This is an issue that I encounter very regularly, particularly here in Chile. The problem I find is that the operators (and I include the metallurgical staff in this) do not believe that a change of 1% is determinable in the plant. This comes from 2 things, first the plant control is often poor and thus daily variations of 1 or 2 percent in recovery are very common, and second they do not understand statistics and how these techniques, when combined with a well-planned and run plant trial, will be more than capable of identifying a 1% difference.

Unfortunately I feel that we are currently working in an industry obsessed with throughput and they consider that by investing in increased throughput, they will have an easily quantifiable increase in metal production. This is the current strategy I suspect is often employed and as such, innovate thinking is not being promoted.

I agree with your point on measuring plant variation. This is a big problem, a lot of it stems from plants not having the basics right and thus don't have stable flotation circuits. By basics I mean functioning control on the air, level and typically density. It's not that they don't have the best-newest-shiniest probe or controller it is that they don't understand and maintain the ones they have.

I don't think its fear of getting it wrong, its more resources and who to believe. Any plant met has limited resources to do work, a key one being time. They are also continually visited/contacted/harassed by vendors with new super products that will do everything from improving recovery/reducing energy/reducing reagent/reducing hair loss etc. They all claim to have the solution. The question is which one to believe. It’s a shame but there are a lot of people out there who are pushing 'snake oil' that don't understand the process and just waste people's time by telling they need something they don't which further reduces the time they (site mets) have to test technology that might work. With this in mind is it surprising that they are sceptical about your claims and reluctant to test? Most sites can tell you what 1% recovery is worth (and it is a lot). You can say that you can improve this but do you have data from a similar float / similar site that backs it up? Bare in mind even if you have this data they still might not believe your claim so it might be time to try another site! Anyway I've had my rant now and for the record Outotec flotation cells not only improve recovery but make anyone who buys them a better lover.

Your rant was much appreciated. Yes the data speaks for itself most of the time, however the technocrats tend to focus on what is achieved in lab trials and not plant trials an odd way about going about things but maybe I should try Outotec's sales pitch of reduced hair loss and improved loving capabilities. 😉

Typically the two main suppliers of equipment seen to have things contractually tied up and therefore mines seem to have their hands tied.

Thanks for your input. I agree with you completely in that increasing through put without improving the metal production is a false economy. But in the end only the fittest survive or so Darwins theory goes.

When working with a flotation chemicals supplier my general finding was that most operations may get 1-3% from a reagent change, but could easily get 5-10% from better operation of the plant and operator training. If operations do not know how to optimise what they have, how can they gain from changes? It is rare for operations to get an audit of operational practices from experienced persons from outside the operation, hence poor practices become entrenched.

This is a good topic, one that many suppliers wrestle with and even in some cases customers. As mentioned a couple times above, the best overall metallurgical improvements in many/most cases will be from operator training, making sure the plant circuits are working well and calibrated. About a decade ago I was involved in a good conversation about just this point and in those discussions operator training and continuing to monitor relative performances was deemed low hanging fruit for improving metallurgy. Differences in how operators run their circuits and even changing the circuit operation starting at the beginning of their shift can impair grade-recoveries on overall average.

What compounds the issue with innovating new products is that the plant circuits truly are quite variable by nature due to ore, plant mechanical issues, etc. and require statistical analyses to assess differences. And it takes a lot of data and proper statistical techniques to define improvements due to the high variability. Unless the change is easy to make and relatively low impact (reversible) and/or the product performance is at a 'wow' level, adoption of new products in the marketplace takes a long time, measured in years.

I am one of those "operators" that you have been mentioning in the above comments. I can assure you that all of you are correct in what you say. I recently finished commissioning a new Flotation plant and the client had added a new froth camera set up in recognition of the new technology- that you have been talking about. It worked very well in parallel with the way we were running the flot circuit. In fact recoveries and grade were far in excess of expectations which was of course good for us as commissioning crew. The day we handed over to the client-AFTER training the operators for the last three months of the commissioning project, the circuit was changed to the way the original old circuit was run, it was then shutdown for 8 hours and all the reagents left on.

So getting your circuit running correctly goes a long way in getting that +1%.

Especially in coal plants around here, you have NO chance of accurately measuring a 1% increase in yield. Any suppliers who claim to be able to do this are taking the piss!

