Personally, I have used a similar (competitor's model) hand-held analyzer a couple of years ago and a "pre-hand held era" analyzer, obviously the first of it's kind ever, the X-MET portable analyzer some 30 years ago. Both hand-held and portable models are very useful tools in almost all concentrators, underpinning if there is an on-stream analyzer like Courier. 

Courier, of course, is inevitable in base metal- and similar flotation plants being the best of it's kind process analyzer, but the number of samples is limited, normally max. 12... 24, for time delay reasons. Therefore, a process metallurgist needs the hand held device for quick spot analysis to learn process dynamics, to detect disturbances etc… Also check / orienting lab testing is much more productive if one can have the quick analysis immediately after the test samples have been dried.

You will need on-stream granulometer PSI (Particle Size Analyzer) and densitometer for the grinding mill discharge control, plus weight measuring devices for mill feed and circulating load, plus mill internal load level. LIBS is for another operation.
For control and reliable calculation of the c/l, continuous measurements are needed.
High level instrumentation and automation to a grinding circuit is also costly. If investment is not that limited, then better to acquire belt scales, flowmeters & control valves for all waters, PSI, density gauges, charge volume indicators, flowmeters for h/c feed and o/f, h/c feed pump kW and rpm measurement etc..., not forgetting the software for optimizing grinding.
In addition, as simple thing as power of the secondary grinding mill tells an operator much. Using the mill power as base for control provides however that the operator has the right to adjust the new feed tonnage and pebble feed rate in case of autogenous grinding AG.
Later, the findings and practices of the operators and mill metallurgists can be converted to control algorithms, making the life more convenient and giving more production.
If you need a clear answer to your circulation load control problem, the case needs to be handled around the same table!

Do not forget ore grindability variations ( short and middle term). Thus it would be reasonable to have online cross belt analyzer upstream the mill (e.g. on mill feed belt conveyor) as well as speed variation mill drive would be also excellent.... So everything really depends on the project budget limits.

Regarding the Courier - there should be a correlation (65-70%) between the mill product size and content of valuable elements in concentrate. It might be a way to use LIBS analyzers on-stream to control the grinding circulation.

A note about LIBS SAFETY:

You have no radiation risk. Also, the laser is regulated in our case by a distance regulator and is either class 3B or in certain cases a class 1M laser, meaning that there is very little chance of user injury from the laser.

Hello guys, the job of an experienced operator mineral processing plant + statistical analysis is the base for an optimal control algorithm (IMHO). 

From the experience of designing for new factories cost automation devices such as level sensor, pressure sensor and flow meter, densitymeter is not too high cost. Now they propose to complete the delivery of almost all the vendors and Outotec and the Russian, Chinese producers. 

I agree that the use of any device or means of automation be justified economically. I understand that in the 70 years of Lynch in his book pointed out the effect of 5% increase in productivity and reduction in size of the product by 2-3%, now in 2015, these figures should be even higher. 

The idea of using LIBS data in grinding circulation is correct or optimal circulation load at which the maximum reovery of useful components. This theoretically allows it will save energy, overgrinding avoid both ore and ore grinding insufficient.

Personally I would prefer to have SAG and pebble mills ( if the ore is hard enough of course) just because of better liberation ( e.g. Lebedinsky ( pebble mills) and Stoilenskij ( ball mills) treat the same ore however first one has 3% higher Fe in concentrate). If you can arrange ore processing according to it types ( one after another) you will definitely get more smooth production. To check the ore type coming to SAG you will need PGNA on the feed belt signaling to the SAG mill control circuit accordingly. Different ore types demands of different grinding circuit settings: feed rate, slurry density, circulation load, classification cut size, consumed motor power, mill filling level, mill weight level, mill sound/vibration level and so on. As far as the ore type was detected a selection of the proper process control mode (among several developed for all available ore types) shall be done.

The mineral liberation in the scheme AG + ore pebble mill for iron ore is worse than the scheme Roller-press (HPGR)+ball mill + tower mill (IMHO). 

The circuit SAG mill + pebble mill less cost-effective because AG not applicable to benefits balls nullified. 

The acoustic characteristics of the mill uses only experienced operator and they do not automatition the grinding cirquit but only add unnecessary information. In fact, at the Lebedinsky and Stoilensky GOKa from acoustic control systems refused.

Technological variety Lebedinsky ore and Stoilensky insignificant.

Promising in my opinion mineralogical evaluation of ore block, on-stream and its liberation in the grinding circuit in the stream. 

Perhaps elemetny analysisLIBS will only intermediate versions of the technology and will not be used to control the grinding cycle.

You are correct Lebedinsky and Stoilensky GOKs treat the same ore but Lebedinsky has a little bit higher Fe content (up to 70%) compare to Stoilensky ( up to 68%). One of the reason is better magnetite liberation in AG+pebble mill circuit. Roller-press (HPGR)+ball mill + tower mill (IMHO) circuit for sure is beneficail compare to 3 stage ball mill circuit however do not forget it demands 3 stage crushing upstream the grinding. In fact AG mill ciruit demands less investments than 3 stage crushing plant. Acustic/vibration control of the grinding mill operation provide a possibility to increase the mill charge more close to the optimum. One of its main target is a preventing of the mill overloading. The common general mistake of many benefication plants here is absence of a pre-blending stockpiles where ore will be mixed properly that to decrease fluctuations of mill feed grindability.
Lebedinsky has a little bit higher Fe content (up to 70%) compare to Stoilensky ( up to 68%) - I ment in the final dispatched concentrate.

Compare Lebedinsky GOK and Stoilensky not entirely correct. Because 70% of the concentrate obtained LGOK 7-8 stages of magnetic separation (mineral processing plant + secondary benefication plant) and Stoilensky has a typical scheme with 5 -6 stages of magnetic separation. Concentrate LGOK best in the country it is a fact - but the main reason for obtaining the quality concentrate is further Electrometallurgical Plant or redistribution of HBI. For Stoilensky further metallurgical conversion does not require the quality concentrate to the maximum possible. 

In world practice, the primary mineral processing plant with 5-6 stages of magnetic separation of iron and receiving more than 69% can be counted on the fingers. A standard delivery to China - a 62% iron concentrate (such as hematite without process plant as the supplies Vele from Carajas). 

Comparative analysis of mineral liberation on each operations flowsheet LGOK /SGOK did not anyone, there are some scattered studies that compare not correct. We spend the calculation of process flowsheet with the liberation minerals (each operations) for process plant - however this requires special sampling and most importantly the customer's request to obtain the most efficient payment scheme. 

Averaging ore is a standard and proven method, but there are negative examples in the Russian climate conditions - one of the GOK northwest Russia newest stockpile very long to start after winter. 

I with colleagues have the opportunity to design an indoor blending stockpile capacity to 100 thousand tons of low-cost designs - but as long as the potential interest of customers for such facilities do not. 

As for the cost of building the scheme with roller presses - it is lower than the scheme with SAG mills, I do know from the experience of both design schemes. 

Acoustic monitoring system boot mills in my opinion are good for experienced operators but they are not a means of automating the grinding circuit and a means of remote control at mills with steel lining. 

I was more attracted to the scheme of selective mining and tuning mineral processing plant technology under raw materials in automatic mode - but it's hard to be solved with current instruments and automation and promising in my opinion for gold and base metals. 

Because the use of LIBS on-stream analyzers to control the grinding circulation - we need practical experiments and the calculation of the efficiency of implementation.