There has been a lot of discussion recently, on this blog and on LinkedIn, of advanced models of mineral processing systems, how effective they are and how to convince operators of their efficacy.
 |
| Jan Cilliers |
A keynote lecture to look forward to at next years Flotation '15 conference is "What have models and measurements ever done for us?" which will be presented by Prof. Jan Cilliers, who leads the Rio Tinto Centre for Advanced Mineral Recovery at Imperial College UK.
Prof. Cilliers will show how the history of flotation has key moments when there were significant advances in understanding. It is notable that these moments are punctuated by advances in the theory or the experiment of flotation. These new techniques in modelling and measurement did not develop independently, but advances in one led directly to advances in the other. Subsequent application of these techniques resulted in improved industrial operation. It is clear that advances in theory and experiment take some time to move from the laboratory to the literature and on to the plant. It is also clear that while some advances have made a significant impact on industrial flotation, there is still much potential for further application.
 |
| Stephen Gay |
It is therefore timely that immediately prior to Flotation '15, Dr. Stephen Gay, an independent consultant who is committed to developing optimisation and simulation software and mathematical algorithms for mineral processing, will be running a 3-day Simulation Course. The course will cover all aspects from basic data accumulation and analysis to advanced software development. Delegates will gain insights into the metallurgical process via advanced mathematical techniques and be able to develop systems for high level decision making using modern techniques of computer simulation methodology.
.JPG) |
| Peter Amelunxen |
In another keynote, Peter Amelunxen, of Aminpro, Chile will ask why, after more than 100 years of application of the flotation process—one of the most important technological advances in the history of extractive metallurgy—we still don’t have a standard test procedure for measuring the floatability of minerals in a given flotation system. Most companies, labs or consulting engineers employ a different set equipment specifications, test procedures, scale-up methods. Why?
Flotation is a complex physicochemical mass transfer process, and it has been difficult to fully understand all of the underlying interactions. While metallurgists can also be complex and difficult, it turns out that this is not the reason for the lack of a suitable standard; rather, the challenges are more practical by nature. They include difficulties in quantifying, at the lab scale, the phenomena that occur in the plant; missing gaps in the phenomenological understanding of the flotation system; knowledge dissemination, particularly knowledge with respect to understanding and mitigating risk; budgetary constraints; and differences among the objectives of the various test alternatives. It is these challenges—along with the metallurgists’ desire to resolve them and get on with the job at hand—that have led to the differences in test procedures that we see in our community today. Peter's keynote will explore some of the key issues that need to be resolved before we can hope to see universally accepted flotation test standards.
So there is much to look forward to next year in Cape Town, but what are your opinions- what have models and measurements ever done for us?
"The journey is more important than the destination"
ReplyDeleteTrying to build models of a process or product, forces you to ask questions about this process/product that you perhaps wouldn't have asked otherwise. Even if your model, once in place, isn't very helpful because of uncertainties etc., you might have learned a thing or two, about the process or product you are trying to build a model of. Measurements is of course a part of this journey. Today I work with engineering simulation not related to mining/minerals, but this rule applies just the same.
Mats Lindqvist , formerly a specialist in cone crusher modeling at Sandvik, today engineering consultant at Avalon Innovation in Sweden.
This is the first time I am learning the simulation of processes especially the concentrator. I feel the models can open clear doors to problem solving and thereby easier to trouble shoot a problem in a process plant. I am eager and looking forward to learn more in process models
ReplyDeleteGabriel Luzendu, Master student in Minerals and Metallurgical Engineering at Luleå University of Technology, Sweden
Dear Barry,
ReplyDeletethese are extremely interesting announcement for the upcoming MEI Flotation conference. We, the 2011 founded Helmholtz-Institute Freiberg for Resource Technology (HIF) in Germany, are very dedicated in improving models using advanced statstics and gaining a better understanding in how to describe the physics and chemistry of flotation processes using advanced measuring strategies and working interdisciplinary. I believe it is very important that all engineers and scientists in the field start speaking the same language and finding common ground (ref. to Peter Amelunxens talk). This is so essential to finally being able to optimize this extremely complex process we all love so much. Be sure that the HIF will participate in the conference and the course of Stephan Gay.
Looking forward to a great event and thank you very much for your dedication.
Best regards from Germany,
Martin Rudolph, prov. head of the processing department of HIF
Great question / comment. If the modeling / simulation OR measurements don't lead to some change - in our plant or knowledge - they accomplish little to nothing!
ReplyDeleteSome useful applications (personal experiences):
1. Improved monitoring and stabilisation of grinding and flotation circuit performance at various operations
2. Improved understanding from modeling leading to dramatically different standard operating conditions and improved recovery.
