations where the analysis of the processed material and the rejection of the waste particles are done in various sections of separators.

1.

Introduction In many mineral processing installations it is of critical importance to provide an efficient material purification. It is carried out by complicated processing steps involving grinding to very fine particle sizes. It is necessary to grind the particles in order to liberate the impurities from a mineral. However, purification of the coarse feed particles before grinding can provide significant benefit in overall energy efficiency and the final product purity. When the particles representing impurities can be separated before grinding they do not contaminate the feed stream for further processing and additionally they are not unnecessarily ground. This can be achieved by optical sorting carried out before the grinding stage when the particles have size in a range of several centimetres. Separating systems based on image analysis must always be tailored for the particular application. Very few of them are flexible enough to be applied to very different conditions without complicated re-programming and re-construction. Consequently, in the majority of possible optical system implementations, its complexity makes such applications cost prohibitive, Wortuba (2000). The newly developed sorting system from Comex makes it possible to separate or identify different particles or objects regarding their colour, shape, pattern, size and 39

indirectly their composition, Kolacz (1998). The optical system is based on the programming platform being very flexible for modifications. The main advantage of this system is connected with its universality and at the same time very sophisticated image processing functions. Low cost of both hardware and software makes it very competitive for other systems.

2.

Economic potential Optical separation represents enormous potential in processing of industrial minerals especially when global activities are considered. There are about 10 GT (gigatons) of various minerals crushed and ground yearly for different purposes (BGS, 2005). This corresponds to about 500-1000 TWh (Terra-Watt-hours) energy required to fulfil this process regarding size reduction only, when 50-100 kWh/T specific energy is assumed for an average crushing and grinding process. When additional energy for further processing is included (like flotation, drying, separation, landfill etc) this energy can be multiplied by 10-20! Fine ground rejects represent a serious environmental threat, where all these particles have to be safely prepared for landfill. Many of these particles, due to former processing stages, are mixed with different complex chemicals. This creates an additional environmental problem. Optical sorting may provide in average about 5-10% reduction of waste in very early processing stages, when the processed material is still large in size. When 10% energy saving is assumed out of 1000 TWh/T for size reduction only, this will correspond to about 100 TWh energy. This energy further corresponds to about 50% of the total annual energy consumption in Norway. This calculation is related only to the crushing and grinding energy and it does not include any further processing cost related to flotation, separation, purification, drying, landfill, etc. Even using these simple assumptions for energy calculations as shown above, it can be clearly noticed that an efficient separation of this kind of minerals carried out by optical sorting may provide enormous benefit for the global economy, environment and it may reduce global environmental threats like extensive CO2 emission. Considering total benefit related to this early stage separation, it can provide completely different evaluation aspects for current individual producers and for the new projected installations.

3.

Separation system configurations The optical separation system can generally be used in two different configurations. In the first configuration the analysis of the processed material is carried out over the transport belt. In the second configuration the analysis is done in a free fall when the particles are in the air, discharged from the belt. Both configurations can be positive or disadvantageous for the separation process depending on an application. 40

3.1.

Separation over the belt The first configuration is illustrated on Figure 1. The image analysis system includes an optical camera installed over the transport belt conveyor with the particle rejection unit at its discharge end. The system is operated by an industrial PC based control system employing a separation algorithm based on selected separation criteria. Each particle is identified as regards the material or combination of materials, its dimension, shape, structure and its position on the conveyor. The particles recognised as the waste fraction are rejected by the air nozzle system or the mechanical flaps (for larger particles over 50-60 mm). The particles related to the concentrate fraction are normally falling in to the belt discharge area. This configuration is used when the amount of waste particles is significantly lower than in the concentrate fraction. This reduces the compressed air consumption for rejecting particles. In the contrary case, if necessary, the system can be easily converted to reject concentrate particles from the waste stream.

Camera with lights

Pneumatic nozzles or flaps

Feed

Splitter

.... ...
Conveyor
PC

Valves communication

Product/Concentrate

Waste

Figure 1. Comex optical sorting system configuration with the analysis over the transport belt.

41

3.2. Separation in free fall

The second configuration (Figure 2) is similar to the first one, however, the camera is installed at the discharge of the belt. In this configuration the particles are analyzed in free fall.

Pneumatic nozzles or flaps Camera with lights

Feed

Splitter

... ....
Conveyor
PC

Valves communication

Product/Concentrate

Waste

Figure 2. Comex optical sorting system configuration with the analysis in free fall.

The particle analysis and the rejection of the waste material are done in a very limited area. This creates a lot of limitations related to processing speed, image analysis, rejection speed etc. 3.3. Main advantages and limitation Both separating techniques have advantages and limitations. The main limitation is related to the processing time and scanning accuracy. In the first configuration, where the particles are analyzed on the belt, they are first scanned and then, after a time period of few hundreds of milliseconds, they are separated by the rejection mechanism. In the meantime, when one image is scanned, the previous image is analyzed in the processing system in an electronic buffer. This allows additional time for complicated processing especially when complicated filtration tools are employed. The main limitation of this system is that the processed material can be scanned only on one side facing the camera and lights.

42

In the second configuration the material is analyzed in a free fall after leaving the discharge roll of the conveyor (or other transport equipment). In this case the processing time is very limited as the rejection of the waste particles takes palace just after the analysis with few ms delay. In such a case there is no time for sophisticated analysis like advanced filtration. The advantage of such a system is related to a possibility for double scanning on both sides of the analyzed particles, when an additional camera can be placed on the other side. The choice of the separator configuration will very much depend on the material properties and the final separation target.

4.

Separation techniques

4.1. General colour and shape analysis Particle recognition used to separate different materials is based on a complex shape and colour analysis carried out by the imaging system. Colour analysis is done by converting video signals to RGB (Red, Green, Blue) components and further to HSL (Hue, Saturation, Light). Hue is the parameter related to the colour frequency, saturation is decisive for colour intensity and the light parameter is connected to the pixel total intensity. This colour analysis is done separately for each pixel creating an image. Results from image processing, carried out in the Comex separator, can be displayed as a number of different parameters describing material particles. In addition to material colour, there are over 50 parameters (like diameter, perimeter, centre of mass, moment of inertia, particle orientation, elongation factor etc.) normally used for particles shape identification. Additional combinations of these parameters can also be used for distinguishing particles of interest. Each particle has a specified position and size so it is possible to use the splitting mechanism (air nozzles) to separate them from the material stream. Figure 3 illustrates some examples of image processing where darker particles are defined and separated as waste. In this case the separated particles are identified on the base of their colour. Darker colour shade refers to the particles containing more iron. Normally, these particles are removed by a magnetic separation process after grinding to fine sizes. However, when such particles are separated initially it may result in simplified grinding and purification process. Therefore, more economical process can be achieved (less material for grinding) and purer material can be produced (purer final product from less contaminated feed).

43

100 mm

Figure 3. Some of the processing steps during the white quartz particle separation (left group), where dark and red minerals (right group) containing impurities are removed.

4.2 Special techniques General colour and shape analysis is a basic separation technique used for optical sorting. This technique is limited to the material properties which are generally used for their visual identification. The technique is similar to manual separation (hand picking) still carried out in many operating plants. However, using other material properties, which are not visible for the human eye, provides new possibilities to provide more sophisticated separation. The typical special separation techniques may utilize: magnetic properties, thermal conductivity, thermal capacity and X-ray attenuation. Especially, thermal material properties can be very useful to provide a sophisticated separation process. An application of thermal cameras can provide additional information about the material properties, which are not achievable by standard optical devices. Figure 4 shows an example of applying two thermal cameras into the separation system. The processed material is pre-conditioned before the separation to achieve a 44

homogenous temperature in the range of 60-100 C. When the particles are then distributed on the transport belt, they are exposed to a cooling effect. An additional fan or blower can be applied to provide a fast cooling effect. Then, the particles are analysed by two thermal cameras in different time intervals. In this way the material can be separated on the base of a relation between specific heat and heat conductivity as well as surface porosity. Particles having larger surface and the same specific heat will reach lower temperature faster than the other ones. Materials with differences in heat conductivity and specific heat will provide even more possibilities to distinguish such particles.
Heating device Feed

Thermal camera 1

Thermal camera 2 (optional)

. ... . . . . .
Valves communication PC

Figure 4. Optical sorting system based on thermal properties of the separated material.

Table 1 shows a few examples for the mentioned thermal properties for some chosen minerals, Clauser and Huenges (1995). It can be noticed that there is a high variation in the heat conductivity for different minerals, while the specific heat for minerals is rather similar and varies from 850 to 1050 J/(kg*K). As an example, for magnetite and hematite the heat conductivity is 4.61 and 12.42 W/(m*K) respectively. Thus, similar colour materials can be separated using this technique. Thermal properties of a material can be very much affected by moisture content. Therefore, it is necessary to consider this parameter especially during separation of rocks and minerals coming directly from a mine.

45

Table 1. Examples for the heat conductivity for chosen minerals. Material Heat Conductivity [W/mK] Muscovite 3.89 Feldspar 2.34 Quartz 6.5-11.3 Magnetite 4.61 Hematite 12.42 Ilmenite 1.49 Corundum 17.7 Pyrite 23.15 Galena 2.76

4.3 Multiple separation stages General optical separation techniques are used in single operating stages. It means the optical separator is often used as an important part of the system and the process is carried out in a single stage. In many cases it is not economical and brings a lot of confusion about optical sorting generally. The result is often not satisfactory when compared to the operating and investment cost. This problem can be solved by a multiple stage separation procedure. Looking at other separation techniques like flotation, gravity separation, magnetic separation, it can be noticed that many of them are applied in multiple stage configuration thus providing a satisfactory result. Optical separation is based on different scanning techniques, which very often perform a single check by which a particle is qualified as a waste or a product. It is therefore very probable that some of the scanned particles can accidentally be analyzed in a wrong way and consequently be separated to a wrong fraction. Multiple separation can solve this problem when there is a extremely small chance that such particle is separated to a wrong fraction in two or three separation stages. The example of the multiple stage separation is shown on Figure 5 where the quartz material has been processed. The feed material contained 65% of the quartz and 35% of different impurities in form of various minerals. The first separation stage was adjusted to very strict separation criteria (high sensitivity threshold) to remove maximum of the impurities from the quartz fraction. The product had a purity of 99.8% which represented a very good result. However, there was a lot of quartz

46

particles also removed from the feed material as the waste because some of the good quartz particles were analyzed as impurities.

Feed material 65% quartz and 35% impurities

Optical Separator A

Product fraction 99.8% quartz and 0.2% impurities

Waste fraction 22% quartz and 78% impurities

Optical Separator B
Final waste fraction 3.5% quartz and 96.5% impurities

Product fraction 99.5% quartz and 0.4% impurities

Final product fraction 99.7% quartz and 0.3% impurities

Figure 5. An example of optical sorting of quartz in multiple stage configuration.

The waste fraction still contained 22% of the quartz particles. In such cases, due to different light reflection conditions or inter-particle collisions, some quartz particles could be detected as the waste fraction. However, it is an extremely low probability that these particles can be analyzed in a wrong way twice. Therefore when the second purification separation stage was applied to the waste fraction a lot of quartz particles were recovered back to the main product fraction. It resulted in very improved and robust separation process with limited losses of quartz particles. The final waste fraction contained only 3.5 % of quartz particles. The similar separation configurations can be applied for the product fraction to provide further purification of the clean fraction, depending on the final purpose and process requirement. Finally, if necessary both fractions from the first separation stage can be processed by the next separators providing very high separation efficiency, high purity and high recovery. In any separation technique employing multiple stages, the cost of the separation equipment will play a decisive role for any process. It is therefore of critical importance that the separation equipment is not cost prohibitive. Optical separation 47

equipment provided by Comex fulfils this main cost requirement for multiple stage separation keeping the total cost at a competitive level.

5.

Conclusion The optical separation provides new potential for continuous separation of particles having different size, colour, shape and pattern. Application areas for such a process are almost unlimited (minerals separation, recycling of metals, plastics, paper, rubber, wood etc.). It can be applied to the mineral industry in many different processing steps. Very significant savings are achievable when the separated material, representing rejects, can be removed from the process in the early stage. By shape and colour analysis the material composition can be estimated indirectly. The similar function can be applied to the recycling area. This may result in new possibilities for optimising processing circuits. Optical sorting technologies must conform to two basic requirements. High separation efficiency is unquestionably the most important requirement. This can be achieved today with a satisfactory result by multiple stage separation and still more advanced systems are under testing. However, such systems must be in the agreement with the second requirement - the investment and installation cost. Efficient separation of materials often requires multiple stage processing. The cost of the separation system from Comex is very competitive when compared to the other ones available on the market today thus providing new possibilities for high efficiency optical separation.

References Clauser C., Huenges E., 1995. Thermal Conductivity of Rocks and Minerals. In: A Handbook of Physical Constants, American Geophysical Union, pp. 105-126 Kolacz, J., 1998. Description of optical sorting systems. In: SINTEF reportConfidential Kolacz J., Chmelar J., 2001. Separation of metals. In: SINTEF report-Confidential. Kolacz, J., 2001. Separation of aluminium particles using X-ray systems and thermal cameras. In: SINTEF report-Confidential Wortuba H., Jungst W., 2000. Optoelecronic Separation Process for the Sand and Gravel Industry. In: Aufbereitungs Technik 41-2000, pp. 71-79 British Geological Survey. 2005. World Mineral Production 2001-2005. 48

NATURAL REFINEMENT OF SOLAR GRADE SILICON RAW MATERIALS Rune B. Larsen, Bjrn E. Srensen & Franck Poitrasson Dep. of Geology, Norwegian University of Science and Technology; Peter M. Ihlen, Geological Survey of Norway

Introduction The single most important component in the production of solar panels is polycrystalline silicon. And not just any type of silicon but a variety with nearly now impurities. Solar grade silicon (SoG-Si) is produced from quartz raw materials during the reaction with carbon under the formation of metallurgical grade silicon (MeSi). Both quartz and carbon is dirty and, therefore MeSi contains thousands of ppm of impurities. Given this inferior raw material, an expensive refinement process (the Siemens process) is required to make SoG-Si. Before 2003, it was possible for the solar cell industry to acquire SoG-Si from scrap metal left over by the silicon chip industry. The actual costs of this electronic grade Si was 50-60 USD/kg but the solar cell industry paid only 20-25 USD for the scrap metal that was too poor for electronic components but more than sufficient for SoG-Si. Already in 2001, it was clear that scrap metal production did not keep up with the exponential growth of the solar cell marked and that a shortage was imminent (Fig. 1). The marked predicted an annual growth of first 15 later 30 % per annum and the solar cell industry, somewhat suddenly realised that the entire marked could collapse out of raw material shortage. Either, the industry had to produce their own silicon at a much higher cost, which would also dampen the marked growth, or they had to develop new technologies to produce inexpensive SoG-Si.

Figure 1 Many new technologies are currently tested but it is beyond the scope of this study to go through all these exciting options. As geologists and processing engineers we became aware of the SoG-Si shortage and in 2001 we decided to follow another avenue in a cross-disciplinary project including geologist, processing engineers and material scientists. The task was to gain a better understanding of 49

the genesis of high purity quartz for the production of SoG-Si. Could it be that natural geological processes did most of the work that today was done in expensive industrial refinement processes and are we able to predict where in the geology these processes are optimalised.

Figure 2. Accordingly. the task was to find a quartz deposits where the quartz raw materials were more expensive, however, where simple low cost processing would produce quartz that in itself had solar grade quality (Fig. 2, yellow route). In other words, it did not matter if the quartz raw material had a much higher price, i.e. 1-2 USD/kg rather than 5 cents, if the very expensive Siemens process (or similar approaches) could be avoided (Fig. 2). The goal was to find quartz with nearly no P and B and a total trace element concentration of max 40 ppm before simple processing such as crushing, magnetic separation, calcination and acid leach reduced the trace element contamination to less than 20 ppm. Geology of quartz After the feldspars, quartz is the most abundant mineral in the Earth crust and it is taken for granted that we can not learn much more about this simple silicate. However, when we began our project it was soon apparent that we neither knew anything about the speciation, nor the concentration of trace elements in quartz; certainly not as a function of the quartz forming environment.

50

Figure 3. The well structured atomic lattice of low-quartz (Fig 3 u. left) (the most common polymorph of quartz) does not accept more than perhaps 1000 ppm of foreign elements in to its lattice structure. When higher concentrations are encountered, the impurities are normally present in microscopic or sub-microscopic inclusions of saline fluid inclusions or minerals. Foreign ions are only included in the crystal lattice if they have near matching charge and ionic radius of the Si4+ ion (Fig. 3). Most commonly, Ti and Ge, both being tetravalent, are included as single substitutions (Fig. 3, u. right). Al3+ Fe3+ and B3+may also substitute for Si. However, having one charge to little they are either charge balanced by pentavalent P5+ at a neighbouring Si-site (Fig 3. l. left), or alternatively, they are charged balanced by a monovalent ion such as Li+, Na+ and K+ (Fig. 3. l. right). The monovalent ion is too large for the quartz lattice but may slip into a neighbouring lattice defect that has left a larger atomic cavity in the crystal lattice. Together, we need to understand the incorporation of Ti, Ge, Al, Fe, B, P, Li, Na and K in order to predict promising geological settings. There may be other elements, but our studies showed that they are inferior to these nine elements.

51

Figure 4. Quartz in the magmatic-hydrothermal continuum The most fertile quartz-forming environments belong to the magmatic hydrothermal continuum of granitic rocks. We determined the trace element concentration in quartz from several hundred geological formations covering T and p of formation from 900 to 200 oC and 7 to 0.5 kb pressure. Many different igneous settings were analysed, but investigations of the 7 km thick layered Kleivan granite in S-Norway (Fig. 4), summarises many of our results. In this rare example of a layered charnokitic granite (i.e. orthopyroxene bearing granite), 6 distinctive zones formed from 840-900 oC in the bottom to c. 600 oC in the top of the

52

intrusion. Most of the crystallization occurred at dry conditions but the upper third of the granite crystallised in the presence of saline aqueous fluids.

Figure 5. Ti and Ge are the elements that show the most consistent behaviour throughout all the igneous settings that were studied. So also in the Kleivan granite (Fig. 5) where the concentration of Ti decreased along a well defined slope from c. 100 ppm at 840 oC to less than 10 ppm at 600 o C. Ge increased from 0.5 to 2.5 ppm in the same T-range. Accordingly, the Ge/Ti ratio of quartz in any igneous environment is increasing as the temperature becomes progressively lower.

Figure 6. Accordingly, the Ge/Ti ratio is a qualitative index of both the temperature and degree of differentiation of the studied igneous system. If Al in quartz is plotted against this index (Fig. 6), we can see for the Kleivan granite that Al is increasing from c. 100 ppm to more than 500 ppm when going towards lower temperatures. The Kleivan granite is peraluminous, hence is charcterised by silicate melts that becomes progressively more enriched in Al. When we analyze quartz formed from metaluminous melts, we see an evolution towards Al-poor quartz and the total concentration of Al is generally much lower.

53

Figure 7. Phosphorous is one of the most unpredictable trace elements in quartz. In the Kleivan granite, it begins to decrease from high to moderate temperatures (Fig. 7). However, at lower T the concentration begin to rise again to a maximum of 25 ppm. At the lowest concentration, it is known that the melt is no longer saturated with apatite. Accordingly, the melt is not any longer buffered by this phosphate and, we assume, the P-content of the melt begin to increase, hence P in quartz also goes up. In metaluminous granites, the P-content of quartz is generally higher than in peraluminous granites. However, the lowest concentrations are found in granitic pegmatite formed from metaaluminous melts. Granitic pegmatites are commonly known for several conspicuous phosphates, for example monazite, and we predict that monazite, typically forming early during pegmatite genesis, efficiently vacuum the melt for P.

Figure 8. Finally, Figure 8 shows the evolution of Li in quartz as an example of the behaviour of one of the monovalent ions. During early crystallization of the Kleivan granite, the Li concentration in quartz is decreasing because amphibole is removing Li from the silicate melts. When this Li-sink cease to form, the Li concentration of quartz goes up to a certain point and, subsequently begin to fall. At the inflection point, the presence of primary aqueous fluid inclusions, imply that the silicate melt is silica oversaturated. Li, together with the other monovalent ions are hydrophile hence is strongly absorbed by the aqueous fluids. Accordingly, the concentration of Li in the silicate melt is decreasing and, therefore, quartz that crystallise in equilibrium with the melt is also Li-poor. Altogether, our studies imply that there is simple buffering relationship between the composition of granitic melts and the quartz that formed from the melts.

54

Figure 9. Aqueous fluids are important during terminal crystallization in attracting the undesired monovalent ions. However, in going from the magmatic to the hydrothermal part of the continuum, aqueous fluids may further refine the quartz raw materials. This is done through a process of hydrothermal alteration of previously formed igneous quartz. Figure 9 shows an extreme example where magmatic aquoues fluids infiltrate primary magmatic quartz with very high concentrations of Ti, Al and K. Quartz 1 (Qz1) which is the primary magmatic quartz, is gradually recrystallized to quartz 2 (Qz2) and quartz 3 (Qz3) by very agressive alkaline aqueous fluids. Ti, that normally is immobile when first incorporated in to the quartz

55

lattice, is decreasing from a plateau around 200 ppm (Fig. 9) to less than 10 ppm and Al is experiencing similar depletion.

Figure 10 At other localities we may not se such a drastic change in the Ti-concentration, however, we nearly always se that monovalent ions, B (Fig. 10), Fe and sometimes Al are efficiently removed from the quartz lattice under the production of high purity quartz varieties. Formation of high purity quartz deposits We conclude that peraluminous granites, form quartz with high Al concentrations, metaluminous granites form quartz low in Al and perakalline quartz is high in K. Igneous differentiation form quartz with low Ti, Fe, K and B contents. Ge and Li is increasing. P and Al may also increase. Largely, the quartz composition is buffered by the melt composition. Li, B, Al and Fe is lowered during hydrothermal infiltration processes whereas Ti, Ge and P normally are immobile during recrystallization unless the altering fluids are strongly alkaline. In general terms, monovalent ions are very mobile due to their loosely bonded character in the quartz lattice, whereas tetravalent ions require more aggressive fluids and partial recrystallization of quartz in order to be removed. Adding all these observations together the most promising geological setting for the formation of solar grade quartz is granitic pegmatite and particularly metaluminous and hydrothermally infiltrated pegmatites

56

Using this model, we managed to find several pegmatites in Norway that meet or at least approach SoG-Si bearing quartz. Another important conclusion is that only 10-20 pegmatites in a pegmatite field counting hundreds of pegmatites, had to be studied before it was safe to say if the pegmatites may or may not carry solar grade quartz.

