A number of people have said that Ai is not that useful as estimator of media wear.Are you able to publically share this development and make us all aware?

Sounds very interesting, particularly the estimate of corrosion. This would be most relevant for mild steel balls with highly sulphidic materials (e.g. either amount [as found in regrind mills] or type [such as pyrrhotite - the 'orange skin' effect]) in the absence of lime (i.e. pH).

It will be interesting to know how the industry jumped from the old Allis-Chalmers procedure to the new SMWT. Someone may have studied it and realized that is not longer applicable.

I read in a paper that the old method would produce differences of about 100% for repeated tests.
Can you share the procedure that JKTEch develops? Is this licensed anywhere else, for example in Canada?

First the old test was a dry test and not really applicable to wet mills. Second, if done blindly it tends to overstate wear for modern media metallurgy. The test has to be done with the same metallurgy coupon as the proposed media. Third, I doubt it applies to large balls in SAG mills. That being said, it was pretty good at comparing a known ore and unknown ore.

This is a good occasion to mention that the metallurgy has changed greatly with finer grained alloys (toughness) with better alloying and heat treatment (hardness and higher corrosion resistance). We have seven (7) grades of SAG balls and as many grades of ball mill balls.

We have some clients that mill in sea water!
One customer has a 19% ball charge and a total charge of 21%
Abrasion attrition is not the only steel consumer, breakage and corrosion can play a significant roll. Numbers at the extremes that I have personally encountered have been 80% breakage and another at 60% corrosion.
As a media designer and manufacturer, we approach this from a different direction. We gather the data variables for a complete milling circuit to assess impact in relation to mill size/speed and particle in/out targets. Following that, media size and grade options are considered; always with maximum efficiencies being the end game.
Looking at the extremes we can make a 6.5" SAG ball for a 40' mill, also we can make a cast medium chrome ball for a small vertimill.
Having said that I do believe that a formula can be devised but the variables will come from a complex set of questions. Today these answers (variables) are usually known - therefore possible.
My thought is that it is better to use the huge data bases of the manufacturers, match up the basic numbers to find a "similar" circuit and evaluate that media and liner/lifter wear.

But that is if you have an ongoing operation. If you are starting from a Green Field study then would you recommend/suggest using the tried and tested Allis-Chambers Abrasion Index for estimating OPEX? Or is there an improvement?

An analysis from some years ago by Peter Radziszewski (Minerals Eng. 2002) comparing the Abrasion test to actual mill data showed that Abrasion test estimates are hit and miss, mostly miss. That was the motivation behind developing the new methodology which gives more accurate media consumption predictions. This test takes into account both the ore type and media type to calculate the consumption.

This new test is quite valid for Greenfield OPEX estimates (as well as ongoing operations if you are considering switching media).
The fact that you get separate estimates of corrosion and abrasion estimates means that you can also choose improvements to operations/media to minimize media consumption depending on the dominant wear mechanism.

Is there any work done to decrease corrosion?

Has anyone quantified what of factors playing a role in steel media consumptions is the most important?

I have used the Molycop database and obtained the following updated Bond Equation

Ball Mill Media (kg/kWh) = 0.0944*Ai^0.299

Aidan Giblet and J. Seidel published a paper on the 2011 SAG conference (MEASURING, PREDICTING AND MANAGING GRINDING MEDIA WEAR) and they obtained

Ball Mill Media (kg/kWh) = 0.0817x(Ai)0.498

the two equations are very similar.
Two or three groups of people are finding updates to the Bond equation in different places using different data bases.

Is it time to officially declare the Bond equation as obsolete?

I think the formula just give a primary estimation while many times is different in practice.

According to my experience, "Mill Amperage" is a good criteria to evaluate grinding media consumption. By this way, we can determine how much balls we have to add to mill during operation periods. The "Amperage Change Figures" is varied according to ore characteristics; however, I experienced the detrimental effects of low filling factor (when the mill is running with low slurry due to temporary upstream problems) on ball consumption which shows quick amperage drop in control system during the period.
In this case, we have to add more balls in same periods.
On the other hand, beside of abrasion, breakage and corrosion, I think the "Ball Coating" by ultrafine particles also is an important factor in evaluation of ball consumption inside the mills.
Sometimes, the coated layer reduces the direct metal-metal collision which could be the source of more breakage.

