Computational Modelling '17 Conference Diary

The conferences in MEI's Computational Modelling series are the smallest, but most specialised of all our events, and attract a core of researchers working in this cutting edge field. Computational Modelling '17, the sixth in the series, was held at the St. Michael's Hotel, Falmouth, UK, from June 13-14, and below is my short report on the event, which I hope captures the flavour of the two days.

 

Tuesday June 13th

Stephen Neethling and Jon Wills

The conference was opened this morning by MEI's Jon Wills, who introduced the 33 delegates from 12 countries to beautiful Falmouth and Cornwall, after which Stephen Neethling, of Imperial College, UK, set the scene for the conference with his keynote lecture on the modelling of multi-phase minerals processing systems.

Minerals processing is characterised by complex multi-phase flows that present significant modelling and simulation challenges. Recent advances in not just computational power, but also the models, algorithms and implementations, especially in terms of parallel processing, have resulted in significant improvements in our ability to directly simulate these systems. Despite the computational resources available, the shear range of scales at which important phenomena occur at in minerals processing means that no one technique can hope to resolve all the complexity of these systems. This means that the approach used needs to be tailored to the system and scale being studied, and Stephen examined a range of different approaches that have been applied within the Imperial College group (more info: s.neethling@imperial.ac.uk).

In his paper "Decoupling the impact of rock properties and operational settings on minerals processing performance: a data-driven approach", Suriadi Suriadi of the Queensland University of Technology, Australia showed that mining operations record a large amount of data from multiple sources (e.g. block model, online processing information) which, currently, is neither effectively nor systematically used to understand and improve operational performance. He proposed a generic semi-automatable data analytics method, the Integrated Analysis Method (IAM) which addresses the disconnection between disparate datasets. IAM enables evidence-based understanding of rock and machine parameters, laying the foundation for a more sophisticated way to model and predict mining processes to deliver financial value (more info: s.suriadi@qut.edu.au).

Adrian Hinde is a well known face at MEI Conferences. Formerly with Mintek, South Africa, he is now an independent consultant. He discussed how, in mining operations, comminution or size reduction usually begins with the use of explosives to generate a run-of-mine product suitable for transport to the metallurgical plant. At the plant, the ore is crushed and milled in stages to liberate and concentrate the valuable mineral species. As a result of laboratory and pilot scale comminution tests it is possible to develop mathematical models of comminution equipment to guide the design, control and optimisation of production scale circuits. Data from these tests can be subject to significant levels of experimental error. Adrian looked at the application of statistical resampling techniques to estimate model parameters and their confidence limits, irrespective of whether the models are linear or nonlinear in structure. Consideration was also given to the application of these techniques for reconciling mass balances around comminution circuits with recycle streams (more info: adrianh@icon.co.za).

Following the coffee break Yuande Zhou, of Tsinghua University, China, asked what can we learn from single spherical particle breakage. The role of single particle breakage has been recognized to be of great significance in many industrial practices as well as in geotechnical and geological events. Many types of single particle breakage test methods have been proposed in the literature and widely used to determine the comminution characteristics of ore particles, such as the twin pendulum test, the JK drop weight test, and the short impact load cell test. Considering the randomness and heterogeneity of ore particles, the breakage distribution function has been more widely adopted for describing the size distribution after crushing or grinding from a statistical aspect. Yuande presented an investigation into the breakage performance of single spherical particle under quasi-static platen compression (more info zhouyd@tsinghua.edu.cn).

Tuande Zhou (centre) and Ahad Harzanagh prepare for their presentations

Vertically stirred mills are widely used for comminution in wet and dry conditions in numerous industrial applications. This type of mill has lower specific energy consumption (kWh/t) compared with the ball mill and part of this efficiency is due to the spectrum of energy involved in grinding media movement. The vertical mill does not waste energy with the lifting of balls, for example, and imposes impacts of lesser magnitude and greater frequency of collisions with the grinding media. Douglas Mazzinghy of the Universidade Federal de Minas Gerais, Brazil, discussed numerical analysis of vertical stirred mills scale-up using the discrete element method to obtain the spectrum of energy of the grinding media for vertical mill and ball mills (more info dmazzinghy@demin.ufmg.br).

Domenico Daraio, of Johnson Matthey Technology Centre, UK also discussed attritor mill- grinding media dynamics, showing how a full understanding of both media motion and medial ball-impeller-grinding chamber interactions are the key features to improve performance and/or product quality. He presented results of a study using the DEM modelling to provide information on the grinding media dynamics and the state of stress inside an attritor mill when changing impeller designs (more info domenico.daraio@matthey.com).

Elizma Ford, of Mintek, South Africa, also used DEM in experimental testing conducted in a batch stirred mill, to investigate the effect of bead size and stirrer speed on product particle size (more info elizmaf@mintek.co.za). 