Figure 11. Two examples are shown in figure 11. Here, A is processed high purity quartz from the famous spruce pine pegmatites in USA, B is processes quartz from Drag in Norway where high purity quartz is currently produced and C and D, comprises to new localities that we found. Before processing these two pegmatites, with less than 20 ppm total impurities, comprises quartz having a better quality than processed high purity quartz from current productions.

57

MODELLERING OCH REGLERING AV EN ZINKFLOTATIONSKRETS Jens Pettersson

ABB Corporate Research, V aster as
<jens.pettersson@se.abb.com>

Henrik Lindvall
Boliden Mineral AB, Garpenberg
<henrik.lindvall@boliden.com>

SAMMANFATTNING r en av dom vanligaste processerna f r s Flotation a or anrikning av malm. Processen a a pass verordnad reglering ofta a r o nskv viktig f or ett gruvomr ades ekonomi att en o ard f or att stabilisera processen. Inom Boliden Mineral AB har man sedan b orjan av 1970-talet i ett antal olika projekt arbetat med otationsstyrning. Tidigare har man, liksom mineralindu vrigt, arbetat mycket med regelbaserade metoder. Inom reglertekniken har dock ett strin i o ren, man har mer och mer b verge regeltrendbrott kunnat sk onjas under de senaste a orjat o vergripande reglerstrategier baserade metoder till f ordel f or modellbaserade metoder d ao tagits fram. Boliden anrikar en malm med betydande zinkinneh all i Garpenberg. Den bentliga zink r d kretsen d ar togs i drift under 2005 och a armed v alutrustad med reglering f or reagenstillsatser, lufttillsatser och skumh ojder. Dessutom har kretsen m anga pulpr ontgenanalyspunkter. I detta bidrag presenteras en styrstrategi f or en otationskrets som anv ander modellbaserad prediktionsreglering (MPC). En dynamisk modell av kretsen har tagits fram med hj alp av avancerade systemidentieringsmetoder. Modellens systemparametrar har tagits fram genom att excitera utvalda styrsignaler under l ang tid. Regulatorn har implementerats i ett industriellt styrsystem (Expert Optimizer) och testk orts vid era tillf allen p a Garpenbergs zinkotatoinskrets. 1. Introduktion r modellering och o vergripande reglering av otation ett Inom mineraltekniken a mne. Flotation a r ju en av de vanligaste metoderna f intressant a or anrikning, samtidigt p averkas ett gruvomr ades ekonomi direkt av otationsprocessens resultat. M anga ansatser till att ta fram otationstyrningar har gjorts, men f a riktigt lyckade f ors ok har rapporterats. Fr an 1990-talet och fram at har m anga av de styrstrategier som tagits fram under dessa projekt varit regelbaserade (fuzzy logic). Inom industrin i allm anhet har l aget varit ungef ar det samma, d a man arbetat med verordnad reglering har man ofta arbetat med regelbaserade metoder. Under de o ren har dock ett trendbrott kunnat ses, inom reglertekniken har man mer senaste a verge de regelbaserade metoderna till f och mer b orjat o ordel f or olika typer av modellbaserade metoder. r ett mer direkt s vergripande styrstrateModellbaserade metoder a att att ta fram o n regelbaserade metoder. En liknelse kan vara att med regelbaserade metoder gier a f ors oker man efterlikna operat orer, medan man med modellbaserade metoder f orst beskriver processen matematiskt, f or att sedan ta fram en styrstrategi med hj alp av den matematiska modellen.
59

1.1

Avancerad processreglering inom Boliden Inom Boliden Mineral AB nns en l ang tradition av att anv anda avancerad processreglering i anrikningsverken. Under b orjan av 2000-talet studerades exempelvis niv areglering av seriekopplade otationstankar i Aitik. En LQ-regulator som ven parametrar kan hantera alla skumniv aer i en otationsserie designades, och a f or s arkoppling med PID-regulatorer togs fram [1]. Alla kvarnkretsar i Bolidens anrikningsverk har v al utvecklad reglering. Regleringen av kvarnkretsen i Bolidenverket arbetar exempelvis aktivt med variabler s asom fyllnadsgrad, last, varvtal och effekt [2]. verordnade otationsstyrningar. ReInom Boliden har mycket arbete gjorts med o dan under b orjan och mitten p a 1970-talet togs en regelbaserad otationsstyrning fram vid Kristinebergs anrikningsverk [3]. Under mitten och slutet av 1990-talet arbetade Boliden tillsammans med CISA fram en otationsstyrning i Aitik. D ar togs en processmodell fram, och Kalmanlter anv andes f or att uppskatta processtillst anden. Tillsammans med ett expertsystem och fuzzy-logic utgjorde detta basen f or en regelbaserad styrstrategi f or r aotationen i Aitik [4]. Under 2007 studerades modellering av otationen i Garpenberg [5]. I samarbete med ABB har arbetet i Garpenberg sedan fortskridit, denna artikel presenterar delresultat fr an detta arbete.

1.2

Syfte Motivationen till en otationsstyrning nns i m ojligheterna att stabilisera proces r m sen. Detta resulterar i f orb attrade eknomiska resultat. Vidare, om det a ojligt att reglera processen p a ett tillfredst allande s att nns m ojligheten att b orja titta p a optimering av processen.

2.

Processbeskrivning I Garpenbergsmalmen nns v ardemineraler som inneh aller fr amst zink, bly och r s silver, men ocks a en del koppar och guld. Processuppkopplingen a adan att alla v ardemineraler utom zinkmineralen oteras i en kopparblykrets, sedan erh alls en koppar- respektive en blyprodukt i en separationskrets. Zinkmineralen, i Garpenbergs fall zinkbl ande (ZnS ), utvinns sedan i en zinkkrets, se gur 1. Denna processuppkoppling g or zinkkretsen l amplig f or expriment, d a f or andringar (f or att kunna modellera processen kr avs att man g or f ors ok av olika slag) i zinkkret vrigt. sen inte p averkar processen i o Efter en ombyggnation togs en ny zinkkrets i Garpenberg i drift under oktober 2005, kretsen best ar av nio otationsceller. Fem 40 m3 Outotec tankceller utg or 3 r aserie/scavenger, medan repeteringen utg ors av tre 20 m Outotec tankceller och 3 en 15 m Metso-cell. Processchemat presenteras i gur 2. D ar ses att kretsen har r det Bolidens egenutvecklade pulpr ontgenanalyser p a sju punkter, naturligtvis a BoxRay som anv ands. En m atning f or BoxRay tar cirka 30 s, vilket betyder att h ogprioterade m atpunkter kan analyseras ungef ar var 5:e minut, medan analyser
60

 Figur 1: Oversiktlig processuppkoppling i Garpenbergs anrikningsverk.

Figur 2: Processchema som beskriver zinkkretsen i Garpenberg. Analyspunkter r markerade. f or pulpr ontgen a r s g ors ungef ar var 10:e minut vid andra m atpunkter. Tack vare att kretsen a a r den a ven i o vrigt v pass modern a alutrustad med m atningar. I kretsen nns tre r varje cell utrustad med en luft volym odesm atningar, dessutom a odesregulator, r utrustad med en skumniv och varje bank a aregulator. Kalk anv ands f or reglering av pH och skumolja1 anv ands f or att stabilisera skumt acket. Tv a reagenser anv ands f or styrning av kretsen: kopparsulfat (CuSO4 , som aktiverar zinkbl ande) och samlare, Ibux2 .

1 2

Diethylenglycol monobutyl ether > 80 % and polypropylenglycol methyl ether < 21 % Natrium isobutyl xantat

61

2.1

Operat orernas strategi f or drift av zinkotationen nd Aven om olika skift styr zinkotationen p a olika s att kan a a n agra generella regler f or hur processen k ors presenteras. De variabler som operat orerna har r: tillg angliga f or styrning av zinkkretsen a Ibux Samlartillsatsen kvotas in mot m angd inkommande zink i kretsen (alterna ndras ofta och anv tivt m angd inkommande malm). Kvotningen a ands aktivt av operat orerna. ndras ocks CuSO4 Kopparsulfaten a a ofta och anv ands aktivt. Ingen kvotering anv ands dock. r ofta konstant o ver l Skumolja Skumoljetillsatsen a angre perioder. Tillsatsen kan dock exempelvis ibland minskas vid problem med slutkoncentrathalten. ndras s Skumtjocklekar Skumtjocklekarna a allan, detta g aller b ade f or r aserie/scavenger och repeteringsstegen. En LQ-regulator enligt [1] anv ands i r aserien. ndras ganska ofta. LuftLufttillsats i r aserie Lufttillsatsen i r aserie/scavenger a tillsatsen minskas exempelvis vid problem med skumstabilitet och vid d aliga kas vid s slutkoncentrathalter. Lufttillsatsen o amre utbyten. Lufttillsats i repetering Operat orerna har tillg ang till en reglerloop som styr den ver ommalningen med hj cirkulerande lasten o alp av lufttillsatsen i repete r l nd ringen. Regleringen a angsam men upplevs a a som v ardefull av operat orerna.

3.

Prediktionsreglering Inom avancerad processreglering har modellbaserad prediktionsreglering (MPC, fr an engelskans Model Predictive Control) blivit en vanligt f orekommande reglermetod [6]. Som namnet antyder anv ander sig metoden av en modell av den styrda processen f or prediktering av processens framtida beteende givet dels de styringrepp som kommer att g oras, dels det tillst and som processen benner sig i just nu. De framtida styringreppen ber aknas genom att l osa ett optimeringsproblem d ar en specik kostnadsfunktion minimeras. Aterkopplande reglering uppn as genom ver tiden l att regelbundet o osa ett nytt optimeringsproblem d ar h ansyn tas till hur processen har svarat p a de styringrepp som gjorts. Enligt beskrivningen ovan spelar matematiska modeller en stor roll i anv andandet av MPC. Vanligtvis anv ands tidsdiskreta dynamiska modeller f or att beskriva sam r av intresse. I banden mellan styrsignaler och de utsignaler fr an processen som a detta projekt har modeller p a tillst andsform anv ants, viz: x(t + 1) = Ax(t) + Bu(t) + Bd d(t) y (t) = Cx(t)
62

(1) (2)

 r en vektor med styrsignaler, y (t) a r en vektor av utsignaler (m d ar u(t) a atsignaler), r framkopplade signaler och x(t) a r tillst r mad(t) a andsvektorn. A, B , Bd och C a r tillst triser av l amplig storlek. Trots den relativt enkla notationen s aa andsmodeller kraftfulla hj alpmedel f or att beskriva komplexa multivariabla system och processer. Tillst andsmodeller har med stor framg ang anv ands inom olika typer av re r dock glerapplikationer. Anv andandet av tillst andmodeller f or MPC-reglering a relativt nytt men har under det senaste artiondet snabbt vunnit terr ang som den mest generella modelleringsmetoden, se [6] f or en mer utt ommande beskrivning av tillst andsbaserad MPC. r f En stegvis beskrivning av MPC-algoritmen a oljande: 1. Observera processen vid tidpunkt t och r akna ut processens tillst and x(t) = x0 . 2. Ber akna en sekvens av styrvektorer Ut = [u(t), . . . , u(t + Nc )] genom att minimera ett kriterium J (Ut ). 3. Styr ut u(t). 4. V anta tills tidpunkt t + 1, s att t := t + 1 och g a till steg 1. r En vanlig matematisk formulering av optimeringskriteriet i MPC-algoritmen a f oljande:
Np Nc

J (Ut ) = min

u(k )

||y (t + k ) ysp ||2 Q+

k=1 k=1

||u(k )||2 R

(3)

r att dels straffa de predikterade utsignalerna y :s kvadraSyftet med kriteriet ovan a tiska avvikelse fr an motsvarande b orv arden ysp , dels straffa styrsignals andringar r att anv u(k ) = u(t + k + 1) u(t + k ). Typiskt f or MPC a anda olika horisonter r Np > Nc . Kriteriet f or predikteringen Np och f or regleringen Nc , vanligtvis a r diagonala matriser av l beror ocks a p a vikterna Q och R, vilka a amplig storlek. Q och R anv ands i praktiken som kalibreringsparametrar f or MPC-regulatorn d a st orre v arden p a Q ger snabbare reglering och st orre v arden p a R ger l angsammare reglering. Som bivilkor i optimeringen anv ands dels sj alva processmodellen men ocks a andra begr ansingar p a styrsignaler och utsignaler, enligt f oljande: x(t + k + 1) = y (t + k ) = u(k ) = |u(k )| < u(k ) = uL < u(t + k ) yL < y (t + k ) Ax(t + k ) + Bu(t + k ) Cx(t + k ) u(t + k + 1) u(t + k ) umax 0, k > Nc < uU < yU
63

k = 0, . . . , Np 1

(4)

Startvilkoren vid k = 0 ges av u(t) = u0 x(t) = x0 (5) (6)

Objektsfunktionen (3) bildar tillsammans med ekvationerna (4-6) ett s.k. kvadratiskt program (eng: Quadratic program, QP) med bivilkor, f or vilken en analytisk r det r l osning oftast saknas. Dock a attframt att med hj alp av iterativa numeriska metoder r akna fram den optimala l osningen. Storleken p a QP-problemet beror av antalet tillst and, styrsignaler och utsignaler samt prediktionshorisonten Np . D a Np r ganska stort, s vanligtvis a ag mellan 30-100, s a blir ocks a QP-problemet stort men r l med modern programvara och h ardvara a osningstiden typiskt ett par sekunder. I den ovan beskrivna MPC-algoritmen antas att man har ett numeriskt v arde av r formulerad s processens tillst and, x0 , i varje iteration. Om modellen a a att alla r detta kanske inget problem men ofta har man inte tillg tillst and m ats a ang till alla tillst and utan de beh over ber aknas p a n agot s att. r att anv En metod f or att ber akna (rekonstruera) tillst and a anda en observat or. En r ett dynamiskt system p observat or a a formen: x (t|t) = x (t|t 1) + Kf (y (t) C x (t|t 1)) x (t + 1|t) = Ax (t|t) + Bu(t) + Bd d(t) (7) (8)

r de uppm d ar y (t), u(t) och d(t) a atta v ardena f or respektive storhet. Matrisen Kf multipliceras med ber akndade prediktionsfelet y (t) C x (t|t 1) f or att f a fram den korrigerade prediktionen x (t|t). Valet av Kf har betydelse f or det slutna systemets beteende, vilket i praktiken g or den till en extra designparameter, se [6]. I MPC-optimeringsproblemet ovan anv ands sedan x0 x (t|t) som startv arde i ekvation (6). 3.1 Implementering F or att kunna anv anda MPC-reglering on-line st alls det speciella krav p a det styrsystem som anv ands. F orst och fr amst m aste det klara av att l osa optimeringsproblemet ovan, det m aste ocks a till ata att man implementerar observat orsekvationerna. Inom ABB anv ands SCADA (Supervisory Control and Data Aquisition) systemet Expert Optimizer (EO) f or MPC-baserade l osningar. EO erbjuder st od och m ojligheter f or att b ade l osa de rent matematiska problemen men ocks a f or hantering av m atdata, larm och fellogik. Det g ar ocks a att g ora presentation och tkomliga via en standard webl anv andargr anssnitt, alla enkelt a asare. Ingenj orsverktygen i EO erbjuder en drag-och-sl app-baserad utvecklingsmilj o d ar man graskt kongurerar sina algoritmer med hj alp av olika funktionsblock. Data och informations odet mellan EO och det bentliga styrsystemet sker via en
64

OPC-koppling3 . Reglerl osningarna som implementeras i EO k ors direkt on-line mot processen. Eventuella fel hanteras automatiskt utan att p averka andra delar av systemet vilket s akerst aller en robust och p alitlig funktion. 4. Processmodellering Anv andandet av MPC kr aver en relativt bra matematisk modell av processens dynamiska betende. Det nns naturligtvis olika s att att h arleda en s adan modell, t.ex. kan man t anka sig att fr an en detaljerad olinj ar processmodell g ora en linj arisering kring en (eller era) arbetspunkter och i MPC-algoritmen anv anda den modellen. I fallet med zinkotationskretsen s a har det visat sig vara sv art att f a fram en bra fysikalisk modell vilket motiverar anv andandet av empiriska modeller, framtagna med hj alp av systemidentieringsmetoder [7]. Dessa metoder bygger p a att man med hj alp av data fr an processen identierar processdynamiken baserat p a statistiska samband i m atdata. r det viktigt att de m F or att kunna g ora bra modeller av processen a atv arden som anv ands f or systemidentiering inneh aller relevant information. Vanligtvis g aller det att m atv arden som samlas in under normal drift inte inneh aller den typ av information om processens betende som beh ovs f or att kunna identifera bra modeller. D arf or brukar det vara n odv andigt att under en period se till att processen exiteras tillr ackligt f or att f a fram denna information. r att o verlagra de normala styrsignalerna Ett vanligt s att att excitera processen a med en excitationssignal med l ampligt frekvensinneh all. P a s a s att s akerst aller man att de signaler som sedan anv ands f or identiering inneh aller information samtidigt som man undviker att l ata processen driva iv ag alltf or l angt fr an sitt normala l age. Denna typ av f ors ok har vid era tillf allen har utf orts i Garpenberg under perioden april-juni 2007. F or att f a tillr ackligt bra data kr avdes det p a grund av processens tskilliga timmar (12-16 h). relativt l angsamma dynamik att f ors oken p agick under a Den l anga f ors okstiden gjorde dock att vissa f ors ok blev misslyckade d a processen st ordes av olika yttre st orningar, men efter en tids idogt utf orda f ors ok fanns det tillr ackligt bra data f or att ta fram anv andbara modeller. I gur 3 visas n agra signaler fr an ett av f ors oken. I princip har alla exitationsignaler sett ut p a liknande s att, dvs en sekvens av steg andringar vid olika tidpunkter. Vanligtvis har tiden mellan steg andringarna leget i intervallet 30-70 minuter, vilket bed omdes vara ett rimligt frekvensomr ade med tanke p a den totala systemdynamiken. 4.1 Identiering av modeller Med hj alp av m atdata insamlat under f ors oken har olika typer av matematiska modeller identierats. M atdatat samplades med 5 minuters sampeltid f or att f a cirka
r en processindustristandard f OPC (OLE for Process Control) a or kommunikation mellan olika system
3

65

 Figur 3: M atdata fr an f ors ok gjorda 28:e juni 2007. Overst: Zinkhalt i avfall fr an andra repeteringssteget. Nedan: Tillsatt m angd kopparsulfat. F or b ada kurvorna r samplat med 5 minuters sampeltid. Tiden p har medelv ardet dragits bort. Data a a r i minuter fr x-axeln a an experiments start. en BoxRay (pulpr ontgenanalys) m atning per m atpunkt, f or att sedan importeras in till Matlab f or analys. I Matlab har System Identication Toolbox anv ants f or att identiera modeller. B ade enkla modeller med en utsignal och era insignaler samt mer komplicerade modeller med era utsignaler har tagits fram och utv arderats. Som kriterium f or att evaluera modellernas godhet har dels deras f orm aga att prediktera utsignalerna, verens med den allm dels hur v al de st ammer o anna uppfattningen om processens beteende. T.ex. s a har modeller d ar f orst akningarna haft fel tecken eller d ar de skattade tidskonstanterna varit alltf or l anga eller korta f orkastats. Som stysignaler (u) har f oljande storheter anv ants (Nm3 st ar f or normalkubikmeter): u1 : summan av luft odena till r aserie [Nm3/min], u2 : summan av luft odena till scavenger [Nm3/min], u3 : summan av luft odena till repetering [Nm3/min], u4 : ode av Ibux [l/min],
66

u5 : ode av kopparsulfat [l/min], vilka motsvarar de signaler som exiterats under de styrda f ors oken. Till utsignaler valdes: y1 : zink i avfall fr an andra repeteringssteget, y2 : zink i avfall fr an r aserie (logaritmerad), y3 : zink i avfall fr an scavenger (logaritmerad), y4 : zink i avfall fr an koppar-bly-kretsen, y5 : j arn i avfall fr an koppar-bly-kretsen. r att dessa norAnledningen till att anv anda logaritmerade signaler f or avfallen a r ganska sm malt a a men strikt positiva. D a en linj ar modell vars utsignaler kan bli negativa ansatts brukar det vara bra att ist allet f or den ursprungliga signa r transformen len anv anda sig av en transformerad signal. I det h ar fallet s a a y = ln(ym ) speciellt l amplig d a den ger b ade positiva och negativa v arden. Processen p averkas ocks a av halten av zink och j arn i den malm som matas in r naturligtvis inte styrbar i sig men genom att i otationsanl aggningen. Halten a anv anda m atv arden fr an provtagningen kan variationer i zink-halterna predikteras. D arf or har m atningen av zink och j arn i malmen in tagits med i modellen som framkopplingsvariabler: d1 : zink till ing aende otation, d2 : j arn till ing aende otation. Sammanfattningsvis har vi ett system med 5 styrsignaler, 5 utsignaler och 2 framkopplingssignaler f or vilken en tillst andsmodell identieras. Efter att ha utv arderat olika modeller och modellstrukturer visade det sig att de b asta modellerna erh olls d a man med hj alp av lite insikt om processen struktur r m ordnade signalerna i tillst andsmodellen (1-2) s a att alla tillst and xj (k ) a atbara, det vill s aga x = y och C = I . P a s a s att kunde varje rad j i tillst andsmodellen skattas var f or sig fr an f oljande modellstruktur:
Ny Nu Nu

xj (k + 1) =
i

Aij xi (k ) +
i

Bij ui (k ) +
i

Bd,ij di (k )

(9)

vars parametrar Aij , Bij och Bd,ij skattades med hj alp av arx-rutinen i System Identication toolbox.
67

5.

