Thursday, 16 January 2014

Is anyone researching liberation enhancement?

With energy costs rising rapidly, the need to reduce energy consumption is one of the big drivers behind current comminution research. Comminution '14 will show how this might be achieved by innovations in grinding technology and circuits, and pre-concentration.
In deciding whether there is any hope for significant improvement in comminution energy efficiency, a keynote lecture from Prof. Tim Napier-Munn, former Director of the JKMRC, will consider the key technical and cultural impediments to progress, and speculate on how the innovation process may yet provide the long-sought paradigm change. The keynote will show that, despite intensive research, and early claims of high energy efficiencies, HPGRs have still only succeeded in niche applications in mineral processing after 30 years of trying.  AG/SAG mills have dominated grinding because of their capital efficiency and operability rather than their energy efficiency.  Stirred mills have been a genuine innovation in fine grinding but even they are prodigious consumers of specific energy.  As Tim will show, we still do not know enough about the physics of the fracture of heterogeneous brittle materials such as mineral ores.  Comminution science is really a branch of materials science, but materials scientists, unlike minerals engineers, are only interested in the fracture event itself (read ‘failure event’), not the nature of the products of fracture.  The main job of comminution is mineral liberation, and recent research has taught us that liberation is a product of the texture of the ore and to only a limited extent the fracture mechanism.
So who is researching the fundamentals of breakage at the moment, with a view to enhancing mineral liberation? There was a lot of interest in liberation enhancement in the 1980s, and in 1993 I co-authored a paper, published in Minerals Engineering, which considered the need to research more deeply the mechanisms of the breakage processes, particularly the promotion of intergranular fracture. In order to do this, control of crack propagation, and the nature and role of grain boundaries, were considered to be areas deserving most attention. My notes below refer to aspects of that paper.
High throughput comminution machines liberate minerals relatively inefficiently and inadequate liberation in itself leads to higher energy consumptions, as finer grinding has to be performed in order to achieve an adequate degree of liberation. This also leads to the generation of ultra-fine slimes particles, which may be lost in the downstream process. More research is needed into the mechanisms of the breakage process in these machines, particularly into the promotion of intergranular, rather than transgranular fracture. The inhomogeneity and non-perfect elasticity of rocks make crack generation and propagation studies in them difficult and tbese difficulties apply equally to ores which contain more than one mineral species. Despite these difficulties, fracture creation in such complex ores needs clearer definition as it must be recognised a priori that the generation of intergranular fractures is the key to successful liberation.

