The volume of grinding media in a mill is directly related to grinding efficiency. The higher the volume of grinding media the more effective the grind. Balls must be added to maintain the media load and mill power draw. The power draw increases as balls are added and decreases as media wears down: add balls.

The main control on the power draw for a BM is the load of grinding media however adding water to the cyclone underflow can also be short term control of BM power.

Higher ball loads will result in finer grind but don’t overcharge. If throughput of slurry to BM decreases the mill power draw will increase because balls grinding against balls are drawing more power: finer grind.

I totally agree with you that the low charge rate directly affects the grinding efficiency, but in here there is another thing we have, that shouldn't be overlooked, it is the power consumption, at low charge rates below 20% the power consumption increase by 6-15% for the ball mills.

What are the dimensions of the ball mill you are looking at? 40% ball charge is quite high (unless you are looking at a small mill).

A difference (duty vs. maximum) of 5% to 10% is fairly typical and you would likely struggle to perceive an energy efficiency difference between the two operating points. Beyond this, the situation may change.

The 40% vs. 20% is a very large difference in ball charge. Under these conditions I would agree your slurry pool is going to be large and this is likely to cause issues from both a power draw and power efficiency perspective. This is not easy to quantify without more detailed analysis/benchmarking.

This shouldn't be a show-stopper, as you say, the mill is oversized so the duty throughput/grind size should be attainable, albeit the exact milling conditions/power efficiency may be difficult to pin-point.

If this is a concern, and if the client has already bought the mill, your options may be limited. One option could be to install a grate possibly with high capacity pulp lifters, to minimize the slurry pool at 20% ball charge. The grate could later be removed to allow for operation at higher ball charges. Obviously this creates its own set of separate issues.

Actually, you need to differentiate between capacity and grinding efficiency. At a lower load capacity will be reduced but grinding efficiency will increase. There are only a few examples of this (most sites go for throughput rather than efficiency) but it has been shown to hold true for overflow mills.

With very low ball charge (less than 20 %) and low slurry density power draw will increase because of increasing toe angle and also more lifting. Also with low ball charge slurry pool will reduce grinding efficiency. Beside lower ball charge will increase P80 of ball mill because of less number of impacts but mean residence time will increase by lower ball charge because of more volume to occupy. I think at last in this situation, power consumption will increase.

You missed the point that the feed rate is lower so the residence time is longer so the product size is not necessarily larger. By the way, circuit product size is not determined by the mill but by the classifier. The mill performance will affect the circulating load.

I don't get why you think that the toe angle will increase and there will be more lifting. I don't understand why the slurry density would be any lower. Why would it change (or in mill that controls it why would the target change)?

All overflow ball mills operate with slurry pooling. Do not confuse with a AG/SAG mill where pooling can be detrimental. In ball mills all the media is balls with a high SG so the impact of pooling is not as large. Additionally, in ball mills grinding is primarily through cascading not cataracting (compared to an AG.SAG) so the slurry pool is not a factor unless density is high, which leads to viscosity issues.

I have been told at a few sites I visited where the ball level is lower (e.g. 25%) that they operate at that level because they don't need the power. What they have observed is that the specific energy is lower and that the media and liner consumption is lower.

Of course all the ball mills operate in slurry pooling conditions but with low ball charge it will be larger than before so it will affect the P80 (Ball mill Product 80 % passing not circuit product) badly and it will increase. He had mentioned adding water to classifier underflow; this work will reduce slurry density. With low slurry density (under 60% solid percentages) because of more charge liberation to lift, the shoulder angle will increase (more lifting). High ball charge will result in developing toe position and because of high ball charge after a certain point of charge level; power draw will decrease so with lower ball charge there will be no toe developing.

You mentioned correctly that most of charge motions in ball mills are cascading but with low ball charge the NUMBER of impacts (most of them with low energy level) will decrease.