Simply due to the sampling used (or lack of) they just can't measure around a circuit that accurately. Nowhere that I am aware of has the correct sampling equipment and regimes required to determine a small yield increase. Some are lucky to get a yield value, period.

It always amazes me with say reagent trials where they take a few grab samples and magically report a 3% increase in flotation yield. I bet this 3% is +/- 10% at best!

The 1% variance is likely to be the error just in the laboratory analysis side of things, negating the errors (often extremely large) in sample collection on site.

I know where you are coming from! However in terms of the 1% I mentioned this was in terms of the value of a 1% increase over and above what is already being achieved and then using this to calculate what value an improvement of greater than 1% could create for an organisation.

It’s all very convoluted. But it is amazing when you say to someone that you would like to trial something at their plant and they say no as the test data was not consistent. In the same vain I would often like to say to them, let’s see you provide plant scale test data that is always consistent and without fault. As you know holding all other variables constant in a plant is nigh impossible.

I think if we could convince the government to give the mines tax breaks for R&D trials of technology you would suddenly see every mine in the country using new and innovative equipment.

Again thanks for your input everyone. I can see a research thesis coming out of all the information here 🙂

AusIndustry do offer tax breaks of 133-150% for the cost of R&D projects, which is great, but you still need the support from the big boys to get on plant and trial your equipment!

In terms of the yield increase, due to such large throughputs even a relatively small increase such as 1% is worth astronomical amounts! This is why every single yield point should be chased and is well worth getting. Sometimes though, it seems just by producing extreme quantities of product, this tends to overshadow the efficiency side and a lot of saleable product is being lost.

As mentioned, an operating plant is seldom at steady state and never 'easy' to sample accurately. This is why anybody looking into trials has to be aware of the variations in the streams to be measured and account for these in the sampling campaign.

During a pilot scale flotation test at a CHPP, we saw a variation in our feed stream (plant Classifying cyclone overflow)of 10% ash (59%-69%) within the space of 2 minutes! Variances of over 15% percentage points in ash were also seen over the 20 minute trial.

Swings of this magnitude were also seen in the solids concentration (which should also be analysed and used to calculate yield, rather than just relying on one method - usually by ash). This was also occurring whilst the plant and cyclones appeared to be operating 100% normally.

These results were obtained using full stream, calibrated, automated slurry samplers, specifically built for the trial, which exceeded Australian Standard designs. In the end, we had to move to 2 second intervals between sample cuts to try and capture the stream variance, as the original 2 minutes was not sufficient.

Imagine what sort of numbers you could've got with just simple grab samples?

My comments were related to sulphide flotation. My experience with coal is very limited at best. Nevertheless, plant fluctuations are certainly not limited to any particular type of processing and they are common regardless of what you are treating or your plant setup. This is why robust metallurgical plant sampling and surveys are so important, and why single spot samples are not good enough, as I am certain you well appreciate.

Recently at Procemin 2013 in Santiago, JK Teck presented a paper related to how they prepare for and carry out flotation plant sampling and survey campaigns. I was little surprised to learn that they typically plan to be onsite for 1 or 2 weeks to ensure that a sufficient number of surveys are completed to most accurately model the plant. In 2 weeks an efficient sampling team can collect A LOT of data! This would go a long way to mitigate the variability in plant feed. Not cheap, but the rewards would easily justify the cost.

I have been through this evolution many number of times as an operator. Considering the number of proposals that were offered, we had to pick and choose which ones that would fit our requirements, ones that would return the most bang for the buck if you will. The plant had a recovery rate in the mid 90% range. A statistical study found that we had to run the plant a total of 44 days alternating normal versus test conditions to have a 95% confidence level of a one percent improvement. At best we could run six test sets per year. In actuality we could run 2 tests a year. This involved set up, training, analysis and reporting. During the six years I was involved we only found one improvement that met the one percent improvement goal. From this you can see why we turned down the vast majority of the proposals offered to us.

“Yes sir, we have this great new product. Atomised mouse milk! Forty dollars an ounce and comes in a five pound bag"

What I would like to know is the best way to approach a Company if you do have a new innovation, and would like it trialed. And I think there are two levels of trials. The first level is where the inventor actually wants to know the improvement so he can tell others. The second level is where the client wants to know the improvement.

Personally I am more at level 1, and the main bottleneck is Companies not wanting to release the results. On the subject of plant monitoring I just don't think that working out whether there has been an improvement of 1% requires as many tests as stated.

