In an operational point of view I'm not sure if the Bond Abrasion Index is an important KPI. Usually the liner consumption have a best relation with the operational times (hours running) and operational conditions. Of course, the kind of measurement produces variations in the BAI but in my experience (four Concentrators with difference type of ore) have a secondary role in the liner consumption.

There is a relationship between Abrasion Index and Life Factor for wearing liners in jaw and cone crushing, and you should make material testing (methods used by Sandvik Rock Processing).

Fully agree in the crusher case. A good example is the behaviour (liner consumption) in the same crusher with different material (ore or slag).

I agree with strongly with both the observations, however, the liner profile especially in the case of SAG mills is worth paying attention to.

May be useful if you put the link for the method you mentioned. Any further information regarding the liner profile relationship with the Abrasion Index!

I have no knowledge of any ready reckoner available for profile relationship; however it should be possible to build one for your application over a period of time with experts in liner design. We attempted this in an earlier organisation where I was working about 15 years back. This resulted in significant savings in cost.

We found a strong correlation between power draw and rock strength, which then was correlated with liner wear rate and change in liner profiles, production capacity, and power.

These points were published in SAG 2001 as one of four papers I co-authored. Based on power draw and through-put we could then predict the change in liner wear and resulting production levels. One such article is: QIU, X., POTAPOV, A., SONG, M., and NORDELL, L. Prediction of wear of mill lifters using discrete element method. Int. AG and SAG Grinding Technology Vol. IV. 2001.

I assume you have read the relevant literature. If so, what additional points are you’re seeking? I do not believe the abrasion index, of milled rock, is as good as rock strength in predicting wear rates although I have not heard of the attempt to sample rock sizes that can be used to measure wear rate from an especially designed sample.

I just cannot find yet any reference relating the Abrasion Index to the wear rate, so that I ask whether anyone has past experience on this.

You will have to develop this for your application. This is normally based on run hours. Most metal liners are Mn Steel. Once you have established the wear rate then you can start tweaking the liner material composition and optimise that. While tweaking the composition, take care to see that you have the new material in one row only. This will give you a very good comparison. Trust this input helps you.

This might help some and have added references:
http://is.gd/HVeBOa

At least in the Cu industry the BAI it not a practical tool! Of course there is an effect of the kind of ore, but usually in the mayor operation of Open Pit and UG there are a good mixing (blending) of the different kind of ore. Then in summary the operator use an average values to estimate the life of liners.

We tend to collect Ai as part of a geo-metallurgy program before doing a mill design, but it has limitations. At best, it gives you a highly approximate definition of relative abrasive properties and is more suited to grinding than crushing.

Bond's original estimation for crushers was developed back in the 1950's and, arguably, is obsolete given the improved metallurgy of modern crusher wear surfaces.

When you get into the details you will find some higher order wear factors that are very relevant:

Liner material and Brinell hardness rating - high chrome vs. Manganese.
Liner shape throughout its life cycle - we increase Cadia 40 ft mill liner life almost 50% when you factor total throughput and liner mass metal loss
Liner shape, mill speed, ball %, and mill end geometry all play a big part in total liner life cycle. Mill end geometries integrated with liner profiles can add many 1000 tons of production to a typical mill. Most mill liner designers do not use modern DEM code to assist, where the code can track historical measures and then propose advances in liner geometries - belly and end lifters.
Rock strength variations was found to be very important - please read the SAG 2001 paper on the same. We divided the rock strength into 3 divisions - soft, medium and hard based on mill draw power and throughput. These simple measures were able to predict mill life close to 99% for various ores, power, and throughput over the wear action and its change to performance for a number of large sag mills with known ball/ore ratios.

Abrasion index, in my opinion, is way down the list of important criteria.

I did some field and lab tests using the AI Bond test. However I used materials equivalent to the applied in the field on the pads of Bond test machine. As result I got between 6% of deviation with respect of the field results. I am still studying a way to get more accuracy.

Do you mean you compare the Bond Index and compare to something in the field?

For new projects or expansions, normally the vendors do some crushing tests. Regarding AI, it is difficult to predict the abrasive action of rocks, minerals, and ores. Often, rocks belonging to the same geological classification will vary widely in abrasiveness from one locality to another. Therefore, in order to select the proper crusher, you must have some reliable measurement of the abrasiveness of the specific material. The Abrasion Index’s denotes the weight loss of the paddle made of clearly defined steel during one hour of rotating in a testing drum with material being investigated. A sample of material to be tested is placed in the drum. The material size used is 12.5-19mm and the quantity is 400g. The drum and paddle rotate for 15 min. The paddle rotates ten times faster than the drum. The drum is emptied and the process is repeated three times so the paddle is subjected to wear for one hour. The paddle is weighed before and after the test. The weight loss in grams is the AI of the material tested.

See the paper by Carlos & Levi is very handy. "Measuring, Predicting and Managing Grinding Media Wear" by A.Giblett and J.Seidel 2011 also contains some very good data.

http://is.gd/KK4H6E