Elizma Ford, with Janco Strydom, Matti Lampinen, Ahad Harzanagh and Stephen Neethling

Following the lunch break, Luis Cisternas, of Universidad de Antofagasta, Chile, discussed how the use of simulation in milling operations is a difficult task because there are uncertainties in operation conditions and mill model parameters, no papers having been published in the literature that propose how to consider these uncertainties. He showed that uncertainty analysis (UA) and global sensitivity analysis (GSA) can be useful tools in the identification of operational conditions for mill systems under uncertainties. A semi-autogenous grinding was used to explain the procedure, UA being used to study the effect of distribution and magnitude of the uncertainties of the input variables on the responses of the grinding process and control of the uncertainty of the significant input variables allowing control of the uncertainty in output variables (more info luis.cisternas@uantof.cl).

This was followed by two papers from Luleå University of Technology, Sweden. Par Jonsén said that modelling of wet grinding in a tumbling mill is an interesting challenge. A key factor is that the pulp fluid and its simultaneous interactions with both the charge and the mill structure have to be handled in a computationally efficient approach. He showed how the pulp fluids can be modelled with a Lagrange based method called incompressible computational fluid dynamics (ICFD) that gives opportunity to model free surface flow (more info par.jonsen@ltu.se).

In the second Luleå paper, Simon Larsson discussed an experimental and numerical study of granular flow using particle methods, and the application in handling of potassium chloride. He showed how advanced optical measurements using digital speckle photography were utilised to obtain the flow characteristics and to support the development of a numerical model of the material flow (more info simon.larsson@ltu.se).

As explained by Christian Ihle, of the University of Chile, slurry pipelines transporting a coarse –the comminution product– and a fine fraction, both in the presence of seawater, can cause an alteration of the liquid phase chemical composition, and he presented results of work on numerical simulation of cation exchange in fine-coarse seawater slurry pipeline flow (more info cihle@ing.uchile.cl).

Christian Isle and Michail Akritopoulos

In the final paper of the day, Chenwen Wang, of Lanzhou University, China, discussed the removal efficiency of industrial particles resulting from aggregation with different amounts of water vapour in a cyclone separator, demonstrating that improving wetting, and choosing the appropriate humidity conditions, in agglomeration and collection could be an effective way to collect ultrafine dust from industry (more info wangchw16@lzu.edu.cn).

We are enjoying glorious weather in Falmouth, ideal conditions for delegates to unwind at the end of the day, for the 3 and a half miles coastal walk to the Quayside Inn in old Falmouth, led by Jon (see also posting of 13th June).

Overlooking the Fal Estuary

Wednesday June 14th

Screening is one of the most widely used unit operations in mineral processing plants. In crushing circuits, the proper selection and sizing, optimization and their operation as efficient as possible is essential in terms of the performance and profitability of crushing circuit and the whole plant. Ahad Harzanagh of Hacettepe University, Turkey, discussed the effects of various design and operating variables on the efficiency of screening, investigated by means of DEM simulation of industrial vibrating screens (more info ahad.aghlmandi@gmail.com).

Wet high intensity magnetic separators (WHIMS) are used in magnetic separation of minerals with low susceptibility. The dynamic process of material built-up in the matrix is influenced, among others, by the matrix geometry, gradient and strength of the magnetic field. These factors, however, do change with the built-up of magnetic material in the matrix. Detecting the built-up of magnetic material is crucial to the continuous operation of WHIMS. Raheel Rasool of the Institute of Mineral Processing Machines, Germany, presented a numerical modelling approach for WHIMS. The electro-magnetic and the fluid flow field are modelled with the FEM, while the material particles are identified and evolved using the Level-set approach. Such a framework retains the influence of magnetic particles on the surrounding magnetic field and can be used to detect and predict material build-up in the matrix (more info holger.lieberwirth@iam.tu-freiberg.de).

Raheel Rasool (right) with Holger Lieberwirth

Jiangang Ku, of Fuzhou University, China, studied the magnetic flux density distribution of particles based on finite element analysis, to produce a simplified model for calculating the magnetic induction field (MIF) of an irregular magnetic particle. The results show that sphericity is the key factor affecting the MIF of an irregular particle, the calculated value of the MIF distribution using the particle volume radius becoming more accurate the higher the sphericity (more info kkcc22@163.com).

Luis Cisternas of Universidad de Antofagasta, Chile, presented what he said was a comprehensive review of all significant research applying computational optimization to flotation design problems. Specifically, the optimization algorithms, superstructures utilized, objective functions, cell and bank models, and the incorporation of uncertainty were considered, and the new knowledge generated using these techniques was highlighted (more info luis.cisternas@uantof.cl).