The effect of low ball% full on grinding efficiency is:

The throughput of mill will decrease with the time;
The P80 in the discharge of the Mill, will increase because of the high time of particle residence in the Mill, and for this reason the grinding will be lower efficient.
The power specific consumption will increase, due to the lower feed rate.

I don't know where you got the low mill density of 60% but I don't see that in the postings above. I don't agree that the lower ball charge will lead to a higher P80 because the feed rate will be lower and pooling does not affect cascading action so low impacts will not be affected. Yes, the number of impacts will be lower but that affects capacity not efficiency. What is important for efficiency is the location of the energy spectra (high impacts vs. low impact). For ball mills low impact is more important because of the smaller particles.

You state that the residence time will increase so what is your evidence that P80 will increase? Yes, the feed rate is lower but so is the power. If the power drops more than the tonnage then the specific energy will be lower.

The only thing I will add to this discussion is that I have been to plants where they operate at low load (-25%) because they don't need to have the capacity. They do it to save energy and reduce media consumption. None of them have complained of low efficiency. So, based on my experience, tell your client that they will be fine.

Quite an interesting discussion here, with many different views! Why is efficiency important anyway? How would you define it? What would be the measured benefit? Would is matter if commodity prices were depressed? What would be considered "high" and what would be considered "low"?

I agree 100% with you regarding low loaded mills with good, even great, efficiency. In fact, some studies point out the importance of reaching this "optimal" load % that necessarily isn’t this common heard 28-30% value.

I just want to add that more than a fixed load %, it’s important to keep the balance between both mill's chambers. In my experience I’ve seen engineers planning and making calculations for a grinding media reload in Chamber 1, not considering at all the load % on Chamber 2. What happens next? : You have a wonderfully filled chamber 1, but your Chamber 2 isn’t able to keep the pace with Chamber 1; in fact, you just unbalanced your mill.
Just want to say that the mill will work as good as your worst chamber.

Low ball ratio is not the same as low mill density. If the slurry density falls below 80% then the slurry will not adhere to the balls and there will be no further abrasion grinding (only impact grinding).

Reducing the ball charge will reduce the grinding capacity, and the comment on installing a grate discharge is a good one as it will let ore out sooner, thus minimizing overgrinding, which will occur if the mill is (temporally) too big, or one chamber is too big in a multi chamber mill. But the reduction is not a direct reduction (50% fewer balls 50% less grinding) as there is a factor from the liner role in grinding.

If one is planning to increase (double) throughput then I'd always suggest two parallel mill trains rather than operating one at 50% capacity unless the reduced throughput was less than 9 months.

If there is a ore shortfall then I'd recommend either filling the vacant space with low grade ore, and reducing the lost energy, or custom milling so that the mill is always full.

It may also be cost effective to look at milling for 6 months then shutting down until the ore pile is refilled. Shorter periods of milling such as week on -week off are seldom effective.

To increase milling capacity put a Derrick screen on close circuit with the mill, so the circulation load reduced substantially.

There is no need to add low grade ore or to change to a grate discharge. There will not be any vacant space because this is (I assume) an overflow mill so by definition will be full. Adjusting the ball charge to meet the capacity demand will ensure that it is not overgrinding (if it is overgrinding then it has more capacity than needed and the ball charge can be reduced). In order to maintain good efficiency the grinding circuit should maintain a good circulating load (e.g. 250%).

I was doing some research trying to see if anything has been published in this area. These papers are more applicable to the cement industry and less applicable to the minerals industry, but they are interesting nonetheless:

D. Longhurst and M. Wilczek (2013): Efficiency and grinding media filling level (http://is.gd/7ZjQaf)

The first paper presents two commercial examples where higher energy efficiency was observed at low ball charges.

Obviously there are numerous potential inaccuracies and uncertainties associated with commercial mineral processing operations and a handful of case studies with conflicting conclusions is not a convincing design basis.