Reglerf ors ok Baserat p a den identierade modellen har f ors ok att reglera zinkotationskretsen i Garpenberg genomf orts under januari 2008. Under f ors oken har b orv arden angetts f or zinkhalter i avfallen fr an andra repeteringssteget och r aserien och som stysignaler anv andes luftm angden till repeteringen (u3 ) och Ibux- odet (u4 ). MPCn regulatorn har till en b orjan trimmats f or att hellre bete sig robust och stabilt a f or att pressa prestanda. I gur 4 visas resultatet av en testk orning med reglering fr an den 16:e januari 2008. Som synes lyckas regulatorn att h alla halterna relativt stabila trots varierande halt av zink i den ing aende malmen. Vid andra tillf allen har dock regleringen inte fungerat lika bra, framf or allt har begr ansningar i m angden reagenser som kan tills attas utan processproblem gjort att regulatorn f att det sv art att h alla de inst allda b orv ardena. Vidare testning och utveckling kommer att kr avas f or att kvantiera regulatorns egenskaper vid olika f oruts attningar.

Figur 4: Resultat fr an testk orning av otationsregulatorn. Fr an ovan: Zinkhalt i avfall fr an andra repeteringssteget (y1 ), zinkhalt i avfall fr an r aserie (y2 ), luft ode till repetering (u3 ), ibux- ode (u4 ) samt zinkhalt i ing aende malm (d1 ). Notera att halterna h aller sig stabila trots den minskande halten av zink i ing aende malm. Vid tidpunkten 400 intr affade ett kortare stopp i matningen av malm, d arav den snabba minskningen av zinkhalt i avfallet fr an r aserien.
68

6.

Sammanfattning och slutsatser I detta bidrag har resultat fr an ett projekt med syftet att reglera en zinkotationkrets redovisats. MPC-reglering har anv ants f or sj alva regleringen, baserat p a dynamiska modeller av processen identierade fr an m atdata. Regulatorn har r det naturligtvis f testk orts med relativt positivt resultat, dock a or tidigt att uttala sig om dess anv andbarhet i ett l angre perspektiv. tg Det ska ocks a p apekas att under projektets g ang har det gjorts olika a arder och f orb attringar vad betr affar basinstrumenteringen i otationsanl aggningen. Nya m atgivare har ocks a satts in i processen och era basreglerloopar (PID-regulatorer) ndrats. har trimmats och i vissa fall har bentliga reglerstrategier a Referenser [1] B Stenlund and A Medvedev. Level control of cascade coupled otation tanks. Control Engineering Practice, 10:443448, 2002. [2] Torbj orn Viklund, Johan Albertsson, Jan Burstedt, Marie Isaksson, and Jens S oderlund. Evolution of AG mill control system at Boliden Mineral AB. In SAG Conference, Vancouver, 2006. [3] Mikael Larsson. Styrning av otationen i Aitikgruvan. Civilingenj orsprogrammet elektroteknik, Lule a Tekniska Universitet, Institutionen f or Systemteknik, Avdelningen f or Reglerteknik, 2008. [4] Bj orn Johansson, Bj orn Bergmark, Olivier Guytot, Christophe Bouch e, and Alain Broussaud. Model-based control of aitik bulk otation. In SME Meeting, Orlando, March 911 1998. Paper presented at the annual SME Meeting. [5] Henrik Lindvall. Flotation modelling at the Garpenberg concentrator using Modelica/Dymola. Civilingenj orsprogrammet teknisk fysik, Uppsala Universitet, June 2007. [6] Jan Maciejowski. Predictive Control with Constraints. Pearson Education Limited, Prentice-Hall, Essex, United Kingdom, 2002. [7] Lennart Ljung. System IdenticationTheory for the user. Prentice Hall, New Jersey, USA, 1987.

69

A characterising of the ore minerals due to mineralogical, chemical and textural properties in Malmberget. Cecilia Lund 1, Olof Martinsson 1 Lule University of Technology1, 971 87 Lule, Sweden (e-mail: cecilia.lund@ltu.se) Abstract This study was an attempt to find a way of characterising an iron ore body both mineralogical and textural in a quantitative manner by using analytical methods like optical microscopy, microprobe (EMPA) and an automatic SEM based system, Particle Texture Analysis (PTA). The source of this study is an iron ore body, called Fabian, located in Malmberget, Sweden. Two types of ores were identified and analysed in this study named orebreccia and ore. The Particle Texture Analysis was made on two fractions of crushed ore. The mineralogy was evaluated and characterized as mineral liberation and mineral association. Magnetite has a simple outline and straight grain boundaries and the gangue minerals have a finer particle size with a more complicated texture. The liberation of magnetite in ore and ore breccia is high. The ore quality for both ore and ore breccia does have similarities in a process technique perspective. Introduction LKAB is today one of the worlds leading producers of upgraded iron ore products. The customers expect good qualities of the products which demand a detailed knowledge of the processes from mine to customer. To achieve this it is important to get a traceability of ore feed and control over the production process but it will also demand a good knowledge of the raw material. Concept of process mineralogy Through studying the process mineralogy of iron ore it is possible to understand the behaviour of the ore feed during the processing plant and also later on the sintering and pelletising processes. It is needed to know what kind of minerals and textures the ore is consisting of due to the fact that the liberation characteristics are intimately related to the mineralogical texture (Lorenzen and van Deventer 1994). Optical microscopy has traditionally been the instrument used for the identification and quantification for both mineralogical and texturally properties, which is a time consuming process (Petruk 2000). A number of different techniques according to image analyses system based on SEM techniques have been developed during the three last decades for a more rapid quantitative description of mineralogy and particle textures (Jones and Gravilovic 1970; Gottlieb, Wilkie et al. 2000; Petruk 2000; Gu 2003). In this study which is a part of a larger thesis, a determination of the mineral properties of one specific iron ore body from LKAB Malmberget was carried out. This was an attempt to find a way of characterise the mineralogy and textures of the iron ore body. Three methods were used; optical microscopy, microprobe and automatic SEM based system, Particle Texture Analysis (PTA) (Moen 2006).

71

The Fabian ore body is major Fe resource at Malmberget since the last drill program has shown an expanded ore body. The ore body plunge in ~ 80 to a depth of at least 1250 meters. The length and width is approximately 700 to 900 m and 50 to 150 m respectively. The ore quality is also better then average for the Malmberget deposit, due to the higher Fe-content (fig. 2). Geological setting Regional geology: The geology of the ore province of northern Norrbotten is compound of bedrock sequences with different ages. The oldest rocks are an Archaean granitoid-gneiss basement with dated ages of 2.8Ga. Unconformably overlaying sequences are supracrustal successions of Palaeoproterozoic age 2.5-2.0 Ga, followed by Svecofennian volcanic and sedimentary units, dated c. 1.9 Ga. Forty apatite-iron deposits are known from the northern Norrbotten ore province and they are hosted and probably genetically related to the volcanic rocks in the Svecofennian succession (Martinsson 2004). The two economically most important deposits are mined by LKAB at Kiruna and Malmberget. These two apatite-iron ore deposits have a similar origin and were formed by magmatic-hydrothermal process at 1.89-1.88 Ga (Bergman, Kbler et al. 2001). However, there are major discrepancies in character between them, due to later overprinting by metamorphoses, deformation and granitic intrusions, which are stronger for the Malmberget ore. Deposit geology: More than 20 different tabular to stock shaped ore bodies are known in the Malmberget ore field, spread over an area of 2.5 x 5km. There exist only one generally description of Malmberget deposit written by (Geijer 1930). The Malmberget deposit were probably from the beginning a continuous ore lens which were exposed for at least two phases of folding and metamorphism. These events torn the ore lenses apart by strong ductile deformation and today they occupy a large-scale fold structure were the individual ore bodies stretch parallel to the fold axis, which plunge 40-50 towards SSW (Bergman, Kbler et al. 2001) (fig. 1).

Fig. 1 A simplified geological map of Malmberget. From Bergman et al. (2001).

72

The iron ore minerals are both magnetite (Fe3O4) and hematite (Fe2O3) but the magnetite is more common of the two. Hematite forms several separate ore bodies and portions of others (Geijer 1930). The main gangue minerals are apatite, amphibole, pyroxene, feldspars, quartz and biotite. Among the accessory minerals are pyrite, chalcopyrite, titanite, zircons and calcite most common. Every ore body is characterised by there own mineral, chemical and textural properties. Due to the strong metamorphic recrystallization of the area, the minerals are recrystallised, coarse grained, and elongated in the direction of the lineation of the rocks.

Fig. 2. A schematic picture of the Fabian ore body.

Analytical methods The analytical methods which were used are optical microscopy, microprobe (EMPA) and Scanning electron microscopy (SEM-PTA). Sample preparation For optical microscopy and microprobe, samples of intact ore were used and for the SEM-(PTA) method, the samples were crushed and sieved into sized fractions. Drill core samples from various parts of the ore were taken to characterise different textures. Polished thin sections were made of both intact ore and two particle sizes from sieved fractions of crushed ore (tab. 1). The sized samples were of a larger volume since the sample should be representative for a specific ore type and not only for a specific texture. 73

Ore types: Two types of ore were identified and used in this study, called ore breccia and ore. The ore breccia is largely consisting of magnetite but do have gangue minerals like quartz, amphibole, pyroxene, apatite and feldspars in different proportion occurring as breccia infill in the wall rocks. The ore is more massive magnetite that contains gangue minerals like amphiboles and apatite in fewer amounts. The ore breccia is bordering the massive ore, but occurs also partly as inclusions in the massive ore (fig. 3).
Table 1. The samples which is used in the study.

Sample
Mbgt 6500 100.22-100.30 (AP1) Mbgt 6500 111.00-111.13 (AP3) Mbgt 6500 112.57-112.63 (AP4) Mbgt 6500 374.23-374.35 (B32585) Mbgt 6500 383.21-383.27 (A32584) Mbgt 6500 435.56-435.64 (C32586) Mbgt 6500 128.7-129.8 (Ore breccia) Mbgt 6500 424.53-426.92 (Ore)

Intact ore

150 m

75 m

x x x x x x x x x x

At the mineral processing laboratory at LTU the samples were crushed in a Retsch jaw crush, +3 mm, split by a Jones splitter and sieved with a Ro-Tap shaker at the fraction 150 m and 75 m. Optical microscopy Polished thin section were optical examined in transmitted and reflected light on a standard petrographical microscopy (Nikon Eclipse E600).

Fig. 3. Two ore types. The red spots are defined textures and parageneses.

74

To characterise all the different mineral associations, textures and parageneses, a mineral identification were made of both the silicates and oxides. 23 different spots were analysed on the major minerals in both ore types, ore breccia and ore (Fig. 3) Microprobe Mineral analyses were performed on a JOEL JXA-8500F electron microprobe at NTNU, Trondheim, Norway. For the microprobe analyses an accelerating voltage at 15.0 kV, a probe current at 95 A and a < 1m beam diameter were used. Totally 166 analyses were made, representing both silicates and oxides on the major minerals in 23 spots to cover all different textural and mineral assemblage variations, observed in the samples. Beside this mineral identification, it also verified that the used sized fraction samples were representative. Scanning electron microscopy The particle analyses were performed on a Hitachi S-4300SE scanning electron microscopy equipped with Oxford Inca Feature software NTNU, Trondheim, Norway. For the particle analyses an accelerating voltage at 20.0 kV and a probe current at ~0.5nA were used. Particle texture analysis PTA By using Back Scattered Electron (BSE), images are analysed by means of grey levels and every grain of interest was analysed by X-rays. Every analyzed grain size fraction is imported to the PTA software. Images analysis is performed offline to process and evaluate if grains occur liberated or in composite particles. Standard queries can be performed from the output results in a new database such as the mineral liberation of any mineral, mineral association of any mineral and miniature images of particles of a certain texture category (Moen 2006). To reduce the unclassified group of minerals an extensive identification of minerals and phases for classification should be done. Results Mineralogy The textures of theore breccia is characterised by magnetite grains with a simple outline and straight grain boundaries, either as single grains or as a particles in a matrix of quartz and feldspar. This matrix has a myrmecitic texture and the grains shows anhedral granular outline with complicated grain boundaries (fig 4).

75

Fig. 4. Photographs of the mineralogy and textures from ore breccia, (upper photos) and ore (lowest photo). Mag, magnetite, ap, apatite, qtz, quarts, amp, amphibole, bt, biotite.

In the ore the texture of magnetite is dominated by grains having simple outlines with straight grain boundaries. The grains are of different size with coarser grains often elongated in the direction of lineation occurring in a finer grained matrix. Gangue minerals like apatite, amphiboles and biotite are elongated in a linear direction (tab. 2).
Table 2. Sample descriptions of the mineralogy and texture.

Mag
Mbgt 6500 100.22-100.30

Hem

Ap

Pl

Mc

Qtz

Amp

Cal

Bt

Py

Texture
Single simple outlined mag grain or aggregate in a myrmecitic matrix of fsp-qtz. Single simple outlined mag grain or aggregate in a myrmecitic matrix of fsp-qtz. Single simple outlined mag grain or aggregate in a myrmecitic matrix of fsp-qtz. Massive mag. fine grained at the contact to amphibole.

x x x x x x x x

x x

x x x

x x x

x x x

x x x x x x x x

Mbgt 6500 111.00-111.13 Mbgt 6500 112.57-112.63 Mbgt 6500 374.23-374.35 Mbgt 6500 383.21-383.27

x x

Massive mag, homogenous, fine grained matrix, elongated coarse grains. Massive mag, elongated coarse grains. Ap and amp grains Mbgt 6500 x x x 435.56-435.64 x elongated in the direction lineation. Mag, magnetite, hem, hematite, ap, apatite, pl, plagioclase, mc, microcline, qtz, quarts, amp, amphibole, cal, calcite, bt, biotite, py, pyrite.

Particle texture analysis (PTA) The modal mineralogy for the ore breccia samples shows slightly different results when comparing the fractions 150 m and 75 m. Magnetite decreases with 8.2 % in the 75 m fraction. The classes Mg(Ca)- silicate and quartz/feldspar are also decreasing in the smaller fraction. Instead the albite and actinolite class increases, about 14 %, however, actinolite is not represented at all in 150 m (fig. 5), (tab. 3). 76

Fig. 5. The modal mineralogy of the different mineral classes divided in particle size fractions.

The two particle fractions in the ore samples do have the same mineral classes represented but in different volume proportion. The magnetite class decreases 5 % from the 150 m to the 75 m fraction. The amphibole - pyroxene minerals (actinolite) and (Mg(Ca)-silicate) constitute together 5 % in the 150 m fraction and increases to 7% in the 75 m fraction which also become the second largest group (fig. 5). In both ore types there is an unclassified class that include particles that could not be determinate because of an incomplete x-ray analysis.
Table 3. Mineral classes and number of grains in each fractions.

Ore breccia
Mineral Apatite Magnetite/hematite Quartz Calcite Biotite Pyrite Titanite Ortoklas/microcline Actinolite/Augit Dolomite Albite Fe-Ti-oxide Quartz/Feldspar Mg(Ca)-silicate Calcite/apatite Zircon Unclassified Total Number 3 10691 592 46 1 3 145 614 0 0 157 133 911 884 7 1700 1385 17272

150 m w% 0,01 64,75 2,48 0,08 0 0 0,08 2,26 0 0 3,32 0,24 9,78 11,04 0 5,98 2,35 100

Ore breccia
Number 8 4213 173 73 48 0 189 827 943 0 1095 79 70 369 3 6 528 8624

75 m w% 0,04 56,52 2,43 0,32 0,41 0 0,34 3,92 13,51 0 17,76 0,24 0,57 3,88 0 0,07 0,71 100

Ore
Number 161 7293 60 1 14 18 131 8 302 0 65 203 8 397 14 0 1420 10095

150 m w% 1,62 91,13 0,84 0 0,02 0,15 0,13 0 3,3 0 0,14 0,55 0 2,1 0 0 0,79 100

Ore
Number 170 10214 77 26 19 108 218 30 963 1 61 433 14 1280 24 2 1700 15340

75 m w% 1,71 87,31 1,17 0,05 0,03 0,11 0,18 0,02 4,68 0 0,6 0,91 0,03 3,18 0,01 0 1,39 100

At the liberation analysis, classes of similar minerals were fit into broader groups. Amphiboles*: actinolite + Mg(Ca)-silicate, Feldspar*: Qtz/Fsp + orthoclase/microcline + albite and the remaining classes are Magnetite och Quartz.

77

Fig. 6. Liberation of minerals. Ore breccia, 1-2 (150m), 2-3 (75m) Ore, 3-4 (150m), 2-3 (75m).

In fig. 6, yellow colour means that 100 % of the magnetite is classified as an apparently liberated grain and orange ditto means that more than 95 % is liberated. The degree of liberation of magnetite is high and quite similar for the both ore types ~ 85 %. It shows a slight increase in the finer fraction. Magnetite with more than 95 % liberated grains will almost be the remaining part in every fraction. The liberation for the different silicate groups is not as high as magnetite. Quartz and Amphibole* classes are more liberated in the finer 75 m fraction (fig. 6).

Fig. 7. Magnetite associated with other minerals. Ore breccia, 1-2 (150m), 2-3 (75m) Ore, 3-4 (150m), 2-3 (75m).

The mineral classes which is associated to magnetite is more diverse in the ore breccia. Amphibole* is associated to magnetite in the ore (fig. 7 & 8). 78

Fig. 8. Amphibole* associated with other minerals. Ore breccia, 1-2 (150m), 2-3 (75m) Ore, 3-4 (150m), 2-3 (75m).

Discussion As pointed out in the introduction this study was an attempt to find a way of characterise the mineralogy and textures of an iron ore body. The different ore types were identified during the geological mapping and interpreted in the optical microscopy. Both these ore type were quantified by the mineral content, mineral liberation and mineral association by the particle texture analysis (PTA). The modal mineralogy for this study shows a decreasing magnetite proportion at the finer fraction. This is probably due to the primary grain size in the samples are larger for magnetite compared to the gangue minerals. The liberation of the silicate groups indicates that the primary grain size is finer. The ore breccia has more and larger mineral classes like quartz, feldspar and amphibole fractionated at 75m particle size. The texture for the silicate matrix is more complicated in comparison to the coarse grained magnetite which has simple grain boundaries. The interpretation is also verified in the optical microscopy This study will be extended to include more samples of the same ore type but also other ore types found in the Fabian ore body. Other ore bodies will also be included to verify that the results could be applicated in a general way for the Malmberget deposit. Conclusions The primary grain size for magnetite is coarser and has a higher proportion at the coarser fraction 150 m. Differences in the modal mineralogy of magnetite and silicates in different particle sizes of the ore breccia are close connected to the texture. The liberation of magnetite in ore and ore breccia is high and being quite similar. The ore quality for both ore and ore breccia does have similarities in a process technique prospective.

79

Acknowledgement Thanks for the financial support by Hjalmar Lundbohm Research Centre (HLRC). I am grateful to Prof. Terje Malvik and Dr. Kari Moen, NTNU for all help during the particle texture analysis and their great knowledge of process mineralogy. References Bergman, S., L. Kbler, et al. (2001). "Description of regional geological and geophysical maps of northern Norrbotten County (east of Caledonian orogen)." SGU Geological Survey of Sweden Ba 56: 110. Geijer, P. (1930). "Geology of the Gllivare Iron Ore field." Geological Survey of Sweden Ca 22: 115. Gottlieb, P., G. Wilkie, et al. (2000). "Using Quantitative Electron Microscopy for Process Mineral Applications." JOM Journal of Metals 52: 24-25. Gu, Y. (2003). "Automated Scanning Electron Microscope Based Mineral Liberation analysis." Journal of Minerals & Materials Characterization & Engineering 2: 3341. Jones, M. P. and J. Gravilovic (1970). "Automatic quantitative mineralogy in mineral technology." Rudy 5: 189-197. Lorenzen, L. and J. S. J. van Deventer (1994). "The interrelationship between mineral liberation and leaching behaviour." International Journal of Mineral Processing 41: 1-15. Martinsson, O. (2004). "Geology and Metallogeny of the Northern Norrbotten Fe-CuAu Province." Society of Economics Geologists, Guidebooks Series 33: 131-148. Moen, K. (2006). Quantitative measurements of mineral microstructure. Department of Geology and Mineral Resources Engineering. Trondheim, Norwegian University of Science and Technology. Doctoral thesis: 194. Petruk, W. (2000). Applied Mineralogy in the Mining. Amsterdam, Elsevier

80

Traceability by multivariate analysis on morphology data from grinding circuit. Pejman Oghazi 1, Bertil Plsson 1, Kent Tano 2, Bjrn Kvarnstrm 1, Lule University of Technology; SE-971 87 Lule, Sweden 1, LKAB, SE-983 32 Malmberget, Sweden 2

ABSTRACT LKAB has started a new pelletization plant at Malmberget, the raw material is a mix from Kiruna and Malmberget. To achieve good products it is important to have a good control over the input material in the concentrators that is why the traceability of the process is a crucial factor. However, creating traceability in continuous processes imply vast challenges: process flows can be parallel, serial and reflux; sub processes can be continuous as well as batch-wise; large buffers; no interruptions in product handling. These challenges imply that loads of data from the material is needed for creating traceability. In this case the grinding sections have been in focus and the data are collected from the old and the new grinding sections. The main task is to find a way to make the traceability easy and practical. One way to reach good traceability would be to find a process mineralogical signature or identification. For having a good traceability we need information from the system. It is important to analyze and look into the variables that have a crucial importance to the process. By using Particle Texture analysis a good overview of how magnetite is liberated or associated to others minerals is shown. More important is that morphological data is produced for each mineral in the process. The number o variables made it difficult to compare the result, and by using multivariate analysis such as Principal Component Analysis (PCA) it is possible to have a better insight from the collected data.