Stress concentration at a crack tip
Brittle materials such as rock and glass break with little plastic deformation, deforming elastically to the instant of fracture. The tensile strength of such materials is much less than that predicted theoretically, which led Griffith to postulate that the low observed strengths were due to the presence of small cracks or flaws, the extremities of these cracks acting as stress raisers . Griffith assumed that the theoretical stress was obtained at the end of the crack, even though the average stress was still far below the theoretical strength. Fracture, according to this concept, occurs when the stress at the ends of the cracks exceeds the theoretical stress, and when this occurs the crack is able to expand catastrophically. The failure of a brittle solid such as rock is therefore caused by the extension of Griffith cracks which are inherent in the material. The energy required to create new crack surface is supplied from the work done by external forces, by release of stored strain energy in the solid, or a combination of these two sources. Plastic deformation in the vicinity of the crack tip and the need to overcome the surface energy of new crack surface can ensure stable growth of a crack under external forces, but above a critical crack length, the stored potential energy becomes greater than the resistance to crack extension (i.e. surface energy) and the crack will accelerate (unstable propagation), and failure will be inevitable, after which excess stored strain energy will be released as heat. As the shock waves of the crack movement have to be dissipated through the material, the maximum theoretical speed of a crack tip is the speed of sound in that material, and it has been shown that if there is sufficient localised stress, the propagation speed of a crack can approach 40 % of the speed of sound.
All rocks contain inherent cracks, derived from many sources, due to the appreciable mechanical, thermal and chemical actions to which they have been subjected over many millions of years. There is also evidence to show that when rock is removed from the earth, particularly from depth, stress relief initiates nucleation of new, and growth of existing cracks, due to the release of the stored compressive strain energy, it being well known that quartz from depth is easier to crush than that from nearer the surface. New cracks are also produced in crushing and grinding, but as the particle size decreases the proportion of cracks above the critical crack length decreases, so that fine particles are characteristically more difficult to comminute.
Crystal boundaries are regions of misfit or disorder between crystals, and it would be expected that such areas should be relatively weak in relation to the ordered crystal lattices. Despite this they can act as barriers to propagation of cracks, in the same manner as grain boundaries in metals are known to impede the motion of dislocations along slip planes. That crystal boundaries act as inhibitors to crack propagation is evidenced by the greater fracture toughness of fine-grained compared with coarse-grained rocks. In most comminution operations, however, cracks appear to be little influenced by grain boundaries, the grains often being cleft across, producing a low degree of liberation. This has led many minerals engineers to conclude that boundaries are a source of strength. However, as has been shown, existing cracks, and others developed from flaws in the matrix, can propagate at very high speed when excess strain energy is put into the lattice. This is the case with all current comminution processes. Cracks moving at such high speeds therefore tend to ignore obstacles in their path, even though these may be paths of lower resistance. This is analogous to an axe moving at high speed through wood, tending to ignore knots and flaws, thus moving in a straight line. A saw moving much more slowly tends to have its blade deflected by such areas.
A crack moving slowly in a rock (stable propagation) would be expected to similarly exploit grain boundary weaknesses as paths of least resistance, particularly as cracks propagate along grain boundaries at low velocities. Evidence for stable crack propagation exploiting grain boundary weakness is the high degree of liberation found in secondary alluvial deposits. The minerals in these deposits have been released from the primary source by millions of years of chemical and mechanical weathering, the minimal strain energies extending cracks in a stable fashion, and breaking the rock eventually around the grain boundaries in a manner similar to the stress-corrosion cracking of metals. It is likely that the strain energy is sufficient only to move existing cracks rather than produce new ones, such that the final result is not only liberated, but also crack free, minerals. Good evidence for this is the high proportion of gem (flawless) diamonds which are found in placer deposits such as those in Namibia, transported by rivers many hundreds of kilometres from the primary source. An ideal comminution process would therefore impart only sufficient strain energy to create stable propagation of existing cracks to liberate minerals at grain boundaries, but would leave the material stress free after comminution, preventing nucleation of further major cracks in the matrix.
However conventional crushing and grinding machines comminute rocks by the application of massive successive stresses. When the rock breaks, the large amount of excess strain energy in the bulk of the lattice is dissipated as heat, and the excess energy causes cracks to propagate in an unstable fashion and accelerate rapidly. There is some evidence to show that mineral liberation can be improved using HPGR. The high compressive stresses, acting on a compacted bed of material, may effectively seal micro-cracks, therefore allowing plastic strain to occur, with the associated flow of dislocations, which could pile up at grain boundaries, coalesce and promote intergranular cracking. It is interesting to note that when the pressure is released, and the rock examined, many microcracks are observed, which substantially reduce the work index of the rock. In some eases, these cracks have been observed at the grain boundaries, and this has been  exploited in some kimberlite processing plants to enhance the liberation of the diamonds without breaking the valuable larger stones.
The method simulating most closely the gentle mechanical action of alluvial formation is autogenous grinding, and there is evidence to show that liberation is improved over steel grinding. Several tests have shown that ores ground autogenously float faster and with better selectivity than if ground conventionally. The arguments often put forward to support enhanced liberation, such as abrasion of the matrix exposing the stronger mineral grains, are not, however, convincing. It seems more likely that slow crack propagation occurs, due to rapid unstable crack acceleration being inhibited by minimisation of the stored lattice strain energy. The cracks are thus able to exploit the inherent boundary weaknesses in finding paths of least resistance.
Returning to the analogy with placer deposits, what is needed is a process similar to physical weathering to weaken the grain boundaries, to facilitate the subsequent liberation. In the mid 80s work was carried out at Camborne School of Mines and the University of Birmingham on the use of prior treatment to weaken the rock, preferably at the crystal boundaries.  It was hoped that heat treatment would promote grain boundary weakening due to differing thermal expansion rates of the different phases in neighbouring grains, which effectively "loosens" the crystals in the matrix. If this is so, then this effect could be enhanced in mineral assemblies, where near neighbours may have widely differing coefficients of thermal expansion. Heat treatment could, therefore, have potential for weakening rock prior to grinding.

Tin ore before (left) and after heat treatment
A number of studies were directed at an evaluation of thermally-assisted liberation as a means to improve mineral recovery. At CSM we heated a hard-rock tin ore to 600°C, before quenching it in water. The differential expansion and contraction promoted significant intergranular cracking and a 55 % reduction in grinding resistance caused by the weakening of the rock matrix. An economic evaluation showed that only a 1% improvement in tin recovery would have made such treatment economically viable, as the increase in smelter revenue, together with the reduction in grinding energy, would offset the energy required to heat the ore. We were obviously excited with this finding, but the results of testwork to assess liberation improvements after conventional steel grinding were disappointing. It was suggested, however, that heat treatment could have potential as a prelude to the gentler action of autogenous grinding, the autogenous mill making use not only of the boundary weakening, but also of the weakened matrix; the gentler action also preventing the critical growth of the transgranular microfractures.
It is over 20 years since the paper was published, and its purpose was to stimulate debate. Maybe 20 years on it will?  I have been out of this field for so long now that I may have missed much which has taken place, but I have a feeling not, as I see little evidence of research into liberation enhancement in recent journal papers, and there is no work in this area scheduled for presentation at Comminution '14. So I have two questions:
  • Is anyone out there working on the fundamentals of mineral breakage, aimed at liberation enhancement?
  • If not, why not?