I am currently developed a software system called MMPlantMonitor. Basically the idea is to linkup analysis from separate audits or plant trials. The whole basis of this development is that I am quite convinced that mineral processors are not taking full advantage of the statistical/properties of the data. MMPlantMonitor goes to a depth of understanding beyond conventional (i.e. it uses a particle property model inferred from conventional measurements) and relies heavily on probability theory rather than statistical theory (the two are not exactly the same).

MMPlantMonitor distinguishes unit behaviour from particle information. Consequently one can identify whether unit behaviour has improved.

Pondering the original question related to getting customer support to evaluate new technologies, my opinion is that a supplier must find a customer focused on resolving or capturing an opportunity their product/technology might offer a solution. There must be strong congruency. This can be related to reagent recoveries-grades, equipment that provides more reliable service, more accurate measurements and so forth. My background covers the bases from operating company R&D and operations to reagents and equipment sales. Since I have been all 'round all sides of the desk' so to speak, I understand that until a customer has identified an issue/opportunity and focuses attention and resources on that matter, suppliers will not get far. A supplier must find alignment with a motivated customer. And this requires market knowledge and finding customer's whose needs and motivation level will support the product/technology evaluations.

Particularly if the technology is very new, no supplier fully and completely understands where their new product provides the best fit. Customers understand that but will work with the situation if they feel there is a potential return. Note ultimately this is issue/opportunity focused, not new product focused. Customers are ultimately agnostic (as much as anyone can be) to what product/technology solves their problem/captures their opportunity efficiently and effectively. As long as the new product/technology offers the potential benefits commiserate with customer resource investment, customers often will be supportive in evaluations and development or at least until a solution is found whether with the supplier's product or alternative solution.

In order to ensure the cost-benefit is balanced in the customer's favour as much as possible, it is imperative the supplier works closely and supports the customer as much as reasonably possible during the testing and subsequent analysis stages. And also takes all efforts to minimize customer risks by proper planning and execution, or as much as can be anticipated. Rare is the product works ideally at every mill or mine due to the multitude system and processing differences and impacts. An important part of any plant study is learning product shortcomings and strengths under different situations and scenarios. And have the integrity to let customers understand what the supplier learned because that goes a long way in establishing credibility and support.

Finally, suppliers need solid and experienced process engineers involved in the testing designs and evaluations. There are very good reasons customers are more readily open to those experienced mining professionals who have good reputations as they know the potential success can provide a return on their time and money investment.

Having been on the receiving end of supplier phone calls, and cold calling salesmen, I, like most people are wary of Salespeople, whether they be telecom, vacuum cleaner or Pump salespeople. Most are full of marketing hype and BS and give you demonstrations that are barely real life scenarios. So it is with reservations that you let someone come into your office and try and sell you something. Keep in mind that his success is with selling the product, the clients is with getting it to improve productivity and higher returns (much more stressful, especially if it is a fail). The last paragraph about having experienced process engineers is VERY true.

I once had an engineer who wanted to sell me better screen decks for our classifying screens. He demonstrated how we were fitting our present screens wrong and showed us how to fit them better and get more life from them. He came back a few days later to check on how they operated and of course they were working much better. This all without any promise of purchase. He then gave us his spiel with samples, references and was also able to recommend what type would suit our very abrasive material. He got his sale and was able to give good back up service.

And even today I would recommend his products. As a commissioning specialist nothing peeves you more than a vendor of new equipment that comes in, sets up the equipment and then leaves with little or no training or equipment run time.

What goes around comes around. The suppliers who get the return service are the ones that invest time into their product success.

Yet another interesting discussion I would like to stick my nose in. I can see a lot of viewpoints from a customer point of view, if you guys do not mind I am going to approach it from a salesman point of view. I fully agree though with the views expressed, as I worked as an Instrument Technician prior to getting into Sales, so I have also seen both sides of the coin.

New Technology is probably the most difficult thing to bring to the market, consultants wants the customer to specify and the customer wants the consultant to specify whether they can use it. Both want to see trials and no one wants to be a guinea pig, which leaves a terrible dilemma. That having been said, although slow progress, this can be overcome once you manage to get a customer that is prepared to test the new technology. Andrew did pose a question which I feel is a very valid and valuable question. Once you have proven the technology, the resistance seem to grow even more.