1 INTRODUCTION At todays market the customer is constantly demanding for higher product quality, to achieve this it is crucial to know and control the production process and the raw material. As a result traceability has become important in any process, since it offers the opportunity to secure process data throughout the process to a certain product. Therefore, traceability offers the possibility to understand how variation origins and learn how to adjusted the process to avoid production of defective products. Hence, traceability is a vital ability for any organisation to produce high quality products. Traceability is defined as: the ability to preserve and access the identity and attributes of a physical supply chains objects (Tyryl, 1999, page 38). Numerous benefits from using traceability systems have been identified, e.g. facilitating investigation of product failures, minimizing the extent of product recalls and assuring lot uniformity in products (Juran & Gryna, 1980). Furthermore, to have better control and knowledge of the final product it is important to know what is coming as input in the system, and also to be able follow it through the process and observe changes. By collecting data from the material during the process flow, and have a good control over the material then it is possible to trace the material trough the process. A traceability system can also be used to identify causes of positive changes in the final product characteristics. Creating traceability in continuous processes imply vast challenges: process flows can be parallel, serial and reflux; sub processes can be continuous as well as batch-wise; large 81

buffers; no interruptions in product handling. A suitable continuous process for exemplifying how traceability could be used were found in the iron ore mining industry, and in this case the grinding circuit has been in focus. 2 MATERIAL

The main ore mineral from Malmberget is magnetite. The samples that were used in this study were from the concentrators section 5 and section 6. Small amounts of material were taken from the process streams entering and leaving each mill in two parallel sections. The samples were taken during seven rounds with 20 minutes intervals. Each sample was sieved by sieveratio 2 , the fraction that was closely examined was at 38 m. Polished, thin and polishedthin section of the selected pieces (which is shown below), is selected for further analysis with microscopes.

All the samples were analysed at NTNU, Trondheim, Norway.

2.1

Flowsheet
Feed Water Water

Water Concentrate

Waste

Figure 1: Flowsheet for grinding circuit, section 5

82

Both sections are very similar to each other the figure above is from the old section 5, there is a wet cobbing stage before the grinding circuit. The new section has larger mills, and it also lacks the wet cobbing stage. The samples that were used in these studies were from the concentrator, from section 5 and section 6. Small amounts of material were taken from the process streams entering and leaving each mill in two parallel sections 2.2 Analysis method To analyse all data from the grinding circuit, Particle Texture analysis (PTA) was used. The PTA system is based on scanning electron microscope and the Oxford Inca Feature software. Using Back Scattered Electrons (BSE) the images are analysed by means of grey level and every grain of interest is analysed with X-rays. When every grain size is analysed, the data will be imported to the PTA software. With the Inca data information the images will be processed and calibrated (grey-scale and binary images); and the grains will be identified and evaluated if they are liberated or in composite particles and which minerals occur in the composite mineral (Moen, 2006). 2.3 Multivariate data analysis For achieving good control and having better overview over the process it is necessary to collect many data with many variables and many properties from the process. By using multivariate data analysis many of these variables will be explained and expressed so it will be easier to understand which variable is the information-rich variable (MVDA). Multivariate data analysis is based on projection methods, it build a model by using data matrix X and containing N observations (samples). The principal component analysis (PCA) is a multivariate projection method that show us how the observation are related to each other and if they are, and also which of the variables supplies with distinctive information about the observation. PCA is a projection method of the original variables onto new ones, orthogonal and arranged according to their eigenvalue. This is carried out as matrix X. X = ABT + E A represent the score matrix and B represent the loading matrix (Prats-Montalbn, 2004). Projections to latent structures by means of partial least squares (PLS), is another way to analyse how different variables are important in the model. By selecting different variables as Y, it will give information how selected variable(s) is explained in the model. PLS model is suitable when the data are symmetrically distributed and not shifting irregular.

2.3.1

Pre-treatment of data

Pre-treatment of data is an important task in most of cases before data analysis. In this case it is very important to pre-processing the data in order to get good models and results. As the goal of this analysis is to find a relation between the materials that goes through the grinding sections and understand the differences between the sections. All the data that is used in multivariate data analysis are from the PTA analysis database. Every single mineral grain that was analysed a lots of information was collected, apart from the morphologic parameters also chemical element was stored in the analysis.

83

Table 1. List of minerals and parameters which were used for analysis. Minerals: Parameters: Magnetite Area; (Area of whole feature in square microns)
Ilmenite Length; (Max feret)

Rutile Plagioclase Quartz Pyroxene/Amphibole (Pyx/Amf) Biotite K-feldspar Enstatite Apatite Titanite Calcite Sulphides Unclassified

Breadth; (Min feret) Perimeter; (Perimeter of whole feature in microns) Aspectratio; (Length/Breadth) Direction; (Angle) Shape; sq (Perimeter) ECD; (Square root of (4 X Area)/Pi Mean grey level; (Mean image grey level for each) % Element (wt %)

Data collected from the PTA analysis were imported to the SIMCA (Soft Independent Modelling of Class Analogy) program and analysed. From a statistical overview a multivariate model explaining the observation was developed. PCA-models were created and further improved investigation was done. By using multivariate tools it will help to find trends, dominating variables and groups among the collected data. In this case mostly the score and loading plots is used which gives important information about variables that are responsible for the pattern seen among the observation and how they are related to each other.

84

RESULT

Multivariate overview
In a first PCA overview with out any improvement to the model the material is clearly spread out in the first direction, there is hardly any trend in Figure 2. Note, the two magnetite population. 3.1 Section 5 - Primary mill

Figure 2. Score plot with mineral identification for incoming and outgoing material in section 5. In the first overview all the minerals, element Wt% and morphology data is included but actually only the morphology parameters is the interesting while they change during grinding circuit. Figure 3 shows that the magnetite and gangue parameters are the ones which dominate in the first direction.

85

Figure 3. Loading plot for the PCA model, section 5. The first analysis was to compare the incoming and outgoing in each mill. By excluding some variables the models was improved, the variables in interest were those which describe the morphology.

Figure 4. List of variables used in model.

86

A PCA model for the data material was created with 4 principal components (PCs) and it had an explanation and predictability value of 98% and 87% respectively cf. Table 1. Most of the information was found in the first and second direction, according to the eigenvalues but it is not necessarily that they give the most interesting information. The best explanation between the sample populations is found if the first and third PC is plotted against each other, see Figure 5. Table 2 . Summary of PCA overview model, section 5.
PC 1 2 3 4 R2X 0,627 0,183 0,117 0,0549 R2X(cum) 0,627 0,810 0,927 0,982 Q2 0,581 0,109 -0,168 0,693 Q2(cum) Eigenvalue 0,581 5,02 0,626 1,46 0,589 0,936 0,874 0,439

As shown in the score plot there are clear differences between the input and output of material in the first mill, the score plot shows also that most data points fall within the 95% confidence interval in Figure 4. It is possible to see two populations.

Figure 5. Score plot for the PCA model of incoming and outgoing material, section 5 . The loading plot in Figure 6 shows that size parameters pull most of the attention in first direction while mean grey have a great influence in the third direction. The first direction is length information.

87

Figure 6. Loading plot for the PCA model, section 5. With the same data coloured according to mineral classification it is possible to see how the minerals differentiate. The main parameter pulling magnetite away appears to be Mean grey. This is what to be expected.

Figure 7. Score plot with mineral identification for incoming and outgoing material, section 5

88

However if the same plot is stripped to leave only magnetite information, sub-populations of magnetite emerges, cf. Figure 8.

Figure 8. Score plot for magnetite for incoming and outgoing material in first mill section 5. To this moment it is not entirely clear what causes the split on sub-populations. The lower value of Mean grey and shorter length, indicates that the two minor groups may constitute of mixed particles. 3.2 Section 6 - Primary mill As it shows in table 2, four PCs were generated for section 6. The explanation value was 97% and predictability 86%. To compare it with section 5 the values are almost the same, but they behave differently because of the parameters affect in model in different direction. In figure 10 it is obvious that mean grey is the reason for that. Table 3. Summary of PCA overview model, section 6.
PC 1 2 3 4 R2X 0,623 0,185 0,113 0,0579 R2X(cum) 0,623 0,808 0,920 0,978 Q2 0,575 0,127 -0,223 0,667 Q2(cum) Eigenvalue 0,575 4,98 0,629 1,48 0,592 0,902 0,865 0,464

Samples from section 6 behaves differently, one reason could be that there is no cobbing stage before the grinding circuit.

89

Figure 9. Score plot for the PCA model of incoming and outgoing material, section 6. Compare the loading plots between section 5 and section 6 there are similarities. PC direction 1 carries length/size information, while PC3 is mostly Mean grey. Both Figure 6 and 10 gives the same information. This proves that the pattern found is systematic. It is also seen in Figures 7 and 11.

Figure 10. Loading plot for the PCA model, section 6.

90

Figure 11. Score plot with mineral identification for incoming and outgoing material in section 6. 3.3 Section 5 - Circuit It is interesting to follow a single mineral comprehensively for the length of a grinding circuit. In Figure 12 the score plot shows how magnetite is behaving in section 5. The previously mentioned sub-populations seem to be large for the discharge streams. It may be interpreted as a generation of mixed particles.

Figure 12. Score plot for magnetite in section 5. To exactly understand how each grinding step affects apatite and how the magnetic separation influence the process Figure 13 was generated.

91

This figure shows us that for each separate stage the apatite is decreasing appreciably as it was expected. It is now interesting to investigate how the same minerals are behaving in section 6. Here shown for apatite and feldspar in Figure 13 and 14 respectively

Figure 13. Score plot for incoming and outgoing apatite in section 5. To compare the feldspar and apatite it is obvious that the apatite continue to exist during the whole circuit, while most of the feldspar is separated after the first separation stage. This is, free particles seems to disappear, and only mixed particles remain.

Figure 14. Score plot for incoming and outgoing feldspar in section 5.

92

3.4 Section 6 - Circuit The same models were generated for section 6, to compare with previous section.

Figure 15. Score plot for incoming and outgoing apatite in section 6. In Figure 15 it is shown that some apatite particles are still present after the third mill. It is probably the result of the missing wet cobbing stage for section 6.

Figure 16. Score plot for incoming and outgoing feldspar in section 6.

93

From figure above is clear that most of the feldspar is separated after the first separation stage, however some particles is found in the second mill.

Figure 17. Score plot for incoming and outgoing magnetite in section 5 and 6. In Figure 17 both section 5 and 6 is combined in the same model. The different subpopulations are much visible in this figure as well, as it shown some of the out going magnetite from section 6 has much larger dimension compare to section 5. It could be some intergrown with some gangue mineral. 4 CONCLUSION The main benefit of multivariate analysis on particle texture data is that it simultaneously compares several thousand particles. This kind of investigation is state of the art in this area and it can provide us with loads of information of how each process step influence the material. By comparing different section and different minerals it can be shown how the cobbing stage affect the circuit. By comparing the feldspar and apatite it is obvious that the apatite continue to exist during the whole circuit, while most of the feldspar is separated after the first separation stage. It depends on that the free particles appear to disappearing while mixed particles exist in circuit. By combining section 5 and 6 a direct similarity is found between the two section but also some divergences is found which can depend on the cobbing stage or other factors that the grinding mill differs from each section. It is interesting to further investigate the material from both sections by using optical analysis. Using optical analysis could intergrown minerals be identified. In some cases, only a small number of particles were studied. In these cases more sample needs to be investigated before certain conclusions could be drawn, as there are a natural variation in the material.

94

5 ACKNOWLEDGEMENT Financial support from LKAB Research and Development is gratefully acknowledged. We would also like to thank Johan Bucht for valuable ideas regarding the multivariate analysis. 6 REFERENCES

Juran, J.M. and Gryna, F.M., Quality Planning and Analysis: from Product Development through use, 2nd edn. 1980, New York, McGraw-Hill Book Company. Moen, K., 2006 Quantitative measurements of mineral microstructure, Doctoral theses at NTNU 2006: 194. Part-Montalbn, A comparison of different discriminant analysis techniques in a steel industry welding process. Chemometrics and intelligent laboratory systems 80:109-119. Reed, S.J., 2005, Electron Microprobe Analysis and Scanning Electron Microscopy in Geology. Cambridge University Press, Cambridge, UK. Tano, K. T., Plsson, B.I., and Sellgren, A., 2005 On-line lifter deflection measurements showing flow resistance effects in grinding mills, Minerals Engineering 18:1077-1085. Tyryl, I., Realising the Potential of Traceability A Case Study Research on Usage and Impacts of Product Traceability. Acta Polytechnica Scandinavia, Mathematics, Computing and Management in Engineering, Helsinki University of Technology, Doctorial thesis No. 97,1999.

95

Mineralteknik Aktuellt om utbildning och forskning Plsson, Bertil I. Lule tekniska universitet, Avdelningen fr mineralteknik, SE-971 87 Lule

SAMMANFATTNING Utbildningsprogrammet i Kemiteknik har genom anpassning till Bolognamodellen ftt en struktur, som gr det mjligt att samverka med andra utbildningsprogram och Master-program inom specialiseringen fr mineralteknik och processmetallurgi. Eftersom gruvbranschen upptckt att det behvs utbildade medarbetare har ven efterfrgan p kortare utbildningar kat. Forskningen brjar kunna utnyttja att branschen har framgngsrikt verkat fr forskningsprogram med relevans fr mineralteknik och metallurgi. Mineralteknik brjar rekrytera nya forskarstuderande och kan med industrins hjlp ter bli ett aktivt kunskapscentrum fr gruvbranschen. 1. Bakgrund Vid tidigare konferenser har det rapporterats om de svrigheter, som utbildningen i kemiteknik och kanske speciellt specialisering mot mineralteknik och processmetallurgi sttt infr (Plsson, 2004, 2005, 2006). Lsningen har varit att skapa ett flexibelt undervisningssystem, som relativt ltt kan acceptera studerande frn olika program och frn industrin. Det har ocks visat sig att det sedan mineralteknik blev ett eget professorsfretrtt mne ca 1940, har det bara producerats 5 15 civilingenjrer per r med kombinerad mineralteknikmetallurgiprofil. Grundutbildning Sktryck Utbildningen inom mineralteknik vid LTU fanns till en brjan inom utbildningslinjen fr Geoteknologi, dr de andra specialiseringarna var bergteknik och prospekteringsteknik.. Nr den linjen lades ner flyttades mineralteknik och processmetallurgi ver till utbildningsprogrammet fr Kemiteknik. Detta har sedan omarbetats och anpassats till Bologna-modellen (Anon, 2000) och r numera programmet fr Kemiteknisk design, som r uppdelat p en grundlggande trerig utbildning med avslutande tvrig Master-utbildning. I praktiken lser dock de som antas till k 1 en normal 5-rig civilingenjrsutbildning. Sktrycket har varierat starkt mellan ren, se figur 1. Det ser nu ut att bli en aning bttre p kemiteknik, men det r framfr allt det nya programmet Naturresursteknik som kan bli en viktig framtida rekryteringsbas fr mineralteknik och processmetallurgi.

2. 2.1.

97

Antal registrerade vid terminsstart k 1 p LTU

35 30 25 20 15 10 5 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Kemiteknik Naturresursteknik

Figur 1.

Antagna till LTU k 1 exkl antagna till hgre rskurs och byte mellan utbildningsprogram. Stapeln fr Naturresursteknik innehller antagna till bde kandidat och civilingenjrsprogram.

Siffrorna i figur 1 inkluderar inte vergngar till Kemiteknik, som grs frn Tekniskt basr och ppen ingng. Dessa har p senare tid varit 2-3 stycken per r, vilket terspeglar att de som vl kommit till LTU sedan upptcker att det finns en intressant utbildning och en industribransch dr det gr att f jobb. 2.2. Naturresursteknik Programmet fr naturresursteknik har till en del mer karaktren av ett geovetarprogram, dr de studerande redan frn rskurs 2, lsperiod 2 mste vlja en av fyra specialiseringar: miljteknik, jord- och bergbyggnad, malmgeologi och miljgeokemi eller mineralteknik och processmetallurgi. Nr det gller den sistnmnda specialiseringen r den nstan identisk med den p kemiteknik. Skillnaden bestr i att naturresurserna lser ngra fler kurser inom mineralogi och andra geovetenskapliga eller bergtekniska mnen. I gengld lser de inte kurser inom organisk kemi och biokemi. Figurerna 2 och 3 visar hur rskurs 2 resp 3 r tnkta att se ut fr Naturresursteknik. Eftersom det i flera av fall r nya kurser eller kurser, som konverterats frn andra program, s saknas ibland de nya kurskoderna. Inom mineralteknik och processmetallurgi r det dock frgan om existerande och inkrda kurser. Nr det gller rskurs 4 och 5 kommer Naturresursteknik att i stort flja utbildningsplanen fr Master inom Minerals and Metallurgical Engineering.

98

M0031M Linjr algebra och differentialekvationer

MILJTEKNIK R0006N Grundkurs i projekt- och industriell ekonomi L0039K Naturliga vattentransportprocesser A0001B Avfallsteknik Valfri kurs L0010K Miljvervakning G0003B Geoteknik T0014B Bergmekanik T0013B Berganlggningsteknik Valfri kurs A0002B Teknisk Mikrobiologi

O0036K Mineralogi De som gr Miljteknikinr lser istllet L0001K Ekologi och miljkunskap

JORD- OCH BERGBYGGNAD R0006N Grundkurs i projekt- och industriell ekonomi B0002B Konstruktionsteknik MALMGEOLOGI OCH MILJGEOKEMI R0006N Grundkurs i projekt- och industriell ekonomi O0008K Petrologi MINERALTEKNIK OCH PROCESSMETALLURGI K0010K Fysikalisk kemi K0006K Vattenkemi

L0041K Kvartrgeologi L0040K Geokemi

O0001K Grundkurs geofysik L0010K Miljvervakning P0001K Hgtemperaturprocesser Valfri kurs

K0011K Oorganisk kemi M0001K Fysikaliska separationsmetoder

Figur 2.

Naturresursteknik rskurs 2 (val av inriktning).

99

MILJTEKNIK S0001M Matematisk statistik Ny kurs Naturvrme, Frnyelsebar energi JORD- OCH BERGBYGGNAD S0001M Matematisk statistik D0002B Drift och underhll MALMGEOLOGI OCH MILJGEOKEMI S0001M Matematisk statistik O0007K Strukturgeologi MINERALTEKNIK OCH PROCESSMETALLURGI S0001M Matematisk statistik B0003K Transportprocesser R0006N Grundkurs i projektoch industriell ekonomi M0032M Flervariabelanalys och datorverktyg A7002B Mekanisk och biologisk avfallsbehandling M0032M Flervariabelanalys och datorverktyg Ny kurs Utv proj From mine to mill Valfri kurs V0002B VA-system och hydraulik Valfri kurs

Exjobb fr Bachelor Praktik fr Civ-ing

Ny kurs Utv proj From mine to mill Valfri kurs

Ny kurs Ingenjrsgeologi Valfri kurs

Ny kurs Miljanalys Ny kurs Malm och mineralresurser T0002K Kemisk reaktionsteknik I P0002K Designer fr hllbar processteknik

Figur 3.

Naturresursteknik rskurs 3, prel 2009/2010.

Den sista lsperioden i rskurs 3 gnas t valfria kurser, som frutom de inom kemiteknik ocks kan omfatta geovetenskapliga och bergtekniska kurser. 2.3. Produktion av civilingenjrer Mineralteknik och Processmetallurgi P grund av svngningarna i det allmnna konjunkturlget och drmed intresset fr utbildning inom gruvmnen har produktionen av civilingenjrer varierat kraftigt ver ren. Detta visas, dels fr det ldre programmet Geoteknologi i figur 4, och dels fr det nya Kemiteknik i figur 5. Notera att produktionen ligger fr bda programmen i genomsnitt p ca 7 civilingenjrer per r. Hr syns ocks tydligt hur produktionen svnger i femrscykler i stort sett tidsfrskjutet i fem r mot konjunkturlget. Eller uttryckt p ett annat stt utbildning och efterfrgan r i motfas. Frdelningen p industriomrden r uppdaterad i januari 2008 fr att terspegla var ingenjrerna arbetar idag. Det r helt naturligt att andelen inom tillverkande industri och konsulter kar desto lngre bak i tiden man gr. 100

Bergsingenjrer LuTH - Mineralteknik/Metallurgi (140 st)

16

14

12

10

Ej examen vrigt Statligt Forsk-Underv Tillv-Konsult IT Kemi Industrimineral Metallurgi Gruv

0 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995

Figur 4.

Produktion av bergsingenjrer Mineralteknik-Processmetallurgi.

Chemical Engineers LTU - Mineral Processing-Metallurgy (67)

16 14 12 10 8 6 4 2 0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

No exam Others Government Res.-Teaching Machine-Consultancy IT Chemistry Industrial Minerals Metallury Mining

Figur 5.

Produktion av civilingenjrer Mineralteknik-Processmetallurgi.