I obtained the equations above from data that Molycop provided and from Aidan Giblet (Newmont). Molycop makes clear that kg/kWh is the parameter that must be analyzed, which I support. I made a quick correlation with what we see at the end of the day, the Ai vs. the gr/tonne

In essence the gr/tonne is not a good indicator as the mill inefficiencies are included in the g/t, which includes mill loads, F80, P80, operation variability, operator’s knowledge about the relationship between circulating loads and efficiency, etc.
The Amps in a mill are a derivative of the Kg/KWh, which is the ultimate most important indicator for steel balls consumption. When one make the integration of Amps vs. time one obtains the kWh, so those parameters are proportional, but also the kWh gives an indication of the mill efficiency.
The balls coating is an interesting concept too. Thanks for sharing. Several operators have tried "grinding aids" which essentially work trying to take advantage of the corrosion chemistry in the balls surface. Certainly the ore and the ions that are produced during grinding have an effect on this surface chemistry, so we cannot under estimate this field of study.
Molycop obtained a new empirical equation for ball mills
KdE= 1.36 [(Ai-0.05)/0.20)^0.166 * (F80/5000)^0.069 * (pH/10)^-0.243]
This equation takes into consideration more parameters so as rule of thumb it should provide more accurate correlations between observed and measured variables.
In any case and to obtain a quick flavor about how well the Bond equations are predicting steel wear the comparison I did is valid as it provides a first comparison for steel consumption. A big majority of people is still using the old Bond equations without asking themselves: do 950 g/t or 1,100 g/t of steel consumption in the ball mills make sense? When we apply our experience from the last 20 years we see right away that steel consumption in the range of 1,000 g/t in a ball mill is not correct, and that is a fact. But people needs a tool to forecast costs and they keep using the old Bond equation because they don't have the authority to say this is the correct equation we should use or even worse they don't even have the data that is needed to update these equations.
The mining industry must fund a project like this. It is in the best interest of everyone. Developing work like this pays itself very quickly. The small investment is already paid for if a mine can save 1 million dollars per year in steel balls. A mine expending 20 to 40 million is steel balls per year may reach gains of this nature.
Having a big budget for steel media consumption, based on assumptions of higher consumption, leads to complacence while the money is worn out in the mill, literally speaking.

I wish that CEO’s in the mining industry would hear of these discussions were mineral processors from all over the world are telling the industry that there is room for more savings and for energy efficient initiatives.
Cost control can help business where there are inefficiencies but I also believe that cutting costs may lead to prevent improvements activities, which from my perspective is not the way of doing business. No one wants to expend the money because everyone is trying to get the lowest budget possible, so pennies are saved while dollars are transformed into noise, heat and waste.
So who is going to help us?

I see some adjusted formulas for ball mill media. How about SAG mill media consumption?

What are people using to get numbers for a green field operation?

There is an equation developed by Newmont for SAG mill

I suggest that you See the "Measuring, predicting and managing grinding media wear" paper in SAG 2011 conference proceedings.

We used the Molycop Non-lineal ball wear model for a SAG Mill with good results for 40 days test period.

We make the test to validate the BM Ball Add Instruction of our BalMet Online suite which we used for grinding circuit optimization.

More info in this link (see page 4):https://www.dropbox.com/sh/ab13suegd9weqto/4hJMagDlCD

we will have the result for full implementation of the ball wear model for 1 SAG and 2 Ball Mills in different plants in March 2014.

I mentioned in an earlier post that JKTech now has a commercial test that will be give accurate predictions of Greenfield steel media consumption. The methodology involves running tests using the ore and target media types generating predictions of consumption for any mill size. There are several papers on the test development.