We have over the past five years jumped through just about every hoop that could be thrown at us to prove that we can provide a stable flotation circuit through accurate level measurement, which reduces bumping and re-working of product etc. Just the stability in the circuit would relate to at least a 1% improvement if not in recovery, at least in stability of the overall plant, which would downstream allow for the improvement (I believe) of at least 1%. Although I realize that this is extremely difficult to quantify without immense amounts of research money going into measuring this, the logic of actually having stability in the circuit, no dart valves jumping up and down and no re-working of product should in itself already equate to massive savings.

This having been said, we were fortunate enough to get a concentrator to install our product on their rougher cells. They were so impressed that they started optimizing their circuit with the stability they achieved. It is still not quantifiable exactly what their gains are, but the stability achieved have given them the grade they wanted to obtain and I believe with having a stable process, they can actually focus on the optimization and achieving the 1 -2 % increase that they would like.

Thus I do fully agree, from a Sales perspective, the salesman that makes promises that he cannot achieve is in my opinion the Salesman that should be chased of site, and there are plenty of them, even from my perspective I do not like them, but once in a while someone tells you that his innovation can bring value to your plant and it is important to allow these opportunity to also bring value to the plant and rest at night instead of call-outs.

This is a great discussion. I identify most with what you said about plants not having the basics right, and with earlier comments as well. Most of the operations I come across are running quite lean on process people, as well as maintenance resources. Thus, the plant staffs are often too overwhelmed just keeping the place running to have time to support any sort of serious plant testing effort. Regardless, I believe the low-hanging fruit in maximizing recovery is almost always doing the "easy things" right. The payback on properly operating your existing equipment (for example, replacing a bad instrument, or making sure an operator is actually looking at it) can be as high as what can be achieved by investing in any sort of trial or capital upgrade, and is practically free by comparison. But it doesn't seem that very many plant superintendents have the time to focus on these things, and their front line metallurgists and supervisors are too busy to tackle them if it's not made a priority for them (or too inexperienced to recognize them on their own). As my old boss used to say, "We’re too busy chopping wood to ever sharpen our axes."

When I go into a plant, I try to point out the low hanging fruit opportunities, firstly because I know that these issues can threaten the success of a plant trial were one to be done, and secondly because hopefully it helps demonstrate that I know what I'm doing and want the operation to be successful. Process Engineer first, sales second!

"Process Engineer first, sales second!"

As someone with a tech sales background, I heartily endorse this. Also, salespeople, you have to be patient - if you go for quick sales, you will, rightly, put the client off. Sales require trust; and trust is bread through sincere relationship; and in a business context, sincere relationship is expressed through honesty and delivering real value, even before the sale has been made.

As for getting trials, with the relationship in place, plants that have circuits with parallel cells are a good way to go; less stats are required, and differences, though still requiring statistical rigour to "prove" are not as difficult to perceive; sometimes you can see the differences in parallel cells almost immediately, and this is somewhat compelling.

And, when proposing trials, equipment, etc, and especially post-sale, don't just leave the client to handle it on their own - you need to be on-site. The client was already busy and stretched, and don't want to be stretched further.

Yes, all this means some investment and risk on the part of the supplier - whether goods or services - but that's just life, I’m afraid. Meet people where they are at, and, within reason, go the extra mile, provide relevant value (to them, not you!) and you will have greater success.

Great conversation!
You definitely have to be in the mining game for the long haul as a supplier, patience is key. As mentioned above, changing suppliers usually involves high risk for not that much reward - particularly in the reagentsgame where margins can be thin on the commodity items.

An answer I am still looking for - If we increased a plant’s production by 1% in recovery what would the effective increase in production be worth in Dollars? I am sure there are a few of you smart people out there who could work it out, based on some real life scenarios?

I'll have a crack at say flotation producing concentrate; and maybe someone can advise - (Increase in recovery as %)*(Concentrate produced per annum in MT)*(average metal price in MT) e.g. (1% increase of copper contained in concentrate)*(200,000 MT)*(7300)=$14.6 Million USD per annum.

However, I believe proving the 1% increase is a result of a certain variable is really the hard part to prove. Statistics certainly was not my favourite subject.

"Millions" of USD gross return is usually about right for a recovery increase of -1%, for the majority of mines. He is right in saying that proving via the stats is difficult; and that's only after getting adequate trial time and trial conditions.

Citation of performance improvement envelopes is, IMHO, a good way to promote a flotation solution - i.e. what tends to be the minimum improvement? This will help to alleviate an engineer's sense of risk at undertaking a trial.