Nr det gller de som inte har examen frn 2006 och 2007 s r det frgan om personer som r p gng, men de har resttentor som gr att de nnu inte kunnat ta ut examen. Siffrorna fr 2008 2010 r prognoser baserat p studenter 101

i examensarbete, de som valt inriktning respektive de som valt kurs som gr det troligt att de vljer avslutningen mot mineralteknik och processmetallurgi. 2.4. Korta utbildningar Det lga antalet studenter frestar givetvis hrt p avdelningarnas ekonomi eftersom vi frn LTU har ett rakt ackord inom utbildningen. Fr att n breakeven behver de flesta kurser ha tminstone 15 deltagare som ocks examineras p kursen. Det hr har gjort att specialiseringarnas huvudkurser har subventionerats frn grundlggande kurser och frn fakultetsanslaget som skall ge en bas fr forskningen. Turligt nog har gruvfretagen insett att, om de rekryterar ingenjrer som inte gtt i Lule, kan fretagen skicka dem p utbildning inom mineralteknik och processmetallurgi p LTU. Under de senaste fem ren har t ex grundkursen i mineralteknik haft i genomsnitt 5 studerande per r som kommit frn gruvfretag. Huvudkursen i mineralteknik, processanalyskursen och simuleringskursen har ocks regelbundet externa studenter. Vad som tillkommit under 2007 r nskeml frn gruvindustrin om kortare utbildningar inom mineralteknik i kombination med geologi/bergteknik eller processmetallurgi. 1. Under vr och hst 2007 gav avdelningarna fr mineralteknik, malmgeologi, bergteknik samt drift och underhll en kurs i gruvproduktionsprocessen fr New Boliden Aitikgruvans samtliga tjnstemn, totalt ca 60 personer. Kursen omfattade 48 lektionstimmar och hade som ett syfte att frska ge deltagarna en bttre samsyn ver processkedjan geologi gruva anrikning underhll milj. 2. Den 28 januari 2008 startade i Storuman en 2-rig processteknisk utbildning (KY-utbildning) fr den lokala processindustrin. Industrin representeras i det hr fallet av Skellefte Kraft (bioenergi) och gruvfretag lngs guldlinjen i Vsterbotten. Utbildningen fick 35 skande till 20 platser. Avdelningarna fr mineralteknik och processmetallurgi kommer att ge en 5-veckors kurs inom mineralteknik och hydrometallurgi fr KY-utbildningen. 3. Det finns ocks frfrgningar frn andra aktrer via Lapplands Kommunalfrbund om distansutbildningar riktade mot stra Norrbotten. I det hr fallet kan det bli frgan om varianter p den 3-riga utbildningen i Naturresursteknik. 2.5. Samarbete med andra universitet och hgskolor Institutionen fr Tillmpad kemi och geovetenskap (TKG) har med uppbackning frn centralt LTU-hll tecknat avtal med andra universitet och hgskolor i Norden. 1. Med Bergsskolan i Filipstad finns nu en verenskommelse som bland annat innebr att de med kandidatexamen frn Bergsskolan skall kunna lsa vidare p TKG:s Master-program. 102

2. Med Universitetet i Uleborg finns ett avtal, som innebr ett utbyte av studenter p Master-niv. I frsta hand kan det bli frgan om ca 5 studenter per r som vljer att lsa mineralteknik-processmetallurgi vid LTU, ev redan frn hsten 2008. Det finns ocks planer p en gemensam professur i mineralentreprenrskap, dvs hur man gr pengar p det som finns i backen. 3. Med Norges Tekniska-Vetenskaplige Universitet och dess Institutt for geologi og bergteknikk finns ett Erasmus-avtal, som underlttar lraroch studentutbyte. 3. Produktion av forskarstuderande Eftersom gruvbranschen har vaknat till liv igen finns det nu ocks forskningsmedel att ska. Institutionen fr Tillmpad kemi och geovetenskap (TKG) har varit framgngsrik nr det gller beviljade anskningar och har ftt flera stora anslag frn Vinnova:s Gruvforskningsprogram och frn HLRC (Hjalmar Lundbohm Research Centre). Det har gjort att vi har ett delikat problem hur fr vi tag i forskarstuderande nr gruvbranschen r s het som den r just nu? I flera fall har vi rekryterat personer som gtt utbildningar vid andra svenska hgskolor eller frn andra specialiseringar n vra egna. Dessutom finns det en stor andel industridoktorander inskrivna vid TKG. Det totala antalet forskarstuderande vid TKG framgr av tabell 1.
Tabell 1. Antal forskarstuderande inkl industridoktorander 4 5 6 7 15 9 7 8 61

Biokemisk och kemisk processteknik Fysikalisk/grnsytors/oorganisk kemi Kemisk teknologi Mineralteknik Processmetallurgi Malmgeologi Tillmpad geofysik Tillmpad geologi (geokemi) Inst fr tillmpad kemi och geovetenskap

Hr r det intressant att gra en internationell jmfrelse. Nyligen publicerades en versikt ver lget vid akademiska institutioner i USA som utbildar fr gruvbranschen (McCarter, 2007). Av den framgr att av ursprungliga 27 Mining Schools terstr nu 14. Hrav r det bara Virginia Tech som har ett program inom Minerals Engineering. Nu r detta en inte helt rttvis jmfrelse. Mineralteknik finns ocks inom en del program fr Chemical and Metallurgical Enginering, t ex vid University och Utah. Ur internationell konkurrenssynvinkel r det intressant att notera att de hr programmen verkar vara rejlt underbemannade. Totalt finns det bara 70 personer med doktorsgrad p de kvarvarande programmen. Det r oklart om det innefattar de fyra som undervisar 103

inom Minerals Eng. I vilket fall verkar lrarttheten vara betnkligt lg. Produktionen av forskarstuderande r inte imponerande, se figur 6.

Figur 6.

Produktion av forskarstuderande inom gruvmnen vid amerikanska universitet (McCarter, 2007).

Prognosen fr den framtida amerikanska forskarproduktionen ser inte heller bra ut, se figur 7. Ur svensk synvinkel borde det innebra att den nordamerikanska marknaden kommer att starkt efterfrga vra ingenjrer och forskare.

Figur 7.

Frvntad examination av forskarstuderande i USA (McCarter, 2007)

104

4.

Forskningsprojekt Vid Avdelningen fr mineralteknik pgr nu flera stora forskningsprojekt, som de flesta har anknytning till den traditionella gruvindustrin. Bara fr ngra r sedan var det i stllet projekt inom tervinning och industrimineral som var helt dominerande. Nu r det en mera balanserad mix av projekt, se tabell 2. Vart och ett av dessa sysselstter en forskare eller en forskarstuderande.
Tabell 2. Name BioMinE Electrostatic beneficiation of Indian thermal coal De-phosphorisation of magnetite fines Traceability of raw materials in process streams Flow and treatment of thick pulps Wet low-intensity magnetic separators Charge dynamics of autogenous mills Recycling of process water and influence of its chemistry on sulphide flotation and flotation selectivity Strre projekt vid Avdelning mineralteknik. Source European Union Swedish Intl. Development Cooperation Agency & Vattenfall HLRC LKAB LKAB LKAB LKAB Vinnova

Mineralteknik r nu i den angenma situationen att vi sker fler forskarstuderande. 5. Framtid Utifrn vad som har sagts hr torde det framg att TKG och dess avdelningar har arbetat mycket hrt med att omstrukturera utbildningen och anpassa forskningen till den nu rdande konjunkturen. Det finns dock ett problem, som vi mste vara uppmrksamma p just nu lskar (nstan) alla gruvindustrin och LTU:s utbildningar. Vad gruvbranschen nog inte frstr, r att Mineralteknik som avdelning och forskningsmne var en hrsmn frn att bli nedlagt p grund av att det ansgs utbilda och forska fr en gammal, omodern och vermogen bransch. Vi klarade oss den hr gngen, men vi kan bara se p exemplet USA och frst vilka konsekvenser kortsiktiga beslut kan f. Det r drfr av strsta vikt att gruvbranschen inte slr sig till ro och tror att hotet mot vr utbildning och forskning r borta. I stllet mste vi redan nu frga oss hur kan vi vidmakthlla en forskningsvolym, som ger stabilitet i utbildning och forskning ver en konjunkturcykel?

105

6.

Referenser Anon., 2000. The Bologna Declaration on the European space for higher education: an explanation. Confederation of EU Rectors Conferences and the Association of European Universities (CRE). http://europa.eu.int/comm/education/policies/educ/bologna/bologna.pdf. 10 p. McCarter, M.K., 2007. Mining faculty in the United States: current status and sustainability. Mining Engineering 59, No. 9 (September), pp. 28-33. Plsson, B., 2004. Mineralteknikutbildning i framtiden en virtuell upplevelse? E. Forssberg & M. Thomaeus (eds.), Meddelande frn MinFo Nr 34, Konferens i mineralteknik, Lule, 3-4 feb., 2004. Freningen Mineralteknisk forskning, Box 5501, 11485 Stockholm. 9 p. Plsson, B., 2005. Frndrad och internationaliserad utbildning i Mineralteknik och Processmetallurgi vid LTU. Konferens i Mineralteknik 2005. In: M. Tomaeus & E. Forssberg (Eds.), Konferens i Mineralteknik 2005, Meddelande frn MinFo Nr 35, 2005. Freningen mineralteknisk forskning, Stockholm. Plsson, B., 2006. Implementing the Bologna process at LTU: a tool for flexibility. Konferens i Mineralteknik 2006. In: M. Tomaeus & E. Forssberg (Eds.), Konferens i Mineralteknik 2006, Meddelande frn MinFo Nr 36, 2006. Freningen mineralteknisk forskning, Stockholm.

106

On lifetime costs of flotation operations

A. Rinne and A. Peltola Outotec, Finland

ABSTRACT Overall economy of a flotation operation bears much more than investment costs. The lifetime operation and maintenance of a flotation machine may affect the economy of a project far more than a million saved in investment. Studying the long-term effects on investments is worthwhile as they often surpass any apparent savings in capital expenditure. The optimal solution does not even need to be more expensive. Besides, with high operating costs, time is never on one s side. In the paper various flotation cell arrangements are compared in a simplified example. The example considers capital equipment investment costs, power delivery arrangements, energy costs, and maintenance costs throughout a twenty-year ownership. Also effects on CO2 emissions are discussed. The paper will illustrate how proper choice of equipment and modern power delivery and control methods will result in significant benefits in lifetime costs and profitability of flotation operations.

1.

INTRODUCTION

When an investment for a new flotation plant is evaluated, the emphasis is often in minimizing the capital expenditures. Until recently, life cycle cost (LCC) as a criterion for selecting beneficiation equipment has played only a small role in the final decision making. A quick analysis discussed below shows that roughly 60-80 percent of the total 25-year cycle costs for a large flotation machine are spent on energy while the initial investment comprises less than 10 percent. As a result, if a small saving in investment is achieved by compromising energy efficiency, it can quickly turn into big losses in operational costs. If one looks for savings in the long run, life cycle cost analysis shows that the importance of the investment cost is almost negligible. Modern technology can offer completely new solutions for optimizing flotation processes both in terms of efficiency and metallurgy. Mechanical flotation machines have traditionally been limited by their relatively narrow range of aeration rate, fixed mechanism dimensions and speed. New flotation machine designs allow much wider adaptability with speed control, shear adjustment and wider range of air feed. More attention should be paid on maintenance of critical components. Cases are known where significant metallurgical losses have been observed due to poor condition of critical wear parts.

107

This paper discusses the economy of flotation projects from the life cycle cost perspective. The examples have been calculated based on generic average values. There is a wide variation in costs between locations and specific processes, but averages give a good starting point to more specific analysis.

2.

LIFE CYCLE COSTS OF FLOTATION OPERATIONS

Life cycle cost analysis simply considers the lifetime operation and maintenance costs of a flotation operation in addition to the initial investment, in selecting the most economical equipment. It may be feasible to pay higher initial cost if one saves in operational expenditures. The relevant cost factors for a flotation plant are investment, energy and reagent consumption, and maintenance. All these should be quantified for the estimated service life of the equipment. Figure 1 presents typical total ownership costs of a large mechanical flotation machine (100-200 m3) over the time span of 25 years using average rates (see Table 1) for electricity, reagents, and maintenance labor. There is high variation in the cost-factors; more specific results can be easily obtained by inserting the actual rates for electricity, reagents and labor as well as for the cost of capital in the calculation model. The breakdown strongly suggests that the most significant life cycle cost item in flotation operations is the cost of electricity. The operational expenditures are heavily influenced by the energy price and the energy efficiency of the equipment used for production.
Table 1. Average rates for electricity, reagents, and maintenance labor. Power draw 138,82 kW Cost of electricity 0,06 /kWh Annual operating hours 8 300,00 h Cost of capital 10,00 % Reagents 13 800,00 /a Hourly rate of maintenance services 50,00 /a

Figure 1. Breakdown of a large flotation cell expenses over the lifespan of 25 years.

108

2.1

Capital expenditures

A given requirement for flotation capacity may be satisfied by several scenarios which may differ significantly in terms of required footprint, investment cost and required maintenance resources etc. The most significant decisions concern the implemented unit size and the principle of operation of the equipment. In general, larger flotation cell units lead to lower investment, energy and maintenance costs as measured in unit price per unit of installed volume. 2.2 Operational expenditures

The operational costs of a flotation machine depend on the efficiency of the equipment. Process efficiency, energy efficiency and availability are discussed below. 2.2.1 Process efficiency The key mechanical aspect for good flotation process efficiency is the proper condition of critical wear components. Missing rotor or stator parts make the cell surface wavy and cause the froth to collapse. Air dispersion is reduced and decreased pumping causes sanding. The use of copied spare parts has often caused problems (see Figure 2). Experience has shown that non-standard spare parts often give a shorter wear life and in some cases decrease the metallurgical efficiency. The real savings that can be achieved by using worn out or low quality parts are negligible when compared to energy costs of the same equipment. If even small metallurgical losses occur because of poorly working equipment the savings in maintenance quickly become expenses. In order to obtain optimal availability performance it is safest to use only original equipment manufacturer s spare parts.

Figure 2. Pirate flotation cell mixer spare parts after two weeks of use.

109

Comparisons of flotation operations using different technologies are published in two recent papers (Froehling et al., 2005 and Coleman et al., 2006). These papers give a good view of the effect of metallurgical performance as well as maintenance aspects on overall economics. 2.2.2 Energy efficiency The traditional drive mechanism of a flotation machine consists of a single-staged V-belt drive connected to a low-speed induction motor. Theoretically the efficiency of the V-belt drive is 97-98 percent provided that the belts are optimally sized, properly aligned and tightened to correct belt tension. Unfortunately in practice this situation is rather rare and the actual efficiency is therefore lower. As the belt wear and stretch readjustments are required. This is often impossible without shutdown of the equipment, which, in turn, would result in reduced availability. Similar challenges occur also with gearbox-driven flotation machines when the power from the electric motor is transferred to the gearbox through a V-belt drive. Figure 3 shows an example of a misalignment.

Figure 3. Misaligned V-belt pulleys.

In addition to the power transfer ratio of the drive mechanism, the rotation speed of the rotor and the air feed equipment are important factor in the electrical energy consumption. Studies on variable-speed drive mechanisms have been conducted indicating a certain range of regulation in the rotor speed where the metallurgical performance of the flotation machine is practically the same. However, the rotation speed has a significant effect on the power draw of the mixing mechanism. The relationship can be simply expressed as P 1 p Q , where

110

P is the power draw, is the hydraulic efficiency of the mechanism, p is the pressure difference generated by the mixing mechanism, and Q is the volume flow rate through the mixing mechanism. The pressure difference over the mixing mechanism is proportional to the rotation speed squared and the volume flow rate is directly proportional to it. Thus the power draw of the mixing mechanism is proportional to the third power of the rotation speed. Consequently, a minor reduction in rotation speed may have no effect on process performance but a significant effect on the energy consumption. For example, 10 percent reduction in rotation speed roughly equals to 27 percent reduction in power draw. A drive mechanism that enables the adjustment of the rotation speed may produce significant savings in electricity consumption. Case studies have shown that a variable speed drive may have payback time of only few months, if the process allows optimization of rotation speed. The same principles apply also to air blowers. Significant savings may be achieved if the air blower is not driven at maximum power against a regulating valve but at a speed that is sufficient to maintain the required airflow. Such optimization can be done using cell-specific air blowers with variable speed drives. 2.2.3 Availability performance One cannot consider operational performance without also considering the availability of the equipment. According to Lyytikinen (1987), the process yield is related to the process performance and availability performance as follows: Process yield = Process performance * Availability performance Availability performance can be further divided into three subcategories, as presented in Figure 4.

Availability Performance

Reliability Performance

Maintainability Performance

Supportability Performance

Figure 4. Factors affecting the availability performanc e. (Lyytikinen 1987)

Reliability of the equipment refers to the probability of a unit functioning normally when used according to specific conditions for at least a given period of time. Reliability can be estimated from the failure rate of the equipment.

111

Maintainability can be defined as the ability of an item to be retained in a state in which it can perform the required function, when maintenance is performed using stated procedures and resources. For flotation machines this means ease of maintenance or replacement of mixing or drive mechanisms. For example, if the drive mechanism can be removed from the cell as a complete unit and there is a reserve drive standing by, the unplanned downtime can be minimized and thus the availability is not compromised even by events of sudden malfunctions. Supportability is the availability of the material required to keep the system operational. In general, standard components are technically and economically optimized structural solutions, which have decent availability. In addition, manufacturers of standard components benefit from economies of scale and thus the costs of standard components are usually lower as compared to craft production.

3. 3.1

THE POTENTIAL OF NEW TECHNOLOGY Use new equipment

Outotec s flotation cells have always been customized for ore and process characteristics to optimize metallurgical results. Anyhow, ore properties change during equipment lifetime or the same equipment may be utilized for a new pit close to original operation. The function of a single cell in the flowsheet can be changed. Mines with different types of ore are also relatively common and change from one ore type to another often causes process hiccups. Especially in this kind of cases optimization of flotation machine parameters such as mixer speed or power consumption would be beneficial. But there is a lot to be done also at flotation operations with no particular troubles. The effect of optimal air dispersion, mixer speed and shear are discussed below. 3.2 Leading edge technology for process optimization

3.2.1 Disperse the right amount of air Optimal air dispersion is one of the basic requirements for good metallurgical performance. Plants operating with forced air cells have often noticed that the best results are achieved using individual and varying air feed rate in each cell. In traditional flotation mechanisms the air feed is limited by the reduction of power draw and mixing, or by reduced dispersion of air making the froth surface unstable and causing the froth to collapse. Outotec s new mechanism design, the FloatForce , pushes further the maximum air feed limit compared to other existing designs. As a result the cell surface is steady in all situations and the pumping rate of the mechanism is only slightly affected by air. Because of the flat power curve less power is needed when the mechanism is operated without air. This allows smaller motors and benefits both in investment and operating costs because of more efficient operation of the motor. Figure 5 presents power draws of OK and FloatForce mechanisms in comparative conditions.

112

Figure 5. Power draw versus air feed -curves of FloatForce

and OK mechanisms.

3.2.2 Adjust mixer speed The easiest solution for adjusting mixer speed is a variable speed drive (VSD). A frequency converter can theoretically be installed in most of existing flotation cell drives but compatibility with existing equipment such as electric motor must be checked. Typical adjustment range of a flotation cell with VSD lies between 75-110% of original shaft speed if there are no special limitations in the existing equipment. Initial process test results indicate two different behaviors with moderate speed adjustments. More often changes in metallurgy are negligible but significant changes in power consumption can be seen. More detailed test results will be reported later. In some cases optimal speed shows optimal metallurgical point of the examined flotation cell. It is still likely that the optimal speed of each cell even within the same plant varies. 3.2.3 Add more shear to boost recovery The role of shear, or rotor-stator gap, has also been investigated. Earlier testwork has shown that smaller rotor-stator gap is beneficial especially for fine and ultra fine particle flotation. However, recent results indicate that also recovery of coarse particles can be increased at the same time. It has also been observed that motor speed adjustment at the same time has a significant effect on the results. (Bilney et al., 2006)

113

3.2.4 The benefits of the recent improvements The potential of the recent improvements can be illustrated in a simple chart. Typical power draw curves of a flotation cell equipped with VSD and FloatForce mechanism and a cell with conventional fixed speed drive with the old OK-mechanism are presented in Figure 6. The main benefit of the new arrangement is the possibility to adjust the cell during normal operation. The optimal flotation cell operation point with VSD and FloatForce rotor can be found in the two-dimensional control space whereas the operation range of the old drive is bound to one curve. For a self-aspirated cell the control range is only one point whose position depends on slurry density. The main benefit of the new arrangement is the possibility to adjust the cell during normal operation. Adjustment of the rotor-stator gap provides another offline parameter that creates a family of curves for each gap value.

Figure 6. Typical power draw curves of VSD with FloatForce mechanism.

mechanism and standard drive with OK-

3.2.5 Which parameters do actually make the difference? A lot of research work and discussion is going on in the search for optimum metallurgy of each ore type. Among the most common tasks is optimization of basic parameters like bubble size and installed power. But how do we actually generate an optimal bubble size in a modern large flotation cell? Controlling bubble size distributions in a laboratory scale flotation machine is a subject for many studies, but research done with large-scale cells is much more limited. Scale up of the findings to industrial cell sizes is not very straightforward either.

114

How is a powerful flotation machine determined? Is installed motor power the same as high measured total power consumption of a mixer drive? By installing a 250 kW motor instead of 100 kW the measurable power consumption of the cell increases for sure due to unfavorable operating point of the motor. When the mixer is updated to correspond the larger motor, what is really sought after with the power increase? Optimal and higher speed for optimal bubble size distribution or more shear? Moving the slurry around in the tank does not necessarily make a big difference if the basic parameters remain the same. In some cases extra power may only make the situation worse by causing unstable cell surface and froth. Today, we have a possibility to find out answers to at least some of these questions by selecting new mechanism design and VSD.

4.

SUMMARY

Let us look at the importance of the energy efficiency and selection of the optimal equipment size by considering a flotation plant requiring, for example, 1800 m3 of flotation volume. This requirement can be fulfilled via five possible scenarios: 18 individual 100 m3 cells in two rows of nine. Nine 200 m3 cells as a single line of nine. Nine 200 m3 cells with variable-speed drive mechanisms enabling optimization of the rotation speed, say, five percent lower than the nominal rotation speed. Six 300 m3 cells as a single line of six. Six 300 m3 cells with variable-speed drive mechanisms enabling five percent lower rotation speed. Relative investment, energy, reagent and maintenance costs are presented in Figure 7. The trend towards larger units is obvious when looking at the comparison. Another leap of similar magnitude can still be taken if the process enables lower mixer speed. The example has been calculated with 5 % speed change; since the power consumption of a mixer is proportional to the third power of the rotational speed this results in 15 % savings of the energy cost.

115

Figure 7. Relative investment costs of different flotation options.