If your project is sensitive to grinding efficiency (I'll let you decide the economics), it would be prudent to assume your project may initially have a lower grinding efficiency and put in controls to help mitigate this risk (if indeed it is an issue). As I said previously, to give a quantitative answer more detailed analysis/benchmarking is required.

If you maintain a high ball charge with low tonnage, you'll overgrind. So you can low the ball charge, these means less interaction mineral particle with steel balls so reduce grinding, but also it means that you bed density will be lower, since you'll reduce the steel (7,75 kg/m) and replaced with mineral slurry (let say 2,0 kg/m for copper mineral) to maintain the volume, thus reducing energy consumption.

But there some worth to point out and it is the consequences of not evaluating the degree of the lowering. As he mentioned paper of David S. Fortsch (2006), when lowering the ball charge to less than 25% volume the energy efficiency goes down for that operation in particularly. And it also make sense, just imagine that you're feeding the ball mill with a high P80, that needs more cataracting than cascading, a you just low the ball mill to a point that cataracting is heavily reduced.

So every operation needs to evaluate particularly the degree of ball charge, and sure there's and optimal minimum, and variables like L/D ratio, Mill Size, and feed particle size had a strong effect on this optimum value.

If the ball charge rate is between 8-12%, it will be SAG mill. If it is more than this, it will be ball mill. SAG Mill's effect is mainly crush the big ores and then use ball mill to grinding the small ores.

We have run our ball mill at 25-30% ball charge. It is 8.3m length and 5.5m Diameter. Operations guys are suggesting we get this up to 40%? What must we expect?

Ball size will also affect the grind P80. Efficiency is in part determined by the Axb value of the ore. A small lab mill can give guidance on the influence of ore strength and ball size. Larger balls will increase P80 in ball mill, thereby reducing overgrinding. If you believe the entry ball size sets the initial point on ball population and exit size is fixed, since the ball wear is near linear, in some camps, this raises the P80 = can only result in grinding larger rock.

Throughput can be optimized from the above plus:

SAG mill optimization to feed ball mill optimized P80. Most SAG mill are not optimized for the combined SAG & Ball mill throughput such as:

SAG & Ball mill % ball content
SAG % ore content
SAG grate size and end mill design including grate geometry, location, shape, pan cavity, recycle % in pan cavity, et. al.
SAG lifter and end cone geometry - can change SAG mill performance more than 20%-50% depending on P80 transfer size and Axb, some of which is published and some will be published at SAG 2015.
SAG mill wear rates and performance changes with wear morphology, grate details, mill speed changes and optimization with speed control
Ball feed size
SAG mill water flow rate
SAG mill power vs. wear geometry - maximizing kW-hr/ton & tons/hour in combination
SAG mill liner change-out life/cycle vs. tons/kg consumed

Ball mill liner efficiency based on liner geometry and its influence on kW-hours/ton, total tons/life cycle, and P80 transfer to cyclones. Ball mill grind is based on different principle; it has some attributes similar to SAG mill optimization. SAG mill breaks with stirring the kidney at optimized kidney specific gravity and maximum stir rate per mill revolution, whereas ball mill breaks down ore with increasing particle-to-ball interaction.

I say this by studying many SAG mill PI archives (equivalent) and a few ball mills with data on full circuit over many liner changes and some changes in ore properties.

To use a potential of your ball mill at the moment when you do not have to feed it with full capacity you could increase its feed size e.g. up to 20 mm.

In dry cement ball mills, there are studies done in the past which clearly shows that energy saving make sense reducing 1°chamber ball mill filling degree. This thread-shore level is considered between 20 and 21%. Below that level mill production/consumption curve do not make more sense increasing the specific energy consumption instead of lower it.

Probably you have similar studies in your industry

Give me actual data of you grinding flowsheet: feed size, product size, power of mill motor, capacity, ball size, numbers & diameter of hydrocyclone ball charge, mill speed. I'm calculate optimum capacity you flowsheet.