Fundamental to any geometallurgical study is to take a variety of samples and submit them for standard characterization. So what should be characterized and how should the resultant data be handled?
There are four key groupings of data that may be considered for sample characterization. These are: chemical, mineralogical, comminution data and process variable data. This list considers only metallurgical response. In a future blog we will discuss use of environmental data in a geometallurgical program.
Firstly, it is worth commenting that it is considered that information saved about samples would typically be numbers that represent that sample’s characteristics (such as percent chalcocite or rate kinetic parameters of chalcocite flotation or Bond ball mill work index). Obviously, other data can be used to characterize a sample, such as text, graphs or pictures but given that we are contemplating mathematical manipulation and modeling of the data, it is therefore considered that the data would be numeric. Hence, it is important to define all relevant parameters that contribute to a rock’s response to metallurgical process.
Chemical characterization
This is always taken care of, as this is the fundamental information for deposit definition. However, some thought should be given to chemical characterization that may indicate some metallurgical process or response. For example, a copper heap leach operation will want to determine sulphuric acid soluble copper. A bacterial assisted copper heap leach operation would also want to characterize cyanide soluble copper. A flotation operation that is sensitive to amount of pyrite present in the ore may also be characterized for iron or sulphur assay.
Careful thought should be given to chemical characterization at an early stage of project exploration to ensure a comprehensive list of elements is included in the desired analytical suite. This will permit more detailed deposit analysis, if and when, the project reaches feasibility stage and detailed economic analysis requiring consideration of key geometallurgical parameters. A simple example would be a deposit that has zones of massive, banded and disseminated sulphides. Such a deposit will have distinct comminution and metallurgical response, due to variability in sulphide mineral type (massive, banded or disseminated). It would be easy to capture this key parameter with sulphur analyses. However, it is common to see that this analysis is not included in the chosen analytical suites.
Mineralogical characterization
Mineralogy is the fundamental bridge between geology and metallurgy. Characterization of key mineralogical parameters will underpin any geometallurgical study. In the future, we should see a whole range of mineralogical information being saved to the block model, much the same way as we treat chemical assays today. For now, the price on mineralogical characterization is still such that we need to be selective in what information and how much mineralogical information (samples tested) we should utilize.
Key modal mineralogical information should be captured and saved. If the deposit is a sulphide type, then key sulphide minerals present and their relative percent occurrence is vital information. However, equally important can be information on key gangue minerals that may critically impact metallurgical response. Examples here include significant occurrences of clays, sericite or graphite and amorphous carbonaceous material (these later two may also be characterized analytically).
Additionally, textural information is often relevant to the way ore responds to metallurgical process. It would be unlikely today (in 2010) to be able to save detailed liberation type data to a block model but indicators of this could include average grain size or parameters such as PSSA (a QEMScan data product – for particle specific surface area).
Comminution characterization
As the comminution process can represent up to half the capital and operating costs of a mineral treatment plant, therefore, its characterization is of critical project importance. As opposed to most other metallurgical processes, this is one area where there are a number of standard and industry accepted tests that are routinely used. These include tests such as the Bond Ball mill work index, the Bond Rod mill work index, the JK Drop weight test, the MacPherson SAG mill test, the SPI test, the Bond Crushing index, the point-load test etc. This is not a complete list of tests available. There are several references that one can find on-line for comprehensive tests in this area. The advantage of using these known and proven tests is that large data bases already exist for these tests with all types of ores. So one can readily characterize a new sample relative to these large databases (and determine that this sample is at such and such percentile point on the data base hardness index).
Process variable data
In order to characterize this, it is obviously fundamental to know what metallurgical process(s) will be utilized for recovery of the valuable metal in the ore. So, an ore that will be treated utilizing flotation will be characterized using some flotation parameter(s). It is possible to have ores that may be diverted to different process routes dependent on their characteristics. For example, several porphyry copper operations have multiple processes of flotation, acid heap leach and bacterial assisted heap leach in one integrated mine site. So, characterization in this case, needs to encompass sufficient parameters to incorporate all these options and facilitate ore process routing selection.
As we are submitting a number of samples for the same test, consideration needs to be given to the selection and design of the standard test. This can be simple, such as a Au cyanide bottle roll test. In the case of flotation the selection is more complex. Will the test be rougher kinetics or a batch cleaner test and what conditions will be applied to the test?