Perhaps a preliminary step could be conducting bench-scale tests in-house or on-site using your solution vs. a control/base that mimics a given site's setup and performance (within reason)? Such an approach would reduce their sense of risk, and actual risk and sense of inconvenience; whilst also providing you and them with more relevant site data, which will be more compelling to them than a claim (obvious, I know, but still worth noting). I won't pretend to be intimately familiar with you sales process, but hopeyou don't mind me throwing around some thoughts.

Unfortunately, since the site engineers hear a lot of general claims from a lot of people, it's often the patient, carefully crafted and deliberate, site-specific sales process that means the difference between an otherwise good solution succeeding or failing.

Another approach could be for you to talk to other non-competing suppliers of flotation equipment and reagents; they are often also on the lookout for general improvement on the behalf of their site customers/clients, that would help improve the certainty of their accounts.

This is an interesting question. It would be nice to think that rational thinking would drive the situation, understanding risk and benefit, but many other things are at play. One day I will write a monograph on BOC Gases' experiences in this challenging space attempting to commercialise various flotation technologies (for example, Actifloat, Maxifloat and Cleanfloat). Before we started on this pathway, BOC Gases commissioned an international interview conducted by McKinsey and Associates who approached most of the flotation plants in the Western world and asked several questions, including what a 1 to 2% improvement in metal recovery would mean to them.

All were very interested and many would 'crawl over broken glass' to obtain this type of outcome if it were economic. Reality proved otherwise one site was not convinced by a 5% improvement in metal recovery statistically significant at the 99.99% level however would not let us remove the oxygen PSA plant from site. And we did have some good successes at the 1% level and people had the patience to stick it out. All a mixed with all sorts of things going on!

As an example of what happens, one potential application that didn't even get to trial was when a variation on marcasite was found in a Mississippi Valley type feed. Low dissolved oxygen levels are bad news for galena and plant recoveries were down 10%. Oxygen conditioning (Actifloat) was the obvious solution - analogous to lowering the oxygen demand of iron sulphide bearing gold ores prior to leaching. But we were told, the site's research money had to be spent wisely and there were many potential 'suppliers' on the radar screen so it went to AMIRA project to investigate mixing in flotation cells. Turned out that the individual had just been invited to sit on the AMIRA board. I was more worried about the shareholders unnecessarily losing revenue when such a simple and cost-effective solution was at hand. Note that the improvement in metal recovery argument is meaningless for a copper porphyry operation where tonnage rules the day and not metallurgy - unless there are issues like surface oxidation (sulphidisation), etc.

And it is not easy to demonstrate 1% or so improvement in recovery when so many other variables are at play - variations in mineralogy, grind size, reagent additions, operating strategies between shifts, and so on. Much easier and quicker to demonstrate a 5% or greater benefit in an operating plant. Of course sampling more often increases the amount of data available for analysis but it is trade-off between various resources - don't want to overwhelm the assay laboratory. Generally you do need chemical assays, however if a site has great confidence (and that means all of the decision makers involved in the assessment) in on-line assay measurements, then removes some of the sampling issues.

The benefit has to be reproducibly demonstrated on the bench scale before anyone will be interested. Then, as pointed out in another commentary, you really need a site champion and a management that is interested in a potential improvement. Application to the plant scale requires considerable planning and commitment (= money and resources). To assist in this process, I engaged Prof. Tim Napier-Munn in the early 90's to help us understand the design and evaluation of plant trials but that doesn't guarantee success

'Identical' parallel circuits is the best and 'quickest' way to get an outcome however as we discovered, there is rarely such a thing as 'identical' circuits due uneven impellor wear, difference in the levels of the flotation bank, uneven air delivery and distribution, unequal performances between level controllers and so on even differences in the particle size ranges [caused by 'imperfections' the splitter] in the feed tothe two parallel circuit (this problem is magnified if the feed for each flotation bank is from a dedicated milling circuit). One can refurbish the flotation banks, etcetera but at the end of the day there will always be a difference or 'bias' between the two identical flotation circuits. One can establish the nature of this bias, hopefully as a function of head grade, by examining operating normal data and including it in your analysis of the new application during the trial.