Robert Seitz, Manager - Crush, Convey, Concentrator at Freeport-McMoRan Copper & Gold Inc., USA
From my experience, proper measurement, analyzed by qualified individuals, leads to useful models. Models not built on real world measurement often lead to costly mistakes. So my answer would be that I would reverse the question to read, "What have measurements and models ever done for us?" and then reply, at least in my case as an energy efficiency consultant with a mineral processing background, that they have allowed me to deliver tens of millions of dollars of cost reduction to my employers and now clients. But beware of models not rooted in real world experience.
ReplyDeleteTony Teske, Energy Consultant, USA
Barry, I am happy that you are shooting it straight.
ReplyDeleteI want to take a layman's view on this. First let us list the parameters we can measure accurately and the instruments and sensors being used for the same. Then let us look at the Models in circulation and see how many of them have the parameters I am talking about above.
This will give a picture of present status and future focus needed.
Rao,T.C., India
Thanks for opening up the discussion on measurement and models
ReplyDeleteThere are a number of issues regarding 'measurement'.
1. measuring the ore variables (i.e. solid flow, sizes, water flow, assays, etc)
2. measuring the operational parameters( feed rate, water addition, dense media, etc.) I use the word 'parameter' here to differentiate from ore variables.
Many years ago a number of the then JKMRC staff recognised the problem that the models for simulation use model parameters that are not operating parameters. Worse still some models have dual function acting as both a design simulator and an operational simulator - rendering the interface confusing.
I discuss these problems in my simulation course (which is workshop-style rather than a lecture). I briefly review some of the unit models and explain that there are many practical issues. This is one reason that simulation systems must be extensible to allow new models (particularly those directly related to operational parameters) to be included.
I am moving toward a machine learning approach (logistic regression) based on operational parameters and ore data. Using data-driven models is more logical than using models that may not be suitable. A presentation I recently gave at Las Vegas IMEX is available on request (StephenG@MIDASTech.net)
This presentation:
provides an overview of the simulation course I will give in MEI Flotation 15.
briefly discusses data-driven machine learning
discusses a new patented method using ore variability to advantage
understanding ore variables in greater detail than what is measured.
I was encouraged by Martin's comments; and hope many great thinkers, industry leaders, researchers and mineral processing professionals will participate in the course as well. Such a discussion is long-overdue.
I spent many years developing an advanced CFD based model of heap leaching. I also worked (less successfully) on a grinding model and I am currently a CFD consultant.
ReplyDeleteI find many of the above comments very interesting, especially the comment about the journey being more important than the destination. A good model requires good data, and model development can very easily highlight areas where available data and knowledge just are not good enough.
The model I produced was used by our commercial partner for some years, and it helped inform best practice. It wasnt cheap to develop but economies of scale meant that small improvements paid for the costs many times over. I could run a simulation of a process that had a timescale of months in a few minutes.
What I had hoped for that model (and its something that should be considered in the development of many of these models) is that it could be developed into a training tool for new engineers. Its a very cheap way of helping new engineers understand the consequences of changes to their processes
There are several on line grinding, flotation and thickening models to predict the behaviour of these processes. These on line models require maintenance. The Cloud is now allowing access to the process data and the models with the right security and capabilities to perform the required maintenance in these remote locations.
ReplyDeleteCompetences are the key support and services these powerful techniques. It is just like the NASA Center in Houston. Now, it is available for everyone.
The mine and mill data are now available to obtain the models parameters required for the online models.
Osvaldo Bascur, OsiSoft Inc, USA
Recently I was talking to a Company and they tended to agree that with appropriate models, and measurements there could be efficiency increases of about 8%.
ReplyDeleteThis was their target, whereas I was using a benchmark of the more conservative 4%.
However using their value of 8%, they claimed 6% would be achieved via quick decisions.
That is, 80% of the potential increase in efficiency is due to rapid ability to change operational parameters.
So I support Osvaldo's comments regarding the importance of online measurement and rapid response systems.
However, the point still remains that the model reliability is also important; similarly there will become more dependence on machine-learning algorithms (i.e. data-driven models)
We have found modeling to be very useful in early assessments on the hunt for optimization. All of our models have been developed in house, as they are adjusted frequently as data is collected from actual results. Mostly they are, for us, a starting point to determine a direction of adjustment.
ReplyDeleteOur modeling is for media size selection to meet mill discharge particle size targeting only. These models have a proven record to have a great deal of accuracy for this application. I do accept that our modeling is far less complex than floatation modeling, as the initial data we require are always hard numbers.
Mark Addison, General Manager - Sino Grinding (Americas) Inc.
What have the Romans ever done for us?
ReplyDeleteThe aqueduct.
And the sanitation!
Irrigation...