More on flotation after the coffee-break, with Pablo Brito-Parada, of Imperial College, UK, presenting a three-dimensional numerical framework for the modelling of the pulp phase in froth flotation. It is important to understand the dynamics of the turbulent three-phase flow in the pulp phase to design efficient flotation tanks. Computational fluid dynamics (CFD) has been used in the past for simulating the hydrodynamics of the pulp phase and flotation rate estimates have been calculated using these models for simple flotation cell geometries. However, the aforementioned simulations have made use of commercially available codes, which provide limited tractability, and have often considered multiphase models that are limited to monosized bubbles and/or particles. To address these limitations, the group at Imperial has implemented polydispersed flow models in an open-source finite element code called Fluidity (more info gaurav.bhutani@gmail.com).

The morning session finished with five papers on pyrometallurgy. Par Jonsén, of Luleå University of Technology, Sweden, discussed the numerical prediction of fracture in iron ore pellets during handling and transportation (more info par.jonsen@ltu.se).

Chris Pickles, of Queen's University, Canada discussed advances in the modelling of pyrometallurgical processes. Pyrometallurgical processes tend to operate at relatively high temperatures and thus equilibrium is usually approached. Modelling is an attractive alternative to performing both costly and time consuming pyrometallurgical experiments, and Chris reviewed some of the software available and its application to pyrometallurgical processes (more info christopher.pickles@queensu.ca).

The refractory linings used in smelting furnaces undergo cooling and heat-up cycles when the furnaces are shut down and restarted. Herman Kotze, of the University of Pretoria, South Africa, described the development of a 3D finite element model of a DC smelting furnace, to reach an improved understanding of furnace refractory linings under transient thermal and mechanical loads (more info ah.kotze@gmail.com).

Herman Kotze, Philip Schwarz and Matti Lampinen

Mineral industries require effective processes of both combustion and mineral transformation. Mostly, general purpose computational fluid dynamics (CFD) models are inadequate and interactive combustion and mineralization calculations are essential in reasonably predicting the gas species, temperature and flow fields as well as emissions and calcination/transformation levels of the minerals. Michalis Akritopoulos, of CINAR Ltd, UK) discussed modelling and experimental results from a Parallel Flow Regenerative Kiln (PFRK). The PFRK kilns are considered the most energy efficient in the lime industry and exhibit energy efficiency of about 85% in most cases (more info tahir@cinar.co.uk).

Rotary kilns are used in several minerals processing operations, as well as related industries such as cement manufacture. They have been found to be ideal for some roasting and calcining processes, as well as some other operations such as nodulization. Examples are alumina calcining, the ilmenite reduction stage of the Becher process, iron ore reduction, and pet coke calcination. Taking us to the lunch-break, Philip Schwarz, of CSIRO Mineral Resources, Australia, described a hybrid simulation methodology for rotary kilns including granular flow and heat transfer (more info phil.Schwarz@csiro.au).

Hydrometallurgy was the theme of the afternoon, and final, session of the conference. Matti Lampinen, of Lappeenranta University of Technology, Finland, stressed that hydrometallurgical reactor leaching is a multiphase reaction system, and research and development of reactor leaching faces many of the challenges typically found with such systems. In view of the complexity of the task, the use of sophisticated modeling and simulation tools is a valid approach for analysis of the important phenomena behind the leaching process, their interactions and relative importance. Matti presented modelling and simulation work that provides good premises for the development of leaching reactors, based on atmospheric direct leaching of zinc concentrates (more info: matti.lampinen@lut.fi).

Heap leaching is a major recovery method for extracting metal from low grade ores. Due to the long leach cycles and solution residence times it is desirable to simulate these systems for both control and optimisation purposes as the impact of an operational change on the performance is very slow. Due to the wide range of length scales involved, the effect of micro-scale mineralogical and textural properties have to be approximated and parametrised when modelling at the heap scale. Normally this is calibrated using experimental data obtained from time-consuming leaching experiments. Francisco Reyes, of Imperial College, UK, showed that by combining X-ray micro-computed-tomography (XMT) and an appropriate simulation environment, leaching rate kinetics can be assessed in a more comprehensive and time efficient way (more info f.reyes-leiva14@imperial.ac.uk).

In the final paper, Janco Strydom, of Stellenbosch University, South Africa, described the simulation of a high-pressure base metal leaching operation, including control layers and abnormal event detection. The critical control layers included sensors, actuators, regulatory controllers, alarm systems, safety interlocks and supervisory control. With the help of expert knowledge, a fault (abnormal event) library was incorporated into the dynamic model. With the use of this dynamic process model, fault detection and identification techniques can be more accurately evaluated for hydrometallurgical industry use. Furthermore, with the use of economic performance functions, an economic case can be made for the use of process monitoring in industry (more info lauret@sun.ac.za).

MEI's Amanda Wills closed the conference, and invited delegates to attend Computational Modelling '19, which is planned to be held in April 2019 in Cape Town, where the series began back in 2005.

A final Cornish cream tea in the hotel gardens

Draft papers presented at the conference are available from MEI Online, and authors have been invited to submit final papers for peer-review to a special Computational Modelling issue of Minerals Engineering.

Twitter @barrywills