We were recently asked by a magnetite processor to look at changing an AG mill to a SAG configuration for throughput improvement. We were limited by bearing and structural design to only have a max of 5% ball charge. Mill design should give owners best flexibility - mills also have a second life.

Do you use JKSimMet or any other simulation software at your University?

You can incorporate three stage crushing. Jaw crusher as your primary crusher followed by secondary and tertiary cone crushers. Crusher product is subjected to ball mills. The ore ground is subjected to primary magnetic separation and the concentrate, after regrinding, can be subjected to secondary magnetic separation. Column floatation (optional) system can be incorporated to improve the quality of magnetic concentrate.

For more information on JKSimMet check http://www.jktech.com.au/jksimmet. I am not sure if you can include LIMS or HGMS in your flowsheet using JKSimMet. However, Metsim (http://www.metsim.com/Applications.htm) can model magnetic separators.

By the way, Citic Pacific's Sino Iron Project will be processing magnetite ore using 6 AG mills, 17 meters high and consuming 44 MW each. Will be keen to follow the process once it commences.

Your production capacity seems to be low (max. 250 t/h). I do not know the required magnetite liberation size and weight recovery level. Based on the power consumption criteria, the best grinding flowsheet concept seems to be jaw crusher, secondary crusher, HPGR and VM (fine grinding). Why HPGR instead tertiary crushers? Because of the maintenance cost: Cone Crusher OPEX are higher than HPGR OPEX). VM fine grinding is a very interesting scenario. I can send our paper prepared for 2013 SME Meeting if you need it.

My question is why didn't you suggest ball mill for this design criteria? You know that for this capacity we must at least apply two parallel ball mill to support this capacity. What are your thoughts about two ball mill in this design?

In principle, the phases of the enrichment of the magnetite & hematite ores are well known, but the selection of the appropriate comminution and separation equipment is the most important aspect. Two criteria must be simultaneously considered: 1) process efficiency (magnetite and hematite recovery); 2) CAPEX and OPEX. The correct responses to your general questions require the knowledge of the following preliminary information: i) Water availability (dry or wet process); ii) Magnetite and Hematite liberation degrees (liberation grinding sizes) quantified by the metallurgical tests; iii) Types of the associated minerals (contaminants). Without the knowledge is difficult to provide you the appropriate solution. In spite of the lack of necessary process requirements and information, I can respond to your question «Why VMs and HPGRs instead BM?»: These two grinding equipment types are characterized by significant lower CAPEX and OPEX in comparison with BM (general response but I do not know if it is valid in your particular situation. As an example, if the required grinding size is very fine, it is preferable to use the VMs. On other side, if the grinding size is very fine, we are obliged to think to efficiency of the hematite separation. It is possible to be in situation to use the High Gradient Magnetic Separators (HGMS) instead High Intensity Magnetic Separators (HIMS). Sorry Reza, without the knowledge of your experimental and process data, is difficult to provide you the accurate responses to your questions.

This is where these forums are wonderful gentlemen. I am very limited to media milling systems but do respect HPGR and cone crushing for alternative size reduction as a process. I am just not educated enough in this equipment. We constantly have projects requesting finer grinding for better product recovery in ore (magnetite included). The ore bodies we are dealing with are all, different in their geology which calls for a range of both sizes and media grades to be deployed. All possible, with solutions found. The old question will always be cost and efficiency comparisons between which circuits to choose from. I believe both are best, subject to the ore body presented. If the ore body through life of mine suits HPGR and cone crushing (or even AG) then that will be the choice. However if the geology changes a ball milling circuit may have the advantage of more flexibility by having greater adjustment capabilities. To perhaps make this personal comment a bit clearer we have done optimization work on countless milling circuits rearranging ball sizes, combinations and ball grades to arrive at the final targeted particle size at maximum throughput. Media consumption is also a target to minimize in this process as is power consumption however, at the end of the day these costs are what they are. (With improvements made as our own ball making technology continues to develop). Media manufacturing over the last 10 years has had several breakthroughs with new designs being purpose built for specific customers, this process continues and that is why my comment regarding "flexibility" comes from. Take these comments only from information and experienced gained from a source without HPGR detailed understanding - it would be great if this exchange could attract a specialist in that field as well!

2 ball mills in parallel or in line as a primary to secondary grind?
It will depend in the front end of your design to reduce the ROM ore to a suitable feed size for BM milling. I would always be in favor of a two stage milling circuit with a higher aspect mill as the primary - but remember that is only the opinion of a grinding guy and without comment from HPGR and multistage crushing. WI will sway the decision as well as future ore body consistency.

The liberation size of the magnetite from silica for instance can be very fine and require grinding of the magnetic product to as low as 20 microns or less which means you may need to consider the UFG mills such as Florin has suggested as part of your processing circuit. The determination of the liberation grind size can be seen using mineralogical methods such as Qemscan. 

