INTRODUCTION
Modeling is a fundamental and quantitative way to understand complex systems
and phenomena; simulation is complementary, as the experiment to theory.
Together they make an approach that can deal with a wide range of physical
problems and at same time exploit the power of large scale computing. All
successful models unashamedly distort the inessentials in order to capture the
features that really matter. At worst, a model is a concise body of data. At best, it
captures the essential physics of the problem and it predicts behavior under
conditions which have not yet been studied (1).
Within the above principles De Re Metallica Company (DRM) has been developing
and applying mathematical model tools to solve many industrial problems, specific
engineering tasks as well as to develop new solutions to modern hydrometallurgy
along the last twenty years. This paper should not be intended neither as a review
on modeling methods nor as a review on hydrometallurgy modeling but as some
relevant contribution to the state-of-the-art in the hydrometallurgy modeling field.
APPROACH BASED ON EXPERIENCE
Why the models are necessary and how to build them? The modeling process
starts with a motivation or requirement within a given scope and boundary
conditions, often of physical and/or economical nature. Fundamental question on
the essential features must be set up at this level. Accordingly, specific
experimentation is planned and performed, usually at small scale level to save
time. Experimental results are then analyzed always keeping in mind a physical
picture of the phenomena and the essential variables with regards to the model
objectives. For instance, if we are dealing with a particulate system changing
someway within a reactor and the objective is to find out optimal conditions for a
standard optimization task, then probably the best approach is the population
balance description provided with appropriate constitutive equations and boundary
conditions. But if we are interested in changing the design of the reactor, then
extended DEM modeling is the best choice. This overall modeling approach is
illustrated in Figure 1.
First proof to our emerging model is to reproduce same data used for modeling.
Lack of flexibility to follow relevant tendencies or biased results should be avoided.
Linear and non-linear behavior should be early identified in order to assure
predictability.
Next step is the scale up of results from the experimental level to the industrial
level of interest. Physical similitude rules must be identified and consequently
tested. The overall scale up procedure should be fully validated often using scale
down strategies. A robust scale up procedure determines robust predictability.
Once the above-mentioned steps have been overcome, we are in conditions to
simulate. Most of the simulations are directed toward: (i) Development of new
process or equipment; (ii) Solution to mass and energy balances for engineering
applications and (iii) Process optimization and automatic control.
Some of the models developed by DRM along the last twenty years are briefly
presented below.
Blasting Model: DRM.Blast
The most popular blasting model is the Kuz-Ram (2) which is an empirical equation
based on the Kuznetsov-Rammler distribution function. DRM applies the
population balance approach with two simultaneous fragmentation phenomena
within a batch frame. Model parameters are linked to rock mechanic
characteristics, blast design and operation conditions. A particular approach to
predict the fines based on reverse-simulation was developed. ROM size
distribution and energy consumption are well predicted. The DRM.Blast code is
currently set in the Vulcan platform at Minera Spence and Minera Gaby.
Crushing Plant Model DRM.Crush
DRM.Crush model is based on fracture kinetic approach plus mass transport for
crushing. Tromp curve and mass transport for screening and mass transport for
conveyor belts are also considered. Again, model-parameter are linked to ore
characteristics, design and operation conditions. Mass balance per size and
energy balance are well predicted within the real capacity constraint. This model is
available in a code working in tandem with the blasting simulator and large periods
can be simulated. It has been used in most of the Chilean hydro plants.
Agglomeration Model DRM.Drum
DRM.Drum is based on population balances with a specific agglomeration kinetics;
two-phase transport approach within the rotary drum is considered. Model
parameters are related to drum design and operation conditions. Agglomerate size
distribution and drum capacity are well predicted. The model is complemented with
a hydrodynamic characterization of the resulting agglomerates leading to identify
permitted irrigation rates in the leaching stage. It can also be used for drum sizing
and automatic control applications. It has also been extensively used in most of the
Chilean hydro plants.
Permeability Model DRM.Flux
DRM.Flux is a model based on hydrodynamics theory through porous media of
variable saturation (3). It includes geotechnical constraints which allow realistic
design of the initial pad soaking. Parameters are linked to soil-mechanics
properties. It provides the calendar to practice optimum wetting ramps as well as to
forecast the industrial irrigation response. It has been extensively validated in most
of the leaching plants in Chile.
Acid Leach Model DRM.AcidLeach
AcidLeach model is based on liquid transport in porous media of variable
saturation, with a particular solution to the Richards/van Genuchten equations (4).
Copper and acid transport, are modeled using dispersion-advection equations.
Different circuit arrangement run in closed circuit with SX-EW plant model can be
simulated. Effluent rate, inventory, solution composition and copper production are
daily predicted in a dynamic way. This is the most used DRM model, either for
engineering problems as well as for optimization and control in existing plants.
Bioleach Model DRM.BioLeach
BioLeach model is based on liquid transport in porous media of variable saturation;
Copper, ferrous/ferric iron and acid reactive transport are fully described by
dispersion/advection equations. Pneumatic air transport with oxygen and carbon
dioxide exchange with the liquid phase are also included; heat transport includes
losses and generation by the leaching of sulphides; strain colony growth with birth
and dead constant rates depending on the changing dynamic conditions are
included. BioLeach has been used at Minera Escondida since the start up in 2006
up to now. There are few similar-level bioleach models worldwide (5).
