You want to have at least 60% liberation of the ore. For gold this can be just a few microns so grinding to 80% 25micron is acceptable, given that operating costs increase with finer grinding. As always it’s a trade-off depending on ore size and gangue components.

Don't you also want to control over grinding by minimizing recirculation of ore passing through grates into grinding chamber and not leaving pan/grate chamber on first rotation?

In addition additional over grinding may take place at the grate, but, still in the grinding chamber. This occurs when ore is prevented from passing into the exit chamber by larger obstacles (balls and rock) that stagnate at the grate face.

If these conditions are improved, the grate size may be reduced and also witness a steeper particle passing curve which is better tailored to control over grinding at the same time.

Some Discrete Element Codes (ROCKY for one) can simulate ore breakage and classify the size distributions in all cases above.

A large part of over grinding occurs when ore passes through the grate, but, is then discharged back into the mill (40-50%) as measured by some Population Balance Models (PBM) and Discrete Element Model (DEM). We observed this detail when modeling a 40 ft. SAG mill for recirculation when ore passes over the grate as gravity pulls the charge in the pan cavity back across the grate, before the main flow either leaves the pan cavity or is recycled within the cavity.

This phenomenon is made worse by higher speeds and straight radial pan shapes.

The phenomenon described in your message has been a great concern for liner suppliers as well. Not only does it influence the grind as you pointed out but it also add to wear in the chamber due to slurry travelling up and down the chamber instead of being discharged in one rotation. We have developed pan lifters that deals with this problem. There is no standard design so each case is treated individually.

One such case I am working on, we have improved slurry discharge by 60% (DEM results). It also seems that the faster the mill speed, the more efficient this pan lifter design becomes. If you have a client in need of such design improvements, we are happy to design and supply optimized pan lifters and grates to suit their needs.

In Platinum ore for example, a specific granulometry is targeted to achieve maximum floatation recovery. If over grind occurs, all the gangue and over-ground chromite float with the mineral which in turn costs the mine millions in smelter penalties. There is a limit to the %Cr in the concentrate allowed by the smelters.

I am very curious about your casting improvements to control fines returns to the grinding chamber and that which recycles within the pan cavity. Can you share any geometries? I believe you are using our ROCKY DEM simulator.

Water split to cyclone underflow is a good indicator of fines being held in the grinding circuit and a system that may be prone to over grinding. Comparing your valuable component size distribution in the grinding circuit product vs. the laboratory baseline will give you some clarity also, if the p80 values are similar.

I believe the art of size reduction is because no ore are really the same across the earth layers so that statistic tools often very important to look at the trends.

To change the grinding conditions is a good way. Some phosphate plant use grinding rods to avoid over grinding.