Depending on the amount of hematite and whether it is a viable proposition to recover it then you would want to consider the use of WHIMS on the LIMS non-mags. Do not use the higher intensity magnetic separation straight off as this equipment will be blinded by the magnetite.
You have a big tonnage variation from around 600ktpa to 2.0Mtpa. This is significant but I would suggest that the critical cost aspect of OPEX in terms of costs per tonne would be less with the higher throughput rates. Often people concentrate on the capital costs but the main critical factor which will enable a project to proceed economically is the operating cost.

I studied all comments related to your project. In this context, I would want to reiterate my previous question: Is it a new project, university work or an expansion of an existing plant? The start basis of your work, its development and steps depend on your project types. If you expect to expand the capacity of an existing plant, you are obliged to correlate the new development with the existing/in operation process facilities. If your project is a new or university work, you are obliged to start from the ore qualification and think to the all phases of the process, including primary crushing, secondary crushing or grinding, primary silica removal, fine grinding (if necessary), magnetite separation and, finally, hematite separation. As previously mentioned, the common steps (all 3 project types) you must develop the following steps:

1) Experimental ore qualification (structure, chemistry, natural size, quantification of magnetite and hematite liberation degrees etc.)

2) Development of the metallurgical testing program, aiming the quantification of the ore crushability and grindability (WI) and the preliminary parameters and efficiency of primary silica removal process (usually by gravity separation) and magnetite and hematite separation

3) Laboratory continuous tests or, better, pilot or mini pilot tests.

Based on the results obtained from these usual tests, you can develop primary technical and financial ways of your project and, depending on your aimed production capacity, select the most appropriate process equipment and comminution and enrichment and circuits. During the development of these works, you are obliged to permanently think to the financial aspects (CAPEX and OPEX, NPV and IRR). The appropriate responses to your questions are based on the aspects mentioned above and the process type, dry or wet, depending on the water availability. Please keep in your mind, the advanced magnetite and hematite separation requires the wet process. The negative impact of the high variability of the production capacity can be minimized by the judicious selection of the process equipment size and type and operation concept, finally by the maximization of the plant flexibility. During the last 10 years, I developed 4 or 5 projects aiming the valorization of the iron ores (mixture magnetite and hematite; Africa, South America, Kazakhstan and Canada), including different process types (dry or wet) and magnetite and hematite liberation degrees. If you expect to develop this project, please provide me further information.

Sounds reasonably straightforward to me: there are milling experts or indeed equipment suppliers if you are serious about purchasing equipment who can help you. You can also calculate the mill power from first principles (a la modified Bond equations) which would be a good exercise for a Masters' student and then select the mill dimensions from standard mills made by equipment suppliers.

250 TPH feed rate would not require not particularly large mill and why two would be needed is not clear.

Naturally you will need to have measured the comminution characteristics of the representative ore types and typical ore grades over the Life of the Mine to that would be fed to the mill to correctly size the mill (e.g. Bond Mill Work Indices, Drop Weight Indices, Abrasion Indices, etc.) as well as specifying the feed size (F80), mill product sizing (P80) and preferred circulating load.

I must say that this project is a industrial work and the calculations for this design is extremely important.

The bond equation gives the power about 5853 HP I must mention that the Bond Ball Mill Work Index for this ore type is about 17.2.
The feed size of ball mill is 20 mm and the production size is 150 microns. So if you put this amounts in Bond equation you will see that the amount of power is 5853 HP. so if we divide this result with 2 we have 2926 HP to match with amount of table. It derived two ball mill with diameter and length of about 5 m. Is this calculations right? (The needed factors was considered in this calculations)

You cannot use the VMs due to F80 20 mm. Max size accepted by VM is 6 mm. Under these conditions you need another grinding phase, before the VM. Concerning your BM sizing: I introduced your input data in the BM Model. The results are the following: D = 5 m; L = 6 m; 2 mills; required net grinding power = 1603 kW/mill; Total power = 2892 kW/mill (including balls, overfilling, slurry and 10% losses. This is a summary of the model results. The model quantified all operational data, including the cyclones.

The task of your future BMs is difficult taking into consideration the reducing ration of F80/P80 = 20000/150 = 133 and ore hardness, Wi = 17.92 kWh/t. Your problem is F80 = 20 mm. usually, in the BM accepts F80 = 15 mm, but your ore is hard. Can you check the possibility to reduce the feed size, in order to diminish the BM power consumption and CAPEX? The BM grinding efficiency will be significantly deteriorated by the high recirculation of the fine particles. On other side, the high variation of the production capacity will have a negative impact on your BM power consumption. Your complete comminution information package is not available, consequently I cannot provide you the most appropriate solutions.

But as you say you have BWI = 17,2 kWh/t instead 17,92. Under these conditions the model results are the following: P = 250 t/h; F80 = 20 mm; P80 = 0,150 mm; Specific Power consumption 12,83 kWh/t; Net Power 1603 kW/mill; Power Installed 2675 kW/mill; Mill number 2 units; D = 5 m; L = 5,6 m.