- To participate in the 911Metallurgist Forums, be sure to JOIN & LOGIN
- Use Add New Topic to ask a New Question/Discussion about Crushing.
- OR Select a Topic that Interests you.
- Use Add Reply = to Reply/Participate in a Topic/Discussion (most frequent).
Using Add Reply allows you to Attach Images or PDF files and provide a more complete input. - Use Add Comment = to comment on someone else’s Reply in an already active Topic/Discussion.
Crusher Discharge T10 PSD Particle Size Distribution (10 replies)
What do you mean by "internal" PSD of a rock stream?
The T10 relates to the PSD from breaking a particular size at a specific energy, but the same definition could in theory be used to describe the in-situ rock mass, if the same could be seen as a distribution of sizes (blocks) starting with an initial top size.
Well my interpretation was that you are looking at preferential breakage.
JKMRC always considered various modelling frameworks for this.
However the specific question 'how would such a model relate to T10' is a bit confusing as T10 was not specifically designed for preferential breakage. I remember discussing preferential breakage with one of the participants asking 'how does this relate to T10 of a Crusher Discharge?
I had hoped others might have commented in the discussion, but there are lots of issues regarding the way comminution is modelled.
The difficulty is that it is currently a compromise between empirical and semi-phenomenological approaches; and it remains my view that comminution modelling could be better (and certainly more consistently) modelled if a bit more 'applied maths' was dedicated to the phenomenological component.
One of the difficulties with empiricism is that once it is successful, researchers tend to stop rather than look for the underlying hidden process that is being manifested.
Now here I use the word 'hidden' which I have mentioned in other mineral processing groups specifically Hidden Markov Models.
I agree that such a description would be useful, especially to primary crusher manufacturers. In this instance the rock stream is the blasted rock/primary crusher feed. One of the major issues with this stream is predicting its size distribution, together with the subsequent distribution of the primary crusher discharge. This is a prediction at which all the crusher manufacturers are very bad at. Ever since I started at JK Tech in the mid 90s through to my time today at AMEC I have continually had arguments with vendors about primary crusher product size distribution. JKMRC, and in more recent times, Steve Morrell have attempted to address this argument through mine-to-mill measurements and through predictive equations based on either ta or DWI measurements.
Perhaps a better prediction method would be to estimate some fines generation factor (Using, for example, the P10 for from crushing core coupled with looking and the natural distribution of fines in the core). This factor could then be applied to the standard size distributions generated by crushers according to the vendors (which are pretty good for scalped feeds) to arrive at an overall primary crusher product size distribution that everyone can sign off on!
I appreciate your comments. I'm using Metsim. The crusher appearance function is due to the USBM, the screen sizing model is due to Karra. It compares almost reasonably with comparable outputs from Bruno, Aggflow and JKSimmet. I've had a good read of the blue book (Mineral Comminution Circuits their Operation and Optimisation). It's the type of book that needs about five re-reads; great job too! Figure 4.12 relates t10 to Specific Comminution Energy and if 'A' were 50 or thereabouts then 'b' or A*b/50 could be used as a measure of what I'm calling 'friability'. I can see A*b has a tenuous relationship to Bond Crusher Work Index in Figure 4.13. Also t10 is related to reduction ratio for the same ore Fig 6.6. All good.
I'm trying simulating a two stage iron ore crushing circuit to produce a maximum of lump and all I know is that the top size of the shot rock is 600 mm and the Crushing Work Index is 8-9 kW-h/tonne. Let's say a very similar ore is giving a lump to fines split of 40:60. (For a more competent ore the split for a similar set-up i.e. reduction ratios and screen sizes might be 70:30). So now I need to parameterise the crushers. I could do this by tweaking the resulting t10's in a consistent manner until the 40:60 is achieved. The ROM feed PSD needs to be adjusted also. My original question could have been better put. For the feed one would expect a relationship between the top size or P80 and the "internal" t10. I am using the word "internal" to refer to the PSD prior-to and not as-a-result-of a crushing process. There probably is a better word. The question is how would one "re-constitute" the feed PSD (by splining a suitable knot) to be consistent with the t10's which result in 40:60.
One of the main results of getting a handle on this would be more confident screen sizing.
There are a few sub issues here.
1. I don't know that much about MetSim, but can MetSim be called as an application from another application. Then it would be more straightforward to use a Solver approach (I have developed independent Solvers)
2. It would seem you are double tweaking; i.e. both the crusher model and the original PSD. This is OK if you have at least some info. About both, but if not I think the problem is not tractable.
3. So it would be good to identify what you know, what you don't know, and what you have an informed guess about.
4. Now suppose the problem is not tractable. Then maybe adjust the PSD as one solution and adjust the t10 as the other solution. The do a simulation with both variations and this can give a range in simulation.
5. In other words don't focus on an exact solution but a range of solutions.
I am sure JKTech will have their own suggestions.
Maybe it would reduce to one problem if someone knows, or could make a reasonable stab at, the feed knot (or PSD) for an iron ore that ends up 40:60. Once through the primary size I'd say the process takes over but I don't know for sure. I'm trying to see if there's some rationale to screen sizing for DSO iron ore or whether its rules of thumb e.g. m²/tonne/h. The Karra method takes into account near size and efficiency of misplaced fines but maybe there's not enough publicly available ore characterisation (feed PSD and t10 vs reduction ratio) for these refinements to be useful.
I've had a good read of the blue book (Mineral Comminution Circuits their Operation and Optimisation). It's the type of book that needs about five re-reads; great job too!
I found the blue book not as consistent as it should be. Just a prompt time for a new book or indeed updated series on mineral processing.
The book is fine as far as the definition of t10 and its use in the JKSimMet models. The issue at hand extends beyond t10, to the fragmentation prediction from blasting rock. The latter needs to be first characterised in terms of insitu hardness and structure, and then we can apply one of several available models to estimate the fragmentation distribution expected from a given drill & blast design. This approach provides a more reliable starting point than any manufacturer or 'rule-of-thumb' for the simulation of crushing and screening, as reported in many Mine-to-Mill studies. The Marandoo iron ore is a good case study in question, which was published in 1998 at the Aus IMM MTM conference.
Blasted ore does not have 'natural' size distribution i.e. the muckpile will contain fragments which resemble the original defect spacing or foliation frequency of the intact rock, plus the pulverized rock distribution which form in the zone immediately adjacent to the blast hole consequently product sizes from open-circuit primary crushing will not display natural size distribution the rosin-rammler form of equation can be fitted to blasted ore and crushed ore but different fits must be applied to the coarse fractions and the fine fractions the inflection point for these fittings will be between 5-50 mm for pilbarahematitesrosin-rammler is a particular form of the more general Weibull distribution pilbarahematites display weaker strength at coarser sizes
in other word individual particles of hematite will show lower average crushing work index as the size of particle increase kw/t
many ore become stronger at coarser sizes
now problem will be to calibrate the fitted size equation parameters for blast conditions
George Boucher has matched blast conditions to many ores, and so obtain sensible explosive/blast design outcomes via the KUZ-RAM equations.
The T10 is useful as a single number which describes the Particle Size Distribution for the product from Crusher Discharge's operation relative to (size by size) feed size i.e. % passing 1/10th of parent size (s). Can T10, or a modified T10-type number, be used to describe qualitatively the "internal" PSD of a rock stream (in isolation from a crushing operation)? Say % passing 1/10th of top size or d80 size (or another fraction perhaps). This could be useful for instance to describe the friability of one rock type relative to another. Is there a common usage in this regard?