Medicine... Education... Health...
Yes... all right, fair enough...
Michael Young, Glencore XT
I think Barry's comment was meant to be semi-rhetorical; and therefore providing opportunity for people to discuss how mathematical models are used in Minerals Engineering.
ReplyDeletePerhaps the question was too broad.
Anyway, I was pondering the question, and Michael Young's response, and considered what was the first applied mathematical model.
The earliest I could find was about 600 BC with Pythagoras' theory of music.
I would suspect that some of the Babylonians may have well have used mathematical models for irrigation. (subject area pointed out by Michael)
However I think the deeper issue, and perhaps what Barry is eluding to is that in our industry there may be many practitioners who do not see that mathematical models are of value.
I had an alternative discussion in Mineral Processing Innovation and asked the simple question 'How many mineral processing engineers can write a computer program'. Naturally the response varied, but many claimed that basic software skills served no purpose to their jobs.
As software is a way by which mathematical models can be applied, I think it is fair inference that those same practitioners did not see the value in implementing mathematical models (other than those that were not already available to them by existing software or consultants.)
One of my pet hates is where a position is advertised 'mathematical modeller' and the requirements do not include much maths. That is a mathematical modeller is someone who implements an already existing mathematical algorithm (or set of algorithms) using a packaged software system.
But at least these people know that what they are running is based on maths.
The main problem I see (and please disagree) is that many people can run complicated models, but have little understanding, nor are required to have understanding of the basis of the models. Thus we have 'ignorant experts'.
My favourite quote in a course I ran on simulation was a participant who asked halfway through the course "by the end of the course will I be an expert?"
This is something I see all too often. People do one week courses and then claim to be experts.
And job advertisements enforce this concept. They don't ask 'do you understand the subject area?' but are you skilled in running the following software....?
Stephen Gay, Australia
There are often real problems with understanding what models can and cant do for you and in interpreting the results. Its all but impossible to get completely accurate results, and so people who dont like modelling can always shoot holes in the method, but used properly it can be very useful.
ReplyDeleteA simulation is just like a real experiment, it is constrained by the assumptions/environment we use to build it,
Having now delved deeply into both Mineral Processing (20 years, gravity separation in particular) and Quantum Physics (10 years), I can confirm to you the harder problem is modelling Mineral Processing.
ReplyDeleteAndrew Jonkers, Australia
Why do you think this is, Andrew? Maybe all the interacting variables in mineral processing operations?
DeleteTwo problems - one practical, one theoretical.
DeleteThe theoretical one is the mathematical challenge of adequately solving the Navier-Stokes in all flow regimes in the face of a solid particulate phase that adds dynamic interactive boundary conditions to the fluid phase.
The practical one is simply one of empirically defining what a particle is! Each particle is unique in composition, shape, porosity, density, and surface properties.
The solution I believe to both problems as far as mineral processing is concerned is not to model the particles themselves but to correctly measure and define the statistical distributions of the particle properties, and then model how these statistical distributions flow through a real device and split into various output streams.
I would like to add my experiences in mineral processing in Australia and modelling gravity separation devices over a 15 year period, 1987 to 2003. I write in general about the “normal” industrial context accepting there are many worthy counter examples.
ReplyDeletea) In academia too little time is left to read other papers exploring new process understanding - the research time scale has shrunk from 5 to 3 to one year and below meaning experienced people are drawn to “consulting research”, and PhD and masters topics cannot be supported or supervised adequately. This has broken the research link between academia, students and industry, of which models were an important part of that technology relationship.
b) Every so often (10-20 years) a decent book or monograph was published that distilled improved knowledge in an easy to understand form. These are well used, but the increasingly private nature of the consulting means such contributions are increasingly rare and no longer comprehensive. Most of my work remains as confidential reports/software to individual companies. Given company turnover, they are quickly lost and never seen in the public domain. What a waste.
c) Mathematical/Computational Models, rather than driving knowledge, should actually be a practical mechanism for summarising and evaluating data. They should encapsulate the most important part of the Scientific Method (confirming theory matches data AND predictions of data). In later years, nobody I knew of had enough research money to come close to adequately empirically verifying such work. Models were often published as likely hypothesis and extrapolated outrageously beyond their defined domain; with many poor outcomes.
d) Nonetheless, the useful state of gravity models did slowly advance. I eventually stopped further development because even routine ore character information needed to drive the models was unavailable; the models were better than the data needed to drive them. The pressure to "estimate instead" was unstoppable (IF the ore was this, what would your model say!). Good grief. Plant designs to sample and measure real time ore character to real process efficiency advantage are proposed by competent engineers and routinely rejected by accountants. (Capital trumps operational efficiency)
e) The out-of-control growth of the safety bureaucracy after a truly great and needed industry reform by the early 90's means ANY non-routine work on site empirical/verification work is now all but impossible to undertake. I know many experienced consultants and researchers who actively turn down value work because compliance costs are just too high.
f) Even simple, robust and completely verified models like the Bill Whiten model based controller are ignored/forgotten by the industry in situations even where they can make a profound difference to plant performance.
g) The modelling knowhow is increasingly held by equipment suppliers, who market products to an industry already sceptical of taking design advice from competing suppliers as compared to independent design engineers (the latter are now either taxi drivers in the current climate and not prone to return to the industry which has turned its back on them in the hard times, or like me went to the equipment dark side.)