It is usually considered that we would want to characterize independent parameters. The question that often arises with process characterization is: “should we save recovery numbers (being recovery from a specific metallurgical process test)?” The answer to this question is “perhaps and depends”. For simpler types of flotation, it is probably preferable to save independent parameters such as, flotation kinetic parameters but for other, more complex and less understood flotation it may be necessary to save recovery information, together with concentrate quality information. For Au recovery from simple cyanidation process, it is probable that recovery will be saved, together with cyanide consumption and lime consumption. Thus, this is an area that needs careful consideration when planning geometallurgical programs.
After the data has been captured it needs to be entered into the block model with the appropriate sample spatial co-ordinates. The data can then be analysed using either zonal or halo type analysis or, using models. These will be discussed in future blogs.
It is has already been mentioned above, that generally the type of sample characterization data that is trapped will be independent type parameters (although not exclusively, for example when a recovery number is saved). It is important to realise that most of these data are independent parameters and may bear little or no statistical co-incidence with other data. For example, chalcocite content in a copper ore may show some or very little relationship to the copper assay of that ore. This may seem a somewhat trite observation, but consider a copper ore where its metallurgical response is influenced by its chalcocite content. In this case, it would be preferable to characterize chalcocite content throughout the deposit and not try to infer it from Cu assay.
Steve Williams
5th August 2010
About Sample Characterization
Blog #2 – about mineralogy
There are a number of reasons why geometallurgy is taking off as a field of study and endeavor now. One of the most important has been the emergence of quantitative, automated mineralogy (QEMScan and MLA technologies). Without doubt the ore mineralogy is a fundamental driver for its metallurgical response. It is the key, at the heart of how an ore responds when it is put through a fixed black box, called a processing plant. Therefore, being able to quantitatively analyse the ore’s mineralogy is very important to geometallurgical studies.
When we set about recovering metal from ore we begin this process by recovering the minerals that carry the metal(s) we seek. Typically, the first step in this road is mineral processing with flotation or gravity concentration or magnetic concentration etc. Yet we analyse ore deposits with metal analyses. Obviously, the more we can know about the nature of the metal in the ore – specifically, its mineral’s, the better we know about the challenges we face when we try to recover that metal. Hence , the importance of mineralogy. Mineralogical studies can tell us about mineral species present, their relative abundance, macro and micro texture, mineral associations, grain size and liberation size(s), and finally about the chemical aspects of the minerals (this, I will discuss in a future blog).
But further – from a geometallurgical perspective – a big part of the variability that a deposit may show will be variability of its mineralogy across the deposit. Consider pyrite. Pyrite (FeS2) is a mineral that has a strong influence on metal sulphide mineral flotation (as an example, copper minerals from a porphyry copper deposit). The variability the pyrite alone can show across a deposit, can be significant. We could see variance in the abundance of the pyrite in the ore. We can see grain size and / or textural variance across the deposit. We could see variance in association – the pyrite can be intimately associated with the valuable metal sulphide or not. We can see variance in the form of the pyrite from blocky or cubic pyrite, to amorphous type of pyrite (the latter being more problematic from a flotation perspective). We could also see variance in valuable precious metal content and its characteristic form, in the pyrite. Au is the most notable case – it could be refractory or finely associated with the pyrite. Finally, we can also see the pyrite carrying deleterious elements, such as arsenic. All of these variances will have an important impact on the metallurgical response and given that the pyrite occurrence will be heterogeneous across the deposit, we would need to understand this variance to fully plan our exploitation strategy and probable expected performance.
Then we add all the other minerals that will play an important part in the metallurgical performance (such as the variety of sulphide minerals and gangue minerals including clays and sericite)
How does quantitative, automated mineralogy help? The technology has a number of key advantages over conventional optical mineralogy:
- It is quantitative (or can be set up to be quantitative dependant on what one is seeking). The machine will pass through tens of thousands to hundreds of thousands of points of analysis. No human eye could undertake this.
- The technology is automated. Typically many samples are analysed though the night, with little supervision. The technology permits bulk handling of many samples at once which in turn has lead to cost per sample analysed coming down. This in turn has permitted much bigger mineralogical sample campaigns to be undertaken than before.