As mentioned above, energy efficiency has a significant effect on the life cycle economy. One aspect of it is that at least 75 percent of the industrial world s energy is produced using fossil fuels that result carbon dioxide emissions. The annual difference in energy consumption between the traditional 100 m3 cells and the latest 300 m3 cells with optimized rotation speed is approximately 4,72 GWh/a which corresponds to approximately 4,32 million kilograms of carbon dioxide if the energy is produced in a fossil fuel power plant. The amount equals to hundred average passenger cars driving approximately 270 thousand kilometers. To prevent climate change, governments are tightening emission standards and emission trading is implemented to control pollution. It is likely that saving energy will be more in the focus in future.

116

REFERENCES Bilney, T., MacKinnon, S., Kok, J., Assessment of High Shear Stator Performance at Kanowna Belle Gold Mine. In: Metallurgical Plant Design and Operating Strategies 2006, Perth, Australia, September 18-19, 2006. Coleman, R.G., Urtubia, H.E., Alexander, D.J., A Comparison of BHP-Billiton s Minera Escondida Flotation Concentrators. In: Canadian Mineral Processors 38th Annual Operators Conference, Ottawa, Canada, January 17-19, 2006. Froehling, M., Mohns, C., Roman, E. Grady, P., A History of Improvements at the Kemess Concentrator. In: Canadian Mineral Processors 37th Annual Operators Conference, Ottawa, Canada, January 18-20, 2005. Grnstrand, S., Niitti, T., Rinne, A., Turunen, J., Enhancement of Flow Dynamics of Existing Flotation Cells. In: Canadian Mineral Processors 38th Annual Operators Conference, Ottawa, Canada, January 17-19, 2006. Lyytikinen, A., Reliability engineering handbook. Technical Research Centre of Finland, Research Notes 678, 147 p. + app. 6p., ISBN 951-38-2633-3, Espoo, Finland, 1987.

117

SYSTEM ASPECTS ON IN-PIT CRUSHING AND COARSE-PARTICLE PUMPING

Anders Sellgren Lule University of Technology

Abstract The feasibility of pumping tertiary crushed raw ore together with mine water with centrifugal slurry pumps in series is briefly demonstrated in a comparison with truck transport to the processing plant. Pumping can also be effective if considered early in the mine planning because of less waste rock handling. Furthermore, in large-scale open-pit mining, only a 0.5 m-diameter pipeline is required for up to 20 Mtonnes per year. Tertiary crushing with an autogeneous type of crusher giving a comparatively wide particle size distribution may gain some effectiveness in the pumping, however the system integration potential can mainly be related to possible energy savings and increased throughput in grinding operations as outlined by Lindqvist (2008). 1. Introduction In open pit mining the blasted rock product is normally loaded on trucks and transported to crushing and further beneficiation and processing. Trucks provide a flexible way of handling and transportation waste rock and raw ore from the pit. The ore may be trucked all the way to crushing at the processing plant on the surface or some distance in the pit to a crushing station. Further transportation with a belt conveying system requires crushing to at least about 0.4 m. The feasibility of in-pit crushing and the transportation to processing and waste dumps is highly site specific. Generally, primary crushing features particles smaller than 250 mm with further reduction to about 70mm in a secondary stage. With tertiary crushing a final product may have max. particle sizes of 10 to 15mm dependent on the closed circuit arrangements. 1.1 Slurry pumping In mining, conversion of particles into a water mixture (slurry),is often part of the normal processing. Therefore, from a systems point of view, hydraulic transportation (slurry pumping) would be considered as a natural alternative, all the way from working face to the final processing. Various applications of hydraulic transportation as a link in integrated systems for mining processing and waste handling are shown schematically in Figure1. Open pit mines often have a drainage water pipeline pumping system to the surface where the water may be used in the processing plant. Coarse particle slurries with sizes of up to and over 100 mm can be pumped with centrifugal slurry pumps, Sellgren and Addie (1995).However, pumping of particles of that size is energy demanding in horizontal pipelines and pump and pipeline wear may be severe, therefore normally only short distances are considered.

119

MINERAL PROCESSING

HYDRAULIC TRANSPORT OF ORE

SYSTEM INTEGRATION

HYDRAULIC TRANSPORT OF WASTE

MINING

Figure1.Hydraulic transportation as a link in integrated systems for mining, processing and waste handling, Sellgren (2000). 1.2 Crushing Impact type of crushers are based on rock against rock, i.e. autogeneous crushing and mainly suitable in the tertiary step in hardrock mining. This type of crushers often produces a finer material with a wider size distribution than cone crushers. In the beneficiation chain from mine to final product the crushed raw ore is ground wet in tumbling mills of ball-category or of semi-autogeneous/autogeneous-type. A finer product with a wide size distribution from a preceding tertiary crushing step may reduce the energy consumption and increase the capacity of the grinding operations, Lindqvist (2008). Lindqvist (2008) estimated schematically the effectiveness of integrating the fine particle generation from a crusher of vertical shaft impact (VSI) type in a grinding schema in terms of energy consumption. He used data presented by Oghazi et al. (2007) for three step iron ore grinding. The resulting estimations together with operating experience from a full scale installation indicated an energy saving potential of the order of 10%. The performance of a cone crusher and an autogeneous crusher of VSI-type has been investigated in a comparison by Lindqvist ( 2008).The crushers operated in close circuit with a 9mm final screen aperture size. The results have here been used to represent the principal differences in particle size distribution between the two types of tertiary crushing, Figure 2.

Figure 2.Comparison of overall particle size distributions for secondary and tertiary cone crushing and a VSI-machine. Schematically from Lindqvist (2008). 120

It follows from Figure 2 how the VSI-machine produced a finer product with a broader size distribution than the cone crusher. The average particle sizes were 1.5 mm and 3mm, respectively. The corresponding content of particles smaller than 0.06 mm were 6 , and 4%, respectively. The recirculating load was slightly higher with the VSI-unit than with the cone crusher. When comparing the two crushers for the same net capacity, the tendency was 6.5% lower energy consumption for the autogeneous crusher, from 1.55 to 1.45 kWh/ton. The average value of 1.50 kWh/tonne will be used in this study. The finer product and the wider size distribution are also beneficial to the power consumption and maintenance in the slurry pumping. The pumping effectiveness increases substantially with a wide and even particle size distribution, for example when pumping tailings from wet milling together with waste rock to one disposal area, so-called co-disposal, see for example Sundqvist and Sellgren(2004). Bengtsson and Evertsson(2006) reports that a VSI-crusher produces less flaky particles which sometimes is desirable, for example for the flow of concrete. 2. Objective and scope The objective is to briefly evaluate the technical-economical feasibility of pumping raw ore from open pit mines based on in-pit tertiary crushing and integration with the mine water drainage system. Any further evaluation of integrating the fine particle generation from the crushing with the grinding operations will not be included here, in addition to what have been covered with reference to Lindqvist ( 2008). 3. Slurry pumping The maximum particle size and the particle size distribution determine together with the concentration of particles in the slurry the choice of pipeline diameter, the required operating velocity, pressure requirement and wear. The energy requirement in horizontal slurry pumping systems can typically be about 25% smaller for the wider size distribution in Figure 2.It is a combined effect of less pipeline friction losses (see for example Wilson et al. 2007) and the influence on the performance of the pumps, see for example Addie et al. (2007). In a pumping system dominated by vertical pipeline transport, the feasibility will be much less affected by the particle size, the symmetry of the flow diminishes the influence of the particle size. Therefore, in a vertical system, it is mainly the effect on the performance of the centrifugal pumps. In that case, the reduction is limited to about 5%.For the system considered here with transportation in an inclined pipeline, the effect will thus be less than 25%.A similar discussion can be related to the wear in pipelines and pumps, however any wear effects related to the different size distributions are not evaluated in this study. 3.1 A schematic example The feasibility will be investigated in a schematic example for 600 tonnes per hour of tertiary crushed raw ore transported in a pipeline a total vertical length of 275 m from the pit to the processing plant located a total horizontal distance of 300m away from the crushing station. The slurry pumping from the crushing station in the pit to the surface is assumed to take place in an inclined pipeline at 45-55o to the rim and then horizontally to the adjacent plant. In order to simplify the hydraulic analysis, the pipeline is conceptually assumed to consist of vertical and horizontal parts. The hydraulic design shows that a pipeline diameter is 0.25 m is sufficient for a concentration by mass of about 55%.The pumping energy consumption were

121

estimated to 2.48 and 2.23 kWh/tonne with the conventional and autogeneous crushing systems, respectively, which corresponds to about 10% less for the autogeneous crusher. It is here decided to use an average value of 2.35 kWh/tonne for the pumping, dependent on the strong influence of vertical transport here resulting in small differences between the crushing methods. No distinction will be made between the two tertiary crushing methods in the example. The resulting operation conditions for the pumping system are summarized in Table 1. Table 1.Operating data for slurry pumping of 600 tonnes/h (4.8 Mtonnes/year) in a 0.25 m diameter pipeline with centrifugal pumps in series from in-pit tertiary crushing to further grinding in the processing plant. The power requirement for drainage water corresponds to with an assumed pipeline diameter of 0.3 m for 490 m3/h (4.3Mm3/year). _________________________________________________________________________ Water requirement 490 m3/h or 4.3 Mm3/year Total number of pumps in series 7 plus 2 boosters along the pipeline Power requirement, for slurry, 1410 kW or 2.35 kWh/tonne Power requirement for drainage water 545 kW or 0.90 kWh/tonne _________________________________________________________________________ Ten percent of the power requirement for drainage water in Table 1 here assumed to be used to deliver water to the crushing station from the water pond at the lowest elevation of the pit. When comparing conventional handling and hydraulic transportation the cost of pumping drainage water should be deduced from the slurry pumping alternative when comparing various alternatives of transportation, see Table 2. Table 2.Energy requirement in kWh/tonne related to the availability of water for the pumping of slurry and water in Table 1 , 275 m vertically and 300 m horizontally to the processing plant. ___________________________________________________________________________ Pumping In-pit drainage water availability 100% 0% Slurry pumping 2.35 2.35 Drainage water pumping -0.80 Total net energy requirement 1.55 2.35

It follow from Table 2 that the net energy consumption with all required water available in the mine in Table 1 is 1.55 kWh/tonne compared to 2.35 with no water available. 4. An overall cost comparison 4.1 Trucking The overall feasibility of in-pit secondary and tertiary crushing, independent on the crushing method, together with slurry pumping can be put into perspective when comparing the cost when using large mining trucks. State-of-the-art information of utilization of trucks was mainly obtained from Renstrm (2007).

122

Transport of 600 tonnes of primary crushed raw ore per hour from a pit depth of 275 m and then 300 m horizontally to the processing plant can be handled with one big truck loading 220 ton in a fleet of several trucks serving raw ore and waste rock handling. With an average utilization time of 67%.i.e. 16h a day, then the capacity is 147 tonne/h on a long time basis. Truck transportation 275 m vertically means 2750 m on inclined roads with slope 1:10.The round time for a truck including loading and unloading was estimated to 14.7 minutes, which corresponds to 600 tonnes per hour. The corresponding fuel consumption was predicted to about 0.26 m3/h, matching a fuel cost of about 1200 SEK/h or 2 SEK/tonne, assuming 4500 SEK/ m3 (1 SEK= 0.11Euro=0.16US$, January 15,2008) For the example considered here the truck fuel and capital cost can each roughly be estimated to be of the total cost per hour. This gives an operating cost of about 6 SEK/tonne. With about 0.5 SEK/tonne for road maintenance, then the total operating cost may end up with 6.50 SEK/tonne or about 31 MSEK per year. The investment cost for a truck of the type considered here may be of the order of 30 MSEK. 4.2 Pumping The economical feasibility of the pumping was evaluated based on investments in fully equipped and wear protected pumps and pipelines( rotated in intervals), power supply, foundations and buildings and pump seal water supply.Furthermore,20% was added for projecting and unforeseen items. The operating costs include power for pumps, heat , light and maintenance and personnel. Yearly maintenance costs for pumps and pipeline was simply based on full replacements after 2 years. 4.3 Overall cost comparison The brief estimation of costs is summarized in Table 3. Table 3.Estimated overall costs for pumping 600 tonnes/h of crushed raw ore compared to truck transportation in the considered example. Assumed electrical energy cost= 0.30 SEK/kWh and diesel fuel cost= 4500 SEK/m3 Slurry pumping Investments 75 MSEK Truck transport 30 MSEK

Operating cost MSEK SEK/tonne MSEK SEK/tonne Energy: 4.8 1.0 (Electrical) 9.6 2.0 (Diesel fuel) Maintenance and labour 4.8 1.0 21.6 4.5 Totally 9.6 2.0 31.2 6.5 ___________________________________________________________________________

5. Discussion and conclusions It follows from Table 3 that slurry pumping can be an interesting alternative for the transportation of raw ore based in in-pit tertiary crushing and use of drainage water, in case of truck transportation of raw ore from an in-pit mine to a nearby processing plant.

123

It can also be seen from Table 3 how the truck transportation is dominated by operating costs while the pumping system means a comparatively large investment. The dependence on local conditions and the essential difference between the two modes of transportation means that it is difficult to draw any far-reaching conclusions of the results in this schematic example. With insufficient inflow of drainage water compared to the planned capacity of solids, reuse of water from the processing plant and make up water will be needed, see Table 2.The corresponding increase in pumping energy cost will be about 0.6 SEK/tonne in Table 3. The cost for the additional secondary and tertiary crushing has not been included here in Table 3. This expresses a situation were the cost balances downstream benefits obtained in the further beneficiation operations. In this integrated system, autogeneous crushing with the wider resulting size distribution may has a special potential through reduced energy consumption and increased throughput, as discussed earlier with reference to Lindquist (2008).Dependent on the effectiveness in this integration, parts of the in-pit additional crushing costs may be associated with the slurry pumping system. The energy requirement for the secondary crushing step was here roughly estimated to 1.2 kWh/tonne. Together with the average consumption of 1.5 kWh/tonne in the tertiary step then the crushing after the primary step requires about 2.7 kWh/tonne which can be compared to the corresponding energy consumption for the slurry pumping in Table 2. In large-scale open pit mining a slurry pumping system has an substantial capacity in relation to the required space. A 0.5 m in diameter pipeline has the capacity of nearly 20 Mtonnes of dry solids per year. Slurry pumping as described here may also be an economically interesting alternative in small and narrow open pit mines, especially if the concept is considered early in the development of the mine. With a substantial part of the raw ore transported in a pipeline, the requirement of wide roads may be smaller which means a smaller amount of waste rock, Figure 3.

Figure 3.The dashed marks represent waste rock volumes which may not be needed to take out for the alternative with slurry pumping of raw ore, due to less requirement for wide roads within the pit. 124

Potential cost savings can be achieved from less required handling, transportation and storage of waste rock as schematically shown in Figure 3.Furthermore,in an open-pit mine planned for underground development ,an extension of an open pit slurry pumping system can be effectively developed to a hydraulic hoisting system, Sellgren et al.(1989). 6. References Addie G, Roudnev.A., Sellgren A.(2007): The new ANSI/HI centrifugal slurry pump standard, The journal of the s. African institute of mining and metallurgi,vol.107,June,,pp 403-409. Bengtsson,M. and Evertsson C.M.(2006):Measuring characteristics of aggregate material from vertical shaft impact crushers, Minerals Engineering, Volume 19,Issue 15,pp 1479-1486 Lindqvist M.,(2008): Energy considerations in compressive-and impact crushing of rock, in press, Minerals Engineering Oghazi P, Plsson B. and Tana K.(2007):An attempt to apply traceability in grinding circuits. Proceedings, conference in mineral processing, February 6-7,Lulea University of technology Renstrm A. (2007):Truck fleet utilization and fuel saving in Aitik, Proceedings, 6th Large Open Pit Mining Conference, Ausimm, Perth, Australia. Sellgren A., Jedborn A., Hansson K. (1989): Hydraulic hoisting: an economic alternative in the deepening of underground mines. Mining Engineering, August, pp 857-859 Sellgren A. (2000): Hydraulic transport in integrated mining systems of tomorrow, Keynote lecture. Proceedings, 10 th. Int. Conf. on transport and sedimentation of solid particles, Wroclaw, Poland. Sundqvist and Sellgren, A (2004):Large-scale testing of coarse waste rock pumping, Proc. 16th,Int.Conference on Hydrotransport, Santiago, Chile. Sellgren, A. and Addie, G. (1995): Cost-Effective Pumping of Coarse Mineral Products Using fine Sands. Proceedings 8th International Freight pipeline Society Symposium, sept. 14-16, Pittsburg, U.S.A. Wilson K.C., Addie G.R., Sellgren A. Clift R.(2006):Slurry Transport using centrifugal pumps, Third Edition, Springer N.Y., U.S.A.,432 pages

125

PROCESSUTVECKLING OCH ENERGIBESPARING I MINERALINDUSTRIN EXEMPEL FRN MINBAS II PROGRAMMET OCH MINFOS ENERGIPROJEKTPAKET Marianne Thomaeus, Research director, MinFo- Swedish Mineral Processing Research Association Drottning Kristinas vg 26 S-100 44 Stockholm, Sweden
Sammanfattning Under r 2007 startade tv strre utvecklingssatsningar inom den svenska industrimineral, bergmaterial och naturstensindustrin. Den ena r utvecklingsprogrammet MinBaS II; ett 4-rigt program riktat mot produktions- och processutveckling, produkt- och applikationsutveckling samt milj - hllbar utveckling. Total budget r 55 Mkr varav staten bidrar med 26 Mkr via Sveriges Geologiska Underskning och industrin med resterande del. Den andra r ett projektpaket fr energibesparing och CO2-reducering inom industrimineralsektorn. Paketet som pgr 2007-2010 bedrivs inom MinFo med en total budget av 37,8 Mkr. Industrin finansierar huvuddelen och Energimyndigheten stder paketet med 13,1 Mkr. Programplanerna bygger p behovsanalyser och fretagen har prioriterat de utvecklingstgrder som behvs fr att strka konkurrenskraften och tillvxten. Ett 50-tal fretag och anlitade forskare vid universitet, hgskolor och institut har varit engagerade i detaljplanering av projekten. Programverksamheterna presenteras vergripande och exempel ges frn pgende projektarbeten inom frmst produktions- och processteknikomrdet. Abstract During 2007 two major industrial development programs were started in the Swedish industrial minerals, aggregates and stone industries. The industrial development program MinBaS II, is planned for 4 years with a total budget of 55 MSEK. The Swedish Geological Survey supports the program with 26 MSEK and the industry will finance the rest. The projects are focusing on production and process development, product development, application technologies and sustainable development. A new project package aiming at increased energy efficiency and CO2 reduction in the industrial minerals sector will be conducted during 2007-2010 with a total budget of 37,8 MSEK. The industrial partners are financing the major part and the Swedish Energy Agency is supporting the work with 13,1 MSEK. The program plans are based on strategic analyses and the industry has set the priorities for the development work needed improve competitiveness and growth. 50 companies and selected research groups at universities and institutes have been engaged in the detailed planning of the projects. Over views of the two programs are presented as well as examples of projects in primarily the production- and process technology areas.

1. Bakgrund Som tidigare informerats om vid Konferensen i Mineralteknik 2006 sammanstlldes under 2005-2006 p initiativ av dvarande regeringen sk Branschprogram fr ett antal tunga industribranscher i Sverige. I programmen identifierades ett antal omrden vilka krver tgrder fr att strka industrins framtida konkurrenskraft och tillvxt. Forskning och utveckling har en central roll bland dessa omrden. Branschprogrammens arbete har drefter lett till att ett antal forskningsprogram statats i landet vilka finansieras av industrin med std av staten.

127

Under r 2007 har startat ett nytt flerrigt utvecklingsprogram fr industrimineral, bergmaterial och naturstensindustrin kallat MinBaS II- programmet och ett 4-rigt projektpaket inom MinFo riktat mot energibesparing vid framstllning av industrimineralprodukter. Som bas fr programarbetet ligger gemensamma behovsanalyser och fretagen har prioriterat vilka utvecklingstgrder som behvs fr att strka konkurrenskraften och tillvxten inom industrimineral-, bergmaterial och naturstensindustrin. Inledningsvis grs en kort presentation av branscherna och de prioriterade utvecklingsomrden. Drefter beskrivs programarbetet vergripande och exempel ges p pgende projekt inom produktions- och processomrdet 1.1 Produktion av bergmaterial, industrimineral och natursten i Sverige Produkterna frn fretagen i dessa delbranscher anvnds som rvara, tillsatsmedel eller frdig produkt i ngon form i praktiskt taget all tillverkning och varukonsumtion i vrt samhlle. Den svenska produktionen av material i olika former uppgr till ver 103 miljoner ton per r till ett ungefrligt vrde av 9 miljarder kronor. Sammanlagt r ca 10 000 personer verksamma inom de tre delbranscherna. Fretagen i branschen r spridda ver hela landet. Bergmaterial i form av krossberg, naturgrus och tervunnet inert material utgr basen i infrastrukturellt byggande. Bergmaterialindustrin frser oss med det stenmaterial, som bygger upp vgar och jrnvgar. Bergmaterial anvnds ocks som insatsrvara i asfalt och betong. Produktionen (2006) var drygt 92 Mton och Sveriges till volymen strsta industriprodukt. Antalet tillstndsgivna tkter r ca 2600. Den strsta produktionen sker i storstadsregionerna i Vstra Gtaland, Stockholms ln och Skne. Industrimineral t.ex. kalksten, kvartsit, fltspat, leror r viktiga insatsvaror vid tillverkning av bl.a. cement, tegel, glas, eldfasta produkter, papper, plast, stl och andra metaller samt inom det miljtekniska omrdet. Produktionen (2006) var ca 11 Mton. Karbonatstenar r de volymmssigt strsta med 9 Mton. Det fanns 44 st producerande tkter 2006. Gotland r det dominerande lnet vad avser kalkstensproduktionen, vriga produktionsstllen finns huvudsakligen i Vstmanland, Nrke, Dalarna, stergtland Vstergtland och Skne samt tv tkter i Norrbotten. Natursten har anvnts fr byggande sedan urminnes tider. Fr husbyggnad anvnds sten bde in- och utvndigt fr golv, trappor och fasader. Fr anlggningsndaml utnyttjas sten i trappor, marksten, kantsten, murar etc. Natursten till gravvrdar och skulpturer r andra produktomrden inom branschen. Produktionen (2006) var 0,9 Mton brutet material. Naturstensindustrin i Sverige utvinner ett 70-tal stensorter varav ett flertal r unika. Dessa r uppskattade internationellt sett och har ett hgt marknadsvrde. Det fanns 57producerande tkter 2006. Stor produktionen sker i Skne, Halland, Bohusln, Kalmar, land, stergtland, Vstergtland. Det finns ett antal tkter i Dalarna, Jmtland och i Vsterbotten. 1.2 Prioriterade utvecklingsomrden Industrins prioritering av utvecklingsomrden har sin grund i ett antal utmaningar som mste mtas om fretagens konkurrenskraft skall kunna bibehllas. Gemensamt fr samtliga branscher r behovet att skerhetsstlla tillgngen till fyndigheter och mjligheter att driva verksamheterna under frutsgbara villkor under lngre tidsperioder. Nedan ges ngra mer specifika exempel frn de olika delbranscherna.