Chloride Leach Model DRM.ChloroLeach
ChloroLeach model is based on liquid transport in porous media of variable
saturation. Reactive transport of cuprous/cupric copper and acid are modeled by
dispersion/convection equations; again, pneumatic air transport with oxygen
exchange with the liquid phase is included as well as heat transport including heat
generation by oxidative dissolution of sulphides. Stability of cuprous ion in highchloride brines makes possible the chemical oxidation of sulphides. It has been
successfully used at column and industrial heap level at Minera Spence.
Salt Heap Leach Model DRM.CalicheLeach
CalicheLeach model is based on liquid transport in porous media of variable
saturation (Richards equation); thermodynamic and kinetic approach are employed
to predict iodine, nitrate and potassium dissolution rates. Model parameters are
related to pad design and operation conditions. Prediction is excellent as shown in
Pampa Blanca and Nueva Victoria SQM heap leach operations.
SELECTED APPLICATIONS
Simultaneous Blasting and Crushing Optimization
Units in the orebody geometallurgical model are first re-classified according to
physical quality associations (ACF in Spanish term). Experimental size
distribution measurements are determined by image analysis techniques. It is well
known this technique cannot see the fine fraction intended as particles less than
1 or 2. This is a strong limitation in hydrometallurgy as this material determines
the physical response of the heap leaching. DRM determines granulometry from
direct blasting simulation and also by reverse simulation from the crushing plant
thus matching the right answer. Figure 5 illustrates this dual approach strategy.
Figure 9. Mass balance per agglomerate size and dual flow throughout the drum.
depends mainly on the initial liquid contents. Then the optimum moisture for a
given system can be identified as the lowest apparent density when moisture levels
are swept. Actual position of this optimum depends on particle size distribution and
plastic deformation. This relationship is illustrated in Figure 15.
It is interesting to note that the scale up factor is here a result instead of an input
to the calculation.
Very Interesting, But How Does It Work?
Some time ago experts from an international copper company came to Chile and
asked how reliable can be this scale up procedure applied from 1 m height and 1
ton cribs to industrial strips, 18 m height and 7 millions tons each. The experts
suggested to run a new validation proof consisting in reproducing all strips
containing more than 80% of a given ore along the whole life of the project, starting
from a small crib in which same ore was processed. More than 10 seven-millionsstrips were well predicted from the 1-ton crib as illustrated in Figures 16 and 17.
Figure 20. Real and estimated copper concentration in effluent industrial flow.
Drm.Flux: Predictor For Soaking & Irrigation Calendar
Modern leaching practice recognizes the initial soaking as a key step to reach
optimum copper extraction results. DRM.Flux model provides an anticipated
calendar to apply not only the initial soaking ramp, but also the application rates
possible to apply within the geotechnical constraints intended as avoiding strong
ponding events as well as any mechanical failure (7). Excessive and optimum
soaking rates drive to worse and optimum copper recovery as shown in Figures 21
and 22.
and (iii) physical boundary constraints related to pad geometry and dimensions.
This is illustrated in Figure 23.
Figure 25. Experimental drop size distribution exiting the mixer in a VSF and a
conventional mixer.
understand and control these emergent behavior patterns will provide the
foundation upon which to build complex systems tomorrow (14).
Continuum mechanics and discrete element modeling applications will continue
growing; the first one focused in the design, optimization and control of existing
industrial processes and DEM or XDEM oriented in the short-term toward research
and innovation in process and equipment. Within the next decade social-like
communication networks will be strongly spreaded into the mining industry and
cloud computing solutions WEB-supported will be standard practice as illustrated in
Figure 32. DRM and WiseConn companies are currently starting up a new
business oriented toward supplying WEB monitoring and automatic control
solutions to the overall mining industry.
REFERENCES
(1)
(2)
(3)
Layton, W. et al., Coupling Fluid Flow with Porous Media Flow, SIAM J.
Numer. Anal. (2008), Vol. 40, N 6, pp 2195-2218.
(4)
(5)
Dixon, D.G., Heap Leach Modeling The Current State of the Art,
Hydrometallurgy, 2003,, C.A. Young et al., Eds., The Mineral & Metals and
Materials Soc., Warrendale, PA, pp 289-314.
(6)
(7)
Menacho, J.M., Troncoso, F. and Domnguez, C., Flux: Predictive Tool for
Optimal Irrigation in Heap Leaching Operations, presentation to
HydroCopper 2009, Hotel del Desierto Enjoy, Antofagasta - Chile, Mayo,
2009.
(8)
(9)
Leahy, M.J. et al., 2007, A Model for Heap Bioleaching of Chalcocite with
Heat Balance, Mesophiles and Moderate Thermophiles, Hydrometallurgy,
Vol. 85, pp 24-41.
(10)
(11)
(12)
(13)
(14)