- The technology is machine and software driven. It is not subject to the extent of human error that optical microscopy can be. However, the technology is not infallible. Behind the technology is a library of minerals that has to be built up and can be ore specific for some minerals. This process requires human intervention to build the library and can have errors or imprecision, although typically not for most common minerals.
- The technology will like all computer based technologies continue to evolve – so it is inevitable that we will see future generations of this technology that will be faster and more task targeted than the current generation. This evolution will in turn permit us to analyse much more samples for mineralogy than we currently do.
So in conclusion, in order to understand key variability that exists in a deposit, one of the key investigations needed is to define the mineralogical variability that exists in the deposit. This is one of the foundation building blocks of a geometallurgical study of a deposit.
Steve Williams
14th July 2010
Blog # 6 – About Geometallurgical Models
Models are used widely in the minerals industry to describe or forecast a result. Firstly though, what do we mean by “model”?
An online dictionary defined many types of models. The two most relevant model definitions were:
“1. a standard or example for imitation or comparison.”
“10. a simplified representation of a system or phenomenon, as in the sciences or economics, with any hypotheses required to describe the system or explain the phenomenon, often mathematically.”
Other definitions provided, related to people posing for art or wearing clothes on a catwalk!
From this definition there are a few key points that one can take to understand a model
- It’s a standard or example
- Representing a system
- For comparison
- Can often be described with mathematics
To me, this describes a way of viewing a complex system (an ore deposit) in a simplified form that we may (but not always), represent mathematically. The model is built to help us understand something complex in a more simple form.
Why are models useful in geometallurgy?
In geometallurgical studies of deposits, we will collect a lot of information that defines key parameters of the variability that exists in the deposit (such as mineralogy, grindability, flotation kinetics etc). This information will come from spatially discrete samples and will be entered into the deposit block model at the appropriate spatial points. We will then infer these characteristics for all blocks, using geostatistical methods. So, we will end up with a lot of information characterizing the whole deposit block model. This then needs to be interpreted into results that are meaningful, such as resultant throughput, recovery and economic return. This is where mathematical models come in.
What is the right type of mathematical model to use?
Obviously there is no one answer to this question, as it is very case specific. However, there are some guidelines to consider in thinking about these mathematical models:
- Keeping it simple is always a good place to start. An example here, is a porphyry copper deposit that has a significant amount of oxide copper and copper flotation recovery is seen to be inversely proportionate to oxide copper mineral content. This would imply a very simple model.
- Look for the most dominant independent parameters and build a model around the key two or three dominant parameters. As an example, a porphyry copper flotation metallurgical performance may be predominantly driven by pyrite content, clay content and copper mineral type. It is conceivable that different parameters will also need to be weighted for their impact on calculated result.
- A corollary to the above is to be alert to mineral systems where the metallurgical performance is heavily dominated by one critical parameter. An example here is the carbonaceous mineral content in a sedimentary copper type deposit.
- Mathematical models can evolve as the project evolves and our technical understanding of cause and effect matures. It is probable, that a model used in a feasibility study will be different to a model used in the producing mine (it will probably be more sophisticated when the mine is in production).
- Always keep in mind the objective of the exercise. In feasibility studies, we are interested in optimizing plant design and forecasting annual production and ultimately IRR. In operation, we will be interested in optimizing production performance by mine plan manipulation. These objectives will inevitably require different models.
We have described mathematical models for metallurgical performance calculations. However, a model can be simply a framework or description of the deposit or some aspect(s) of the deposit. Geologists use geological models to describe the different types of mineralization that we find in Plant Earth (such as Kuroko type volcanogenic, massive sulphide or porphyry copper deposits). These models are merely descriptive and general, but they are widely used to aid geologists in exploration. As an aside – because of this, when we use the word model, it inevitably means different things to geologists and metallurgists. Geologists think about descriptive models whereas metallurgists think about mathematical models.
I think it is possible to see, in the future, the evolution of descriptive models for metallurgy that describe the type of metallurgical response that discrete geological / mineralogical domains will imply. This type of model will be useful for communicating quickly what we have and what that may imply in terms of metallurgical response, technology, product(s) and challenges.
Steve Williams
13th August 2010