128

Bergmaterialindustrin Byggmaterialsektorn str infr en stor utmaning nr naturgrusmlet (ett delml av det nationella miljmlet God bebyggd milj) skall uppns. Uttaget av naturgrus skall minska frn dagens ca 20 miljoner ton till 12 miljoner ton r 2010 och till 1 miljon ton r 2020. Den enda rimliga ersttningen fr naturgrus r krossgrus (stenmjl). Vidare str bergmaterialleverantrerna infr en stor omdaning i och med att mnga sm lokala grustkter kommer att stngas efterhand som nya tkttillstnd inte lngre beviljas och behovet av krossgrus skall tillfredsstllas genom etablering av nya strre bergtkter. Detta medfr att transportavstnden till betongindustrin kommer ka med kad drivmedelsfrbrukning som fljd. Detta krver ett tekniksprng och kat behov av infrandet av nya, effektiva produktionsmetoder inom bergmaterialbranschen. Fr betongbranschen gller framfr allt att kunna erstta grusfraktionen dr man idag nstan uteslutande anvnder naturgrus. Betongbranschen frbrukar ca 40 % av uttagen volym naturgrus. Krossgrus har visat sig ge en generell kning av cementbehovet. Fr att uppn miljmlet mste drfr betongbranschen lra sig att p ett ekonomiskt och miljmedvetet stt tillverka betong med krossgrus. Industrimineral En stor utmaning i mineralbranschen r kande energipriser, oskerheten om tillgngen p energi samt utformningen av den framtida handeln med utslppsrtter fr CO2. Framstllning av cementklinker och kalk sker genom energikrvande brnnprocesser vid hga temperaturer under samtidig emission av CO2 frn tv olika kllor; frn brnslet(fossila brnslen dominerar idag) och frn rmaterialet kalciumkarbonat (kalksten). Frsk med att introducera biobrnslen och avfallsbaserade brnslen har redan startat i cementindustrin; till en mindre del i kalkproduktionen. Det krvs dock en kombination av processfrndringar och ny teknik fr att kunna ka andelen alternativa brnslen i processerna. Skillnaderna i att anvnda biobrnslen i jmfrelse med fossila brnslen r inte klart utredda. Inom cement- och kalkproduktionen r inverkan p produkternas kvalitet en viktig del. De nya teknikerna fr avskiljning av CO2 frn rkgaserna mste utvecklas och testas. De idag bst tillgngliga teknikerna fr CO2 separation frn rkgaser, som kemisk absorption med aminer, krver mycket stora energiinsatser per ton borttaget CO2. Den industriella implementeringen r lg idag, men infrandet av denna typ av installationer kan bli en ndvndighet fr att kunna mta framtida miljkrav, vilket skulle medfra ett kraftigt kat energibehov. Industrimineralindustrin frbrukar ocks stora mngder energi vid framstllningen av avancerade fillerprodukter genom finmalning, klassering och i separationsstegen. Naturstenindustrin Den hgsta prioriteringen inom naturstensbranschen r att ka anvndningen av sten i Sverige, som ligger klart under per capitafrbrukningen i vriga Europa. Fr att mta konkurrensen frn importen r av yttersta vikt att intressenter som arkitekter, konstruktrer och myndigheter fr tillgng till bsta tnkbara tekniska informationsverktyg och utbildningsmaterial rrande applikation av natursten. Vidare krvs att den hgre utbildningen fr blivande arkitekter och byggare fr tillgng till undervisningsmaterial och att samarbete kan utvecklas mellan naturstensbranschen och befintliga utbildningscenter.

129

2. MinBaS II - programmet MinFo tillsammans med Sveriges Bergmaterialindustri SBMI och Stenindustrins Forskningsinstitut AB - SFI AB driver i det gemensamma bolaget MinBaS AB det 4riga utvecklingsprogrammet MinBaS II. Bolaget MinBaS AB: s styrelse br det vergripande ansvaret fr programmets genomfrande gentemot SGU. Till bas fr arbetet ligger en framtagen av industrin framtagen programplan med budget och de direktiv som ges i beslutbrevet fr den statliga medfinansieringen. Programmet har en total budgetomfattning p 55 miljoner kronor, varav SGU har beviljat totalt 26 miljoner kronor i std till utvecklingsarbetet. Industrins egna insatser utgr resterande finansiering. Programmet startade den 15 juni 2007 och slutrapporteras den 1 juli 2011. Nrmare ca 50 talet fretag har redan under programmets frsta halvr varit engagerade i samarbete med universitet/hgskolor, branschinstitut och organisationer, varav smfretag frn stenindustrin r ett 20-tal. Ett antal kommuner, lnsstyrelser och statliga institutioner har involverats i samband med regionala projektarbeten. MinBaS II-programmet r uppdelat i fem programomrden enligt fljande:
MinBaS AB STYRELSE PROGRAMSEKRETARIAT

Programomrde 5 Planering, teknikspridning, implementering

Budget 10,0 Mkr 5,0 fr MinBaS 5,0 Mkr fr industrin

STYRGRUPPER FR PROGRAMOMRDEN

Programomrde 1 Produktions- / processutveckling

1. Nya frunderskningsmetoder, materialkarakterisering och tktplanering 2. Bergteknik - Brytning 3. Mineralteknik 4. Krossat ballastmaterial i Framtida betong 5. Modellering, optimering av produktionsprocessen 6. Transporter och logistik

Programomrde 2 Produktutveckling
1. Utveckling av nya anvndningsomrden - Vgbyggnad, Deponibyggande, Filter - vattenrening 2. Utveckling av industrimineralbaserade produkter 3. Europastandardisering Budget 9,9 Mkr 4,7 fr. MinBaS 5,2 Mkr fr. industrin

Budget 14,5 Mkr 7,0 fr. MinBaS 7,5 Mkr fr. industrin

Programomrde 3 Milj - Hllbar utveckling

1. a Miljpverkan frn anlggningar 1. b Arbetsmilj 2. Efterbehandling 3. Sustainable Development Sustainable Dev. Indicators (SDI) mm. Budget 4,0 Mkr 2,0 fr. MinBaS 2,0 Mkr fr. industrin

Programomrde 4 Applikationsutveckling Stenindustrin

1. Natursten och industrimineral inom Sveaskogs marker i sdra Sverige 2. Teknikverfring till freskrivande led 3. Avjmningsmassor, fstmassor, fr vissa kritiska applikationer 4. Applikationsteknik: Stenhandboken. 5.a Naturstens tekniska egenskapers betydelse fr olika applikationer 5.b Fogar i markbelggning 6. Mallar fr egenkontroll Material resp. utfrande 7. Goda exempel med natursten inom arkitekturen 8. Prisvrda konstruktioner med natursten 9. Naturstensmurar i det offentliga rummet 10. Internat. harmonisering av standarder & krav 11. Stenakademien Budget 16,55 Mkr 7,3 fr. MinBaS 9,25 Mkr fr. industrin

130

2.1 Omrde 1 Produktions och processutveckling Prioriterade huvudprojektomrden Nya frunderskningsmetoder, materialkarakterisering och tktplanering Bergteknik Brytning Mineralteknik Krossat material i Framtida betong Modellering, optimering av produktionsprocessen Transporter och logistik Projekten r riktade mot utveckling av processer, metoder och system som leder till resurseffektivisering och bttre produktkvalitet. En viktig del i programarbetet r introduktion av produktionsteknik i bergmaterialindustrin som underlttar vergngen frn produktion av naturgrus till hgvrdiga krossprodukter. Speciell vikt lggs vid att introducera ny teknik och system som gr att fragmenteringsprocessen kan effektiviseras och kontrolleras och slutprodukternas kvalitet frbttras. De startade projekten, svl utveckling av underskningsmetoder som produktionsutvecklingsprojekten, rr fragmenteringskedjan sprngning krossning. De projekt som rr modellering och styrning r ven de riktade mot krossprocessen. Inom den processtekniska delen dominerar utvecklingen av torra processer; frmst klassering och separation av fina partiklar. Detta gller svl bergmaterial som industrimineral. Utvecklingen av processteknik fr krossat ballastmaterial till betong har givits speciellt projektutrymme. Energieffektivisering r en viktig del i arbetet. Industrimineralindustrin har prioriterat projekt mot nya energieffektivare tekniker fr finmalning och klassering. En studie grs fr att belysa mjligheterna att energieffektivisera de interna transporterna i tkterna. Exempel p pgende projekt: 2.1.1 Nya frunderskningsmetoder, materialkarakterisering och tktplanering Karakterisering av mikrosprickor orsakade av fragmenteringsprocesser Utfrare: SP CBI Betonginstitutet AB Syfte och ml: Att detektera eventuell mikrospricktillvxt vid olika processer i ballastproduktionen. Genom att ha bra kontroll p vad som hnder med bergmaterialet vid sprngning och krossning har man i sin tur bttre frutsttningar fr styrning av produktionen vid frndringar i en tkt. Arbetet har ocks intresse fr industrimineralfretagen och stenindustrin. De nya karaktriseringsmetoder som skall utvecklas i projektet kan ven bli anvndbara fr sdan produktion. Mlet r att etablera nya metoder fr materialkaraktrisering, ka frstelsen fr hur olika parametrar i bergmaterialet pverkar fragmenteringsegenskaperna och i frlngningen produktens egenskaper. Genomfrande: Projektet kommer att genomfras i en av NCCs bergtkter. Projektet har anknytning till det sprngtekniska projektet. 131

Andra projekt under planering: - Utveckling av markradar som prospekteringsverktyg inom MinBaS-omrdet, dr syftet r att testa markradarmetoden i olika situationer fr att kunna utreda anvndbarheten i MinBaS-branscherna, - Utveckling av MWDtekniken, dr syftet r att primrt underska i vilken grad MWD-tekniken kan anvndas som kvalitetsskring och kvalitetskartlggning i industrimineralfrekomster, Inventering av tktplaneringsverktyg, dr syftet kunna initiera arbete kring en nedskalning av de kommersiella gruvplaneringsprogrammen och anpassa dem fr ppna tkter och berg- och industrimineralindustrins behov. 2.1.2 Bergteknik Brytning Optimal fragmentering genom sprngning II Utfrare, Swebrec, LTU Detta kommer att utgra det huvudsakliga arbetet inom projektomrdet under programperioden. Resultat frn tidigare forskning har kat frstelsen fr den mekaniska snderdelningen av berg. Detta i kombination med utvecklingen av sprngtekniken, nya mtmetoder samt modern informationsteknik fr mtning, verfring, tolkning och styrning har kat mjligheterna att utveckla brytningstekniken. Syfte och ml: Att skapa en mer selektiv brytning med lgre finandel, mindre vibrationer, och kostnadseffektiva metoder. Projektet har som mlsttning att validera de arbeten med designkurvor i Vndletkten som gjordes under MinBaS I programmet. Vidare skall utvrderas anvndning av elektroniksprngkapslar med syfte att f bde finare styckefall och andra frbttringar vid pallsprngning. Valideringsarbetet utgr frsk i en tkt. Det r ven mjligt att behandla lastbarhet och lastcykeltider i projektet. Genomfrande: Arbetet skall genomfras i tkter med annan geologisk struktur. En omfattande frsksplan har utarbetats dr specifika laddning och hlmnster varieras samt sprngtekniken. Valideringsarbetet utgr frsk i en tkt. Ett antal tkter har identifierats dr frsksarbetet kan genomfras. Planerade vriga arbeten inom omrdet r att gra en state of the art studie kring icke explosiva metoder. 2.1.3 Mineralteknik Processteknisk utveckling fr exploatering av vanadinfyndigheten Sumssjn Utfrare: Fretaget Svenska Vandin AB i samarbete med processexperter och forskare. Syfte och ml: Den vergripande mlsttningen med projektet r att exploatera fyndigheten Sumssjn i Hlsingland fr att starta en ny produktion av vanadin i landet. Syftet med projektarbetet, vilket berr programomrde 1 och 2 inom MinBaSprogrammet, r att skapa tekniska och ekonomiskt underlag till en frdjupad lnsamhetsstudie inkl. detaljplaneringsunderlag fr produktionsanlggningarna. Genomfrande: Svenska Vanadin AB har under 2006 genomfrt en genomfrandestudie. I samarbete med institutet MeFos har det utvecklats en ny direktreduktionsprocess, dr ferrovanadin framstlls genom en smltprocess direkt ur ett vanadininnehllande magnetitkoncentrat. Denna process frvntas ge en mycket energieffektiv produktion jmfrt med konventionella framstllningsmetoder. Ingen avkylning av mellanprodukter sker fre slutprodukt. Energi i form av vrme frn 132

smltprocessen skall ocks anvndas fr att frvrma magnetitkoncentratet. Mjligheter att ocks tervinna vanadin ur vanadinrika slagger frn stlverk ges genom denna process. Det processtekniska utvecklingsarbetet inriktas mot att utveckla energieffektiva malningsmetoder och magnetsepareringsmetoder som kan ge optimala koncentrathalter och utbyten. I arbetet ingr ocks att framstlla ett strre tonnage av ett representativt magnetitkoncentrat fr de direktreduktionstester som skall genomfras. Den nyutvecklade direktreduktionsprocessen skall verifieras genom smltningsfrsk i strre skala av vanadininnehllande magnetit. Kommersiella produkter skall utvecklas av alla de produkter som kommer frn direktreduktionsprocessen svl ferrovanadin, tackjrn som nya produkter ur slaggen, ett aluminiumkalcinat. Mineralteknik - Finmalning, klassering och separation Beslut har ocks tagit om att detaljplanera frsk med nya tekniker fr finmalning samt nya tekniker fr klassering och separation av fina partiklar. Arbetena r riktade mot industrimineralindustrin. Till grund fr planeringen ligger en inventeringsstudie som genomfrts 2007, dr intressanta tekniker identifierats. Ansvarig fr arbetet r prof. Eric Forssberg i samarbete forskare, leverantrer och producenter. Nya tekniker fr finmalning Sicomanttekniken Syfte oh ml: att underska hur Sicomant-kvarnen kan anvndas fr malning av mineral och bergarter med olika egenskaper. Av speciellt intresse r att underska sambandet mellan energifrbrukning och partikelstorlek eller specifik yta. Av stor betydelse r att faststlla kapaciteten fr kvarnen vid malning till olika partikelstorlekar. I samband hrmed r det ocks av intresse att studera partiklarnas egenskaper ssom form och specifik yta. Det freslagna projektet r av inledande karaktr och fortsttningsprojekt frutstts kunna genomfras i samarbete mellan Sicomant och fretag. Frstudie - Nya tekniker fr klassering och separation av fina partiklar Syfte och ml: Mlsttningen r att belysa ett antal teknikers tillmpning och anvndning fr beredning av industrimineral. Hypotesen r att frsk med ett begrnsat antal modellmaterial skall ge generiska resultat som mer eller mindre direkt kan tillmpas fr andra material vilkas egenskaper inte allt fr mycket skiljer sig frn modellmaterialen. Projektet genomfrs i etapper med beslut om budget, tidplan och genomfrande infr varje etapp. Tekniker av intresse: Virvelkvarnen Tornado samt olika vindsiktningstekniker frn t.ex. Comex A/S och Alpine AG. 2.1.4 Krossat material i Framtida betong Projektomrdet har hg prioritet inom bergmaterialindustrin med speciell inriktning mot processutveckling fr framtagande av helkrossad ballast till betong som ersttning fr naturgrus. Genom att koppla betongtillverkarnas kvalitetskrav till produktionen skall metoder utvrderas som ger bergmaterialproducenterna frutsttningar att kunna bedma olika bergarters och fyndigheters frutsttningar fr produktion av helkrossad finballast enligt kundernas/betongtillverkarnas krav. Arbetet genomfrs vid Chalmers, Tillmpad mekanik. Bgge projekten samarbetar med andra projekt inom MinBaS II och STEM fr att erhlla synergieffekter och bsta mjliga resultat. Medverkande fretag r producenter av ballastmaterial till betong och maskinleverantrer. 133

Utveckling av lmpliga krosstekniker fr fina ballastfraktioner till betong Syfte och ml: Att ka frstelsen om kornformsgenerering i fina fraktioner i mikrometeromrdet under 500-250 m. Frgestllningen r hur olika bergarter skall krossas fr att skapa erforderlig kornform i dessa sm storleksfraktioner. Projektets ml r att ge ballastproducenterna kunskap att vlja krossutrustning beroende p bergartstyp, mineralsammansttning, nskad kornkurva, kornformskrav, energitgng, eventuell efterfljande separering. Metoder fr produktions- och kvalitetskontroll ingr i uppgiften. Torr klassering av fina bergmaterialfraktioner (<4 mm) fr betong Syfte och ml: Projektets syfte r att ka kunskapen om vindsiktars funktion samt hur olika mineraler och storleksfraktioner kostnadseffektivt, ur ballastproduktionsperspektiv, kan separeras med utrustning som utnyttjar torra luftstrmmar. Projektets vergripande ml r att svara p frgan om vilken utrustning som br vljas beroende p bergartstyp, mineralsammansttning, kornformskrav, rensiktningskrav, energitgng, kornkurva, antal fraktioner/sorteringar, eventuell efterfljande proportionering. 2.1.5 Modellering, optimering av produktionsprocessen Tv doktorandprojekt pgr vid Chalmers, Tillmpad mekanik. Sedan MinBaS I programmets avslut har de drivits vidare av MinFo och finansieras av MinFo, SBUF samt deltagande fretag. Dessa projekt ingr i MinBaS II-programmet men finansieras f.n. helt av industrin. Realtidsoptimering av krossanlggningar Syftet och ml: Att utveckla teknik och metoder som medger terkoppling och styrning i syfte att minimera produktionskostnader, maximera processutbytet eller minimera energitgng. I arbetet ingr att utreda vilka mjliga givare och sensorer som kan appliceras i en krossprocess, att utreda olika processparametrars regleringspotential och att utveckla hrd- och programvara som stder on-line monitoring och som omfattar reglering av processen i realtid. Genomfrt arbete: I projektet har redan utvecklats en sensor som utgr ett billigt alternativ till bandvg, vilken baseras p effektmtning av starkt lutande transportband. Modellerutvecklingen fr styrning av processen via krosspalt pgr och hrd- och mjukvaror har tagits fram som medger en styrning i realtid via Internet. Fortsatt arbete r inriktat mot att fortsatta studier av nya typer av sensorer, ssom optiska dito. Modeller fr styrning av krossprocessen via en reglering av krossens varvtal r nu under utveckling. Utvrderingstester i industrianlggningar av de utvecklade styrmodellerna skall genomfras och grnssnitt och anvndarvnligheten hos den Internetbaserade styrsystemet skall frbttras. Doktoranden E Hulthn avlgger licentiatexamen i februari 2008. Arbetet planeras pg ytterligare 2,5 r fram till en doktorsexamen. Arbetet utfrs i nra samarbete med bergmaterialfretag och all frsksutrustning r installerad i befintliga anlggningar. Optimal fragmentering genom krossning Syfte och ml: Att ta fram ett optimalt tillvgagngsstt fr krossning av olika bergmaterial, beroende p deras fragmenteringsegenskaper och krossproduktens 134

kvalitetskrav. I mlsttning ingr ven att kunna identifiera srskilt goda kombinationer av krossmaskiner och urskiljningsmaskiner. Det r av stort intresse att studera en optimal sekvens av krosszoner helt befriad frn dagens konkrossars fysiska randvillkor. Det innebr att man optimerar antal krossteg, kompressionsgrad i varje steg, och eventuell mellanliggande urskiljning (siktning) med avseende p materialutnyttjande, kraftminimering eller energifrbrukning. Med utgngspunkt frn optimeringarna kan man sedan analysera befintliga krossmaskiner och ge frslag till nyutveckling. Genomfrt arbete: Inledande studier har indikerat att dagens moderna krossmaskiner utformade fr maximal reduktion inte alltid r de optimala fr applikationer som krossning av gnejs (ballast) eller kalksten (styckekalk fr brnning). Nu pgr utveckling av i frsta hand dagens befintliga snderdelningsmodeller. Selektionsfunktionen skall studeras djupare med ett strre antal partiklar (individer) fr att ka den statistiska signifikansen i modelleringen. Fortsatt teststarbete och optimeringar skall gras fr att klarlgga t.ex. inverkan av materialens petrografiska sammansttning samt fler optimeringar med andra intressanta fitnessfunktioner. Studier av mjligheterna att simulera siktning och materialextrahering under pgende krossekvens skall genomfras. P detta stt skulle inte bara energi kunna sparas, utan ven skulle en bttre produkt kunna genereras. Projektet r avsett att i frsta hand leda fram till en licentiatexamen under r 2009 fr doktoranden Elisabeth Lee. 2.1.6 Transporter och logistik En frstudie har startats kring Energitgng & kostnad fr interna transporter i tkter. Arbetet genomfrs av Sandvik M & C i samarbete med forskare vid Chalmers, I dagslget anvnds olika fordon och bandtransportrer fr att frflytta material mellan olika maskiner och upplag. Tidigare underskningar har visat att en betydande del av energifrbrukningen i en anlggning utgrs av materialfrflyttning med hjlp av olika typer av fordon. Problemet r att det r tmligen svrt att dels skifta mellan teknikerna eftersom det krver ganska omfattande frndringar i anlggningens infrastruktur. Det r ocks svrt att avgra vad frndringen fr fr konsekvenser. Fr att underltta denna typ av studier behvs modeller som kan prediktera prestandan hos olika typer av transportsystem. Syftet med frstudien r att utreda vilka olika verktyg och modeller som idag finns tillgngliga fr att gra dessa typer av studier. 2.2 Omrde 2 Produktutveckling Prioriterade huvudprojektomrden Nya anvndningsomrden fr MinBaS branschernas material Utveckling av industrimineralbaserade produkter Europastandardisering Exempel p pgende projekt: 2.2.1 Nya anvndningsomrden fr MinBaS branschernas material Smskaliga lsningar fr avloppsrening med mineralbaserade filtermaterial Utfrare: KTH mark och vattenteknik i samarbete med mineralproducenter och teknikleverantrer. 135