15 comments:

  1. Again you picked a nice topic; we do not want any more models(mathematics) on comminution; we want energy efficient mechanisms/machines for comminution,
    I see a day when we have to break particles to submicron range to be able to recover the last grain of value from an ore mined with so much capital invested.
    So, from my perspective, efficient fine grinding and super efficient and cost effective dewatering mechanisms are the future need.
    Tadimety Rao, India

    ReplyDelete
    Replies
    1. Thanks for getting the ball rolling on this discussion, TC. Interesting to hear what the modellers might have to say!

      Delete
    2. Thanks Barry for opening up discussion. Sorry, but 100% disagreement with Tadimety above.

      The lack of utilisation of mathematical approaches is the greatest hindrance to holistic simulation and optimisation.

      Once we can simulate/predict the profit value of improved technology, we then have a driver for making those improvements.

      Fundamentally the main issue with why liberation is not being utilised is a lack of interest in profit potential.
      Stephen Gay, Australia

      Delete
    3. I think that Stephen may be putting the horse before the cart. One needs robust formulae for liberation as a function of size reduction mechanisms, mineralogy and petrology as well as proposed duty and machine type. Didn't think that we are quite there yet - so the 'mathematics' has little to work with to generate reliable, reproducible forecasts of size reduction performances for specific ores.

      One of the primary issues with the lack of advances in step-change improvements in comminution, as discussed in the pre-conference 2010 IMPC workshop, is IP and commercial control over the innovation to recoup the investment in time and resources. I think that there is actually a high concern for profit; when technologies are readily copied by competitors there is no driver for investigating and developing step-changes in technology by commercial entities.

      No doubt someone will say protect it with patents; this is an expensive business and basically provides the world with full details of the innovation (ready for copying). Unless it requires a special manufacturing or assembly method it can be hard and expensive to protect. In addition, if engineering drawings are submitted with the patent or mechanical relationships described, for example, simple changes in the design angles or the like are considered non-infringements. This assumes of course, that the competitor and their country has a culture of respecting and enforcing patents.
      Andrew Newell, Australia

      Delete
    4. This interesting debate continues on LinkedIn

      Delete
  2. We are looking at using accelerated mineral carbonation as a means to simultaneously alter and liberate minerals.

    We are working on this together with Innovation Concepts B.V. (Gorinchem, The Netherlands). Pol Knops (@GreenOlivine) is doing his doctoral studies on this subject.

    I will give a related talk at the Hydrometallurgy Conference 2014 in Victoria, BC, on "Enhanced Nickel Extraction from Ultrabasic Silicate Ores Using Mineral Carbonation Pre-treatment".

    Rafael Santos, KU Leuven

    ReplyDelete
  3. In 1988 Barbery and Pelletier made comment regarding data accessibility - particularly the pixels data.

    Any serious liberation analysis requires the pixel data rather than simplistic interpretations.

    It has became harder and harder to access the pixel data, and it is understandable most of the momentum in this area of research has died.


    It also doesn't help hat papers currently being published ignore the earlier work, so that there is endless an endless recycling of research adding to the demoralisation.

    The industry desperately needs a service provider to enter the market who is dedicated to rejuvenating the subject area.


    ReplyDelete
    Replies
    1. Klaas P. van der Wielen, SELFRAG AG Switzerland16 January 2014 at 20:37

      The accessibility of pixel data is a good point. From my experience with QEMSCAN+iDiscover you can export nice BSE images/mineral maps, but there is an algorithm in iDiscover that bins BSE values into the nearest 5, so you go from an image with 255 brightness values to one with 51 at best. I was doing some very interesting weakening+fracture+liberation investigations using QEMSCAN but this data reduction in the end killed this work as I could not extract the data quality for further fracture modelling.

      Delete
  4. Hi Barry, Dr. Jakobs GmbH is focusing in ultrasonic processing in order to add this process where attrition scrubbing is reaching its limit and also we are looking into that topic regarding enhancement of chemical leaching. We mainly focus on silica sand/quartz, but also did learn, that we can improve results of bleaching kaolin. sending my best!
    Udo Dr. Jakobs, Germany

    ReplyDelete
  5. Klaas P. van der Wielen, SELFRAG AG Switzerland16 January 2014 at 20:34

    We're getting very good liberation results here at SELFRAG, using our high voltage breakage technology. This was very well demonstrated by Wang et al. (2012), who reported considerably better liberation after SELFRAG treatment at the same specific energy as a conventionally comminuted control sample. In addition, in my PhD data I demonstrated both better liberation, as well as preferential concentration of sulphides into the -355 micron fraction at energies less than 5 kWh/t, and I found some very good examples of completely liberated minerals inside a weakened rock matrix.