If you don't have parallel circuits, then you have to use the ON/OFF approach, and carefully monitor changes in the feed (grades, size distribution, degree of oxidation, etc.) assuming that the same operating conditions are being applied through the trial. By the way, there are some great champions out there, who support innovations and plant trials. Rob Dunne at Newmont comes to mind.

Just add small humble thing. Statistical Process Control system is a great help to back up your conclusion. Because we are looking at trends over time, and we deal with ore that is never homogeneous!

Great comments! I cannot speak for him, but I can certainly agree that our experience with grinding media industrial trials is very similar to what you have just discussed.

I am interested to know how many metallurgists out there have experience in performing industrial trials with the aim of obtaining a statistically significant outcome (either positive or negative). I look back at my early years as a graduate and a plant met, and my experience was limited, although I suspect better than most as I was exposed to 2 plant trials (one to trial an additional cleaner bank and the second a media trial). It was not until I started to work for Magotteaux that a focus on statistically robust test work become paramount. I consider this to be a failing of the our industry, especially when you consider that diminishing feed grades and ore availability will one day require the effort to achieve that extra 1%. 100 and 200,000 tpd mines are not sustainable and the prehistoric notion of tonnes is the only target will not last forever.

Like many metallurgists working with low grade Cu/Mo deposits I am always looking to optimize the profitability of the plant. It has been my guide throughout 15-20 plant reagent trials and 4 ball trials that the statistically meaningful net cost benefit is the first deciding factor. i.e. does the net new product value exceed the added cost to produce that new metal. Corporate goals of NPV and IRR are the final hurdles that must be met. Plant trials can be as short as a day (back to the drawing board) and as long as 5 months (for low grade ores with multiple ore types).

I completely agree with you. My question to you is, during those reagent and ball trials, did you simply look at the immediate effect of the change (whether it is grinding ball wear, or reagent addition rates) or did you also analyse the plant performance data and attribute an improvement (or a deleterious effect) to the change?

It has been my experiencethat statistics is often accepted when analysing ball wear data (as an example) however not to determine the changes in recovery (as it is generally thought that seeing a change in recovery, considering daily plant fluctuations, is not possible).

Ball tests were predicated on wear rates, quality (spauling), and incremental costs. Importantly, the trial should be long enough to have mill inventory turnover 3 or 4 times. Reagent trials likewise will have a high confidence level over 1 - 6 months depending on the complexity of the orebody. I think, considering most plants have a historical database, to include all variables in the trial spreadsheet would be wise.

After a month or so, and at intervals thereafter, run a cross correlation vs. the objective function(s) to see if variables other than the usual dosage, throughput, grades, pH, particle size, and % solids are having a significant effect.

I have only been a flotation operator for just over a year and have had the pleasure of working with some great Mets. One of the Mets Dr Matt had never worked with copper flotation (previous experience on sulphates) and the site had decided that the new ore from a pit 60kms away (which was oxide not sulphide) still at a high grade was going to be floated.With multiple techniques including setting optimum density, multiple reagents and my favourite lime! Lime was my secret weapon of choice to get the grade! But this was not the case in the trial.

It was a very long and trying week and the results we miserable and grade was terrible. Reagents include collector, frother, msbs and nash. Any ideas on how the trials could have been improved?

It sounds like the sulphidisation is not selective and to be effective, it does need to be added in a certain fashion i.e. selected Es or ORP. How did you add it? Secondly, why are you using SMBS? A real 'sulphide' type reagent. I presume that the lime was for pH (>7 to 10) rather than depression (otherwise you are suing as a 'sulphide' type reagent).

I guess you need to go back to basics - mineralogy, bench scale tests - have these been done? Be interesting to know what is in the ore and how fine you are grinding it...could be making lots of slimes which chew up reagents like it is going out of fashion. I presume you are using a strong xanthate collector. And what sort of frother - a sticky froth or 'dry' well-draining one e.g. MIBC.

This is a very interesting topic. It is always challenging to prove that a particular change increases recoveries by 1 or 2%. There is always a great amount of noise due to random plant variations. I am a big fan of histograms because they are simple to generate and read. I recently presented a paper at CMP 2014 in which I reported 1.3% increase in recovery. This I calculated by plotting two histograms, one with two years of data before changes and the other with two years after changes. The difference is visually evident.

Below is the reference:

Necessity Driving Change and Improvement to the Cleaner Circuit at Lumwana Copper Concentrator, 46th Canadian Mineral Processors Conference (CMP), Jan 21-23, 2014, Ottawa, Canada.