Syfte och ml Att systematiskt belysa olika filtermaterials avskiljningsfrmga samt vxttillgnglighet avseende fosfor. Fltfrsk skall genomfras i ett mobilt laboratorium i anslutning till lmplig avloppsklla. Olika system med relevanta filtermaterial skall studeras fr att jmfra olika systems avskiljning av fosfor, kvve och BOD under en tvrsperiod. Studier av vxttillgnglighet och nyttjandeaspekter av terfrt filtermaterial nr det blandas med jorden ingr i konceptet. Drnerande markstensbelggningar fr frbttrad milj Utfrare Cementa AB i samverkan med intressenter inom kommuner och bergmaterialfretag. Syfte och ml Att i samarbete med kommunerna, tillverkare av prefabricerade betongbelggningar och bindemedelsleverantrer utreda miljpotential och konstruktionsmetodik fr drnerande markstensbelggningar. En frutsttning fr verbyggnader med drnerande markstensbelggningar r utveckling av ballastmaterial fr ppna och drnerande verbyggnadslager med bibehllen brighet. Syftet r vidare att etablera demonstrationsprojekt i kommunal regi dr miljpotential, dimensioneringsmetodik, ballastkvaliter och utfrande etableras och fljs upp. Det r viktigt att betongens bestndighet kan garanteras vid en vergng till krossad ballast fr att kunna ka anvndningen av krossberg i betongindustrin. Medel har reserverats fr tre delprojekt som rr Bestndighet i betong baserad p krossad ballast. Det ena rr krympning och genomfrs vid LTU, det andra rr frostbestndighet och leds av Cementa AB och det tredje rr studier av alkalikiselreaktioner och genomfrs vid CBI. b Ett pgende projekt, deltagande i Vrmeforsks Askprogram, vilket finansieras av MinFo, kommer att avrapporteras inom projektomrdet. Syftet r allmn bevakning och att ka medlemmarnas mjligheter att kunna samverka i projektarbeten dr restmaterial anvnds i nya applikationsomrden. 2.1.2. Utveckling av industrimineralbaserade produkter Produktutvecklingsarbetet fr exploatering av vandinfyndigheten Sumssjn finansieras frn detta programomrde. Planering av ett projekt benmnt Partikelteknologi pgr. Syftet r att analysera utvalda restmaterials sammansttning och vrdera de skillnader i partiklarnas kemiska och fysikaliska egenskaper som kan anvndas fr separationsteknik fr att kunna upparbeta materialen till sljbara produkter. 2.2.3 Europastandardisering Fr de sm- och medelstora fretagen inom bergmaterialindustrin r tv startade projekt kring uppfljning och implementering av nya EU-produktstandarder av stor vikt fr den framtida marknadsutvecklingen. Projektarbetena inriktas mot Implementering efter 5rs versyn av produktstandarder och Farliga mnen Bevakning och Implementering.

136

2.3 Omrde 3 Milj - Hllbar utveckling Prioriterade huvudprojektomrden Miljpverkan frn anlggningar Efterbehandling Sustainable Development Indicators Omrde 3 r nytt och fanns inte med i det tidigare MinBaS-programmet. Drfr kommer ett omfattande planeringsarbete att lggas ned innan projekten sjstts i full omfattning. Omrdet samlar alla MinBaS-branscher. Fr att n resultat i utvecklingsarbetet krvs att samverkan kan sker med myndigheter p svl nationell som regional niv. Under hsten 2007 har startats ett frprojekt, Miljpverkan frn anlggningar. Syftet r vergripande att projektet skall bidra till att konkreta och objektiva, vetenskapligt underbyggda kriterier tas fram runt utvalda miljomrden som buller, vibrationerluftsttvg, bortledande av vatten och diffus damning, som kan anvndas av fretag och myndigheter vid tillstndsbedmningar. Framtagande av en s.k. "Best Practice Guide planeras inom delomrdet Arbetsmilj, srskilt riktad mot tgrder mot respirabel silika. Inom projektet Efterbehandling r syftet r att sammanstlla en Best Practice Guide med lmpliga tekniker som skall kunna anvndas av fretag och myndigheter. Projektfrslag bearbetas som rr karakterisering av MinBaS-branschernas avfall med hnsyn tagen till implemen-teringen av det nya gruvavfallsdirektivet 2008. Projekt planeras rrande en utveckling och infrande av Sustainable Development Indicators som r anpassade fr MinBaS-branschernas fretag. 2.4 Omrde 4 Applikationsutveckling -Stenindustrin Prioriterade huvudprojektomrden Teknikverfring till freskrivande led. Applikationsteknik: Stenhandboken Naturstens tekniska egenskapers betydelse fr olika applikationer. Naturstensmurar i det offentliga rummet Internationell harmonisering av standarder och krav STENAKADEMIEN Under ret har arbetet pbrjats med samordning och detaljplanering av de prioriterade delprojekten, 18 st. tta delprojekt har pbrjat sitt arbete. Ett tjugotal fretag r involverade tillsammans med forskargrupper vid Sveriges Lantbruksuniversitet, Alnarp, SP och Chalmers samt Lunds tekniska hgskola. Syftet r att ka anvndningen av sten i Sverige. Identifierade projekt har hgsta prioritet och r anpassade till smfretagen i denna bransch. Stenindustrin har systematiskt under flera r prioriterat och drivit utvecklingsarbetet mot att framstlla bsta tnkbara tekniska informationsverktyg och utbildningsmaterial rrande applikation av natursten. Detta fr att ka kunskapen i bestllar- och anvndarledet och 137

internt i branschen, vilket r en viktig del av marknadsutvecklingen. Samma krav p underlagsmaterial stlls frn utbildningsanstalter fr blivande arkitekter och byggare. Inom projektet skall sdana verktyg och material utvecklas. Speciella insatser skall gras fr att vidareutveckla och samordna verksamheterna vid befintliga utbildningscenter. 2.5 Omrde 5 Programadministration, Planering, teknikspridning, implementering Arbetet inom syftar till att a) planera och leda verksamheten i hela MinBaS programmet, b) sprida information om kunskaper och resultat, c) arbeta fr att erhllna resultat implementeras i branscherna och d) stmma av och rapportera programarbetet, samt lta genomfra en utvrdering. Fr programomrdena har utsetts Ansvariga kanslier som skter styrgruppernas arbete, avtal med forskningsutfrare samt den lpande ekonomin inom programomrdena. Ansvarigt kansli fr omrde 1, 2, 3 och 5 r MinFo, fr omrde 4 r SFI AB ansvariga. Fr de fyra tekniska programomrdena har tillsatts Styrgrupper med en ordfrande som representerar industrin. Styrgrupperna planerar och fljer upp projektarbetet och ansvarar fr att inom ramen fr av styrelsen tilldelad budget fr programomrdet besluta om medelstilldelning till enskilda projekt. Styrgruppernas ledamter r personer frn deltagande fretag och adjungerade experter frn SGU och forskningsorganisationer. 3. MinFos Projektpaket 2007-2010 med Energimyndigheten Under 2007 startade ett 4 riga projektpaket inom MinFo. Budget r totalt 37,8 MSEK. Energimyndigheten stder paketet med 13,1 Mkr. Industrin medfinansierar projekten med kontanta bidrag och egna arbeten. MinFo funderar som vergripande projektkoordinatr och medelsmottagare. Energifrbrukning vid produktion av bergarter och industrimineral samt vid utveckling av mineralbaserade produkter uppgr sammantaget till 5 TWh per r och utgr nra 3,3 % av industrins totala frbrukning. Cementframstllningen frbrukar ca 3.5 TWh och kalkindustrin 1,3 TWh rligen. Projektpaketet bestr av tre av industrin prioriterade teknikprojekt och ett projekt fr uppfljning, teknikspridning och administration som drivs av MinFo. Fr de tre tekniska projekten har utsetts ansvariga forskningsutfrare med vilka MinFo slutit avtal fr projektens genomfrande. Projekten r i total kostnadsomfattning lika stora eller strre n de enskilda programomrdena inom MinBaS-programmet. Utvecklingsinsatserna riktas mot infrandet av ny frbrnningsteknik och effektivare CO2-avskiljning i cement- och kalkindustrin, energieffektivare framstllning av betong baserad p krossballast och dito framstllning av brnda karbonatprodukter. Projektpaketet berknas reducera energibehovet i industrimineral- och byggmaterialsektorn med sammanlagt 1,2 TWh/r. Besparingen skall uppns genom ett reducerat energibehov i produktionsprocesserna och fr interna transporter samt kade utbyten av sljbara produkter.

138

3.1 Energibesparing vid framstllning av betong baserad p krossat bergmaterial Budget: Total budget r 10,8 Mkr varav Energimyndigheten bidrar med 4,0 mkr och industriintressenterna resten. Utfrare: Arbetet drivs vid CBI Betonginstitutet AB. Industriintressenterna r Cementa AB, CBI-s A-finansirer samt betongfretag som Betongindustri, A-betong, Strngbetong, bergmaterialleverantrer som NCC Roads, Jehander, Swerock samt industrimineralproducenter som Nordkalk AB. Samplanering sker med produktionsprojekten inom MinBaS II programmets projekt 1.4 Krossad ballast i betong. Syfte och ml: Mlet r att helt motverka den kning av cementbehovet som sker med dagens betong-framstllningsteknik nr krossgrus anvnds i betong p grund av ogynnsam kornform och kornkurva hos krossgruset i jmfrelse med naturgrus. kningen r ca 5% och ytterligare 3% kning sker p grund av minskad tryckhllfasthet. Detta motsvarar en rlig energifrbrukning vid cementproduktionen p 200 GWh. Syftet r ocks att vidareutveckla tekniken s att en generell minskning med 5 % cement sker trots vergng till krossad ballast, vilket ger en ytterligare reduktion p 125 GWh. vergngen frn sm lokala naturgrustkter till stora krossanlggningar kommer ofta att medfra strre transportavstnd. 2 mils kning av transportlngden fr endast grusfraktionen motsvarar ca 15 GWh. I projektet skapas underlag fr att motverka denna energikning. Reduceringspotentialen fr CO2 r ca 230 kton per r. Genomfrande: Baserat p omfattande karakterisering av olika krossgrus skall riktlinjer utarbetas kring lmpligheten hos krossgruset (stenmjlet) som fingrus i betong Kravspecifikationer frn betongtillverkarna fr ballast till betong skall utvecklas. Ett proportioneringsprogram som tar hnsyn till krossgrusets kornfrdelning och kornform skall utvecklas. Case-studies i betongindustrin skall genomfras dr konsekvenser av en vergng frn naturgrus till krossgrus studeras och utvecklade verktyg testas. Studierna omfattar ven effekterna av transportkostnader fr olika tktalternativ. Fr att implementera kunskapen genomfrs utbildning av minst 200 personer inom bergmaterial- & betongtillverkningssektorn om produktion och anvndning av bergmaterial i betong p ett energieffektivt stt. Vid projekts slut finns i genomfrda Case-studies ett bra underlag fr att bedma mjligheterna att n besparingsmlet 3.2 Energieffektiv framstllning av brnda karbonatprodukter Budget: Total budget r 9,3 Mkr varav Energimyndigheten bidrar med 3,05 Mkr och industriintressenterna resten. Utfrare: Arbetet drivs vid Lunds universitet, Geologiska institutionen. Industriintressenterna r SMA Mineral samt andra kalk- och dolomitproducenter som Nordkalk AB och Bjrka Mineral. Syfte och ml: Att reducera energibehovet vid produktion av brnda karbonatprodukter med totalt 251 GWh/r genom att reducera snderfallet att produkterna under brnnprocessen. Genom att utveckla underskningsmetoder och verktyg fr prospektering, produktion och styrning av brnnprocessen frvntas snderfallsfrlusterna i brnnprocessen kunna minskas med 20%. Till detta kan ocks lggas potentialen fr minskat energibehov och kostnader fr brikettering, transporter samt deponering. Direkt potential fr CO2 reduktion r 169 kton/r.

139

Genomfrande: Frsta delen av projektet syftar till att ge basen fr metodutvecklingen genom att klarlgga orsakerna till snderfallet. Grundlggande studier genomfrs fr att relatera mineralogiska, strukturella och kemiska egenskaper hos rvarorna till snderfallet. I denna del av projektet finns en beslutspunkt om inriktningen p fortsatt utvecklingsarbete. Utvald metodik utvrderas genom en pilotstudie. Drmed kan en bedmning gras om mjligheterna att uppn mlsttningen med att utveckla enkla metoder och verktyg. Drefter skall en enkel fltmetod utvecklas som kan anvndas ocks inom prospektering fr att skilja bra stenar frn dliga. Vidare skall modeller fr termiskt beteende utarbetas som kan vara en grund fr verktyg fr planering av en verksamhet. Som avslutning skall basen lggas fr att implementera utvecklade metoder och verktyg i produktionsapparaten. Metoder och modeller skall anpassas fr industri- och branschspecifika verksamheter och underlag skall utarbetas till metodval och praxis. Vid projektets slut finns underlag i form av minst en case study i en produktionsanlggning som kan anvndas fr bedmning av potentialen samt modeller och verktygs anvndbarhet. 3.3 Energieffektiv oxy-frbrnningsteknik och CO2 avskiljning i cement- och kalkindustrin Budget: Total budget r 12,7 Mkr varav Energimyndigheten bidrar med 4,5 Mkr och industriintressenterna resten. Utfrare: Arbetet drivs vid Ume universitet, Tillmpad fysik och elektronik som tv industridoktorandprojekt. Industriintressenterna r Cementa AB, Cementa Research AB och Nordkalk Oy Abp. Syfte och ml: Mlet r att reducera energifrbrukningen vid tillverkning av cementoch kalk med totalt 562 GWh/r, frmst genom att reducera det frvntade kande energibehovet fr efterfljande processteg, dr CO2 separeras ur rkgaserna. Introduktion av en ny frbrnningsteknik, oxy-fuel frbrnning frvntas medge en 50 % ig kad anvndning av CO2 neutrala brnslen i jmfrelse med dagens frbrnningsteknik. Detta minskar den fossila CO2 emissionen med 38 kton/r i cementproduktionen. Applikationen inom cement- och kalkindustrin, skiljer sig frn andra applikationer dr oxy-tekniken introducerats, genom att slutprodukternas kvalitet starkt pverkas av brnslekvalit och CO2-balansen. Ett betydligt energieffektivare system, jmfrt med dagens standardteknik, skall utvecklas fr att avskilja CO2 ur rkgaserna. Detta berknas reducera CO2 utslppen i rkgaserna med 20%, vilket ger en rlig reduktion av cement- och kalkindustrins CO2 med 550 kton. En indirekt energieffektivisering stadkoms i cementprocessen genom en 10 %ig minskning av elfrbrukningen som en fljd av minskade rkgasvolymer. Genomfrande: Projektet r indelat i tv doktorandprojekt med olika delml. Delprojekt 1 CO2-avskiljningsmetoder ur rkgaser i cement- och kalkproduktion: Delprojektet har som ml att utveckla energieffektivare koncept fr CO2-separation i cement- och kalkproduktionen. Delprojekt 2 Modellering av kemin i cement- och kalkprocessen : Delprojektet har som ml att utvrdera oxyfrbrnningstekniken och de CO2- neutrala brnslenas inverkan p klinker och kalkprodukternas kvalitet.

140

En presentation kommer att gras av detta delprojekt vid rets konferens av doktoranden Bodil Wilhelmsson-Hkfors. Vid projektet slut kommer det att finnas utvecklat ett modellverktyg som ger ett bra underlag fr att bedma mjligheterna att n besparingsmlet i form av minskad CO2- emission och reducerad energifrbrukning samt strategier och metoder utvecklade som medger mjligheter att bedma energibesparingpotentialen i nya CO2-avskiljningsmetoder. 3. 4. Uppfljning, teknikspridning och administration inom MinFo projektpaket 2007-2010 Syftet med projektet r att fr industrins rkning ta fram relevanta syntesrapporter, flja upp arbetet, administrera och sprida resultaten frn MinFos projektpaket. Speciell vikt skall lggas vid slutrapporteringar och mluppfljningar fr implementering av resultaten. Inom MinFo r det styrgruppen Processteknik som har den vergripande ledningen av projektpaketet och formellt utser ansvariga projektledare och arbetsgrupper till std fr projekten. Referenser:
Programplan inkl programbudget MinBaS II programmet 2007-2010, Bilaga till anskan till SGU, juni 2007 Prel. rsrapport 2007 MinBaS programmet, 2007-12-01 Programbeskrivning MinFos programpaket 2007-2010, Energimyndigheten, 2007-03-29, www.energimynidhgeten/industriforskning/FoU vid MinFo och rsrapporter 2007 till Energimyndigheten Bergverkstatistik 2006 samt Grus, sand och krossberg 2006, SGU:s periodiska publikationer 2007.

141

Conference in Minerals Engineering 5 6 February 2008

LIST OF PARTICIPANTS
Adolfsson Gran Aitahmedali Hassna Alatalo Johanna Alldn berg Eva Andersson Mats Andrasson Benny Bengtsson Joakim Berggren Andreas Berhan Siv Bida Jan Bjurling Kenneth Bucht Johan Degerstedt Erik Dwari Ranjan Ekenstedt Fredrik Famurewa Stephen Flanagan Ian Forssberg Eric Furberg Ronnie Gustafsson Jan-Olof Hagemalm Malin Hallberg Dan Henriksson AnnSophie Hokka Harri Holm Therese Holmqvist Tore Hgglund Anders Hmn Ilkka Johansson Bjrn Jonsson Olof Kappel Ravi Kivenne Timo Kleiv Rolf-Arne Klemo Stina Kolacz Jacek Kongas Matti Kota Hanumantha Rao Kruger Paul Larsen Rune Lejon-Isaksson Linus Lindqvist Mats Lindqvist Per-Arne Lindvall Henrik Lund Cecilia Majamki Reijo Mohammad Seyed Niiranen Kari Odn Pr Oghazi Pejman Olsson Maria Poromaa Ramona Plsson Bertil Riyahi Kamrooz Ruong Urban LKAB Lule tekniska universitet Lule tekniska universitet LKAB Lule tekniska universitet LKAB Lule tekniska universitet Boliden Mineral AB Lule tekniska universitet Bergsskolan Ciba LKAB Lule tekniska universitet Lule tekniska universitet Boliden Mineral AB Lule tekniska universitet CIBA Lule tekniska universitet Boliden Mineral AB Akzo Nobel Surface Chemistry AB Lule tekniska universitet LKAB LKAB Outotec Minerals Lule tekniska universitet Ciba LKAB LKAB Boliden Svenska Sililkaverken AB Moventas AB TK Slurry Consulting Ky NTNU LKAB Comex AS Outotec Minerals Lule tekniska universitet Boliden Mineral AB NTNU Lule tekniska universitet Sandvik SMC Lule tekniska universitet Boliden Mineral AB Lule tekniska universitet Teknikum Lule tekniska universitet LKAB Nordkalk AB Lule tekniska universitet LKAB LKAB Lule tekniska universitet Lule tekniska universitet LKAB KIRUNA LULE LULE MALMBERGET LULE MALMBERGET LULE BOLIDEN LULE FILIPSTAD VSTRA FRLUNDA MALMBERGET LULE LULE BOLIDEN LULE BRADFORD, England LULE BOLIDEN STENUNGSUND LULE MALMBERGET MALMBERGET ESPOO, Finland LULE VSTRA FRLUNDA MALMBERGET MALMBERGET BOLIDEN KPING LULE LAHTI, Finland TRONDHEIM, Norway KIRUNA RUD, Norway ESPOO, Finland LULE BOLIDEN TRONDHEIM LULE SVEDALA LULE GARPENBERG LULE VAMMALA, Finland LULE KIRUNA KPING LULE KIRUNA KIRUNA LULE LULE KIRUNA

Conference in Minerals Engineering 5 6 February 2008

LIST OF PARTICIPANTS
Saarinen Markku Samskog Per-Olof Sellgren Anders Stocks Paul Strm Elena Sundvall ke Sderlund Jens Taavoniku Krister Tano Kent Thomaeus Marianne Thornhill Maria Wanhainen Christina Weihed Pr Wennerhorn Claes Wettainen Hans Olof Wilhelmsson Hkfors Bodil Vilinska Annamaria kerhammar Peter strm Lars nal Gven Weir Minerals Finland Oy LKAB Lule tekniska universitet CIBA LKAB LKAB Boliden Mineral AB LKAB LKAB MinFo NTNU Lule tekniska universitet Lule tekniska universitet Svenska Sililkaverken AB LKAB Ume Universitet Lule tekniska universitet SGU Metso Minerals Sala Istanbul Technical University LAHTI, Finland KIRUNA MRSTA BRADFORD, England MALMBERGET MALMBERGET BOLIDEN MALMBERGET MALMBERGET STOCKHOLM TRONDHEIM, Norway LULE LULE KPING KIRUNA UME LULE UPPSALA SALA ISTANBUL, Turkey

Universitetstryckeriet, Lule 2008