    Coming back to Barry's question, two things come to mind when considering liberation research:
    1. Cost of liberation studies; I did about 150 QEM blocks for my PhD, which would have cost an extortionate amount had it been done commercially.
    2. What happens once you have achieved better liberation? Will it actually translate into better recovery or do you still have to grind well-liberated 300 micron sulphide particles to 100 micron to float them (this is ore-dependent as well of course). This comes back to other recent min-eng blog postings about coarse particle flotation.

    ReplyDelete
    Replies
    1. I look forward to your presentation at Comminution '14, Klaas, and hope that you will highlight the liberation aspects of your work.
      Regarding your answers to my question:
      1. I sincerely hope that people will not be put off liberation enhancement work by feeling that they need to buy, or have access to, automated electron microscopes such as QEM*SEM. They don't- the traditional methodology of recovery-grade comparisons will show if liberation has been enhanced. If it has, then I am not sure why it is necessary to quantify it by using sophisticated and expensive instruments.
      2. Yes, in the vast majority of cases, enhanced liberation will lead to better recovery. Where flotation is used, most modern ores need to be ground well below the current optimum size for flotation to achieve adequate liberation (see Prof. Rao's comment above) and in the rare cases where liberation might be achieved at 300 microns or above, then this might promote more interest in the recent developments in fluidised bed flotation machines, such as the Eriez Hydrofloat cell. Liberation at coarse sizes will always benefit those ores treated by physical separation processes such as gravity concentration of course.

      Delete
    2. Klaas P. van der Wielen18 January 2014 at 13:44

      You are right, grade-recovery comparisons will mostly do the job. However, for ores with a complex mineralogy the modal mineralogy, mineral association data etc. from automated mineralogy in my opinion do really add value. It would be interesting to know if anyone has developed a method of quantifying ore texture data from automated mineralogy for use as an input in breakage modelling, for instance to look at textural effects on stress distributions at <mm-scale.

      I'm following the Hydrofloat and similar coarse recovery technologies with great interest, and I hope they will become more commonplace.

      Sorry, no mention of liberation in the upcoming presentation, just a weakening assessment. I am working on the liberation data for a later publication.

      Delete
  6. Liberating the waste before it gets to the concentrator shouldn't be overlooked. Blast Movement Monitoring System has been commercially available for a while now.

    I actually did my UG thesis on the enhanced liberation of gravity recoverable gold by using higher powder factors in sub-level caving. The material science seemed to point to some possibilities (to a wide eyed 20 year old), but the lab work just showed noise ...although that was a long time ago and was only UG research. Tony Partridge supervised the work and we had some support from Dyno.
    Thomas Rivett, Outotec, Australia

    ReplyDelete
  7. Klass (re: 16 Jan.)

    Yes and that is the problem with intepretitive software. It takes data with great potential for analysis and reduces it to a form that is of limited value.


    All we need is mineralogy images in a standard formatso that we can used Matlab, Python or R (etc.etc.) . and do professional analysis. That is let the market provide the level of service required.



    To be fair, this capability (getting the images in pixel format) is available, but it costs alot (and is prohibitive ).

    There is also a principle involved, why should getting pixel data actually cost alot? particularly when most of the rest of the world uses standard image formats.

    My view here is that mineral processing/mineralogy is at least 30 years behind where it should be. (I recently gave a seminar and claimed mineral processing was about 100 years behind where it should be, and I understand noone disagreed with me from the feedback).

    However I would expect a number of people here to disagree, although I can definitely explain why I hold my position.

    So what we really need is a Service Group/Uni. who is prepared to make the revolutionary changes so desperately needed to modernise the industry - particularly mineralogy.











    ReplyDelete
    Replies
    1. Re. Stephen.

      With regards to interpretative/standard analysis software, have you tried ImageJ? I can absolutely recommend it, it is a very powerful, freeware (!) package with lots of customisability, ability to do automated processing of images using macros etc. I had great experiences with it to produce fracture patterns from QEMSCAN BSE images. The only problem, as outlined before, was exporting a .raw/similar uncompressed format from I-Discover. The raw image is on the screen in I-Discover to do SIP development, and all the relevant data is there (i.e. exact BSE value), you just need it to come out that way. That said, this was 1.5-2 years ago and maybe FEI have amended this with newer versions of this software.

      Delete

If you have difficulty posting a comment, please email the comment, and any photos that you might like to add, to bwills@min-eng.com and I will submit on your behalf