--------------------------------------------------------------------
In summary there is an impressively complete disconnect between industry and academia in Australia with no sign of a turnaround. Models should be core value to the heart of this collaboration.
I weep in despair at what might have been. I received several awards/commendations for routine work but when I presented what I consider my best work in the gravity modelling game, the state of understanding was such that the only feedback I got was ‘Gee that would make a great screen saver”
Regards,
Andrew Jonkers
Mainly response to Andrew.
ReplyDeleteBased on my knowledge of your work, I understand exactly where you are coming from, but would disagree on a few key points.
Firstly I remember your presentation at JKMRC; it was very impressive. I also remember the comment about 'ScreenSaver' which was actually meant as a humorous compliment.
The main issue I see is that you may have had a perception that the audience were the gatekeepers to implementation. This was not the case.
As you are probably aware, I assume very low success probabilities.
i.e. 10% of PhD students may actually do something original. 10% of these may have opportunity to show their work (to an audience made up of potential implementors; and about 10% of these might actually see implementation. So about 0.1% of PhD students actually provide true implemented innovations.
Given (my view of low probability) then it means that developing a good idea is only a single small step. PhD students and other researchers need to be trained in implementation strategies.
It is a cruel world we live in; and you can almost dismiss immediately the industry providing any assistance on this. Similarly the Unis and other research institutions should be helpful. but aren't. Instead they will bring in Business Develop Managers. And herein lies a major disconnect, BDMs often have little passion in the work, and seek easy money products; so the passionate researcher is left no closer to his objective.
Therefore the only alternative is for the passionate researcher to go independent. This is what both you and I have done and many other ex-researchers. Many have been very successful; and many are struggling; but at least they are trying.
The Australian Government up until last year had Commercialisation Australia providing assistance to early innovative startups. The current Govt. abolished it, and in theory a new granting system will appear in November (fingers crossed).
Once you go independent; even though the world is cruel at some levels; it is also very helpful at other levels. And in time you can develop a totally different set of contacts from what you would obtain through conventional University environment. It is amazing how many people will actually help you.
So let us not despair about the lost opportunities of the past (which are totally valid) but find the true mechanisms for moving forward.
Further discussion on this posting is also taking place on LinkedIn
ReplyDeleteBarry,
ReplyDeletePl keep the discussions on Modelling and Measurements going as long as you feel necessary. For me, so far, this is one of the most important topics of the future. I am extremely happy the way experts and knowledge seekers are expressing themselves freely and frankly which in turn is showing all facets of the issues concerned.
I am sure that at the end, we will have so much information which I am sure will set the road map for the future direction of work to the academicians/R&D personnel and would also enthuse youngsters to see the large knowledge gaps and the challenges which lie in Mineral Processing; make the profession more exciting.
My request to you, Barry, is to summarise the salient points from all that is being expressed(only you can do it nicely) as and when the discussions end.
Thank you again, Barry for all you are doing.
Prof.(Dr.)T.C.Rao, India
This discussion is very personal for me. I chose computer modeling in chemistry as my specialization in Moscow university 36 years ago. I am still doing this, but now in Canada. I must admit that the dream to design the entire mineral process plant based on computer simulation is still mostly a dream. I am not talking about material and energy balances that are mandatory for any process design. An average engineer using METSIM or SysCAD or else can do it. All engineering companies that I worked with claimed that they do plant optimization based on the modeling results and save a client whatever % of capital and operating costs. And there is certain truth in this statement. But material balance should be only a beginning of the plant design and optimization.
ReplyDeleteOnly recently I started see a beginning of dynamic modeling in mineral processing design. But I did it 30 years ago too.
Your discussion is very important. I would like to see some good or bad examples of mineral process optimization/design based on computer modeling.
Another important topic that was discussed is that the equipment manufacturers are often the main developers and users of advanced computer models. These models are often companies secret as well as all test data. A design engineer should reinvent these models when compiling different pieces of equipment in the plant design. Wouldn’t it be nice if equipment manufacturing companies will sell the computer models of their equipment?